Data Physicalization on Visualizing.JP

Shape for the Item, Count for the Quantity: Mesopotamian Clay Tokens, as Schmandt-Besserat Reads Them

Sun, 01 Nov 2026 00:00:00 +0900

In the decipherment by the researcher Denise Schmandt-Besserat, clay tokens are records that expressed the item by shape and the quantity by count. In her reading, one cone stands for one small unit of grain, and two ovoids stand for two jars of oil. She explains that the tokens were used for calculation and could be organized by lining them up in columns and the like, and she sees the marks pressed onto the surface of envelopes as the beginning of writing.

Clay tokens is a general name for small clay counters excavated from prehistoric sites in West Asia. The collection dataphys.org lists them as “5500 BC – Mesopotamian Clay Tokens,” but Schmandt-Besserat dates their appearance to around 7500 BC and gives their range as Syria to Iran. Her decipherment has also drawn criticism from archaeology. Below, I read it alongside her writing hosted at the University of Texas at Austin and the Louvre’s records.

Cones for Grain, Ovoids for Oil: The Convention of Shape and Count

Schmandt-Besserat’s main works are the two-volume Before Writing (1992) and How Writing Came About (1996). In a 2017 conversation, she says she came across small cones and similar objects in the storerooms of Middle Eastern museums, and saw the shapes of the tokens everywhere in the published volumes of the archaic texts from Uruk. In her decipherment, the basic mechanism is this.

The fundamental principle of the token system was the substitution of a small clay counter for each unit of goods to be counted.

In other words, one small clay counter stood in for each unit of the goods being counted.

Six tokens in a horizontal row, from the left: a small lump, a tetrahedron, a cone, a flat disc, a sphere, and a black sphere

Source: University Museum, University of Pennsylvania / Denise Schmandt-Besserat, “Tokens: their Significance for the Origin of Counting and Writing”

The photo shows tokens from Tepe Gawra in present-day Iraq, dated to around 4000 BC. According to the figure’s caption, the cone, sphere, and flat disc are the small, medium, and large units of grain, and the tetrahedron is a unit of labor (with a question mark: “one man-day?”).

Quantity was expressed one-to-one.

Two jars of oil were shown by two ovoids and three jars of oil were marked by three ovoid tokens.

Two jars of oil are two ovoids, and three jars are three ovoids. Oil is counted only with ovoids. Size also carried meaning: she writes that many shapes were made in two sizes, about 1 cm and about 3 cm in diameter, to express a difference in amount.

The table below summarizes how data corresponds to the properties of the object, drawn from her writing. The ancient side left no legend; everything here is her decipherment.

Data	Property of the object	Example (Schmandt-Besserat’s explanation)
Item	Shape	Cone, sphere = grain; ovoid = jar of oil; cylinder, lenticular disc = livestock; tetrahedron = labor
Quantity	Count (one-to-one)	3 ovoids = 3 jars of oil
Large vs. small unit	Size	Many shapes come in two sizes, about 1 cm and about 3 cm
Quality, subcategory	Incised lines, dots	Discs with different numbers of parallel lines = quality of textiles
One debt	Container, string	Sealed inside a hollow envelope; pierced and tied with string
Contents of the envelope	Impressions on the surface	Cone → wedge-shaped mark; sphere → circular mark
Classification, totals	Arrangement (columns, above/below)	Said to have been divided into columns by item, income and expenditure, and donor and recipient
Calculation	Moving, removing	The four arithmetic operations

Her own texts also waver. One page treats the flat disc as the largest unit of grain, while “From Accounting to Writing” writes “a disc for sheep.” The table is not a settled dictionary; it is a reading assembled by a researcher.

From 12 Types to 350, with Incised Lines Distinguishing Quality

According to Schmandt-Besserat, the roughly 12 shapes of around 7500 BC grew to roughly 350 by around 3500 BC. She explains that this was to count the goods made in the workshops of the cities. In the urban period of 3500–3100 BC, complex tokens appeared with lines and dots incised on the surface, and the lines and dots added information about quality.

Seven complex tokens with different shapes and lines, including a rectangle with incised lines, a crescent, an ovoid, and a disc with a cross

Source: Musée du Louvre, Département des Antiquités Orientales / Denise Schmandt-Besserat, “Tokens and Writing: the Cognitive Development”

The photo shows complex tokens from Susa in Iran, dated to around 3300 BC. The caption reads the top row, from the right, as sheep, jar of oil, metal ingot, and garment, and the bottom row, from the right, as garment, unknown, and beehive. The “?” attached to the one unknown piece shows the nature of the work of decipherment.

On their use, she is explicit.

There is no evidence that tokens were used for trade. Instead they were central to administration.

The claim is that there is no evidence of use in trade, and that the tokens were central to administration. Her theory is that they were used to manage a redistributive economy, in which the community’s surplus was collected and handed out again at festivals and the like.

Put in the vocabulary of visualization, this can be read as follows: shape corresponds to category, count to a quantity obtained by repeating the unit symbol and counting, size to the level of the unit, and incised lines to a subcategory.

They Could Be Moved by Hand to Calculate and Arranged in Columns

Tokens were tactile records.

Data processing with tokens was tactile. The counters were meant to be grasped and manipulated with the fingers. Addition, subtraction, multiplication and division of quantities of commodities were done by moving or removing counters by hand.

They were meant to be picked up and moved with the fingers, and the four arithmetic operations were done by moving or removing counters by hand. On how they were arranged, her 2013 paper “Tokens and Writing: the Cognitive Development” contains the following.

With tokens, the Near Eastern accountants could organize the budget for a festival in columns according to – Types of goods – Entries and expenditures – Donors or recipients

Goods symbolized by tokens could be lined up according to their relative value – Large units above – Small units below

The statement is that a festival budget could be set out in columns by item, income and expenditure, and donor and recipient, and that units of higher value could be placed above and smaller units below.

The word used is “could.” It is not a record that they were excavated lined up in columns. The footnote at this passage cites research in linguistics and the history of mathematics on counting and measurement (Justus, Høyrup). In the writing of hers that I read, in the Louvre’s records, and in the review discussed later, I found no description of any excavated example arranged in columns.

The dataphys.org entry quotes her 1999 text and summarizes it this way.

Clay tokens suggest that physical objects were used to externalize information, support visual thinking and enhance cognition way before paper and writing were invented.

This places the tokens as an example of objects externalizing information and aiding thought, long before paper and writing. The entry is tagged rearrangeable.

Even if we cannot confirm that they were lined up in columns, they can be read as a record that could be rearranged. Because one piece corresponds to one item of goods, the grouping can be changed just by moving the pieces. The rows and columns of a table can be read as one way of using that freedom.

Marks Pressed onto the Envelope Become Writing

Around 3300 BC, hollow clay spheres appear, the so-called envelopes. According to Schmandt-Besserat, tokens representing the amounts not yet delivered were placed inside and kept until payment.

A round clay envelope with seal impressions on its surface sits on a stand, with small clay tokens displayed in a row below it

Source: Photo by Marie-Lan Nguyen, Musée du Louvre (Sb 1932) / Wikimedia Commons (photo: CC BY 2.5)

Sb 1932 at the Louvre is an envelope from Susa, Proto-Urban period (around 3500–2900 BC), 5.1 cm in diameter. Its surface bears several cylinder-seal impressions. The museum’s record describes “six deliberately drawn vertical lines” in the design and adds a question mark: “numbers?”

In Schmandt-Besserat’s explanation, once the envelope was closed the contents could not be seen, so the users, whom she calls accountants, pressed the tokens onto the damp surface before sealing to indicate what was inside. A cone becomes a wedge-shaped mark, and a sphere becomes a circular mark.

A clay sphere with a row of round depressions on its surface, with a flat disc and a cone-shaped token lined up in front

Source: Musée du Louvre, Département des Antiquités Orientales / Denise Schmandt-Besserat, “Tokens: their Significance for the Origin of Counting and Writing”

The figure of another envelope from Susa carries captions explaining that the lenticular discs each stand for a “flock” (10 animals?), the large cone for a very large unit of grain, and the small cone for a small unit. By her account, within about a century envelopes were replaced by clay tablets bearing only the impressions.

Reducing the three–dimensional tokens to two–dimensional signs proved revolutionary: It was the invention of writing.

The claim is that replacing three-dimensional tokens with two-dimensional signs was the invention of writing. The envelope can be read as one of the earliest examples in which the surface shown outside and the substance held inside were separated.

One-to-one counting does not suit large quantities. On tablets of around 3100 BC, the wedge-shaped mark that had been the unit of grain became the numeral “1,” and the circular mark became “10.”

A clay tablet with three round depressions, three elongated wedge-shaped marks, and incised signs in a row

Source: T. Cuyler Young Jr. and Royal Ontario Museum / Denise Schmandt-Besserat, “Tokens and Writing: the Cognitive Development”

On the tablet from Godin Tepe in Iran, three circles (10×3) and three wedges (1×3) are followed by the sign for “jar of oil,” expressing 33 jars of oil. Schmandt-Besserat sees this as a change in which numbers were detached from items and became usable for anything.

How Did Third Parties See This Theory?

The harshest is the review of Before Writing that the archaeologist Paul Zimansky wrote in 1993 in the Journal of Field Archaeology.

For the first time the author has put most of her cards on the table… and made it possible to review the substance and validity of her work. The result is devastating.

The assessment is that the author had for the first time laid out her hand, making it possible to examine her claims, and that the result was devastating. Zimansky points to the vagueness of the token classification and to the problem with a method that relies on shapes being “similar” to connect tokens with written signs, and he also points out errors in the catalog. He also counts that, of the 50 complex tokens linked to written signs, 18 have only a single example. Against her claim that all the main tokens of the urban period were found in envelopes, he counters that the shapes of particular complex tokens have hardly ever come from envelopes. The table’s “correspondence of shape and item” is a reading that has received this kind of criticism.

Research on data physicalization put this theory at the starting point of the field. The CHI 2015 paper “Opportunities and Challenges for Data Physicalization” by Yvonne Jansen and colleagues opens with this.

More than 7000 years ago, the Sumerians used clay tokens to represent quantitative data well before inventing written language

The sentence says that more than 7,000 years ago the Sumerians used clay tokens to represent quantitative data, well before inventing written language. The source is How Writing Came About, but “the Sumerians” and “7000 years ago” are broad phrasings that do not match the region and dates she gives.

The 2017 conversation in the journal Exchanges raises another point. The anthropologist Bill Maurer set the tokens beside Bitcoin and said that both are in essence accounting systems. This can be summarized as a view that reads them as tools of bookkeeping and social guarantee rather than as representations for viewing. Schmandt-Besserat herself says that some researchers think numbers were abstract from the start, so interpretations are divided.

Takeaways

The first is to check who wrote the legend. The correspondence between shapes and items was not left by the ancient users; it is a reading assembled by twentieth-century researchers. The account that places the tokens at the origin of the lineage of data physicalization also stands on top of it. When citing old sources, it is better to cite where the correspondence table comes from, together with the criticism of it.

The second is the design of matching one piece to one item of goods. Because one piece is one unit, the pieces can be regrouped just by moving them, whether by item, by income and expenditure, or by size. The arrangement in columns and above/below is not a confirmed fact, but we can read from it that making each unit a separate object makes rearrangement free.

The third is the limit of one-to-one counting. Thirty-three jars of oil are 33 ovoids with tokens, and with a clay tablet are three circles, three wedges, and the sign for the item. Expressing by repeating the unit conveys small numbers intuitively, but becomes unwieldy as the quantity grows. In physicalization where the count itself becomes the number of objects, you need to decide in advance up to what scale you will count.

References

Primary sources

dataphys.org, “5500 BC – Mesopotamian Clay Tokens” (added by Pierre Dragicevic): https://dataphys.org/list/mesopotamian-clay-tokens/
Denise Schmandt-Besserat, “Tokens and Writing: the Cognitive Development,” Symbols at ‘Ain Ghazal, ex oriente, Berlin, 2013, pp.56–62: https://sites.utexas.edu/dsb/tokens/tokens-and-writing-the-cognitive-development/ (PDF: https://sites.utexas.edu/dsb/wp-content/uploads/sites/3043/2014/01/TokensWriting_the_Cognitive_Development.pdf )
Denise Schmandt-Besserat, “Tokens and Writing: The Cognitive Development,” ‘Ain Ghazal Excavation Reports 3, Chapter 2.3, 2018: https://sites.utexas.edu/dsb/?p=578 (Texas ScholarWorks: https://hdl.handle.net/2152/79833 )
Denise Schmandt-Besserat, “Tokens: their Significance for the Origin of Counting and Writing”: https://sites.utexas.edu/dsb/tokens/tokens/
Denise Schmandt-Besserat, “From Accounting to Writing”: https://sites.utexas.edu/dsb/tokens/from-accounting-to-writing/
Denise Schmandt-Besserat, “Tokens: The Cognitive Significance,” 1999 (the link target from dataphys.org; the original could not be retrieved and is unverified): http://www.laits.utexas.edu/ghazal/Chap1/dsb/chapter1.html
Musée du Louvre, Sb 1932 “Bulle avec empreinte et jetons”: https://collections.louvre.fr/ark:/53355/cl010175963
Wikimedia Commons, “Accountancy clay envelope Louvre Sb1932.jpg” (photo by Marie-Lan Nguyen, CC BY 2.5): https://commons.wikimedia.org/wiki/File:Accountancy_clay_envelope_Louvre_Sb1932.jpg

Secondary sources

Paul Zimansky, Review of Before Writing, Vols. I–II, Journal of Field Archaeology 20(4), 1993, pp.513–517: https://urkesh.org/attach/Zimansky1993.pdf (JSTOR: https://www.jstor.org/stable/530080 )
Yvonne Jansen, Pierre Dragicevic, Petra Isenberg, Jason Alexander, Abhijit Karnik, Johan Kildal, Sriram Subramanian, Kasper Hornbæk, “Opportunities and Challenges for Data Physicalization,” CHI 2015: https://doi.org/10.1145/2702123.2702180 (author version: https://ndownloader.figshare.com/files/41160233 )
Wilding, D., Rowan, C., Maurer, B., Schmandt-Besserat, D., “Tokens, Writing and (Ac)counting: A Conversation with Denise Schmandt-Besserat and Bill Maurer,” Exchanges 5(1), 2017: https://doi.org/10.31273/eirj.v5i1.196 (https://exchanges.warwick.ac.uk/index.php/exchanges/article/view/196 )

Hollow or Solid Stems Sorted the Votes, and a Counting Board Laid Them Out: Voting Tokens in Athenian Courts

Tue, 03 Nov 2026 00:00:00 +0900

In the Athenian courts of the 4th century BC, bronze voting tokens separated plaintiff from defendant by whether the stem was hollow or solid. At the count, the tokens were set into a reckoning-board and rearranged into a form that was “visible and easy to count.” This procedure appears in chapters 68–69 of The Constitution of the Athenians (Athenaion Politeia), attributed to Aristotle (or his school).

The case-collection site dataphys.org lists it as “500 BC – Pebble Voting.” But this is not the work of a single author: it is an entry that bundles together 5th-century BC vase painting and courts, 4th-century BC courts, and the ballot boxes of modern societies. The dataphys.org description, “a white pebble meant yes and a black pebble meant no,” cannot be found in The Constitution of the Athenians, in Boegehold, in the Getty’s commentary, or in the Agora records. What follows reads the sources one by one, following The Constitution of the Athenians, the records of the objects excavated in the Agora, the cup in the Getty Museum, and the 1963 paper by Alan L. Boegehold.

A Cup with Two Mounds of Pebbles

The oldest source is a red-figure kylix (wine cup), 86.AE.286, in the J. Paul Getty Museum. It was made in Athens, is attributed to the Brygos Painter, and dates to around 490 BC. One side of the exterior shows the quarrel between Ajax and Odysseus; the other shows the Greek leaders voting with stones.

Part of the cup’s exterior. Mounds of small dots are drawn to the left and right of a pedestal, and the figure at the far right holds his head in his hands

Source: J. Paul Getty Museum (86.AE.286), from Getty, “Voting with the Ancient Greeks”

The number of pebbles on Ajax’s side, at right, falls short of the more politically savvy Odysseus’s by one, causing Ajax to grasp his head in despair.

The Getty’s commentary takes the dots on either side of the pedestal to be two mounds of stones, one for Odysseus and one for Ajax. The mound on the right, Ajax’s, is one short, and Ajax clutches his head. One pebble was one vote; which mound it sat in stood for the candidate, and the size of the mound stood for the number of votes.

Boegehold saw the distinctive feature of such scenes in the fact that pebbles are used and yet no secrecy is kept at all. The pebbles are simply “counting stones,” which can be counted slowly and accurately after everyone has voted, an advantage that shouts and show of hands do not have. This is his reading. He also writes, however, that it would be rash to claim that the painter copied the procedure of the time in every detail.

In the 5th-century BC courts, a juror held a single pebble (or shell) and put it into one of two urns, one for the plaintiff and one for the defendant (Boegehold; Getty). The vote was distinguished by position: which urn it went into. The contents of the urn were emptied onto a board and counted.

Hiding the Stem Hole Under the Fingers

In the 4th century BC, the voting tokens used in the courts changed into bronze discs. Boegehold dates the court described in The Constitution of the Athenians to “around 325 B.C.” The description is in chapter 68 (Kenyon translation).

The ballot balls are made of brass with stems running through the centre, half of them having the stem pierced and the other half solid.

A stem runs through the center; half of the tokens have a stem with a hole through it, hollow, and the other half have a solid stem. Hollow meant a vote for the plaintiff, solid a vote for the defendant.

Seven bronze voting tokens, each a disc with a stem at its center. Only the one at upper left has a hole through the center of its stem

Source: Marsyas (Wikimedia Commons contributor), Ancient Agora Museum of Athens, from Wikimedia Commons, “AGMA Jetons vote.jpg” (CC BY-SA 2.5)

Actual specimens have come out of the Agora in Athens. In the excavation records, B 728 is hollow with a diameter of 0.063 m, and B 1055 is solid with a diameter of 0.065 m. B 1055 bears the inscription “ΨΗΦΟΣ ΔΗΜΟΣΙΑ” on its face. According to Boegehold, all the published tokens are dated to the 4th century BC on the basis of letter forms.

A photograph of five bronze voting tokens laid side by side. The one in the middle has a hole in its stem, and in the two lying on their sides the stems project from both sides of the disc

Photo: Craig Mauzy. Source: American School of Classical Studies at Athens: Agora Excavations, “Agora Image 2007.10.0080” (© ASCSA)

The photograph shows, from left, B 1056, B 146, B 728, B 1058, and B 1055. The discs look almost the same; only the inside of the stem differs.

the juror, taking his two ballot balls from the stand, with his hand closed over the stem so as not to show either the pierced or the solid ballot to the litigants, casts the one which is to count into the brazen urn, and the other into the wooden urn.

The procedure is this: the juror receives two tokens and, keeping the stem covered by his hand so that the litigants see nothing, drops the one that is to count into the bronze urn and the other into the wooden urn. The hole in the lid of the bronze urn admitted only one token, so that no one could put in two (Rackham translation).

dataphys.org describes these tokens as ones “whose difference could only be felt by touch.” The center of the design can be read as making the tokens look alike and hiding the difference under the fingers. Because both tokens always go into one urn or the other, the hand’s movement does not reveal the choice either, which can also be read from the procedure.

Setting the Tokens into a Counting Board and Laying Them Out

The counting procedure is in chapter 69 (Rackham translation).

empty it out on to a reckoning-board with as many holes in it as there are pebbles, in order that they may be set out visibly and be easy to count, and that the perforated and the whole ones may be clearly seen by the litigants.

The contents of the bronze urn are emptied onto a reckoning-board with as many holes as there are tokens. The explanation is that this lets them be set out clearly and counted easily, and lets the litigants clearly see the hollow ones and the solid ones.

According to Boegehold, on the board the stem is pushed into a hole and keeps the disc from falling. He calls this “a second function” of the tokens’ peculiar shape. The stem is the place that hides the type of vote at the time of voting, and at the count it becomes the foot that stands the token on the board.

The counting was done by jurors chosen by lot. A herald announced the totals; the side with more votes won, and a tie went to the defendant. Penalties were decided by the same method, with a second vote in which hollow stood for the plaintiff’s proposal and solid for the defendant’s. The number of jurors was normally 201, 401, or 501 (Boegehold). If several hundred tokens are sorted into hollow and solid and stood in the holes, the two totals can be compared by the lengths of the rows; this is a reading.

The table below summarizes, source by source, how data corresponds to the properties of objects.

Scene (source)	Data	Property of the object	How it is read
Cup of around 490 BC (Getty)	Votes per candidate	Position (which mound) and size of the mound	Compare the two mounds
5th-century BC courts (Boegehold, Getty)	Plaintiff or defendant	Position (which urn)	Empty onto a board and count
4th-century BC courts, voting (ch. 68)	Plaintiff or defendant	Shape (stem hollow or solid), hidden by the fingers	The counted token goes into the bronze urn
4th-century BC courts, counting (ch. 69)	Number of plaintiff and defendant votes	Arrangement stood in the holes of the counting board	Sort, lay out, and announce the totals
Metamorphoses, book 15	Guilty or not guilty	Color (white = not guilty, black = guilty)	Empty the urn and count
Society ballot boxes (Mackey)	Admit or not	Color (white ball = yes, black ball or black cube = no)	Whether even one black is present
RPSL ballot box (late 19th century)	Yes or no	Position (left and right drawers); black balls only	Open the drawers and count

In the 4th-century BC courts, the difference in shape is hidden at the vote and laid out in view at the count. This can be read as a design that switches the encoding of the same vote to a different way of showing at each stage.

White Pebbles and Black Balls Come from Other Sources

Ovid’s White and Black Pebbles

The dataphys.org entry says “a white pebble meant “yes” and a black pebble meant “no”.” But a description of telling votes apart by color appears in none of The Constitution of the Athenians, Boegehold, the Getty’s commentary, or the Agora records.

White and black pebbles appear in Roman-era Latin poetry, in Ovid’s Metamorphoses, book 15.

Mos erat antiquus niveis atrisque lapillis, / his damnare reos, illis absolvere culpa

The line says that there was an old custom of using white and black pebbles, black to condemn the accused and white to acquit. It comes from the story of the trial of Myscelus of Argos, in which a black stone in the urn is changed to white by divine power. The meaning is “guilty / not guilty,” not “yes / no.” The explanation in circulation seems closer to something a Roman poem told as an old custom of Argos; this is a reading.

Furthermore, votes in the Assembly were normally taken by show of hands (χειροτονία) (Carbon et al.). Pebble voting was not voting in general.

A Society’s Box Is Checked for “Clear” or “Foul,” Not Counted

Voting by putting white and black balls into a box belongs to later societies and clubs. The Encyclopedia Masonica, which carries the text of Mackey’s encyclopedia, writes that lodges voted with white and black balls, and that some Grand Lodges also admitted black cubes. The box was divided into two compartments, and its opening was built so that the hand could not be seen.

At the count, the Senior Deacon shows the box to the Junior Warden, the Senior Warden, and the Master in that order, and each reports either “the box is clear” or “the box is foul.” If there is no black at all it is clear; if there is even one it is foul. What is read is not the number of votes but whether any black is mixed in. Regulation VI of the General Regulations in the 1723 edition of the Constitutions of the Free-Masons provides that no one is to be admitted without the unanimous consent of the members present, but does not specify balls or a box.

A wooden ballot box with YES and NO on top, a large round hole on the front, and the letters Y and N on the drawers below

Source: Photo by Philafrenzy, Royal Philatelic Society London collection, from Wikimedia Commons, “Royal Philatelic Society London committee voting box 1880s” (CC BY-SA 4.0)

The box of the Royal Philatelic Society London (RPSL) from the late 19th century was used to elect committee members. According to the Society’s description, a Victorian box normally had one drawer into which white (yes) or black (no) balls were put, to be counted later. This box uses only black balls, and the vote is divided by whether the ball is dropped into the left or the right drawer. The opening is large enough to admit a hand up to the wrist.

A receiving slot on top of a red wooden box, a compartment holding white balls, and a drawer pulled out at the front

Source: Smithsonian National Museum of American History (exhibition “Vote: The Machinery of Democracy”), from Wikimedia Commons, “Ballotboxballs.jpg” (public domain)

dataphys.org writes that when the box was opened, everyone could see the result at once. In Mackey’s procedure, the box is viewed by three officers and the result is conveyed orally. When the white-and-black-ball box began is also something that the sources used here could not establish.

How Third Parties Have Seen This Voting

Boegehold’s 1963 paper placed the pebble first as a tool for counting, before it was a tool for secrecy.

the men who first conceived of polling an assembly needed to have an accurate count before the more sophisticated notion of perverting the count occurred to them.

His conclusion is that the people who first thought of taking a vote needed an accurate count before the more elaborate idea of corruption occurred to them. He also notes that the 5th-century BC “two urns and one pebble” and the 4th-century BC “two tokens and two urns” have been confused.

The 2026 paper by Carbon, Isager, and Pedersen states that accurate vote counts seem to have come only from secret ballots in which the votes were counted. The Assembly normally voted by show of hands, and cases in which vote counts survive in inscriptions are few, mostly from the Hellenistic period onward. To the view that the pebble was a “countable vote,” they add a limit on the settings in which it was used.

Annelisa Stephan of the Getty introduced the cup in a 2012 article as “one of the earliest depictions of voting in art.” She also notes that the pebble, psephos, is the root of psephology (the study of elections), and that ballot comes from the medieval French ballotte (a small ball).

Uses

The first is assigning a single vote a different way of showing at each stage. What was needed at the time of voting was secrecy, and the discs looked the same. What was needed at the time of counting was an appearance that the litigants could accept, and the same discs were stood on a board and laid out. This can be read as an example of changing the property used between input and display.

The second is choosing the form to suit the task of reading. The counting board is an arrangement for comparing two numbers, and the society’s box is built to judge “is even one black present?” In data physicalization too, deciding first what the viewer is to judge makes it easier to decide whether to use color, position, or shape.

The third is checking where a well-known explanation comes from. “White means yes, black means no” seems to be told in a way that overlaps not with Athenian sources but with a Roman poem (guilty / not guilty) and with the white and black balls of later societies. Writing about an old case by separating which period and which source it comes from keeps one from mixing up the correspondence between the properties of objects and their meanings.

References

Primary sources

Aristotle, The Constitution of the Athenians, ch. 68 (Rackham translation, Perseus): https://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%3A1999.01.0046%3Achapter%3D68
Same, ch. 69 (Rackham translation, Perseus): https://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%3A1999.01.0046%3Achapter%3D69
Same (Kenyon translation, MIT Internet Classics Archive): https://classics.mit.edu/Aristotle/athenian_const.mb.txt
American School of Classical Studies at Athens, Agora Excavations, B 728: https://agora.ascsa.net/id/agora/object/b%20728
Same, B 1055: https://agora.ascsa.net/id/agora/object/b%201055
Same, Agora Image 2007.10.0080 (photo by Craig Mauzy): https://agora.ascsa.net/id/agora/image/2007.10.0080
J. Paul Getty Museum, “Wine Cup with the Suicide of Ajax” (86.AE.286, Google Arts & Culture): https://artsandculture.google.com/asset/wine-cup-with-the-suicide-of-ajax/tAEB2kodFlg3DA?hl=en (object page: https://www.getty.edu/art/collection/object/103W7A )
Ovid, Metamorphoses, book 15 (Latin, Perseus): https://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%3A1999.02.0029%3Abook%3D15%3Acard%3D1 (English translation: https://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%3A1999.02.0028%3Abook%3D15%3Acard%3D1 )
The Constitutions of the Free-Masons, 1723 edition, General Regulations VI (full text): https://distantreader.org/stacks/trust/uc2/uc2_ark+=13960=t6ww7bd0n.txt
Encyclopedia Masonica (text of Mackey’s encyclopedia), “Ballot”: https://www.universalfreemasonry.org/en/entry/ballot
Same, “Ballot-box”: https://www.universalfreemasonry.org/en/entry/ballot-box
Royal Philatelic Society London, “The Ballot Box”: https://www.rpsl.org.uk/Museum/Explore/U3A-Project/emodule/798/egallery/28
dataphys.org, “500 BC – Pebble Voting” (added by Pierre Dragicevic): https://dataphys.org/list/visualizing-opinions-with-pebbles/
Wikimedia Commons, “AGMA Jetons vote.jpg” (Marsyas (contributor), CC BY-SA 2.5): https://commons.wikimedia.org/wiki/File:AGMA_Jetons_vote.jpg
Wikimedia Commons, “Royal Philatelic Society London committee voting box 1880s.jpg” (Philafrenzy, CC BY-SA 4.0): https://commons.wikimedia.org/wiki/File:Royal_Philatelic_Society_London_committee_voting_box_1880s.jpg
Wikimedia Commons, “Ballotboxballs.jpg” (public domain): https://commons.wikimedia.org/wiki/File:Ballotboxballs.jpg

Secondary sources

Alan L. Boegehold, “Toward a Study of Athenian Voting Procedure,” Hesperia 32(4), 1963, pp. 366–374: https://www.ascsa.edu.gr/uploads/media/hesperia/147360.pdf
Jan-Mathieu Carbon, Signe Isager, Poul Pedersen, “Vote Counts, Quorum, and Unanimous Decisions at Halikarnassos,” C&M Supplementum 4, 2026, pp. 177–201: https://tidsskrift.dk/classicaetmediaevalia/article/download/169641/210370/380240
Annelisa Stephan, “Voting with the Ancient Greeks,” Getty, 2012-11-06: https://www.getty.edu/news/voting-with-the-ancient-greeks/
Wikipedia, “Blackballing” (as a supplement): https://en.wikipedia.org/wiki/Blackballing

Easter Tables Become Turning Disks, Layers of the Body Become Lifting Paper: Movable Books of the Medieval and Early Modern Eras

Tue, 03 Nov 2026 00:00:00 +0900

The 13th-century English monk Matthew Paris turned a circular table for finding the date of Easter into a disk that rotates on a metal pin at its center. Single-sheet anatomical prints made in Strasbourg and Augsburg from 1538 showed the layers of the body, from skin to spine, through stacked paper flaps that the reader lifts one by one. The turning mechanism is called a volvelle, and the lifting mechanism a flap.

The entry point is the item “1200 – Pop-Up and Movable Books” on dataphys.org, the timeline of data physicalization. This article follows it back to the records of the holding libraries: manuscript MS 26 at the Parker Library, Corpus Christi College, Cambridge, and an anatomical sheet of 1544 held by the Wellcome Collection. In the second half, we also look at Johann Remmelin’s anatomy book of 1619, as an example of the lifting mechanism entering the bound book.

In the Front Matter of the Chronica Majora, a Table Turns

Matthew Paris (c. 1200–59) was a Benedictine monk of St Albans Abbey and a chronicler. The Parker Library’s description introduces him as follows.

“Matthew was also a talented artist who was both scribe and illustrator of his own chronicles.”

That is, he was a man who handled both the copying and the illustrations of his own chronicles. His major work, the Chronica Majora, is a manuscript in two volumes: volume I is Parker Library MS 26, and volume II is MS 16. It was made c. 1240–55. MS 26 covers the period from the Creation to 1188.

The turning table is in the front matter of this MS 26. The front matter holds, among other things, an itinerary map from London to Jerusalem (seven pages), a royal genealogy, a diagram of the winds, and a calendar. M. R. James’s catalogue lists f. v recto as a “Circular Paschal table”, that is, a circular Easter table, and describes the mechanism like this.

“On f. vr the circular Paschal table is made to revolve: head of metal pin in C.”

The circular Easter table is made to revolve, and the head of a metal pin sits at its center. The catalogue also says that another table or diagram that was once pasted at the bottom of the page is now lost.

On the back, f. v verso, there is a table that does not turn. The catalogue gives it the title “Table of Concurrentes etc. from 1116 to 1620”. Concurrentes are numbers used to determine the day of the week for each year. A static table lists 505 years, from 1116 to 1620. The front side has a turning disk, and the back side has a static table.

The date of Easter is found by a calculation called computus. Gianfranco Crupi explains that this calculation rests on the 19-year cycle of the Moon across solar years. As for the number of concentric rings in the disk and what is written on each ring, I could not confirm them from textual primary sources. The material of the disk is not confirmed either.

The image of f. vr can be seen on Parker Library on the Web (CC BY-NC 4.0).

1538: Stacking the Body on a Single Sheet

Anatomical illustrations with lifting flaps began not in books but as single sheets. Andrea Carlino calls this kind of print an anatomical fugitive sheet, and catalogued them in Paper Bodies (1999), which covers 1538 to 1687. The first edition dates from 1538. Heinrich Vogtherr in Strasbourg and Jost de Negker in Augsburg issued their sheets in the same year. Between 1538 and 1545, at least twenty editions appeared.

The one we look at here is the 1544 edition held by the Wellcome Collection, Anathomia oder abconterfettung eynes Mans leib, wie er inwendig gestaltet ist. It is paired with a sheet showing a woman. It was printed by Jacob Frölich of Strasbourg. It is a woodcut measuring 55 × 25 cm, and the Wellcome copy is hand-colored.

A single-sheet print of a man, 1544 edition. Around a seated man, small drawings of the stomach, liver, lungs and other organs and German text are arranged in three tiers

Source: Wellcome Collection (Public Domain Mark) / Wellcome Collection, Anathomia oder abconterfettung eynes Mans leib… (1544)

According to Carlino, when Vogtherr left Strasbourg in 1544, he seems to have left the woodblocks of the sheet with Frölich. Frölich printed from those blocks in the same year and added color. Carlino writes that the color softened the “austerity” of the image.

According to the Wellcome record, flaps are pasted over the chest and abdomen of the seated male and female figures: six on the man and nine on the woman. Each time a flap is lifted, the anatomical structure appears one layer deeper. Carlino describes how this genre is built. The whole torso lifts as a single flap. The internal organs are printed on separate paper, cut out, and pasted in layers. The last layer is the back of the rib cage and the spinal cord.

The torso of the man in the 1544 edition, arranged in order from Wellcome images 1 to 7. From the layers of lungs, heart and stomach, ribs and spine appear at the far right

Source: Wellcome Collection (Public Domain Mark) / Wellcome Collection, Anathomia oder abconterfettung eynes Mans leib… (1544); the torso area is cropped from images 1–7 and placed side by side

The figure above arranges Wellcome’s images 1 to 7 of the male figure from left to right. Along the way, the same layer appears twice. In the second image, PULMO (lung), COR (heart), DIAFRAGMA (diaphragm), STOMACHUS (stomach) and SPLEN (spleen) are printed over the organs. At the far right, the ribs and spine appear.

According to Carlino, in Vogtherr’s edition the name of each organ is printed on top of it, mainly in Latin. Some organs are printed once more in the surrounding text, where they come with a German name and a short explanation. The male and female figures were printed from the same blocks, and only the head and the flaps for the chest and the genitals were swapped. Carlino reads this as a way of keeping costs down.

In 1539, Vogtherr also issued, from the same blocks, a small booklet of easy anatomy in German. From this, Carlino concludes that the readers were neither specialists nor scholars.

From a Single Sheet to a Book of About 120 Flaps

The lifting mechanism eventually entered the book. Carlino writes that Andreas Vesalius used this technique in the Epitome (1543) and the Fabrica, which included figures for the reader to cut out and paste together.

The work that lies beyond them is Johann Remmelin’s Catoptrum microcosmicum (Mirror of the Microcosm). It was published in Augsburg in 1619. It is a folio 47 cm high, and its anatomical figures are three, on pp. 9, 15 and 21. The plates were engraved on copper by Lucas Kilian after Remmelin’s own drawings. The flaps on the three figures number about 120, according to the bibliographic record of the Getty Research Institute copy.

of Remmelin’s Catoptrum microcosmicum. A copper engraving with standing male and female figures, a torso in the center, and small drawings of eyes, ears and organs around them

Source: Getty Research Institute (shown as NOT_IN_COPYRIGHT on the Internet Archive) / Internet Archive, Catoptrum microcosmicum (1619)

The figure above is the first figure on p. 9, “Visio prima”, and it bears the signature “L. K. sculptor” at lower left. McGill University Library explains the order of the layers as “surface → blood vessels → muscles → nerves → spine”. The organs of the ears, eyes and torso are divided into separate small figures. In Visio prima, the last layer of the female torso in the center shows a fetus in the womb.

The year of publication is disputed. McGill calls the 1619 edition the “first authorized edition”, while Columbia University gives 1613 for the first Latin edition. Columbia writes that it was a bestseller of its day, aimed at “curious laypeople rather than medical students or physicians”. The book was translated into German, Dutch, English and French.

What Was Mapped to Which Movement

The table below sums up how the data corresponds to the properties of paper.

Source	Data (what it represents)	Mapping to a physical property	Sources
Paris’s Easter table (MS 26, f. vr)	The calculation for finding the date of Easter, based on the 19-year lunar cycle	A circular table made into a disk that turns on a metal pin at its center	James’s catalogue, Crupi 2019
Same (f. vv)	Concurrentes for determining the day of the week (1116–1620)	Laid out as a static table on the back	James’s catalogue
1544 single sheet	The order of the organs of the chest and abdomen, from front to back	The order in which the paper is stacked. Six flaps on the man, nine on the woman. The last layer is the back of the rib cage and the spinal cord	Wellcome, Carlino 1999
Same	Names of organs	Printed over each organ in Latin. German names and explanations in the text	Carlino 1999
Same	Difference between male and female	The same blocks are used, and only the head and the flaps for the chest and the genitals are swapped	Carlino 1999
Remmelin 1619	Layers of the body	Stacked in the order surface → blood vessels → muscles → nerves → spine. About 120 flaps across three figures	McGill 2019, Getty bibliographic record

With a static circular table, the reader had to rotate the manuscript to read the numbers. The ArchBook site summarizes this point from the research of Daniel K. Connolly as follows.

“required the reader to reorient the manuscript in order to read the numbers. Matthew Paris’s innovation allowed readers to turn the disk and align it with their bodies”

In other words, Paris’s innovation let the reader turn the disk and align it with their own body.

Looking at the table, the turning mechanism can be read as a calculator that looks up the correspondence between periodic quantities by rotation angle, and the lifting mechanism as a hierarchical display that shows spatial order through overlap. In the language of visualization, the former can be read as a circular layout with a means for the reader to manipulate its alignment, and the latter as the paper version of drill-down (a display that opens detail step by step).

Note that the body text of the dataphys entry contains points that do not agree with the holding libraries’ records.

The title is given as “Chronica Majorca”, but the correct form is Chronica Majora (maiora).
The axis of rotation is given as a string or rivet, but the MS 26 catalogue says it is a metal pin.
It says movable books began to be used for entertainment and for children in the 19th century, but ArchBook cites a volvelle for a fortune-telling lottery game from 1546.

How Third Parties Saw These Mechanisms

The first is Crupi, who focuses on the reader’s hand. In his paper “Volvelles of knowledge” (2019), he writes of Paris’s disk as follows.

“transformed the reader’s manual gesture into an intellectual experience of knowledge, thus enabling them to become not merely the interpreter of the text but also its architect, whose intervention enables the algorithm that generates meaning.”

This is an assessment that the reader’s hand movement became an experience of knowledge, and that the reader became not merely an interpreter of the text but a designer who sets in motion the procedure that produces meaning. Crupi also places the volvelle and the flap as the two devices most widely used in the scientific books of the 16th and 17th centuries.

The second is Carlino, who focuses on how the procedure of dissection was transferred to paper.

“This technique of illustration provided a virtual three-dimensional representation … In short, it translated on to paper the whole concept of anatomical dissection, mimicking the progressive unveiling of the body, from skin to guts.”

This is an assessment that the technique gave a virtual three-dimensional representation, and carried over onto paper the whole idea of dissection, in which the body is opened step by step from skin to innards. Carlino also touches on the commercial side, writing that the success of the single sheets was “immeasurably greater” than that of any anatomical treatise.

The third is Connolly, cited in the previous section. His view is that Paris’s disk solved the trouble of rotating the manuscript. ArchBook introduces this disk as “the first known volvelle”, and Crupi calls it “one of the earliest”.

Valeria Finucci, who curated the Duke University exhibition, says of such books in Duke Today:

“They allow for a material reading of their medical content.”

In other words, they make it possible to read medical content as a physical object.

Uses

From these two mechanisms, three things can be read as worth taking back to visualization work.

The first is to choose a way of moving that fits the shape of the data. Paris’s table was a calculation on a periodic calendar, and the mechanism was rotation. The anatomical sheet was an order from front to back, and the mechanism was stacked paper that lifts. When deciding on interaction on a screen too, checking first whether the data is periodic or hierarchical narrows the kinds of operation, so it can be read.

The second is to make the number and order of layers the number and order of operations. The 1544 male figure has six flaps stacked, and the last layer is said to be the back of the rib cage and the spinal cord. The number and order of lifts can be read as corresponding to the order of depth in the body. When designing a drill-down, make it predictable to the reader how many times they must open to reach the deepest level. The paper anatomical sheet can be read as an example of this.

The third is to place names and explanations separately. Over the organs, mainly Latin names were printed, and for some organs the text gave a German name and a short explanation. Keep the annotations inside the figure short, and leave the explanation to the surrounding text. This can be read as the same idea as how annotations are divided in charts today.

Both mechanisms show nothing unless the reader moves their hands. In works of data physicalization, deciding what to make the reader’s hands do is itself part of the encoding. The movable books of the medieval and early modern eras can be read as showing exactly that.

References

Primary sources

dataphys.org, “1200 – Pop-Up and Movable Books” (Added by Pierre Dragicevic, sent by Benjamin Bach). https://dataphys.org/list/pop-up-and-movable-books/
Parker Library, Corpus Christi College, Cambridge, MS 26 (Matthew Paris, Chronica maiora I), IIIF manifest (description and license). https://dms-data.stanford.edu/data/manifests/Parker/rf352tc5448/manifest.json
Parker Library on the Web, MS 26. https://parker.stanford.edu/parker/catalog/rf352tc5448
M. R. James’s description of the contents of MS 26 (via Biblissima IIIF Collections). https://iiif.biblissima.fr/collections/manifest/82acae38b80a68174ccdf8d7ca902548f7c32f36
Wellcome Collection, Anathomia oder abconterfettung eynes Mans leib, wie er inwendig gestaltet ist., Strassburg: Jacob Frölich, 1544. https://wellcomecollection.org/works/w2mgwvb3 / IIIF https://iiif.wellcomecollection.org/presentation/v2/b33552964
Wellcome Collection, Anatomical fugitive sheets, c. 1540. https://wellcomecollection.org/works/anmb7n8h
Andrea Carlino, Paper Bodies, Catalogue (Cat. 1 = Vogtherr 1538). https://doi.org/10.1017/S0025727300073129
Johann Remmelin, Catoptrum microcosmicum, Augsburg, 1619 (Getty Research Institute). https://archive.org/details/gri_33125015238948
Same (McGill University Library, Osler Library). https://archive.org/details/McGillLibrary-osl_elf_cum-deo-johannis-remmelini_elfWZ250R388c1619-19996
Same, German edition of 1661, Kleiner welt Spiegel (Columbia University Libraries). https://archive.org/details/ldpd_11497246_000
Ruoshan Gu & Mary Hague-Yearl, “A Mirror into the Microcosm: Digitizing Johann Remmelin’s 1619 Flap Anatomy,” McGill Library Matters, 2019-12-17. https://news.library.mcgill.ca/a-mirror-into-the-microcosm-digitizing-johann-remmelins-1619-flap-anatomy-catoptrum-microcosmicum/
Columbia University Irving Medical Center, “A Medical Pop-Up Book From the 17th Century.” https://www.cuimc.columbia.edu/news/medical-pop-book-17th-century
Duke University Libraries, “Animated Anatomies” (2011). https://exhibits.library.duke.edu/exhibits/show/anatomy/intro / list of exhibited items https://exhibits.library.duke.edu/exhibits/show/anatomy/intro/bib

Secondary sources

Gianfranco Crupi, “Volvelles of knowledge,” JLIS.it 10(2), 2019. doi:10.4403/jlis.it-12534. https://iris.uniroma1.it/handle/11573/1269189
Andrea Carlino, Paper Bodies: A Catalogue of Anatomical Fugitive Sheets 1538–1687, Medical History Supplement 19, 1999, ch. 2 “Anatomical Fugitive Sheets.” https://doi.org/10.1017/S0025727300073099
Gravelle, Mustapha, Leroux, “Volvelles,” ArchBook (2012, revised 2022; includes a summary of Daniel K. Connolly, Art Bulletin 1999). https://drc.usask.ca/projects/archbook/volvelles.php
Duke Today, 2011-04-07 (remarks by Valeria Finucci). https://today.duke.edu/2011/04/flapbook.html
Laughing Squid (E.D.W. Lynch), “Animated Anatomies, An Exhibition of Antique Medical Pop-Up Books,” 2011-05-18. https://laughingsquid.com/animated-anatomies-an-exhibition-of-antique-medical-pop-up-books/
Wikipedia, “Chronica Majora.” https://en.wikipedia.org/wiki/Chronica_Majora / “Matthew Paris.” https://en.wikipedia.org/wiki/Matthew_Paris

Recording Numbers by Knot Type and Position on a Cord: The Andean Quipu

Sun, 01 Nov 2026 00:00:00 +0900

The quipu (also spelled khipu) was a recording device that distinguished numbers by “the type and number of knots × their position on the cord (decimal place value)” and items by “the color and order of the cords.” Dozens of cords hang from a main cord, and a number is knotted into each one. Garcilaso testifies that the color and order of the cords showed what the numbers were counting. However, the meaning of the colors cannot be generalized beyond the specimens studied.

The materials here are two primary sources. One is the Royal Commentaries of the Incas (Comentarios Reales), published in 1609 by Inca Garcilaso de la Vega, born in Cusco, who writes about his own experience of reading quipus. The other is a paper by L. Leland Locke in American Anthropologist in 1912, which demonstrated the decimal place-value system on actual specimens. The quipu is also listed in dataphys.org’s collection of data physicalization examples. What follows reads the quipu along these two sources.

Position on the cord gives the place; the shape of the knot gives the value

A quipu has a thick main cord with thinner cords hanging from it like a fringe. According to Locke, pendent cords rarely exceed 0.5 meters, and their number ranges from one or two to over a hundred. The materials are cotton and wool.

An Inca quipu with dozens of brown, white, and black cords hanging radially from a main cord, with knots here and there

Source: The Cleveland Museum of Art, Gift of John Wise, 1940.469 / Cleveland Museum of Art Open Access (CC0)

Garcilaso had already described how numbers were written. The topmost knot is the ten-thousands place, followed downward by thousands and hundreds, and the lowest is the ones place.

The knots of each number, and each thread, were placed in a line with each other, exactly in the way a good accountant places his figures to make a long addition sum.

Summary: Knots of the same place were lined up in a row. Garcilaso compares this to a skilled accountant aligning the digits of figures for a long addition.

Locke confirms this on actual specimens. In the general description, a single knot in the row farthest from the main cord stands for 1, in the next row for 10, and in the one after that for 100. A group holds at most nine single knots, and he writes “the number system being strictly decimal.” In the Huando type, such as B8713, only the ones place is written with a different kind of knot. The numbers 2 to 9 are “long knots,” and the number of turns wound is the value. The number 1 is a figure-eight knot (this is a reading of Locke’s illustration by the Khipu Field Guide). Zero is expressed by leaving that place without a knot.

A photograph showing sample knots for 1 through 9, how cords are attached to the main cord, and how long knots, figure-eight knots, and single knots are tied

Source: L. Leland Locke, “The Ancient Quipu, a Peruvian Knot Record” (American Anthropologist, 1912), Plate XXI (published in 1912; public domain in the United States)

In the upper row of samples, the larger the number, the more turns the knot has.

Changing the form only for the ones place has a practical benefit. In researchers’ notation (s = single knot, L = long knot, E = figure-eight knot, X = blank), 731 is written “7s, 3s, E” and 804 is “8s, X, 4L.” When a ones-place knot appears, the number ends there. This is why 107 and 51 can be tied one after the other on a single cord (1s, X, 7L, 5s, E). It can be read as changing the form of the value itself to mark where a number ends, without using a delimiter.

1912: the upward cord reveals the sum

Locke took B8713 of the American Museum of Natural History as his representative specimen for confirming the decimal reading. It was excavated at Huando, on the coast north of Lima. Its pendent cords are divided into six groups of four, and each group has one upward cord (top cord) passed through it.

The top cord sums the numbers on the pendent cords through which it is looped, thus giving an accurate key to the numerical character of the knots.

Summary: The top cord adds up the numbers on the pendent cords it passes through. That, Locke argues, is a reliable key to the knots being numbers.

A diagram redrawing the knots of B8713 place by place in symbols. For each of the six groups, the sum of the four lower cords is shown alongside the value of the upward cord

Source: L. Leland Locke, “The Ancient Quipu, a Peruvian Knot Record” (American Anthropologist, 1912), Plate XXIII (published in 1912; public domain in the United States)

In the leftmost group, a, the pendent cords are 0, 10, 6, and 1, and the upward cord is 17. The sums of each group, shown below the figure, are 17, 1517, 805, 612, 88, and 658. The values of the upward cords are 17, 1417, 805, 612, 135, and 658. They agree in four of the six groups. Group b differs by 100, and group e has a missing cord. In the text, Locke writes that the fourth cord of b shows traces of having had a subsidiary cord attached, and that there is a discrepancy in the numbers (p. 328).

Locke estimated that this specimen is a record of “six periods (or six years) × four kinds of things.” His conclusions include the following.

These knots were used purely for numerical purposes.

The quipu was not used for counting or calculating but for record keeping.

Summary: The knots were used only for numbers. The quipu was not a tool for counting or calculating but a tool for keeping records.

David Eugene Smith, who contributed an introduction, wrote that this is probably the oldest decimal notation known to the Western world, at least the oldest portable one.

The sum cord is not a trick limited to this one specimen. Manuel Medrano and Ashok Khosla reported in 2024 that at least 74% of 650 quipus have summation relationships between cords. It can be read as meaning that the key for checking whether a reading was correct was built into the data itself. Garcilaso also writes that every village, however small, had at least four record keepers, who checked one another’s work. This was a mechanism for verifying one record through multiple readers.

Color and order show what is being counted

What showed what was being counted was color and order. Garcilaso writes that some cords were of a single color and others combined two or three colors, and that every color had a meaning.

The thing to which a string referred was understood by its colour: for instance, a yellow string referred to gold, a white one to silver, and a red one to soldiers.

Summary: What a cord referred to was known by its color. For example, a yellow cord referred to gold, a white one to silver, and a red one to soldiers.

What could not be shown by color was shown by order. For grain, the order was wheat, barley, beans, and millet. In a census, the first cord was men aged 60 and over, the second men in their fifties, and so on, ending with infants. The convention was to arrange items from the most important.

Breakdowns were shown with thin subsidiary cords of the same color. Garcilaso gives the example of attaching one to the cord for married men to record the number of widowers, and Locke also confirms that most old quipus have subsidiary cords.

However, no table of color correspondences survives. For B8713, Locke concludes that if the six groups record the same four kinds of things, the colors have no special meaning and were chosen by the maker’s preference or convenience. He does not rule out that there may have been a rough rule for colors. The Khipu Field Guide writes that color information is clearly important, but that its symbolic meaning has been lost.

Of traditions of history, law, and ritual, Garcilaso writes as follows.

But the Indian who had not received the traditions by means of his memory was as uninformed concerning them as a Spaniard or any other stranger.

Summary: An Indian who had not received the traditions through memory was as uninformed about them as a Spaniard. On the other hand, regarding tribute accounts, Garcilaso writes that he himself soon became able to read them “as well as the Indians themselves.”

The people who made and read quipus were specialists in charge of records. In the English translation of Garcilaso they are the Quipu-camayu, “He who has charge of the accounts.” Garcilaso does not write about a legend for the colors, but it can be read that such a legend likewise lay in these people’s memories. In 1583, the Third Council of Lima prohibited quipus that recorded offerings to non-Christian gods as idolatrous objects and ordered them burned. Because the legend existed only in the readers’ memories, it can be read that only the numerical part remained readable.

A line drawing of a figure holding both ends of a main cord and spreading out a quipu with many cords hanging from it. A grid-like counting device is at the lower left

Source: Felipe Guaman Poma de Ayala, c.1615 (the original is a 17th-century work; 1936 facsimile, p. 360) / Wikimedia Commons

A quipu record keeper drawn by Guaman Poma around 1615.

Listing the variables on the cord

The correspondences so far are laid out in a table.

Data	Physical property on the cord	Example	Certainty
Value of each digit (1-9)	Number of knots gathered in one place	Above the tens place, the number of single knots; at most 9 per group	Established (Locke)
Value of the ones place	Type of knot	2-9 are the number of turns of a long knot; 1 is a figure-eight knot	Established (Locke, KFG)
Place (1, 10, 100…)	Position on the cord (distance from the main cord)	The row farthest from the main cord is the ones place	Established (Locke, Garcilaso)
0	Absence of a knot	Leave that place empty	Established (KFG)
Sum	Upward cord passed through a group	B8713: 17 = 0 + 10 + 6 + 1	Established (in B8713). A minority across the corpus
Item	Color of the cord (single or multicolor)	Yellow = gold, white = silver, red = soldiers	Contemporary testimony. Cannot be generalized
Item not expressible by color	Order of the cords	Wheat, barley, beans, millet	Contemporary testimony
Breakdown, exception	Thin subsidiary cord of the same color	Number of widowers on the cord for married men	Contemporary testimony; confirmed by Locke on specimens
Grouping	Bundle of cords of the same color	May represent a kin group (ayllu)	Hypothesis (KFG)
Binary information	Direction of twist (S/Z), way of attachment to the main cord	May represent moieties (hanan/hurin), etc.	Hypothesis (KFG, Medrano & Urton)

dataphys.org lists “rope length” alongside color, knot position, and knot type. I could not find a primary source that supports this. Garcilaso writes that cords were of a fixed length of about two feet, and Locke also says that what determines the place is the distance from the main cord.

Dating also calls for care. The heading “2600 BC” on dataphys.org rests on a 2005 announcement that the “oldest quipu” had been found at the Caral site, but this is not a well-verified claim. The oldest quipu confirmed by radiocarbon dating is from Wari, dated 780-1024. Most of the 31 Inca quipus date to 1500-1600.

How third parties have seen the quipu

On the visualization research side, Yvonne Jansen, Pierre Dragicevic, and others cite the quipu as a historical example of data physicalization in their CHI 2015 paper.

The Incas never even developed a writing system but relied on Quipus – complex assemblies of knotted ropes – for data storage and retrieval.

Summary: The Incas had no writing system and used quipus, assemblies of knotted ropes, to store and retrieve data. It can be read as treating the quipu not as a display for showing, but as a device for storage and retrieval.

In anthropology, Sabine Hyland announced in 2017 that two letter quipus preserved in the village of Collata, dating to the 18th century, contain 95 kinds of signs. Her claim is that family names are expressed by combinations of color, fiber type (vicuña, alpaca, deer, and others), and direction of twist.

The reader must often feel the cords by hand to distinguish the fibre sources of these three-dimensional texts.

Summary: To tell the fibers apart, the reader often has to feel the cords by hand. However, the subject is colonial-period quipus, and the possibility that they were influenced by European writing has also been pointed out.

On the data science side, Medrano and Khosla look for coding conventions across the whole corpus. The abstract of their 2024 paper says that the summation relationships carried by upward cords are a minority, and suggests that white pendent cords may have been used as boundaries of summation groups. The Khipu Field Guide, created by Khosla, is a database of 703 quipus redrawn from the specimens. It says:

It’s difficult to understand a khipu from tables. A rendering is necessary.

Summary: A quipu is hard to understand from tables, and a redrawn rendering is needed. It can be read that Locke redrew B8713 as a diagram of symbols in 1912 out of the same need. To read quipus, researchers visualize them anew.

Takeaways

There are four things to take away from the quipu.

The first is assigning quantity and category to different variables. Applied to Bertin’s visual variables, it can be read as corresponding to the separate use of position, shape, hue, and arrangement. However, I found no research that connects the two. Rather than declaring it a “prototype of the visual variables,” it is more accurate to say that “a corresponding division of roles can be seen.”

The second is changing the form for each kind of value. By writing only the ones place with a different knot, the quipu can be read as marking where a number ends without a delimiter. It is a clue for asking, before adding a legend or dividing lines, whether the form itself can make the distinction.

The third is building a structure that allows checking into the data. The sum cord works like the total row of a spreadsheet, and Locke took it as a reliable key to the numerical character of the knots.

The fourth is keeping the legend outside the maker’s head. Numbers written under the shared convention of the decimal system can still be read today. Because the meaning of the colors lay only in the record keepers’ memories, it can be read that they are unreadable now. The quipu shows what happens to a record whose legend was never written down.

References

Primary sources

Garcilaso de la Vega, First Part of the Royal Commentaries of the Yncas, Vol. II, trans. Clements R. Markham, Hakluyt Society, 1871, Book VI, ch. VIII–IX (original Comentarios Reales, 1609). archive.org
L. Leland Locke, “The Ancient Quipu, a Peruvian Knot Record,” American Anthropologist n.s. 14(2): 325–332, 1912. archive.org
A. Cherkinsky & G. Urton, “Radiocarbon chronology of Andean khipus,” Open Journal of Archaeometry 2(1), 2014. doi:10.4081/arc.2014.5260 OpenAlex
S. Hyland, “Writing with Twisted Cords,” Current Anthropology 58(3), 2017. doi:10.1086/691682 / University of St Andrews, “Lost writing of the Inkas discovered,” 2017-04-20. news.st-andrews.ac.uk
M. Medrano & G. Urton, Ethnohistory 65(1): 1–23, 2018. doi:10.1215/00141801-4260638
M. Medrano & A. Khosla, “How Can Data Science Contribute to Understanding the Khipu Code?,” Latin American Antiquity, 2024. Cambridge Core
The Cleveland Museum of Art, Inka Khipu (Fiber Recording Device), 1940.469. clevelandart.org / Open Access API
Reuters report on the announcement of the “oldest quipu,” NBC News, 2005-07-19. nbcnews.com
dataphys.org, “2600 BC – Quipus,” 2013-09-22 (Added by Pierre Dragicevic). dataphys.org

Secondary sources

Y. Jansen, P. Dragicevic, P. Isenberg et al., “Opportunities and Challenges for Data Physicalization,” CHI 2015. doi:10.1145/2702123.2702180 HAL
Ashok Khosla, Khipu Field Guide. khipufieldguide.com / Introduction / Khipu Signs / Khipu Data Visualization
Wikipedia, “Quipu” (chronology dispute, the 1583 ban, number of surviving specimens, examples of numerical notation). en.wikipedia.org
Felipe Guaman Poma de Ayala, El primer nueva corónica y buen gobierno, c.1615, p. 360 (illustration). Wikimedia Commons

Winding 6,500 Years at 0.1 Inch per Year: Barbeu-Dubourg's Machine chronologique (1753)

Sun, 01 Nov 2026 00:00:00 +0900

The Machine chronologique (“chronological machine”) is a timeline that lays out about 6,500 years, from the Creation to the 1750s, on a single strip of paper at an even scale of 0.1 inch per year. The strip is about 16.5 m (54 feet) long. It is wound on two rollers and advanced by turning a handle. The reader sees only about 140 years at a time, the part spread on the stand. It was published in Paris in 1753 by the physician Jacques Barbeu-Dubourg.

This article follows a paper (1991) by Stephen Ferguson, Curator of Rare Books at Princeton University Library; the Denis Diderot entry “Chronologique (Machine)” in the Encyclopédie, which Ferguson reprints in an appendix; and writing by the design historian Stephen Boyd Davis. dataphys.org, which collects data physicalization works, lists this timeline with the note “Not a physical visualization.” This article takes the same premise and reads it as an example of a paper visualization operated by a machine.

1753: Making Chronology Look Like Geography

Barbeu-Dubourg (1709–1779) was a doctor of medicine of the University of Paris and a professor of pharmacy. He was also a friend of Benjamin Franklin. The approval for the timeline is dated May 2, 1752, and the royal privilege for publication May 28, 1753 (Ferguson p.192).

The timeline is a single strip made of 35 copperplate engravings pasted end to end. A 16-page booklet, Chronographie, ou Description des tems, came with it. The printed paper alone cost 12 livres; the version “with the machine” cost 15 or 18 livres (Ferguson p.195, p.204).

In the booklet, the author explains why he made it by comparing it with geography. The English translation below is Boyd Davis’s.

Geography has as its object the extent of the earth; Chronology has as its object the succession of time. May not duration be imitated and represented as effectively to the senses, as distinctly as space, and may not intervals of time be as easily counted in degrees?

In short, he asks whether lengths of time can be shown to the senses as clearly as space, and whether intervals of time can be counted off as easily as degrees.

The same booklet contrasts geography as “lively, convenient, attractive” with chronology as “dry, laborious, unprofitable” (Ferguson p.196). The author called his chart a “tableau mouvant & animé” (a moving, living picture).

The same year, in an entry in volume 3 of the Encyclopédie, Diderot introduced the chart as one that “saves time for those who know and effort for those who learn” (Ferguson p.219).

0.1 Inch per Year, 16.5 m for 6,500 Years

The backbone of the timeline is a single scale running along the length of the paper. According to Diderot, the scale consists of small equal divisions alternating white and black, with as many divisions as there are years from the Creation to the present. One division is one year, and it is 0.1 inch long (Ferguson p.192 n.4).

Drawing the whole span at the same scale is what produces a length of about 16.5 m. The figures agree across the sources. According to the Journal de Trévoux, the author counted 4,700 years from the Creation to the birth of Christ (Ferguson p.199 n.27). Adding 1753 gives about 6,450 years. Multiplied by 0.1 inch, that is about 645 inches, or about 16.4 m. This is close to the stated total length of “54 feet / 16.5 m.” This calculation is mine, not the author’s explanation.

Vertically, bands for countries and regions are stacked. In the early part the divisions are labeled “Patriarchs,” “Descendants of Cain,” and the like; after the Flood, bands for countries such as Egypt and China are added. The bottom row is for “événements mémorables” (memorable events). Diderot called lines perpendicular to the scale “lines of contemporaneity” and lines parallel to it “lines of duration.”

les perpendiculaires à l’échelle chronologique sont des lignes de contemporanéité ; les paralleles à l’échelle chronologique sont des lignes de durée.

That is, things placed on the same vertical line are at the same moment, and the length along a horizontal line is how long something lasted. The correspondence between data and appearance is summarized below.

Data	Appearance / mechanism	Specifics
Year	Position along the length of the paper, at the same scale throughout	0.1 inch per year; alternating white and black divisions
Decade boundaries	Dotted vertical lines across the full height of the paper	Every 10 years
Major eras	The scale is divided into three	Creation to the founding of Rome, to the birth of Christ, to the present
Contemporaneity	Placed on the same vertical line	“Lines of contemporaneity”
Duration (reign, lifetime)	Length of a line parallel to the time axis	Start and end read at the ends against the vertical lines
Place / category	Vertical bands	A band per country; “memorable events” in the bottom row
People and events	Names and short phrases engraved at the corresponding year	Kings, queens, assassins, sages, etc.
Year of beginning	A dotted line dropped from the initial of a name to the scale	Abolished in the 1838 edition
Occupation, character, moral judgment	Small symbols beside the name	About 80 types, about 83 symbols
Range visible at once	The part spread on the stand of the machine	At least 140 years
Moving through time	Winding between rollers with the handle	Winding left moves toward the present, right toward the past

The symbols borrow from astrology, alchemy, pharmacy, Christianity, and other systems. An assassin is an upward-pointing dagger, a person assassinated a downward-pointing dagger, a physician a snake, and a poet a lyre.

The even scale had another effect. Ferguson writes that the number of events becomes density on the paper as it is.

A glance at the beginning of the line shows long stretches of time with no recorded events; the end of the line (the 1700s) is dense with them.

The observation is that the start of the timeline has long empty stretches with no recorded events, while the end, the 1700s, is crowded with them.

About 140 Years at a Time

The author put the 16.5 m of paper in a box. According to Diderot’s description, the machine is made of two identical halves. Each has a stand on which the timeline is spread (board B, 16 pouces long and 5½ pouces wide), a thin roller to reduce friction, and a roller with a handle. The two ends of the timeline are pasted to the rollers with handles, and turning either one moves the paper from one side to the other.

A wooden box opened, with part of the timeline spread between two rollers. Names and symbols fill the vertical lines and the country bands

Source: Rare Book Division, Department of Rare Books and Special Collections, Princeton University Library / dataphys.org

Diderot wrote that the range that can be spread on the stand is “au moins cent quarante ans” (at least 140 years). In the photograph in Ferguson’s paper, 150 years, A.D. 360 to 510, are open. 140 years is 14 inches, about 35.6 cm, which fits within the length of board B (about 43 cm). This conversion is also my own calculation.

How to use it is described in the booklet’s “Avertissement.”

Are you reading a book of history? Then to see at a glance world events covered by the book, just roll the Carte to the relevant years and there it is!

That is, if you are reading a history book, you can roll the chart to the relevant years and see the world events of the same period at a glance. The booklet goes on to say that you can roll back to Adam and advance to the present to follow the whole axis of time (Ferguson p.197–198).

When not in use, the two halves are folded together, closed, and fastened with clasps.

The machine closed. The covers of the two cylinders, the folded stand, and the handles at the ends are visible

Source: Rare Book Division, Department of Rare Books and Special Collections, Princeton University Library / dataphys.org

Diderot writes that when closed, the timeline is “à couvert de toutes parts, & fort en sûreté” (covered on all sides and very safe). dataphys.org says it is “folded for transport,” but I could not confirm in the primary sources that carrying it was the purpose.

The Note “Not a Physical Visualization”

The dataphys.org entry was registered by Pierre Dragicevic. Its description begins with the following sentence, and the note follows.

Before Joseph Priestley published his famous timelines, Frenchman Jacques Barbeu-Dubourg built a 16-meter long timeline showing 140 years of world history, which can be mechanically scrolled and folded for transport. Not a physical visualization but maybe the first “interactive” timeline representation in history.

It says that before Joseph Priestley, a 16 m timeline that could be scrolled mechanically was built; that it is not a physical visualization; but that it may be the first “interactive” timeline representation in history.

“Before Priestley” is correct. Barbeu-Dubourg’s timeline dates from 1753, and Priestley’s Chart of Biography from 1765 (Ferguson p.202 n.29). In a 1768 letter, the author himself wrote that Priestley’s chart works on “almost the same principle.”

On the other hand, “140 years of world history” is not the period the timeline covers. The timeline depicts the whole span from the Creation to the 1750s; 140 years is the range visible at one time when the machine is opened.

The note “Not a physical visualization” agrees with the distinction in the table. Year, duration, place, and a person’s character each correspond to a position, line, text, or symbol on the paper. Data values are not turned into the shape or quantity of an object. All the machine does is the operation of choosing which part of the long paper to show. This series includes the work not as a piece that turns data into the shape of an object, but as a device for operating a paper visualization by hand.

How Third Parties Saw This Timeline

Boyd Davis evaluates it as a forerunner of the evenly spaced time axis.

It may be the first to use a constant numerical scale for an entire historical period … And it is interactive - slightly.

He says it may be the first to use a constant numerical scale for an entire historical period, and that it is, slightly, interactive. In a 2015 paper he compares it with Barbeau de la Bruyère’s Mappemonde of 1750. In the Mappemonde, the number of years per division changes going back in time: 100, then 200, 250, and 500. Barbeu-Dubourg and Priestley, by contrast, “speak with one voice” in holding that an evenly spaced time axis is needed, he argues.

The art historian Astrit Schmidt-Burkhardt gives more weight to the machine side. Her essay has a section titled “CINEMATOGRAPHIC VISION MACHINE.”

The irreversible timeline on which all historical accounts had hitherto rested could at last be recast as a reversible conceptual model. … The idea of the interactive user was born.

Her conclusion is that the irreversible line of time on which historical accounts had rested could at last be recast as a reversible conceptual model, and that the idea of the interactive user was born. The same essay, however, gives the length of the paper as about 6.5 m and the number of symbols as 65, which do not agree with the figures of Ferguson and Boyd Davis.

Ferguson reads it as a product of Enlightenment thought.

Obviously, the ‘machine’ or apparatus itself reinforces the ideological message of the time-line. Its significance lies not only in its content, but also in its physical form.

The machine itself reinforces the thinking behind the timeline; its significance lies not only in its content but in its physical form. He also writes that the practice of morally judging people with symbols connects to the tradition of allegory. Sales appear to have been poor, and Ferguson writes, “Apparently only a few machines were sold.”

Takeaways

What we can take from this timeline is a design that keeps the scale fixed and moves only the visible range. To fit the whole on one sheet, you would have to change the scale for older periods. Barbeu-Dubourg kept the scale and chose a form in which the lengthened paper is fed through rollers.

The second is not erasing the blank. Because the scale is even, the earlier periods with few records look empty and the 1700s look packed. The difference in density is itself information, I read. When data is skewed and you are tempted to compress the axis, it is worth first considering what the blank is saying.

The third is to think separately about how the data is mapped and how the mechanism is operated. In this work all the data is on the paper, and the machine only moves the range being viewed. When you make a work with physical objects, putting into words whether the object represents the data or lets people operate how the data is viewed keeps the design from wavering.

Finally, there is a mechanism for accepting readers’ corrections. According to Diderot, the author offered a free copy to scholars who wrote in corrections or opinions and sent them back. The Princeton copy also has handwritten additions thought to be by the author himself, such as Franklin’s kite experiment (1752) (Ferguson p.207). It can be read as showing that he did not intend to finish a long timeline alone but to correct it through readers’ opinions.

References

Primary sources

Jacques Barbeu-Dubourg, Chronographie, ou Description des tems, Paris, 1753 (the booklet and the 35 sheets of the timeline). BnF Gallica: http://gallica.bnf.fr/ark:/12148/bpt6k1314025 (main text not checked; the content is taken from quotations in Ferguson and Boyd Davis)
Denis Diderot, “Chronologique (Machine),” Encyclopédie, vol. 3, 1753. The French text and English translation are from Ferguson (1991), appendix, pp.218–226. English translation by the Collaborative Translation Project: http://hdl.handle.net/2027/spo.did2222.0001.081 (not checked)
Letter from Barbeu-Dubourg to Franklin, May 8, 1768: https://founders.archives.gov/documents/Franklin/01-15-02-0067 (main text not checked; the English translation is from Ferguson pp.202–203)
Stephen Ferguson, “The 1753 Carte chronographique of Jacques Barbeu-Dubourg,” Princeton University Library Chronicle 52(2), 1991, pp.190–230: https://www.princeton.edu/~ferguson/PULC_1991_duBourg.pdf
Princeton Alumni Weekly, “From Princeton’s vault: A chronographic machine,” 2011-04-06: https://paw.princeton.edu/article/princetons-vault-chronographic-machine
dataphys.org, “1753 – Barbeu-Dubourg’s Machine Chronologique” (registered by Pierre Dragicevic): https://dataphys.org/list/barbeu-dubourgs-machine-chronologique/

Secondary sources

Stephen Boyd Davis, “The first modern timeline?,” Chronographics, 2009: https://chronographics.blogspot.com/2009/08/first-modern-timeline.html
Stephen Boyd Davis, “Chronographics at Princeton,” 2009: https://stephenbd.blogspot.com/2009/08/chronographics-at-princeton.html
Stephen Boyd Davis, “French copy of Barbeu-Dubourg now online,” 2011: https://chronographics.blogspot.com/2011/09/french-copy-of-barbeu-dubourg-now.html
Stephen Boyd Davis, “Beholder of All Ages: The History of the World in a French Mappemonde,” Textimage n°7, 2015: https://researchonline.rca.ac.uk/1701/1/boyd-davis.pdf
Astrit Schmidt-Burkhardt, “Learning in the Age of Enlightenment,” George Maciunas Foundation (first published in Bildwelten des Wissens 7(1), 2009): https://georgemaciunas.com/?p=2803
Wikipedia, “Jacques Barbeu-Dubourg” (consulted as a supplement for his biography): https://en.wikipedia.org/wiki/Jacques_Barbeu-Dubourg

Producing a Frequency Distribution by Dropping, Not Drawing: Galton's Quincunx

Sun, 01 Nov 2026 00:00:00 +0900

Francis Galton’s quincunx is the device now also known as the Galton board. One lead shot corresponds to one individual, and one row of pins corresponds to one small chance event. Rather than drawing a frequency distribution, it lets shot fall from the top and produces the distribution inside the apparatus, as the outline of the columns of shot that collect in the compartments below.

The original was made in 1873 by Tisley and Spiller of London to Galton’s design, and is now in the Galton Collection at UCL (University College London). The dataphys.org entry gives the year as 1893, but UCL’s record dates its manufacture to 1873. What follows is a reading based on Galton’s own 1877 lecture and his 1889 book Natural Inheritance.

1874: dropping lead shot at the Royal Institution

The device first appeared in public on 27 February 1874. Galton lectured at the Royal Institution in London under the title “On men of science, their nature and their nurture.” The lecture abstract adds “[this was illustrated by many experiments]” at the passage on the “law of frequency of error.” In his 1877 lecture he also recalls using the device “three years ago in this very theatre.”

According to UCL’s record (LDUGC-063), the original measures 360 × 290 × 52 mm. The historian of statistics Stephen M. Stigler writes that it has 17 rows of staggered pins.

The original carries an inscription explaining how to use it. Kunert et al. quote it from Stigler’s The History of Statistics (1986).

“Charge the instrument by reversing it, to send all the shots into the pocket. Then sharply re-reverse and immediately set it upright on a level table. The shot will all drop into the funnel, and running thence through its mouth, will pursue devious courses through the harrow and will accumulate in the vertical compartments at the bottom, there affording a representation of the law of dispersion.”

In short: turn the body upside down to gather the shot in the pocket at the top, quickly flip it back and stand it on a level table. The shot runs from the funnel through the “harrow” (the rows of pins), collects in the compartments below, and so represents the law of dispersion.

A replica: on the left, inverted with the shot gathered at the top; on the right, set upright with a heap formed in the compartments below

Exhibit: Estes Objethos Atelier. Photo: Rodrigo Tetsuo Argenton. Collection: Matemateca (IME/USP). From Wikimedia Commons (CC BY-SA 4.0)

One shot is one individual, one row of pins is one chance event

Galton described the device in detail in Chapter 5 of Natural Inheritance (1889).

“I shall now illustrate the origin of the Curve of Frequency, by means of an apparatus shown in Fig. 7, that mimics in a very pretty way the conditions on which Deviation depends.” (p.63)

This announces that he will show the “origin” of the frequency curve with a device that cleverly imitates the conditions on which deviation depends. It can be read as an attempt to depict not the curve itself but the conditions under which the curve arises.

Fig. 7, with pins in quincunx arrangement beneath a funnel and shot piling into a mound in the compartments below, and its variants Figs. 8 and 9

Source: Francis Galton, Natural Inheritance (1889), Figs. 7–9. From Wikimedia Commons (public domain)

The pins are set in “quincunx fashion,” and a falling shot is bound to strike a pin in every row. Each time, it goes “darting a step to the right or left,” that is, it shifts one step to either side (p.64). Galton states the principle of the device as follows.

“The principle on which the action of the apparatus depends is, that a number of small and independent accidents befall each shot in its career. … This illustrates and explains the reason why mediocrity is so common.” (pp.64–65)

Many small, independent accidents befall each shot. Shifts to the right and to the left mostly cancel out, so most shot land close to directly beneath the funnel. That, he explains, is why mediocrity is so common.

The table below summarizes the correspondence between parts and meanings, drawn from Galton’s text.

Part of the device	What it stands for	Galton’s words and source
One lead shot	One individual (in 1877, one member of a population of seeds)	“pellets which serve as representatives of a population of seeds” (1877, p.292)
One row of pins	One small, independent chance event	“small and independent accidents” (1889, pp.64–65)
Number of rows of pins	Number of independent factors that determine a value (height, for example)	“corresponds to an equal number of sets of rows of pins” (1889, p.84)
The vertical line through the funnel’s mouth	The mean (the point of no deviation)	The vertical line through the mouth of the funnel (1889, p.65)
Left-right position of a compartment	Size of the deviation from the mean	The side walls of the box mark the limits of the range (1877, p.292)
Height of a column of accumulated shot	Frequency	“the columns of shot that accumulate” (1889, p.64)
Outline joining the tops of the columns	Frequency curve	“approximates to the Curve of Frequency” (1889, p.64)

The core of the encoding can be read as two points: position is the value, and accumulated height is the frequency. Because one shot is one unit, the finished mound can also be read as a histogram built by stacking units.

More rows change the width; more shot change the height

On 9 February 1877, Galton lectured in the same theatre on “Typical Laws of Heredity.” He says it is easiest to “begin by showing it in action,” and first sets the device going. He then speaks separately about the movement of individual shot and the shape of the whole.

“The courses of the pellets are extremely irregular; it rarely happens that any two starting from the same point will pursue the same path from beginning to end; yet, notwithstanding this, you will observe the regularity of the outline of the heap formed by the accumulation of pellets.” (1877, p.286)

Two shot starting from the same point will rarely follow the same path, and yet the outline of the heap is regular. The 1889 book likewise says the outline is “closely of the same shape however often the experiment is repeated” (p.64).

A replica with white shot piled in the compartments below the rows of pins, highest at the center

Exhibit: Estes Objethos Atelier. Photo: Rodrigo Tetsuo Argenton. Collection: Matemateca (IME/USP). From Wikimedia Commons (CC BY-SA 4.0)

What changes the shape of the mound? Galton distinguishes two factors.

Number of rows of pins: with few rows, almost all the shot drop into the compartments directly below and form something like a “column.” With a great many rows, the shot scatter widely and form something like a “horizontal bar” (1877, p.286).
Quantity of shot: adding shot makes the mound higher. But the left-right distance that contains half of the shot does not change. This distance is the probable error, “Prob: Error,” written Q in the book (1889, p.65).

In the 1877 lecture, he explains that if the curve is drawn on a rubber sheet, stretching it sideways increases the scatter and stretching it vertically increases the number. It can be read as the device separating the parts that determine a distribution’s “shape” from the parts that determine its “quantity.”

Rearranging pins and boxes to explain heredity

The main subject of the 1877 lecture was heredity. Galton had been comparing the weights of sweet pea seeds between parents and offspring (Galton 1877, p.290). He kept the 1874 device as the lower stage and added a mechanism above it.

The upper box and trap doors: the distribution of the parental generation. The shape of the mound is trimmed by a curved slit cut in the back board, and overflow shot are discarded to the back (p.292).
Slanted chutes (shoots) inclined toward the center: reversion (regression). Every chute leads toward the central vertical line, so the width of the mound shrinks (p.292).
The rows of pins below them: variation within a single family. Opening one trap door produces a small mound directly beneath it (p.292).

When all the trap doors are opened, the small mounds overlap and return to a mound just like the original. This showed, through a combination of parts, that the width of the distribution is preserved from one generation to the next. The lecture record describes Galton opening the trap doors one after another (p.292).

Stigler reads the text of the same lecture as indicating that there was a working model of the device representing natural selection (explanation of Fig. 3, p.474). The model has not survived. Kunert et al. write that the two-stage quincunx described in a letter of 12 January 1877 apparently was never actually built (p.150). Whether the two-stage device in the letter and the device in the lecture are the same has not been established.

In the 1889 book, the same structure is reused for thought experiments. In Fig. 8, shown above, the device is cut horizontally at an intermediate height and a box AA is inserted in the middle. If the probable error down to the middle is a, that from there to the bottom is b, and the total is q, then q² = a² + b², which is the composition of variances (pp.67–68). In Fig. 9, the shot are first gathered toward the center and then dropped, so that only the width shrinks while the shape is kept (p.69).

What Galton wanted to talk about with this device was heredity and the distribution of human ability. Kunert et al. write that Galton’s main interest was “the improvement of mankind by selective breeding,” and that for this purpose he coined the word eugenics (p.144). The mathematics of the device and the claims made with it need to be read separately.

How third parties have seen the device

As a tool for thinking: in his 2010 paper, Stigler reads the device as Galton’s own instrument for arriving at regression. The slanted chutes of 1877 look at first like a “just-so story” with the conclusion built in (p.472). The decisive step was the idea of “running his quincunx backwards” (p.477): the shot in the lower boxes come, on average, not from directly above but from nearer the center.
Both the mathematics and the context: Kunert, Montag and Pöhlmann, in their 2001 paper, describe the device as “a device, which generates a histogram which is similar to the normal distribution” (p.146). At the same time they point out that Galton’s method of scaling was used to generalize about the abilities of races and “corresponded to the prevailing prejudice of that time” (pp.145–146).
An objection from physics: in a 2006 manuscript, Kevin Judd of the University of Western Australia questions the very premise of “independent accidents.” He ran a mechanical model, built with the original as a guide, on a computer and concluded that the motion of the shot does not follow a binomial process.

“The internet abounds with so called simulations of the quincunx that are not simulations of the mechanical device, but simply simulations of a Binomial process, or random walk.”

That is, many online “simulations” reproduce not the device but merely a binomial process. Judd nevertheless acknowledges the conceptual value of the statistical model.

Today the device is used to illustrate the central limit theorem (per Judd’s abstract), but Galton himself did not use that term.

Takeaways

Three things from this device can be carried back into visualization work.

The first is to assign the mechanism that produces a result, not the result itself, to parts. Viewers can look at the shot and the pins and explain, in the names of the parts, why the center ends up highest.

The second is to separate the factor that determines shape from the factor that determines quantity into different parts. Change the number of rows and the width changes; change the amount of shot and the height changes. In a chart too, keeping apart the appearance that depends on the amount of data and the appearance that expresses a property of the distribution can be read as making charts easier for people to compare.

The third is that a physical model gives form even to its premises. The moment one row of pins is called an “independent accident,” that premise is built into the device. Judd’s objection shows that the premise can diverge from the motion of real shot. When showing people how a visualization works, one also needs to be able to say which premise is being shown, as I read it.

Galton himself wrote that even after moving to formulas, one should not let go of the picture.

“The visible Schemes are not, however, to be removed too soon from our imagination. It is always well to retain a clear geometric view of the facts when we are dealing with statistical problems, which abound with dangerous pitfalls …” (1889, pp.66–67)

The visible diagrams should not be removed from the imagination too soon; statistical problems abound with pitfalls, so it is wise to keep a clear geometric view of the facts. The quincunx can be read as a device that follows this attitude of “keeping it in view.”

References

Primary sources

dataphys.org, “1893 – Galton Board” https://dataphys.org/list/galton-board/
Francis Galton, Natural Inheritance, Macmillan, 1889, pp.62–69, 84. Internet Archive / PDF at galton.org
Francis Galton, “Typical Laws of Heredity”, Proceedings of the Royal Institution 8, 1877, pp.282–301 (lecture of 9 February 1877). PDF at galton.org
Francis Galton, “On men of science, their nature and their nurture”, Proceedings of the Royal Institution 7, 1874, pp.227–236 (lecture of 27 February 1874). PDF at galton.org
Francis Galton, “Statistics by intercomparison, with remarks on the Law of Frequency of Error”, Philosophical Magazine, 1875. PDF at galton.org
UCL Collections Online, Quincunx, LDUGC-063 (maker, date of manufacture, dimensions, inscription). https://collections.ucl.ac.uk/Details/science/60000062
Galton bibliography. https://galton.org/bibnew/all.html
Images: Wikimedia Commons Quincunx (Galton Box) - Galton 1889 diagram / Galton box / Tabuleiros de Galton (antes e depois)

Secondary sources

Stephen M. Stigler, “Darwin, Galton and the Statistical Enlightenment”, Journal of the Royal Statistical Society: Series A 173(3), 2010, pp.469–482. PDF / Bibliographic record
Joachim Kunert, Astrid Montag, Sigrid Pöhlmann, “The quincunx: history and mathematics”, Statistical Papers 42(2), 2001, pp.143–169. PDF / doi:10.1007/s003620100048
Kevin Judd, “Galton’s Quincunx: random walk or chaos?”, 2006. PDF
Stephen M. Stigler, The History of Statistics: The Measurement of Uncertainty before 1900, Harvard University Press, 1986, pp.275–281 (second-hand, via p.147 of Kunert et al.; the book itself was not consulted)

Filling the Gaps Between 106 Measurements with Plaster: Thurston's Copper-Tin-Zinc Alloy Model

Sun, 01 Nov 2026 00:00:00 +0900

Robert Henry Thurston’s alloy model shows the proportions of copper, tin and zinc as a position on an equilateral triangle, and the tensile strength of the alloy at that position as the height of a brass wire standing there. The 106 values that were actually tested became the lengths of the wires, and plaster poured between the wires turned them into a single curved surface. On the finished surface, the tips of the wires show through as points.

The model was made by M. I. Coster during the alloy research (1875-78) carried out for the U.S. Board appointed to test metals. It was first published in Thurston’s paper at the Nashville meeting of the American Association for the Advancement of Science (AAAS) in August 1877. This article reads along that paper, the Board’s report dated June 1878, Part III of Thurston’s textbook Materials of Engineering, and his 1898 paper “Graphic Diagrams and Glyptic Models.”

1875: The combinations of three metals are beyond counting

On March 25, 1875, the President appointed the U.S. Board to test iron, steel and other metals. Its first meeting was held on April 15 of the same year. A Committee on Alloys was set up within it, and Thurston served as its chairman.

Thurston was then a professor of mechanical engineering at the Stevens Institute of Technology in Hoboken, and the tests were carried out there. He later became the first president of the American Society of Mechanical Engineers (ASME).

The Committee on Alloys was charged with experimenting on the properties of alloys and investigating the laws of their combination. For an alloy of two metals, an ordinary flat graph with composition on the horizontal axis and strength on the vertical axis is enough. With three metals, in the words of the 1877 paper, there is an “infinite number of possible combinations.” In 1899 Thurston recalled that no simple method had been found, and that ternary alloys had long been left untouched.

The 1877 paper states the goal of the research as follows.

… any one studying the results of the work should be able to say, with confidence, what would be the precise character of any possible alloy of the three metals which he might propose to make.

In short, once the maximum and minimum points have been found, anyone should be able to say with confidence what properties any alloy they propose to make will have. The task of the model, as I read it, is to give an answer even for alloys that were never tested.

Surveying the triangle as a territory

The title of the 1877 paper puts “planning researches” and “representing to the eye” side by side (Planning Researches / Representing to the Eye).

The method rests on an equilateral triangle. If you drop perpendiculars from a point inside the triangle to its three sides, the sum of the three lengths equals the height of the triangle at every point. Divide the height into 100 equal parts, and the lengths of the three perpendiculars become, as they are, the percentages of the three metals. The vertices are the pure metals, and points on a side are alloys of two metals.

Thurston treated this triangle as a piece of land to be surveyed.

Its topography may be studied as systematically and completely as that of any other territory of which the exact boundaries have been determined and marked out.

That is, the topography of the triangle can be studied as systematically and completely as any other territory whose boundaries have been fixed.

An equilateral triangle containing five concentric triangles, each shifted inward by 10%, with the letters A, B and C at the vertices

Source: R. H. Thurston, Proceedings of the AAAS 26 (1878), p.118, Figure 2 / Internet Archive (public domain)

This figure also determines how the sample points are chosen. Figure 2 shows concentric triangles, each shifted inward by 10%. Taking points every 10% along their sides gives the list of ternary alloys to be tested. The combinations that contain at least 10% of each of the three metals number 36 in the paper’s table.

According to Materials of Engineering, Part III, in the first series of 36 alloys at 10% steps, more than 5/6 of the alloys turned out to be “comparatively worthless.” The following series narrowed down to the strong region, testing 24 alloys at 5% steps and 12 alloys at 2.5% steps.

The 1877 paper also gives the reason for making the steps finer.

at sharp culminations, in points or lines, forming peaks or ridges in our topography, it would be necessary to take another set of points nearer together

Where the values change sharply, as at the summits and ridges of the terrain, the points have to be taken again at closer spacing. So the sample points are not spread evenly over the triangle; they become denser in the region of strong alloys, as I read it. The triangle was a stage for drawing results, and at the same time a stage for deciding what to test next.

A forest of wires, filled in with plaster

The Board’s report records the dimensions of the model. The base is an equilateral triangle with sides of 10 inches (about 25 cm), divided into 100 parts. At the vertices stand vertical rods that represent the tensile strength of pure copper, tin and zinc. The rods are brass wires, at a scale of 25,000 pounds per square inch (psi) per inch of length.

For example, 1 inch from the copper vertex stands the wire for gun metal, 90 copper and 10 tin. In the same way wires were set up at the points of bronzes, brasses and ternary alloys, and the number of points came to 106.

The 1877 paper describes the method of construction as a procedure. A triangle is drawn on a brass plate, and at the measured points wires cut to lengths proportional to the strength are fixed. A frame is built of boards, and plaster is poured in until the wires are covered. Once it has set, it is cut down from above, stopping just where the tips of the wires emerge.

An engraving of a plaster mass on a triangular base, with a high ridge along the back edge and a slope descending toward the front vertex

Source: R. H. Thurston (model by M. I. Coster), Proceedings of the AAAS 26 (1878), p.119, Figure 3 / dataphys.org (original illustration in the public domain)

According to the paper, this engraving (Figure 3) was made from a photograph of Coster’s model, and was included to illustrate the method.

Materials of Engineering describes the finished surface this way.

the tops of all the wires are just visible, shining points in the now smooth surface of the model

The tips of the wires are just visible as shining points on the smoothed surface. In this model, the values that were actually measured (the shining points) and the values supplied by hand in between (the plaster surface) can be told apart on the same surface, as I read it.

The table below sets out how the data corresponds to the materials.

Data	Physical property / material	Values and scale
Proportions of the three metals (% of copper, tin, zinc)	Position on the equilateral base. The lengths of the perpendiculars to the three sides are the % of each metal	Side of 10 inches, divided into 100. Vertices are pure metals
Alloys of two metals (copper-zinc, copper-tin, tin-zinc)	Position on a side of the triangle	Example: 1 inch from the copper vertex = 90 copper, 10 tin
Tensile strength of an alloy	Length (height) of a brass wire	1 inch = 25,000 psi
Alloys actually measured	One wire each. After completion, the wire tip showing through the plaster surface	106 points
Compositions not measured	The curved plaster surface filling the space between the wires	Poured in, then cut down to the wire tips
Compositions of equal strength	Contour lines from horizontal slices	0.1-inch intervals = steps of 2,500 psi
Strongest alloy	Highest point of the surface	Around 55 copper, 43 zinc, 2 tin; inside the 65,000 psi contour
The strongest alloy itself	Material of the model cast later	Cast using the plaster mold as the pattern

From the scale, the height of the 65,000 psi summit works out to about 2.6 inches (about 6.6 cm), a little under three tenths of the side of the base. This is not a dimension given in the sources; it is my own calculation.

A contour map, and the alloy at the summit

According to the Board’s report, slicing the model with horizontal planes at 0.1-inch intervals gives contour lines. From that map, the report says, the tensile strength of any alloy can be known to within 2,500 pounds. The map was attached to the report as “Plate I.” The same kind of map appears in Materials of Engineering as Figure 24.

A triangle with copper at the top vertex, tin at lower left and zinc at lower right, with strength contours drawn inside and a closed 65000 contour near the copper-zinc side

Source: R. H. Thurston, Materials of Engineering, Part III, p.425, Fig. 24 / Internet Archive (public domain)

In Figure 24, toward the right and near the copper-zinc side, the contour lines close into concentric rings. According to Materials of Engineering, however, the height scale of this map was converted by treating 100 foot-pounds in the torsion test as 25,000 psi in tension.

Thurston describes the location of the strong alloys as follows. The alloys of greatest strength cluster around a point near 55 copper, 43 zinc and 2 tin. On the map that point is enclosed by the 65,000 psi line, and on the model it appears as “the peak of the mountain seen at the farthest side — the copper-zinc side.”

An engraving of a model on a triangular base, labeled Cu at back left, Zn at back right and Sn at the front vertex, with the summit near the copper-zinc side at the back

Source: R. H. Thurston, Materials of Engineering, Part III, p.427, Fig. 25 / Internet Archive (public domain)

Figure 25 shows a model in a different material. According to Materials of Engineering, this model was made using the plaster mold as the pattern and “cast in an alloy of maximum tenacity,” that is, in the strongest alloy. Figure 25 was made from a photograph of that model. I could not confirm the year of the casting, or where the plaster mold and the casting are now held.

Thurston also noted the limits of the data. The alloys were made under the same conditions as in an ordinary foundry, without a deoxidizing flux, so the values scattered, and he admits “Some variations in the model are probably due to such accidental circumstances.” In a later follow-up with pure metals and flux, he identified the strongest bronze as 57 copper, 42 zinc and 1 tin. The summit of the model also served as the starting point for the next round of tests, as I read it.

In 1898, in his ASME paper “Graphic Diagrams and Glyptic Models,” Thurston named this method the “Glyptic System.” The 1877 paper does not use the term. On the value of the method he writes:

we are thus enabled to represent an infinite number of cases from the results of study of a few, and to find a minimum or a maximum, if it exists, whether it happens to be one of the cases selected or not

That is, from the results of studying a few cases one can represent an infinite number of cases, and find a minimum or a maximum whether or not it is among the cases chosen.

How did third parties see this model?

Assessments diverge among contemporary reports, later metallurgy, and the history of diagrams.

T. T. S. Laidley, chairman of the Board, introduced the model in his report dated June 8, 1878, as follows.

… is shown at a glance by a model ingeniously devised by Professor Thurston

That is, the tensile strength of all the alloys that can be made from two or three metals is seen at a glance in Thurston’s model. The report records even the dimensions, the scale, the number of points and the accuracy of the contours. It treated the model as a summary of the research results, as I read it.

The metallurgist Samuel L. Hoyt, in his 1914 doctoral thesis at Columbia University, excluded Thurston’s work, which deals mainly with mechanical properties, from his survey.

… since it deals mainly with the mechanical properties, it (and other similar work) will not be considered here.

His judgment was that although Thurston’s work also offers speculation on the structure (phases) of alloys, it deals mainly with mechanical properties and so is not taken up here. On the other hand, Hoyt cites Thurston’s result that the strength rises sharply from 50,000 to 65,000 and then falls to 5,000 within a narrow range, as an indication of a phase boundary.

In the history of diagrams, the ternary diagram itself is said to be older than Thurston. Richard J. Howarth, in a 1996 paper, placed its origin in 18th-century studies of color mixing. Thurston’s name does not appear in the published list of references. Thurston himself wrote that Gibbs proposed this principle in a different field in 1876. Yet in 1899 he wrote of the three-axis diagram, “I think, first employed by me.” It is safest to read this “first” as limited to its use in planning alloy research.

Takeaways

What this model offers is a way to show measured points and interpolated surface on the same surface, distinguishably. In Thurston’s model, the shining wire tips are the measurements, and the plaster between them is the interpolation.

Next, the sample points were decided with a figure before any chart was drawn. A first look at the whole at 10% steps, then finer steps of 5% and 2.5% only in the strong region. The concentric triangles of Figure 2 were a test plan. The visualization is used as a tool for deciding what to measure next.

Finally, how precision is stated. The Board’s report left behind the scale (1 inch = 25,000 psi) and the contour interval (2,500 psi). Thurston also admitted that scatter from the casting conditions shows in the model. Even when showing a solid form, add what can be read from it and to what precision. This model and its report can be read as an early example of that.

References

Primary sources

R. H. Thurston, “On a New Method of Planning Researches and of Representing to the Eye the Results of Combination of three or more Elements in Varying Proportions,” Proceedings of the American Association for the Advancement of Science, 26th Meeting, Nashville, Aug. 1877 (Salem, 1878), pp.114-120. Internet Archive
Report of the United States Board appointed to test Iron, Steel and other Metals, vol. I (Washington: GPO, 1881). Report dated June 8, 1878 by the chairman T. T. S. Laidley, pp.1-4. Internet Archive
R. H. Thurston, Materials of Engineering, Part III (A treatise on brasses, bronzes, and other alloys), 4th ed. revised, §§248-255, pp.416-432. Internet Archive
R. H. Thurston, “Graphic Diagrams and Glyptic Models,” Transactions of the American Society of Mechanical Engineers, vol. XIX (1898), pp.514-550. Internet Archive
R. H. Thurston, Discussion of the paper of Prof. Howe on the Use of the Tri-Axial Diagram and Triangular Pyramid (AIME, 1899). OneMine

Secondary sources

Samuel L. Hoyt, On the Copper Rich Kalchoids (Copper Tin Zinc Alloys), PhD thesis, Columbia University, 1914. Internet Archive
Richard J. Howarth, “Sources for a history of the ternary diagram,” The British Journal for the History of Science 29(3), 1996, pp.337-356. doi:10.1017/S000708740003449X Cambridge Core

Other (image sources and confirmation of biographical details)

dataphys.org, “1875 - Robert Henry Thurston’s Glyptic Models” (source of the Figure 3 image; registered by Pierre Dragicevic). dataphys.org
Wikipedia, “Robert Henry Thurston” (for confirming biographical details). Wikipedia

Press the Keys in Order, and Territory Plates Cover the Tribal Map: Ives's Historical Map

Sun, 01 Nov 2026 00:00:00 +0900

James T. B. Ives’s Historical Map, Showing the Successive Acquisitions of Territory by the United States of America maps the order in which the United States acquired territory onto the order of key presses, and the area acquired onto colored plates that slide into view. The plates overlay a layer showing where Native American tribes lived, one after another. The first key brings up a plate for the 16 states that correspond to the 13 colonies, and the last key brings up a plate for the Gadsden Purchase of 1853.

The map dates from around 1897, and its frame measures 34 by 24 inches. Below, I read it in the order a viewer operates it, following the introduction in the educational journal The School Journal of October 2, 1897, a 1911 catalogue of teaching materials, the catalogue of the holding institution, the Boston Public Library, and the description by the antiquarian map dealer Boston Rare Maps.

1897: A “Mechanical Map” in an Educational Journal

The earliest article on this map that I could confirm is in The School Journal of October 2, 1897. It ran on p. 322, in the “School Equipment” column, under the headline “Growth of the United States Shown by a Mechanical Map.” The column states that it carries no paid advertising. The article calls the map a mechanism that shows the growth of the United States, and says it is set under glass in an oak frame.

The maker was James Thomas Bostock Ives. He was born in London in 1839 (1840 in another biography) and was a Fellow of the Geological Society of London (F.G.S.). He moved to Toronto in 1886, then made maps in New York and Philadelphia, where he taught geology and, later, history. In 1904 he moved to Andover, New Hampshire, and he died in 1915.

In the 1911 catalogue of the New England History Teachers’ Association, the map was sold as “The Ives Historical Map” at a price of $15. Boston Rare Maps notes that this was $15 at a time when the 32-sheet set of Foster Historical Charts cost $20.

Item	Details
Date	Around 1897 (the BPL catalogue says “approximately 1897”)
Publication	Depending on the printing, Philadelphia (Academy of Natural Sciences) or New York (340 Fourth Avenue)
Size	34 x 24 inches, framed
Mechanism	Several map layers stacked, connected to cut-out tin pulleys and moved by a wooden lever
Holdings	Boston Public Library (Leventhal Map & Education Center). According to Boston Rare Maps, also the Clements Library, the Peabody Essex Museum, and a private collection in the Midwest

dataphys.org lists it as “1896 – James Ive’s Mechanical Teaching Map.” The correct spelling of the name is Ives, and “Mechanical Teaching Map” is a descriptive phrase. I could not confirm the basis for the year 1896.

Tribal Colors Placed on a Gray River Map

With all the plates pulled out, what you see is the bottom map. The School Journal describes its contents as follows.

Printed in neutral gray is shown, first, a simple river map of the United States. On this, grouped in distinctive colors, and arranged linguistically according to the map accompanying the eleventh census, are shown the locations of the various Indian tribes originally occupying the United States.

In short: first there is a simple river map of the United States, printed in neutral gray. On it, the locations of the tribes that originally lived in what is now the United States are shown in grouped colors, arranged by language family following the map that accompanied the eleventh census (1890).

The area of the United States with the tribal group names printed large, a legend of acquisitions at lower right, and the levers lined up along the lower edge of the frame

From James T. B. Ives / Boston Rare Maps

This layer also carries early settlements. English, French, and Spanish settlements are shown as colored dots, with the color indicating nationality. Each dot has a number, and the accompanying legend gives its name and year. The 1911 catalogue describes the state with the plates pulled out as “the virgin country before European occupation,” that is, the land before European occupation.

According to Boston Rare Maps, the layers consist of “eight sets of cut-outs.” The lower layer of each set is printed with the main river systems and the names of Native peoples. The upper layer is printed with the state borders as of 1893, major cities, Indian reservations, and so on. The upper layer slides in on the lever and covers the lower one.

Each Key Press Slides In a Territory Plate

The map is operated with keys lined up along the lower edge of the frame. Boston Rare Maps writes of a wooden lever, and the BPL catalogue writes of levers.

On touching the first of a row of keys below the lower edge of the frame, a colored segment slides into view, representing the 16 states corresponding to the 13 original colonies.

That is: touching the first of the row of keys below the lower edge of the frame slides a colored plate into view, representing the 16 states that correspond to the 13 original colonies.

Only the area of the 16 eastern states has been replaced by a plate printed with state borders and place names, while tribal names are still visible in the west

From James T. B. Ives / Boston Rare Maps

The first plate shows one more thing: how far the borders of the present 16 states extend beyond the watershed that France, and later Britain, claimed as the “proper limits of the colonies.” The second key brings up the region between the 13 states and the Mississippi River. The 1911 catalogue says this plate is of a “different color” from the first. The third is Louisiana.

Plates for the East, the region up to the Mississippi, Louisiana, and Florida are in place, while tribal names remain in the West

From James T. B. Ives / Boston Rare Maps

From here on, The School Journal describes the rest in one passage.

So each accession of territory down to the Gadsden purchase in 1853 is shown on the map by pressing a different key. Eleven red dots indicate the western movement of the center of population.

In other words, each acquisition of territory, down to the Gadsden Purchase of 1853, appears on the map when a different key is pressed, and eleven red dots show the westward movement of the center of population. The number 11 can be read as corresponding to the eleven censuses from 1790 to 1890. The article, however, does not say so.

Boston Rare Maps counts the acquisitions as nine stages, citing among others the 1783 Treaty of Paris, Florida in 1819, and the annexation of Texas in 1845. The plate for the 1848 Treaty of Guadalupe Hidalgo is said to be misprinted “1850 Texan Cession.”

With all plates in place, the continental United States is covered by plates printed with state borders, the northwest in blue and the southwest in reddish brown

From James T. B. Ives / Boston Rare Maps

Based on these contemporary descriptions, the correspondence between data and appearance or material can be summarized as follows.

Data	Appearance / material	Specifics	Source
Terrain (river network)	The bottom layer, a river map printed in gray	The 1911 catalogue calls it a “government river map”	School Journal, 1911 catalogue
Locations of Native tribes	Printed on a lower layer, grouped by color	Color groups = language families (after the map accompanying the 1890 census)	School Journal
Early settlements	Colored dots; color = nationality	The dot number is looked up in the legend for name and year	School Journal
Territorial acquisition (where)	Colored plates that slide in and cover the lower layer	Adjacent plates change color	School Journal, 1911 catalogue
Order of acquisition (when)	The order of the keys (levers) along the lower edge of the frame	The first is the 16 states, the last is the 1853 Gadsden Purchase	School Journal, 1911 catalogue
Gap from the colonies’ original extent	The offset between the first plate’s border and the watershed	-	School Journal
Westward shift of the center of population	A row of red dots	11	School Journal
Present-day (around 1893) geography	State borders, major cities, and reservations printed on the plates	-	Boston Rare Maps, BPL catalogue

What carries the order of acquisition is neither color nor position but the order in which the viewer presses the keys (only the westward shift of the center of population is shown by the positions of the red dots). The map stays a single frame, while the time of acquisition has been moved to the side of the hand’s operation, as I read it.

From a Map of Strata to a Map of Territory

Before the historical map, Ives had made geological maps. Advertisements for the Ives Strata Map and the Ives Altitude Map appear in the October 1895 issue of The American Naturalist.

According to the advertisement, the strata map shows the superposition of strata and outcrops, and can also show folds if the cards are bent. The altitude map has relief worked into its surface, with heights painted in strongly contrasting colors.

The advertisement says both received a diploma and a medal at the World’s Columbian Exposition in Chicago in 1893. The catalogue of the Pennsylvania exhibits at the Exposition also lists “Strata and Altitude Maps.” According to Boston Rare Maps, a label for this medal is pasted in the upper margin of the historical map.

The 1923 Encyclopedia of Connecticut Biography introduces Ives as “the inventor of the Ives Strata Map and the Ives Historical Map.” This can be read as moving the mechanism that shows strata stacked up to a mechanism that shows territory stacked up. However, I have found no source in which Ives himself says so. I could not confirm the text of any explanation Ives wrote himself.

How Third Parties Saw the Map

The School Journal of 1897 wrote that the map would help “make interesting the teaching of United States history.” The mechanism is introduced as a classroom tool.

The holding institution, the Leventhal Map & Education Center, pointed out the bias of the story in an article for its 2023 event “From The Vault: Mechanical Maps.”

The map told a story of the seemingly inevitable process of expansion and emphasized nation-building by treaties rather than the violence of war.

The assessment is that the map told a story in which expansion seemed an inevitable process, and emphasized nation-building by treaty over the violence of war.

Boston Rare Maps, the antiquarian map dealer, says the maker’s intent is unknown, and then reads the mechanism itself.

However the basic design of the mechanism conveys an impression, in keeping with the myth of Manifest Destiny, of a process both orderly and inevitable. It also conveys—at least to our modern sensibility—a sense of the tragedy that befell a civilization that was literally effaced from the map.

The reading is that the basic design of the mechanism, in keeping with the myth of Manifest Destiny, conveys an impression of an orderly and inevitable process, while also conveying, at least to modern sensibilities, the tragedy of a civilization literally effaced from the map.

The historical map displayed in a case at a 2019 exhibition. All plates are in place, and a legend of dates and acquisition names is visible at lower right

Source: Courtesy Aram Boghosian / WBUR

In 2019, the map appeared in the Leventhal exhibition “America Transformed: Mapping the 19th Century” (on loan from a private collection in the Midwest). WBUR’s article on the exhibition, by Frank Redner, opens with a photograph of this map. Redner is a member of the Upper Sioux Community.

In the exhibition, “Viewpoint” texts from the Native side were displayed beside the institution’s own captions. The article summarizes their point as “Native concepts about land are not two-dimensional.” These words are not aimed at Ives’s map alone, but they can be read as a clue for looking at this map, which stacks flat layers.

Takeaways

The first is that the order of operation becomes the axis of time. The sequence of keys is the sequence of dates, and each press advances one step. This can be read as connecting to today’s step-by-step timeline maps. However, there is no source showing a direct line of influence.

The second is that where you place the divisions becomes, as it is, a way of telling history. In this map, one step corresponds to one territorial acquisition (one plate). Leventhal’s remark about the emphasis on “nation-building by treaties” can be read as aimed at this way of dividing. When you divide a timeline into stages, deciding what the unit of a stage is is an editorial judgment, not data.

The third is that the stacking order carries meaning. The tribal layer is at the bottom, and the territory plates cover it. In a digital map too, deciding which layer is the base and which goes on top is a claim before it is a display setting. Boston Rare Maps’s phrase “effaced from the map” can be read as aimed at this stacking order.

The fourth is that what was covered can be brought back. As the 1911 catalogue writes, pulling out the plates makes the tribal layer appear. If you choose a design that covers, you should prepare an operation for returning to what was covered with equal weight. This map held both in a single frame.

References

Primary sources

Boston Public Library catalogue record, “Historical Map, Showing the Successive Acquisitions of Territory by the United States of America”: https://bpl.bibliocommons.com/v2/record/S75C8129750
The School Journal, Vol. 55, No. 12 (October 2, 1897), p. 322, “Growth of the United States Shown by a Mechanical Map”: https://archive.org/details/sim_school-journal_1897-10-02_55_12 (full text: https://archive.org/download/sim_school-journal_1897-10-02_55_12/sim_school-journal_1897-10-02_55_12_djvu.txt / Google Books edition: https://books.google.com/books?id=ov1KAAAAYAAJ&pg=PA322 )
New England History Teachers’ Association, A Catalogue of the Collection of Historical Material, Simmons College, Boston, 1911, entry “The Ives Historical Map”: https://archive.org/details/catalogueofcolle00newerich (the same text also appears in The History Teacher’s Magazine, April 1911: https://archive.org/details/sim_social-studies_1911-04_2_8 )
The American Naturalist, October 1895, advertisement (Ives Strata Map / Ives Altitude Map): https://archive.org/details/american_naturalist_189510
Encyclopedia of Connecticut Biography, 1923, entry on the Ives family: https://archive.org/details/encyclo07unse
Catalogue of the Exhibits of the State of Pennsylvania and of Pennsylvanians at the World’s Columbian Exposition, 1893: https://archive.org/details/catalogueofexhib00penn
Leventhal Map & Education Center, “Highlights from From The Vault: Mechanical Maps” (LMEC Staff, May 30, 2023): https://www.leventhalmap.org/articles/highlights-from-from-the-vault-mechanical-maps/
Leventhal Map & Education Center post on X (Twitter), May 23, 2019: https://twitter.com/bplmaps/status/1131571980770193409
dataphys.org, “1896 – James Ive’s Mechanical Teaching Map” (registered by Pierre Dragicevic, contributed by Jason Forrest): https://dataphys.org/list/james-ives-mechanical-teaching-map/

Secondary sources

Boston Rare Maps, “Ingenious mechanical map of United States territorial expansion” (Inventory ID: BRM3857): https://bostonraremaps.com/inventory/mechanical-map-of-united-states-territorial-expansion/
Frank Redner, “This Map Exhibit Draws A Darker U.S. History — Of Expansion Into Native Lands,” WBUR, October 14, 2019: https://www.wbur.org/news/2019/10/14/boston-public-library-native-americans-maps-united-states
Jonathan Crowe, “New Leventhal Exhibition: America Transformed,” The Map Room, May 9, 2019 (for the exhibition’s dates): https://www.maproomblog.com/2019/05/new-leventhal-exhibition-america-transformed/

Rearranging the Same 40 People into a Different Graph: Davenport's "Frequency Polygon Without Paper"

Sun, 01 Nov 2026 00:00:00 +0900

Line the same students up by height, and the tops of their heads trace a sequence of rank and value. Line them up in columns of two-inch classes, and the length of each column becomes a frequency. In 1901 the biologist Charles Benedict Davenport published a paper with photographs of 40 University of Chicago students arranged in these two ways. The data and the people are the same; only the arrangement changes the graph.

The source is Davenport’s paper “The Statistical Study of Evolution” (pp.447–460), published in the general-interest science magazine Popular Science Monthly, Vol.59 (September 1901). It is also listed as “Davenport’s Physical Distributions” on dataphys.org, a catalog of data physicalization examples. What follows is based on the Wikisource transcription and the page scans at the Internet Archive.

1901: An Introduction to Counting Evolution

The paper is signed “Professor C. B. DAVENPORT, UNIVERSITY OF CHICAGO.” Its subject is an introduction to the statistical study of evolution. It explains the mean, median, mode, frequency polygon, standard deviation, coefficient of variation, and correlation to general readers, one after another.

It opens with a declaration that science is becoming quantitative.

As science advances its methods become more and more quantitative.

In short: the more science advances, the more quantitative its methods become. Davenport also writes that individual differences among people cannot be captured in adjectives, so they are measured and expressed in numbers.

The photographs of people and shells appear in the first half of the paper. They are examples explaining how to make a “frequency polygon.” There are two photographs of students and two of shells. The photographer, date, and place are not stated in the magazine.

The ending of the paper also deserves a mention. In the last paragraph, Davenport writes that rapid and thorough improvement of the human race can be achieved only through understanding and applying these methods. He also laments, using the word “appalling,” that people are ignorant of the comparison between race mixture and pure breeding. Davenport was a eugenicist who went on to found the Eugenics Record Office in 1910. The photographs of students lined up by height appear in a paper that closes with such claims. This article reads only the “arrangement” part of the photographs.

One Line by Height: The Curve Traced by the Tops of Heads

The caption of Fig. 1 reads “File of Forty University of Chicago Students arranged (approximately) in order of Height.” Forty students stand in a single line, roughly in order of height.

A photograph in front of a board fence: 40 University of Chicago students stand in one line in roughly height order, with heads getting higher toward the right

Source: C. B. Davenport (photographer unknown), Popular Science Monthly 59 (1901), p.449, from the Internet Archive (public domain)

In the text, Davenport describes how to read this line.

Imagine a file of 40 men arranged in order of stature—the tallest at the head of the column. The crown of their heads will form a flowing curve, nearly level in the middle of its course and becoming more oblique at the ends. The reason for this is that the middle statures are much more common than the extreme ones.

If 40 people are arranged by height, the line of their crowns forms a smooth curve, nearly level in the middle and steeper at both ends. The reason is that middling heights are far more common than extreme ones. He continues: “Half the people have a stature within two inches of the average.” That is, half of the people fall within two inches of the mean.

In this arrangement, one person is one unit. The height is the person’s actual physical height. The horizontal position is the rank in height. The line connecting the tops of the heads becomes a curve of rank against value. In today’s terms, this can be read as close to a quantile plot, or a bar chart sorted by value.

In this arrangement the distribution is not directly visible; what we see is the slope. It can be read that the gentler the slope in a stretch, the more people of similar height there are. Each person’s value is preserved, but “how many people there are” can only be inferred from the slope.

One Column per Class: Photographed from Directly Above

In the next paragraph, Davenport explains how to make a frequency polygon on paper.

The frequency polygon is got first by sorting out the data into a number of equally extensive ‘classes’; then by laying off these classes as a series of points at equal intervals of space along a horizontal base-line; by erecting perpendiculars proportional to the frequency of each class, and by joining with a line the tops of all these perpendiculars.

The steps are: divide the data into classes of equal width; place the classes at equal intervals along a horizontal baseline; erect perpendiculars proportional to the frequency of each class; and join the tops of the perpendiculars with a line. Right after that comes the sentence at the heart of this article.

Such frequency polygons may also be obtained, without drawing on paper, by putting the individuals belonging to the same class in the same vertical column and arranging the columns in order along a common base-line.

The claim is that a frequency polygon can be made without drawing on paper: put individuals of the same class in the same column, and arrange the columns in order along a common baseline.

The students’ classes are eight: 56–57.9, 58–59.9, 60–61.9, 62–63.9, 64–65.9, 66–67.9, 68–69.9, and 70–71.9 inches. This divides roughly 142 to 183 cm in steps of about 5 cm. Students of the same class form one line, and the lines are arranged in order from the baseline. Fig. 2 is that arrangement taken as a “bird’s-eye view,” that is, photographed from directly above.

A photograph looking down from a building: students wearing hats stand in several lines running from the baseline in the foreground toward the back, and the line slightly right of center is the longest

Source: C. B. Davenport (photographer unknown), Popular Science Monthly 59 (1901), p.450, from the Internet Archive (public domain)

Even with the same 40 people as in Fig. 1, the way height is expressed has changed. Height is shown not by how tall the person is, but by which column the person stands in. The length of a column is the number of people. Because the photograph is taken from above, each person’s physical height does not appear. In exchange for giving up individual values, frequencies become visible as lengths, as can be read.

The number of students in each class is not given in the body text of the magazine. In the reading by Brinton, who reprinted the photographs in 1914, the lowest class has one person, and each of the two highest classes also has one. The class with the most people is the 64–65.9 inch column.

This format can be read as a good match for the mode, which Davenport valued. He wrote that the mean is “over-rated and over-used,” and described the mode as follows.

it is not the result of calculation and is not an ideal value merely, but is the prevailing or typical actual condition.

The claim is that the mode is neither the result of a calculation nor an ideal value, but the actual condition that prevails most. In the columns by class, that condition can be read as appearing, without any calculation, as the longest column.

Piles of Shells and Three Arrangements

Davenport shows the same construction with shells. In Atlantic coast scallops (Pecten), the number of ribs on the shell differs from individual to individual. According to the text, among 100 individuals from one locality, the number of ribs varies from 15 to 21.

If we put in one pile the shells having the same number of ribs and arrange the piles in order, from 15 to 20, upon a level base we shall get a figure which is the frequency polygon for the ribs of Pecten shells from the given locality (Fig. 4).

Put shells with the same number of ribs in one pile, and arrange the piles in order from 15 to 20 on a level base. The result is the frequency polygon of the ribs of scallops from that locality.

On the left, two scallop shells with 15 and 20 ribs; on the right, six piles of shells stacked by number of ribs, with the third pile from the left the tallest

Source: C. B. Davenport (photographer unknown), Popular Science Monthly 59 (1901), p.451, from the Internet Archive (public domain)

The two shells on the left are Fig. 3, with 15 and 20 ribs. They correspond to the two ends (15 and 20 ribs) of the piles arranged in Fig. 4. The right side is Fig. 4, which Davenport named “Self-formed Frequency Polygon of Pecten Ribs.” According to Brinton, the leftmost pile (15 ribs) has three shells, the rightmost pile (20 ribs) has one, and the pile of 17 ribs is the tallest.

The table below sets out how data corresponds to appearance in the three arrangements.

Arrangement	One unit	Horizontal position	Length / height	What is visible
Fig. 1: one line by height	One student	Rank in height	The height itself (the person’s actual stature)	Each individual’s value, and a curve of rank against value
Fig. 2: columns by class (bird’s-eye view)	One student	Height class (2-inch width, 8 classes)	Column length = number of people in the class	Frequency polygon. The mode is the longest column
Fig. 4: piles of shells	One shell	Number of ribs (15–20, 6 divisions)	Pile height = number of shells with that rib count	Frequency polygon. The mode is the tallest pile

Fig. 1 and Fig. 2 use the same 40 students as material. What differs is what is assigned to the horizontal position, and what is not photographed. Fig. 2 and Fig. 4 have different materials but form the same type of figure.

Davenport’s own term is “frequency polygon,” and the word “histogram” does not appear in the magazine. Calling it a “unit chart” can also be read as a later reclassification.

How Third Parties Viewed These Photographs

Assessments divide by point of interest across visualization research, practical guides to charts, and statistics education.

Pierre Dragicevic, who registered this entry on dataphys.org, summarizes its purpose as “to explain the notion of statistical distribution to a lay audience.” He then writes the following.

His examples also show that spatial arrangement is a key feature of data visualizations, and that a good spatial arrangement can augment cognition even without the use of abstract mappings and encodings.

The assessment is that a good spatial arrangement can aid cognition even without abstract mappings or encodings. On eugenics, he mentions only briefly that it “may be controversial,” and says the educational value of the explanation is clear.

Willard C. Brinton reprinted the piles of shells and the line of students in Chapter 10 of his 1914 chart textbook Graphic Methods for Presenting Facts. At the start of the chapter, Brinton writes that engineers had hardly used the frequency curves that biologists use. Davenport’s photographs were what he chose as the introduction. At the same time, he points out the following.

Though it is unfortunate that no horizontal scale or vertical scale is shown in the picture, the reader will nevertheless be able to see from this illustration the general scheme on which frequency curves are based.

The assessment is that it is a pity there is no scale on either axis, but the idea behind frequency curves still comes across. For the students’ photograph, he even reads off an asymmetry: there are more people below the mode than above it.

The paper “Is Human Height Bimodal?” (2002) by Schilling, Watkins, and Watkins says that a “living histogram” of height has become a standard classroom demonstration. The oldest example they found was a 1914 photograph of 175 male students at Connecticut Agricultural College lined up. The paper does not mention Davenport, but the 1901 photographs can be read as an example 13 years earlier. The same paper warns that even when a histogram appears to have two peaks, one cannot say the underlying distribution is bimodal unless the sample is very large. It can be read that the unevenness in a line of 40 should likewise not be over-interpreted.

Takeaways

What these photographs leave for visualization work is this: deciding the arrangement is deciding the type of graph. From the same 40 people, Fig. 1 and Fig. 2 made a chart of rank and value and a chart of frequency.

The same holds for on-screen charts. A bar chart sorted by value keeps each individual value and shows the distribution only through its slope. A histogram that counts by class discards individual values and shows frequency as length. Which to show depends on whether you want to answer the reader’s question “who has how much” or “how many are at what level.”

A format in which one individual is one unit shows the mode without counting. It is a form suited to the first picture in an explanation.

Brinton’s point also applies. Even with physical objects lined up, without a scale you cannot read the number of people or the classes. In both physical arrangements and on-screen charts, the axes and the definitions of classes need to be shown separately.

When you line up participants in a class or workshop, making both a single line and columns by class lets them experience the same people becoming different graphs. However, in a small group, unevenness in the lines can arise by chance alone.

References

Primary sources

Davenport, C. B. “The Statistical Study of Evolution.” Popular Science Monthly, Vol.59, Sept. 1901, pp.447–460.
- Wikisource (transcription): https://en.wikisource.org/wiki/Popular_Science_Monthly/Volume_59/September_1901/The_Statistical_Study_of_Evolution
- Pages with figures (Fig. 1–4): https://en.wikisource.org/wiki/Page:Popular_Science_Monthly_Volume_59.djvu/459 , /460 , /461
- Page scans (Internet Archive): https://archive.org/details/popularsciencemo59newy
Davenport, C. B. Statistical Methods, with Special Reference to Biological Variation, 2nd ed., 1904: https://archive.org/details/statisticalmetho00daverich
Wikimedia Commons file pages (figure metadata and licenses)
dataphys.org “Davenport’s Physical Distributions”: https://dataphys.org/list/davenports-physical-distributions/

Secondary sources

Pierre Dragicevic, “Davenport’s Physical Distributions,” dataphys.org: https://dataphys.org/list/davenports-physical-distributions/
Willard C. Brinton, Graphic Methods for Presenting Facts, Engineering Magazine Co., 1914, pp.164–165 (Fig. 140, 141): https://archive.org/details/graphicmethodsfo00brinrich
Schilling, M. F., Watkins, A. E., & Watkins, W., “Is Human Height Bimodal?” The American Statistician 56(3), 2002, pp.223–229: https://www.jstor.org/stable/3087302 (PDF: https://staff.washington.edu/tamre/IsHumanHeightBimodal.pdf )
Wikipedia “Charles Davenport” (biography and eugenics activities): https://en.wikipedia.org/wiki/Charles_Davenport

Stand a Frequency Table Up as Columns and the Ridge Shows Its Direction: Pearson and Lee's Stature Model

Sun, 01 Nov 2026 00:00:00 +0900

In the Department of Statistical Science at University College London (UCL), there is a framed model. It replaces a frequency table, in which the statures of 1,078 fathers and sons were counted on a grid in one-inch steps, with the heights of rectangular columns. Looking at the arrangement of the columns, you can see the peak of the distribution and the diagonal direction of its ridge (correlation coefficient r = 0.514) as a shape, without reading a table of numbers.

The data were collected by Karl Pearson and Alice Lee, and the original table is Table XXII of their 1903 paper “On the Laws of Inheritance in Man.” No record has been found of who made the model or when. What follows is a reading based on the dataphys.org entry, which collects works of data physicalization, and on the 1903 paper.

1893: Handing Out Cards to Measure Families

Pearson opens the paper by explaining why the measurements were begun: to supplement the family data Francis Galton used in Natural Inheritance with wider measurements (p.357).

The instructions were drawn up around 1893 with his colleague W. F. R. Weldon. The paper prints the circular asking for help, “FAMILY MEASUREMENTS.” The families to be measured were a father, a mother, and at least one child aged 18 or over. The number of families needed is given as “1000—2000 families.” Stature was to be measured with shoes off and a book held vertically against the head (p.359).

Students did much of the measuring, and it was Lee who supported the work of gathering the data.

“In the course of four to five years about 1100 cards were filled in. The tabling of the data on these cards and the calculation of the statistical constants, some 78 tables in all, are due entirely to Dr Lee, and occupied her spare time for nearly two years."（p.358）

In short: about 1,100 cards were collected over four to five years, and Lee did all of the tabling into 78 tables and the calculation of the statistical constants, which took nearly two years of her spare time. The paper’s byline is likewise “By KARL PEARSON, F.R.S., assisted by ALICE LEE, D.Sc.”

Stature was recorded to the quarter inch and tabulated to the inch. For people measured with their shoes on, one inch was subtracted from the record (p.361).

Counting 1,078 Pairs into 340 Cells

The father-son statures are compiled in Table XXII, “Father’s Stature and Son’s Stature” (p.415), in the paper’s appendix.

A frequency table with the father’s stature across and the son’s stature down in one-inch steps, with the number of pairs written in each cell. The grand total at the bottom right is 1078

Source: Karl Pearson and Alice Lee, Biometrika vol. 2 (1903), p.415 (published 1903)

Item	Content
Number of pairs	1,078 pairs (the grand total at the bottom right of the table)
Classes of father’s stature	17 classes, from 58.5–59.5 to 74.5–75.5 inches
Classes of son’s stature	20 classes, from 59.5–60.5 to 78.5–79.5 inches
Number of cells	17 × 20 = 340. 179 have nonzero frequencies
Most populated cell	31.5 pairs, father 66.5–67.5, son 67.5–68.5
Correlation coefficient	.514 ± .015 (Table IV, p.378)

The frequencies include fractions such as 0.25 and 0.5, and 129 of the 179 nonzero cells have fractions. This can be read as the result of splitting a pair equally between neighboring cells when a quarter-inch record fell on a class boundary. However, in the range I examined (pp.357–361 and the pages of the table), I have not found a passage that says so explicitly.

The shape of the distribution is already there in the table of numbers. But to read the position and direction of the peak from 340 cells, you have to compare the numbers one by one.

Making Frequencies the Heights of Columns

The model turns this table into a solid object. Pierre Dragicevic, editor of dataphys.org, lists it as a bivariate histogram.

Gray and gold-tinged rectangular columns stand in a grid on a base inside a frame, with the tops of some columns red. Below is written MODEL TO ILLUSTRATE THE INHERITANCE OF STATURE IN MAN

Photo: Richard Chandler (UCL Department of Statistical Science) / From dataphys.org, “1900 – Pearson and Lee’s Height Correlation Chart”

The lower part of the frame reads “MODEL TO ILLUSTRATE THE INHERITANCE OF STATURE IN MAN,” the top edge reads “Stature of Fathers,” and the left edge reads “Stature of Sons.” The correspondence between the data and the form is as follows.

Data element	Form in the model	Certainty
Class of father’s stature	Position along the horizontal direction. The scale on the top edge runs 76→58 from the left, so the right is lower	dataphys.org and the photo
Class of son’s stature	Position along the depth direction. The scale on the left edge runs 59→79 from bottom to top	dataphys.org and the photo
Number of pairs in each cell	Height of a square column	dataphys.org (reproduced in R)
Cell with frequency 0	No column; the surface of the base is visible	Reading from the photo
Red tops on some columns	Meaning unknown	No source explains it
Material	Not recorded. In the photo: gray, gold-tinged, and red	Photo only

Dragicevic took the father-son pairs from the Pearson and Lee data in the R package HistData and drew them with plot3D::hist3D.

“The following R code replicates the older model exactly, confirming that it shows the original dataset from Pearson and Lee”

That is, the code reproduces the old model exactly, confirming that it shows Pearson and Lee’s original data.

In the photo, tall columns gather in the center and become lower and sparser toward the edges. The band of tall columns runs from the lower right to the upper left. Since the father’s scale gets lower toward the right and the son’s scale gets higher toward the top, this direction can be read as extending toward the side where the taller the father, the taller the son. The position of the peak and the tilt of the ridge catch the eye without comparing the numbers in the table.

Neither the maker nor the year of making is known. dataphys.org only estimates that it may have been made under Pearson’s supervision around the same time as the data collection or a little later. The “1900” in the title is also an estimated year.

The rest of the history is in a letter from Richard Chandler, a former head of the Department of Statistical Science. He says he found it in a desk drawer when he arrived at UCL in 1994. It had broken into a great many pieces, and he writes that he had it restored.

The Paper’s Line and the Model’s Peak

The paper shows the same 1,078 pairs in another form as well. Pearson writes that it was not possible to draw all 78 correlation tables, so three were chosen as representatives and drawn as diagrams. The first of them is Diagram I (p.362).

A diagram with circles marking the mean son’s stature for each father’s stature and a rising regression line drawn through them. The title is Probable Stature of Son for given Father’s Stature

Source: Karl Pearson and Alice Lee, Biometrika vol. 2 (1903), p.362 (published 1903)

The diagram reads “Regression Line: S = 33·73 + ·516 F. 1078 Cases.” Plugging into the formula, a father of 60 inches gives a son of about 64.7 inches, and a father of 74 inches gives about 71.9 inches. For each inch taller the father is, the son’s expected height is about 0.5 inch greater.

What Diagram I shows is the mean for each class of father’s stature and the line connecting them. What the model shows is the frequencies themselves, including the scatter around the mean. A relationship that the line condenses into a single slope can be read, in the model, as the width of the peak and the tilt of the ridge.

In his 1886 paper, Galton smoothed the frequency table of parents’ and children’s statures and joined points of the same value with lines.

“The figures in it were first “smoothed” as described in the memoir, then lines were drawn through points corresponding to the same values, just as isobars or isotherms are drawn."（Galton 1886, p.263）

That is, points of the same value were joined, as isobars or isotherms are drawn. Galton writes that the lines came out as concentric ellipses.

A diagram in which a tilted ellipse and lines for its major and minor axes are drawn over a frequency table of parents’ and children’s statures

Source: Francis Galton, Journal of the Anthropological Institute vol. 15 (1886), Plate X (public domain)

From here on is my own reading. Galton’s ellipses can be read as contour lines that treat frequency as height. The UCL model can be read as turning that height into actual columns.

dataphys.org also carries a photo of another model, a later one from a different period and with different data.

A model of rows of cream-colored columns, with a red dashed line drawn on the sides of the columns and a legend on the base reading REGRESSION LINE and MEAN

Photo: Pierre Dragicevic (mathematical model collection of the Institut Henri Poincaré) / From dataphys.org, “1900 – Pearson and Lee’s Height Correlation Chart”

It is a model held by the Institut Henri Poincaré in Paris, and its label reads “Model Prepared by Dept. of Statistics, University College London.” Chandler writes that it resembles an exhibit made for a Royal Statistical Society event in the 1980s (he adds “I think” himself). Dragicevic examined it closely and writes that its data differ from the old model’s.

From the photo, one can make out the red dashed line on the sides of the columns and the legend “REGRESSION LINE (MEAN STATURE OF SON FOR GIVEN STATURE OF FATHER).” The total in the margin reads “1000,” which may mean different data, adjusted to 1,000 cases. Whereas the old model shows only the peak of frequencies, the later model can be read as laying a regression line over the peak.

How Third Parties Have Seen This Model

Evaluations can be found in three directions, each with a different point.

Placing it in the history of data physicalization

Dragicevic listed this object on dataphys.org as an example of a “Passive physical visualization.” He recomputed the data to confirm that it matches the model, and also distinguished the similar later model as using different data. Chandler, on the side of the holder, writes:

“It is quite a nice piece of 19th-century data visualisation, I think!”

That is, he thinks it is a rather fine piece of 19th-century data visualization.

Recasting it as a one-point-per-pair scatter plot

In statistics courses, the same 1,078 pairs are taught as a scatter plot. A University of Chicago handout puts it this way.

“The following scatter diagram shows the heights of 1,078 fathers and their full-grown sons, in England, circa 1900. There is one dot for each father-son pair.”

That is, it explains the plot as a scatter diagram showing 1,078 father-son pairs in England around 1900, one dot per pair. A Brown University lecture slide describes the shape as like an American football and computes r = 0.5011627.

But the original source is a frequency table. Karl Broman’s repository states the procedure of interpolating values from Table XXII and randomizing them slightly to create the pairs. The scatter-plot version’s r of 0.501 differs from the paper’s 0.514, and this can probably be read as due to that reconstruction.

Reusing it as tables split by gender

Galton’s data had aggregated parents and their sons and daughters together. To check the cause of the slight nonlinearity seen there (Galton’s bend), Wachsmuth and colleagues used Pearson and Lee’s tables in 2003. The HistData description says:

“they used analogous data from English families at about the same time, tabulated by Karl Pearson and Alice Lee (1896, 1903), but where the heights of parents and children were each classified by gender of the parent.”

That is, they used data from English families of about the same period, tabulated separately by the gender of the parent. The same Pearson and Lee tables are still being used today as a column model, a scatter plot, and frequency tables split by the parent’s gender.

Takeaways

What we can take away from this model is the idea of choosing, from the purpose, how to show a bivariate frequency table.

At least four forms have been made from the same kind of table: Galton’s ellipses, Pearson’s regression diagram, this model’s columns, and the textbook scatter plot. Scatter plots, heatmaps, and 3D histograms can all be made from a “bivariate frequency table.” What differs is what the frequencies are mapped to.

Making frequencies the heights of columns shows the position of the peak and the direction of the ridge together. Even frequencies with fractions can be turned into shape as they are.
Making frequencies the intensity of color (a heatmap) shows the same table on a flat plane, with no part hidden behind a column in front.
To make a one-point-per-pair scatter plot, values must be reconstructed from the frequency table, and the statistics shift slightly.
Turning only the means into a line (a regression diagram) condenses the relationship into a single slope and hides the scatter.

It is also instructive that the later model wrote a regression line on top of the columns. Overlaying the peak and the line lets you compare the movement of the mean and the actual scatter at once.

When showing a correlation, we sometimes stop after writing the value of r. This model puts in your hands the shape of the peak that r = 0.514 came from. Show the shape of the frequencies before the number. That is the takeaway.

References

Primary sources

Pierre Dragicevic, “1900 – Pearson and Lee’s Height Correlation Chart,” List of Physical Visualizations (dataphys.org). Includes the quotation from Richard Chandler’s reply, the R code, and the photo credits. dataphys.org (mirror: GitHub)
Karl Pearson (assisted by Alice Lee), “On the Laws of Inheritance in Man: I. Inheritance of Physical Characters,” Biometrika 2(4): 357–462, 1903. pp.357–361 (background and measurement procedure), p.362 (Diagram I), p.378 (Table IV), p.415 (Table XXII). JSTOR (scanned PDF: kbroman/PearsonData)
HistData package (PearsonLee data and description). CRAN
Francis Galton, “Regression towards Mediocrity in Hereditary Stature,” Journal of the Anthropological Institute 15: 246–263, 1886. galton.org

Secondary sources

Karl Broman, PearsonData (transcription of Table XXII, and the procedure for interpolating and randomizing it into per-pair values). GitHub
University of Chicago Stat 200 handout “Pearson’s Father-Son Data”. PDF
Brown University CS100 “A Brief History of Regression”. PDF
A. W. Wachsmuth, L. Wilkinson, G. E. Dallal, “Galton’s bend,” The American Statistician 57: 190–192, 2003 (introduced in the HistData description). doi:10.1198/0003130031874

Strip Counts Set the Band Width, and a Photograph Inverts Black and White: IBM's Cosmograph

Tue, 03 Nov 2026 00:00:00 +0900

The Cosmograph (International Cosmograph) was a charting device sold by IBM. It expressed the width of a band by the number of paper strips laid side by side. One thousand white strips made 100%. On a black work board, the strips were bundled, divided, and held with clamps. The finished board was photographed with a photostat, and the negative print, with black and white inverted, was used as the graph.

The sources are IBM’s sales manual (including a page revised on January 31, 1934) and Chapter 8, “Flow Charts,” of Willard Cope Brinton’s textbook on charts, Graphic Presentation (1939). The manual was found in IBM’s archives in 2012 by Jim Strickland, a volunteer at the Computer History Museum, who presented it with images and a transcription. It also appears as “1933 – IBM’s Cosmograph” in dataphys.org, a catalog of examples of data physicalization. What follows sticks to these materials.

A Device in IBM’s 1933 Sales Manual

The 1933 U.S. copyright catalog lists a chapter of IBM’s “Manual of sales instruction, 1933” as “International cosmograph. © June 9; AA 130049.” What was registered was a document, a chapter of a sales manual, not rights to the device itself.

What the device was can be learned from the 1934 sales manual that Strickland found. Its preface says:

The drawing of graphs usually requires the time and skill of an accomplished draughtsman. The Cosmograph, however, due to its simplicity of construction and operation, requires no skill or foreknowledge beyond that imparted in a few moments’ instruction.

In short: drawing graphs takes the time and skill of a trained draftsman, but the Cosmograph requires no skill beyond a few moments of explanation.

A page from IBM’s sales manual. At the top is a photograph of the device, in which white paper strips split left and right on a black board; below it are the price, uses, components, dimensions, and weight

Source: IBM, Manual of Business Instruction (Rev. 1-31-34), photographed by Jim Strickland, from the Computer History Museum “Volunteer Information Exchange” Vol.2 No.16

The “SPECIAL DEVICES” chapter of the same manual has a page with a photograph of the device and its price. The price is “$50.00 each, f.o.b. Endicott, N. Y.,” that is, 50 dollars per unit. However, the upper right of this page reads “Rev. 1-31-34,” a revision of January 31, 1934. Whether this was the 1933 price has not been confirmed.

The components are a grooved work board with Bakelite fasteners, 1,000 paper strips, wedges and clamps, edge guides, and a carrying case. The strips are described as “1000 strips of white charting paper divided by blue strips at each 1% mark and red strips at each 5% mark.” The dimensions are 30 × 23 × 2 inches (about 76 × 58 × 5 cm, converted by the author).

A note on the names used at the time. IBM’s wording is “International Cosmograph,” and Brinton explains “Cosmograph” as a trade name for a kind of flow chart. A flow chart that shows quantity by band width is now classified as a Sankey diagram, and the dataphys.org tag uses that name. But this is a later classification; it appears in neither the manual nor Brinton’s book.

Who devised it, how many were sold, and where an actual unit survives could not be confirmed.

Count 1,000 Strips to Make the Band Width

Page 78 of Brinton’s book shows how it was used, with photographs provided by IBM.

Two photographs: on the left, hands pressing and holding a bundle of white strips on a black board; on the right, a tool being inserted between the bundles

Source: Willard C. Brinton, Graphic Presentation (1939), p.78 (plate: International Business Machines Corp.) / from the Internet Archive (Internet Archive rights status: NOT_IN_COPYRIGHT)

One thousand strips of paper are set on edge to represent 100%, and are separated into component parts of 100%.

That is, 1,000 paper strips are stood on edge to make 100% and then divided into its components. The left photograph shows the stage of positioning the strips and holding them with clamps; the right shows wedges and bar spacers being inserted between the bundles.

The manual describes how quantity is laid out:

It requires no measurements or computing for measurement. … Each minimum percentage unit of data is represented by one white strip.

That is, no measuring and no calculation for measuring is needed; the smallest unit of a percentage is represented by one white strip. One strip is equivalent to 0.1% (the author’s calculation).

The width of a band is decided not by measuring with a ruler and drawing, but by counting strips. The correspondence between data and device is summarized below.

Data side	Device side	Source
The whole (100%)	1,000 strips	Brinton p.78
Smallest unit of a percentage	One white strip (0.1%, the author’s calculation)	Manual
Share of each item	Number of strips in a bundle = band width	Manual
Scale mark every 1%	Blue strip	Manual
Scale mark every 5%	Red strip (20 strips)	Manual, Brinton p.78
Divisions between items	Wedges and bars between bundles	Brinton p.78
Path of a flow	Position where strips are bent and held with clamps	Brinton p.78
Frame of the figure	Edge guides	Brinton p.78

The scale is inside the band, not outside it. The blue and red strips enter the same bundle as the white strips, at fixed positions. This can be read as a design that gives the band itself its scale, without placing an axis outside.

Because the smallest unit is one strip, proportions are placed in steps of 0.1% (the author’s calculation). In exchange, the accuracy of band width is supported by the number of strips, not by drafting skill. The manual could say that no skill was required, and this mechanism can be read as the reason.

Invert the White Strips on the Black Board with a Photograph

The finished board is not the finished product. Brinton writes that once the edge guides hide the leftover ends of the strips, “the Cosmograph is ready for photostatting.” That is, it is ready to go to the photostat.

On the left, a photograph of the completed device with bands of white strips spreading up and down on a black board; on the right, a negative print in which the same arrangement appears as black bands on a white ground

Source: Willard C. Brinton, Graphic Presentation (1939), p.78 (plate: International Business Machines Corp.) / from the Internet Archive (Internet Archive rights status: NOT_IN_COPYRIGHT)

Note that all black portions of the device fail to reproduce. Of the one thousand strips of paper, twenty are red and are set at each 5% mark. In the negative photostat, these red strips of paper reproduce as white.

That is, all the black parts of the device fail to reproduce, and the 20 red strips, set at every 5% mark, come out white in the negative. The arrangement of white strips on the black board becomes black bands on a white ground in the print. The red strips remain as thin white lines inside the black bands. How the blue strips appeared in the negative could not be confirmed.

The manual also says that what is used as the graph is the print.

When all groups have been arranged a photostat of the board is made. The negative print, on which the groups appear in black against a white background, is used as the graph.

That is, once all the groups are arranged, a photostat of the board is made, and the negative print, in which the groups appear black on white, is used as the graph. The strips on the board were not the final form for people to read but an intermediate form for producing a printed piece. The black board was meaningful precisely because it did not reproduce.

Two Figures Built for Reparations and for Income and Outgo

Page 80 of Brinton’s book carries two figures made with the Cosmograph. The first is the distribution of German reparations.

A flow chart in which bands split from the total on the left to France, Britain, Italy, and others; some stop partway, and the rest gather at the United States on the right

Source: Willard C. Brinton, Graphic Presentation (1939), p.80 (plate: International Business Machines Corp.) / from the Internet Archive (Internet Archive rights status: NOT_IN_COPYRIGHT)

On the left are the total Germany paid and the countries that received it; in the center is the amount each country kept; on the right is the amount paid from each country to the United States. The amount kept is shown by branches that stop partway. According to Brinton, this is made by pulling the strips back and aligning their ends at the center of the figure. They are only pulled back, and cutting is not mentioned, so the total number can be read as unchanged.

The second is “Simple Income and Outgo,” a figure of plain income and expenditure.

A flow chart in which income bands entering from the left merge into a thick central trunk, which then splits to the right and downward into expenditure bands such as salaries, taxes, dividends, and savings interest

Source: Willard C. Brinton, Graphic Presentation (1939), p.80 (plate: International Business Machines Corp.) / from the Internet Archive (Internet Archive rights status: NOT_IN_COPYRIGHT)

The central trunk is fixed first, the income side is assembled, and then the board is turned to assemble the outgo side. About the space for the amount on the trunk, Brinton writes:

A short strip of black paper is pasted across the trunk to provide a white block on the negative photostatic print.

That is, pasting a short strip of black paper across the trunk makes a white block in the negative, where the total amount is entered in type. Even the label space was made on the assumption of inversion. On the printed page, the white block reads “$2,193,427,000.”

The figure also has a vertical note, which the author read from the scan of the page.

EACH INTERVAL BETWEEN THE WHITE LINES INDICATES 5 PER CENT OF THE TOTAL … IT IS NECESSARY ONLY TO COUNT THE INTERVALS AND FRACTIONAL INTERVALS.

The note says that each interval between the white lines is 5% of the total, and that to know a proportion one need only count the intervals and fractional intervals. It can be read as a mechanism in which the reader counts the quantity the maker placed by counting strips.

How Third Parties Viewed the Device

Jim Strickland saw “cosmograph” on an exhibit poster at the Computer History Museum, looked into what it was, and in October 2012 found the sales manual in IBM’s archives.

In 1933, for $50.00, ($850.00 in 2012) a business could buy a device to illustrate financial inputs and outputs by separating and grouping strips of white paper against a black background.

That is, in 1933, for 50 dollars, a business could buy a device that illustrated the flow of money in and out by grouping strips of white paper against a black background. The article ends with “Eat your heart out Excel!”, treating the device as an ancestor of the charting features of spreadsheet software. However, the page cited for the price is the January 1934 revision.

The Thomas Fisher Rare Book Library at the University of Toronto, in its 2023 exhibition “Emerging Patterns,” placed Brinton’s figure in the history of the Sankey diagram and wrote:

The replacement of ‘cosmograph’ with ‘Sankey’ is a loss for the world of data visualization.

That is, the replacement of the name Cosmograph with Sankey is a loss. The exhibition describes the method as “using pins to guide the paths of the strips,” but what Brinton and the manual describe are clamps and wedges; pins do not appear.

Phineas of sankey-diagrams.com wrote in 2009 that Brinton’s Chapter 8 puts its emphasis on “Cosmographs,” which seems to have been an IBM trade name of the time. Another article on the same site reports that, in a discussion among Spanish-speaking makers, “cosmografías” was raised as a term in James Glenn Stovall’s Infographics (1997). This can be read as an example of a trade name surviving as a generic name for flow charts, but I have not checked Stovall’s original.

What to Take From It

The first is a design that “counts” quantity instead of “measuring” it. When you draw by length, accuracy depends on the skill of the person drawing. When you place by number of units, accuracy can be read as supported by the counting. In exchange, precision is set by the size of the unit, which for the Cosmograph works out to steps of 0.1%. Choosing the unit also chooses the precision.

The second is a design that puts the scale inside the mark. The red strips became white lines inside the bands in the print, and counting those lines gave the proportion. Even without an axis outside, clues for reading can be left inside the band.

The third is a view that treats a physical object as an intermediate form for output. The black board was meaningful because it did not reproduce, and even the label space assumed inversion. Deciding first what the result will pass through at the end changes how colors and materials are chosen.

Finally, the name. At the time, these figures were called Cosmograph or flow chart. If you search for past figures only under the classification “Sankey diagram,” you may overlook figures that went by different names.

References

Primary sources

IBM sales manual “Manual of Business Instruction” (1934): the preface, the explanatory text, and the “SPECIAL DEVICES / COSMOGRAPH” page (Rev. 1-31-34). Reproduced as images and transcription in Jim Strickland’s article: http://s3data.computerhistory.org.s3.amazonaws.com/chmedu/VIE_02_016.pdf
Catalog of Copyright Entries, New Series, Part 1 Group 2: Pamphlets, Leaflets…, Vol.30 (1933), entry “International business machines corp.” with “International cosmograph. © June 9; AA 130049.”: https://archive.org/details/catalogofcopyri301libr (relevant page: https://archive.org/details/catalogofcopyri301libr/page/n769/mode/1up )
Willard Cope Brinton, Graphic Presentation, New York: Brinton Associates, 1939, Chapter 8 “Flow Charts,” pp.73–80, and the glossary at the end: https://archive.org/details/graphicpresentat00brinrich (p.78: https://archive.org/details/graphicpresentat00brinrich/page/78/mode/1up / p.80: https://archive.org/details/graphicpresentat00brinrich/page/80/mode/1up )
dataphys.org, “1933 – IBM’s Cosmograph” (entered by Pierre Dragicevic): https://dataphys.org/list/ibms-cosmograph/

Secondary sources

Jim Strickland, “Cosmograph? What’s a Cosmograph?”, Volunteer Information Exchange Vol.2 No.16, Computer History Museum, 2012-11-23: http://s3data.computerhistory.org.s3.amazonaws.com/chmedu/VIE_02_016.pdf
Thomas Fisher Rare Book Library, University of Toronto, “Emerging Patterns: Data Visualization Through History” (2023), “Sankey and Venn Diagrams”: https://fisherdigitus.library.utoronto.ca/exhibits/show/emerging-patterns--data-visual/network-diagrams/network-diagrams--sankey-and-v
Phineas, “1939 Graphic Presentation Book”, sankey-diagrams.com, 2009-09-16: https://www.sankey-diagrams.com/1939-graphic-presentation-book/
Phineas, “Infographics Experts on Sankey Diagrams (Part 2)”, sankey-diagrams.com, 2009-08-18: https://www.sankey-diagrams.com/infographics-experts-on-sankey-diagrams-part-2/

160 Years of Paris Emptying Out at Its Center, Filmed in 10,000-Person Cubes: Chronique Démographique de la Région Parisienne

Sun, 01 Nov 2026 00:00:00 +0900

Chronique Démographique de la Région Parisienne 1801 - 1961 is a film in which columns of small cubes, one cube per 10,000 people, are stacked for each commune and arranged on a map, then shot in stop motion, year by year, as the narrator reads out the years. As the columns rise and shrink, you can see the movement of population between 1801 and 1961: the center of Paris empties while the suburbs swell.

The film is listed on dataphys.org, in the “List of Physical Visualizations” that collects examples of data physicalization, as an educational film from the 1960s. According to dataphys, it was made between 1962 and 1967 by ISHA and CMSS, in collaboration with Jacques Bertin’s cartography laboratory. This article reads the film through the dataphys entry and a transcript of the film’s narration.

1960s: White Columns Standing on a Map

The dataphys.org entry is titled “1965 – Stop Motion Animation of Physical 3D Map.” It was added on April 16, 2018, by Pierre Dragicevic, a researcher of data physicalization.

This educational movie from the 1960s uses physical bars and stop motion animation to show the evolution of population in the Paris area between 1801 and 1961. It was made between 1962 and 1967 by the Institut des Sciences Humaines Appliquées (ISHA) and the Centre de Mathématique Sociale et de Statistique (CMSS), in collaboration with the Laboratoire de Cartographie directed by Jacques Bertin.

In short: an educational film that uses physical bars and stop motion to show how the population of the Paris area changed, made by ISHA (Institute of Applied Human Sciences) and CMSS (Center for Social Mathematics and Statistics) in collaboration with Jacques Bertin’s cartography laboratory.

The “1965” in the entry title is a representative year chosen by dataphys; the body text does not pin the production down to a single year. The producers are ISHA and CMSS, and Bertin’s laboratory took part as a collaborator.

A film frame showing three columns of stacked white cubes standing on a map with drawn boundary lines.

Source: produced by ISHA and CMSS, held by the Archives nationales, digitized by dataphys.org / From dataphys.org “List of Physical Visualizations”

The film was provided by the French National Archives (Archives nationales), digitized by the dataphys curators, and uploaded to YouTube by Dragicevic.

I have found no reliable source for the names of the director or cameraperson, or for the running time. What follows is limited to what the narration tells us.

One Cube = 10,000 People, One Column = One Commune

The film opens with “Paris et sa banlieue.” (Paris and its suburbs), and the next two sentences serve as the legend.

Chaque colonne de petits cubes représente la population d’une commune.

Chaque petit cube représente 10 000 habitants.

Each column of small cubes represents the population of one commune. Each cube represents 10,000 inhabitants. dataphys writes “physical bars,” but according to the narration, each column is not a single block of wood but a stack of small cubes. The height of a column is determined by the number of cubes stacked, that is, by the population.

The table below summarizes how the data maps to what you see.

Data	Visual form / technique	Confirmed in
Population (quantity)	Number of stacked cubes. One cube = 10,000 people; this sets the column height	Narration
Place (commune)	One column = one commune, placed at that commune’s position on the map	Narration; dataphys title and tags
Time (1801–1961)	Columns are rebuilt for each year the narrator reads out and shot one frame at a time (stop motion)	Narration; dataphys
Pace of time	The speed of frame-by-frame shooting (cadence) changes from scene to scene	Narration
Regional grouping	Scenes are divided: suburbs only, all of Paris, groups of arrondissements, three rings	Narration
Viewpoint	Can be seen from all sides (dataphys.org, 18:40); seen in cross-section (narration)	dataphys; narration
Material, color	Not confirmed (in the film frames the columns look white)	—

The narration does not tell us how fractions smaller than 10,000 people were handled, or what the columns were made of.

This form, showing a quantity per area as the height of a solid, touches on Bertin’s own theory. In 1966 Bertin wrote the paper “La cartographie statistique automatique” for issue 17 of the journal Mathématiques et sciences humaines, published by CMSS. The paper is bilingual, in French and English.

only a histogram in three dimensions, a “relief” can properly transcribe quantities by zones.

That is: only a three-dimensional histogram, a “relief,” can properly transcribe quantities by zone. Yet the same paper chooses, as the form to use on a paper map, a regular scatter of circles of increasing size (semis régulier de cercles croissants). The reasons were that the whole can be read at once, that it can be reduced a great deal, and that it is easy to automate.

Bertin’s original drawing of regional densities in France, rendered as a solid built up in tiers.

Source: Jacques Bertin, BnF GE CC-5541 (19-46) / DESIGNSHS “Le laboratoire graphique de Jacques Bertin”

Bertin’s laboratory also left original drawings of maps that depict quantities as solids (stéréogramme). The figure above is one of them, and it appeared in Sémiologie graphique. The film can be read as building, out of real cubes, the relief that was not chosen on paper, and adding time, which is hard to draw on paper, through stop motion. However, no source shows that Bertin himself decided the design of the film.

One Frame per Year Read Aloud, Repeated at Changing Speeds

In the film’s first scene, the narration simply reads out years. After “Paris et sa banlieue en 1801.” it continues with 1817, 1821, 1826, 31, 36, 41, 46, and so on, ending with “Et 1961.” The sequence of years can be read as following census years, but the narration does not say what kind of statistics they are.

The twentieth-century part proceeds through 1911, 1921, 26, 31, 36, 46, 51, 56. The years 1916 and 1941 are not read out. These can be read as years in which no census was held, in the midst of the two world wars. Note that the transcript was generated automatically and contains places that seem to be mishearings.

The order of scenes that can be read from the transcript is as follows.

Paris and its suburbs (1801→1961)
Suburbs only
All of Paris
The inner 11 arrondissements (Les 11 arrondissements intérieurs)
Arrondissements 1 to 4, then group by group
Arrondissements 5 to 7
The outer 9 arrondissements (Les 9 arrondissements périphériques)
Cross-section from Neuilly to Montreuil
Comparison at 40-year intervals (1801, 1846, 1886, 1926, 1956)
Averages of three rings (couronnes): central, middle, outer

After the scene of arrondissements 1 to 4, the narration announces a change in how the data is shown.

Examinons le mouvement démographique à des cadences cinématographiques de plus en plus rapides.

It means: let us examine the population movement at ever faster cinematic cadences. In later scenes the years are thinned out, as in 1801, 1856, 1876, 1921, 1946, 1956. In the scene of arrondissements 5 to 7, “cadence de prise de vue accélérée” (accelerated shooting cadence) is followed by “Cadence lente” (slow cadence).

Played fast, only the direction of increase and decrease remains; played slowly, you can follow what happened in which year. The structure can be read as showing both views of the same data, one after the other.

From Neuilly to Montreuil, the Center Empties

After the scene of the outer 9 arrondissements, “Démonstration rapide de l’évolution démographique.” (a rapid demonstration of the demographic evolution) is inserted, and the cross-section scene follows.

Vue en coupe de Neuilly à Montreuil. Alors que les côtés grandissent, le centre se vide.

A cross-section from Neuilly to Montreuil: while the sides grow, the center empties. This is the sentence in the film that states its answer most plainly. A cross-section removes the overlap in depth, and can be read as letting the heights of the columns on the section be compared from the side.

The narration then says “Reprenons plus lentement.” (Let us resume more slowly) and reads through the years from 1801 to 1956 again. The structure of the previous section, in which a fast playback shows the direction and a slow playback confirms it, repeats here.

A film frame showing many columns of stacked white cubes in the middle of a map, with a river and green areas around them.

Source: produced by ISHA and CMSS, held by the Archives nationales / From YouTube “Chronique Démographique de la Région Parisienne 1801 - 1961”

With solid columns, depending on the viewing angle, columns in front hide those behind. According to dataphys, the scene from 18:40 lets you see the model from all sides.

The final scene goes through the contrast of “Paris en 1801.” and “Paris en 1956.”, then the 40-year comparison, and ends with the averages of the three rings.

moyenne des trois couronnes centrales, médianes et extérieures en 1956, en 1801, et de nouveau en 1956.

That is: the average of the central, middle, and outer rings, shown for 1956, for 1801, and again for 1956. For viewers who have watched dozens of columns, this can be read as a way of closing by handing over the conclusion as three values.

How Third Parties Saw the Film

I have found no criticism or paper that discusses this film itself. Here are three items from different standpoints.

The first is its place as a means of telling stories with data. Dragicevic cites this film as an example in dataphys’s “Stop Motion Animation” entry.

For an example of how stop motion animation can be used for telling stories about data, see our entry 1965 - Stop Motion Animation of Physical 3D Map.

This points readers to the entry as an example of using stop motion to tell a story about data. dataphys classifies the film as a “Passive physical visualization” and tags it “storytelling.” It treats the film as a physicalization meant to be shown, not one the viewer touches and manipulates.

The second is its place as part of a laboratory’s work. The historian of science Jessica Martin describes the structure of Bertin’s laboratory in a 2023 paper. In 1961 the laboratory had a “laboratoire cinéma” (film laboratory) alongside the cartography, documentation and analysis, and photography sections, and Bertin called the whole a « centre graphique ». The paper does not mention the film. Even so, it can be read as work done in collaboration with a laboratory that had the setup to make films.

The third is the readability of physical 3D bar charts. “Data Physicalization” (2021) by Dragicevic, Yvonne Jansen, and Andrew Vande Moere introduces the CHI 2013 experiment by Jansen and colleagues.

They found that participants performed best with the 2D visualizations, but when a 3D representation was used, the physical condition outperformed even the stereoscopic display. A second experiment suggested that the likely reason was that physicalizations allowed participants to quickly and efficiently “bookmark” data points (bars) of interest with their fingers.

The result: performance was best with 2D charts, but among 3D representations, physical bars beat even a stereoscopic screen, probably because people could “bookmark” bars of interest with their fingers.

Viewers of the film cannot put a finger on the columns. dataphys writes, “You can see the physical visualization from all sides by jumping at 18:40.” The film also has a cross-section scene. These can be read as devices that reduce occlusion. Set beside the study’s conclusion, the film can be read as drawing out the advantage of physicalization not through touch but through the way it is shot.

Takeaways

What we can take from this film is how it handles the legend, time, and viewpoint.

With a unit of one cube = 10,000 people, you can count the cubes instead of reading the column height off a scale. Because the first two sentences state the unit outright, the viewer can afterward watch the columns while hearing only the years.

Time is marked not at equal intervals but by the years the narrator reads out (this can be read as following census years, though the kind of statistics is not stated). When showing change over time in an animation, matching the years you read out to the years of the data tells the audience which points are actual measurements.

Play fast to show the direction, play slowly to follow the years, and finally compress everything into the averages of three rings. It can be read as handing the audience three ways of seeing one dataset, the overall flow, the detail, and the summary, in a fixed order.

The occlusion problem can be read as answered with the cross-section. In the cross-section from Neuilly to Montreuil, the movement “the sides grow and the center empties” is visible without the front columns getting in the way. If you know what a solid figure is meant to say, you can look for the cut that shows that one sentence best.

References

Primary sources

dataphys.org “1965 – Stop Motion Animation of Physical 3D Map” (added by Pierre Dragicevic, 2018-04-16) — https://dataphys.org/list/stop-motion-animation-of-physical-bar-charts/
The film itself, “Chronique Démographique de la Région Parisienne 1801 - 1961” (YouTube, uploaded by Pierre Dragicevic) — https://www.youtube.com/watch?v=IYq9Nw9Izxk (from 18:40: https://www.youtube.com/watch?v=IYq9Nw9Izxk&t=18m40s). Title and uploader confirmed via oEmbed: https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=IYq9Nw9Izxk&format=json. Quotations from the narration are based on a transcript of this video.
CAMS (2015) Histoire du CAMS (a source cited by dataphys; not readable at the time of writing) — http://cams.ehess.fr/histoire-du-cams/ (Wayback: http://web.archive.org/web/20230529205235/http://cams.ehess.fr/histoire-du-cams/)
Jacques Bertin, “La cartographie statistique automatique”, Mathématiques et sciences humaines, 17, 1966, pp.71-76 — https://www.numdam.org/item/MSH_1966__17__71_0/ (PDF: https://www.numdam.org/item/MSH_1966__17__71_0.pdf)
Jacques Bertin, Carte de France des densités exprimées par stéréogramme (BnF GE CC-5541 (19-46)), DESIGNSHS — https://laboratoire-graphique.fr/plateforme/document/?id=9041__carte-de-france-des-densites-exprimees-par-stereogramme

Secondary sources

Pierre Dragicevic, dataphys.org “1897 – Stop Motion Animation” — https://dataphys.org/list/stop-motion-animation/
Jessica Martin, “Le laboratoire de graphique, au-delà de Jacques Bertin ? Les femmes du laboratoire de graphique”, Design Arts Medias, 2023 — https://journal.dampress.org/issues/femmes-de-sciences/le-laboratoire-de-graphique-au-dela-de-jacques-bertin-les-femmes-du-laboratoire-de-graphique
Pierre Dragicevic, Yvonne Jansen, Andrew Vande Moere, “Data Physicalization”, Springer Handbook of Human Computer Interaction, 2021 (HAL) — https://hal.science/hal-02113248v3/document
Yvonne Jansen, Pierre Dragicevic, Jean-Daniel Fekete, “Evaluating the Efficiency of Physical Visualizations”, CHI ‘13 — http://doi.acm.org/10.1145/2470654.2481359

Showing Yes and No Twice, in Paper and in Numbers: Hans Haacke's MoMA Poll (1970)

Sun, 01 Nov 2026 00:00:00 +0900

Hans Haacke’s MoMA Poll showed visitors’ votes in two ways at once. One was the volume of paper piling up in transparent ballot boxes, where you could compare the heights of the left and right heaps. The other was the number on a photoelectric counter attached to each box. On the last day of the exhibition, the counters read YES 25,566 and NO 11,563. The accumulating volume conveys the lean of the vote, and the number keeps the exact value. This double way of showing data can be read as a pattern that still works for physical visualization.

The work was shown in 1970 in the exhibition “Information” at the Museum of Modern Art (MoMA) in New York. What follows relies mainly on Haacke’s proposal on p.57 of the exhibition catalogue Information (1970), and on pp.9–11 of Framing and Being Framed: 7 Works 1970–75 (1975), where Haacke himself documented the work. I also use the later Tate Papers (2009) and Artforum (2004), and where they disagree I note it source by source.

A question handed in on the eve of the opening

The curator of “Information” was Kynaston L. McShine. The catalogue’s title page gives the dates as July 2 to September 20, 1970, but Haacke’s 1975 text says “From June 20 through September 20, 1970.” The Museo Reina Sofía’s commentary also says July 2 to September 20, so this article follows the catalogue.

Asked to take part, Haacke proposed that visitors vote on a two-choice question about a current social and political issue.

“The proposal was accepted, but I did not hand in the question until the evening before the opening of the exhibition."（Haacke 1975, p.9）

In other words, the proposal was accepted, but the question was not handed over until the night before the opening.

About two months before the opening, the U.S. military had bombed and invaded Cambodia, and at Kent State University the National Guard had shot and killed four protesting students. The president of MoMA’s board was David Rockefeller. His brother Nelson Rockefeller was the sitting governor of New York, running for re-election, and also a MoMA trustee.

The ballot boxes stood in the first room of the exhibition, with the question posted above them. The installation photograph in the 1975 book lets us read the text.

“Question: Would the fact that Governor Rockefeller has not denounced President Nixon’s Indochina policy be a reason for you not to vote for him in November? Answer: If ‘yes’ please cast your ballot into the left box, if ’no’ into the right box.”

The sign asks: does the Governor’s failure to denounce Nixon’s Indochina policy give you a reason not to vote for him in November? If yes, the left box; if no, the right box.

A black-and-white installation photograph: a sign with the question above the ballot boxes, and a woman putting a slip into the transparent box on the left

The wording differs by source. dataphys.org and Wikipedia quote “…be a reason for your not voting for him in November?” In David Rockefeller’s memoirs, as quoted in Tate Papers (2009), it reads “had not denounced.” This article uses the wording of the 1975 book, which is legible in the photograph.

According to Haacke, the day after the opening, an emissary of David came to the museum and demanded the work’s removal. Haacke reports this as hearsay, “Word has it” (Tate 2009) and “I was later told” (Artforum 2004). The museum’s director, John Hightower, did not comply, and the work stayed until the end of the exhibition.

The left box, the right box, and the numbers

The catalogue proposal describes the mechanism this way.

“Two transparent ballot boxes are positioned in the exhibition, one for each answer to an either-or question referring to a current socio-political issue. The question is posted with the ballot boxes. The ballots cast in each box are counted photo-electrically and the state of the poll at any given time during the exhibition is available in absolute figures."（McShine ed. 1970, p.57）

It places a transparent box for each answer, counts the votes photoelectrically, and makes the state of the poll at any moment available in absolute figures. The 1975 book also says that each box had a photoelectric counting device.

What matters is that the proposal specified both the volume of paper and the absolute figures from the start. The height of the heaps shows which side has more, and the counters keep the exact count. Two roles are layered onto one apparatus.

A color photograph of two transparent ballot boxes. The left box is mostly filled with colorful slips of paper, while the right box is only about half full

Hans Haacke (the source credits the photograph to Irmgard Emmelhainz’s e-flux article) / Source: dataphys.org

In the photograph, of the two identically shaped boxes, the left holds more paper. If the left is YES, as the sign says, this difference in heaps points the same way as the final tally (roughly 2.2 to 1).

The correspondence between data and physical properties is as follows.

Data	Physical property	Concrete form	Source where confirmed
Answer (YES / NO)	Position (left / right)	YES is the left box, NO is the right box	Sign in the photograph, 1975 book p.11
Votes per answer (the lean)	Volume (height of paper piled in the transparent box)	Compare the heaps’ heights across identical boxes	Installation photograph (1975 book p.11)
Votes per answer (exact value)	Number (photoelectric counter)	Counts each time a slip drops. Final tally YES 25,566 / NO 11,563	Catalogue p.57, 1975 book p.9
Visitor category	Color of the ballot	Plan to vary the color for full-price visitors, members, holders of invitation tickets, and free Monday visitors. Actual colors unknown	1975 book p.9
Day-by-day change, breakdown by category	Table (a tally sheet posted beside the boxes)	Plan for staff to write in the number of visitors and each box’s votes every evening	Catalogue p.57, 1975 book p.9

Note that YES meant “this is a reason not to vote for him.” The left heap therefore represents opposition to Rockefeller. The position mapping is simple, but what a heap meant was determined by the wording of the question, as I read it.

Calling this a “bar chart” comes from dataphys.org’s description of it as a “participatory physical bar chart.” Haacke himself calls it a “poll” and the containers “ballot boxes.”

How far the planned colors and tables actually worked

Haacke’s design had two more layers besides volume and numbers.

One was the color of the ballots. According to the 1975 book, the plan was to vary the color of the slips for full-price visitors, members, holders of invitation tickets, and visitors on free Mondays. However, the 1970 catalogue proposal contains no mention of color coding; it says the ballots would be serially numbered.

The other was the tally table. Every evening, staff were to write the number of visitors per category and each box’s votes into a table beside the boxes. The sample in the catalogue has columns for tickets sold, ballots handed out, and ballots in box 1 and box 2, plus a column for “certification of accuracy of figures by MoMA.”

The Haacke page of the catalogue Information. The proposal “POLL OF MOMA VISITORS” and a sample tally table with columns for the date and the number of tickets sold

In practice, things did not go as planned.

“In effect, ballots were not handed out regularly or according to the color code as directed. Consequently some visitors cast improvised ballots of torn paper."（Haacke 1975, p.9）

Ballots were not handed out as directed, nor according to the color code, and some visitors cast ballots of torn paper.

A heap of colorful ballots. Among the red, blue, yellow and white scraps is one with “yes” written by hand

The layer of colors marking categories did not function, and once handmade scraps were mixed in, the link between volume of paper and number of votes loosened too. In participatory physical visualization, whether the designed correspondence is realized depends on how it is run on site, as I read it. I could not confirm how far the tally table was updated during the exhibition.

37,129 votes, and the percentages that disagree

Here are the figures recorded by the automatic counters on the last day, September 20, copied from the 1975 book, p.9.

Item	Value
YES	25,566 (68.7%)
NO	11,563 (31.3%)
Total votes	37,129
MoMA visitors in the same period (including people who did not see Information)	299,057
Turnout	12.4%

Checking the arithmetic, the total of 37,129 and the turnout of 12.4% are consistent. Recomputing the ratio, though, gives YES 68.9% and NO 31.1%, which is 0.2 points off Haacke’s figures.

Later accounts vary further. Artforum (2004) summarizes it as “69% of the votes were against Rockefeller,” and Tate Papers (2009) writes this.

“68.7 per cent dropped their ballots into the ‘No’ box, indicating their opposition to Nelson Rockefeller; and 37.3 per cent voted in his favour.”

It says 68.7% put their ballots in the “No” box, showing opposition to Rockefeller, and 37.3% voted in his favour.

Under the 1975 record and the direction of the question, the box meaning opposition is the YES box. Also, 68.7 plus 37.3 makes 106. I read 37.3% as a misprint of 31.3%, and “No” as a slip for “Yes.” The same text also calls 299,057 the number of visitors to the Information exhibition, which differs from the 1975 figure for MoMA as a whole.

Haacke described the nature of the results in 2004 as follows.

“The results cannot claim to meet professional standards. However, they probably reflect the sentiments of the world of contemporary art in 1970. My piece gave these sentiments visibility in one of the most prominent institutions of modern art.”

The gist: the results do not meet professional standards, but probably reflect the sentiments of the contemporary art world in 1970, and the piece made them visible in a prominent museum.

How third parties saw the work

The sociologists Howard S. Becker and John Walton contributed an essay to Framing and Being Framed.

“It is clear enough in the case of the polls that Haacke has not followed standard procedures of questionnaire construction, sampling or analysis."（p.147）

This is a criticism that Haacke’s polls follow standard procedure in none of questionnaire construction, sampling, or analysis. At the same time, the two authors write that the flaws are not serious enough to cast major doubt on the conclusions, and that laying out every circumstance of the survey is faithful to scientific practice. The criticism targets “the polls” as a whole, mainly the later visitor surveys, and is not about the MoMA Poll alone, as I read it.

Jack Burnham, a critic and longtime friend of Haacke, called the work “Haacke’s Rockefeller Poll.”

“He noted that the poll served as a safety-valve for the political tempers of a surprisingly large number of museum visitors."（p.135）

This is Haacke’s own view that the vote served as a “safety valve” for the political feelings of a surprisingly large number of visitors. Burnham also relays the view that the museum did not make an issue of it, for reasons of public relations.

The “List of Physical Visualizations” at dataphys.org, compiled by researchers of data physicalization, lists the work as “1970 – MoMA Poll: Participatory Bar Chart” and describes it as a “participatory physical bar chart.” It does not touch on institutional critique, and I read it as treating the work as an early example of a visualization type.

The reading in this article is close to dataphys.org’s, but its entry does not mention the counters’ numbers.

Takeaways

The first takeaway is to layer volume and numbers in one apparatus.

The heap of paper is a volume representation that shows the lean without counting, and the counter’s number works as a record that keeps the exact value, as I read it (I could not confirm how the numbers were displayed). The heap loosens once scraps get mixed in. In physical visualization, it helps to prepare separately the volume that is shown from a distance and the number that is kept on record.

The second is that the question’s wording decides the meaning of the axis. In Haacke’s text in Tate Papers (2009), the box showing opposition is written as “No.” That conflicts with the direction of the question and looks like a slip. I read it as the question’s wording having made the meaning of the boxes easy to mix up. Writing what a heap represents on the sign as well should reduce misreadings.

The third is operation. Neither the color coding nor the tally table could exist unless staff acted every day. In participatory works, the encoding has to be designed to include the on-site procedures.

References

Primary sources

Kynaston L. McShine (ed.), Information, exh. cat., The Museum of Modern Art, New York, 1970. p.57 is Haacke’s proposal “POLL OF MOMA VISITORS”. PDF (Monoskop) / Overview page
Hans Haacke, “MOMA-Poll”, in Framing and Being Framed: 7 Works 1970–75, Halifax: Press of the Nova Scotia College of Art and Design / New York University Press, 1975, pp.9–11. PDF (Monoskop)
Hans Haacke, “Lessons Learned”, Tate Papers, no.12, Autumn 2009. Tate
Tim Griffin, “Historical Survey: An Interview with Hans Haacke”, Artforum, vol.43, no.1, September 2004. Artforum
David Rockefeller, Memoirs, New York, 2002, pp.452–3 (as quoted in Haacke, Tate Papers 2009). Tate

Secondary sources

Howard S. Becker & John Walton, “Social Science and the Work of Hans Haacke”, in Framing and Being Framed, 1975, p.145 ff. PDF (Monoskop)
Jack Burnham, “Steps in the Formulation of Real-Time Political Art”, in Framing and Being Framed, 1975, p.127 ff. PDF (Monoskop)
dataphys.org, “1970 – MoMA Poll: Participatory Bar Chart” (registered by Pierre Dragicevic). dataphys.org
Museo Reina Sofía, Repensar Guernica, “‘Proposal: Poll of MoMA Visitors’”. Reina Sofía
Wikipedia, “Hans Haacke”. Wikipedia
Monoskop, “Information (1970 exhibition)”. Monoskop

One Grain of Rice Is One Person, and Every Pile Uses the Same Conversion: Stan's Cafe, "Of All The People In All The World"

Fri, 06 Nov 2026 00:00:00 +0900

Of All The People In All The World is a piece that Stan’s Cafe, a theatre company from Birmingham, England, has been performing since 2003. One grain of rice stands for one person, and every statistic is weighed out into a pile at about 60 grains per gram, so the size ratio between labelled piles is exactly the ratio between the numbers of people. The paper under each pile carries only the name of the statistic, never a number.

What follows draws on the company’s official pages (the overview, the FAQ, and the page for each performance) and on two essays, from 2005 and 2018, by James Yarker, the originator and director. As third-party records, it uses articles from VOA and the Christian Science Monitor, and dataphys.org, a catalogue of physical visualizations. It differs from Hans Rosling’s demonstration, in which one IKEA box stood for one billion people, in the size of its unit and in how it treats numbers.

Gathering a number like 6.2 billion out of things you can hold

The idea came during a tour in 2002. In the 2005 essay, Yarker writes that every European city was full of people who would not have missed the company if it had not been there. He wanted to understand how many people there are on Earth. In the 2018 script of his TEDx Bonn talk, he looks back:

6.2 billion, what does that mean? Even written out in full the number was ungraspable.

Even written out, 6.2 billion was a number he could not grasp. He also writes that if you gather 6.2 billion things you can hold, you can pick one up as yourself and look back at everyone else.

The material was sand at first. According to the 2005 essay, a student at Manchester Metropolitan University told him that “sand grains vary in size, and there is even dust.” Yarker went around builders’ merchants and a spice shop on Ladypool Road in Birmingham, and settled on rice. He counted out 100 grams on a kitchen scale and did the arithmetic, which showed that the world’s population would need “100 odd tons.”

The official FAQ lists the conditions for choosing rice in order. The grains had to be “regular size grains,” so that numbers could be calculated from grains per gram. They had to be small, cheap, durable and unable to roll. Beyond that, rice is a staple food across much of the world, and up close it looks slightly like a person. The first condition is that numbers can be weighed out.

One conversion of 60 grains per gram, for every statistic

According to the official page, the piece began in May 2003 in the foyer of Warwick Arts Centre in Coventry. It was the “UK” version, in which Britain’s population of 60 million at the time was represented by 1,000 kg of rice.

According to the FAQ, the company put many grains on a scale and took the average per gram. Numbers up to 200 are counted by hand, and anything larger is weighed. The next sentence is the idea behind the piece:

Although this is inevitably an approximation so are most of the statistics themselves. Ultimately the show is about proportions, this beside that, using the same formula for all statistics keeps these proportions accurate.

It is an approximation, but so are the statistics themselves. The piece is about proportions, this pile beside that one, and using the same formula for every statistic keeps those proportions accurate.

Put together from the sources, the correspondence between data and rice looks like this.

Data	How it is shown	Specific value or example	Source
One person	One grain of rice	“Each grain of rice = one person”	Official overview
Number of people (201 or more)	Weight of rice, shown as pile size	Long-grain rice, 60 grains per gram	FAQ, Tokyo performance page
Number of people (up to 200)	Grains counted by hand	“count numbers up to 200 by hand”	FAQ
Name of the statistic	Label on a white sheet of paper under the pile	No number is written	2018 script
Relations between statistics	Piles placed side by side	One grain for Michael Jackson next to the 12 people who walked on the Moon	2018 script
Regions	Places within the venue	In 2008: the Americas, then Europe, Africa and Asia, in that order	2008 performance page
World population over time	Three piles and lighting in one room	AD 1, 1500 and 1900	2008 performance page
Population density	Number of grains per country	Canada has 3 people per km², so 3 grains	VOA
The visitor themselves	One grain handed to them	A child is given a grain at the entrance	CS Monitor

The scale changes from one edition to the next.

Edition and venue	Year	People represented	Amount of rice	Source
UK (Warwick Arts Centre)	2003	Britain, 60 million	1,000 kg	Official overview
World edition (Stuttgart)	2005	World, 6.2 billion	104 tonnes	Official overview
World edition (Birmingham)	2008	World, 6.7 billion	112 tonnes / 120 tonnes	Overview / performance page
If the same edition were staged in 2022	2022	n/a	133 tonnes	Official overview

Dividing 6.2 billion by 104 tonnes gives about 59.6 grains per gram (my calculation). That matches the conversion closely.

A factory floor with a pile of rice far taller than a person, a small pile on white paper in the foreground, and a child standing at the lower right

Source: Stan’s Cafe, Stan’s Cafe “Birmingham 2”

The conversion changes with the kind of rice. According to the page for the September 2010 Tokyo performance (Setagaya Public Theatre), Japanese rice runs to 52 grains per gram, fewer than the 60 of the usual long-grain rice. The same number of people weighs more, so the company hurried to buy more rice.

The sources disagree on some numbers.

Amount of rice in 2008: the overview page says 112 tonnes, while the performance page says “120 tons.” The performance page also mentions an “extra 8 tonnes,” and 104 + 8 = 112. A photo of the announcement sign on the performance page also reads “112 TONNES.” The 120 can be read as a typo.
Population of Britain: the overview and the 2018 script say 60 million, while the 2005 essay says “59 million.”
CS Monitor writes “50 million” for the pile for England, which is a different subject from the figure for Britain as a whole.
Scale in CS Monitor: it says the MASS MoCA performance used more than 15 tonnes, about 250 million grains, and that one 100-pound bag holds about 1 million grains. That works out to roughly 17 to 22 grains per gram (my calculation), which does not match 60 grains. I could not confirm the reason.

No numbers: let viewers read piles beside their labels

In the 2018 script, Yarker writes the rule of the piece like this:

We never show the numbers because the numbers get in the way; in standing in for the people numbers erase the people.

Numbers are never shown. The reason is that numbers get in the way, and by standing in for the people they erase the people.

What is put down instead is the name of the statistic and what sits next to it. The script gives examples such as placing the world’s refugees next to the world’s dollar billionaires, and a joke about placing one grain for Michael Jackson next to the 12 grains for the people who walked on the Moon.

A white sheet of paper with three lines of Japanese in the upper half, “私たち、落ちてみました。ナイアガラ滝から。樽や樽に類する物に乗って。” (We tried going over. Niagara Falls. In barrels and things like barrels.), and two lines of English in the lower half, “PEOPLE WHO HAVE GONE OVER NIAGARA FALLS IN BARRELS AND OTHER DEVICES”, with a dozen or so grains of rice beneath it

Photo: Graeme Rose. Source: Stan’s Cafe “Tokyo”

The labels in Tokyo were bilingual, in Japanese and English, with no numbers. On the largest piles, Yarker writes:

even in vast piles the granularity of the rice remains clear, every pile is demonstrably a collection of individuals

Even in the biggest piles the grains stay visible, so each pile is plainly a collection of individuals. Asked about the contradiction of one person appearing in two piles, the FAQ answers that the grains are “actors,” and that two grains may play the same person at the same time. This can be read as saying that the piles are independent for each statistic and are not meant to be added together.

Compared with Rosling’s IKEA boxes, the differences are as follows.

	Rosling’s boxes (TED@Cannes, 2010)	Of All The People In All The World
One unit	One box = 1 billion people	One grain = one person
Numbers	The speaker says them aloud	Not written on labels
How things are compared	Lined up along income levels, and moved by advancing the years	Labelled piles placed side by side
Time	A 9 minute 47 second talk	A run of several days to several weeks

You cannot see people inside a box, so Rosling can be read as making up for it with spoken numbers. With rice, one grain is one person, so individuals are visible even inside a pile. Bringing the unit down to one person can be read as the precondition for getting by without writing numbers. Yarker called Stuttgart in 2005 “a map as big as the territory.”

Weighing, placing and swapping during the run

The company calls the piece a “performance installation.” According to the official overview, the performers spend several days weighing out rice and laying down labelled piles. The piles become an “ever changing landscape.”

Three people in brown jackets weigh rice with a scale and a metal bowl at a long table covered with a green cloth

Source: Stan’s Cafe, Stan’s Cafe “Of All The People In All The World: UK”

This was the 2022 performance, back at Warwick Arts Centre, where the piece first played. According to the performance page, statistics were taken up and swapped out throughout the run.

According to the FAQ, statistics are collected from newspapers, academic journals, television, the web and other places. Official and unofficial figures are sometimes placed side by side. Where to put what is decided by the performers in discussion with their colleagues, and visitors’ suggestions are taken as well.

A wide factory floor with large and small piles of rice on white paper, and visitors walking and crouching to look

Source: Stan’s Cafe, Stan’s Cafe “Birmingham 2”

In Birmingham in 2008, the route through the factory rooms followed roughly the order of geography. The three piles for AD 1, 1500 and 1900 were lit with slow crossfades.

The official list of performances includes “Adelaide 2026,” so the piece is still being performed.

How third parties saw the piece

The first record is a reaction to the no-numbers rule. VOA spoke to visitors at the performance at the Skirball Cultural Center in Los Angeles. The date in the URL is 2007-12-21, and the publication date shown on the page is 2009-11-01. Rick Brous says:

We would love to get quantity on these things as we walk around and see big piles, we would like to know if that is hundreds of thousands or millions.

He would like to know whether a big pile is hundreds of thousands or millions. In the same article, Hillary Schmitt says the opposite:

I like the fact that the numbers aren’t listed on here and it’s just a visual comparison.

She likes that there are no numbers and it is only a visual comparison. It is a record of responses split over the author’s rule.

The second is about what is being compared. The Christian Science Monitor reported on the MASS MoCA performance in April 2007. The pile for the US population stood about 6 feet high. It also writes that the piles for residents of US gated communities and for prison inmates were nearly the same size.

The third is a researcher’s classification. dataphys.org, run by Pierre Dragicevic and Yvonne Jansen, classifies the piece as a “Passive physical visualization.”

Rice is weighted manually in small quantities and manually poured over piles. This labor-intensive process is part of the show.

The view is that the labour of weighing small quantities by hand and pouring them is part of the work. The date given is “Since around 2004,” but according to the official pages the premiere was in May 2003.

Takeaways

There are four judgements to take back to visualization work.

The first is that a smaller unit makes quantity and the individual visible at once. With one box = 1 billion people, you cannot see individuals. With one grain = one person, you can see the grains even in a big pile. It can be read as an option when you want to show both quantity and the individual.

The second is a way of thinking that puts a consistent conversion ahead of the precision of individual values. The FAQ admits the approximation, then says that using the same formula for everything keeps the proportions. For a chart meant to make people compare, it can be read as saying that applying the same scale to every series is a precondition.

The third is the cost of removing the numbers. As with the VOA visitor, people who want to know the order of magnitude are left dissatisfied. Whether to favour ease of comparison or readable information is decided by the purpose of the chart.

The fourth is to re-weigh the conversion constant for each material and state it. Just as Japanese rice changed it to 52 grains, a different material means a different constant. The reason for the discrepancy has not been confirmed, but there is also the CS Monitor figure that does not match 60 grains, and it can be read as safer to state the conversion explicitly.

References

Primary sources

Stan’s Cafe, “Of All The People In All The World” (overview, FAQ, list of performances). Official page
Stan’s Cafe, “Of All The People In All The World: UK” (Warwick Arts Centre, 4-9 October 2022). Performance page
Stan’s Cafe, “Stuttgart” (Wagenhalle, 20 June to 10 July 2005). Performance page
Stan’s Cafe, “Birmingham 2” (A.E. Harris Factory, 13 September to 5 October 2008). Performance page
Stan’s Cafe, “Tokyo” (Setagaya Public Theatre, 14-22 September 2010, photo: Graeme Rose). Performance page
James Yarker, “Making Of All The People In All The World,” 2005. Essay
James Yarker, “Human Empathy In A Granular World,” script of a talk at TEDx Bonn, 2018-11-17. Essay

Secondary sources

Mike O’Sullivan, “Performance Artists Make Statistics Real Through Grains of Rice” (Skirball Cultural Center, Los Angeles), VOA, date in the URL 2007-12-21, publication date shown on the page 2009-11-01. Article
“Sizing up the world with rice” (North Adams, MASS MoCA), The Christian Science Monitor, 2007-04-03. Article
“2004 – Of All the People in All the World: Stats with Rice,” List of Physical Visualizations (dataphys.org, run by Pierre Dragicevic and Yvonne Jansen). Item list (rice tag) / Item page

Transcribing a Storm's Numbers into a Score, Then Weaving a Basket from It: Nathalie Miebach's Weather Data Sculptures

Fri, 06 Nov 2026 00:00:00 +0900

Nathalie Miebach makes sculptures that map weather data such as temperature, pressure, and wind speed onto the colors and positions of basket-weaving materials, and has musicians perform musical scores that transcribe the same data. She assigns the time of day and a temperature range to the basket’s vertical members, then weaves tide, water temperature, air temperature, and the moon phase into that grid. For her sculpture about Hurricane Noel in 2007, CSMonitor (2010) writes that temperature is red, pressure is green, and wind speed is blue, “and so on.”

What follows draws on Miebach’s TED talk “Art made of storms” (TEDGlobal 2011), the work descriptions and performance records on her own website, and a 2010 interview and newspaper article. The order of the process follows her talk: first the data is transcribed into a score, then that score serves as the blueprint for assembling the sculpture.

In 2000, Learning Basket Weaving and Astronomy at the Same Time

It began in 2000. Miebach took a local basket-weaving class and an astronomy course at Harvard University at the same time. Instead of a written final paper for the astronomy course, she handed in a basket representing the Hertzsprung-Russell diagram, the Christian Science Monitor (2010) reports.

She moved to weather in 2006, during a residency at the Fine Arts Work Center in Cape Cod. Zach Smith, a climate educator at the Wright Center for Science Education, proposed a climate change project to her. She says she learned how to collect data while trying out the center’s instruments on the beach (ArtSake, 2010).

She describes how she collects data in her TED talk (recorded July 11, 2011, about 4 minutes). With instruments of the kind you can buy at a hardware store, she gathers information about a specific place. She checks it against satellite images and data from weather stations and offshore buoys, and gathers the numbers on a clipboard. From there, she starts with only two or three variables. The video of the talk can be watched on the TED page.

Assigning Time and Temperature to the Basket’s Grid

Miebach calls the basket a “medium of translation.” A basket is made of vertical and horizontal members. In her talk she explains that if you assign values to them, the change of data points over time takes shape as it is.

She assigns the time of day to the vertical members, so that one full turn is 24 hours. She assigns a temperature range to the same members. Into that grid she weaves high-tide readings, water temperature, air temperature, and the moon phase. On why she uses natural reed as her material, she says:

“I use natural reed, because natural reed has a lot of tension in it that I cannot fully control. That means that it is the numbers that control the form, not me.”

The reed has a tension she cannot fully control, so it is the numbers, not she herself, that decide the form: that is her explanation.

The correspondence between colors and variables is not described in the talk. The talk says only that “each colored bead or thread is one weather element.” There are two sources that spell out a specific correspondence, and both concern the work “Hurricane Noel,” about Hurricane Noel in November 2007.

The Christian Science Monitor (2010-04-09, Kendra Nordin) writes, of the sculpture “Hurricane Noel,” that in the vertical direction temperature is red, pressure is green, and wind speed is blue, “and so on.”
The two-dimensional score “Hurricane Noel” (2009) on her website gives, in a handwritten legend, the temperature at the Hyannis station as red, pressure as green, average wind speed as light blue, and gusts as dark blue.

In the score’s legend, red indicates only the temperature at Hyannis. I could not confirm the color schemes of her other works.

Data	What it is mapped to	Source
Time of day (24 hours)	The basket’s vertical members; one turn is one day	TED talk
Temperature range	The same vertical members	TED talk
High tide, water temperature, air temperature, moon phase	Members woven into the grid	TED talk
Temperature, pressure, wind speed	Color of the reed in the vertical direction (red, green, blue)	CSMonitor 2010
Rises and falls of readings	Distortions of the colored reed, that is, the shape of the sculpture	CSMonitor 2010
Time periods of the storm’s cycle	Spokes radiating from the central cone	CSMonitor 2010
Type of weather element	Range of pitch in the score (one octave per element)	CSMonitor 2010
Temperature values	Pitch in the score (one pitch per temperature)	CSMonitor 2010

The data came from stations in Hyannis and in Natashquan, Quebec, and from a buoy in the Gulf of Maine (CSMonitor).

Transcribing into a Score, Then Assembling from It as a Blueprint

This work is sometimes introduced as “a sculpture that also becomes a score,” and the TED description likewise says the sculpture “becomes a score for string quartet.” In her own talk the order is the reverse: she says she first transcribes the weather data into a score.

“Meanwhile, I use these scores as blueprints to translate into sculptural forms like this, … so it can actually be read as a musical score.”

She uses that score as a blueprint and turns it into a sculpture, she explains. The sculpture contains the score’s grid inside it, so it can also be read as a score.

Other sources give the same order. ArtSake (2010-03-10) describes the project as translating data “first into a musical score, then into a sculpture.” In her artist statement (published by the Schick Art Gallery at Skidmore College), she says she transcribes the numbers onto score paper in pictorial form. The musicians interpret it as a piece of music, and Miebach interprets it as a sculpture. The score’s title also reads “Score for Musicians and Sculptor.”

A handwritten score on graph paper, with time on the horizontal axis and scales for wind, temperature, and pressure on the vertical axis, and red, green, and blue square notes lined up in bands

Source: Nathalie Miebach, “Hurricane Noel” (2009), from nathaliemiebach.com, Musical Scores 2D

The horizontal axis of this score is marked with times from November 2 to 5, 2007. The legend separates “Fixed Notes” from “Flexible Regions.” The distinction is that the notes representing observed values are not to be changed, while regions such as cloud cover may be interpreted by the performers.

A three-dimensional work in which red, blue, and light-blue rods and beads and number tags such as “65°F” are attached around a woven reed basket

Source: Nathalie Miebach, “Hurricane Noel III” (2015), from nathaliemiebach.com, Musical Scores 3D

Her website calls such sculptures “3D musical score.”

Work	Year	Description on her website
Hurricane Noel	2010	Passage through the Gulf of Maine, November 6-8, 2007
Hurricane Noel III	2015	Passage through New England, November 2-5, 2007. Two stations and one ocean buoy
Sun Gnomon for Stormy Night	2009	A three-dimensional score based on the two-dimensional score “Stormy Weather, Internal Storm”

A two-page score in which red and green dots, blue wavy lines, wind-direction arrows, and handwritten words are drawn over staff lines and vertical time lines

Source: Nathalie Miebach, “Stormy Weather, Internal Storm” (2012), from nathaliemiebach.com, Musical Scores 2D

A three-dimensional work in which clusters of orange beads, a green net, and yellow ladder-like legs are attached around a cone and a disc woven from reed

Source: Nathalie Miebach, “Sun Gnomon for Stormy Night – Nov 13-19, 2008” (2009), from nathaliemiebach.com, Musical Scores 3D

The years and periods differ among the sources.

Point	What each source says
Period of Hurricane Noel	The two-dimensional score and “III” give November 2-5. The 2010 sculpture, and “Musical Buoy in Search Towards a New Shore” (a work with a different title) in The Marginalian (2011), give November 6-8
Year of Hurricane Noel	The Chipstone Foundation says “Noel, 2006.” Her website gives 2009 for the score and 2010 and 2015 for the sculptures

Since the storm took place in November 2007, 2006 appears to be an error.

How Much to Leave to the Performers

Miebach tells ArtSake that she started making scores because the language of sculpture no longer expressed how she read her data. In her talk she also says that a score lets her translate in detail without losing information.

She also explains in ArtSake what she fixes and what she leaves to the performers.

“When I give musicians the score, I tell them what it’s about, what portions of the score are flexible and those that aren’t. Then I pretty much withdraw …”

Item	Who decides	Source
The notes themselves	Determined by the numbers the artist collected	ArtSake 2010
Tempo and rhythm	The artist uses them to add nuance to the translation	ArtSake 2010
Rhythm, tempo, number of instruments	Largely left to the performers	ArtSake 2010
The form of the piece (in the case of Axis Ensemble)	The performers made it a 15-minute piece. The only condition was “being faithful to the numbers”	CSMonitor 2010

On March 14, 2010, Axis Ensemble performed “Hurricane Noel” at the Lilypad in Cambridge. The lineup was piano, violin, accordion, clarinet, cello, and double bass (CSMonitor). Miebach tells ArtSake that after watching the rehearsal she spent two hours in her studio staring at her own sculpture.

Her website gives two purposes for the collaboration: to convey a layer of emotion that the sculpture has difficulty reaching, and to learn about the patterns and stories in the data that the musicians find.

This effort has continued since 2009 as “The Weather Score Project,” which, according to her website, comprises 20 commissioned works and more than 15 concerts.

The instrumentation also differs by source. The TED description says “string quartet,” but what appears in the talk is the trio “1913 Trio.” Who performed the Hurricane Noel music at the start of the talk is not stated in the talk or on the TED page.

How Third Parties Saw the Work

Here are three views, each with a different point. The first is the work profile “Sculpting The Floods” in Issues in Science and Technology (Winter 2026).

“Rather than visualizing data for clarity or communication, Miebach’s goal is to provide a different way of experiencing data—to make the complexities of natural systems physically and emotionally perceptible.”

The view is that the work is not a visualization for clarity or communication, but a way to let people sense the complexity of nature with body and emotion.

The artist herself stresses accuracy, in words quoted by Colossal (2016).

“Nothing is put on there for purely aesthetic reasons.”

That is, nothing is attached merely for looks. In the same article she also says she wants people to be able to read the weather from the sculpture. The third party’s positioning and the artist’s emphasis can be read as differing.

The second is the Christian Science Monitor (2010) interview article.

“But behind the joyful chaos is a precise order, one of surprising complexity and magnitude.”

Behind the toy-like appearance lies a precise order, the review says. In the same article Miebach herself says the playful elements are there to draw viewers in, and that on closer look they realize everything is numbers.

The third is a catalog from the visualization research side. The “List of Physical Visualizations,” edited by Pierre Dragicevic and Yvonne Jansen on dataphys.org, lists Miebach’s basket-weaving sculptures as an entry for 2006 and classifies them as “Passive physical visualization.” Jansen et al.’s evaluation experiments are covered in another article.

Takeaways

There are four judgments that can be carried over to visualization work.

The first is to narrow down the variables before putting them onto a material. From a bundle of numbers, Miebach first selects only two or three variables. When the amount a material can carry is limited, the act of selecting can itself be read as design.

The second is to make an intermediate notation the blueprint. Miebach first transcribed the data into a score and made that the source of both the sculpture and the performance. Deriving a form to look at and a form to listen to separately from one notation can be read as keeping the two pointed at the same data.

The third is to separate, within the notation, the parts that are fixed from the parts that are left to others. The Hurricane Noel score separates “Flexible Regions” in its legend and instructs the performers not to change the notes for observed values (Fixed Notes).

The fourth is to write the legend into the work itself. Hurricane Noel III carries number tags such as “65°F,” and the score has a color legend and the names of the stations. In her talk, Miebach said that in an art museum it becomes a sculpture, in a science museum a three-dimensional visualization, and in a concert hall a score. That it can be read wherever it is placed can be read as owing to the legend being inside the work.

As examples of passing the same data to sound or form, it can be read alongside Chemo Singing Bowl and The Tempestry Project.

References

Primary sources

Nathalie Miebach, “Art made of storms,” TEDGlobal 2011 (recorded 2011-07-11, published 2011-10-21). Talk page / Transcript
Nathalie Miebach, The Weather Score Project (performance records). nathaliemiebach.com
Nathalie Miebach, Musical Scores 2D. nathaliemiebach.com
Nathalie Miebach, Musical Scores 3D. nathaliemiebach.com
Nathalie Miebach, Storms, Gales and Blizzards. nathaliemiebach.com
Nathalie Miebach, The Sandy Rides. nathaliemiebach.com
Nathalie Miebach, Weather Sound Drawings. nathaliemiebach.com
Nathalie Miebach, Artist Statement, Schick Art Gallery, Skidmore College. Statement
“Nathalie Miebach: weaving science, sculpture, and music,” ArtSake (Mass Cultural Council), 2010-03-10. Interview
Nathalie Miebach, Bruce D. Campbell, Francesca Samsel, “Nathalie Miebach: Sculpted Data Infused With Craftsmanship,” IEEE Computer Graphics and Applications 42(1): 7–16, 2022 (bibliographic record only checked). doi:10.1109/MCG.2021.3132092 / dblp

Secondary sources

Kendra Nordin, “Climate change as art,” Christian Science Monitor, 2010-04-09. Article
“Sculpting The Floods,” Issues in Science and Technology, Winter 2026. PDF
Pierre Dragicevic, Yvonne Jansen (eds.), “Nathalie Miebach’s Woven Sculptures,” List of Physical Visualizations. dataphys.org
Maria Popova, The Marginalian, 2011-07-12. Article
Colossal, 2016-03. Article
Chipstone Foundation, work description for the exhibition “The New Materiality.” Exhibition page

Showing Web Activity with Four Bars That Rise and Fall on a Table: Wable

Tue, 03 Nov 2026 00:00:00 +0900

Wable was a prototype table that showed activity on web services through the height of bars that raise and lower automatically. There are four bars, and they move up and down according to activity on services such as Plazes, Flickr and Last.fm. The activity is pulled in through RSS feeds. According to infosthetics, the display can be configured in a web application. It was made by Physical Interaction Lab, built in 2007 and made public in February 2008.

The project’s official page no longer loads. What follows relies on the text of the official site as quoted by the data physicalization catalog dataphys.org and others, on the press coverage of February 2008 (infosthetics, Core77, Engadget, GeekAlerts), and on the research paper “Physecology” (2022). I have found no source that describes the mechanism, the dimensions, or the scale of the bars. This article does not go into those either.

February 2008: reported as a bar chart in a coffee table

The first to report on it was the visualization blog information aesthetics (infosthetics). Its article of February 1, 2008, filed under “aesthetic / physical”, introduced it in one sentence:

“a physical bar graph integrated in a coffee table, which visualizes the changes of one’s web identity over time, & creates a physical link between your virtual & real identity.”

Summary: a physical bar graph built into a coffee table that shows how one’s web identity changes over time and physically links the virtual and real identities.

On the same day, the design magazine Core77 covered it as a coffee-table piece, “via infosthetics”. On February 2, Engadget ran a post “Via Core77”, and GeekAlerts also published an article.

Photo of four prisms in green, purple, blue and orange standing at different heights on a black tabletop

Source: Physical Interaction Lab / GeekAlerts, “Wable – Cool Interactive Table”

In the photo, the green bar is the tallest and the blue one the shortest.

The year it was made is 2007 according to dataphys.org. The same photo is also reproduced on EduTech Wiki (University of Geneva) as “Wable, version 1 prototype”. The EXIF data of that image file records a digitization date of November 7, 2007 and a file modification date of December 30, 2007. It can be read as consistent with the press dates if the table was photographed and built between autumn and the end of 2007 and made public in early February 2008.

The makers’ words: change over time, and a physical link

The makers’ own text was on the official site of Physical Interaction Lab. Four sources (dataphys.org, Core77, GeekAlerts and EduTech Wiki) quote the same passage.

“The personal feeds from webapplications like Plazes, Flickr, and Last.fm tell much about the activity of an individual on the internet. In this project we aim to explore how you can visualize the changes of your web identity over time and create a physical link between your virtual and real identity. The interface consists of both a physical table and a web application. This direct feedback from your web identity is customizable and can be connected to any RSS feed.”

Summary: Personal feeds from web applications such as Plazes, Flickr and Last.fm say a lot about what a person does online. The project explores how to show the changes of a web identity over time and how to link the virtual and real identities physically. The interface has two parts, a physical table and a web application. The display is customizable and can be connected to any RSS feed.

Plazes was a location-sharing service of the time, Flickr a photo service, and Last.fm a service for music listening history. Two aims are stated. One is to show how activity on the web changes over time. The other is to link the virtual and real identities physically.

The official site also carried a note:

“Update: What you see is the first prototype version of Wable. We are currently building version 2 so keep an eye on our blog or sign up below to find out when it’s released and ready for sale.”

Summary: What you see is the first prototype of Wable. Version 2 is being built, and the blog or the sign-up form will announce when it is released and ready for sale.

What was shown publicly was the first prototype. Version 2 was under construction, with a sale in view. I have found no source saying whether version 2 ever appeared.

Little is known about the makers. dataphys.org gives the source as “Physical Interaction Lab (2007). Wable.” and quotes the official text as “From the company’s website”, so it treats the Lab as a company. The video “Wable” on YouTube was uploaded by David Kjelkerud. The paper Physecology also lists the work in its references as “David Kjelkerud. 2007. Wable.” What role he had in the Lab, who else was involved, and where the Lab was located are not known.

The official page (physicalinteractionlab.com/projects/wable) no longer loads. In a 2018 update, GeekAlerts added a note that it had removed the link because the page was no longer public.

What was mapped to which bar

The correspondence between data and appearance is summarized in a table.

Data	Appearance / material	What is known	What is not known
A person’s web services	Number and arrangement of bars. I read one bar as corresponding to one service	There are four bars (Engadget). The services named are Plazes, Flickr and Last.fm (official text)	Which service each of the four bars stands for
Activity on each service	Height of the bar. It rises and falls automatically with the data	“raise and lower automatically based on the activity” (Engadget). Physecology gives it as an example of countable data (online activity)	The scale of the heights, units, maximum, update interval
Color of the bars	Material color, which does not change	Physecology gives it as an example of “fixed optical properties, such as material color”. In the photo: green, purple, blue and orange	Whether color was used to tell the series apart is not known. I read it as distinguishing the series. The mapping between colors and services is also unknown
Time	A slider. Past activity is seen through the bars	“explore personal online activity over time through a slider button” (Physecology)	The position of the slider, how far back it could go
Data source and settings	A web application (screen)	Can be connected to any RSS feed (official text). infosthetics writes “a customizable web application”	What the settings screen contained, how feeds are assigned to bars

dataphys.org classifies this work as an “Active physical visualization”. Its tags are extruded 2d, online community data, personal information management, pragmatic and self-actuated. Extruded 2d means a two-dimensional chart extruded in the height direction, and self-actuated means that the device moves by itself.

A small component is visible at the front right of the tabletop in the photo. No source says whether it is the slider.

Video thumbnail showing four prisms in purple, green, orange and blue standing on a table in front of a window

Source: David Kjelkerud / YouTube, “Wable” (video thumbnail)

In the video thumbnail, four prisms stand on a table in front of a window. The shooting angle and the combination of heights differ from the photo above, and the blue bar looks clearly taller than in the photo above. The difference in heights between the two images can be read as consistent with the reports that the bars move up and down. The video is available on the YouTube page for “Wable”.

From here on, these are my own readings. In that quantity is shown by length (height) and series by position and color, the encoding can be read as a textbook bar chart. Making it physical adds two things. The bars move by themselves, so change over time is seen as motion. And the bars sit inside a piece of furniture, so they catch the eye without opening a screen, as I read it.

How third parties saw this table

Contemporary press, blog readers and later research each looked at a different aspect.

Darren Murph of Engadget kept his distance in his article of February 2, 2008. The headline, “Wable graphically represents web activity, ain’t much of a table”, means that it is not much of a table.

“This unique “table,” as it were, sports four bars that raise and lower automatically based on the activity from web applications such as Plazes, Flickr and Last.fm. Beyond that, it doesn’t seem to add too much more value to one’s life, but if it’s the simple things that keep you smiling, check out a remarkably unremarkable video of it after the jump.”

Summary: This “table” has four bars that rise and fall automatically with activity from Plazes, Flickr and Last.fm. It does not seem to add much value to one’s life, but if simple things make you smile, watch the remarkably unremarkable video.

In the opening of the article, he writes that he is not sure anyone really wants what the makers advertise, a physical link between the virtual and real identities.

The infosthetics article has three reader comments. The first says “Wow. Beautiful.” and points out that the music in the video is by Kraftwerk. The next comment is negative:

“Ugh … ridiculous … too BIG … I mean, where are you going to put it?”

Summary: Too big. Where would you even put it?

The third comment is a rebuttal to this one. Readers’ opinions are not expert evaluations, but they can be read as showing that making the chart physical raised questions a screen visualization does not face: its size and where to put it.

On the research side, the paper “Physecology” by Kim Sauvé, Miriam Sturdee and Steven Houben (ACM TOCHI, 2022) discusses Wable. The paper analyzes 60 physicalization examples, and Wable is third in its list. For data type, it gives “Wable [56] (online activity)” as an example of countable data. For data subject, it is an example of personal data. It cites color as an example of the fixed color of a material rather than of a dynamic display. For interaction, it gives the following as an example of exploring data:

“Example physicalizations are Wable [56] in which you can explore personal online activity over time through a slider button”

Summary: An example physicalization is Wable, in which personal online activity can be explored over time with a slider button.

What the press saw as a “moving table”, this paper organizes along design dimensions: the relations among data type, interaction, and input and output.

There is one more early record. The paper “Embodiment in data sculpture” by Jack Zhao and Andrew Vande Moere (DIMEA 2008) cites the official Wable page in its references, with an access date of May 2008. I have not been able to confirm how the body of the paper treats it, but it means that within a few months of its release, Wable was in the references of a paper on data sculpture.

Takeaways for practice

I read three things from this work that can be carried back into visualization work.

Separate the data source from the display. Wable’s entry point is RSS feeds, and the official text says it can be connected to any feed. infosthetics writes that the display can be configured in a web application (“a customizable web application”). I read it as a design in which the thing displayed stays four bars while its content can be swapped.
Treat how readable the mapping is as separate from how new the work is. A mapping that shows quantity by height is understood at a glance. Yet Engadget called the video “remarkably unremarkable”. If you make a bar chart physical, I read it that the value has to lie in the motion and the placement, not in a novel encoding.
Treat placement as part of the design. For a screen visualization, size is settled inside the screen. A bar chart on a table takes up space in a room. The infosthetics reader’s “where are you going to put it?” can be read as showing that, in physicalization, where to put it is itself in question.

References

Primary sources

Physical Interaction Lab, “Wable” official page (link dead): http://www.physicalinteractionlab.com/projects/wable / Wayback Machine snapshot (2019-06-09): http://web.archive.org/web/20190609053241/http://www.physicalinteractionlab.com/projects/wable
Pierre Dragicevic, Yvonne Jansen, “2007 – Wable: Web Behavior Shown with a Dynamic Bar Chart”, List of Physical Visualizations (quotes the official site’s text): https://dataphys.org/list/wable-visualize-web-behavior-as-a-physical-bar-chart/
David Kjelkerud, “Wable”, YouTube: https://www.youtube.com/watch?v=e6G5YlICVRg (uploader confirmed at: https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=e6G5YlICVRg&format=json )

Secondary sources

Darren Murph, “Wable graphically represents web activity, ain’t much of a table”, Engadget, 2008-02-02: https://www.engadget.com/2008-02-02-wable-graphically-represents-web-activity-aint-much-of-a-table.html
infosthetics, “physical bar graph table”, information aesthetics, 2008-02-01 (includes the statement “the Wable interface consists of both a physical table & a customizable web application” and three reader comments): https://infosthetics.com/archives/2008/02/physical_bar_graph_table_visualization.html
squirrelbait, “Wable physical web activity visualization table”, Core77, 2008-02-01: https://www.core77.com/posts/8843/Wable-physical-web-activity-visualization-table
Robert Birming, “Wable – Cool Interactive Table”, GeekAlerts, 2008-02-02 (updated 2018): https://www.geekalerts.com/wable-cool-interactive-table/
Kim Sauvé, Miriam Sturdee, Steven Houben, “Physecology: A Conceptual Framework to describe Data Physicalizations in their Real-World Context”, ACM TOCHI 29(3), 2022: https://eprints.lancs.ac.uk/id/eprint/161277/1/TOCHI_Physecology.pdf (https://dl.acm.org/doi/10.1145/3505590 )
Jack Zhao, Andrew Vande Moere, “Embodiment in data sculpture: a model of the physical visualization of information”, DIMEA 2008: https://dl.acm.org/doi/10.1145/1413634.1413696 (the citation of Wable was confirmed in Crossref’s reference data: https://api.crossref.org/works/10.1145/1413634.1413696 )
EduTech Wiki, “Physical visualization”: https://edutechwiki.unige.ch/en/Physical_visualization / image file page: https://edutechwiki.unige.ch/en/File:Wable.jpg

100 Residents Become a Pie Chart on Stage: Rimini Protokoll's 100% City

Thu, 05 Nov 2026 00:00:00 +0900

Rimini Protokoll’s 100% City is a theatre piece in which each resident stands for 1% of a city’s population. One hundred residents, gathered to match the city’s statistics, stand on stage. They move around the stage according to their answers to questions, and in front of the audience they form a pie chart out of human bodies.

The creators are three people: Helgard Haug, Stefan Kaegi and Daniel Wetzel. The piece premiered as 100% Berlin at HAU 1 in Berlin on 1 February 2008. It was then remade for each city and staged in many places, and in Japan 100% Tokyo (100％トーキョー) was performed in 2013. What follows draws on the official pages, interviews with the creators, the F/T day-of program, reviews and papers.

2008: Starting as a Celebration of a Theatre’s 100th Year

100% Berlin was created to celebrate the 100th anniversary of the Hebbel Theater. Its subtitle is “A Statistical Chain Reaction.” It ran at HAU 1 on 1–3 February and 16–18 May 2008.

Haug recalled in a 2018 interview that touring had not been planned.

In the beginning, the show was not intended to be a touring project. … finally, we realised the potential of “working locally but comparing globally”

In other words: at first it was not meant to tour, and in the end they realised the potential of working locally and comparing globally. The number of cities where the piece has been staged differs from source to source.

When	Number given	Source
2013	“more than 10 places in Europe, Australia and North America” (Japanese: 「欧・豪・北米で10カ所以上」)	Katsura Mana, F/T13 day-of program (F/T13 tōjitsu panfuretto)
July 2013	“nearly 20 cities”	Peter Crawley, The Irish Times
2015	“23 cities”	Contemporary Theatre Review
February 2018	“34 different cities”	Haug, LitNet interview
2020	“over 40 cities”	HAU, “100% Berlin reloaded”
2024	“41 Städten” (41 cities)	Anja Quickert, paper
Accessed November 2026	“more than 35 cities”	Rimini Protokoll official, “100% City”
November 2026, counted by the author	40 cities (Berlin in 2008 to Oulu in 2026)	Performance list on the official “100% City” page

The last row is the 42 entries on the official list minus Berlin reloaded (the 2020 revival) and Uni Marburg (a university version planned for 2027).

According to the tour materials, the running time is 110 minutes, and the piece needs a motorised revolving stage at least 10 m in diameter. A local team spends four to five months on casting, and the piece is normally performed three times.

Many performers stand on a round green stage, and a circular screen above shows the arrangement of the performers filmed from directly overhead

Source: © Nurith Wagner-Strauss / Rimini Protokoll, “100% Vienna”

The creators’ own writing also appears in the book Making with Data (edited by Huron, Nagel, Oehlberg and Willett; A K Peters/CRC Press, 2022). According to the book website’s structure page, the second part, “Participation,” includes a chapter “100% [City]” as the fourth piece after Cairn, SeeBoat and Let’s Play with Data (chapter 2.4, pp. 147–162, doi:10.1201/9781003264903-9). I have not been able to read the chapter text, so its content is not covered here.

One Person Is 1%, and a 24-Hour Chain Gathers the 100

One person on stage represents 1% of the city’s population. How many people that is varies from city to city: about 34,000 in Berlin, about 16,935 in Vienna and about 41,000 in Melbourne.

There are rules for how the 100 are gathered. In the first Berlin production, the directors chose only the first person. Within 24 hours, that person introduced the next person from among their acquaintances, and so on until there were 100. In Stellenbosch, because of large gaps between regions, three chains were run in parallel. In some cities, the head of a statistics bureau or a statistics researcher was chosen as the first person.

The creators explain that they build the frame from five census criteria: age, gender, district, place of birth or origin, and household size or marital status. In the script for the first production, marital status was 46 single, 39 married, 6 widowed and 9 divorced.

Beyond those criteria, there are also hidden conditions. In Melbourne, eight conditions were added, including education, income, Indigenous status and sexual orientation. In a 2015 conversation, the creators also said they made sure that actors and theatre enthusiasts did not exceed the census proportion (1–2%).

The more the frame fills up, the harder the search becomes. The official Berlin page says that the closer they got to the end, the harder it was to find the next “link in the chain.” In London, at the 37th person, they realised that there was not a single person of Pakistani heritage (The Guardian, 2012). When a chain breaks, they approach local organisations, schools and care homes for older people, and place advertisements in newspapers. The creators say they talk about this struggle on stage too.

On stage we address the struggle to find people because it tells us something about the city, too.

That is, the struggle to find people also tells us something about the city.

A woman performer speaks at a microphone, with other performers standing in a row behind her

Source: © Maja Nydal Eriksen / Rimini Protokoll, “100% Copenhagen”

At the start of the performance, the performers step up to the microphone one by one, in the order in which they were introduced, and introduce themselves. The order in which they line up on stage can be read as itself being the route along which the sample was gathered.

Walking Beneath a Placard to Become a Pie Chart

The core of the performance is a yes-or-no question. Richard C. Windeyer recorded 100% Montréal (2017) like this.

In full view of the audience, each percentile stands beneath one of two placards—“Me/Not Me” and “Moi/Pas moi”— according to their answer for each question.

In short, for each question the performers stand beneath one of the two “Me/Not Me” placards in full view of the audience. For example, to the question “Who doesn’t pay taxes?”, four people stood on the “Me” side. The critic Katsura Mana wrote about 100% London (2012) that the circle on the floor was divided into left and right along a diameter, and the performers walked to the right for Yes and to the left for No.

The number of people who gather becomes the proportion directly. The official page describes this scene as follows.

A series of statistical groupings enacted onstage, like a giant pie chart brought to life.

In a 2015 conversation, the creators say that the answers to yes-or-no questions are turned into a “pie-chart” that becomes the starting point for political discussion.

On a green revolving floor, children run toward the front across a stretched yellow tape. Adult performers stand in rows behind them

Source: © Pigi Psimenou / Rimini Protokoll, “100% Melbourne”

Footage shot from directly overhead is also layered onto the stage. According to Windeyer’s record of 100% Montréal, a green platform on the floor was filmed by a camera directly above, and image processing replaced the green parts with radial charts, scatter plots and district maps, which were shown on a circular screen.

Like in a Petri dish, you can look at how the individuals are moving and forming majorities and minorities.

Haug compares this footage to a Petri dish, saying you can see how individuals move and form majorities and minorities.

A circular screen shows footage, shot from directly above, of performers scattered across a green floor, and the performers stand on the round stage beneath it

Source: © Nurith Wagner-Strauss / Rimini Protokoll, “100% Vienna”

For questions that are hard to answer in front of others, a different method is used. The lights are dimmed and each person raises a flashlight. The number of points of light is the number of people it applies to. When the same question, “Who doesn’t pay taxes?”, was asked in the dark in 100% Montréal, more than ten lights came on. On the brightly lit stage, four people had stood on “Me.” Windeyer calls this image a “luminous scatterplot of anonymized confession.” In London, people earning £3,000 or more a month lit up their mobile phone screens.

The correspondences are summarised in a table.

Data	How it appears on stage	Source
Population of the city	The 100 people on stage; one person is 1% of the population	Official pages
Population composition (five criteria)	Number of people cast; the number in each category is its proportion	Official pages, CTR 2015
How the sample was gathered	Order of appearance; people introduce themselves in the order they were introduced	Haug 2018, Windeyer 2018
Answers to yes-or-no questions	Standing position: the side of the “Me / Not Me” placards, or right and left of the circle	Windeyer 2018, Katsura 2013
Proportion of answers	The number of people who gather; seen from directly above, a sector-shaped cluster	Official Melbourne page, CTR 2015
Reading as a chart	Footage that replaces the green floor with a pie chart, scatter plot or district map	Windeyer 2018
Answers that are hard to say in public	The number of flashlight or phone lights raised in the dark	Windeyer 2018, The Guardian 2012

Laid out as a table, the correspondences fall into three stages. At the casting stage, proportions are turned into numbers of people; during the performance, answers are turned into standing positions and clusters of people; and the footage lets the audience read those clusters as a chart. Because of that last stage, the same body can be seen as both a “person” and a “data point,” or so it can be read.

2013: 100% Tokyo

It was performed as part of Festival/Tokyo 13, from 29 November to 1 December 2013 at the Tokyo Metropolitan Theatre Playhouse. It was directed by Daniel Wetzel and performed in Japanese.

The F/T introduction describes the work as a “live opinion-survey theatre for 100 citizens gathered on the basis of Tokyo’s population statistics” (Japanese: 「東京の人口統計に基づいて集められた100人の市民への、ライブ意識調査演劇」). The 100 people were said to “embody the ‘standard’ of the 9,055,257 people living in Tokyo’s 23 wards” (Japanese: 「東京２３区に住む9055257人の『標準』を体現する」). One person corresponds to about 90,553 people. The official English page says:

this cross-section of the society could tell the truth of modern Tokyo life in a way graphs or pie-charts never could.

That is, this cross-section of society could tell the truth of modern Tokyo life in a way that graphs or pie charts never could.

The statistical adviser was Nakamura Kazuyuki. In his interview in the program, Nakamura said he chose items by whether they “could be looked up.” He left out smartphone penetration because “penetration” is hard to define, and left out educational background because there are many “no answer” responses. He also said that, from the standpoint of statistics, 100 people is not enough of a sample.

According to Katsura Mana’s contribution, 3% of the residents of the 23 wards are foreign nationals, so 3 of the 100 people were foreigners. The breakdown was 1% each for “Koreans, Chinese and others.” Katsura calls the Tokyo production “the Asian premiere of the series on the theme of cities” (Japanese: 「都市をテーマに置くシリーズのアジア初演」). The only Japanese city on the official list is Tokyo.

How Third Parties Viewed the Work

Eddie Paterson discussed 100% Melbourne in his 2016 paper “Bodies of Data.” He reads the Petri dish footage as an expression of scale that moves between census figures, bodies on stage and a cell-like swarm. At the same time, he writes that the work traces the grid of the census.

by staging official population data onstage, Rimini’s performance visualises the grid-logic of the State.

That is, by staging official population data, the work makes visible the grid-logic of the State. Paterson notes that although there is a place near the venue where people sleeping rough gather, there were no rough sleepers on stage.

Windeyer questioned the veracity of the data. In 100% Montréal, more than ten people raised their hands to the question “Who lied tonight?” The question is how the veracity of statistics can be guaranteed if the answers themselves cannot be trusted. He also records that Indigenous people make up less than 1% of the population, so they were not given a dedicated slot and were not visible on stage.

In his review in The Irish Times, Peter Crawley enjoyed the staging technique but criticised the sameness of the form. There is no message and no commentary, only questions. The form has been honed over five years and nearly 20 cities, and is hardly different in Dresden or in Tokyo, he says. He also writes that in Cork they were unable to get participants from wealthy districts or from the large Polish community.

Douglas Eacho, in the abstract of his 2018 paper, reads the work as nostalgia for nineteenth-century statistical reason and contrasts it with the algorithmic counting of big data. I have not read the main text.

Takeaways

The first thing to take away is to attach one face to one mark. A sector on the stage is a cluster of people, and each of them introduced themselves at the start. The creators say that the audience follows a chain of a few people’s answers rather than the whole city. A design that shows the aggregate and the individual on the same screen can be tried in visualizations of people too.

Next is to show how the sample was made. In 100% City, the 24-hour chain and the people who could not be found are told as part of the performance. In a visualization too, you can put near the chart where the numbers were collected from and who is missing.

Third, the number changes when the way of answering changes. In Montréal, an answer given by four people on the lit stage became more than ten in the dark. From the two numbers, it can be read that the survey method is part of the result.

Finally, keep a count of what cannot be put in a chart. Nakamura Kazuyuki said he chose only the items that could be looked up, and that 100 people is not enough of a sample. Paterson and Windeyer pointed to the people who are not on stage. Noting what was counted and what was not can be read as a clue for reading a chart made with human bodies.

References

Primary sources

Rimini Protokoll, “100% City” (overview and performance list): https://www.rimini-protokoll.de/website/en/projects/100-stadt-7-1
Rimini Protokoll, “100% CITY” tour portfolio PDF: https://www.rimini-protokoll.de/website/media/Touringportfolios/Portfolio_100%25.pdf
Rimini Protokoll, “100% Berlin”: https://www.rimini-protokoll.de/website/en/project/100-prozent-berlin
Rimini Protokoll, “100% Berlin reloaded”: https://www.rimini-protokoll.de/website/en/project/100-berlin-reloaded
Rimini Protokoll, “100% Vienna”: https://www.rimini-protokoll.de/website/en/project/100-prozent-wien
Rimini Protokoll, “100% Melbourne”: https://www.rimini-protokoll.de/website/en/project/100-percent-melbourne (making-of video: https://vimeo.com/53318534 )
Rimini Protokoll, “100% Copenhagen”: https://www.rimini-protokoll.de/website/en/project/100-kopenhagen
Rimini Protokoll, “100% London”: https://www.rimini-protokoll.de/website/en/project/100-london
Rimini Protokoll, “100% Tokyo”: https://www.rimini-protokoll.de/website/en/project/100-tokyo
Rimini Protokoll, “100% Montréal”: https://www.rimini-protokoll.de/website/en/project/100-montreal (documentation video: https://vimeo.com/224155970 )
Interview with Helgard Haug, “100% Stellenbosch: A whole town on a stage,” LitNet, 2018-02-23 (reprinted by RP): https://www.rimini-protokoll.de/website/en/text/100-stellenbosch-a-whole-town-on-a-stage
Conversation between Rimini Protokoll and Daniel Koczy, “’…faces behind the numbers’,” Contemporary Theatre Review, 2015: https://contemporarytheatrereview.org/2015/rimini-protokoll-100-percent-city/
Interview with Haug and Kaegi, “Tableaux vivants,” Goethe-Institut Canada, 2017 (reprinted by RP): https://www.rimini-protokoll.de/website/en/text/tableaux-vivants
HAU Hebbel am Ufer, “100% Berlin reloaded”: https://www.hebbel-am-ufer.de/en/programme/pdetail/rimini-protokoll-haug-kaegi-wetzel-100-berlin-reloaded/
Festival/Tokyo 13, 100％トーキョー (100% Tokyo) program page: https://www.festival-tokyo.jp/13/program/13/100percent_tokyo/
Same, day-of program PDF (Tōjitsu panfuretto; includes the interview with Nakamura Kazuyuki and the contribution by Katsura Mana): https://www.festival-tokyo.jp/13/program/13/pdf/FT13_TP_100%25tokyo_web.pdf
Rimini Protokoll, “100% [City],” in Huron, Nagel, Oehlberg, Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, A K Peters/CRC Press, 2022, ch. 2.4, pp. 147–162 (chapter text not checked): https://doi.org/10.1201/9781003264903-9
Making with Data, “Structure” (the book’s parts and chapters): https://makingwithdata.org/blog/structure
dataphys.org, “2008 – 100% City”: https://dataphys.org/list/100-city/

Secondary sources

Eddie Paterson, “Bodies of Data: Informatics in Contemporary Performance,” Performance Paradigm 12, 2016: https://performanceparadigm.net/index.php/journal/article/download/178/174
Richard C. Windeyer, “Petri Dish Deceptions: The Search for 100% Veracity in Rimini Protokoll’s Statistical Portrait of Montreal,” Canadian Theatre Review, 2018 (reprinted by RP): https://www.rimini-protokoll.de/website/en/text/petri-dish-deceptions-the-search-for-100-veracity-in-rimini-protokoll-s-statistical-portrait-of-montreal
Peter Crawley, “Cork’s vital statistics,” The Irish Times, 2013-07-02 (reprinted by RP): https://www.rimini-protokoll.de/website/en/text/cork-s-vital-statistics-1
Sarfraz Manzoor, “100% London tells the capital’s story through its inhabitants,” The Guardian, 2012 (reprinted by RP): https://www.rimini-protokoll.de/website/en/text/100-london-tells-the-capital-s-story-through-its-inhabitants
Douglas Eacho, “Serial Nostalgia: Rimini Protokoll’s 100% City and the Numbers We No Longer Are,” Theatre Research International 43(2), 2018, pp. 185–200 (abstract only checked). doi:10.1017/S0307883318000299 / https://www.cambridge.org/core/journals/theatre-research-international/article/abs/serial-nostalgia-rimini-protokolls-100-city-and-the-numbers-we-no-longer-are/DA0D36242D6CB84BE8C8789935EAB8D6
Anja Quickert, “Theatrale Feldforschungen: Welt,” in Struktur und Ästhetik, Narr Francke Attempto, 2024, pp. 149–176: https://www.rimini-protokoll.de/website/media/100prozent/strukturundaesthetik_Anja%20Quickert.pdf

Shrinking the Whole Budget to 8,880 Pennies, and Showing a $100 Million Cut as a Quarter of One: Matt Shapiro's Budget Video

Tue, 03 Nov 2026 00:00:00 +0900

In 2009, Matt Shapiro shrank the entire US federal budget to 8,880 pennies (one penny = $400 million). He then showed the $100 million spending cut announced by President Obama as one quarter of a single penny. The video is titled “Obama Budget Cuts Visualization.”

The work is listed in the List of Physical Visualizations (dataphys.org), a collection of data physicalization works, under the heading “Federal Budget Explained with 10,000 Pennies.” That heading was given by dataphys.org; the “10,000” comes from the name of the channel that posted it, “10000Pennies.” I have not been able to watch the original video. What follows relies on what FactCheck.org confirmed with the author by email, the author’s blog, and contemporary press coverage.

April 2009: An Order to Cut $100 Million

According to Joe Miller of FactCheck.org, on April 20, 2009, President Obama gathered his Cabinet and directed them to cut a total of $100 million in expenses within 90 days.

On April 20, President Barack Obama caused a bit of a splash when he gathered members of his Cabinet and directed them to cut (collectively) $100 million in expenses within the next 90 days.

The same article introduced Shapiro’s work as a video that helps put this $100 million in perspective.

Shapiro, the author, was a 28-year-old software developer living in Salt Lake City (FactCheck.org). The local station KSL spelled his name Matthias Shapiro and reported that he lived in nearby Holladay. The YouTube channel he posted to, “10000Pennies,” says in its self-introduction that it visualizes policy so that anyone can understand it. The author later started a blog under the name “Political Math.” However, the blog’s first post is dated May 15, 2009, after this video. At the time of this work, the author was less a blogger than a video poster.

I have not been able to confirm the posting date on YouTube. Articles embedding the video appear from April 28 (GovExec), so it can be read as posted in the few days before that.

A screen showing pennies spread across a wooden desk, with white and blue lines running vertically over the coins and a right hand touching a coin in the foreground

Source: Political Math (10000Pennies) / List of Physical Visualizations (dataphys.org)

dataphys.org presents this screen as one from “Obama Budget Cuts Visualization.” What it shows is pennies spread on a desk and two lines, one white and one blue. I could not confirm from written sources how the coins were rearranged during the video or what the lines mark off.

The Budget as 8,880 Pennies, the Cut as a Quarter of One

FactCheck.org confirmed the correspondence of numbers with the author by email.

Matt, whose YouTube channel is entitled 10000 pennies, told us by e-mail that he actually uses 8,880 pennies in his video. The math checks out. Each penny represents $400 million in government spending, so that works out to a budget of $3.552 trillion — pretty close to Obama’s proposed 2010 budget. Obama’s pledge amounts to cutting one-fourth of one of those pennies.

In short: the video uses 8,880 pennies, not 10,000. Each penny stands for $400 million in government spending, the total comes to $3.552 trillion, and that is close to the 2010 fiscal year budget the Obama administration proposed. The President’s $100 million is one quarter of one such penny. FactCheck.org wrote that “the math checks out.”

The correspondence between the data and the objects is as follows.

Data	Physical counterpart	Value
Total federal budget	Number of penny coins (quantity)	8,880 pennies = $3.552 trillion
Conversion unit	One penny	1 penny = $400 million
Cut announced by the President	A fraction of one penny	$100 million = one quarter of a penny
Share of the cut in the whole	Ratio of 8,880 pennies to a quarter of a penny	About 0.0028% (my calculation)

Dividing $100 million by $3.552 trillion gives about 0.0028%, which equals a quarter of a penny divided by 8,880 pennies.

Here, the penny is not used at its original value of one cent. One penny becomes a symbol for $400 million. What the author replaced was not the unit of the $100 million sum but the ratio of the part to the whole, it can be read. The whole budget is shrunk by a single scale, and the cut is shown on that same scale. It can be read as the same idea as a scale model.

Around the same time, the economist Greg Mankiw explained the same ratio in another form. According to the Christian Science Monitor, Mankiw put the cut at 0.003% of the budget and wrote that for a household spending $100,000 a year it would be $3, about one latte at Starbucks. It can be read that Mankiw turned the ratio into “household units,” while Shapiro turned it into “a scale that makes the whole a heap of coins.”

The author wrote by email about the concern behind this form.

I know that we have a really hard time intuitively grasping scale when we’re dealing with two numbers that are orders of magnitude away from each other.

His concern is that it is very hard to grasp intuitively the difference in scale between two numbers that are many orders of magnitude apart. According to FactCheck.org, the author made this video as a way to see numbers “a little fresh,” compared with the usual bar chart.

A video thumbnail with a large reverse side reading ONE CENT and an obverse showing Lincoln’s profile, overlaid on pennies spread across a desk

Source: 10000Pennies (Political Math) / YouTube, “Obama Budget Cuts Visualization”

The video is available on YouTube (https://www.youtube.com/watch?v=cWt8hTayupE ). According to dataphys.org, it has 1.7 million views.

I do not know whether the quarter penny was shown by cutting a coin or by a diagram. dataphys.org says “videos often end with a dramatic action,” and lists examples such as cutting a penny in two or spilling several hundred coins off a table. But this describes the series as a whole and does not say which action belongs to this work.

From Coins to Water and Ice: Repeating the Same Procedure

The author later made videos following the same procedure with water and ice. dataphys.org says that “not only coins, but also bills, water, and whiskey represent amounts of money and numbers of jobs,” mentioning whiskey as well, and the channel also has a Lego video. However, I could not confirm the contents of these two.

On May 13, 2009, he released “Obama’s $17 Billion Budget Cuts In Context.” This time the $17 billion cut proposal is shown as a quantity of water. According to Brooks Jackson of FactCheck.org, the cut amounts to 2 liters. The video had initially given the difference between the Obama budget and the Bush budget ($3.107 trillion) as 9 gallons.

A video thumbnail of a park lawn, with a hand moving water from a large blue container into a small container

Source: 10000Pennies (Political Math) / YouTube, “Obama’s $17 Billion Budget Cuts In Context”

The 9 gallons was an error, which FactCheck.org found. The correct figure was about 12 gallons. The author added a correction in text to the video and thanked FactCheck.org in the description. In the same description, he also wrote that a $100 million cut would not even be “a drop in the bucket.”

In “What Does The Federal Budget Freeze Look Like?” (January 30, 2010), he showed a budget freeze with water and ice. The author devotes the first half of the blog post to explaining the data used in the video. He explains that if the budget is taken as 192 ounces of water divided into 4-ounce cups, all the calculations in the video work out. The first-year savings of $15 billion are 0.75 ounces of water frozen into ice, measured with one and a half tablespoons.

A table covered with a white cloth, with transparent cups of blue and green water in a row and a small piece of ice a little apart, labeled “Budget Freeze Savings”

Source: Political Math (10000Pennies) / List of Physical Visualizations (dataphys.org)

Putting the three videos side by side gives the following.

Video	Whole	Part taken out
Obama Budget Cuts Visualization (April 2009)	Budget $3.552 trillion = 8,880 pennies	$100 million cut = a quarter of a penny
Obama’s $17 Billion Budget Cuts In Context (May 2009)	Budget = a quantity of water	$17 billion cut = 2 liters
What Does The Federal Budget Freeze Look Like? (January 2010)	Budget $3.7 trillion = 192 ounces of water	$15 billion savings = 0.75 ounces of ice

Dividing $3.7 trillion by 192 ounces gives about $19.3 billion per ounce. 0.75 ounces is about $14.5 billion, which nearly matches $15 billion. Even as the material changes, the procedure appears to stay the same: fix the quantity of the whole first, then take the policy amount out as a part of it.

In one of his earliest blog posts, “The National Debt Road Trip” (May 15, 2009), the author writes that he tilted the numbers in Obama’s favor so as to leave no room for the charge of fudging. This is not a statement about this work, but it offers a clue to how he handled numbers.

How Third Parties Saw the Video

The first point is the verification of the numbers. FactCheck.org confirmed the number of coins and the unit value with the author and wrote that the math was right. In the follow-up, it found an error, confirmed that the author had fixed it, and then commented as follows.

Thanks to Matt’s helpful images, you don’t have to be an accountant to grasp the size of the budget problem.

The assessment is that, thanks to Matt’s clear images, you do not need to be an accountant to grasp the size of the budget problem. The first article also notes that the liberal economist Paul Krugman made the same point on his blog, that $100 million is small.

The second point is the response to the difficulty of large numbers. Megan McArdle of The Atlantic introduced this work in an April 2009 blog post.

Numbers a million times bigger than our total lifetime earnings literally boggle the imagination.

In short, a number a million times our lifetime earnings is literally beyond imagination. McArdle introduced the author as an “ambitious video blogger” who took on that problem. Alyssa Rosenberg of GovExec also cited that post, embedded the video, and called it a “fun, educational video.”

The third is a skeptical view. Adam Paul of The Think Tankard wrote on May 4, 2009, linking to McArdle’s post.

Video bloggers have taken to using piles of pennies to represent the budget … The concern is great and if the goal is simply to inform people it’s doing a good job. Yet any hopes that people might apply, or that that matter retain, their knowledge suffers from the same problems that “awareness” movements always have: how to turn knowledge into action.

His point: video bloggers have taken to representing the budget with piles of pennies, and if the aim is to inform, they do a good job. But whether the knowledge is retained or leads to action runs into the same problem that awareness campaigns always have.

dataphys.org introduces the author as someone who “criticizes the US policy” with statistics and handmade physical visualizations. The author himself wrote in the description of the follow-up that a $100 million cut would not even be “a drop in a bucket” (“weren’t even a drop in a bucket”). According to KSL, in July 2009 Republican National Committee Chairman Michael Steele presented the author with a “Grassroots Logic Award.” From these, it can be read that the video also had a partisan aim of criticizing the administration. At the same time, placing FactCheck.org’s verification next to Krugman’s identical point suggests that the way of presenting the ratio itself holds across party lines.

Takeaways

The first thing to take from this video is to put the whole and the part on the same scale. Showing only $100 million does not settle whether it is large or small. The design of preparing the whole as a quantity of 8,880 pennies and assigning the cut a quarter of one penny can be read as a way of making the denominator of the ratio visible.

The second is to choose the unit’s value from the size at which the whole fits. The value of $400 million per penny can be read as chosen so that the whole budget fits within a manageable quantity. At $100 million per penny the cut would take one penny, but the whole would exceed 35,000. The unit is decided by whether you want the whole or the part to be the more manageable quantity.

The third is to make the basis of the conversion checkable from outside. FactCheck.org confirmed the calculation just by asking for the number of coins and the unit value, and found an error in the follow-up. For the freeze video, the author published the conversion himself on his blog. If the basis is open, outsiders can give you the occasion to correct it.

Finally, the limits. As Adam Paul wrote, understanding scale and changing behavior are different things. The question this video answered was how to compare two numbers that differ by many orders of magnitude on a single scale. What comes after that is left to the viewer and to the places where the debate happens.

References

Primary sources

List of Physical Visualizations (dataphys.org), “2009 – Federal Budget Explained with 10,000 Pennies” (contributed by Pierre Dragicevic): https://dataphys.org/list/federal-budget-explained-with-10000-pennies/
dataphys.org tag pages, storytelling: https://dataphys.org/list/tags/storytelling/ / budget: https://dataphys.org/list/tags/budget/
10000Pennies, “Obama Budget Cuts Visualization” (YouTube, April 2009): https://www.youtube.com/watch?v=cWt8hTayupE (title and poster confirmed via oEmbed: https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=cWt8hTayupE&format=json )
YouTube channel 10000Pennies (video list, channel description): https://www.youtube.com/user/10000Pennies/videos
10000Pennies, “Obama’s $17 Billion Budget Cuts In Context” (YouTube, 2009-05-13, description): https://www.youtube.com/watch?v=miIBp7ufxjk
10000Pennies, “What Does The Federal Budget Freeze Look Like?” (YouTube, 2010-01-30): https://www.youtube.com/watch?v=gMH9QZdjS2Y
Political Math blog: https://www.politicalmathblog.com/
Political Math, “What Does the Federal Budget Freeze Look Like?”, 2010-01-31: https://www.politicalmathblog.com/2010/01/31/what-does-the-federal-budget-freeze-look-like/
Political Math, “The National Debt Road Trip”, 2009-05-15: https://www.politicalmathblog.com/2009/05/15/the-national-debt-road-trip/

Secondary sources

Joe Miller, “What’s in a Number?”, FactCheck.org, 2009-05-07: https://www.factcheck.org/2009/05/whats-in-a-number/
Brooks Jackson, “Drop in the Bucket? See for Yourself.”, FactCheck.org, 2009-05-15: https://www.factcheck.org/2009/05/drop-in-the-bucket-see-for-yourself/
Megan McArdle, “Penny Wise, Pound Foolish”, The Atlantic, April 2009 (the original URL, http://meganmcardle.theatlantic.com/archives/2009/04/penny_wise_pound_foolish_1.php, no longer opens). Full-text reprint: https://thinkingonthemargin.blogspot.com/2009_04_26_archive.html
Alyssa Rosenberg, “What $100 Million Looks Like”, GovExec, 2009-04-28: https://www.govexec.com/federal-news/2009/04/what-100-million-looks-like/38913/
Adam Paul, “Learning Large Numbers”, The Think Tankard, 2009-05-04: https://thinktankard.blogspot.com/2009/05/learning-large-numbers.html
Andrew Adams, “Utahn’s YouTube video catches RNC’s attention”, KSL, 2009-07-08: https://www.ksl.com/article/7084161
Legal Insurrection, “Obama Budget Cuts Visualization”, 2009-04-29: https://legalinsurrection.com/2009/04/obama-budget-cuts-visualization/
“Obama’s $100 million budget cut amounts to a cup of coffee”, Christian Science Monitor, 2009-04-21 (Mankiw’s analogy; it does not mention this work): https://www.csmonitor.com/USA/Politics/The-Vote/2009/0421/obamas-100-million-budget-cut-amounts-to-a-cup-of-coffee

A Century of Search Terms as the Height of Ten Moving Bars: Centograph

Tue, 03 Nov 2026 00:00:00 +0900

Centograph is a moving bar graph made in 2009 by the London studio Tinker. It shows the relative number of Google News Archive articles that mention a search term, as the height of ten moving bars. Each bar is assigned to one of the ten decades from 1909 to 2009.

The client was St Paul’s School for Boys in London. The work was made for the school’s 500th anniversary celebration and was later installed permanently in its Computing Department. What follows draws on the Tinker London project page, the description of the video Tinker posted on Vimeo, and the description of the Flickr album. All of them are short texts, and I could not find any technical documentation that explains how the work operates.

May 2009, at the 500th-anniversary Open Day

The Tinker London project page gives the date of the work as “May 2009” and the client as “St Paul’s School for Boys”. The collaborator is Ben Burry, credited with web programming. No individual designer is named. In the Vimeo description, the maker is given as “Tinker.it!”.

The school was founded in 1509, so 2009 was its 500th anniversary (Wikipedia). The project page states the purpose of the work as follows.

Centograph is a kinetic bar graph that was developed for the 500th anniversary celebration of St. Paul’s School for Boys. It was designed to encourage users to reflect on possible connections between new and emerging technologies and other academic and social disciplines.

In short, it is a moving bar graph made for the school’s 500th-anniversary celebration, designed to get users thinking about connections between new technologies and other academic and social fields.

According to the Flickr album description, the commission was for the 500th-anniversary “Open Day”.

A black box-shaped Centograph in a room. Year labels line its front, a laptop sits on a desk beside it, and two people stand in front

Source: Tinker.it! (Flickr: tinker_it) / Tinker London, “Centograph” project page

The Flickr photo was taken between May 20 and 25, 2009. The Vimeo video (49 seconds) was posted on June 2, 2009. The project page says “Centograph is now a permanent fixture in St Paul’s School for Boys Computing Department.”, so the work was installed at the school permanently.

Type a search term, and ten bars rise and fall

The project page describes how it works as follows.

When users enter a search term on an online interface, Google News Archive is queried for related news articles from the past 100 years and returns a histogram of articles sorted according to date. That graph is then made physical by pushing and pulling the bars on the graph.

When a term is entered, Google News Archive is queried for related articles from the past 100 years, and what comes back is a histogram of the articles sorted by date. That graph is then made physical by pushing and pulling the bars.

The Vimeo description is clearer about what the height of a bar means.

The bars on the graph then change height to display a histogram of the relative number of news articles for each decade.

The bars change height to show a histogram of the “relative number” of news articles for each decade. The Flickr photo description says how many years one bar covers.

Each bar represents a decade from 1909 to 2009.

Each bar represents one decade between 1909 and 2009. The number of bars is given as “Ten” in the dataphys.org introduction (quoted in the section on third-party views).

On the front of a black base, year labels such as 1929 and 1939 are lined up, and plank-like bars stand on top of the base

Source: Tinker London, “Centograph” project page

Taken together with the opening photo, labels with the years 1919, 1929 and 1939, ten years apart, line the front of the base (from 1949 on, the labels cannot be read in the photos). The Flickr description says “from 1909”, but the leftmost label legible in the photos is 1919. The sources do not say whether a label’s year marks the start or the end of its decade.

Plank-like bars lit in red rise from slits in a black base

Source: Tinker London, “Centograph” project page

The bars come up through slits cut in the base. You can see them move in Tinker’s Vimeo video.

The correspondence between data and object that the sources let us establish is summarized below.

Data	Property of the object	Description	Source
Decades (1909 to 2009, ten years each)	The row of bars; one bar = one decade	“Each bar represents a decade from 1909 to 2009.” The count is “Ten” in dataphys.org	Flickr, dataphys.org
The relative number of news articles related to the search term in that decade	Height of the bar	“the relative number of news articles for each decade”	Vimeo
Switching the search term	Up-and-down movement of the same ten bars	“pushing and pulling the bars”, “moving the bars up and down”	Tinker London, Flickr
The whole period	The row of ten bars as a whole	“over the past 100 years”	Tinker London, Vimeo
Marker for each decade	Year labels on the front of the base	1919, 1929, 1939 (1949 onward cannot be read in the photos)	Photos of the work
Input	A search screen separate from the object; Twitter	“an online interface”, “twitter enabled”	Tinker London, Flickr

Some things are not in the sources. They do not say how the relative number was calculated, which value was mapped to the maximum height, what range the bar heights cover, or how the bars are moved. Because the normalization is unknown, you cannot read the article count back from the height of a bar.

Terms also arrive from Twitter

The place where search terms are entered was separate from the object. The Vimeo description says “When you enter a search term into the computer”, and dataphys.org says “A separate search interface is provided on a regular desktop computer”. In the opening photo, too, a laptop sits on a desk beside the base.

The Flickr photo description mentions another way in.

Physical web and twitter enabled bar graph that looks up the frequency of keywords on Google News Archives and displays them by moving the bars up and down.

It is a physical bar graph connected to the web and Twitter, which looks up how often keywords appear in Google News Archive and shows the result by moving the bars up and down.

The screen of the centograph account on Twitter. Short posts such as thomas, st paul’s, 2nd world war and great war are listed

Source: Tinker London, “Centograph” project page

A screenshot on the project page shows a Twitter account named centograph. Its location is “St Paul’s School, London”, and its bio is the same as the Flickr sentence above. The posts include terms such as “thomas”, “nicholas”, “st paul’s”, “2nd world war”, “great war” and “pineapple”, all posted from Twitter4J on the evening of May 23.

The sources do not say whether these posts are terms that reached the work from Twitter, or a record of terms the work had received. Nor do they say who entered these terms, or on what occasion.

Not the numbers, but the “shape” of a century

The project page and the Vimeo description contain almost the same sentence.

This allows users to view the ‘shape’ of the past century in relation to different key words—from progression in computing technology to times of war and peace to changing sources of energy.

Users can view the “shape” of the past century for different keywords, from progress in computing technology, to times of war and peace, to changing sources of energy.

All three subjects the makers name are terms whose article counts would probably vary a great deal from decade to decade. Centograph can be read as a tool conceived less for reading off numbers than for comparing where each term’s peaks and valleys fall among the decades. The ten decades along the horizontal axis stay fixed, and only the heights change. Tinker writes the height as a “relative number” and names what it wants viewers to see as the “shape”. This suggests a design that put comparing shapes ahead of letting viewers read exact counts.

The data source, Google News Archive, is a newspaper-archive search that opened in September 2006. A blog post in Information World Review at the time introduced one of its features this way.

Another striking innovation is a Show Timeline option, which categorises results according to their publication dates.

The archive had a feature called “Show Timeline” that sorted search results by publication date. The project page’s “returns a histogram of articles sorted according to date” can also be read as describing counts received per decade in this way. Centograph can be read as moving the on-screen timeline into the height of bars. That said, it has not been confirmed that the work used this feature.

The sources also do not say why the period was set at the 100 years from 1909 to 2009. It can be read as fitting the framing of looking back on “the past century” at a 500th-anniversary celebration.

How third parties saw the work

The researcher Pierre Dragicevic has listed this work in the List of Physical Visualizations on dataphys.org. It is categorized as “Active physical visualization”, with the tags bar charts, news and temporal data. The description reads as follows.

Ten actuated bar charts that show the popularity of keywords of interest in news articles over time, made by the company Tinker from London.

It introduces the work as ten moving bar charts showing how the popularity of keywords changes in news articles over time. It differs from the primary sources in two points, however. Tinker writes “a physical bar graph”, a single bar graph with ten bars. And Tinker’s own words are “frequency” and “relative number of news articles”; it does not use “popularity”. dataphys cites an infosthetics article (August 2009) as its source, but I have not been able to check its text.

A student blog from Carnegie Mellon University’s course 15-104 (Fall 2017) draws attention to where the work is placed.

By turning data from something esoteric into something tangible and cool, it helps to incite a level of interest in the areas of computation and data visualization, especially in students.

The evaluation is that turning obscure data into something touchable raises interest in computation and data visualization, especially among students. The author counts it as a strength that the work is a permanent installation made for an educational setting. Since this is a student assignment, it is weak as an expert assessment. Still, the factual part, bars every ten years starting in 1909, agrees with the primary sources.

I found no newspaper or magazine coverage. The 2009 press items listed on Tinker London’s Press page do not mention Centograph either.

What to take from it

The first thing to take from this work is fixing the axis and swapping only the heights. The bars are always assigned to the same decades. Because the row does not change when the term changes, viewers can compare differences in shape.

Second is separating input from display. Search terms come in from a separate screen or from Twitter, and what moves on the object’s side is the height of ten bars. Each time a viewer tries a term, the same object becomes a different graph.

Last is the lifespan of a physicalization that relies on web data. According to Wikipedia, Google announced in 2011 that it would stop adding material to News Archive. Whether the work keeps running in the same form as back then has not been confirmed. For a work whose data source lies outside it, you need to consider what happens after the source stops being updated, as I read it.

References

Primary sources

Tinker London, “Centograph” project page: https://tinkerlondon.com/what-we-do-centograph.html
Tinker, “Centograph” (Vimeo, 2009-06-02, 49 seconds): https://vimeo.com/4961482 (description and metadata: https://vimeo.com/api/oembed.json?url=https://vimeo.com/4961482 )
Tinker.it!, “Centograph (work)” (Flickr album): https://www.flickr.com/photos/tinker_it/sets/72157618662102020/
Tinker.it!, “Centograph” (Flickr photo): https://www.flickr.com/photos/tinker_it/3555602327/
Tinker London, “Press”: https://tinkerlondon.com/press.html

Secondary sources

dataphys.org, “2009 – Centograph: Dynamic Bar Charts Show Keyword Popularity” (listed by Pierre Dragicevic): https://dataphys.org/list/centograph-dynamic-bar-charts-show-keyword-popularity/
mmiller5, “Looking Outwards-07” (Carnegie Mellon University 15-104, Fall 2017): https://courses.ideate.cmu.edu/15-104/f2017/index.html%3Fp=17742.html
infosthetics, “centograph: physical bar graph” (August 2009; text not checked): http://infosthetics.com/archives/2009/08/centograph_physical_bar_graph.html

Background sources

Information World Review, “Historic perspectives in Google News” (2006-09-11): https://informationworldreview.blogspot.com/2006/09/historic-perspectives-in-google-news.html
Wikipedia, “Google News Archive”: https://en.wikipedia.org/wiki/Google_News_Archive
Wikipedia, “St Paul’s School, London”: https://en.wikipedia.org/wiki/St_Paul%27s_School,_London

One IKEA Box for a Billion People: Hans Rosling's Physical Visualizations

Tue, 03 Nov 2026 00:00:00 +0900

On June 21, 2010, at TED@Cannes in Cannes, Hans Rosling represented the world’s population with boxes, using the rule “one IKEA box = one billion people.” He lined the boxes up horizontally along levels of income and moved them by hand as the years advanced: 1960, 2010, 2050. In the second half of the same talk, from 6:08, he switched to an animated bubble chart on the screen.

The source is the transcript of the talk “Global population growth, box by box” (9 min 47 s), published by TED. dataphys.org, a catalog of physical visualizations collected by researchers, treats this talk as the starting point and gathers the later talks and videos in which Rosling used physical objects into a single entry. The entry was submitted by Pierre Dragicevic and Yvonne Jansen. Below, I follow the 2010 talk in chronological order and cover the later examples only to the extent they could be confirmed.

June 2010: IKEA boxes on the stage in Cannes

Rosling was a Swedish physician, a professor of international health at the Karolinska Institute, and a co-founder of Gapminder. According to dataphys, the technology and staging of the talk were handled by his son, Ola Rosling.

The 2010 talk begins with a joke.

I’m going to talk now about how world population has changed from that year and into the future, but I will not use digital technology, as I’ve done during my first five TEDTalks. Instead, I have progressed, and I am, today, launching a brand new analog teaching technology that I picked up from IKEA: this box.

The preamble: he used digital technology in his first five TED talks, but today he will use a “new analog teaching technology” he found at IKEA, namely this box.

He then says, “This box contains one billion people,” and recreates the world of 1960 as he heard about it from a schoolteacher. The developed countries hold one billion people and the developing countries two billion. Families in the developed countries save up to buy a Volvo; families in the developing countries save for the day’s food and for a single pair of shoes. What a family wants to get becomes the marker for where a box is placed.

Rosling in front of a screen reading 2010, speaking beside a pile of yellow-green boxes and a blue box, with miniature sandals, a car, and an airplane

Source: TED, “Global population growth, box by box”

Moving the boxes from 1960 to 2050

A box’s horizontal position is a level of living. Small objects mark the levels: shoes, a bicycle, a car, an airplane. infosthetics (July 2010) describes these as “miniature icons of living conditions.”

Advancing to 2010, there are seven boxes, because four billion people were added in 50 years. The West’s one billion moves to the airplane position. The developing countries that have grown move to the car position. The biggest pile is the three billion at the bicycle position. And at the shoe position, two billion people remain, about as poor as they were 50 years earlier. Rosling says:

But there is a continuous world from walking, biking, driving, flying – there are people on all levels, and most people tend to be somewhere in the middle.

In short: the world is continuous from walking through biking and driving to flying, and most people are somewhere in the middle.

He moves the boxes for the 2050 outlook as well. China and others move to the airplane position. The poorest two billion grow to three or four billion. But if they escape poverty and children survive, population growth stops in 2050. The number and position of the boxes show this conditional outlook.

Rosling leaning in behind miniature bicycle, car, and airplane models, in front of a screen reading 2050, with yellow-green boxes stacked on a blue box

Source: TED (talk video) / dataphys.org, “Hans Rosling Adopts Physical Visualizations”

The colors of the boxes carry meaning too. At the end of the talk, Rosling calls the developed countries “this blue box.” The CHI 2015 paper by Jansen et al. describes this scene as “rich countries = blue box, poor countries = green box.”

At 6:08, the bubble chart on the screen

In the second half of the talk, Rosling moves from the boxes to the screen.

And let me show you why by converting back to the old-time digital technology. Here I have on the screen my country bubbles. Every bubble is a country. The size is population.

He says he will show why this happens using “old-time digital technology,” turning the opening joke around. Each bubble on the screen is one country; its size is population and its color is continent. The vertical axis is the number of children per woman, and the horizontal axis is the share of children who survive to school age. Advancing the years from 1960, the motion shows child survival rising and family sizes shrinking.

At 7:47, Rosling ties the boxes to the bubbles.

Can you see the poorest billion, those [two] boxes I had over here? They are still up here.

The point: the boxes of the poorest people, which he placed over there earlier, are still up here (at the position with many children). His conclusion was that raising child survival to 90% is the only way to stop population growth.

Putting the encodings of the first and second halves side by side gives the following.

	First half: IKEA boxes	Second half: bubble chart
What one mark represents	One billion people	One country
Quantity	Number of boxes	Bubble size (population)
Horizontal position	Level of living (marked by small objects: shoes, bicycle, car, airplane)	Child survival rate
Vertical position	None (boxes at the same level are stacked)	Children per woman
Color	Blue = developed (rich) countries, green = poor countries (as described in the CHI 2015 paper)	Continent
Time	Boxes moved by hand (1960 → 2010 → 2050)	Animation (1960 → present)

What the boxes in the first half show is the magnitude and composition: how many billions of people are at which level. The bubbles in the second half track 50 years of change country by country along two axes. Within a single talk, he can be read as dividing the roles between the physical visualization and the animation. The sentence at 7:47 can be read as a bridge that hands the “billion-person blocks” counted in the first half over to the cloud of country dots in the second.

Later boxes, and other materials

dataphys’s description begins with stacking IKEA boxes in 2010 and says that he later “switched” to pebbles, building toys, juice, snowballs, and toilet paper. He did not, however, give up boxes.

On April 16, 2012, at TEDxSummit in Doha, Qatar, Rosling gave “Religions and babies.” This time the order is reversed. At 2:20 he shows the bubble chart first and changes the color to “the majority religion of each country.” The boxes appear at 10:25.

I will use these card boxes in which your notebooks came. They are quite useful for educational purposes. Each card box is one billion people.

He declares that he will use the cardboard boxes the notebooks handed out at the venue came in, with one box standing for one billion people. The boxes are stacked vertically by age group: 0 to 15 years, two billion; 15 to 30, two billion; 30 to 45, one billion; 45 to 60, almost one billion; 60 and over, one billion. Rosling notes, “These are rounded numbers.” The roughly three billion people missing from the upper age groups will fill in over time. He calls this “the big fill-up” and says the world will reach 10 billion.

Rosling on the TEDxSummit stage, lifting the top box of a vertical stack of cardboard boxes

Source: TED, “Religions and babies”

Here is a summary of how each material was used, as far as could be confirmed.

Material	Occasion	What one piece represents	Second variable
IKEA storage box	TED@Cannes, 2010	One billion people	Horizontal position = level of living
Cardboard box	TEDxSummit (Doha), 2012	One billion people	Vertical stacking = age group
Stones	Interview during TEDxSummit, 2012	Not confirmed	Income levels compared by means of transport
LEGO (DUPLO) blocks and buttons	The Guardian video, 2013	Not confirmed	Not confirmed
Toilet paper	No specific talk or program identified	One billion people (as described by an eyewitness)	Age group (as described by an eyewitness)

The stone example was improvised in an interview setting during TEDxSummit, and Herbert Reininger posted a video of under a minute on YouTube.

A hand arranging white stones on a tiled ground, with a black stone set to the side

Source: Herbert Reininger (YouTube video) / dataphys.org, “Hans Rosling Adopts Physical Visualizations”

In a The Guardian video of May 17, 2013, he uses LEGO blocks to explain population growth, child mortality, and CO2 emissions, according to the video’s description. The correspondence between the blocks and the numbers has not been confirmed. dataphys also lists a LEGO video from the Brookings Institution as part of the same stream as Rosling’s storytelling (related article).

For toilet paper, a blog post by Statistics Finland (2020) describes a scene in which one roll stood for one billion people, stacked by age group. But which talk or program the scene comes from has not been identified. Snowballs appear in the title of a January 2015 BBC Newsnight video, but what they represented has not been confirmed. Juice is said to have been used by dataphys and others, but the scene and what it represented have not been identified.

What the confirmed examples have in common can be read as this: count one piece as one billion people, and let position carry a second variable. They can be read as visualizations that convey magnitude and composition rather than exact values.

How did third parties see these demonstrations?

For a researchers’ placement of the work, there is the paper “Opportunities and Challenges for Data Physicalization” (CHI 2015) by Jansen, Dragicevic, Petra Isenberg, and others. In its Figure 2, the paper cites Rosling’s boxes as an example of a physicalization “updated by hand.” On the left of the same figure is a quality board built from LEGO by a GM engineer.

His physicalizations are useful and effective in communicating information which individuals may find difficult to comprehend due to unfamiliar magnitudes and scales. It can also carry a shock value.

The assessment: they help convey unfamiliar magnitudes and can also carry a shock value. The paper goes on to say that the strength of Rosling’s physicalizations lies in their “pedagogical and persuasive power,” which conventional evaluation by time or error count cannot measure. This points to the need for a different yardstick than the experiment that compared real bar charts on speed.

The visualization researcher Robert Kosara wrote a tribute on eagereyes on February 7, 2017, after Rosling’s death.

Rosling demonstrated the importance of a human voice in visualization. He also used props: apples, rolls of toilet paper – he even swallowed a sword. This is how you make people watch, and this is how you leave an impression.

The assessment: props are a means of making people watch and leaving an impression. Kosara reads Rosling as someone who opened up visualization for explaining, teaching, and entertaining.

There are also assessments that include criticism. On an NPR program on February 11, 2017, the Stanford mathematician Keith Devlin acknowledged the power of animation to overturn how we see the world. He then pointed to the risk that it is taken to mean “don’t worry,” and to problems when the data are used for prediction. This is an assessment of animated charts, not of the boxes.

Uses

Rosling told the dataphys editors:

physical props reach the mind of a large part of the public because thats how you represent data in daily life at the kitchen table, at the cafe and in the bar.

In other words: physical props reach many people because that is how people ordinarily represent data, at the kitchen table, at the cafe, and in the bar.

Three things can be carried back into visualization work.

Fix one unit and declare it. If you say at the start, “this box is one billion,” the audience can follow the magnitude of the world’s population just by counting boxes.
Give the second variable to position, and only one. In 2010 the horizontal position carried income; in 2012 the vertical stacking carried age group. With small-object markers, the levels are clear without reading axis ticks.
Divide the roles between objects and the screen. Show the overall magnitude and composition with boxes, and leave country-by-country change to the bubble chart. As at 7:47, pointing within the later chart to the objects from the first half connects the two representations.

Physical visualizations are not suited to reading fine values. Rosling can be read as having chosen the coarseness of billion-person units and handing the fine changes over to the screen.

References

Primary sources

dataphys.org, “2010 – Hans Rosling Adopts Physical Visualizations” (submitted by Pierre Dragicevic and Yvonne Jansen). https://dataphys.org/list/hans-rosling-adopts-physical-visualizations/ (tag list: https://dataphys.org/list/tags/ted/ )
Hans Rosling, “Global population growth, box by box,” TED@Cannes, recorded 2010-06-21. Transcript: https://www.ted.com/talks/hans_rosling_global_population_growth_box_by_box/transcript
Hans Rosling, “Religions and babies,” TEDxSummit (Doha), recorded 2012-04-16. Transcript: https://www.ted.com/talks/hans_rosling_religions_and_babies/transcript
TED, list of talks by Hans Rosling. https://www.ted.com/speakers/hans_rosling
Gapminder, “History of Gapminder.” https://www.gapminder.org/about/about-gapminder/history/
The Guardian, “Population growth and climate change explained by Hans Rosling,” 2013-05-17. https://www.theguardian.com/global-development/video/2013/may/17/population-climate-change-hans-rosling-video
Herbert Reininger, “Hans Rosling’s shortest TED talk,” YouTube. https://www.youtube.com/watch?v=UNs-ziziPyo
BBC Newsnight, “Hans Rosling on global income disparity (and snowballs),” YouTube, 2015. https://www.youtube.com/watch?v=DoSTNRhoceY

Secondary sources

Yvonne Jansen, Pierre Dragicevic, Petra Isenberg, et al., “Opportunities and Challenges for Data Physicalization,” CHI 2015. https://eprints.lancs.ac.uk/id/eprint/72691/1/OA_2015_04_CHI_DataPhys.pdf
Robert Kosara, “Hans Rosling: An Appreciation,” eagereyes, 2017-02-07. https://eagereyes.org/blog/2017/hans-rosling-an-appreciation
Keith Devlin, NPR Weekend Edition, 2017-02-11 (reprinted by WUSF). https://www.wusf.org/2017-02-11/remembering-the-edutainer-who-made-statistics-come-to-life
infosthetics, “More Recent Talks from Hans Rosling (with Original Physical Props),” 2010-07-19. https://infosthetics.com/archives/2010/07/more_recent_talks_from_hans_rosling_with_original_physical_props.html
Tilastokeskus (Statistics Finland), Tieto&trendit blog, 2020-03-18 (in Finnish). https://stat.fi/tietotrendit/blogit/2020/kayttoa-vessapaperille-kotiopetuksessa
NPR Goats and Soda, 2017-02-08. https://www.npr.org/sections/goatsandsoda/2017/02/08/514074090/video-edutainer-hans-rosling-who-taught-us-about-the-world-has-died

31 Years of Snow in 31 Drawers: Adrien Segal's Snow Water Equivalent

Thu, 05 Nov 2026 00:00:00 +0900

Adrien Segal’s Snow Water Equivalent is a storage cabinet that turns 31 years (1980–2010) of snowpack data into 31 drawers. Each drawer is one year, with 1980 at the bottom and 2010 at the top. The height of a drawer shows that year’s precipitation, and the relief carved into its plywood front panel shows how the snow water equivalent (SWE) changed through that year. The data were recorded by the NRCS (Natural Resources Conservation Service) at the Ebbetts Pass station in California’s Sierra Nevada.

What follows is based on the work page on the artist’s official website and on the artist’s own making-of account on the official website of the book Making with Data. In the book, the first chapter of Part 1, “Handcraft” (1.1 “Snow Water Equivalent,” pp. 37–50, doi:10.1201/9781003264903-1), is devoted to this work. I have not been able to read the chapter itself, so I use the making-of account, which is organized under the same headings as the chapter.

From childhood memories of Colorado snow to SNOTEL data

Segal is an artist based in Oakland, California, and teaches at the California College of the Arts (CCA). She studied furniture design at CCA and grew up in Colorado. According to the making-of account, the work grew out of her childhood memories of snow in the Colorado mountains. This is how Segal sees snow:

Snow is the perfect seasonal water storage system, one that captures excess precipitation as frozen ice crystals during the winter months.

In other words, snow is a seasonal water-storage mechanism that holds the surplus precipitation of winter as ice crystals.

Looking for data, Segal came across SNOTEL, the NRCS monitoring network. It is an automated network in the western United States that publishes precipitation, snow depth, SWE and other measurements. Segal chose a station in the Sierra Nevada, where she now lives. SWE is the depth of water you would get if the whole snowpack were melted at once.

Item	Details	Source
Station	SNOTEL Site: Ebbetts Pass, CA (Site Number 462)	The artist’s work page
Location and elevation	38°33′N, 119°48′W, elevation 8765 feet	The artist’s work page (the NRCS’s current site information gives 8660 ft, which does not match)
Period	31 years (1980–2010) of water years	The artist’s work page
Materials	Solid ash dyed black (frame), carved plywood (front panels)	Making-of account
Year made	2011	Heading on dataphys.org
Year made	2012	Caption in The Finch & Pea (2014)
Year made	The account says “1980 through 2010 when the sculpture was made,” which could also be read as 2010	Making-of account

The sources disagree on the year the work was made, so this article does not settle on one. I could not confirm the dimensions from any primary source.

A tall cabinet standing on the floor. On one face of the black frame, carved light-colored plywood front panels run in waves from top to bottom

Source: Adrien Segal / “Snow Water Equivalent” on Adrien Segal’s official website

Stacking 31 years of graphs into a template

Segal first explores the form on paper. The “Paper mockups” section of the making-of account says:

The annual charts of snow water equivalent were stacked to compare year to year snowpack levels.

That is, the yearly SWE charts were stacked to compare snowpack from one year to the next.

The sketch in the “Paper mockups” section of the making-of account (makingwithdata.org) has the axes labeled. Height is “SWE,” depth is “DAYS (255) OR 365” with OCT and SEP at the ends, and the length direction is “1979–2010,” with “WIDTH = Accum Precip TOTAL.” At this stage the yearly layers were lined up side by side. The one-year offset from “1980–2010” can be read as the result of the water year starting in October of the previous year.

A blue-pen sketch. Mountain-shaped layers for each year are lined up side by side, with annotations: SWE for height, DAYS for depth, 1979–2010 for length, and Accum Precip TOTAL for width

Source: Adrien Segal / “Snow Water Equivalent” on Making with Data

Once the proportions and scale were set, Segal fitted in the actual data points and, at this point, decided to orient the layers vertically. The making-of account says, “At this point I decided to orient the layers of data vertically.” From here she switches to digital tools.

I took each annual chart of snowpack into the computer program Adobe Illustrator and indexed them in relation to each other.

In short, she brought each year’s snowpack chart into Illustrator and aligned them with one another. The work page includes a set of the yearly SWE charts.

A figure of small blue mountain-shaped charts in three columns. In each, snow starts to accumulate on the left, builds a peak, and returns to zero on the right

Source: Adrien Segal / “Snow Water Equivalent” on Adrien Segal’s official website

Using this figure as a template, she cuts the plywood with a band saw, laminates and glues the layers, and shapes them by hand with an angle grinder and other tools. The making-of account says, “I grind away the plywood until it becomes a single fluid form.” After finishing with sandpaper, she cuts the form into its layers with a table saw and a dedicated jig. During the work she kept a spreadsheet of each layer’s dimensions in view.

The figure of overlaid small multiples becomes the blueprint for the three-dimensional object as it is, or so it can be read. Aligning the graphs is the same job as making the template for cutting the wood.

Drawer height and front-panel relief

The frame is solid ash, and its dimensions differ from year to year. Each drawer also has its own dimensions, and the making-of account says both were laborious. The cut plywood pieces are attached to the drawer fronts, and when the drawers are slid into the frame, the form of the original sculpture is restored. Finally, the drawers are assembled in chronological order, with 1980 at the bottom and 2010 at the top.

The table below sums up the correspondence between data and object as the artist describes it.

Data	Property of the object	Artist’s words (source)
Year (water year, 31 years)	One drawer. 1980 at the bottom, 2010 at the top	Making-of account
That year’s total precipitation	Drawer height	Work page
Day-by-day SWE through that year	Shape of the front panel. A cross-section of the drawer seen from above is that year’s SWE graph	Making-of account
Amount of water in that year’s snowpack	Drawer size (how much it can hold)	Former work page (quoted on dataphys.org)
Whether it is data or not	Front panel is light plywood, frame is ash dyed black	Making-of account
Snow as natural water storage	Type of furniture (a storage cabinet)	Making-of account

The former work page describes the front panels as a three-dimensional graph showing “the first snowfall, peak amount of water, and final snowmelt.”

The artist’s text does not say which direction on the front panel is the date and which is SWE. In the earlier sketch, SWE was height, days were depth, and precipitation was width. By orienting the layers vertically, precipitation, which had been the width, became the height of the drawer, and days and SWE moved onto the horizontal plane. That is how it can be read.

A section where light-colored plywood front panels project from the black frame, with drawers of different heights stacked up row by row

Source: Adrien Segal / “Snow Water Equivalent” on Adrien Segal’s official website

To get a sense of the scale of the source data, during research I re-downloaded the daily data for SNOTEL 462 (SWE and accumulated precipitation) from the NRCS Report Generator and aggregated it by water year. The values below are reference values, not the ones the artist used. The NRCS notes that values may be revised, so they may differ from what the artist obtained at the time of making.

Item (WY1980–2010, in inches)	Min	Max	Mean
Annual peak SWE	17.2 (WY1987)	76.8 (WY1995)	39.9
Water-year accumulated precipitation	31.0 (WY1987)	90.6 (WY1983)	53.7

In these reference values, the correlation coefficient between annual peak SWE and annual precipitation is about 0.94. The artist distinguishes “height = precipitation” from “size = water in the snowpack,” but as objects they are the same drawer dimension. Because the two quantities move together this closely, viewers probably take them in as nearly one quantity, or so it can be read.

If drawer height is assumed to be proportional to precipitation, the tallest drawer would be about 2.9 times the shortest. But the artist does not say whether it is proportional or whether a base height was added. The actual dimensions of the drawers have not been published either, so nothing about dimensions can be inferred from this. In the reference values, low-snow years run from 1987 to 1992, but I have not checked in the photographs how that appears as a pinch in the front panels.

Snow’s storage and human storage

This work matches the meaning of the data with the function of the object. In the Reflections section of the making-of account, Segal explains why she made it a cabinet:

The Snow Water Equivalent sculpture ultimately took the form of a cabinet to make a connection between snow as a natural device for storing water and the devices humans use for storage.

That is, she made it a cabinet to connect snow, a natural device for storing water, with the tools people use for storage. The same section also says that in dry years the drawers hold less.

One drawer is open and the box inside is visible. The front panel is a curved surface on which the stripes of the plywood layers can be seen

Source: Adrien Segal / “Snow Water Equivalent” on Adrien Segal’s official website

The artist also writes about what she hopes viewers will do.

They may ask themselves “what am I looking at and what is it telling me? Why is this drawer so small?”

In other words, viewers may ask what they are looking at and why this drawer is so small.

The smallness of a drawer is a quantity you notice when you try to put something in it, without reading any scale. The way the object is used is itself part of the encoding, or so it can be read. On the other hand, the station and the period do not appear in the work itself. The work page adds a table of station information and an explanatory text.

She explained her choice of plywood in 2017 on Data Stories #106. It can be cut into flat, complex shapes and the layers are easy to align. Solid wood expands and contracts and is prone to cracking and warping. The frame was dyed black for contrast with the front panels. It can be read as separating the surface that carries the data from the surface that supports it, by material and color.

The curved surfaces between years are not data points. They are joins made by sanding the stacked plywood by hand, so it can be read that the artist’s hand is interpolating between the 31 graphs.

How third parties saw the work

The first is a survey by researchers. “Data to Physicalization” by Djavaherpour et al. (EuroVis 2021 STAR) cites this work as an example of an approach distinctive to its maker. The section on environmental data includes this sentence:

Segal transformed the amount of water stored as snow throughout a season into furniture, where the choice of forms and materials connected back to the origins of the data

In short, she turned the amount of water stored as snow into furniture, and tied form and materials to where the data came from. In the section on the fabrication process, on the other hand, the survey says that the method of cutting and stacking layers, which includes Segal’s work, is time-consuming.

The second is a science-art blog. Michele Banks, in The Finch & Pea (2014), picked up the contrast of cold data rendered in a warm material.

Data are hard. Snow is cold. And yet artist Adrien Segal chose wood, a warm, yielding material, to visualize snowpack data, to stunning effect.

That is, data are hard and snow is cold, yet Segal chose warm, yielding wood. Banks introduces SNOTEL as data for forecasting the water supply under a changing climate.

The third is a local newspaper. Michael Armstrong of the Homer News (2014) opens with the observation that the XY graph dominates how information is shown, and stresses that this work is a real cabinet in which things can actually be stored. The article states that the volume of the drawers represents the amount of water to scale, but this cannot be confirmed from the artist’s own text. In this interview Segal says, “Can I touch that? Can I feel that? That’s the response I want.”

As a side note, on Data Stories #106 Moritz Stefaner says to Segal, “the shape is the data, which is so abstract and non tangible.” But he touched on this work itself only briefly.

Takeaways

There are three things from this work that you can take back to visualization work.

The first is to match the meaning of the data with the function of the object. Snow stores water, so she made the data into furniture that stores things. The artist writes, “The concept behind the data drives every decision.” Even for charts on a screen, you can ask whether the form and interactions behave the way the thing the data represents behaves.

The second is that aligning small multiples can become the design of a three-dimensional object. Because the 31 yearly graphs were laid out on the same baseline, they could be cut, stacked and made into a single form. How you align baselines also decides how easy comparison is on a flat page.

The third is to separate the parts that carry data from the parts that do not. Segal separated the front panels and the frame by material and color. It is also worth keeping in mind the difference between surfaces interpolated by hand and forms that come from data points. Context that cannot be carried by the object, such as the station and the period, needs to be supplied separately.

References

Primary sources

Adrien Segal, “Snow Water Equivalent,” the artist’s official website: https://www.adriensegal.com/snow-water-equivalent
Adrien Segal, “About”: https://www.adriensegal.com/about
Adrien Segal, “Press”: https://www.adriensegal.com/press
Adrien Segal, “Snow Water Equivalent,” Making with Data project page (the artist’s making-of account): https://makingwithdata.org/project/cabinet
Adrien Segal, “Snow Water Equivalent,” in S. Huron, T. Nagel, L. Oehlberg, W. Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, A K Peters/CRC Press, Part 1 “Handcraft”, Chapter 1.1, pp. 37–50. (The publisher’s page gives Copyright 2023; the Crossref registration date is 2022-11-16.) doi:10.1201/9781003264903-1 / Book page: https://www.routledge.com/Making-With-Data-Physical-Design-and-Craft-in-a-Data-Driven-World/Huron-Nagel-Oehlberg-Willett/p/book/9781032182223
Data Stories #106 “Data Sculptures with Adrien Segal” (Enrico Bertini, Moritz Stefaner, 2017): https://archive.datastori.es/episode/106
NRCS Report Generator: daily SWE and precipitation for SNOTEL 462 Ebbetts Pass (used for the reference-value aggregation): https://wcc.sc.egov.usda.gov/reportGenerator/view_csv/customSingleStationReport/daily/462:CA:SNTL%7Cid=%22%22%7Cname/1979-10-01,2010-09-30/WTEQ::value,PREC::value
dataphys.org, “2011 – Adrien Segal’s Data Furniture” (added by Yvonne Jansen; quotes the former work page): https://dataphys.org/list/data-furniture/

Secondary sources

H. Djavaherpour, F. Samavati, A. Mahdavi-Amiri, F. Yazdanbakhsh, S. Huron, R. Levy, Y. Jansen, L. Oehlberg, “Data to Physicalization: A Survey of the Physical Rendering Process,” EuroVis 2021 STAR, Computer Graphics Forum 40(3): https://arxiv.org/abs/2102.11175
Michele Banks, “The Art of Science: Cold, Hard Data in Warm, Soft Wood,” The Finch & Pea, 2014-05-14: https://thefinchandpea.com/2014/05/14/the-art-of-science-cold-hard-data-in-warm-soft-wood/
Michael Armstrong, “From Science to Sculpture,” Homer News, 2014-05-07: https://www.homernews.com/2014/05/07/from-science-to-sculpture/
Wikipedia, “Adrien Segal”: https://en.wikipedia.org/wiki/Adrien_Segal

Eating Local Data as Local Food: Data Cuisine by Moritz Stefaner and Susanne Jaschko

Fri, 06 Nov 2026 00:00:00 +0900

Data Cuisine is a workshop that expresses a place’s open data through that place’s food. The first one was held in Helsinki in September 2012, and after Barcelona in 2014 it was held in many other places. In the dishes, data was mapped onto the amount of sugar, the size of a dish, and combinations of ingredients. Spain’s cut to its science budget became a cake with 34% less sugar. The destinations of Spain’s young emigrants became a plate of fish dishes, each evoking its country, served in portions sized to match the ranking.

The project was planned by the visualization designer Moritz Stefaner and the Berlin-based curator Susanne Jaschko. What follows draws on the organizers’ announcement pages, an interview with the two (We Make Money Not Art, 2014), and reports in WIRED, Core77, and Engadget. The official site, data-cuisine.net, would not open from my environment, so I could not read the official descriptions of individual dishes. I write about the contents of the dishes only to the extent I could confirm them in the interview and the press coverage.

2012: an open-data cooking workshop in Helsinki

The first event was called “The Art of Data Cuisine: The Open Data Cooking Workshop.” It was held on 15-16 September 2012, produced by Pixelache in Helsinki and organized by Jaschko’s prozessagenten. It was timed to coincide with the Open Knowledge Festival, held the same month. Pixelache’s announcement poses these questions.

“Have you ever tried to imagine how a fish soup tastes whose recipe is based on publicly available local fishing data? Or what a pizza would be like if it was based on Helsinki’s population mix?”

In other words: what would a fish soup made from local fishing data taste like, or a pizza made from Helsinki’s population mix? The description of the work at the 2016 Information is Beautiful Awards states the rule: express “local open data in local food” through properties such as color, shape, texture, smell, taste, nutrition, and origin.

In Helsinki there was a public tasting on the evening of the second day, and the chef was Antti Nurkka. The second event was held in Barcelona in June 2014, co-produced by the CCCB (Centre de Cultura Contemporània de Barcelona) and Sónar+D. For the first two days participants researched topics and data, and in the second half they cooked the dishes with chefs in a professional kitchen.

The number of host cities differs depending on the source and the date.

Source	Date	Host cities
WIRED, Core77, We Make Money Not Art	2014	Two events: Helsinki and Barcelona
Information is Beautiful Awards description	2016	“Helsinki, Barcelona, Leeuwarden, Basel, Berlin, Gembloux, Belgium and Boston”

For the events in the 2016 list other than Helsinki and Barcelona, I could not confirm the years or the dishes. In the same year, Data Cuisine won Gold in the Dataviz Project category of the Information is Beautiful Awards. Stefaner also gave a talk on the project at eyeo 2016 (video).

Expressing numbers through sugar and the size of a dish

Requiem for Science in Barcelona is a dish by Antonija Kuzmanic. It expressed the cut to Spain’s science budget with two cakes.

Two round cakes sit left and right on a black slate plate, with an atom symbol drawn in powdered sugar between them

Source: Data Cuisine / We Make Money Not Art

The reports differ on what the two cakes were compared on.

Source	What was compared
WIRED (2014-07-24)	A cake made from an ordinary recipe and a cake with 34% less sugar. The 34% is the size of this year’s cut to Spain’s science budget
Engadget (2014-07-24)	The cake for the 2013 budget has 34% less sugar than the cake for the 2005 budget

Both map the proportion of the cut onto the amount of sugar.

Emigration Fish is a dish by Samuel Boucher and Jahn Schlosser. Its subject was the countries to which many young Spaniards had emigrated. According to WIRED, the fish was prepared differently for each country, and the size was set to match the ranking.

Two white square plates. On the front plate, a red fish fillet topped with three lemon slices; on the back plate, fish prepared another way

Source: Data Cuisine / We Make Money Not Art

France is fish with a wine sauce, Germany parsley and beer, the United Kingdom battered fish, Ecuador ceviche, and the United States fish fried in bacon fat. WIRED writes of the top six countries, but only five dishes are listed.

First Date Noodles takes as its subject whether people have sex on a first date. According to Core77, in an informal survey of Facebook acquaintances, 59% of women and 86% of men answered yes. Pink and blue noodles stand for gender; the tangled lump is the people who said yes, and the noodles sticking out are those who said no.

On a black slate plate, pink and blue noodles lie in straight lines on the left, and a tangled lump of pink and blue noodles sits on the right. A hand is arranging them

Source: Moritz Stefaner / Truth & Beauty

Stefaner’s page does not label this photo with a dish name, but it can be read as the same form that Core77 describes for First Date Noodles. The correspondences I could confirm are summarized below.

Dish	Data	How it is expressed in food	Source
Requiem for Science	Rate of cut to the science budget	Difference in sugar between two cakes (34%)	WIRED, Engadget
Emigration Fish	Countries where young people emigrate	Cooking method of the fish, evoking the country	WIRED
Emigration Fish	Ranking of destinations	Size of the fish dish	WIRED
First Date Noodles	Gender	Color of the noodles (pink and blue)	Core77
First Date Noodles	Share of “yes” and “no”	Tangled lump of noodles and noodles sticking out	Core77
Sugarsm	Methods named in a survey, and their shares	One lump of sugar per method, in that quantity	WIRED caption
Kippis!	Alcohol consumption by region of Finland	Shape of the map of Finland	Core77
Alcohol consumption map (Helsinki, dish name not given)	Alcohol consumption within the country	Wine in varying amounts, and regional dishes	Engadget
Name unconfirmed (Helsinki)	Types of crime	Different kinds of fish stacked up	Engadget

Quantity is mapped to the amount of sugar or noodles, category to cooking method or color, and rank to size. This can be read as a search for the food-world equivalents of length, color, and area in a diagram on paper.

Bad taste and culinary associations can also be codes

Some dishes deliberately use bad taste. Jaschko answers as follows in We Make Money Not Art.

“Creating a dish that tastes terrible is sometimes the best way to communicate a negative development or a problematic situation.”

In other words, a dish that tastes bad can sometimes be the best way to convey a negative change or a problem. Her examples are Suicide Cocktail, which dealt with alcohol consumption and suicide rates in Finland, and Unemployed Pan con Tomate!, which expresses the sharp rise in youth unemployment. The sources I could read did not say which ingredients were changed or how. Stefaner also gives as an example of a dish that need not taste good one that expressed noise with salt. According to Core77, it is Take it with a Pinch of Salt!, which dealt with a day of noise on Barcelona’s streets.

Stefaner told WIRED what can be used in food: three-dimensional form, sweet, sour, salty, bitter, umami, and spiciness, texture, the cultural associations of ingredients and dishes, cooking conditions, serving temperature, and plating. The country-by-country cooking methods of Emigration Fish can be read as falling under “cultural connotations.”

On how the dishes are made, Jaschko says:

“In most cases, a ‘data’ dish is created by either remixing, altering or re-interpreting existing recipes.”

That is, a data dish is usually made by remixing, altering, or reinterpreting existing recipes. Chefs encourage participants to try other techniques and ingredients, and also take part in the plating. The data mapping is built not as a wholly new dish but as a deviation from a familiar recipe, as I read it.

“Twice as sweet” is hard to tell apart, says its creator

On how precisely taste can convey information, Stefaner answers in We Make Money Not Art:

“It is simply much harder to determine what is “twice as sweet” as opposed to as twice as long line in a graphic. Then again, taste is a much more emotional and temporally complex experience that just looking at a dot on a screen. So, the mechanisms to encode information might be more fuzzy, but potentially much deeper.”

That is, telling “twice as sweet” is much harder than telling a line twice as long. On the other hand, taste is a more emotional experience than a dot on a screen, and one that changes over time. The code may be vague, but it may reach deeper. In the same answer he also says that reading the data accurately is not the primary concern.

Jaschko says that making a visual diagram out of food is relatively easy, and that the aim is a dish that communicates through both appearance and taste. But she also says that there are so many possibilities that most participants settle on something simple.

Stefaner describes what he found interesting from the visualization side:

“From a data visualization point of view, I found it really interesting to watch how deeply people meditate on very simple data points, when they think about turning them into food experiences. In a way, this is a very needed counterpoint to the current trend of consuming lots of data in a very quick and superficial way.”

That is, when people try to turn data into a dish, they think hard about very simple data points, which is a needed counterpoint to the trend of consuming data quickly and superficially. The number Requiem for Science dealt with was a single figure, 34%. Its weight seems to have lain less in making people read accurately than in creating time to keep thinking about one number.

On differences between places, Jaschko says that in Helsinki many participants were immigrants or travelers, and there was less local cuisine. The rule of “using the local food” can be read as having been followed differently depending on the participants.

How third parties saw the workshop

The assessments each point to a different aspect. The first is the view that quantity reaches people as taste. WIRED’s Kyle Vanhemert wrote of the science-budget cake:

“the dish turned an abstract budget cut into something much more palpable—the drop in funding literally left a bad taste in your mouth.”

That is, an abstract budget cut became something tangible. The same article, however, mocks the choice of fish fried in bacon fat for the United States: “Ha, ha, Euros. We get it. Americans are fat.” This can be read as pointing out that when cultural associations are used as codes, stereotypes land on the plate as they are.

The second is the lack of precision. Core77 wrote that, through the “geotagging” of regional taste, eaters can taste the differences between data points. It then calls Data Cuisine a “not entirely precise way to present information.”

The third is its position as a code that uses the five senses. In 2014, Andy Kirk of Visualising Data introduced the Barcelona event as “a truly multi-sensory approach to encoding data.” In 2016 a CHI extended abstract, “Data Edibilization: Representing Data with Food” by Yun Wang et al., also appeared, but I have not read its full text.

Takeaways

There are four judgments that can be taken back into visualization work.

The first is to recognize situations where accuracy is not the goal. Stefaner admits that “twice as sweet” is hard to tell apart, and said that readability is not the primary concern. A chart that demands precise comparison and a chart that makes people think about a single number call for different designs, as I read it.

The second is to narrow down the numbers. Requiem for Science used 34%, and First Date Noodles used 59% and 86%. The fewer the numbers, the more time the audience seems to have to think.

The third is to separate which property is used for quantity and which for category. Quantity was the amount of sugar or noodles, category the cooking method or color, and rank the size. Even when the material changes to food, the same thinking as in a paper diagram can be used.

The fourth is the price of cultural association. If countries are expressed by cooking methods, stereotypes enter the code, as WIRED mocked. When you use associations, you need to decide on the view they carry as well.

A work by the same Stefaner that uses physical material is Perpetual Plastic. An example that leaves the rendering to sound is Chemo Singing Bowl.

References

Primary sources

Moritz Stefaner, “Data Cuisine,” Truth & Beauty (project page). Page / Markdown version
Pixelache, “The Art of Data Cuisine: The Open Data Cooking Workshop,” 2012-09-15 to 16. Announcement page
CCCB, “Sonar+D: Data Cuisine Workshop with Moritz Stefaner and Susanne Jaschko,” 2014-06-10 to 13. Announcement page
Information is Beautiful Awards, “Data Cuisine” (description of the work, list of host cities). Showcase page
Truth & Beauty, Information is Beautiful Awards 2016. Page
Truth & Beauty, talk at eyeo 2016 (2016-06-10). Page / Video (Vimeo)
Truth & Beauty, 1 2 3 data (Fondation EDF, Paris, 2018). Page
Truth & Beauty, Data Cuisine exhibited in Zürich (Museum für Gestaltung Zürich, 2019). Page
Régine Debatty, “Data Cuisine, food as data expression,” We Make Money Not Art, 2014-07-02 (interview with Stefaner and Jaschko). Article
Data Cuisine official site (would not open from my environment; text not checked). data-cuisine.net

Secondary sources

Kyle Vanhemert, “These Weird Meals Are Actually Clever Data Visualizations,” WIRED, 2014-07-24. Article
“Data Cuisine Lets You Have Your Data and Eat It, Too,” Core77, 2014-06-30. Article
Sarah Silbert, “Data Cuisine creates meals based on cultural statistics,” Engadget, 2014-07-24. Article
Andy Kirk, “Tasty visualisations from the Barcelona Data Cuisine,” Visualising Data, 2014-06-19. Article
Yun Wang, Xiaojuan Ma, Qiong Luo, Huamin Qu, “Data Edibilization: Representing Data with Food,” CHI EA 2016 (full text not checked). doi:10.1145/2851581.2892570

One Quality Issue, One Block: The GM and WellStar LEGO Board

Sun, 01 Nov 2026 00:00:00 +0900

General Motors’ (GM) LEGO board turns each quality issue into one LEGO block. The color of the block shows the part of the vehicle, its size shows the severity of the problem, and its position on the board shows the stage the issue has reached, from root cause to solution to outcome. Each block carries an identification number and a date of discovery. GM and WellStar Health System, a hospital group in Georgia, built it together and announced it on March 26, 2012. The two companies call the method “3-D Visualization.”

The main source is GM’s press release (March 26, 2012). The original page is now gone, so I read the text from the full reprint in the Naylor Network, an industry association newsletter. ThoughtWorks’ P2 Magazine (2013) describes the same board in a way that differs from the release, so I treat the release’s account and P2’s account in separate sections. I could not check the text of the Fast Company article that the catalog dataphys.org cites as its source.

March 2012: a car maker and a hospital announce the same tool

According to the release, the collaboration between the two organizations began in a hospital emergency department. GM’s crossover vehicle team, as part of leadership development, invited WellStar to talk about triage in the emergency department. GM then donated a mock assembly line to WellStar’s training center so that doctors and nurses could learn lean methods.

WellStar is a nonprofit health system of five hospitals, headquartered in Marietta, Georgia. The board came out of a brainstorm between GM chief engineer Tim Herrick and Dennis Pastor, who leads process improvement at WellStar. Pastor used to work at GM, where he was a production superintendent for assembly at the Lordstown plant in Ohio. The release says the prototype took one week.

A man in glasses places a block on a black LEGO baseplate set in a wooden frame. In the foreground, yellow and orange blocks with white labels are stacked in many layers.

Probably a GM/WellStar press photo (photographer unknown). From: List of Physical Visualizations (dataphys.org)

The release lists these uses.

Organization	Uses (as described in the release)
GM	Tracking problems that lead to warranty repairs. Also being rolled out worldwide for tracking design changes and budgets. A digital version has also been developed
WellStar	On-time starts at clinics, the foundation’s five-year fundraising plan, healthcare quality metrics, and improving the process for enrolling physicians with insurers (expected to save $1 million a year from this one board)

The release says the effect on the GM side is that the time to implement changes that prevent future warranty repairs “could” be cut by 33%. The original wording is “could reduce,” so this is an expectation, not a measurement.

One issue, one block, with color, size, and position assigned

The release explains how the board works in a single paragraph.

“3-D Visualization builds on GM’s Problem Resolution Tracking System. If a transmission case breaks on a durability test vehicle, a problem resolution report documents the problem, and its corresponding LEGO block goes on a LEGO board. The block color identifies the area on the vehicle and the block size denotes severity; the bigger the block, the bigger the problem. Each block has an identification number and date of discovery, and the board shows its progress from root cause to solution to outcome. 3-D Visualization is applicable to any process that has volume and aging.”

The foundation is GM’s problem resolution tracking system. If the transmission case of a durability test vehicle breaks, one report is written and one corresponding block is placed on the board. Color is the part of the vehicle and size is severity: the bigger the block, the bigger the problem. Each block carries an identification number and a date of discovery, and the board shows progress from cause to solution to outcome. The paragraph closes by saying the method applies to any process that has volume and aging.

Data	How the board shows it	Data type
One issue (one report)	One LEGO block	Unit (one issue = one block)
Part of the vehicle where the problem occurred	Block color	Categorical
Severity of the problem	Block size	Ordinal
Identity of the problem	Identification number on the block	Nominal (text)
Date of discovery	Date on the block	Time (text)
Stage: cause → solution → outcome	Position on the board	Ordinal (process stage)
Resolved	Block removed from the board	State (based on the remark in the next section)

Some things are not stated in the public sources. I could not find the table mapping colors to parts, the number of size levels, the dimensions of the board, or the total number of blocks.

On a wooden-framed board, yellow, orange, blue, and green blocks with white labels stand in columns of different heights. A man in the background handles a block.

Probably a GM/WellStar press photo (photographer unknown). From: Torque News

In the press photos, you can see white labels on the blocks with printed numbers. This can be read as the identification numbers the release mentions. Many of the blocks are stacked vertically into columns of different heights. The release does not say whether its “size” means the footprint on the plane or the height of the stack.

The volume and aging in the last sentence can be read as a hint for how to read this table. The number of blocks tells you how many problems there are. Position and date of discovery tell you which problem is at which stage and since when it has been stuck there. It is an assignment designed to show volume and stagnation on one board.

One block on a board, not one line in a spreadsheet

Jamie Pickett, a GM senior manager in charge of the quality of current vehicles, says this in the release.

“3-D Visualization allows us to look at issues more holistically. By visually representing a problem it is easier to see the whole scope of it and where the opportunities lie. And, unlike a line of data in a spread sheet, seeing a problem as a block on a board is a strong motivator for finding a solution to get it off the board.”

In essence: it lets people see issues as a whole, makes the scope and the places to act easier to see, and, unlike a line in a spreadsheet, a block on a board is a strong motivation to solve the problem and take it off.

The comparison is with a line in a spreadsheet. Unlike a table row, a block stays on the board until it is removed. From Pickett’s words, taking the block off the board can be read as the very goal of resolution. It can be read as an encoding that values surveying the volume and getting people to act over reading precise values.

Pastor is also reported to have told Fast Company that the process was three-dimensional while the reports were two-dimensional. However, this is secondhand, quoted by SmartData Collective from Fast Company, and I could not check the original. Nor did the board replace screens: GM also built a digital version.

The catalog’s description needs a correction. dataphys.org introduces this board as “LEGOs help business executives log and explore data.” But the release does not use the word “executives,” and the users it names are a chief engineer and a quality manager. It is closer to the release to read it as a tool for engineering and quality teams.

ThoughtWorks’ article reads it by rows, columns, and height

The article “3D Card Walls” in ThoughtWorks’ P2 Magazine Issue 4 (October 2013), by Nick Thorpe and Paige Mulholland, describes the board differently from the release.

“On the General Motors LEGO board, each block represents a car which needed repairs under warranty. The rows represent the faulty area of the vehicle, and there is a separate section of the board for issues which have been contained or solved on the assembly line. The height of the block itself represents the severity of the problem, and the board also uses columns to denote the time it took for the fault to appear, and colours and labels to further define each block.”

One block is one car that needed warranty repair. Rows are the faulty part, the height of the block is severity, and columns are the time until the fault appeared. There is a separate section for problems contained on the assembly line, and colors and labels are used to distinguish blocks further.

Item	GM release (2012)	P2 Magazine (2013)
Unit of one block	One issue (one report)	One car that needed warranty repair
Part of the vehicle	Block color	Board rows (color is auxiliary)
Severity	Block size	Block height
Time	Date of discovery on the block	Board columns (time until the fault appeared)
State	Position moving from cause to solution to outcome	Separate section for contained or solved problems

The P2 article does not give the basis for this reading. It is unclear whether it was taken from photos or from interviews. It may also be looking at a board from a different period or for a different use. It is possible that the release’s “size” was actually the height of the stack, but there is no source to check this. In this article I take the release, the account of the people who built it, as the main axis, and treat P2 as one third-party reading.

How third parties saw this board

Assessments differ depending on what they use as a clue.

Seeing it as an extension of kanban. The P2 article opens by saying that even with card size, color, position, and shape, there are cases where a card wall cannot express all the variables. Thorpe says of the GM board, “The problem they’re solving here is ‘how do we decide what should we start next?’” The reading is that it is a board for deciding what to take up next.
Moved to another setting, it did not work well. In a paper at EIAH 2015, Daniel K. Schneider of the University of Geneva writes that, inspired by the Fast Company article, he built a structured LEGO board showing the progress and seriousness of doctoral theses. The result: “Cette première variante n’a eu que peu de succès puisque, par définition, le LEGO appelle à « jouer ».” In other words, the first version had little success because LEGO invites people to play.
Valuing the three-dimensionality itself. Erica Driver of SmartData Collective writes “Humans think in 3D.” and extends it to making software BI three-dimensional and physical.

Automotive media treated it lightly. The Truth About Cars wrote, a little wryly, “Problem solving becomes child’s play.” Torque News cites being able to move it by hand as an advantage: “It can also be touched and felt and moved by hand.”

Takeaways

What you can take from this board lies less in whether you build it from physical objects than in how you decide the encoding.

First, make one issue one block. You can tell the volume just by counting, and the pile-up is visible as it is, without writing a total separately.

Second, assign process stages to position. If you divide the board into sections, you can see at which stage blocks are piling up, and together with the date of discovery you can follow how long they have been stuck.

Third, decide the exit in advance. “Remove from the board” is a design in which the end of the display coincides with the end of the work. Even on a screen dashboard, you can make a place where resolved items disappear.

Fourth, keep a legend. The color-to-part table is not public, and P2, looking from outside, wrote a different account from the release. The more a board is shared, the more its encoding conventions need to be written down and attached.

Schneider’s report says that because LEGO invites play, the first version built with a fixed encoding did not go well. It can be read as an example of a fixed encoding colliding with free building. The GM board worked, it can be read, because it was built on top of an existing system, the report, and the one-issue-one-block convention was already settled on the business side.

References

Primary sources

General Motors, press release, 2012-03-26 (original URL; it now redirects to news.gm.com and the text no longer remains). http://media.gm.com/media/us/en/gm/news.detail.html/content/Pages/news/us/en/2012/Mar/0326_lego
Full reprint of the same release: “GM, WellStar Develop LEGO-Based Visual Management Tool,” Naylor Network. https://www.naylornetwork.com/naf-nwl/articles/print-v2.asp?aid=171511
Pierre Dragicevic (information provided by Fanny Chevalier), “2012 – General Motors’ 3D LEGO Visualizations,” List of Physical Visualizations (dataphys.org). https://dataphys.org/list/general-motors-3d-lego-visualizations/
Mark Wilson, “How GM Is Saving Cash Using Legos As A Data Viz Tool,” Fast Company / Co.Design, 2012 (text not checked because of bot blocking). https://www.fastcompany.com/1669468/how-gm-is-saving-cash-using-legos-as-a-data-viz-tool
The Daily Drive, “GM Plays with Legos (Video),” 2012 (not checked; the domain is gone). http://cgdailydrive.com/gm-plays-with-legos/

Secondary sources

Nick Thorpe & Paige Mulholland, “3D Card Walls,” P2 Magazine Issue 4, ThoughtWorks, 2013-10. https://thoughtworks.github.io/p2/issue04/3d-cardwall
Daniel K. Schneider, “Visualisation de projets de thèse avec un dispositif physique et constructif,” EIAH 2015. https://tecfa.unige.ch/tecfa/talks/schneide/eiah-2015/Schneider_EIAH2015-lego-final.pdf (related wiki: https://edutechwiki.unige.ch/en/Lego-compatible_thesis_project_board )
Erica Driver, “Visualizing Data in 3D with Lego,” SmartData Collective, 2012-05-09 (includes secondhand quotation of Pastor’s remark). https://www.smartdatacollective.com/visualizing-data-3d-lego/
Bertel Schmitt, “Latest GM Invention: Management By LEGO,” The Truth About Cars, 2012-03-26. https://www.thetruthaboutcars.com/2012/03/latest-gm-invention-management-by-lego/
John Goreham, “GM’s adaptation of Legos could help protect your car from problems,” Torque News, 2012-03-27. https://www.torquenews.com/1083/gms-adaptation-childs-toy-could-help-protect-your-car-problems
Jason Siu, “General Motors Develops Lego Based Visual Management Tool,” AutoGuide, 2012-03-26. https://www.autoguide.com/auto-news/2012/03/gm-and-wellstar-develop-lego-based-visual-management-tool.html

Carving Tweet Counts into Height, Projecting Topic Heatmaps: emoto

Sun, 01 Nov 2026 00:00:00 +0900

The emoto data sculpture takes English-language tweets about the London 2012 Olympics, counts them by hour and sentiment level, and carves those counts into height. Each day of the Games is one plate, and 17 plates stand side by side. Topic-specific heatmaps are projected on top. With a knob in hand, visitors switch topics and move through time.

It was made by three parties: Moritz Stefaner, Drew Hemment, and Studio NAND. What follows draws on the official site emoto2012.org, the paper Hemment wrote for ISEA2013, Stefaner’s page at Truth & Beauty, and The Creators Project’s interview with the makers.

12.5 Million Tweets, Shown in Three Forms

emoto was commissioned as a FutureEverything project for the Cultural Olympiad and the London 2012 Festival. Hemment conceived it and was project director, and Stefaner was lead designer. Studio NAND (Stephan Thiel, Steffen Fiedler, Jonas Loh) led the design of the physical sculpture.

The data are English-language tweets collected through Twitter’s public API from the opening ceremony on July 27, 2012, to the closing ceremony on August 12. The count is “12.5 million” in Hemment’s paper and “12 million English tweets” on the official site. Lexalytics’ Salience Engine gave each tweet a sentiment score, which was sorted into equal-width bins from −6 (very negative) to +6 (very positive). Topics (events, athletes, countries, and so on) were separated using a predefined list.

Hemment describes three ways of presenting the work.

Presentation was in three forms: a real-time sentiment visualisation web application, data journalistic analysis of the Games, and a physical interactive data sculpture.

That is, a real-time web application, data-journalism analysis, and a physical, interactive data sculpture.

A flow diagram: photos of the Games lead to tweets and sentiment classification, then split into three outputs: real-time visualization, data journalism, and the data sculpture

Source: Moritz Stefaner / Truth & Beauty, “emoto”

The web version showed the Games as they happened. The sculpture was made after the Games ended and was first exhibited at WePlay in Preston (part of the London 2012 Festival), September 7-9, 2012. In 2013 it also appeared in the Information in Style exhibition at the CAFA Art Museum in Beijing.

Blue-lit milled plates arranged on a dark platform. In the foreground the text DAY 3, 15:00, DAY (the right end is cut off) is visible, and a long horizontal sentigraph hangs on the wall behind

Source: Moritz Stefaner, Drew Hemment, Studio NAND / Truth & Beauty, “emoto” (the same photo appears as Fig. 2, “emoto data sculpture and sentigraph,” in the ISEA2013 paper, CC BY-NC-SA 3.0)

Acquiring the data nearly fell apart just before launch. According to the paper, the team had expected to use an extended version of the Firehose, which streams all tweets, through a data reseller. When Twitter tightened its terms, the reseller told them shortly before launch that neither the data nor any derived insight could be displayed publicly in any form. The team rebuilt its infrastructure in two weeks and switched to the public API, which covers about 1% of the roughly 400 million tweets sent per day.

Hemment also writes about the limits of this data. It covers only English-language tweets, and more than 65% of the web version’s users came from four countries: the US, UK, Canada, and Australia.

Such a limited perspective does not represent ‘what the world thinks.’

In other words, a limited perspective does not represent what the world thinks. He adds that evaluating the accuracy of the sentiment analysis was outside the project’s scope, and that mix-ups occurred, such as the BBC receiving a negative score when it reported bad news.

Carving a Gray Heatmap into Height

Stefaner describes concretely in an interview how the sculpture was made.

The basis for its creation process are 2D-graphics, basically gray-tone ‘heat-maps,’ which show the aggregated number of tweets per hour for each sentiment-level in a small rectangle. The lighter the rectangle, the higher the number of tweets for this hour and sentiment-level and the higher the elevation of the sculpture at this point. This results in the rollercoster-like bands you see in the sculpture.

The starting point is a flat gray heatmap. The tweet count for each hour and each sentiment level is summarized in a small rectangle. The lighter the rectangle, the higher the count, and the higher the sculpture at that point. The result is the roller-coaster-like bands.

What the flat heatmap carried in the lightness of its color, the sculpture carries in height. In the words of Studio NAND’s Stephan Thiel, the height is “a third dimension in addition to time and sentiment-score.” Time and sentiment level remain as positions on the plate. Stefaner also says the sculpture contains “all tweets we have collected for emoto.”

The paper and other sources say the height shows “emotional highs and lows.” This can be read as a metaphor. By the description of how it was made, height is the count, and sentiment is the position of the band.

The material is Chemiwood, a polyurethane foam, cut by CNC. According to Thiel, they preferred the traces left by the cutting head to the clean surface of a 3D print. The surface is finished with a two-component paint in a matte gray suited to projection.

Combining the makers’ descriptions with my own reading, the correspondence between data, appearance, and material is as follows.

Data / variable	Physical property	Specifics / notes
One day of the Games	One plate	17 plates side by side
Time of day	Horizontal position on the plate (my reading)	Hourly. 17 plates form the time axis of the whole Games
Sentiment level	Vertical position on the plate (the bands, my reading)	Bins from −6 to +6
Tweet count for that hour and level	Height of the carved surface	The lighter in the original heatmap, the higher
Tweets on all topics	The shape of the sculpture itself	All collected tweets
Stories by topic	Projection onto the sculpture (heatmap)	Switched with the knob
Most retweeted tweet of that hour	Text shown beside the sculpture	Chosen from the hour under the cursor
Material and finish	Chemiwood cut by CNC, matte gray paint	Cutting marks left visible

How many sentiment steps there were (and whether 0 was included) could not be confirmed from the primary sources.

Projecting Topics on Top, Switching with a Knob

The shape of the sculpture is determined by all the tweets. Differences between topics are carried by the light projected on top. According to the official site, heatmaps for themes found during the Games were projected, and visitors moved between themes with a controller and could “explore our archive.”

Blue-green light projected onto the milled plates, with tweets cheering on Tom Daley shown as text behind the plates

Source: Moritz Stefaner, Drew Hemment, Studio NAND / Drew Hemment, “emoto – Visualising the online response to London 2012” (ISEA2013), Fig. 6 (CC BY-NC-SA 3.0)

According to Stefaner, many topics were grouped into “high-level stories.” The Team GB story, for example, includes topics about individual athletes. To choose them, he ran a “data-driven ‘hunt’” with the sentigraph, keeping both topics that showed up in the data as they did in real life and amusing data artifacts such as One Direction.

The interaction is a single rotating knob, a Griffin PowerMate. In Thiel’s description, pressing it switches the theme and turning it moves the cursor along the timeline. The most retweeted tweet of the hour under the cursor appears beside the sculpture.

The tweets and topics shown in the exhibition can also be explored in the interactive heatmap on Stefaner’s site. Horizontal bands are lined up by topic, and hovering over a row shows the most retweeted tweet at that time.

A figure titled “Olympic sentiment heatmaps.” Horizontal heatmap bands are stacked vertically by topic: all topics, Team GB, Phelps/Lochte, Tom Daley, and others

Source: Moritz Stefaner / well-formed-data.net, “emoto data sculpture”

This can be read as a division of labor: form fixes the whole, and light shows the parts. Switching the projection does not change the relief beneath. On the same platform, you can compare where one topic’s surge falls within the volume of the whole Games. Whether the color of the projection stood for positive or negative could not be confirmed from the primary sources.

What Changed by Making It a Solid

Stefaner describes the sculpture’s place in the project this way.

In many ways the crowning piece of the project, and a conceptual counterpoint to the ephemeral web activities, our data sculpture preserved the more than 12 million tweets in physical form.

That is, it was the crowning piece of the project, a conceptual counterpoint to ephemeral web activity, preserving more than 12 million tweets in physical form.

Thiel lists several advantages of the three-dimensional form.

Creating a three-dimensional sculpture helped people understand the displayed dimensions in the data more quickly and naturally. … This is often more immediate than encoding multiple variables in two-dimensional space.

Visitors grasped the dimensions of the data faster and more naturally, and this is often more immediate than packing several variables into a plane. He also names the ability to emphasize quantity through physical size and the way people can look and talk together, and says “if all the lights are off, it’s still an object which represents information.”

Thiel also writes about the difficulty. On a screen the view can be rearranged freely, but a large sculpture constrains how you can work with it.

If you have a massive sculpture, this sets a limitation on how you can work with it, also in terms of interactivity

The projection and the knob can be read as answers to this constraint. The carved form is made once and cannot change. What needed to change was left to light and text.

According to a note in the paper, the team argued to Twitter that the carved piece was “far removed from the data stream” and could not be used to reconstruct the original information (the negotiation did not succeed).

Andy Kirk’s article before the opening said it would be carved into “19 plates.” The finished work has 17, and I found no source explaining the change.

How Third Parties Saw the Work

On July 19, 2012, before the Games opened, Andy Kirk introduced emoto on Visualising Data as “One of the most ambitious, unique, multi-faceted and fundamentally exciting projects.” This was written before the work was finished.

In an article dated September 20, 2012, Julia Kaganskiy of The Creators Project summarized the sculpture as follows.

Peaks and valleys are etched into the plates to indicate the volume of tweets per hour, and a heat map of color gradations is projected on top to show the public’s feelings towards a particular topic

The relief of the plates shows the volume per hour, and the gradations of projected color show feelings toward a topic. The same article calls the sculpture “9.5 meter long,” but on the official site and in the paper, 9.5 m (9 m in the paper) is the length of the wall sentigraph. The sculpture’s dimensions could not be confirmed from the primary sources.

At the Information is Beautiful Awards 2013, the installation was among the entries judged, and its showcase page carries the labels Bronze and Data visualization. The description calls the sculpture “its core component.”

Takeaways

First, how to translate a flat heatmap into a solid. Keep the two axes as positions and move only the count, which the lightness of color used to carry, into height. Because it maps one-to-one onto the original figure, it is harder to mistake what became height.

Second, separate what stays fixed from what changes. emoto carved a form determined by all tweets and layered topic differences on top with light. Leaving the whole to the material and the parts to projection and text allows many ways of viewing on a single platform.

Third, give each medium its own role. The web version streamed the Games as they happened, the sentigraph told the analysis, and the sculpture preserved the Games after they ended.

Finally, write down the bias in the data outside the work. From the relief alone you cannot tell that it is a 1% English-language sample. A record like Hemment’s paper becomes the premise for reading the work.

References

Primary sources

emoto official site: https://emoto2012.org/
Drew Hemment, “emoto – Visualising the online response to London 2012,” ISEA2013 Proceedings (PDF): https://emoto2012.org/downloads/emoto_published.pdf (submitted version: https://www.isea-symposium-archives.org/wp-content/uploads/2021/07/ISEA2013_108_DrewHemment-Emotovisualising.pdf , bibliographic record: https://www.research.ed.ac.uk/en/publications/emoto-visualising-the-online-response-to-london-2012/ )
Moritz Stefaner, Truth & Beauty, “emoto”: https://truth-and-beauty.net/projects/emoto (Markdown version: https://truth-and-beauty.net/projects/emoto.md )
Truth & Beauty, interactive heatmap: https://truth-and-beauty.net/assets/extras/emoto/heatmap/
Moritz Stefaner, well-formed-data.net, “emoto” (2012-07-31) and “emoto data sculpture” (2012-09-20): https://well-formed-data.net/archives/tag/emoto/
Moritz Stefaner, Drew Hemment, Gerrit Kaiser, talk slides (Strata London, October 2012): https://speakerdeck.com/moritzstefaner/emoto-visualising-the-online-response-to-london-2012
Studio NAND, “emoto — The interface explained” (Vimeo, 2012-07-30): https://vimeo.com/46615102
Lancaster EPrints bibliographic record: https://eprints.lancs.ac.uk/id/eprint/61698
Interview with the makers (remarks by Stefaner and Thiel), The Creators Project / VICE: https://www.vice.com/en/article/data-visualization-sculpture-shows-twitter-reactions-to-london-2012-olympic-games/
dataphys.org, “Emoto: Projection Augmented Heatmaps of Twitter Data”: https://dataphys.org/list/emoto-projection-augmented-heatmaps-of-twitter-data/

Secondary sources

Andy Kirk, “’emoto’: Visualising the emotional response to London 2012,” Visualising Data, 2012-07-19: https://visualisingdata.com/2012/07/emoto-visualising-the-emotional-response-to-london-2012/
Julia Kaganskiy, “Data Visualization Sculpture Shows Twitter Reactions To London 2012 Olympic Games,” The Creators Project / VICE, 2012-09-20: https://www.vice.com/en/article/data-visualization-sculpture-shows-twitter-reactions-to-london-2012-olympic-games/
Information is Beautiful Awards 2013, “emoto installation”: https://www.informationisbeautifulawards.com/showcase/68-emoto-installation

One Row a Day, a New Yarn Every 5°C: The Temperature Scarf and Blanket

Sun, 01 Nov 2026 00:00:00 +0900

The Temperature Scarf is a year-long time series on a stepped color scale. “One day = one row” gives the length, and “the day’s high temperature, binned in 5°C steps” gives the color. Knit one row a day, and by the end of the year you are left with 365 stripes. The order of the stripes is the date, and the color of each stripe is the temperature class. The color card that serves as the legend is made by each knitter to fit the temperature range of their own region.

The original is a knitting pattern that Kristen Cooper, who lives in northern British Columbia, Canada, published on her blog “Honey Nutbrown’s” on January 2, 2013. dataphys.org lists it as “2013 – Temperature Scarves and Afghans”. What follows draws on the text, images, and comments of Kristen’s blog, and on a post by Arlene Cline, who made one of the earliest blanket (afghan) versions.

January 2, 2013: A “Conceptual Knitting Pattern”

Kristen’s post, “My Year in Temperatures -Scarf-”, begins by defining the pattern.

“This is a conceptual knitting pattern that will look different for everyone who knits it. In this pattern, you record the day’s highest temperature by knitting one row in the colour designated for each temperature. At the end of the year, you will have a visual colourful graph of the temperatures of your area for a whole year.”

She calls it a “conceptual knitting pattern” that looks different for every knitter. You record the day’s high temperature by knitting one row in an assigned color. After a year, you have a colored graph of your region’s temperatures. Kristen herself calls it a “graph”.

The instructions are short. Cast on 35 stitches in white and knit 5 rows in garter stitch. After that, knit one row a day in the color for that day’s temperature. At the start of each month you may knit a white row as a divider. At the end of the year, knit 5 rows of white and finish. The Ravelry description of the same pattern says that each color stands for a 5°C range, and that you may widen the ranges to use fewer colors. The only fixed part is the rule for transferring data onto fabric. In 2017, when someone asked about needle size, she answered that choosing the yarn, needles, and stitches yourself is what makes it a “conceptual knitting pattern”.

The source of inspiration was Sky Scarf by Lea Redmond, a project in which you add one stripe a day in the color of that day’s sky. In a comment dated April 25, 2013, Kristen wrote that Sky Scarf inspired her and that she thought temperature would be easier to record. It can be read as a judgment that, once numbers are binned into classes, a yarn color can be looked up from a table.

It spread quickly. The day after publication, Kristen wrote that many people on Ravelry were knitting along. The yarn maker Bernat (now Yarnspirations) also ran a knit-along that year and made a color guide.

Reading the Handwritten Color Card as a Legend

The blog post includes a photo of Kristen’s handwritten notes. They give the knitting rules and the color for each temperature class, with yarn snippets attached.

Kristen Cooper’s handwritten color card. It divides 30°C to −40°C into 14 classes in 5°C steps, with a yarn sample attached next to each color name

Source: Kristen Cooper / Honey Nutbrown’s, “Knitting!: My Year in Temperatures -Scarf-”

This card is, as it stands, the legend of a chart. The mapping between data and fabric is as follows.

Data	How it appears in the fabric	Specific values
Date (one day)	Position along the length; 1 day = 1 row	With thin yarn, 2 rows a day is fine (her reply)
The day’s high temperature	Row color	14 classes in 5°C steps, −40°C to 30°C
Start and end of the year	White rows	5 rows each
Month divider (optional)	White rows	2 rows in the text, 3 on the card
—	Width (no data)	35 stitches, garter stitch

The 14 classes and colors are written on the card as follows.

High temperature	Color	High temperature	Color
30 to 26°C	Pink	−5 to −9°C	Dark Green
25 to 21°C	Red	−10 to −14°C	Teal
20 to 16°C	Orange	−15 to −19°C	Bright Blue
15 to 11°C	Gold	−20 to −24°C	Navy Blue
10 to 6°C	Yellow	−25 to −29°C	Heather Blue
5 to 1°C	Light Yellow	−30 to −34°C	Grey
0 to −4°C	Green	−35 to −40°C	Light Grey

There is no class above 30°C, and only the last class, −35 to −40°C, spans 6°C. Seven classes are spent on the cold side, which suggests that this is a legend for a place where winter highs drop a long way.

The hot side is red and orange, and the cold side is blue and grey. The hottest class, however, is pink. Lightness does not change in one direction, either. On the cold side it goes from Navy Blue to Grey to Light Grey, getting lighter. Around 0°C the color switches sharply from yellow to green. This can be read as a palette that first makes you tell apart above and below freezing.

White is used in none of the 14 classes. It is reserved for information that is not temperature: the start and end of the year, and the monthly dividers. This can be read as assigning a color not used for data to dividers that do not represent data. Note that the number of rows in a monthly divider differs between the blog text, “Knit two rows in white”, and the card, “work 3 rows in white”.

The Legend Is Remade for Each Knitter’s Region

This card is a legend for Kristen’s region. To readers in Paris and Dubai, she answered that where the temperature range is narrow, they can use finer steps and more colors. When asked on January 20, 2013 for a Fahrenheit legend, she pointed to Bernat’s legend in Fahrenheit and wrote:

“Everyone seems to be making up their own.”

Everyone, it seems, is making their own legend. A local legend uses every color at hand to show a region’s changes over the year. The cost is that, it can be read, you cannot line the stripes up against a scarf from another region to compare them.

That summer, blanket versions appeared. On July 15, 2013, Arlene Cline posted the January to June section of a crocheted blanket on the knitting forum Knitting Paradise.

“Each row is a day. I did the temperatures in 5 degree increments,with black separating each month. This is based on my city.”

One row per day, 5-degree steps, black between months, and the temperatures of her own city, she writes.

Arlene Cline’s crocheted blanket (January to June 2013). Stripes of light blue, green, yellow, red, and pink are separated month by month with black rows

Source: Arlene Cline / Knitting Paradise, “temperature afghan-c”

From the cold side, the colors ran blue, green, yellow, orange, pink, wine, red, and anything below 0°C was white. Only the black monthly divider rows were worked in double crochet, so the stitch type changes too. January to June came to 50 inches wide and 63 inches long, and after completion she split it into two pieces of six months each. On the forum, another user wrote that she should attach an explanation to the work so that the meaning is not lost. It can be read as the point that once the legend is separated from the work, the stripes lose their meaning.

Other makers have varied the rules a little.

Case	What one row represents	How temperature is taken	Month divider
Kristen Cooper (2013, knitting)	1 day = 1 row	Daily high, 14 classes in 5°C steps	White rows
Arlene Cline (2013, crochet)	1 day = 1 row	5°C steps (whether high or average is not stated)	Black rows + double crochet
SleepyEyes (2014, Ravelry)	1 day = 2 rows	Temperature at the time of knitting, 18 colors	Garter stitch and seed stitch alternating by month

The temperature scarf SleepyEyes knitted for the whole of 2014. The stripe colors shift from purple and blue to green and yellow, then orange and red

Source: Ravelry user SleepyEyes / dataphys.org, “Temperature Scarves and Afghans”

SleepyEyes marked the month divisions not with color but with the texture of the fabric. Because color is not used for the dividers, it can be read that color is freed up to express temperature. dataphys.org also lists a blanket for 2016. It was knitted by Sharon MacDermaid and submitted by the meteorologist Ellen Bacca.

The 2016 temperature blanket knitted by Sharon MacDermaid. It is large enough to cover a bed, and rainbow stripes run from blue and green rows to yellow and red rows

Source: Made by Sharon MacDermaid, submitted by Ellen Bacca / dataphys.org, “Temperature Scarves and Afghans”

How Third Parties Saw This Way of Knitting

From visualization research

In their IEEE VIS 2025 paper “Stitching Meaning”, Sydney Purdue, Melanie Tory, and colleagues studied the encodings used by data textile creators. Of 159 pieces, 45 were temperature blankets, and the paper dates their beginning to Cooper’s 2013 post. Twenty-five creators used a rainbow colormap. The 22 who represented temperature with stepped colors used between 8 and 49 colors, with a median of 14.5. On intervals left open at both ends, the paper says:

“Some creators who used these kinds of catch-all intervals expressed regret that extreme temperatures were grouped with less extreme temperatures.”

The point is that with “or above” and “or below” intervals, extreme temperatures got the same color as milder ones, and some creators regretted it. One person who had assumed temperatures would never exceed 109 degrees found they did, and added two colors. The paper classifies the white rows between months as “annotation”, which does not represent data.

From the press

In 2018, Maddie Stone of Earther covered the Tempestry Project, which began in 2017. Everyone knits NOAA’s daily high temperatures on the same color scale. Co-founder Justin Connelly says:

“We want people to take our framework and develop their own collections, this large mosaic of international climate data that can be compared and contrasted because it’s all the same color scale.”

The claim is that, because the scale is the same, climate data from around the world can be lined up and compared. The article describes a case in which a multi-year collection from Utqiagvik, Alaska, made the rise in temperatures from autumn to early winter clearly visible. It can be read that a local legend shows change within a year, while a shared legend lets you compare places and years.

From a climate scientist

In 2017, the climate scientist Ellie Highwood adapted this format to global mean temperature anomalies, with one row per year. She then reflected on her own color choice:

“rainbow colour scales aren’t great as they can distort data and render it meaningless if you are colour blind”

Her reflection is that rainbow scales can distort data and, for some kinds of color vision, become meaningless. Canary Media confirms that Ed Hawkins’s warming stripes came out of this blanket.

Takeaways

The rules of the temperature scarf contain decisions that can be carried back into visualization work. Each of them can be read as coming from the nature of the craft: data arrives one day at a time, and a knitted row is hard to undo.

The first is to decide the legend first. Kristen matched the colors on the color card to temperature classes. If you knit one row a day, you need the legend before you start. This is the reverse of the usual order, in which you decide the range once the data is in. If you decide first, you must also decide then how to draw values outside the range. The regret about open-ended intervals that Purdue and colleagues recorded can be read as bearing on this point as well.

The second is to choose the scale according to what you want people to compare. Kristen recommended changing the steps for each region, while Tempestry has everyone use the same colors. It can be read that the former shows the seasons within one piece, and the latter shows differences between places and years.

The third is to use a means separate from the data for dividers. White, black with a different stitch, texture. Because they do not overlap with the temperature colors, the dividers are not misread as temperature.

The fourth is to keep the legend together with the work. Only with the color card do the 365 stripes become a chart.

I found no statement by Kristen or Arlene that they knitted in order to learn color scale design. Even so, a year of stripes makes the maker decide, one at a time, the width of the steps, the order of the colors, the handling of the two ends, and how to insert dividers. It can also be read as practice in the steps of designing a color scale, checked by hand one row a day.

References

Primary sources

dataphys.org, “2013 – Temperature Scarves and Afghans” (added by Pierre Dragicevic, submitted by Andy South). dataphys.org
Kristen Cooper, “Knitting!: My Year in Temperatures -Scarf-,” Honey Nutbrown’s, 2013-01-02 (text, images, and comments). Blog / Archive
Kristen Cooper, “My Year in Temperatures -Scarf-,” Ravelry pattern, January 2013. Ravelry
Arlene Cline (arlenecc), “temperature afghan-c,” Knitting Paradise, 2013-07-15. Knitting Paradise
Arlene Cline, “temperature afghan,” Knitting Paradise, 2014-01-13 (finished piece). Knitting Paradise
SleepyEyes, “My Year in Temperatures -Scarf-,” Ravelry project, 2014. Ravelry
Yarnspirations (formerly Bernat), “Crochet Temperature Afghans & Blankets,” 2016-01-05. Yarnspirations
Tempestry Project official website. tempestryproject.com
Lea Redmond, “Knit The Sky.” Leafcutter Designs

Secondary sources

Sydney Purdue, Eduardo Puerta, Enrico Bertini, Melanie Tory, “Stitching Meaning: Practices of Data Textile Creators,” IEEE TVCG (VIS 2025). Paper page / Preprint
Peter Rubinstein, “Data Visualizations You Can Wear? The Rise of Data Textiles,” Northeastern Global News, 2026-09-10. Article
Maddie Stone, “Meet the Knitters Who Are Turning Climate Change Data Into a Fashion Statement,” Earther (Gizmodo), 2018-12-22. Article
Ellie Highwood, “#climatechangecrochet – The global warming blanket,” 2017-06-12. Blog
“Who really invented the climate stripes?,” Canary Media, 2022-06-03. Article

Faster Than the 3D Bar Chart on Screen, Short of 2D: Jansen et al.'s Evaluation of Physical Visualizations

Sun, 01 Nov 2026 00:00:00 +0900

When the same 3D bar chart was moved from a screen to a physical object, reading data became about 20% faster. It still did not match an on-screen display that combined a flat matrix view with 2D bar charts. This is the result of an experiment published in 2013 by Yvonne Jansen and her colleagues. The “about 20%” is my rough estimate read from the paper’s figures, and what was compared was speed, not accuracy.

The source is the paper “Evaluating the Efficiency of Physical Visualizations” (CHI 2013) by Jansen, Pierre Dragicevic, and Jean-Daniel Fekete. In dataphys.org, the catalog of data physicalization examples, it appears as a 2012 entry titled “Putting Physical Visualizations to the Test.” What follows tracks the body of the paper. Values I read from figures are marked “approximate” each time.

The Debate over 3D Bar Charts Was Only About Screens

In information visualization, the common view is that 3D displays are not recommended. The paper acknowledges this at the start of its background section.

Overall 3D visualizations are not recommended since they are subject to serious occlusion, distortion, and navigation issues.

In short, 3D is discouraged because of occlusion, distortion, and the difficulty of controlling the viewpoint. The authors continue right after:

However, the controversy of 2D versus 3D only concerns the on-screen modality.

That is, the 2D-versus-3D debate has only ever been held about screens. 3D is common in physical visualizations, yet problematic on screen; the abstract gives this as the reason for choosing 3D. In the background are digital fabrication machines and research on shape-changing interfaces. The authors expect that, as either advances, building physical visualizations will become practical. The way of comparing is also distinctive.

The vast majority of infovis studies involve comparisons across visual mappings only. We focus on comparisons across modalities only

Most studies have compared visual mappings; this one declares that it compares only the medium. Height, color, and layout are held constant, and only whether the chart is physical or on screen changes. The authors describe it as the first study to compare physical and on-screen visualizations on information visualization tasks. The paper itself cites earlier work that compared physical and screen media, such as research on spatial memory.

Thirteen Bar Charts Built from Acrylic

The data are 16 sets built from Gapminder country indicators. One set consists of 100 values, 10 countries by 10 years, and the indicators include electricity consumption, birth rate, and car-accident death rate. They were chosen because the indicators are easy to understand, the values vary enough, and the years are evenly spaced.

One acrylic 3D bar chart. Bars for 10 countries by 10 years are color-coded by country, with transparent scale plates and a base engraved with country names and years; a euro coin sits next to it

Source: a work by Yvonne Jansen, Pierre Dragicevic, and Jean-Daniel Fekete (no photographer credited) / from dataphys.org, “2012 – Putting Physical Visualizations to the Test”

The physical charts are made of laser-cut acrylic. For each country, one cross-section of a 2D bar chart is cut out, painted, and assembled. The templates were generated automatically from the data. Each chart measures 8 × 8 cm outside and weighs 270 to 350 g, so it can be held and turned in the hand. Thirteen were built. The mapping is as follows.

Data / variable	Appearance / material	Details
Indicator value	Bar height	Proportional to the value. Each bar is 5 × 5 mm, with 2 mm gaps
Country	Row position and bar color	ColorBrewer qualitative palette. Order sorted by similarity
Year	Column position	10 evenly spaced years
Aid for reading values	Tick lines engraved on the sides of the bars	Engraved on 2 sides in the physical version
Axis scale	Engraved on transparent acrylic plates	2 plates per chart. Tick intervals computed automatically
Country and year labels	Engraved vertically on the 4 sides of the base	Requires turning the chart slightly to read, but allows a compact size

As countermeasures against occlusion, transparent bars, thin bars, and top-face-only display were also tried. The result, in the paper’s words, was “They all turned out to have poor legibility.” All were hard to read, and the authors went back to ordinary bars.

More than ten physical 3D bar charts laid out on a floor. All share the same format; only the pattern of bar heights differs

Source: a work by Yvonne Jansen, Pierre Dragicevic, and Jean-Daniel Fekete (no photographer credited) / from dataphys.org, “2012 – Putting Physical Visualizations to the Test”

Physical Was Faster, and 2D Was Faster Still

In the first experiment, the same data were shown under four conditions.

Condition	Visual mapping	Medium	Interaction
physical	3D bar chart	Physical object	Hold, turn, and touch by hand
stereo	3D bar chart	Screen (stereoscopic)	23-inch polarized 3D monitor and glasses
mono	3D bar chart	Screen	Rotate with the mouse; click to highlight a bar
2D	Matrix view + 2D bar chart	Screen	Selecting a row or column label shows a bar chart of that cross-section

The 2D condition shows values by the size of a square in each matrix cell, and selecting a row or column displays a bar chart of that cross-section. The authors settled on this form after also trying overlaid line charts and small multiples.

the matrix view is analogous to a top view of the 3D model and the 2D bar chart view is analogous to a side cut.

The matrix corresponds to a top view of the 3D model, and the 2D bar chart to a cross-section. The paper also states the role of the 2D condition.

Although our goal is not to compare 2D with 3D visualizations, we included an on-screen interactive 2D condition as a comparison baseline.

In short, comparing 2D with 3D was not the goal; the 2D condition was included as a baseline for comparison.

There were three kinds of tasks: Range, which asks for the range of values for one country; Order, which asks participants to sort the 10 countries for one year by value; and Compare, which asks them to pick the smallest value among three country-year combinations. Sixteen participants answered 576 questions in total.

Errors were few overall and did not differ by condition. The analysis was run on completion time, and the effect of condition was clear (F(3,24) = 165.18, p < 0.001). The authors summarize:

Overall, we can safely conclude that the physical 3D visualization is more efficient than its 3D on-screen counterpart, mono or stereo alike.

The physical 3D bar chart is more efficient than the on-screen 3D bar chart, whether monocular or stereoscopic. By my rough estimate from the paper’s figure, the physical condition’s completion time was about 0.82 times that of the on-screen monocular condition, and the 2D condition’s about 0.63 times that of the physical condition. The “about 20%” in the opening is this value.

Stereoscopy did not help. “Contrary to H3, there is no evidence of stereo outperforming mono.” The hypothesis that stereo is faster than mono was not supported.

2D was fastest on all three tasks. However, the size of the gap varies by task.

So although 2D beats all 3D conditions, the effect is weaker if a task cannot be solved by one 2D cut.

2D beat every 3D condition, but the advantage is smaller when a task cannot be solved with a single cross-section. Reading mean times per task from the figure gives the following.

Condition	Range	Order	Compare
physical	about 24 s	about 47 s	about 39 s
stereo	about 27 s	about 62 s	about 52 s
mono	about 27 s	about 62 s	about 48 s
2D	about 15 s	about 24 s	about 31 s

(All values in the table are my rough estimates read from Figure 5 of the paper.)

Preference rankings matched the speed rankings. Some participants said of on-screen 3D, “I would loose my ‘mark’ easily during rotation,” meaning they lost their place as they rotated the chart. The recordings capture this behavior:

Most subjects used their fingers to temporarily “mark” rows and columns of interest. Eventually, almost everyone switched to putting their fingers directly on the bars relevant to the task.

Most people used their fingers to mark rows and columns, and in the end almost everyone put their fingers directly on the bars relevant to the task.

Part of the Speed Came from Fingers Placed on the Bars

Why is the physical chart faster? The second experiment separated the reasons and measured them. There were four conditions.

Condition	Description
mouse	On-screen 3D bar chart rotated with the mouse. No highlighting
prop	On-screen 3D bar chart rotated with a physical prop containing a rotation sensor
no touch	Physical 3D bar chart. Marking with fingers prohibited
touch	Physical 3D bar chart. Touching encouraged

From mouse onward, each condition adds one element: “rotating directly by hand,” “the look of the physical object,” and “touching.” Only the Compare task was used, with newly made problems in which no viewpoint shows all three bars at once. There were again 16 participants. The paper reports the results numerically in the text.

touch is 15% shorter than no touch (confidence interval 7 to 22%)
no touch is 13% shorter than prop (same, 5 to 20%)
prop is 3% shorter than mouse (same, −4 to 10%)

So the effect of rotating directly, if any, is small. In numbers, touching cut time by 15%, the look of the physical object by 13%, and rotating directly by hand made a difference of 3%.

The authors limit the meaning of touch as follows.

We do not refer to the use of tactile sensory information but to the action of placing fingers to use them as visual or memory aids.

It refers not to tactile information but to the act of placing fingers as a visual and memory aid. For the 13% gap between no touch and prop, the authors write only with a hedge, “Visual realism might also play a role.” This suggests that even with the same way of rotating, how the chart looks may have affected speed.

Dragicevic and Jansen later reanalyzed the second experiment with 56 different analysis methods and concluded that the results do not depend on the choice of analysis. For the first experiment, the authors themselves caution that, once corrected for multiple comparisons, some effects may become less certain.

How Third Parties Viewed This Study

The assessments fall into three groups: replication, positioning, and reactions at the time of publication.

Replication: In 2020, K. Danyluk and colleagues compared a physical Lego chart with a VR version, using the same format, data, and tasks (18 participants). On the Compare task, physical took 33.6 seconds, VR at palm size 56.6 seconds, and VR at table size 65.7 seconds. They report that every participant traced the bars with their fingers in the physical version, and summarize this as consistent with Jansen et al.’s results.
Positioning: In 2016, S. Stusak and colleagues introduced this study as the first to compare physical and on-screen displays on reading tasks. They also note that Jansen et al. themselves said their tasks and media were limited. Stusak and colleagues then moved their own focus to memory.
Reaction at the time: In a May 2013 blog post, Patrick Durusau asked how closely a VR simulation of touch could approach the real object.

Danyluk and colleagues also list “physical manipulation” as an advantage, but the original paper says rotating directly is not important, so this can be read as slightly off. Durusau’s question can be read as having been tested seven years later in VR by Danyluk and colleagues. Danyluk and colleagues also point out a limitation:

Ultimately, static 3D barcharts like those we used are unlikely to outperform either 2D or 3D visualizations that include interactive features like dynamic filtering or tooltips.

Static 3D bar charts are unlikely to beat displays with interactive features such as filtering or tooltips.

Takeaways

There are three takeaways about choosing a medium and one about evaluation methods.

First, if you use 3D, making it physical is an option. As a design implication, the authors write:

i) if an analyst wishes to use a 3D visualization, making it physical is an option worth considering; ii) physical visualizations should be built to support direct touch and not enclose data.

Making a 3D chart physical is worth considering, and physical visualizations should be built to allow direct touch and not enclose the data. Putting the model in a case would take away the “fingers on the bars” way of reading that mattered in this study.

Second, for data with two categories and one quantity, first consider the pair of a top view and a cross-section. This pair was fastest in the study. However, the study was not designed to settle which of 2D and 3D is better. For comparisons that a single cross-section cannot solve, 2D’s advantage also shrank. These two points need to accompany any citation of this work as grounds for avoiding 3D.

Third, provide a “place to put a finger” on screens too. What affected speed appears to have been less the rotation than the ability to leave a mark. The authors also write that 2D displays would benefit from paper or touch screens. A feature that keeps the selected row or column highlighted can be read as helping reading regardless of medium.

Finally, on evaluation methods: hold the visual mapping constant and change only the medium. Because of this design, “physical is faster” and “2D is faster” could both be stated, without contradiction, within the same experiment.

References

Primary sources

Yvonne Jansen, Pierre Dragicevic, Jean-Daniel Fekete, “Evaluating the Efficiency of Physical Visualizations,” Proceedings of CHI 2013, pp. 2593–2602. doi:10.1145/2470654.2481359 / HAL / PDF
dataphys.org, “2012 – Putting Physical Visualizations to the Test” (Added by Pierre Dragicevic). Entry page
The authors’ project page (the paper says it hosts the datasets, experiment data, and videos; I have not verified this). aviz.fr/phys
Pierre Dragicevic, Yvonne Jansen, “Re-Evaluating the Efficiency of Physical Visualizations: A Simple Multiverse Analysis”. Explorable Multiverse Analyses / Reproduction materials for the R multiverse package
Yvonne Jansen, Physical and Tangible Information Visualization, doctoral dissertation, 2014. HAL

Secondary sources

K. Danyluk, T. Ulusoy, W. Wei, W. Willett, “Touch and Beyond: Comparing Physical and Virtual Reality Visualizations,” IEEE TVCG, 2020. doi:10.1109/TVCG.2020.3023336 / Author version / Abstract
S. Stusak, M. Hobe, A. Butz, “If Your Mind Can Grasp It, Your Hands Will Help,” TEI 2016. PDF
Patrick Durusau, “Evaluating the Efficiency of Physical Visualizations,” Another Word For It, 2013-05-12. Blog

Pulling Out and Rearranging One Chart per Country: The Rearrangeable 3D Bar Chart

Sun, 01 Nov 2026 00:00:00 +0900

The Rearrangeable 3D Bar Chart is a physical 3D bar chart showing unemployment rates for 10 countries over 10 years. Each country’s 2D bar chart is a separate part that can be pulled out and pushed back in. People move the parts by hand to sort the countries, pull them out, and superimpose them to compare.

The creators are Yvonne Jansen and Pierre Dragicevic, then at the AVIZ team of Inria. The work is presented as one of eight case studies in their paper “An Interaction Model for Visualizations Beyond The Desktop” (IEEE TVCG, 2013). In dataphys.org, the catalog of data physicalization examples, it is listed as a 2012 example. What follows is based on this paper and the dataphys.org entry.

Ten 2D Bar Charts in One Piece

The dataphys.org entry describes the work in a single sentence.

A modular physical visualization like this rearrangeable 3D bar chart allows people to sort, filter, compare and examine data by direct physical manipulation.

In other words, a physical visualization made of separate parts lets people sort, filter, compare, and examine data by moving the parts directly with their hands. The entry was added by Dragicevic, and it is classified as “Passive physical visualization,” meaning it has no electronic components.

According to the paper, the chart shows “unemployment rates over 10 years for 10 countries.” With 10 countries and 10 years, that makes 100 bars by calculation. The parts were cut automatically from the data with a laser cutter, and painting and assembly were done by hand.

Left: a photo of a finger pinching a blue row and lifting it up. Right: a photo of a hand holding a whole pink row pulled out of the chart

Source: Yvonne Jansen and Pierre Dragicevic, dataphys.org, “2012 – Rearrangeable 3D Bar Chart”

In the left photo, a finger lifts a blue row. On the right, a whole pink row has been pulled out. Each row has a different color, and year labels beginning with “2001” line up on the red base at the front. The paper does not explain what the colors or the base mean, so what follows here is a reading from the photos. The table below summarizes how data maps to physical properties.

Data	Physical property	Basis
Unemployment rate (quantity)	Bar height	Paper. Scale values and proportions are unknown
Country (10 countries)	A row in the depth direction. One country is one part that can be pulled out and pushed back in	Paper
Year (10 years)	The horizontal arrangement of bars within each row. Fixed within the part	The 10 years are from the paper; the arrangement is a reading from the photos
Order of countries	Not determined by the data. Can be changed by hand	Paper
Distinguishing countries	Color per row	A reading from the photos
Year labels	The base at the front	A reading from the photos

As the source for how it was built, the paper cites the same authors’ CHI 2013 paper. That paper’s physical 3D bar chart and its evaluation experiment are covered in another article, “Faster than On-Screen 3D Bar Charts, but Not Faster than 2D.” I could not confirm whether the two are the same object or different specimens made in the same way. I also could not confirm the materials or dimensions for this work.

Pull Out, Move Apart, Superimpose

The paper divides the operations possible with this bar chart into two levels. The first is operations that change how the chart is seen, such as rotating the chart or pressing a finger on a data point. The second is operations that move the parts.

Each country is a 2D bar chart that can be taken out and manipulated separately. This simple design choice enables a user to perform a range of additional tasks (2): she can sort countries, filter them out by moving them away, or compare countries by superimposing them.

The paper contrasts this with an example of a single-piece model: a 3D-printed coral model used as a tool for manipulating a screen.

Left: a photo of the pink row being pulled out. Right: a diagram drawn with the paper’s model, with the labels “rotate & mark” at (1) and “sort & filter” at (2)

Source: Yvonne Jansen and Pierre Dragicevic, “An Interaction Model for Visualizations Beyond The Desktop” (IEEE TVCG, 2013), Fig. 9

The right side of Figure 9 is a diagram that describes these operations with the paper’s model. The arrow at (1) is labeled “rotate & mark” and the one at (2) “sort & filter.”

In the paper’s terms, filtering means “moving the parts away,” and comparing means “superimposing” them. Filtering here is not a process that deletes data but one that takes parts out of view, which can be read this way. Both can be read as leaving the data itself untouched.

Sorting Rewrites a “Free” Variable

The core of the paper is a model of the visualization pipeline. It divides the path into stages: raw data becomes a figure, the figure takes on a physical form, and a person reads it. The key to the explanation of sorting is free visual variables.

We call those free visual variables, to contrast them with encoding visual variables that are constrained by the visual mapping.

Visual variables that are not constrained by the mapping from data are called free visual variables, as opposed to encoding visual variables. In the paper’s example, the box plot as a mapping does not determine where along the horizontal axis each box is placed.

Free visual variables are fixed at the stage of the presentation mapping. The paper calls assigning them so as to make the chart easier to read optimization, and gives the example of ordering box plots by participant ID. It also distinguishes operations that locally override a uniform process: moving the nodes of a graph by hand, for instance, is an optimization-overriding operation. For this bar chart, the paper writes:

rearranging the visualization amounts to modifying free visual variables at the presentation mapping level, or more specifically, performing optimization overriding operations.

The order of countries is a variable the data never determined in the first place. Sorting changes neither the unemployment values nor the bar heights; only the placement changes. This work can be read as turning that free variable into a physical part and handing it to the user’s hands. The order of years, on the other hand, is fixed inside each part and cannot be changed by hand. The paper does not state this explicitly; it is a reading from the structure.

On a Screen, Separate Instruments; by Hand, One Motion

The paper describes the difference from a screen as follows.

Such operations are supported through true direct manipulation and are very versatile. In contrast, on a desktop system these tasks would typically be considered separately and supported by different instruments.

In the paper, an instrument is something that mediates between a person and a visualization; a sort button or a filter slider are examples. With this bar chart, every task comes down to one motion: pinching a part and moving it. It can be read as not needing to add an instrument for each task.

The same paper also describes the limits of manual operation in its conclusion.

Still we will always need instrumental operations as those allow to carry out complex tasks that have no real world counterpart (e.g., automatic sorting, brushing & linking).

Tasks with no real-world counterpart, such as automatic sorting or brushing and linking, will always need operations through instruments. Elsewhere, the paper also writes that tasks too complex or too repetitive to do by hand, such as reordering a matrix, are better left to the computer.

Another limit is the link to the data. The paper gives printed matter and 3D fabrication as examples of one-shot propagation: if the source data changes after the object is made, the form is not updated. This bar chart, too, is described as built by “semi-automatic one-shot propagation.”

How Third Parties Saw This Work

The evaluations fall into three: what the way of dividing parts means, users’ hesitation, and rearrangement by machine.

The first is a survey of the physicalization fabrication process presented at EuroVis in 2021 by H. Djavaherpour et al.

the way in which a physicalization is designed to be assembled can determine what types of interactions end-users can have with its represented data.

The point is that how a piece is designed to be assembled determines what kinds of operations users can perform on the data. The survey goes on to write that if Jansen et al.’s assembled 3D bar chart had been glued together, users could not have engaged with the data so deeply. However, the survey cites the CHI 2013 paper, and I have not confirmed whether it refers to the same object as this work. Also, Jansen is among the co-authors, so this is not a fully independent third party.

The second is Simon Stusak of LMU Munich. A different group built a physical visualization aimed at sorting and filtering by separating layers. A 2013 poster says that with a model whose layers can each be rotated, similar items can be grouped only along the time axis, and values within a layer cannot be reordered. In his 2016 doctoral thesis, he built a modular version whose parts are joined with hook-and-loop fasteners and compared it with a fixed version.

Four participants did not disassemble the modular physical visualization, two because it seemed impractical, two because of fear to break something.

Four participants did not take the modular version apart: two because it seemed troublesome, and two out of fear of breaking it. One participant who tried to take it apart broke a part. This was not Jansen et al.’s bar chart, but it can be read as showing that being able to pull parts out and push them back is different from people actually doing so.

The third is EMERGE by Faisal Taher et al. It is a physical bar chart with 10 x 10 bars raised and lowered by motors. The 2015 paper names the drawbacks of passive physical visualizations as being static and, once made, cut off from the data source. The IEEE VIS 2016 paper reports that in a presentation setting, some participants swapped rows and columns and built elaborate ways of showing the data. Jansen is a co-author of both.

Takeaways

In the 2015 paper “Opportunities and Challenges for Data Physicalization,” the same group wrote about the degree of freedom in rearranging parts as follows.

However, limited reconfigurability can be desirable. Information as to which states and operations are meaningful can be conveyed as physical constraints

Limiting how a piece can be reconfigured can be desirable, and which states and operations are meaningful can be conveyed through physical constraints. What can be taken from this work overlaps with this passage.

The first is to decide where to split the parts starting from the operations you want to allow. This bar chart is divided between countries, so the order of countries can change and the order of years is preserved. The position of the split decides at once which operations are allowed and which are not. Whether to glue, to make parts pull-out, or to make them snap in is part of the same decision.

The second is to think of sorting as separate from changing the data. In the paper’s framing, sorting is overriding a free visual variable, and no value changes. On a screen, too, if you treat the order as a variable the data does not determine, you can add operations like dragging rows without strain.

The third is to build ease of pulling out and putting back into the design. In Stusak’s observations, some people did not disassemble the piece for fear of breaking it. The design needs a form that shows a part may be pulled out and is easy to return.

The fourth is to identify the operations not to leave to the hand. With 10 countries, sorting by hand works, but as rows increase it is better left to computation. The fact that data is not updated once the object is made is also worth considering before choosing physicalization.

References

Primary sources

dataphys.org, “2012 – Rearrangeable 3D Bar Chart” (Added by Pierre Dragicevic): https://dataphys.org/list/rearrangeable-3d-bar-chart/
Yvonne Jansen, Pierre Dragicevic, “An Interaction Model for Visualizations Beyond The Desktop,” IEEE TVCG 19(12): 2396–2405, 2013. doi:10.1109/TVCG.2013.134 / Full text I read (mirror): https://typeset.io/pdf/an-interaction-model-for-visualizations-beyond-the-desktop-3p3eh2iru9.pdf / HAL (not checked by the author): https://hal.science/hal-00847218 / Authors’ page (the author could not connect): http://www.aviz.fr/beyond
dataphys.org, “2012 – Putting Physical Visualizations to the Test”: https://dataphys.org/list/testing-physical-visualizations/
Yvonne Jansen, Pierre Dragicevic, Petra Isenberg, Jason Alexander, Abhijit Karnik, Johan Kildal, Sriram Subramanian, Kasper Hornbæk, “Opportunities and Challenges for Data Physicalization,” CHI ‘15, pp. 3227–3236: https://eprints.lancs.ac.uk/id/eprint/72691/1/OA_2015_04_CHI_DataPhys.pdf
Yvonne Jansen, Pierre Dragicevic, Jean-Daniel Fekete, “Evaluating the Efficiency of Physical Visualizations,” CHI ‘13, pp. 2593–2602 (cited by the paper as the source for how it was built; the author has not read the full text): https://hal.archives-ouvertes.fr/hal-00781831

Secondary sources

H. Djavaherpour, F. Samavati, A. Mahdavi-Amiri, F. Yazdanbakhsh, S. Huron, R. Levy, Y. Jansen, L. Oehlberg, “Data to Physicalization: A Survey of the Physical Rendering Process,” Computer Graphics Forum 40(3), 2021 (EuroVis STAR): https://arxiv.org/pdf/2102.11175
Simon Stusak, Aurélien Tabard, Andreas Butz, “Can Physical Visualizations Support Analytical Tasks?” IEEE VIS 2013 poster: https://www.medien.ifi.lmu.de/pubdb/publications/pub/stusak2013visPoster/stusak2013visPoster.pdf
Simon Stusak, Exploring the Potential of Physical Visualizations, Dissertation, LMU München, 2016: https://edoc.ub.uni-muenchen.de/20190/1/Stusak_Simon.pdf
Faisal Taher, John Hardy, Abhijit Karnik, Christian Weichel, Yvonne Jansen, Kasper Hornbæk, Jason Alexander, “Exploring Interactions with Physically Dynamic Bar Charts,” CHI ‘15: https://www.kasperhornbaek.dk/papers/CHI2015_Emerge.pdf
Faisal Taher, Yvonne Jansen, Jonathan Woodruff, John Hardy, Kasper Hornbæk, Jason Alexander, “Investigating the Use of a Dynamic Physical Bar Chart for Data Exploration and Presentation,” IEEE TVCG 23(1), 2017: https://yvonnejansen.fr/articles/VIS2016%20-%20Investigating%20the%20Use%20of%20a%20Dynamic%20Physical%20Bar%20Chart%20for%20Data%20Exploration%20and%20Presentation.pdf

Five Stacks as a Base: Remove, Move, Compare in Brookings' LEGO Videos

Sun, 01 Nov 2026 00:00:00 +0900

The US think tank Brookings Institution represented five income tiers with five LEGO stacks. In a 2014 video, a LEGO figure was moved up and down the rungs of society. In a 2015 video, bricks were removed to account for taxes, and the heights before and after tax were compared. Both are videos of about three minutes in which a researcher moves the LEGO in front of the camera.

This article looks at one entry in the dataphys.org catalog of physicalizations, “2014 – Data Storytelling with LEGOs”. The entry is not a single work but a bundle of two separate Brookings videos. The first is Richard V. Reeves’s video on social mobility (August 2014); the second is David Wessel’s video on taxes and income inequality (April 2015). I have not been able to check the videos themselves, so what follows is a reading based on what Brookings’ own articles and the press coverage of the time say.

One dataphys.org entry, two videos

dataphys.org is a catalog of physical visualizations compiled by researchers. This entry was registered by Fanny Chevalier, and its tags are “LEGO”, “rearrangeable”, and “storytelling”. The description begins as follows.

LEGO blocks are convenient tools to progressively construct a physical visualization while unfolding a story.

In short, LEGO is a handy tool for building a physical visualization step by step as the story unfolds. The entry goes on to say that this follows the same flow as Hans Rosling’s talks, and names two subjects: how much taxes reduce inequality, and the prospects of someone born on the bottom rung.

The “2014” in the entry title is the year of the social mobility video. The tax video was released in 2015, so it does not match the year in the title.

	Video A	Video B
Title	Is America Dreaming?: Understanding Social Mobility	Do Federal Taxes Reduce Income Inequality?
Presenter	Richard V. Reeves	David Wessel
Released	August 20, 2014	April 9, 2015
What is compared	The tier a person is born into, and the tier they reach as an adult	Average income before and after taxes
Data	NLSY79 and C-NLSY (surveys by the US Bureau of Labor Statistics)	Estimates from the Urban-Brookings Tax Policy Center

2014: Where the figure ends up

Video A was published on August 20, 2014, embedded in the Brookings article “Watch: A Snapshot of Social Mobility in America” (Fred Dews). Dews’s article introduces it as a video to watch before reading Reeves’s long essay “Saving Horatio Alger,” published the same month.

Green LEGO stacks stand in front of Reeves, labeled 10%, 14%, 20%, and 52% from the left

Source: Brookings Institution / Brookings, “Watch: A Snapshot of Social Mobility in America”

According to the local outlet Baltimore Brew, Reeves stands behind five stacks of green LEGO. Each of the five is a tier covering 20% of the population. The story starts from a perfectly mobile society.

In a perfectly mobile society, an opportunity utopia, being born down here in the bottom quintile would have no effect on where you ended up. … But now I want to show you what it’s really like…

In a perfectly mobile society, being born in the bottom tier would have no effect on where you end up; now he will show what it is really like (transcript by Baltimore Brew). The order is to place the baseline state first, then rearrange the pieces into reality.

A LEGO figure stands on top of five stacks, labeled 36%, 24%, 18%, 13%, and 10% from the left

Source: Brookings Institution (video screenshot) / Baltimore Brew, “A lesson in Legos about social mobility in America”

Baltimore Brew writes that Reeves moves a LEGO figure up and down the rungs of society. According to press reports, of the people born in the bottom tier, a little over one in three stays in the bottom tier, and one in ten reaches the top tier (Gizmodo, Yahoo Finance). The 36% at the far left and the 10% at the far right of the image above match these figures.

The video then rearranges the lineup of figures by group. According to Yahoo Finance, 3% of Black Americans born in the bottom tier reach the top tier, compared with 1% of high school dropouts and 20% of college graduates.

The source of the data is in an article that Reeves and Joanna Venator wrote the next day.

In both the video and the essay, we’ve created a series of ‘mobility matrices’ showing how income status in one generation influences income status in the next.

That is, in both the video and the essay they built “mobility matrices” showing how a parent generation’s income position affects the children’s. The sample was 5,783 children, and the matrices came in three kinds: by race, by education, and by mother’s marital status.

Five bars for children whose mothers were continuously married, each stacking the shares of adult quintiles for one birth quintile

Source: Richard V. Reeves & Joanna Venator / Brookings, “Saving Horatio Alger: The Data Behind the Words (and the Lego Bricks)”

The same article presents the mobility matrices as stacked bar charts. The figure above is the matrix for children whose mothers were continuously married. In the bar for those born in the bottom tier, 17% stay in the bottom tier and 19% reach the top tier.

The underlying data is a 5×5 table. The video can be read as taking just one row of it, “those born in the bottom tier,” and showing it by group through the lineup of figures. Since I have not watched the video, I have not checked whether it touched on the other rows.

2015: Removing bricks for taxes

Video B was published on April 9, 2015, embedded in Wessel’s article “How much does the tax code reduce inequality?” It came just before April 15, the US tax filing deadline.

And you can tell it with tables of numbers, charts and graphs—or you can tell with Legos.

You can tell it with tables of numbers, charts, and graphs, or you can tell it with LEGO. Wessel goes on to write that he made a three-minute video with LEGO.

Wessel picks up a green LEGO brick, with five stacks rising from left to right in front of him and the text AVERAGE INCOME BEFORE TAXES at left

Source: Brookings Institution / Brookings, “How much does the tax code reduce inequality?”

Upworthy writes that each stack represents the average pre-tax income of each 20% of the population. In the still, the leftmost stack is about one brick tall, and the rightmost rises above Wessel’s shoulder.

According to Wessel’s article, the figures are 2014 estimates from the Urban-Brookings Tax Policy Center. The average pre-tax income of the top 20% is $306,320, more than 21 times that of the bottom 20%. After federal taxes, it is $229,360, about 17 times the bottom.

Taxes were shown by removing bricks from the stacks. The Week writes that 70 bricks are taken from the top 1%, while only one thin brick is taken from the bottom 20%. For the top 1%, Deseret News writes that showing them on screen would take more than 200 bricks. In Wessel’s article, the top 1%’s income is a little over $2 million before taxes and $1.34 million after.

The two encodings are summarized below.

Data	Physical representation	Video	Source
Income quintiles (20% of the population each)	Five stacks of green LEGO	A, B	Baltimore Brew, Upworthy
Children by group	LEGO figures	A	Baltimore Brew, Gizmodo
Adult tier	Position of the stack (rung) the figure is placed on	A	Baltimore Brew
Group (race, education, etc.)	Rearranging the lineup of figures	A	Gizmodo, Yahoo Finance
Average pre-tax income	Height of the stack	B	Upworthy
Federal taxes	Bricks removed from the stack	B	The Week
Average post-tax income	Height remaining after removal	B	Wessel’s article, The Week
Income of the top 1%	Not made into a stack; the video says it would take more than 200 bricks	B	Deseret News

Going by the press coverage, both videos can be read as using five stacks, differing in what moves: a figure in one, bricks in the other. In Video A the figure moves; in Video B the bricks decrease. The dataphys.org tag “rearrangeable” can be read as pointing to this stacking, removing, and moving.

One caution. Wessel’s article gives the bottom 20%’s average income as $13,809 both before and after taxes, which also does not line up with the figures in the press reports. For that reason, I use only the ratios and the top-tier amounts here.

“Animation would be too sterile”

The director of Video A was George Burroughs of Brookings’ communications office. Video B is listed as one of the productions of the Brookings Creative Lab, which Burroughs leads. Baltimore Brew quotes from Burroughs’s production notes.

Animation would be too sterile, Richard explaining two-dimensional charts would be visually mind-numbing, nothing felt right,

Animation would be too sterile, Reeves explaining two-dimensional charts would be visually mind-numbing, and nothing felt right. Burroughs goes back to why people watch videos online, and writes that it is “emotion and story.”

The credits for Video A that Baltimore Brew published include a director of photography and graphics, as well as someone who operated a motion control slider. This can be read as footage shot with controlled camera movement.

On July 20, 2015, Brookings launched the “Brookings Creative Lab,” led by Burroughs, and listed the tax video among its productions. Burroughs’s words are “We’re trying to give a face to the numbers.”

The same data was not released only as a video. The social mobility data came out in three forms: the essay (a long text with an interactive matrix), a blog post explaining the data, and the LEGO video. Reeves and colleagues present the video as something you can watch in “three minutes.” The LEGO video can be read as placed there as an entry point that leads people to the more detailed materials.

How third parties saw the videos

The press praised different things.

Baltimore Brew (Fern Shen, October 16, 2014) called the social mobility video “a clever approach to visual storytelling” and spent much of its space introducing the production team. It is an article that looks at how the video was made rather than at what it says.

The Week (Peter Weber, April 14, 2015) wrote that the tax video is “a pretty helpful way to visualize a fairly dry data set.” The point is that it helps to show plain data. It reads taxes as “the number of bricks removed,” contrasting 70 bricks with a single thin one.

Upworthy (Parker Molloy, April 13, 2015) used the tax video as material for its own argument. It writes that for the tax system to have a large effect on the distribution of wealth, it would have to become much more progressive. The Wessel in the video went no further than saying that taxes narrow the gap but that things remain quite unequal after taxes (as quoted by The Week). This can be read as showing that even when a physical form makes things easier to grasp, room for interpretation remains.

As for scholarly assessment, I found no literature that discusses these two videos individually. Mentions from the research side are mainly the dataphys.org entry.

Takeaways

What we can take from these two videos is, first, to fix the base. Going by the press coverage, both videos can be read as using five stacks, differing in what moves: a figure in one, bricks in the other. With the same base, viewers can compare the next state while remembering the previous one.

Second, decide the order of comparison first. Video A proceeds from a “perfectly mobile society” to “reality,” and then to groups. Video B proceeds from “before taxes” to “after taxes.” In a visualization that uses motion, which state moves to which becomes the argument itself.

Third, do not turn everything into objects. For the top 1%, Deseret News writes that making a stack would take more than 200 bricks. Decide the range to show with objects, and hand numbers that do not fit over to words and articles. Using the video as an entry point and having people read the rest in the essay and blog can be read as an extension of this.

Finally, checking the numbers. As in Wessel’s article, even the original text can have figures that do not line up. Before converting values into the height of an object, it is worth confirming which values to use.

References

Primary sources

dataphys.org, “2014 – Data Storytelling with LEGOs” (registered by Fanny Chevalier): https://dataphys.org/list/data-storytelling-with-lego/
Fred Dews, “Watch: A Snapshot of Social Mobility in America,” Brookings, 2014-08-20: https://www.brookings.edu/articles/watch-a-snapshot-of-social-mobility-in-america/
Richard V. Reeves & Joanna Venator, “Saving Horatio Alger: The Data Behind the Words (and the Lego Bricks),” Brookings Social Mobility Memos, 2014-08-21: https://www.brookings.edu/articles/saving-horatio-alger-the-data-behind-the-words-and-the-lego-bricks/
Richard V. Reeves, “Saving Horatio Alger: Equality, Opportunity, and the American Dream,” Brookings Essay, 2014 (full text not checked): https://www.brookings.edu/essay/saving-horatio-alger-equality-opportunity-and-the-american-dream/
David Wessel, “How much does the tax code reduce inequality?,” Brookings, 2015-04-09: https://www.brookings.edu/articles/how-much-does-the-tax-code-reduce-inequality/
Fred Dews, “Brookings launches “Brookings Creative Lab” to show the stories behind the research,” Brookings, 2015-07-20: https://www.brookings.edu/articles/brookings-launches-brookings-creative-lab-to-show-the-stories-behind-the-research/
Brookings Institution, “Is America Dreaming?: Understanding Social Mobility” (YouTube, video itself not checked): https://www.youtube.com/watch?v=t2XFh_tD2RA
Brookings Institution, “Do Federal Taxes Reduce Income Inequality?” (YouTube, video itself not checked): https://www.youtube.com/watch?v=7fgtE-AJs-A

Secondary sources

Fern Shen, “A lesson in Legos about social mobility in America,” Baltimore Brew, 2014-10-16: https://www.baltimorebrew.com/2014/10/16/a-lesson-in-legos-about-social-mobility-in-america
“Economist Uses Lego Bricks to Show the Ugly Truth of Social Mobility,” Gizmodo, 2014: https://gizmodo.com/economist-uses-lego-bricks-to-show-the-ugly-truth-of-so-1627881169
Mandi Woodruff, “The dying American dream, perfectly illustrated with Legos,” Yahoo Finance, 2014-08-22: https://ca.finance.yahoo.com/news/american-dream-illustrated-with-legos-brookings-200611923.html
Peter Weber, “Watch an economist use Legos to explain how taxes affect U.S. income distribution,” The Week, 2015-04-14: https://theweek.com/speedreads/549534/watch-economist-use-legos-explain-how-taxes-affect-income-distribution
Parker Molloy, “Using Lego bricks, an economist demonstrates how taxes affect income inequality.,” Upworthy, 2015-04-13: https://www.upworthy.com/using-lego-bricks-an-economist-demonstrates-how-taxes-affect-income-inequality
Herb Scribner, “Do taxes help income inequality? These Legos hold the answer,” Great Bend Tribune (reprint of Deseret News), 2015-04-15: https://www.gbtribune.com/news/trending-topics/do-taxes-help-income-inequality-these-legos-hold-the-answer/

Moving a Familiar Pie Chart in a Street-Facing Window: Physical Charts on Tenison Road

Sun, 01 Nov 2026 00:00:00 +0900

Physical Charts is a work that moves the pie charts and bar graphs everyone knows by machine. David Sweeney and colleagues at Microsoft Research Cambridge made it in 2014. They placed it in the lab’s window facing Tenison Road in Cambridge, UK, and used it to show locally collected car traffic and the results of votes that residents answered.

The work is part of the “Tenison Road project,” in which residents and researchers handle local data together. It is listed on dataphys.org under the name “Dynamic Physical Charts Display Community Data.” What follows reads it alongside the authors’ papers “A Pixel is Not a Pixel” (RTD 2015) and “Designing Engaging Data in Communities” (CHI EA 2015), and a CSCW 2017 paper that reports on the deployment in the window.

2013: Collecting Street Data with Residents

Tenison Road is a street of about half a mile that runs from Cambridge station toward Anglia Ruskin University. The Microsoft Research lab faces this street. In October 2013, the lab’s Human Experience & Design group started a project with residents. They handed out flyers to about 80 homes and held a launch meeting, and since then have met almost monthly with 20 to 30 residents.

The data collected falls into three main kinds.

Data	How it is collected
Car traffic (volume, speed, direction)	A thermal camera on the lab’s roof counts passing cars. 7,314 counts in a single day, February 25, 2014
Air quality	Readings from Air Quality Egg sensors placed in residents’ homes are aggregated
Local votes	The web vote BallotShare, the push-button voting poster PosterVote, and the home voting device Bullfrog

Before building the charts, in 2014, the authors wrote down the idea in a paper in Big Data & Society.

we are envisioning the data driving remotely controlled, mechanical pie-charts and graphs, creating a spectacle for bystanders on the road to view the ‘data’ they are themselves producing, as well as ways for them to engage in and comment on it.

In short: use the data to drive mechanical pie charts and graphs, make a spectacle through which people on the street can see the “data” they themselves produce, and also give them a way to engage with it and comment on it.

Putting Physical Charts in a Window Full of Screens

On where to show the work, the abstract of RTD 2015 says this.

We chose to use our highly visible window-frontage, visible from Tenison Road itself. As a technology company, our street facing window is full of screens, which could easily have been used for our purposes. Instead we have designed and built physical visualizations of a pie chart and a line graph.

The location is the lab’s window, which is easy to see from the street. A technology company’s window has plenty of screens that could have been used, but the authors instead built a pie chart and a line graph (in the finished work, a bar graph) as physical objects.

The same paper says that simple charts are so common that they fail to catch the eye. The way out was to keep the simplicity while presenting the data physically rather than on a screen or on paper. The paper is also explicit about its reaction against the pursuit of novelty.

In response to the increasing sophistication of data visualisations and the seemingly unquestioning quest for novelty, the charts make playful use of long established and highly familiar representations like pie charts and bar graphs. Rather than estrange viewers, the objective is to enable them to, at a glance, engage with and comprehend data.

In short: visualizations are becoming more elaborate, and novelty is pursued without question. In response, the charts make playful use of familiar pie charts and bar graphs, so that viewers can understand the data at a glance. Sweeney’s portfolio also lists not using glowing screens or LEDs as one of the design conditions.

In the author’s view, physicalization is a technique often used to add a new dimension, such as volume or touch. Here, by contrast, it can be read as a means of keeping intact the way 2D pie charts and bar graphs are read.

The Length of a Tape Measure, the Angle of a Disc

The finished work consists of two pie charts and one bar graph. The pie chart comes in a card version and an acrylic version.

Three charts: a card pie chart with white, light gray, and dark gray sectors; a color-coded acrylic pie chart; and a bar graph with red and blue bars in two rows

Source: David Sweeney, Tim Regan, Alex Taylor, Siân Lindley / RTD 2015 “A Pixel is Not a Pixel” (CC BY 4.0)

The pie chart can show up to five categories. Motors and gears turn five discs one at a time to make the size of each sector. Each sector emerges from behind the front plate along a 3D-printed spiral guide. The legend sits on a separate stand: category names appear on e-paper, with swatches of the same material as the pie chart placed beside them.

A legend with five material swatches of different patterns on the left, and on the right e-paper listing the category names Walking, Cycling, Public Transport, Taxi, and Car / Motorcycle

Source: David Sweeney, Tim Regan, Alex Taylor, Siân Lindley / RTD 2015 “A Pixel is Not a Pixel” (CC BY 4.0)

The bar graph has 2 rows of 12 bars, and the bars are commercially available DIY tape measures. Each is wrapped in colored masking tape and fed out one by one by a motor and a 3D-printed rubber wheel. A pair of bars, one above the other, shares a single e-paper label.

The inside of the bar graph: tape measures and motors mounted in a row beneath a black top panel

Source: David Sweeney, Tim Regan, Alex Taylor, Siân Lindley / RTD 2015 “A Pixel is Not a Pixel” (CC BY 4.0)

The correspondence between data and appearance, as summarized from the paper’s description, is as follows.

Data	Chart	Appearance / material
Proportions (vote results, share of speeding cars, etc.)	Pie chart	Angle of the sector
Telling categories apart	Pie chart	Color and material of the sector (card, acrylic)
Category names	Pie chart legend	Material swatches and e-paper text
Quantities (traffic, speed, rainfall, number of votes, etc.)	Bar graph	Length of the extended tape measure
Two datasets	Bar graph	Upper and lower rows
Time periods / categories	Bar graph labels	E-paper text
Switching between datasets	Both	Motion

The design policy is stated in RTD 2015 as follows.

To ensure maximum clarity of the displayed data, the charts have been designed to be as visually similar to their 2D versions from which they were inspired. This requires that all the mechanics and control systems are hidden when viewing from the front.

In short: to show the data as clearly as possible, the charts were made to look as similar as possible to the 2D charts they are based on, and for that reason the mechanics and control systems are hidden from the front. The paper also says paper, card, and tape measures were chosen to add warmth and a tactile quality to the data. The charts use power only when a value changes.

The cues the charts use stay within the basic ones: length, angle, and color. Hiding the mechanics from the front can also be read as a way of not breaking the familiar form.

From a Street Party to Five Weeks in the Window

The first public showing was at a street party in the summer of 2014, which about 150 people attended. The dates disagree: the project blog says July 27, while the CHI 2015 paper says “August 2014.” In September the work was exhibited at the V&A’s Digital Design Weekend in London, where about 13,000 visitors came over the weekend.

On November 20, 2014, the charts were placed in the lab’s ground-floor window. According to the announcement, the exhibit ran “until Christmas.”

Inside the lab’s glass window: two pie charts on tripods and a bar graph on a stand, with tall signs inviting people to vote

Source: Tenison Road project (Microsoft Research) / tenisonroad.com “VOTE NOW!”

According to the CSCW 2017 paper, the window display ran at the same time as a five-week study in which Bullfrog and PosterVote were distributed. The pie charts showed, in real time, the results of BallotShare’s “question of the day.” The bar graph showed the weekly question about the neighborhood, combining votes from PosterVote and BallotShare.

CHI EA 2015 lists three findings from the showings at the street party and the V&A. After reporting that people enjoyed seeing their votes appear in the graph straight away, it describes the cost of simplicity.

However, their simplicity also losses something of the fluid and emergent qualities of data, how data comes to mean different things to different people, at different times. The comments, stories and criticisms get lost in pairing down the results.

In short: simplicity loses the fact that data means different things to different people at different times. In paring the results down, comments, stories, and criticisms are lost.

The results after the charts went into the window are in CSCW 2017.

the potential to find Special Cards at the posters, or view results displayed on the Physical Charts, was acted upon by only a minority of Bullfrog participants. Most participants simply did not encounter the posters when taking their usual routes through the area, and seemed reluctant to change their habits to accommodate them.

In short: only a minority of Bullfrog participants looked for the Special Cards at the posters or viewed the results on the charts in the window. Most did not come across the posters on their usual routes, and did not seem willing to change their habits.

How Third Parties Saw the Work

Physical Charts was shortlisted in the “Data visualization” category of the Information is Beautiful Awards 2014. There is no record of a win, and no jury comments could be found. So a physical version of the familiar pie chart and bar graph remained among the finalists of a visualization award.

In the research literature, it is cited as an example of dynamic bar charts and pie charts. Taher, Jansen, and colleagues write the following in their IEEE TVCG (2017) paper.

For example, Microsoft’s Physical Charts [26] demonstrates dynamic bars and pie charts

That is, Physical Charts is placed as an example of shape-changing technology that shows moving bars and pie charts. It does not mention the community or participation context.

The authors themselves report the doubt about the effort involved. The conclusion of RTD 2015 says that viewers were certainly pleased but the effect is hard to state, and quotes the visualization designer Moritz Stefaner.

As Moritz Stefaner said “that is a lot of work for a pie chart”.

In short: that is a lot of work for one pie chart. This is hearsay quoted in the authors’ paper, and the author of this article has not been able to confirm when or where Stefaner said it.

Takeaways

The first thing to take from this work is the option of using physicalization not for novelty but to keep a form people are used to reading. It can be read as a design that makes a chart readable without instruction stand out away from screens.

The second is to build the legend out of materials. The legend swatches are the same material as the pie chart, so it can be read as making it easy to check by eye how they correspond to the chart itself.

The third is that choosing a location does not let you choose people’s routes. The authors chose a window visible from the street. Even so, only a minority of Bullfrog participants reached the results, and most did not come across the posters on their usual routes. The same can be said of the charts in the window, as far as one can read. For visualizations placed in public spaces, it is worth finding out even who walks past them.

The fourth is to prepare a place to receive the voices that dropped out because of the simplification. The authors write that residents need a light-weight way to add explanations and questions to the data. It can be read as: the easier a pie chart is to read, the more a mechanism to pick up voices outside it is needed.

References

Primary sources

Sweeney, D., Regan, T., Taylor, A., Lindley, S. (2015). “A Pixel is Not a Pixel: Simple Data Charting Made Physical.” Proc. RTD 2015, Article 24. doi:10.6084/m9.figshare.1327998 (CC BY 4.0): https://figshare.com/articles/journal_contribution/RTD2015_24_A_Pixel_is_Not_a_Pixel_Simple_Data_Charting_Made_Physical/1327998 / https://ast.io/archive/download/2228/pixel-not-pixel-2015.pdf
Regan, T., Sweeney, D., Helmes, J., Vlachokyriakos, V., Lindley, S., Taylor, A. (2015). “Designing Engaging Data in Communities.” CHI EA ‘15, pp. 271–274. doi:10.1145/2702613.2725432: https://ast.io/archive/download/2226/designing-engaging-data-2015.pdf
Taylor, A. S., Lindley, S., Regan, T., Sweeney, D. (2014). “Data and life on the street.” Big Data & Society 1(2). doi:10.1177/2053951714539278: https://ast.io/archive/download/2231/data-life-street-2014.pdf
Taylor, A. S., Lindley, S., Regan, T., Sweeney, D., Vlachokyriakos, V., Grainger, L., Lingel, J. (2015). “Data-in-Place: Thinking through the Relations Between Data and Community.” CHI ‘15, pp. 2863–2872. doi:10.1145/2702123.2702558: https://ast.io/archive/download/2070/data-in-place-2015.pdf
Lindley, S. E., Thieme, A., Taylor, A., Vlachokyriakos, V., Regan, T., Sweeney, D. (2017). “Surfacing Small Worlds through Data-In-Place.” Computer Supported Cooperative Work 26(1-2), 135–163. doi:10.1007/s10606-017-9263-3: https://openaccess.city.ac.uk/id/eprint/18505/1/surfacing-small-worlds-2017.pdf
Regan, T., Sweeney, D. “Reclaiming Charts: of Physicality and Small Data.” (contribution to V&A Digital Design Weekend 2014): https://digitransglasgow.github.io/OpenCollaborativeMaking/contributions/01.html
David Sweeney, “Visualisation Machines” (portfolio): https://davidasweeney.myportfolio.com/visualisation-machines
Tenison Road project (site restored from the Wayback Machine): https://tenisonroad.com/ / https://tenisonroad.com/project.html
Alex Taylor, “Announcing Tenison Road launch” (2013-09-27): https://ast.io/archive/tenison-road/
Microsoft Research HXD Design, “Physical pie charts” / “Physical bar charts” / “Physical Charts workings” (Vimeo, 2014-09-13): https://vimeo.com/106053695 / https://vimeo.com/106053696 / https://vimeo.com/106053697
dataphys.org, “Dynamic Physical Charts Display Community Data” (contributed by David Sweeney): https://dataphys.org/list/dynamic-physical-charts-to-communicate-community-data/

Secondary sources

Information is Beautiful Awards 2014, “Physical Charts” (longlist and shortlist in the Data visualization category): https://www.informationisbeautifulawards.com/showcase/504-physical-charts
Taher, F., Jansen, Y., Woodruff, J., Hardy, J., Hornbæk, K., Alexander, J. (2017). “Investigating the Use of a Dynamic Physical Bar Chart for Data Exploration and Presentation.” IEEE TVCG 23(1), 451–460. doi:10.1109/TVCG.2016.2598498: https://purehost.bath.ac.uk/ws/files/211082383/VIS16_PhysicalBarChart.pdf (In this public version, the reference number [26] in the quoted passage points to a different work, so confirmation in the published version is needed.)
Moritz Stefaner’s remark “that is a lot of work for a pie chart” (hearsay quoted by the authors in the conclusion of RTD 2015): https://ast.io/archive/download/2228/pixel-not-pixel-2015.pdf

Which Data Goes on 100 Bars? EMERGE, a Dynamic Physical Bar Chart

Thu, 05 Nov 2026 00:00:00 +0900

EMERGE is a physical bar chart whose 10×10 bars move up and down. It was built by a research group at Lancaster University in the UK. Because it can show only 100 bars at a time, operations that decide which data goes onto the bars, such as hiding, reordering, and scrolling, became a large part of the design.

The first paper is “Exploring Interactions with Physically Dynamic Bar Charts” (CHI 2015) by seven authors including Faisal Taher and Jason Alexander. A follow-up observational study appeared in IEEE TVCG (2017; presented at VIS 2016), and a study of multi-person use appeared at CHI 2023. What follows is based on these three papers and the university’s press release.

A Platform of 100 Rising and Falling Bars

EMERGE’s design follows inFORM, the shape display from MIT. According to the 2015 paper (CHI15 below), the differences are the addition of LEDs, the control mechanism, and the software model.

Plastic bars are arranged in a 10×10 grid, each connected to a motorized slide potentiometer. The travel is 100 mm. The slider below and the bar above are joined by a push rod running through a hollow tube. The sliders are stacked in two tiers, which keeps the footprint to 0.55 × 0.44 × 1.20 m. An RGB LED under each bar lights it up.

A dark room where a 10×10 grid of transparent bars glows at different heights, with year labels and the text “Exports of Goods and Services (% of GDP)” shown on the surface in front

Source: Lancaster University, “Shape changing display could spell the end for the 2D graph” (press release, April 13, 2015)

When a person pushes or pulls a bar, the potentiometer value changes. CHI15 used this to detect interaction. Axis labels and control buttons were projected from above onto the surface around the bars, and a Microsoft Kinect picked up touches. In the 2017 paper (TVCG17 below), touchscreens were placed along the four sides of the top surface. Opposite sides show the same information, so the display can be operated from any side.

Left, a hand pressing a bar with a finger; center, a structural diagram showing the bars, push rods, the tiers of sliders, and the touchscreens on four sides; right, a photo of the unit from the side

Source: Miriam Sturdee, Hayat Kara, Jason Alexander, “Exploring Co-located Interactions with a Shape-Changing Bar Chart” (CHI 2023, CC BY). The central diagram is reprinted in that paper from CHI15 (© 2015 ACM).

The mapping between data and appearance is as follows.

Data / state	Physical property	Details
Value	Bar height	0–100 mm
Row category (country, region)	Bar color (RGB LED)	A different color for each row
Column category (year, topic)	Depth position and axis label	Labels projected; later on touchscreens
Vertical axis scale	None	Deliberately omitted in CHI15
A selected point	Light	Dims the others and gives the intersection a different color
Emphasis	Pulling a bar up	Dims all bars other than the pulled one
Hide	Pushing a bar down	Hides the pressed point or row
Visible range	A 10×10 window	Moved by scrolling (CHI15, TVCG17)
Locked row	Position within the window	Stays when scrolling (TVCG17)
Saved view	The whole arrangement of bars	Saved and recalled later (TVCG17)

This suggests that color is used only to show which row a bar belongs to, and quantity is shown by height alone. As for the vertical axis, CHI15 itself writes that having one could change interaction and behavior.

This research also has a chapter in the book Making with Data (edited by Samuel Huron, Till Nagel, Lora Oehlberg, and Wesley Willett; A K Peters/CRC Press, 2022). It is section 4.4, “EMERGE,” in Part 4, “Actuation,” which Crossref lists as pp. 287–298, doi:10.1201/9781003264903-19. The chapter’s four authors are Alexander, Taher, John Hardy, and John Vidler. This is a different lineup from the seven CHI15 authors, and Vidler appears in CHI15 only in the acknowledgments.

Trying 14 Interactions: The 2015 CHI Paper

CHI15 opens like this:

Visualizations such as bar charts help users reason about data, but are mostly screen-based, rarely physical, and almost never physical and dynamic.

In short: most bar charts live on screens, and physical, moving ones are almost nonexistent. The question is how such bar charts should be operated to make exploration work well. Because the field is young, the authors avoided a comparative experiment and looked at participants’ behavior and impressions rather than accuracy. There were 17 participants.

The tasks were four types based on the taxonomy of Heer and Shneiderman, with 14 interactions in total.

Task	Interactions tried	Most preferred interaction (mean on a 5-point scale)
Mark	Point, Pull, Press	Press (4.29)
Filter	Swipe away, Manual press, Assisted press, Press to compare, Press shortcut	Press shortcut (4). Pressing the first bar of a row hides the whole row
Sort	Drag and drop, Press to swap (each in two movement styles)	Dragging a label (3.88)
Navigate	Scroll bar, directional arrows, Directional press, paging	Scroll bar (4)

Of the four tasks, filtering, sorting, and navigating can be read as interactions that change which bars show what. A reason is also given for “hiding.” On paper bar charts, neighboring bars distort height judgments the most (Zacks et al., 1998), so removing unneeded bars should make comparison easier.

When asked beforehand how they would do each task, all participants answered with touchscreen-style interactions such as swiping or tapping a label. For filtering, however, once they tried interactions that touch the bars directly, their ratings shifted toward the physical side. Even so, the paper concludes that each has its place.

smaller interactions such as annotation afforded physical interaction, whereas larger motions such as row organization afforded touch-screen style swipe interactions.

In short: small interactions like marking suit physical interaction, while large motions like reordering rows suit touchscreen style.

The animation in which rows slowly move during sorting was not liked. When scrolling, the values for the years in between continued to appear on the bars, and every participant who viewed UK regional rainfall (1920–2012) noticed that the north gets more rain. Three of the 17 participants moved around the unit while operating it.

As a limitation, the paper notes that 10×10 shows little data. The country data came from the materials page of Jansen et al.’s CHI 2013 paper (aviz.fr/phys). Yvonne Jansen is also a co-author of CHI15.

Viewing 31×46 Data Through a 10×10 Window

TVCG17 is an observational study by the same group. The 16 participants had never used a dynamic physical display or an interactive on-screen visualization tool. They explored European values survey data for about 16 minutes on average, then presented for about 5 minutes on average.

Six interactions were implemented.

Highlighting: pull a bar up to emphasize it
Swapping: drag labels to exchange rows
Scrolling: move through data larger than 10×10
Locking: fix a row so it stays in the window while scrolling
Hiding: push a bar down to hide it. Pressing two points at the edge at the same time keeps only those two rows so they can be compared
Snapshot: save the current view and recall it later

The data was 31 topics × 46 countries, and EMERGE can show only 10×10 at once. As a result, nobody looked at the data as a whole.

Heatmaps for each of the 16 participants showing, in shades of blue and red, which parts of the 31×46 data they viewed. The square at the lower left is the first 10×10 window

Source: Faisal Taher et al., “Investigating the Use of a Dynamic Physical Bar Chart for Data Exploration and Presentation” (IEEE TVCG, 2017, author’s final version), Fig. 4

Seven of the 16 felt there was “too much data.” On the other hand, 15 said a 10×10 window was enough. The authors acknowledge this discrepancy and list, as future work, bringing the focus+context techniques used in on-screen visualization to physicalization.

In the presentations, 9 of the 16 showed their saved snapshots in sequence. Nine also hid unneeded data to emphasize what remained, or used hidden bars as dividers between groups. This suggests that a state in which they had chosen and arranged what appeared in the window became the material for the presentation itself.

How people used their bodies was recorded too. Everyone moved between at least two sides, and four went around three sides. Movements such as peering in, tilting the head, and crouching appeared often right after scrolling. The authors write that because a 3D object looks different from different angles, multiple viewpoints help in checking relationships.

Four photos: a person pointing at a label and a bar, a person crouching to view the bars from the side, a person holding a palm over the bars, and an arrangement divided into three by hidden bars

Source: Faisal Taher et al., “Investigating the Use of a Dynamic Physical Bar Chart for Data Exploration and Presentation” (IEEE TVCG, 2017, author’s final version), Fig. 1

Bars in the center were hard to reach when surrounded by tall bars, so interaction concentrated on the edge bars. Fifteen participants said the data was easy to read, while one said it was hard to present without exact numbers.

It is not known what the body movements are for. The authors write:

It remains an open question as to whether body movements are correlated with better insights from the data, if they allow participants to read data more accurately, or if they compensate for occlusion.

In short: whether they lead to better insights, serve accurate reading, or make up for hidden parts is unknown. The authors add that the unit should be placed where people can walk around it, not against a wall.

The 2023 paper (Miriam Sturdee, Hayat Kara, Alexander) repeated the same procedure with 24 participants in pairs and groups of four. Their exploration covered a narrower range than with one person, and they stayed in one window longer. The paper says that a single person could compensate for hidden parts by moving around, but groups and pairs could not.

“The End of the 2D Graph” and the Paper’s Self-Assessment

On April 13, 2015, Lancaster University issued a press release about EMERGE. Its headline was “Shape changing display could spell the end for the 2D graph,” which says it could be the end of the 2D graph. The body text is stronger, saying it is “sounding the death knell for the two dimensional bar chart,” that is, announcing the end of the 2D bar chart.

The papers are written differently. TVCG17 describes its own study this way:

Our study was not a comparative experiment, and we therefore cannot draw any conclusions about how visualizations and physicalizations compare.

In short: because it was not a comparative experiment, no conclusion can be drawn about which is better, on-screen visualization or physicalization. This suggests the paper does not claim to be better than 2D.

The comparative experiment was done by Jansen et al. at CHI 2013. A static 3D bar chart was read faster as a physical object than on screen, but did not match a 2D on-screen display (Jansen et al.’s evaluation experiment). EMERGE can be read as the next step, asking “if it can move, how should it be operated?” Physical Charts in the same series is also a moving bar chart, but it sits inside a lab’s glass window and is shown to people on the street. EMERGE is a lab prototype that viewers operate by pushing and pulling the bars.

How Others Viewed EMERGE

The points of evaluation differ by source.

Bae et al. (CHI 2022) place EMERGE alongside Zooids, under the argument that the breadth of interaction determines the depth of analysis. It is an example that builds filtering, reconfiguring, abstraction and elaboration, and selection into a 2.5D display.
Djavaherpour et al. (Computer Graphics Forum 2021) put EMERGE in the “statistical data” category of their physicalization taxonomy, describing it as a responsive bar chart with bars and RGB LEDs. Jansen is a co-author.
Sauvé, Potts, Alexander, and Houben (CHI 2020) cite EMERGE as motivation for an experiment on orientation and perception: it is unclear whether walking around compensates for occlusion or serves accurate reading. Alexander is a co-author.

The latter two include people from the creators’ side, so they are not entirely third-party evaluations. In press coverage, ScienceDaily reposted the press release.

Takeaways

EMERGE suggests that in physical visualization, the upper limit on how much can be shown is set first. This prototype has 100 bars. So the design turned toward operations for deciding what to assign to the limited bars.

The same applies on screens when a lot of data is placed in a small area. I read three takeaways. First, pair a means of looking outside the window with a means of keeping things fixed inside it. Second, choose the means by the size of the interaction: to select one point, touch that point; to move a row, move its label. Third, place a 3D display where viewers can walk around it.

References

Primary sources

Faisal Taher, John Hardy, Abhijit Karnik, Christian Weichel, Yvonne Jansen, Kasper Hornbæk, Jason Alexander, “Exploring Interactions with Physically Dynamic Bar Charts,” CHI 2015, pp. 3237–3246. PDF / Lancaster eprints / doi:10.1145/2702123.2702604
Faisal Taher, Yvonne Jansen, Jonathan Woodruff, John Hardy, Kasper Hornbæk, Jason Alexander, “Investigating the Use of a Dynamic Physical Bar Chart for Data Exploration and Presentation,” IEEE TVCG 23(1), 2017, pp. 451–460. Author’s final version / Lancaster eprints / doi:10.1109/TVCG.2016.2598498
Miriam Sturdee, Hayat Kara, Jason Alexander, “Exploring Co-located Interactions with a Shape-Changing Bar Chart,” CHI 2023. PDF / University of Bath / doi:10.1145/3544548.3581214
Lancaster University, “Shape changing display could spell the end for the 2D graph,” 2015-04-13. Press release
J. Heer, B. Shneiderman, “Interactive Dynamics for Visual Analysis,” Communications of the ACM 55(4), 2012, pp. 45–54 (cited by CHI15; the source of the task taxonomy)
J. Zacks, E. Levy, B. Tversky, D. J. Schiano, “Reading bar graphs: Effects of extraneous depth cues and graphical context,” Journal of Experimental Psychology: Applied 4, 1998, pp. 119–138 (cited by CHI15; the rationale for “hiding”)
Yvonne Jansen, Pierre Dragicevic, Jean-Daniel Fekete, “Evaluating the Efficiency of Physical Visualizations,” materials page for CHI 2013 (cited by CHI15 as the source of the country data). aviz.fr/phys
Jason Alexander, “EMERGE Capabilities Demonstration” (YouTube). Video
Jason Alexander, Faisal Taher, John Hardy, John Vidler, “EMERGE,” Samuel Huron, Till Nagel, Lora Oehlberg, Wesley Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, A K Peters/CRC Press, 2022, 4.4. Taylor & Francis / Crossref / Routledge (table of contents)

Secondary sources

S. Sandra Bae, Clement Zheng, Mary Etta West, Ellen Yi-Luen Do, Samuel Huron, Danielle Albers Szafir, “Making Data Tangible: A Cross-disciplinary Design Space for Data Physicalization,” CHI 2022. arXiv
H. Djavaherpour et al. (Y. Jansen is a co-author), “Data to Physicalization: A Survey of the Physical Rendering Process,” Computer Graphics Forum 40(3), 2021. arXiv
Kim Sauvé, Dominic Potts, Jason Alexander, Steven Houben, “A Change of Perspective: How User Orientation Influences the Perception of Physicalizations,” CHI 2020 (Alexander is a co-author). PDF
ScienceDaily, “Shape changing display could spell the end for the 2-D graph,” 2015-04-13 (reprint of the press release). Article

Turning Housing Costs Upside Down and Sinking Them in Water: Ekene Ijeoma's Wage Islands

Thu, 05 Nov 2026 00:00:00 +0900

Ekene Ijeoma’s Wage Islands is a map of New York City sunk in water, with elevation set so that the cheaper the housing, the higher the ground. The part that rises above the water surface, which stands for a wage, shows the area where housing is within reach at that wage. Press a button and the wage rises, lifting the map out of the water.

The work was shown in 2015 in the group exhibition “Measure” at the Storefront for Art and Architecture in SoHo, New York. What follows draws on the introductions in Domus and designboom and on what Ijeoma said in an interview with Vice. The artist’s own description of the work was on his old official site, which would not open for me, so I could only check it through a reposted copy. I also could not confirm the source of the data used for the map (the survey name, the year, the geographic unit).

From a minimum-wage campaign to a question about where you can live

It began at Re3Storyhack, a hackathon on social issues. In an interview with Vice (The Creators Project, September 8, 2015), Ijeoma says:

At Re3Storyhack I worked with Fast Food Forward who have been fighting to raise minimum wages for fast food workers to 15 dollars an hour. I thought, what if the fight was framed around access not just wages?

He teamed up with Fast Food Forward, which was campaigning to raise the minimum wage for fast food workers to $15 an hour, and wondered whether the fight could be framed as a matter of “access” and not only of wages. He goes on to recall the cover of New York magazine after Hurricane Sandy, with Manhattan in the dark, and says he wanted to produce that intuitive feeling in a work based on data.

The artist’s description appears to have been on his old official site, studioijeoma.com, but I could not open it. The passage below is from a repost on the “artresearch” board at scrapbox.io.

We wanted to see how their access to the city would change based on housing costs. What would the city look like if its’ islands were based on wages? How much would it take for them to have access to all of the city?

These are the questions: how does the part of the city people can live in change with housing costs? What would the city look like if its islands were determined by wages? How much would it take to reach the whole city? The same reposted text also says the work was commissioned for the “Measure” exhibition.

Making housing costs elevation, cheapest highest

The Domus introduction (September 15, 2015) describes how the map works:

At the base is the full city where an hourly wage of $77 is needed for monthly housing costs of $4001. At the peaks are only parts of the city where an hourly wage of $8.75 is needed for monthly housing costs of $271.

The base is the whole city, at an hourly wage of $77, corresponding to monthly housing costs of $4,001. The top is the level at $8.75 an hour, corresponding to $271 a month, and it covers only part of the city. The source of the elevation is “median monthly housing costs,” and the range runs from $271 to $4,001 (Domus, designboom). The reposted text by the artist sums this up in one line: “The higher the level the lower the wages.”

The $8.75 an hour matches the minimum wage in New York State at the time. The state’s general minimum wage became $8.75 on December 31, 2014, and rose to $9.00 on December 31, 2015 (New York State Department of Labor).

The correspondences the sources let me establish are as follows.

Data	Appearance / material	Value	Source
Geography of New York City	Planar shape of a terrain model made of stacked transparent acrylic	500+ laser-cut parts	Vice, Domus, repost of the artist’s text
Median monthly housing costs	Elevation (inverted): cheaper areas are higher	$271 at the top, $4,001 at the base	Domus, designboom
Hourly wage corresponding to housing costs	Scale of the elevation	$8.75 at the top, $77 at the base	Domus
Current wage (set by the viewer)	Height of the map relative to the water surface	From $8.75 to $77	Domus, Vice, repost of the artist’s text
Areas where housing is affordable at that wage	The part above the water surface (islands)	At minimum wage, only the peaks	Domus
Areas that are out of reach	The part hidden under the dark water	—	designboom, Vice, repost of the artist’s text
Wage and area figures	LCD display	—	Domus, Vice

On an ordinary topographic map, high ground is read as “more.” This work reverses the direction and makes the cheaper areas higher. This can be read as a way of making everything above the water surface mean “within reach.”

Only the peaks of a terrain built from stacked transparent acrylic rise like islands out of dark water

Courtesy of Ekene Ijeoma. From designboom, “wage islands installation by ekene ijeoma highlights new york’s salary disparity”

I could not confirm how the correspondences between $271 and $8.75, and between $4,001 and $77, were decided. What follows is my own calculation. Assume 40 hours a week for 52 weeks (2,080 hours) and 30% of income going to housing. Then $77 an hour gives about $4,004 a month, which nearly matches the $4,001 at the base. The same formula gives about $455 a month for $8.75, which does not match the $271 at the top. This can be read as the top end being set by a different method.

The source of the data is also unknown. “Median monthly housing costs” is the same as the name of a table in the American Community Survey (ACS), but I found no statement that ACS was used. In Vice, Ijeoma says Richard Dunks and Juan Francisco Saldarriaga helped with the modeling of the geographic data.

Press the button and the map rises from the water

The work consists of a tank filled with dark water and a box with a button in front of it. designboom calls the water “black water,” and the artist’s reposted text calls it “dark blue water.”

A transparent tank of dark water on a white stand. The upper part of a terrain model rises above the surface; in front is a black box with a button and a display, and on the wall behind are the artist’s name and a description

Courtesy of Ekene Ijeoma. From designboom, “wage islands installation by ekene ijeoma highlights new york’s salary disparity”

What moves is not the water but the map. Vice writes, “A button moves the physical map up and down in the water.” The reposted text by the artist explains the mechanism as follows.

Inside was a lift and outside was an display with a button. Holding down on the button slowly raised the wages shown on the display and lifted the map to show the areas of access to housing above the water. When let go it lowered the map to the bottom showing only minimum wage.

There is a lift inside the tank and a display with a button outside. Holding the button slowly raises the wage on the display and lifts the map, so the areas where housing is within reach come out above the water. Letting go lowers the map back to the minimum-wage state.

The LCD on a black box reads “HOURLY WAGE: $23.50” and “AREA: 46.00 SQ MI”, with a red button below

Courtesy of Ekene Ijeoma. From designboom, “wage islands installation by ekene ijeoma highlights new york’s salary disparity”

The LCD in this photo shows an hourly wage of $23.50 and an area of 46.00 square miles. Domus says the display shows the wage and the area. The height of the map alone does not let you read the numbers, so the display takes on that part. That can be read as the division of labor here.

On the making, Ijeoma told Vice that he glued together about 500 laser-cut parts himself, and that the hardest part was keeping water from leaking into the motor. A video of the work is on Vimeo (https://vimeo.com/138549946), but I could not open it and have not checked its contents.

On how it looks, Ijeoma says (Vice):

There’s something about seeing the clear abstracted forms of the city coming out of the dark clouded water, and feeling the concreteness of the message

Seeing the clear, abstracted forms of the city come out of dark, clouded water, he says, makes the message feel concrete. The areas that are out of reach are not erased but sunk under the water. This can be read as using hiding itself as the means of display.

In front of a wall panel, a transparent terrain model emerges largely from the dark water of the tank

Courtesy of Ekene Ijeoma. From designboom, “wage islands installation by ekene ijeoma highlights new york’s salary disparity”

designboom captions this photo “as the wages increase the terrain emerges from the water.”

Other editions, and a book chapter

Wage Islands: Immigrants (2017) appears in “Imagining Data,” a virtual exhibition at the Museum of Craft and Design in San Francisco. Its materials are acrylic, ink, and water, and it measures 18 × 3 × 18 inches. According to the museum’s description, it is based on data about low-wage immigrant workers, and the peaks above the water show where they can live.

In print, there is a chapter in Making with Data: Physical Design and Craft in a Data-Driven World (edited by Samuel Huron, Till Nagel, Lora Oehlberg, and Wesley Willett, CRC Press). It is section 3.2, “Wage Islands,” in Part 3, “Digital Production.” The author is Ijeoma, the pages are 199–210, and the DOI is 10.1201/9781003264903-12. I have not been able to read the text of the chapter.

The abstract listed in OpenAlex as publisher metadata describes the work as “a series of data-driven sculptures.” It also says the data source, the number of map layers, the dimensions, and the mechanism of operation differ from edition to edition. This suggests that Wage Islands is treated as a series built on the same method.

How third parties saw the work

dataphys.org, a list of physical visualizations by data physicalization researchers (run by Pierre Dragicevic and Yvonne Jansen), introduces the work this way:

The “Wage Islands” installation by Ekene Ijeoma makes clever use of water as a data query device.

It says the work makes clever use of water as a data query device. It is classified as “Active physical visualization.” This is a remark that sees the water not as a material but as an operation that filters by a condition.

Marina Garcia-Vasquez of Vice sees it as a tool for civic debate.

The exhibit is thus a heady display of information turned into a healthy experiment in visualizing complex civic issues.

It is a healthy experiment in visualizing complex civic issues. The same article adds that the production process itself, built with many collaborators, can be seen as “a physical collective bargaining.”

The Museum of Craft and Design’s description (guest curator: c2-curatorsquared) sees it as an entry point to the subject.

By transforming the impersonal data into interactive sculptural form, Ijeoma is creating an alluring entry point for deeper discussion.

By turning impersonal data into a form you can touch, the work creates an entry point to discussion.

ArchDaily (January 24, 2016) presented it as a city that sinks, under the title “Watch New York City Sink…”. The biography on the artist’s current site also says “submerged in water at rising rent costs.” The mechanism lifts the map, yet the way it is talked about leans toward “sinking.” That is how it can be read.

Uses

There are three things to take away from this work.

First, it turned filtering by a threshold into a single physical movement. Showing only what meets a condition, which on a screen would be a filter, becomes raising and lowering a map. The dataphys.org remark about a “query device” can be read as pointing to this.

Second, it chose the direction of the scale to suit what it wants to show. Making cheaper areas higher means that everything above the water is “within reach.” But because this reverses how an ordinary topographic map is read, it needs an explanation.

Third, it leaves the source of the data with the work. In this case, neither the name of the housing-cost survey nor how the $271 and $8.75 were matched could be confirmed from public materials. A physical work is hard to re-check for its numbers afterward. It is worth writing the source and the method of calculation into the description of the work.

References

Primary sources

dataphys.org, “Wage Islands” (entry in the List of Physical Visualizations): https://dataphys.org/list/wage-islands/
Ekene Ijeoma, “Wage Islands” (old official page; I could not connect and have not checked it): https://studioijeoma.com/Wage-Islands
Ekene Ijeoma official site (biography, CV): https://ekeneijeoma.com/about
Repost of the artist’s description (scrapbox.io “artresearch”; original not checked): https://scrapbox.io/artresearch/Wage_Islands
Marina Garcia-Vasquez, “Turning New York’s Salary Gap into an Interactive Sculpture,” Vice / The Creators Project, 2015-09-08 (interview with the artist): https://www.vice.com/en/article/turning-new-yorks-salary-gap-into-an-interactive-sculpture/
Domus, “Wage Islands,” 2015-09-15: https://www.domusweb.it/en/news/2015/09/15/ekene_ijeoma_wage_islands.html
Philip Stevens, “wage islands installation by ekene ijeoma highlights new york’s salary disparity,” designboom, 2015-11-03: https://www.designboom.com/architecture/ekene-ijeoma-wage-islands-storefront-for-art-and-architecture-chicago-architecture-biennial-11-03-2015/
“Wage Islands: Ekene Ijeoma,” WNW Magazine, 2015-09-15: https://wnwblog.squarespace.com/magazine/2015/9/15/g89og36ne7cxoacofbu8nl9zxtneah
Press release for Storefront for Art and Architecture’s “Measure” (repost by World Architecture Community): https://worldarchitecture.org/architecture-news/ccemz/storefront-for-art-and-architecture-presents-measure-opening-august-13th.html
Museum of Craft and Design, “Ekene Ijeoma” (Wage Islands: Immigrants): https://sfmcd.org/project/ekene-ijeoma/ / exhibition “Imagining Data”: https://sfmcd.org/exhibitions/imagining-data/
Ekene Ijeoma, “Wage Islands,” in Samuel Huron, Till Nagel, Lora Oehlberg, Wesley Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, CRC Press, pp. 199–210. doi:10.1201/9781003264903-12 / book page: https://www.routledge.com/Making-with-Data-Physical-Design-and-Craft-in-a-Data-Driven-World/Huron-Nagel-Oehlberg-Willett/p/book/9781032182223 / table of contents: https://makingwithdata.org/press / https://makingwithdata.org/about
Chapter metadata and abstract: https://api.crossref.org/works/10.1201/9781003264903-12 / https://api.openalex.org/works/doi:10.1201/9781003264903-12 / chapter page (not checked): https://www.taylorfrancis.com/chapters/edit/10.1201/9781003264903-12/wage-islands-ekene-ijeoma
New York State Department of Labor, “History of the Minimum Wage in New York State”: https://dol.ny.gov/history-minimum-wage-new-york-state
Video of the work (Vimeo; not checked): https://vimeo.com/138549946

Secondary sources

dataphys.org, “Wage Islands” (added by Cedric Honnet, contributed by Cheng Xu): https://dataphys.org/list/wage-islands/
Marina Garcia-Vasquez, Vice / The Creators Project, 2015-09-08: https://www.vice.com/en/article/turning-new-yorks-salary-gap-into-an-interactive-sculpture/
Museum of Craft and Design, “Imagining Data” (c2-curatorsquared): https://sfmcd.org/project/ekene-ijeoma/
Sabrina Santos, “Watch New York City Sink in This Commentary on Affordable Housing,” ArchDaily, 2016-01-24: https://www.archdaily.com/780344/watch-new-york-city-sink-in-this-commentary-on-affordable-housing

White Beads for Satisfied, Red for Unsatisfied: The Bar Chart in the Antell Cafeteria at the University of Helsinki

Tue, 03 Nov 2026 00:00:00 +0900

At the exit of an Antell cafeteria on the University of Helsinki campus, there was a physical bar chart. Customers put in a white bead (satisfied) or a red bead (unsatisfied).

The source is the entry “Physical Customer Satisfaction Survey” on dataphys.org (List of Physical Visualizations), a site that collects physical visualizations. Pierre Dragicevic added it in March 2015. The text is two sentences long, and the photo is a single image taken by Luana Micallef. This article follows those two sentences and the photo.

What stood on the table by the exit

The text of the entry reads as follows.

Participatory physical bar charts at the exit of the Antell cafeteria at the University of Helsinki. White beads represent statisfied customers, red beads are unsatisfied ones.

In short: participatory physical bar charts at the exit of the Antell cafeteria at the University of Helsinki, where white beads stand for satisfied customers and red beads for unsatisfied ones. “statisfied” is the spelling in the original.

The Source field says “Antell”, and the link goes to Antell’s home page. It contains no description of the device. Since the Source field names Antell, the device can be read as belonging to the Antell outlet rather than being run by the university. Micallef, who took the photo, was in 2015 a visualization researcher affiliated with the Helsinki Institute for Information Technology (HIIT).

On the table at the exit stand a slender glass tube with white beads collected at the bottom, a square glass bowl holding white and red beads, a card showing a smiling face and a frowning face, and a box for Antell’s customer survey

Photo by Luana Micallef / From dataphys.org

Here is what appears in the photo, starting from the front.

There are two square glass bowls: the left one holds white beads, the right one red beads.
On the card in the transparent stand behind them, a white smiling face and a red frowning face sit side by side under the Antell logo.
On the left stands a slender glass tube, with white beads collected at the bottom.
The sign behind the tube poses a question in Finnish and English. Part of it is hidden by the tube, but the English can be read as “How did we serve you today?”
At the back is a white box labeled “ASIAKASPALAUTE”, and a stack of paper lies at the front left.

The two faces on the card can be read as a legend for the bead colors. The two bowls in front can be read as the place where customers pick up their beads. Only one tube appears in the photo, and the beads in it are all white. The text says “bar charts” in the plural, but the photo does not show where any other tubes were.

What was mapped to color and height

The table below lays out the mapping between data and physical properties, together with the basis for each.

Data	Physical property	Basis
Customer’s rating (satisfied / unsatisfied)	Bead color (white / red)	Stated in the text
Meaning of the colors	The card’s white smiling face and red frowning face	Visible in the photo
Who answers	Customers put in the beads themselves	“Participatory” in the text
When they answer	Placement (the cafeteria exit)	Stated in the text
Number of responses	Height of the beads piled in the tube	Read from “bar charts” in the text and the tube in the photo
One respondent	One bead	Not described in the source

The only mapping stated in the text is color. Reading counts from height comes from the name “bar chart” and from the tube in the photo. The source does not say how many beads one person puts in, or what the bars were divided by. The installation period, the number of beads collected, the ratio of satisfied to unsatisfied, and who designed the device are also unknown from the source.

dataphys classifies this device as a Passive physical visualization, a category of physical visualizations that use no power or electronic components. The only thing that updates the display can be read as the hands of the customers dropping in beads.

dataphys lists it alongside devices of the same kind

The entry’s Related field lists three other works alongside this device.

Related: Also see similar contraptions by Hans Haacke (1970), Lucy Kimbell (2006), and Jennifer Payne (2014).

It is a single sentence asking readers to also see similar contraptions by Haacke (1970), Kimbell (2006), and Payne (2014). The Related field calls the three “similar contraptions”, which suggests that dataphys lists this case as the same kind. This device can be read as a tool the outlet set out for a customer survey (the source does not describe its purpose).

The entry also carries the participatory tag. The tag page begins with voting by pebbles in ancient Greece and says this.

The earliest participatory visualizations were probably voting systems.

In short: the earliest participatory visualizations were probably voting systems. Here is how the dataphys entries divide up the answers.

Device	Year	How answers are divided
Pebble voting (ancient Greece)	—	Pebble color (white = for, black = against). Sometimes two urns were used
Hans Haacke, MoMA Poll	1970	Position of the box (left box = Yes, right box = No)
Lucy Kimbell, Physical Bar Charts	2006	Badges are taken out, so the more votes, the lower the bar
Jennifer Payne, Physical Visual Sedimentation	2014	Gumball color separates groups within the university
Antell’s device	2015 (year of registration)	Bead color (white = satisfied, red = unsatisfied)

Haacke’s MoMA Poll divided yes and no by the position of the box (article). Antell’s device, like pebble voting, divides answers by color. Dividing by color suggests that the position of the bar could be assigned to something else. However, what this device divided by position cannot be confirmed.

dataphys’s treatment is a classification, not an evaluation. I could not find any criticism or press coverage of this device itself in a form I could read.

How it could be used

In this device, the act of answering and the act of showing the result happen on the same tube. The next customer walking out can also see the beads that have piled up so far, it can be read.

In this photo, the card with the smiling and frowning faces can be read as a legend. It can be read as a design that makes the meaning of white and red clear without reading any text.

The materials in the photo are beads, glass bowls and a tube, a sign, and a card. There are no electronic components. A device of this kind can be read as easy to assemble from materials at hand.

The source does not describe any tallying or record-keeping for this device. MoMA Poll kept the final day’s vote count with a photoelectric counter (article). As a general point, when collecting on a physical object, it helps later to decide in advance how the counts will be taken and recorded.

References

Primary sources

dataphys.org “2015 – Physical Customer Satisfaction Survey” (Added by Pierre Dragicevic, registered 2015-03-12): https://dataphys.org/list/physical-customer-satisfaction-survey/
Photo (Photo by Luana Micallef): https://dataphys.org/list/images/uploads/2015/03/antell.jpg
Antell (the link target of the entry’s Source field; it contains no description of the device): http://www.antell.fi/
The photographer’s 2015 affiliation (HIIT): https://research.aalto.fi/en/activities/creating-high-quality-figures-for-your-publications/

Comparison and background sources (related pages on dataphys.org; these are not evaluations of this device itself)

dataphys.org participatory tag: https://dataphys.org/list/tags/participatory/
dataphys.org Passive physical visualization category: https://dataphys.org/list/categories/passive-physical-visualization/
dataphys.org Pebble Voting: https://dataphys.org/list/visualizing-opinions-with-pebbles/
dataphys.org MoMA Poll (Haacke, 1970): https://dataphys.org/list/moma-poll-haackes-participatory-bar-chart/
dataphys.org Inverted Participatory Bar Charts (Kimbell): https://dataphys.org/list/physical-bar-charts/
dataphys.org Physical Visual Sedimentation (Payne, 2014): https://dataphys.org/list/physical-visual-sedimentation/

Changing the Color of a Boat's Light on the Water It Measures: SeeBoat

Thu, 05 Nov 2026 00:00:00 +0900

SeeBoat is a device that turns water-quality sensor readings into the LED color of a small boat running over the very water being measured, as it measures. Sensors for water temperature, conductivity, turbidity, and pH are mounted on a remote-controlled boat, and one chosen value is assigned to the hue of the hull’s light. In a long-exposure photograph, the boat’s path remains as lines of color.

The creator is Laura Perovich, and the work dates from her time as a doctoral student in the Object-Based Media group at the MIT Media Lab. Its predecessor is the float “Thermal Fishing Bob,” born at Public Lab, and Perovich joined in improving it from 2015. What follows draws on the repository Perovich published on GitHub (README, Arduino code, parts list), her research notes that remain on Public Lab, and exhibition captions.

Throwing in floats to draw a river with light

The starting point is the Thermal Fishing Bob on the Public Lab wiki. A cheap thermistor and an LED are put in a waterproof container, thrown into a waterway, and photographed with a long exposure. The wiki describes the result like this.

This will create a “light painting” where the colors describe the temperature of the water at that location.

In short, the result is a “light painting” in which the colors represent the water temperature at each location. It can be read as saying that the tool that measures and the tool that shows were, from the start, one and the same float.

In April 2014, Sara Wylie posted the result of drawing the edge of a pond with floats that students had built in a class. Her note says “the different groups had the Thermal Bobs set to different temperature ranges.” The temperature range assigned to the colors differed from group to group.

From June to August 2015, Perovich posted a series of notes on improvements: adding more sensors and LEDs, making the temperature sensor faster, and putting the housing in a white sphere. On August 25 she also released a version whose color is not continuous.

Thermal fishing bobs with this new code will light up red when they are above the set threshold value and will be off when the temperature is below the set threshold value.

In short, the float lights up red above the threshold and goes dark below it. The note gives, as a use for this version, searching for hot spots in a river that might exceed a regulatory limit. Only this threshold version ties red to “hot spots.” It can be read as showing that the same author tried both a continuous hue scale and an on/off binary display.

From floats to remote-controlled boats

On November 20, 2015, an early SeeBoat test was photographed near the Kendall Cogeneration Station on the Charles River.

A long-exposure photograph of a night river, with the tracks of boats left as several lines of green and yellowish light. City lights and a cable-stayed bridge are visible in the background

Source: Laura Perovich & collaborators (photo: Jorge Valdez) / untold possibilities at the last minute, “ArtBoat and SeeBoat”

The exhibition caption says the following.

The change in color from green to yellow is tied to the temperature sensor readings and indicates a potential thermal pollution source near the outflow pipe.

In short, the change from green to yellow corresponds to the temperature sensor readings and indicates a possible source of thermal pollution near the outflow pipe. The caption does not say whether green or yellow is the higher value.

The hull is a Horizon Hobby remote-controlled boat, the “Stealthwake 23” Brushed Deep-V RTR." The controller is an Adafruit Feather M0 with RFM95 LoRa Radio (900MHz), and the light source is a DotStar LED strip (white, 60 LEDs/m). The README opens by describing the work like this.

SeeBoat is a fleet of remote controlled boats that measure and display water quality in real time. It allows communities to explore water quality in their local lakes, rivers, and streams by changing the color of the boat in response to changes to the water quality.

In short, it is a fleet of remote-controlled boats that measure and show water quality in real time, changing the boat’s color as the water quality changes so that local people can explore their nearby rivers and lakes.

The Feather version of the code reads “Started June 2017.” In autumn 2017, Rod Bayliss added turbidity, pH, and conductivity, and in 2019 Talia Spitz integrated everything and made it support multiple boats.

Writing the color scale into each place

The values in the table are the defaults in the version published on GitHub (master, commit b61f5c3 of 2020-11-12). I have not checked the settings of other versions or other periods.

Data	Appearance / material	How it is mapped	Source
Water-quality value (one of four)	Hue of all the hull’s LEDs	Uses only the red-to-green hues (about 0–120°). Saturation and brightness are at maximum. All LEDs are the same color	`LED_updated.ino`
Water temperature (default)	Same as above	62.0–72.0°F (about 16.7–22.2°C) sets the two ends of the color range	`seeBoatFeather.ino`
Conductivity	Same as above	18,000–21,000 µS/cm	Same as above
pH	Same as above	7.50–8.50	Same as above
Turbidity	Same as above	50–300 NTU	Same as above
Values outside the range	The color at the end	Converted linearly to 0–255, with anything beyond clipped	`normf()` in `LED_updated.ino`
Where it was measured	The water surface where the boat is	Position is not converted into a different appearance	README
Time and path	Lines of light in a long-exposure photograph	The photographer records them	README, caption
Exact numbers	Numbers sent by radio	Viewed in an Android app. A receiving-side Python script saves them to CSV	README, `readMe.txt`
Hull	A boat painted black	Blends into the night background	Parts list
Window of light	White milky acrylic	Light shows through holes cut in the hull	Parts list

Only one value lights the LEDs per boat. It is chosen by writing one of “temperature”, “conductivity”, “turbidity”, or “pH” into the variable whichDataControlsLEDs. In the same code, the turbidity sensor is disabled by default (turbsensor = 0).

The values at the ends of the color range are not an absolute standard. Just before the place where the range is set, Perovich writes the following.

You want to choose values that are the bounds of what data you expect to see in a particular location so the color ranges well.

In short, if you choose the upper and lower limits of the values you expect to see in that location, the colors spread out well. The comment on conductivity also lists rough guides of about 700 µS/cm for fresh water, about 50,000 for brackish water, and about 53,000 for seawater. The configured 18,000–21,000 falls between the fresh-water figure of about 700 and the brackish figure of about 50,000.

The color can be read as meant for seeing “what is high or low within this river on this night.” Values outside the range take the color of the end, so the colors of boats set to different ranges cannot be compared on the same scale. Note also that which of red and green a low value becomes is reversed between the latest published code (LED_updated.ino) and the older version (archive/LED_functions.ino). It has not been confirmed which direction the 2015 photograph used. For that reason, this article does not state the direction.

Black hull, white window, numbers by radio

The parts list gives the use of the black spray paint as “to paint the boat so it blends into the background better at night.” That is, the boat is painted so that it melts into the background at night. The light comes out through “1/8” satin ice white acrylic strips," a window of milky acrylic. The list also says this replaced the earlier method of attaching the LED strip to the outside of the boat with one in which the light shows through holes cut in the hull.

The numbers are sent separately from the light. The README says “Numeric sensor data is also transmitted via radio, visible through an Android app.” The receiving-side Python script graphs the values and saves them to a CSV with timestamps.

The onboard GPS was, in the README’s words, “in practice this doesn’t seem to work (it never gets a fix).” As a workaround, the README describes a procedure in which a separate board records position to an SD card, which is later matched against the sensor CSV to make a map of the boat’s path and the data.

From this, SeeBoat can be read as handling data through two routes. The light on the water can be read as a design that, without carrying coordinates, shows people on the spot how the values change. The exact numbers are sent by radio and saved to CSV on the receiving side. The map is described as built by matching the SD-card position record against the sensor CSV.

In the abstract of her 2020 doctoral thesis, Perovich places SeeBoat among “citizen science water quality tools for data collection and in situ display.” In the book Making With Data (edited by Huron, Nagel, Oehlberg, and Willett; A K Peters/CRC Press), Part 2, “Participation,” has a chapter by Perovich, 2.2 “SeeBoat” (pp. 123–132, DOI 10.1201/9781003264903-7). I have not been able to check the text of the chapter for this article.

How third parties saw the work

I found no criticism or news coverage devoted to SeeBoat alone, within what I was able to trace this time. What follows is what third parties have written around the work.

The first is the 2019 Cambridge Arts exhibition “Untold Possibilities at the Last Minute: Artists Respond to Climate Change.” “ArtBoat and SeeBoat” was shown there, and the overview says the following.

SeeBoat invites us to experience our nearby rivers through the act of “drawing” with light attached to the boats.

In short, it lets us experience our nearby rivers through the act of “drawing” with the light attached to the boats. The overview also notes that water quality can change as water temperature rises, placing the work in the context of climate change.

A composite long-exposure photograph of a river at dusk, with lines of green, blue, and pink light drawn by many boats overlapping

Source: Laura Perovich (photo: Jorge Valdez) / untold possibilities at the last minute, “ArtBoat and SeeBoat”

The photograph above is of ArtBoat, the parent of SeeBoat. The exhibition site describes SeeBoat as ArtBoat combined with water-quality sensors. ArtBoat was an event at Magazine Beach in 2018 in which participants steered the boats’ colors with a color-mixing board and drew pictures of light on the river. Below the work’s page, an explanation of the city’s stormwater and combined sewer overflows (CSO) is attached. It can be read as showing that the work was exhibited alongside an explanation of municipal water-quality policy.

The second is the record of a night test at the Mystic Generator Station in September 2015, using the predecessor floats. It was written by Erik Hanley of Northeastern, and Perovich was among the participants.

It was difficult to see from the dock whether there was any color change from the floaties.

In short, it was hard to tell from the dock whether the floats’ color had changed. It can be read that this problem may be related to SeeBoat’s later painting of the hull black and lining LEDs up behind milky windows. I have not found any words from the author that show a causal link.

The third is Claudia Berger’s review of Making With Data (College & Research Libraries). The review criticizes the book as a whole like this.

These projects were designed by an overwhelmingly white audience of practitioners; only a few are led by practitioners who are Black, Indigenous, or people of color.

In short, most of the examples are by white practitioners, and only a few are led by Black, Indigenous, or people-of-color practitioners. The review also asks what about projects co-led by communities that tend to be shut out of data. A youth group from the environmental justice organization GreenRoots joined SeeBoat for testing, but the creator is a university researcher. This question can be read as an axis of tension for reading SeeBoat.

Takeaways

What can be taken away from SeeBoat can be read as follows.

Do not encode place. The place where the measurement was taken is the place of display, and this can be read as a design that shows people on the spot how the values change. The map is made separately by matching the position record against the sensor CSV.
Set the color range on site. If the two ends are fitted to the expected span, the colors spread widely. Colors from boats with different range settings cannot be used for comparison.
Continuous or threshold. Depending on what you want to find, you can choose between a hue scale and a binary display.
Separate light and numbers. On site, show only the color of one variable, and send the exact numbers by radio.
Erase the ground. Paint the hull black so that everything except the light you want to show blends into the background.

A long-exposure photograph combines these in a single image. The shape of the lines of light is where the boat went, and the color of the lines is the measured value.

References

Primary sources

Laura Perovich et al., SeeBoat (GitHub repository lperovich/seeboat. README, software/main/seeBoatFeather/seeBoatFeather.ino, LED_updated.ino, software/main/archive/LED_functions/LED_functions.ino, software/main/readMe.txt, hardware/SeeBoat_fullPartsList.xlsx, tutorials/, app/fontsColors.pdf). GitHub
Exhibition site work page and caption “ArtBoat and SeeBoat,” untold possibilities at the last minute (Cambridge Arts, 2019). untoldpossibilities.org
Laura J. Perovich, “From data physicalization to data experiences: combining art, science, technology, and community to move towards collective action on environmental challenges,” Ph.D. thesis, MIT Program in Media Arts and Sciences, 2020 (bibliographic record and abstract). DSpace@MIT
Laura Perovich, “SeeBoat,” in Huron, Nagel, Oehlberg, Willett (eds.), Making With Data: Physical Design and Craft in a Data-Driven World, A K Peters/CRC Press, pp. 123–132 (chapter text not checked). doi:10.1201/9781003264903-7 / Routledge / Crossref
Public Lab, “Thermal fishing bob” (wiki). publiclab.org
Laura Perovich, Public Lab research notes (2015): multiple inputs outputs / faster waterproofed temperature sensors / in search of floating spheres / threshold single thermal fishing bob (text checked in publiclab-archive)
Sara Wylie, “Successful Thermal Fishing Bob Maps,” Public Lab, 2014. publiclab.org
Sara Wylie et al., “Successful thermal fishing bob test: mapping the Kendall power plant thermal plume,” Public Lab, 2014. publiclab.org
MIT Media Lab, “SeeBoat (Thermal Fishing Bob): In-place environmental data visualization” (project page. Could not be reached, so its text has not been checked). media.mit.edu

Secondary sources

Cambridge Arts, “Untold Possibilities at the Last Minute: Artists Respond to Climate Change” (exhibition overview, 2019). untoldpossibilities.org
Erik Hanley et al., “Mystic Generator Station Thermal Fishing Bob Event with Mare Liberum,” Public Lab, 2015. publiclab.org
Claudia Berger, review of Making With Data, College & Research Libraries. PDF

Turning a Year of Chemotherapy Blood Pressure into a Bowl That Rings: Stephen Barrass's Chemo Singing Bowl

Thu, 05 Nov 2026 00:00:00 +0900

Stephen Barrass’s Chemo Singing Bowl maps a year of blood pressure readings from a person undergoing chemotherapy onto the shape of a bowl. The data is conveyed through the sound the bowl makes when it is struck. The bowl is 3D-printed in stainless steel, and you sound it by tapping it or rubbing its rim. The artist calls a sounding object whose form is determined by data Acoustic Sonification.

The work was published on the artist’s blog on January 18, 2016, and in the same year was shown in an exhibition at the Museum of Applied Arts and Sciences (MAAS) in Sydney. What follows draws on the artist’s blog, the installation paper at ICAD 2016, a 2014 paper about the predecessor, and a 2019 paper on which Barrass is a co-author. Where the sources differ in how they describe things, I show the difference as it is.

2013: Turning his own blood pressure into a “singing bowl”

Chemo Singing Bowl has a predecessor: the Hypertension Singing Bowl, whose form was determined by the artist’s own blood pressure. It was published on the blog in July 2013 and written up as a paper at the SoniHED conference in 2014.

It began with a diagnosis of high blood pressure. Barrass started measuring his systolic and diastolic pressure with a cuff and logging it in an app. In the paper, he writes this about the time-series graph the app drew:

The plots of my blood pressure are generic and could have been share prices or CO2 emissions. Surely there was a way to make this personal dataset more personal, and more engaging.

In short: the graph of his blood pressure is generic, and could just as well show share prices or CO2 emissions; could this personal data be made more personal and more engaging?

For the sounding object he chose the Tibetan singing bowl, a bowl that “sings” with a sustained tone when its rim is rubbed. Singing bowls are associated with meditation and relaxation therapy, and these can lower blood pressure, which was his reason.

The first prototype (v0.0) varied wall thickness with the data, up to 5 mm. When struck it rang at 3628 Hz for 3 seconds, but it did not sing when rubbed. In the next version (v1.0) he made the wall a uniform 2 mm and placed the time axis around the circumference of the bowl. Each pair of readings was assigned one “spoke” joining the rim to the base, with the radius of the upper half varying with the systolic value and the radius of the lower half with the diastolic value.

The predecessor, Hypertension Singing Bowl. A dark metallic bowl with thin vertical bands lined up at regular intervals on its lower half

Source: Stephen Barrass, “Research Through Design into Acoustic Sonification”, ICAD 2016 (CC BY-NC 4.0)

v1.0 is 100 mm in diameter and weighs 162 g. Struck, it rings at 609 Hz for 10 seconds, and rubbed on the rim it sang like a traditional bowl. It also produced a 2 Hz tremolo and a hiss, which the artist says “may be caused by the data spokes.” A 2014 video of the predecessor being played is on YouTube. The ICAD 2016 paper sums up the predecessor this way:

The bowl sings like a traditional singing bowl, but with a timbre that is changed by the dataset.

In short: the bowl sings like a traditional bowl, but its timbre is changed by the data.

A year of chemotherapy, written into curved tines

The data in Chemo Singing Bowl is not the artist’s own. It is the blood pressure that a person treated with chemotherapy for breast cancer recorded over the year of treatment. The artist’s blog (2016) introduces it like this:

This version of the acoustic sonification singing bowl introduces 3D splines as a way of mapping the data into the form. The irregularity of the dataset makes for inharmonic and dissonant acoustic effects.

In short: this version introduces 3D splines (smooth curves drawn through control points) as the way to map data into form, and the irregularity of the data produces inharmonic, dissonant sound. The ICAD 2016 paper states the intent behind the mapping:

This mapping aims to amplify the effect of the dataset on the acoustics of the bowl by mapping the diastolic/systolic pressure to control points for the curvature of 2D splines amped to the tines.

In short: diastolic and systolic pressure become the control points that set the curvature of splines, and those splines are mapped onto the tines (the bands), in order to amplify the data’s effect on the acoustics (“amped” appears to be a typo for “mapped”). So the predecessor mapped data to thickness, while Chemo maps it to how the tines curve.

A photo of a gold bowl seen obliquely from above. Dozens of thin tines curve outward in arcs between the rim and the base

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

In the photo, the tines are not uniform in length or in how they curve.

A photo of the bowl seen from directly above. Tines of uneven length and curvature stick out radially beyond the rim

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

The correspondences that can be confirmed from the sources are as follows.

Data	Property of the object	Source
Blood pressure recorded over a year of chemotherapy	The shape of the whole bowl (3D-printed stainless steel)	Artist’s blog (2016)
Diastolic and systolic pressure values	Control points that set the curvature of splines; the splines are mapped onto the tines	ICAD 2016
A year of blood pressure readings	Variation of the splines around the bowl	Wang et al. (2019)
Irregularity of the data	Inharmonic, dissonant sound	Artist’s blog (2016)
How it is read	Sounded by striking or rubbing; the sound follows from the form	Wang et al. (2019)

No source states explicitly that the time sequence was laid out around the circumference. Since the predecessor v1.0 did so, and since the source speaks of “variation of the splines around the bowl,” Chemo can be read as also arranged around the circumference. The number of readings, the number of tines, the dimensions, and the frequencies were not found in any source.

A photo of the bowl held in a hand, seen from the side. The tines project outward from the rim and wrap around toward the base

Source: Stephen Barrass / SONIFICATION.COM, “Chemo Singing Bowl”

Does the bowl sing? Where the sources disagree

The sources differ on two points: whether the bowl sings, and whether the splines are 2D or 3D.

Point	Source	What it says
Does it sing?	ICAD 2016	“This bowl does not sing very well”
Does it sing?	Wang et al. (2019)	“not sing harmonically”
Does it sing?	Artist’s blog (2018)	“3D printed in stainless steep [steel] so it sings”
Dimension	Artist’s blog (2016)	“3D splines”
Dimension	ICAD 2016	“2D splines”

The 2018 blog post is a list of demos the artist brought to a seminar at Schloss Dagstuhl in Germany, and it uses the same wording for the predecessor.

The explanations for why it does not sing also differ. ICAD 2016 says “possibly due to the curved shape of the tines.” The blog and the 2019 paper point to the irregularity of the data. ICAD 2016 goes on to say that the next version would explore improving the singing effect using acoustic theory, but I found no source describing that version.

I found no source that says whether the shape of the tines or the irregularity of the data is the main cause. Because the data is mapped to curvature, the more irregular the data, the more irregular the tine shapes become, so the effect of the form and the effect of the data come out together in a single sound; this can be read that way. Set side by side, the predecessor mapped data to thickness and sang, while Chemo mapped it to curvature, for amplification, and sings less well; this suggests that both the character of the data and the choice of mapping show up in the sound.

“How I felt at the time”: the words of the person who supplied the data

According to the 2019 paper, the bowl was made as a gift for the data provider’s mother, in thanks for caring for her during the chemotherapy. The data provider’s reaction on hearing the sound is worded differently in three sources.

Source	Reaction
ICAD 2016	“it sounds as sick as I felt at the time”
Wang et al. (2019)	“that’s exactly how I felt at the time”
Artist’s blog (2016)	“… quite accurate – the sound is dry and heavy, without much resonance, just like I felt!”

In short: it sounds as sick as I felt then; that’s exactly how I felt then. The blog version continues: quite accurate, the sound is dry and heavy, without much resonance, just like I felt.

The predecessor was a bowl in which the artist listened to his own data. The SoniHED 2014 abstract says it invites a meditative reflection on the data and reminds one of a healthy life, and calls it a “poetic alternative” to the generic graphs of self-tracking.

Chemo was made from someone else’s data, and the owner of the data listened to it. The predecessor’s question of making personal data more personal came back, with another person’s data, to that very person; this can be read that way. The words the owner used were “accurate” and “exactly.” They can be read as pointing not to the correctness of reading a graph but to a match with a remembered bodily feeling.

How did third parties see this work?

I found no independent critique or press coverage of Chemo Singing Bowl. It is treated in three ways.

The first is the 2019 paper in IEEE Computer Graphics and Applications, “An Emotional Response to the Value of Visualization.” Co-authored by Yun Wang and others, with Barrass among them, it grew out of a working group at the 2018 Dagstuhl seminar.

Informed by the role of emotion in psychology, art, design, marketing, and HCI, we argue for an expanded definition of value, applicable to all forms of data visualization.

In short: drawing on the role of emotion, the authors argue for broadening the definition of visualization’s value. The paper writes that existing frameworks that measure efficiency (such as ICE-T) overlook the value of data physicalization, and it cites the Chemo bowl as an example. Because Barrass is a co-author, this is not a fully independent assessment.

The second is dataphys.org, a catalog of data physicalization works. It has no entry for Chemo alone, but has an entry for the predecessor group, “2011 – Stephen Barrass’ Physical Data Sonifications,” added by Pierre Dragicevic and recommended by Yvonne Jansen.

Physical representations of data can target other senses than vision or touch.

In short: physical representations of data can address senses other than vision or touch. About the HRTF Bells (bells whose shapes are determined by the characteristics of the left and right ears), the entry says the differences in form are barely visible, but the pitch and timbre can be compared.

The third is selection by a museum. MAAS curator Matthew Connell chose this bowl for “Out of Hand: Materialising the Digital,” which ran from September 3, 2016 to June 25, 2017. According to the artist’s blog, it was an exhibition on the role of digital fabrication, and the bowl was shown alongside works by Zaha Hadid, Iris van Herpen, and others. It was also presented at the DESIGN Canberra Festival that year, and was placed as a hands-on demo you could pick up and sound at ICAD 2016 and at Dagstuhl in 2018.

People picking up and sounding bells and a medal on a table. A dark bowl is on the right, and part of a bowl with gold tines can be seen at front right

Source: Stephen Barrass / SONIFICATION.COM, “Data Physicalisation @ Schloss Dagstuhl”

In book form, there is a chapter by Barrass, “Chemo Singing Bowl,” in Making with Data: Physical Design and Craft in a Data-Driven World, edited by Samuel Huron, Till Nagel, Lora Oehlberg, and Wesley Willett (AK Peters / CRC Press, 2022). It is chapter 3.1 in Part 3, “Digital Production,” pp. 185–198, DOI 10.1201/9781003264903-11. I have not been able to read the text of the chapter.

Takeaways

What this bowl lets us take away is that changing the mapping changes how the object itself behaves. With a graph on a screen, switching from showing a value as thickness to showing it as curvature does not change how the screen behaves. With a sounding object, the mapping directly determines the sound.

In the paper on the predecessor, as a reflection on v0.0 not singing, Barrass writes that it would have been more efficient to print a baseline bowl with no data in it first and test that, before moving on to the mapping. For physicalization where the material is the medium, this procedure can be used as is.

Chemo also shows that a mapping which “amplifies” the effect can undermine the foundation of a metaphor. The metaphor of a singing bowl depends on the bowl singing. Even so, the owner of the data called the muddy sound “accurate.” The collapse of the metaphor itself can be read as having conveyed the data.

The other point is separating causes. When you change the mapping, also make a version with no data in the same form and compare by ear. Then you can say how much of the audible difference is due to the data. That is what I think.

References

Primary sources

Stephen Barrass, “Chemo Singing Bowl” (SONIFICATION.COM, 2016-01-18) https://sonification.com/2016/01/18/chemo-singing-bowl/
Stephen Barrass, “Research Through Design into Acoustic Sonification”, ICAD 2016 installation paper (CC BY-NC 4.0) https://www.icad.org/icad2016/proceedings2/installations/ICAD2016_paper_52.pdf
University of Canberra Research Portal, “Chemo Singing Bowl: Out of Hand - Materialising the Digital” https://researchprofiles.canberra.edu.au/en/publications/chemo-singing-bowl-out-of-hand-materialising-the-digital/
Stephen Barrass, “3D Printed Acoustics @ Out Of Hand, Sydney” (2016-09-03) https://sonification.com/2016/09/03/out-of-hand-materialising-the-digital/
Stephen Barrass, “Chemo Singing Bowl @ DESIGN Canberra” (2016-10-28) https://sonification.com/2016/10/28/design-canberra/
Stephen Barrass, “Data Physicalisation @ Schloss Dagstuhl” (2018-10-30) https://sonification.com/2018/10/30/data-physicalisation-schloss-dagstuhl/
Stephen Barrass, “Making With Data” (2022-12-13, reprint of the table of contents) https://sonification.com/2022/12/13/making-with-data/
Stephen Barrass, “Chemo Singing Bowl”, in S. Huron, T. Nagel, L. Oehlberg, W. Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, AK Peters / CRC Press, 2022, pp. 185–198 (text not read) https://doi.org/10.1201/9781003264903-11
Stephen Barrass, “Acoustic Sonification of Blood Pressure in the Form of a Singing Bowl”, SoniHED 2014, York https://www.york.ac.uk/media/c2d2/media/sonihedconference/Barrass_SoniHED_2014.pdf
University of Canberra Research Portal, “Acoustic Sonification of Blood Pressure in the Form of a Singing Bowl” (abstract) https://researchprofiles.canberra.edu.au/en/publications/acoustic-sonification-of-blood-pressure-in-the-form-of-a-singing-/
Stephen Barrass, “Hypertension Singing Bowl” (2013-07-10) https://sonification.com/2013/07/10/hypertension/
Stephen Barrass, demo video of the Hypertension Singing Bowl (YouTube, 2014) https://www.youtube.com/watch?v=DE3tolD5Krg
Stephen Barrass, “Sound in Data Physicalisation”, Dagstuhl Reports 8(10), 2019, §5.4 https://drops.dagstuhl.de/storage/04dagstuhl-reports/volume08/issue10/18441/DagRep.8.10.127/DagRep.8.10.127.pdf

Secondary sources

Yun Wang, Adrien Segal, Roberta Klatzky, Daniel F. Keefe, Petra Isenberg, Jörn Hurtienne, Eva Hornecker, Tim Dwyer, Stephen Barrass, Theresa-Marie Rhyne, “An Emotional Response to the Value of Visualization”, IEEE Computer Graphics and Applications 39(5), 2019 (Barrass is a co-author) https://petra.isenberg.cc/publications/papers/Wang_2019_AER.pdf / doi:10.1109/MCG.2019.2923483
dataphys.org, “2011 – Stephen Barrass’ Physical Data Sonifications” (added by Pierre Dragicevic, recommended by Yvonne Jansen) https://dataphys.org/list/barrass-physical-data-sonifications/
Jason Alexander, Petra Isenberg, Yvonne Jansen, Bernice E. Rogowitz, Andrew Vande Moere (eds.), Dagstuhl Seminar 18441 “Data Physicalization” report, Dagstuhl Reports 8(10), 2019 https://drops.dagstuhl.de/storage/04dagstuhl-reports/volume08/issue10/18441/DagRep.8.10.127/DagRep.8.10.127.pdf
Out of Hand: Materialising the Digital (MAAS, curator Matthew Connell, 2016-09-03 to 2017-06-25). The selection is confirmed on the artist’s blog https://sonification.com/2016/09/03/out-of-hand-materialising-the-digital/
Making with Data book site https://makingwithdata.org/

One Data Point, One Robot: Zooids

Thu, 05 Nov 2026 00:00:00 +0900

Zooids assigns each data point to one tiny robot and builds scatterplots and maps from the robots’ positions. The axes are made of robots too, and you can pick up and move any point by hand. Each robot is 26 mm in diameter, 21 mm tall and weighs about 12 g, and runs on two wheels across a flat table.

It was developed by the Shape Lab at Stanford University (Sean Follmer) and the Aviz team at Inria (Jean-Daniel Fekete, Pierre Dragicevic). The foundational paper was presented at UIST in 2016, and a second paper, presented at IEEE VIS in 2018 and published in the January 2019 issue of IEEE TVCG, summarizes the application to data visualization. The author lists of the two papers differ slightly. What follows draws on the full text of the UIST 2016 paper, the abstract of the TVCG paper and the Shape Lab page, and the source code the authors published on GitHub. I have not been able to read the full text of the TVCG paper, so details of the 2018 encodings are written as things read off the source code.

2016: A swarm of robots that fit in the palm of your hand

The first paper is “Zooids: Building Blocks for Swarm User Interfaces.” It was presented at UIST ‘16 in Tokyo in October 2016 (pp. 97–109). The name comes from the biological term zooid: each of the individual animals that make up a colony.

As starting points, the authors cite Ivan Sutherland’s “Ultimate Display” and Hiroshi Ishii’s “Radical Atoms.” The former is the idea of a room in which a computer controls the existence of matter.

Swarm user interfaces can be seen as a coarse-grained version of Sutherland’s and Ishii’s futuristic visions of user interfaces based on programmable matter.

That is, interfaces built from a swarm of robots (swarm UIs) are positioned as a coarse-grained version of the vision of programmable matter. The authors also write about how this differs from earlier tangible UIs.

We are interested in user interfaces where the tangibles are used not only as controllers, but also as representations of digital content.

The point is that graspable objects should serve not only as tools for input but also as representations of digital content.

Five photos: a robot resting on a palm, a clump of robots pushed together by hand, and robots arranged in a wavy line and in a rectangle

From Mathieu Le Goc et al. (Shape Lab, Stanford / Inria Aviz), GitHub “ShapeLab/SwarmUI” (CC BY-SA 4.0)

Each robot finds its position by reading a Gray-code pattern projected from above with photodiodes on its upper side. The resolution is about 1.1 mm. Electrodes wrapped around the body sense touch, and a color LED sits on top. Parts and assembly cost about $50 per robot, and the paper says that a swarm beyond 30 to 40 robots would be too expensive outside of research.

One robot per point, in line charts and scatterplots alike

The UIST 2016 paper presents several example applications. Two of them are visualizations.

One is a line chart of time-series data. It shows a computer’s CPU usage as the heights of a row of robots along the top. Each time new data arrives, the line appears to flow to the left. The axes and labels do not move, so they are printed on paper. Two robots at the bottom are the two handles of a range slider and change the time range displayed. One more robot switches between CPU and memory.

The other is an example with several scatterplots side by side. Each scatterplot is assigned a different pair of variables. If you pick up one robot by hand and place it in another scatterplot, you can see that point along different dimensions.

As each zooid embodies a specific data point, moving it into a different chart allows users to probe different dimensions.

Because each robot embodies one data point, moving it to another chart lets you examine a different dimension. Placing a robot on a tablet or similar device brings up detailed information about that point. Robots for display and robots for control are told apart by the color of their LEDs. The authors themselves write that this color assignment is arbitrary and presumes a user who is used to the system.

The paper’s “design principles” section starts from the difference from screen pixels. Pixels are fixed to a grid, and only their color can change. Zooids’ elements can be placed anywhere, but there are few of them. The authors write that, just as designers in the 1980s thought about how to use each pixel, we should think about how to use each robot.

in systems where each zooid has a stable meaning (e.g., a visualization system where a zooid represent a data point), swapping zooids can produce confusing or misleading animations.

In a system where one robot stands for a data point, swapping robots can produce confusing or misleading movement. From this the authors conclude that the designer should think about which elements can be made interchangeable and which should be given a fixed identity. This can be read as saying that the promise of one robot per point also determines how the movement is read.

The authors also list limitations: the robots can move only on a flat surface, they lose their position when a hand blocks the light, and charging has to be done by hand, one robot at a time. The section on future work contains a sentence calling information visualization “an exciting area.”

2018: The axes are made of robots too

The paper two years later is “Dynamic Composite Data Physicalization Using Wheeled Micro-Robots.” It was presented at IEEE VIS 2018 and published in IEEE TVCG vol. 25, no. 1 (January 2019, pp. 737–747). The authors changed as follows.

Paper	Authors
UIST 2016 “Zooids: Building Blocks for Swarm User Interfaces”	Mathieu Le Goc, Lawrence H. Kim, Ali Parsaei, Jean-Daniel Fekete, Pierre Dragicevic, Sean Follmer
IEEE TVCG 2019 “Dynamic Composite Data Physicalization Using Wheeled Micro-Robots”	Mathieu Le Goc, Charles Perin, Sean Follmer, Jean-Daniel Fekete, Pierre Dragicevic

Charles Perin appears only on the TVCG paper. His affiliations are the University of Victoria and City, University of London. The abstract reads as follows.

This paper introduces dynamic composite physicalizations, a new class of physical visualizations that use collections of self-propelled objects to represent data.

It proposes a new class of physical visualization that represents data with a collection of objects that move on their own. The Shape Lab page presents two scenarios. In one, four faculty members choose participants for a summer school. “Each Zooid embodies an applicant.” One robot is one applicant.

Four people around a table, moving scattered robots by hand and looking at a tablet

From Le Goc, Perin, Follmer, Fekete, Dragicevic / Shape Lab, “Dynamic composite data physicalization using wheeled micro-robots”

The description of the other scenario is about a scatterplot turning into a map.

By placing Lattitude and Longitude magnets in the sandbox, the scatterplot become a map, showing the city of origin of applicants.

Placing latitude and longitude “magnets” in the sandbox (the frame of the scatterplot) turns the scatterplot into a map that shows the applicants’ cities of origin.

A world map laid under a transparent panel, with robots scattered over it and more robots placed at the panel’s four corners

From Le Goc, Perin, Follmer, Fekete, Dragicevic / Shape Lab, “Dynamic composite data physicalization using wheeled micro-robots”

From here on, what I describe is read not from the paper’s text but from the source code of ShapeLab/ZooidsCompositePhysicalizations on GitHub (the files under Zooid_Vis/src). The dataset is a fictional “Genius University dataset” of 306 rows, and it displays up to 24 applicants who meet the grade conditions. In the code, applicant robots are called Dust and robots that represent variables are called Magnet.

Data	Physical property	Where I checked
One applicant	One robot	Shape Lab page, code
A variable (grades in English, science and math, recommendation rating, age, etc.)	Another robot serves as a magnet. The magnet pulls applicant robots toward it according to their values	Code
Strength of the magnet	The angle to which the magnet robot is turned by hand (180° maps to 0–20)	Code
Strength of the magnet (visual)	Brightness of the magnet’s LED	Code
Type of variable	Hue of the LED (English = red, science = blue, math = green, etc.)	Code
Value of that variable when touching the magnet	Intensity of the applicant robot’s LED	Code
Axes of the scatterplot	Three robots are placed at a right angle. If the two end robots are magnets, their variables become the x and y axes	Code
Latitude and longitude of the city of origin	The x and y positions in the scatterplot (fixed ranges: longitude −180 to 180, latitude −90 to 90)	Shape Lab page, code
Details of an applicant	Shown on the tablet screen when a robot is touched	Code
Removing / excluding	Placed on the strip at the edge of the table, a magnet disappears and applicants are no longer moved	Code

The terms and numbers used in the paper’s text may differ from those in the code. The other scenario, a family looking for a holiday destination, has only empty dummy data in the public code, so I could not check its contents.

In its related-work section, the UIST 2016 paper cited Dust & Magnet (Yi et al., 2005) and Microsoft SandDance as visualizations that swarm robots could physicalize. The 2018 code is built around the names Dust and Magnet. This can be read as the TVCG paper actually physicalizing a technique that the UIST paper cited. However, I have not confirmed whether the TVCG text explains it that way.

The work appears in the book Making with Data: Physical Design and Craft in a Data-Driven World (edited by Samuel Huron, Till Nagel, Lora Oehlberg and Wesley Willett, A K Peters/CRC Press, 2022), in Part 4, “Actuation,” as chapter 4.5, “Zooids” (pp. 299–309, doi:10.1201/9781003264903-20). The authors are the same five as on the TVCG paper. I have not been able to read the text of the chapter, so I do not cover its contents here.

How outsiders saw it

From the visualization research side, Djavaherpour et al. discuss it in their survey paper at EuroVis 2021, “Data to Physicalization.” Comparing it with devices that build static physicalizations with robot arms, they write about Zooids as follows.

zooids is a truly active physicalization system capable of showing many different data sets and types

This is an assessment that it is a truly active physicalization that can show many kinds of data and that people can pick up and move. The same paper also names a problem of scale as a weakness of active physicalizations: they usually need one actuator per data point.

From the HCI side, ShapeBots by Suzuki et al. (UIST 2019) is a follow-up. It points out that a swarm UI can change shape only by rearranging units of roughly the same form. It then argues that if each unit could change its own shape, it could do the following.

support representations that are not limited to moving points, but also lines, and other shapes on a 2D surface

The claim is that it can represent not only moving points but also lines and other shapes. This can be read as an assessment that, from the other direction, points out a limit of Zooids: its representation is confined to the positions of points and the colors of LEDs.

General-audience coverage looked elsewhere. Silicon Republic (November 2016) introduced it as “dozens of miniature, robotic helpers,” and Hackaday (February 2017) picked up the scene in which the swarm brings you a smartphone. The data visualization side, it seems, barely came across in the coverage.

How to use this

The first is that if you assign one object to each point, you should keep its identity. As the UIST paper writes, if the robot representing a data point gets swapped midway, the movement misleads. For animations on screen too, it turns out that whether to keep track of which point is the same one is a design decision.

The second is how to narrow down when the number of elements is limited. The paper gives 30 to 40 robots as a rough upper limit, and the 2018 code narrows applicants by conditions to at most 24. With physicalization, you need to decide how many items to show before you display anything.

The third is to make axes and legends objects that can be manipulated, too. In UIST 2016 the axes were printed on paper, but in the 2018 code the axes are robots as well, and how they are placed determines the variables. Switching variables becomes a hand motion of placing, turning and removing robots.

The fourth is to know the range of what can be expressed. The expressions in Zooids can be read as centered on the position and color of points. ShapeBots argues that to express lines and shapes, the elements themselves have to change. Before choosing a chart, you can check what the material is able to express.

References

Primary sources

Mathieu Le Goc, Lawrence H. Kim, Ali Parsaei, Jean-Daniel Fekete, Pierre Dragicevic, Sean Follmer, “Zooids: Building Blocks for Swarm User Interfaces,” UIST ‘16, pp. 97–109, 2016. https://doi.org/10.1145/2984511.2984547 / PDF: https://shape.stanford.edu/research/swarm/SwarmUIs_no_copyright.pdf / HAL: https://inria.hal.science/hal-01391281
Shape Lab, “Zooids: Building Block for Swarm User Interface”: https://shape.stanford.edu/research/swarm/
Mathieu Le Goc, Charles Perin, Sean Follmer, Jean-Daniel Fekete, Pierre Dragicevic, “Dynamic Composite Data Physicalization Using Wheeled Micro-Robots,” IEEE TVCG 25(1), pp. 737–747, 2019. https://doi.org/10.1109/TVCG.2018.2865159 / Abstract (PubMed): https://pubmed.ncbi.nlm.nih.gov/30136993 / HAL (full text not checked): https://inria.hal.science/hal-01848436
Shape Lab, “Dynamic composite data physicalization using wheeled micro-robots”: https://shape.stanford.edu/research/dynamicCompositePhysicalizations/
GitHub, ShapeLab/SwarmUI (CC BY-SA 4.0): https://github.com/ShapeLab/SwarmUI
GitHub, ShapeLab/ZooidsCompositePhysicalizations (MIT License. Files consulted: Zooid_Vis/src/ofApp.cpp, Sandbox.cpp, Magnet.cpp, parameters.h, Zooid_Vis/bin/data/student-dataset.csv, zooids-students-app/js/constants.js): https://github.com/ShapeLab/ZooidsCompositePhysicalizations
Le Goc, Perin, Follmer, Fekete, Dragicevic, “Zooids,” in Huron, Nagel, Oehlberg, Willett (eds.), Making with Data, A K Peters/CRC Press, 2022, ch. 4.5, pp. 299–309. https://doi.org/10.1201/9781003264903-20 / HAL (abstract only checked): https://api.archives-ouvertes.fr/search/?q=halId_s:hal-03978632&fl=* / Book page: https://www.routledge.com/Making-with-Data-Physical-Design-and-Craft-in-a-Data-Driven-World/Huron-Nagel-Oehlberg-Willett/p/book/9781032182223
Mathieu Le Goc, “Supporting Versatility in Tangible User Interfaces Using Collections of Small Actuated Objects,” PhD thesis, Université Paris-Saclay, 2016 (bibliographic record only): https://theses.fr/2016SACLS563
Videos (YouTube, posted by Mathieu Le Goc): UIST version https://www.youtube.com/watch?v=ZVdAfDMP3m0 / TVCG version https://www.youtube.com/watch?v=2GmWR617YeI

Secondary sources

H. Djavaherpour, F. Samavati, A. Mahdavi-Amiri, F. Yazdanbakhsh, S. Huron, R. Levy, Y. Jansen, L. Oehlberg, “Data to Physicalization: A Survey of the Physical Rendering Process,” Computer Graphics Forum 40(3) (EuroVis 2021 STAR): https://arxiv.org/pdf/2102.11175
Ryo Suzuki et al., “ShapeBots: Shape-changing Swarm Robots,” UIST ‘19: https://arxiv.org/pdf/1909.03372
Gordon Hunt, “Swarms of tiny robots could soon be at our beck and call,” Silicon Republic, 2016-11-29: https://www.siliconrepublic.com/machines/swarm-robots-zooids-robotics
Pedro Umbelino, “Zooids — Swarm User Interface,” Hackaday, 2017-02-17: https://hackaday.com/2017/02/17/zooids-swarm-user-interface/

A 32-Color Legend Shared by Everyone, One Strip per Year Hung Side by Side: The Tempestry Project

Thu, 05 Nov 2026 00:00:00 +0900

The Tempestry Project fixed everything from −30°F and below to 121°F and above to 32 colors in 5°F steps, and had everyone use the same yarn. One tapestry represents the daily high temperatures of one place for one year. One day is one row: January 1 is at the bottom and December 31 at the top. At exhibitions, tapestries of the same place in different years, or of different places, were hung side by side so that they could be compared.

The project began in Anacortes, Washington, in 2017. What follows draws on the knitting instructions and color cards (PDF) the project distributed, an interview with a co-founder, and records of exhibitions. As a sequel to temperature scarves and blankets, where each maker sets the color steps to suit their own climate, this article reads a case in which everyone shares the same legend.

2017: Knitting before the data disappears

The name is a coinage from temperature and tapestry. The trigger was the 2016 US presidential election: climate data might be removed from government sites such as NOAA (the US National Oceanic and Atmospheric Administration). Reading the situation that way is how it began, according to an article from Barnard College. Co-founder Emily McNeil says:

“We were joking about using knitting to create wall hangings [that would record] this data for posterity … Because unlike online data storage, tapestries can last for centuries or even millennia.”

In other words: they joked about knitting wall hangings to preserve the data, because tapestries last for hundreds of years.

The founders’ names differ depending on the source.

Source	Founders listed
Barnard College (2020)	Two: McNeil and her partner Justin Connelly
English Wikipedia, Hindustan Times (2019), Irish Examiner (2025)	Three: Emily McNeil, Marissa Connelly, and Justin Connelly
Exhibition text, Berman Museum (Ursinus College, 2021)	Three: Asy Connelly, Marissa Connelly, and Emily McNeil

I could not confirm whether Justin Connelly and Asy Connelly are the same person.

The project sold kits containing yarn, instructions, and a table of daily temperatures for the buyer’s location. As of April 2020, Barnard reports that participation had spread to 49 states and more than 30 countries.

One 32-color legend, shared by everyone

The instructions explain why they assign yarn colors to temperature ranges:

“The idea is to have everyone use the same colors in order to facilitate comparisons from year to year and place to place.”

That is, if everyone uses the same colors, it becomes easier to compare across years and across places. The knitting and crochet versions of the instructions contain the same sentence.

The instructions map data to the fabric as follows.

Data	How it appears in the fabric	Specific values
Date (one day)	Position along the length	One day = one row. Cast on with the color of January 1 and bind off with the color of December 31
One year	One piece	“January at the bottom and December at the top”
That day’s high temperature	Color of the row	32 colors in 5°F steps
Place	One weather station per piece	The station’s name is written on a card and enclosed
Precipitation (optional)	Beads on the right edge	Clear = 0.1 inch, light blue = 1 inch, sapphire = 5 inches
Anniversaries (optional)	Beads across the full width of a row	―
―	Width and length (no data)	About 8–12 inches wide, about 4–5 feet long. “gauge is not important”

The temperatures come from NOAA data, and gaps are filled from sources such as Weather Underground.

The color card lists all 32 classes. The color names are those of KnitPicks’ Wool of the Andes yarn, which the project used at first.

Two color cards, one in °F and one in °C. They list the names and temperature ranges of 32 colors, from Currant ≥121° to Coal ≤ −30°, with a yarn sample tied through a hole for each

Source: Justin Connelly / Wikimedia Commons (CC BY-SA 4.0)

The 30 classes other than the two end classes are all 5°F wide. The boundaries are whole numbers in Fahrenheit, and the Celsius values are converted, fractions included. This suggests Fahrenheit is the primary unit. 0°C falls inside Tranquil (31–35°F), so the freezing point is not a color boundary. The 2022 color chart changed the yarn and the color names, but the temperature breaks are the same.

How the ends of each range are written differs between sources. The instructions say “Less than −30°F,” which leaves the treatment of exactly −30°F unstated. The color card says “≤ −30°.” Barnard writes “less than −30°F to more than 121°F,” and the Berman Museum writes “less than −31°F.”

Kristen Cooper’s card in the temperature scarf article had 14 classes in 5°C steps, adapted to her own climate. Tempestry has 32 classes in 5°F steps, from −30°F and below to 121°F and above. It can be read as taking a range wide enough to hold the daily highs of any place. It can be read as a legend that decided on a common axis for comparison before it decided on legibility for each locale.

Not one strip, but side by side

The instructions say:

“Multiple Tempestries for one location create a striking historical display of shifting temperatures over time.”

That is, several strips of the same place make a historical display of changing temperatures. A 2018 Yale Climate Connections article states in its own narration that “the goal of the project is not to look at each piece individually,” and says the point is to view different years of the same place side by side. McNeil’s own words:

“If you see a lot of them together, it’s a very visually striking representation of changing temperature over time.”

In other words, when many are shown together, the change in temperature becomes clearly visible.

The exhibition records show two ways of arranging them.

Year	Venue	What was hung	Source
2018	FutureFest (Anacortes)	Utqiagvik, Alaska: 1925, 2010, 2016; Death Valley, California: 1950, 2016	Photo caption on Wikimedia Commons
Fall 2018	Museum of Northwest Art, “Surge” exhibition	22 strips for Deception Pass, Washington; older years at the left	Photo caption on Wikimedia Commons
2019	Pacific Northwest Quilt & Fiber Arts Museum	One Orcas Island station, the past 100 years at 4-year intervals; the count disagrees: 29 (Islands’ Sounder) vs. 27 (Wikipedia)	Islands’ Sounder, Wikipedia
2019	National Parks project	Two strips per park; mostly 2016, the National Park Service centennial, plus an earlier year	Adobe Express project page

On a white wall, 22 tapestries with tags from 1950 to 2016 hang in a single row, with color-reference yarn hanging at both sides

Source: Justin Connelly / Wikimedia Commons (CC BY-SA 4.0)

One way is to hang the years of a single place in a row. The Commons caption gives the Deception Pass range as 1950–2017, but the tags in the photo run from 1950 to 2016. Color-reference yarn hangs at both sides.

Three tapestries with blue, purple, and green stripes on the left, and two with red and yellow stripes on the right

Source: Justin Connelly / Wikimedia Commons (CC BY-SA 4.0)

The other way is to hang different places, each in chronological order. According to the Commons caption, the three on the left are Utqiagvik and the two on the right are Death Valley. In the photo, the two use different parts of the 32-color range. Both ways of hanging can be read as working only because everyone uses the same 32 colors.

In the National Parks project, many examples put the older year on the left, but Glacier Bay and Death Valley have 2016 on the left, and Grand Canyon is arranged top and bottom. This suggests that while the legend was shared, the direction of arrangement was not fixed.

Hung alongside the Knitting Map

In 2021, the Berman Museum of Art at Ursinus College opened “Mapping Climate Change: The Knitting Map and The Tempestry Project.” It was the first exhibition to show Jools Gilson and Richard Povall’s Knitting Map next to Tempestry. It later toured, reaching Cork in 2025.

The Knitting Map was a project of Cork 2005, European Capital of Culture. Over the course of 2005, more than 2,500 people knitted a single cloth. According to the Berman Museum’s exhibition text and the Irish Examiner (2021), the stitches encode how crowded the city was, measured by surveillance cameras, and color encodes temperature, precipitation, and wind speed. Early reports (Irish Times, 2004; RTÉ) described the source of the stitches as “satellite images.”

The Berman Museum’s exhibition text called the Knitting Map “a conceptual and artistic progenitor, of sorts” of Tempestry. In 2025, Gilson also described Tempestry as “this later project it helped inspire.” However, the claim that the Knitting Map was a source of inspiration comes only from the people who organized the exhibition and from the Knitting Map’s creators. English Wikipedia gives temperature blankets as the source of inspiration. The materials from the Tempestry side that I was able to read (the instructions and the Barnard article) do not mention the Knitting Map. I have not been able to read the About page of the official site.

The Knitting Map is one cloth for one city and one year; it has two data streams, and its colors are, in Gilson’s word, “muted.” Tempestry gathers many strips, one per place and year; it has one data stream, and its legend is a public 32 colors. The difference can be read as that between a work read within a single cloth and a work compared across many strips.

How outsiders saw the project

The assessments each address a different point. The first is skepticism about the project’s role in preserving data. The Hindustan Times (2019) quotes Sagnik Dey, an associate professor at IIT Delhi:

“… Individual efforts are okay, but in climate science everything should be in network mode and government funded.”

His opinion is that individual efforts are fine, but climate science should be done in network mode with government funding. It can be read as a rebuttal to McNeil’s motivation that tapestries last for centuries.

The second is museum curation. The Berman Museum’s exhibition text says that the data is interpreted through the knitter’s “personal calligraphy.”

“The process of handknitting and the resultant textiles thus map the motions of makers’ hands, as much as they transmit data relating to climate and locale.”

The view is that hand-knitted cloth records the movements of the knitter’s hands as much as it conveys data.

The third is the researcher’s positioning. In a 2020 essay in electronic book review, Anastasia Salter cited a Tempestry scarf as an example of the fusion of craft and computation. She places it within “data-driven craftivism” spreading on social media.

Takeaways

There are five judgments to carry back into visualization work.

First, if you want people to compare, fix the legend outside the makers’ hands, in advance. Tempestry decided the range, the steps, the colors, and even the yarn brand, and distributed them to everyone. It can be read that not letting each maker decide the legend was the precondition for side-by-side comparison.

Second, a shared range has a cost. Covering −30°F to 121°F means that one place in one year uses only part of the 32 colors. This can be read as a choice that put differences between places and between years ahead of the seasons within a place.

Third, separate what is compared from what is not. The instructions say “gauge is not important,” and width and length vary by knitter. Only color is compared, not size; that can be read as the design.

Fourth, decide the arrangement too. Within one strip, “January at the bottom” is fixed, but the left-to-right order of strips was not consistent. If you line up small figures, it is better to decide the direction of arrangement as well, just as you decide the legend. That is how I read it.

Fifth, attach the legend and the source to the work. The kit includes a card with the station’s name, and the color correspondence is published on the color card. At Barnard, McNeil says that 30 strips for one place would become a record of climate change. What makes those 30 strips readable is the 32-color legend that everyone shared.

References

Primary sources

Tempestry Project™ Instructions (PDF, uploaded September 2018). PDF
Tempestry Crochet Instructions (PDF, uploaded April 2019). PDF
The Tempestry Project Color Card Template v1 (PDF). PDF
Tempestry Yarn Color Temperature Beads DMC Thread Chart (PDF, March 2022). PDF
Veronica Suchodolski ’19, “Way Back Wednesday | Environmental Activist Emily McNeil ’98,” Barnard College, 2020-04-22. Article
Ariel Hansen, “Knitters turn climate data into tapestries,” Yale Climate Connections, 2018-11-09. Article
National Parks Tempestry Project (edited by Erika Zambello), Adobe Express. Project page
Wikimedia Commons (photos and captions by Justin Connelly, CC BY-SA 4.0). Color card / 22 strips for Deception Pass / Utqiagvik and Death Valley
“Island artists’ textile work on display in La Conner,” Islands’ Sounder, 2019-05-15. Article
Berman Museum of Art, Ursinus College, “Mapping Climate Change: The Knitting Map and The Tempestry Project,” 2021-06-11 to 11-30. Exhibition page
Lawrence University, Wriston Art Galleries, Past Exhibitions (Mapping Climate Change, Fall 2023). Exhibition page
Marc O’Sullivan Vallig, “Cork In 50 Artworks, No 33: The Knitting Map for European Capital of Culture, 2005,” Irish Examiner, 2021-12-13. Article
Marc O’Sullivan Vallig, “‘I do feel really humbled’: Cork’s Knitting Map gets tangled up in debate on American values,” Irish Examiner, 2024-11-04. Article
Marc O’Sullivan Vallig, “Jools Gilson of UCC: Mapping Climate Change at Cork’s newest art venue,” Irish Examiner, 2025-11-10. Article
“Knitting together a new stitch pattern of Cork,” Irish Times, 2004-10-30. Article
“Cork knitters make giant textile record,” RTÉ Archives (broadcast 2005-08-24). Archive
English Wikipedia, “The Tempestry Project” (version of 2026-04-16 consulted). Wikipedia

Secondary sources

“A knitting movement of weather data to document climate change,” Hindustan Times, 2019-01-06. Article
Berman Museum of Art exhibition text. Exhibition page
Anastasia Salter, “Re:traced Threads: Generating Feminist Textile Art with Tracery,” electronic book review, 2020-05-03. doi:10.7273/dped-gd56

Counting Animal Tool Use in Layers of Colored Paper: National Geographic's "Animal Hacks"

Sun, 01 Nov 2026 00:00:00 +0900

“Animal Hacks” is a graphic that shows how animals use tools, built as a three-dimensional stack of colored paper. It maps 22 modes of tool use to layers of colored paper, and whether each animal group has a recorded instance to whether a layer’s leg is visible or hidden. The magazine’s credits list a photographer for the “photo of paper construction,” and the author writes that he made the graphic with tools and real materials. It can be read that the photograph, overlaid with labels and text, became the printed graphic as it was. The paper model can be read as the raw material of the finished graphic, not a draft for it.

It ran in the March 2017 issue of National Geographic Magazine. The graphic was made by Alberto Lucas López and others. What follows draws on the magazine-spread image and description in the author’s portfolio, the GIF of the making process, and the credit lines on the page. dataphys.org, which collects works of data physicalization, lists it under the name “Hand-Crafted Magazine Infographics.”

A spread in the EXPLORE section, March 2017

The graphic appeared in the EXPLORE section of the March 2017 issue. The header band reads “TOOLS AND TECHNOLOGY” and the article is titled “ANIMAL HACKS.” The text was written by Rachel Hartigan Shea. The graphic’s title is “Differences in animal tool use.”

A spread on a black background: in the center, layers of colored paper nest in the shape of a gate, with animal-group icons and names lined up on the left and right

The text begins with the point that humans are not the only ones who use tools.

Humans aren’t the only creatures to use tools. Archerfish shoot water droplets from their mouths to fell insects. Octopuses carry coconut shells to serve as shelter. Orangutans borrow canoes to forage for aquatic plants.

These are the examples of archerfish, octopuses, and orangutans. The text then quotes the biologist Robert W. Shumaker, “Tool use is widespread and diverse,” and goes on to say that tool use is not necessarily a sign of intelligence.

Beneath the graphic there are two credit lines.

ALBERTO LUCAS LÓPEZ, NGM STAFF; KELSEY NOWAKOWSKI / PHOTO OF PAPER CONSTRUCTION: REBECCA HALE, NGM STAFF

SOURCE: ROBERT W. SHUMAKER, INDIANAPOLIS ZOO

Two names are listed as the graphic’s authors, and Rebecca Hale’s name appears as the photographer of the “photo of paper construction.” This line shows that the paper model made it into the magazine. The sources do not say how the two authors divided their roles. López is a Spanish designer who, after serving as graphics director at the South China Morning Post, moved to National Geographic in 2016 as Senior Graphics Editor.

22 layers and 22 animal groups

The data is whether each animal group has a recorded mode of tool use in the scientific literature. There are 22 modes of use and 22 animal groups. The description on the page reads as follows.

Each layer represents a mode of tool use. Colored bars next to the animals’ names indicate which behaviors have been recorded for that group in the scientific literature. Animal groupings can have one species (bonobo), a few (Old World monkeys), or hundreds (insects).

Each layer stands for one mode of use, and the colored bars beside the animal names show the behaviors recorded for that group. A group may contain a single species or several hundred.

As the source of the data, the magazine names only Shumaker. The number 22 matches the “22 modes of tool use” classified in Shumaker, Walkup & Beck’s Animal Tool Behavior (2011). The magazine does not cite the book, so it cannot be confirmed that the book’s classification was used as is. Because the name and the number match, it can be read as likely based on this book.

The correspondence between the data and the appearance is as follows.

Data	Appearance in paper and photo	How to check
Mode of tool use (22)	One layer of colored paper, cut out in a gate shape and nested. Color distinguishes the modes	Description on the page, magazine-spread image
Order of modes	From the outside of the layers to the inside. The outermost is Affix (attach to the body, wear), the innermost is Cut	Label positions in the magazine-spread image
Animal group (22)	A horizontal band (row). 11 groups each on the left and right. Icon and name, with the number of recorded modes in parentheses	Magazine-spread image
The mode is recorded	In that row, the leg (vertical color band) of that layer is visible	Description on the page
Not recorded	In that row, the leg of that layer is hidden in black	Magazine-spread image
Number of recorded modes	A numeric note and the number of color bands visible in the row	Magazine-spread image
Stacking of layers	White board-like spacers are placed between the sheets of colored paper to create steps and shadows	GIF of the making process

The left column runs from New World monkeys (22) to Amphibians (1), and the right column from Chimpanzees, orangutans (22) to Echinoids (1), both in descending order of the number. The lower the row, the more black there is, and the dark area in the center appears to widen in steps.

The left leg of the gate is used as the column of the 11 groups on the left, and the right leg as the column of the 11 groups on the right. A single layer can be read as doubling as two columns, left and right. The order of the layers also shows a tendency. The outer layers appear to be modes recorded in many groups, and the inner layers appear to be modes found only in great apes and New World monkeys. In the Amphibians (1) row, only the outermost layer, Affix, reaches the bottom. The page gives no basis for this ordering. On the other hand, the steps and shadows of the layers do not represent quantity, and can be read as there to tell the layers apart.

From paper on a desk to a magazine spread

The author’s portfolio includes a 13-frame GIF of the making process. On March 24, 2017, the author tweeted the same GIF with the words “Handmade #dataviz.”

The first frame shows the structure on a white desk, shot from the side. Colored paper and white spacers are stacked, with scissors, a pencil, a ruler, and a glue stick in front. A National Geographic mug is also in the frame.

A structure of stacked colored paper and white spacers on a white desk, with scissors, a pencil, a ruler, and glue in front

Source: Alberto Lucas López, Kelsey Nowakowski (graphic), Rebecca Hale (photo of paper construction) © 2017 National Geographic Partners / Alberto Lucas López, from the making-process GIF of “Animal tools”

From there the camera moves to directly overhead. The gate-shaped layers nest and begin to look like a passage leading to the back. In the top-down frame, the tips of the legs have already been cut to uneven lengths. Roughly, the outer layers have longer legs and the inner layers shorter ones. On both sides the leg tips line up in steps.

Colored paper layers seen from directly above. The gate-shaped layers are nested, and the leg tips are cut in steps, longer on the outside and shorter on the inside

Each leg can be read as cut so that it reaches the lowest row among the groups in which that mode is recorded. In the next frame, a large black section is added beneath the legs. The sources do not confirm whether this black part is paper or was painted on the image.

In the frame after that, the layout carries labels and text, and the last frame shows the printed magazine spread. The paper construction was photographed and became the graphic itself. The order of the GIF and the credit lines suggest this reading.

The author himself writes this in his portfolio.

To support the information shown, the infographic was made with tools and real materials, looking for a parallelism between the visual side and the exposed theme. This was the first papercraft infographic published in National Geographic Magazine.

In other words, he made the graphic with tools and real materials to set the visuals parallel to the theme, and it was the first papercraft infographic in the magazine. A graphic about animals’ tools was made with scissors, a ruler, and glue.

The same data, as a table on the web

The web version was published on February 15, 2017, under the title “The Tools Animals Use.” The credits are Alberto Lucas Lopez and Kennedy Elliott. It shows the same “animal groups x 22 modes of use” data as a colored matrix.

A matrix with the 22 modes of tool use in columns and animal groups in rows, with recorded cells filled in color and callouts of concrete examples attached

Source: Alberto Lucas Lopez, Kennedy Elliott © National Geographic / Information is Beautiful Awards, “The Tools Animals Use”

The order of the columns runs from Affix to Cut, the same as the outside-to-inside order of the layers in the magazine. Chimpanzees and Orangutans, combined into one in the magazine, are split into separate rows in the web version.

Callouts add concrete examples. For instance, “Elephants have been seen draping plants and soil over the carcasses of dead peers.” The magazine graphic includes no such examples; the text carries them instead.

The web version is a 2D table with columns and rows at right angles. The magazine version turns those columns into layers and the rows into bands, and makes it a photograph of a three-dimensional object. The structure of the data is the same, and it can be read as only the form being chosen anew to suit each medium.

There is also an example of researchers charting the same kind of data. The paper by Johnston et al. (iScience, 2020) used Shumaker et al.’s 22 classifications and made a black-and-white pixel table of animal groups x modes of use. Black is observed in the wild, gray is observed with human influence not ruled out, and white is no record. The table was used to estimate the order in which modes of use are acquired. The same type of data thus takes three forms, depending on purpose.

How third parties saw this work

dataphys.org is an entry added by Aurélien Tabard and Pierre Dragicevic on April 30, 2017. It is classified as Passive physical visualization. The description reads as follows.

Infographic designers at the National Geographic crafted a 3D physical visualization out of paper and colored cardboard in order to prepare a 2D infographic.

The summary says the 3D object was made of paper and colored cardboard in order to prepare a 2D graphic. Dragicevic also reposted the author’s GIF on YouTube.

The phrase “in order to prepare” conflicts with the author-side materials. The magazine’s credits list a “photo of paper construction,” and the author writes that he made the graphic “with tools and real materials.” The paper model can be read as the content of the graphic itself, not a stage before the 2D graphic. “Colored cardboard” also cannot be confirmed in the author-side materials, which say “papercraft,” and the magazine says “paper construction.”

Gestalten’s profile of the author, “The Infographic Is Mightier Than The Sword,” mentions it as a challenge in materials.

He recently decided to create the magazine’s first infographic using Kraft paper.

It then quotes López.

If we are not willing to create content beyond what is conventional, we will end up killing journalism.

That is, if one is not willing to make something beyond the conventional, one ends up killing journalism. The article does not name the work. But because “the magazine’s first paper infographic” matches the portfolio, it can be read as referring to this work.

At the Kantar Information is Beautiful Awards 2017, the web version “The Tools Animals Use” made the longlist in the Science, Technology & Health category. What was judged was the web version’s matrix, not the paper magazine version.

Takeaways

What this work offers is a way of treating a physical model as “material for the finished piece.” A three-dimensional object may be built for prototyping, or it may be photographed and become the final graphic. In the latter case, even the steps and shadows of the paper reach the reader’s eye as they are.

In that case, decide in advance which properties carry data and which do not. In “Animal Hacks,” layers correspond to modes of use and whether a leg is visible corresponds to whether a mode is recorded, while depth and shadow can be read as left to the job of telling the layers apart.

The arrangement also seems to matter. The modes recorded in many groups appear to sit in the outer layers, and by ordering the groups from the largest number down, the unrecorded parts gather in the center and take the shape of a staircase.

Another point is matching theme and method. The author gave making a graphic about tools with tools as his reason for the expression. The same data is a table on the web and a photograph of paper in the magazine. It is an example of choosing the form anew for each medium.

Finally, the credits. The magazine prints the name of the person who photographed the paper construction alongside the graphic’s authors. If a physical graphic is photographed and printed, the photography is part of making the graphic.

References

Primary sources

Alberto Lucas López, “Animal tools” (portfolio; description, magazine-spread image, making-process GIF, rights notice). lucasinfografia.com
Magazine-spread image “ANIMAL HACKS” (including the credit lines, source line, and description). Image
Making-process GIF. Image
Alberto Lucas López, “Handmade #dataviz. National Geographic Magazine - March, 2017,” X (Twitter), 2017-03-24. Post
National Geographic, “The Tools Animals Use” (web version), 2017-02-15. National Geographic
National Geographic Magazine, March 2017, “ANIMAL HACKS” (issue table of contents and article excerpt). Zinio
Shumaker, R. W., Walkup, K. R., & Beck, B. B., Animal Tool Behavior: The Use and Manufacture of Tools by Animals, Johns Hopkins University Press, 2011. Johns Hopkins University Press

Secondary sources

Aurélien Tabard & Pierre Dragicevic, “2017 – Hand-Crafted Magazine Infographics,” List of Physical Visualizations, dataphys.org, 2017-04-30. dataphys.org
Pierre Dragicevic, “animal hacks” (repost of the author’s GIF), YouTube. Video
Gestalten, “The Infographic Is Mightier Than The Sword,” Gestalten Journal. Article
Kantar Information is Beautiful Awards 2017, “The Tools Animals Use” (Science, Technology & Health category longlist). Showcase
Johnston et al., “Data-Driven Inference Reveals Distinct and Conserved Dynamic Pathways of Tool Use Emergence across Animal Taxa,” iScience, 2020. doi:10.1016/j.isci.2020.101245 / PMC

Patterns Without a Legend, Painted on Everyday Tableware: Alice Thudt's Life in Clay

Thu, 05 Nov 2026 00:00:00 +0900

Alice Thudt turned data from her own life and the lives of people close to her into patterns with no legend and no axes, and painted them on tableware used every day. The series is called Life in Clay. Her Skype call history with her parents went onto the cereal bowls she used each morning. The messages containing “love” that she and her partner exchanged went onto a teapot the two of them use. Thudt writes that she chose each vessel’s shape and function to suit the experience she wanted to evoke and the person who would receive it.

At the time, Thudt was a PhD student in Computational Media Design at the University of Calgary. The work began as a pottery hobby and later became part of her doctoral research. She presented it in 2017 in the CHI workshop paper “Data Craft,” and the same year it received an Honourable Mention at the Kantar Information is Beautiful Awards. Below, I read it vessel by vessel, following that paper, her doctoral dissertation, and the podcast Data Stories #112, on which she appeared.

A Pottery Hobby Meets Personal Data Research

Data Stories #112 was published on January 3, 2018, with Moritz Stefaner as the host. The quotes from the episode in this article come from the automatic transcript on the show’s site (archive.datastori.es), so they may differ in small ways from the audio.

Thudt says she started pottery about two years before the recording. As her skills improved she was drawn to decoration, and that connected with personal data, the topic of her doctoral research. She made “maybe about 15” pieces and gave most of them to family and friends. The artist’s official website states the aim of the work as giving personal data a presence as an object that you can touch, grow attached to, and use every day.

Her research presentation is the 2017 workshop paper “Data Craft: Integrating Data into Daily Practices and Shared Reflections.” It is a five-page paper by Thudt, Uta Hinrichs, and Sheelagh Carpendale, submitted to a CHI 2017 workshop. It defines data craft as follows.

We define data craft as a practice at the intersection of personal visualization and manual crafting to create physical and functional mementos

In other words, it is a practice that sits where personal visualization meets handcraft, and that makes mementos you can use.

The paper was incorporated into Chapter 9, “Data Craft” (pp.157–168), of her 2018 dissertation, Visualizations for Personal Reflection and Expression. Thudt also has a chapter in the book Making with Data (edited by Huron, Nagel, Oehlberg, and Willett; CRC Press): section 1.2, “Life in Clay” (pp.51–62, DOI 10.1201/9781003264903-2), in the Handcraft part. I have not read the text of that chapter.

Her Parents’ Calls, Painted on Morning Cereal Bowls

Thudt was a doctoral student in Canada while her parents lived in Germany, and they kept in touch mostly over Skype. On the show, the first work she mentioned was the pair of cereal bowls depicting this call history. dataphys.org dates the work to 2015.

Two cereal bowls: the front brown bowl has white vertical lines, and the one behind it has rows of white dots

Source: Alice Thudt / dataphys.org, “2015 – Life in Clay: Sharing Memories through Data Pottery”

According to the paper, both bowls lay out three years of calls, counted from when she left home, along a timeline.

on one bowl, each call is depicted as a line; on the other bowl calls are shown as dots with one dot representing 5 minutes of call time.

On one bowl each call is a single line; on the other the call time is shown as dots, with one dot standing for 5 minutes. The dissertation says the same.

The description on dataphys.org disagrees with this. It says one line is “a day on which we Skyped” and one dot is “10 minutes of calls.” This article follows the paper and dissertation, which the artist wrote herself.

The paper also gives the reason for choosing cereal bowls. They are used every morning, so she meets the data and its meaning every day. She gave one to her parents and kept one herself. The intention, she says, was to remind them to stay in touch. The kind of vessel decides when the data is seen, and giving one bowl to each side decides who sees it; that is how I read it.

A Teapot for Two, a Bowl Like a Journey

The Love Teapot (2016) is a memento of her relationship with her partner. It depicts the text messages containing the word “love” that the two exchanged since they started dating. One dotted line is one year, and one dot is 10 messages.

A white teapot with several vertical dotted lines of black dots drawn on its body and lid

Source: Alice Thudt / dataphys.org, “2015 – Life in Clay: Sharing Memories through Data Pottery”

The paper explains that there are many dots during their long-distance period, and fewer after they moved in together, since they could talk in person. She made it a teapot because it is a shared tool: the two can look back on how the relationship has changed while drinking tea together.

The Moray Coastal Trail Bowl commemorates a trail she hiked with a friend in Scotland, and it was a gift for that friend. The boundary between two glazes represents the coastline they walked, and the white dots on the boundary mark the places where they camped. Its shape resembles the bowl they used during the trip.

A bowl with a blue glaze inside and a gray glaze outside, the wavy boundary between them lined with white dots

Source: Alice Thudt, Uta Hinrichs, Sheelagh Carpendale / “Data Craft: Integrating Data into Daily Practices and Shared Reflections” (2017), Figure 7

In the paper, Thudt describes how she decides on a vessel as follows.

I decide on the associations and encounters I want to enable and take the personality of the pot’s recipient into consideration. This influences the shape and function of the pot I make as well as how it visually reflects the data.

She decides which associations and encounters she wants to bring about and also considers the recipient’s personality, and this shapes the vessel’s form and function and how the data appears.

The table below includes the other vessels as well. The last three appear only on dataphys.org or on the show.

Vessel	Data	Mapping to the visuals	Why this vessel	Source
Parents’ Call Bowls (2015)	Skype calls with her parents (three years)	One call = one line; on the other bowl, one dot = 5 minutes of calls	Cereal bowls used every morning; one for her parents, one for herself	Paper, dataphys.org (year)
Love Teapot (2016)	Messages containing “love”	One dotted line = one year, one dot = 10 messages	A shared tool for drinking tea together	Paper, dataphys.org (year)
Moray Coastal Trail Bowl	A trail hiked with a friend	Glaze boundary = coastline, white dots = campsites	A shape like the bowl used on the trip	Paper
Productivity bowl (2016)	A partner’s year of work	Outside = git commits, inside = emails sent per month	A birthday present	dataphys.org
Dissertation mug	The structure of a friend’s dissertation	How the chapters divide, and the page count of each chapter	For a friend who loves tea, as a “trophy” for the work	Data Stories
Father’s bicycle vessel	Three years of her father’s cycling	A pattern that looks like turning wheels; about three kinds of data	A gift for her father	Data Stories

No Legend, and the Kiln Takes Over

None of the vessels comes with a legend. The paper says this.

I do not include legends so the pots reveal their meaning only to insiders or through storytelling.

Without a legend, a vessel’s meaning is clear only to people who know the background or have heard the story. On the show she says she puts neither axes nor numbers on it (“I don’t put any axes or numbers on it.”).

The dissertation also addresses this ambiguity from the standpoint of privacy.

Keeping the data patterns consciously ambiguous allows me to maintain control over the aspects of a story I want to share.

By keeping the patterns deliberately ambiguous, she can decide which aspects of the story to show.

The dissertation’s Figure 42 summarizes the production process in ten photographs. She chooses a data source, works out the mapping with digital tools and sketches, and creates the pattern and refines it by hand. Then she tries the form on the potter’s wheel, transfers the pattern with a stencil, decorates, and finishes.

Ten photographs: a table-of-contents page, a bar chart of page counts, a sketch, a pattern of red circles, the potter’s wheel, stencil cutting, a mug with the pattern transferred, decorating, and the finished mug

Source: Alice Thudt / doctoral dissertation Visualizations for Personal Reflection and Expression (University of Calgary, 2018), Figure 42

According to the show, she makes the basic shapes in Excel or Tableau, prints them at actual size to fit the vessel’s dimensions, and transfers them.

In a high-temperature kiln the glaze flows, so fine detail is not possible. According to Thudt, her first trials used geometric patterns.

It was always moving and organic in the end. So it makes sense to sometimes just go for the more organic patterns in the first place.

Since the results always came out moving and organic, she says, it makes sense to choose organic patterns from the start. Elsewhere she also says, “I would not want it to be more precise in some ways.” This can be read as a record of a maker who chose how to draw on the premise that the material is imprecise.

Thudt does not expect anyone to read the vessels in detail.

I think a lot of these turn out to be more symbolic than anything else.

Many of them, she says, work as symbols more than anything else. She also says she does not expect anyone to sit down and read out the details. The purpose of the visualization appears to have shifted from getting people to read accurately to reminding them of a relationship or an event each time they use it; that is how I read it.

How Third Parties Saw the Vessels

On the same episode, Stefaner describes the patterns as lying “between decoration and chart.”

I think it works as just an organic decoration pattern, but you can clearly see, like, size differences or arrangement differences. So it’s nicely in between decoration and readable chart, I think.

It works as an organic decoration, but differences in size and arrangement are clearly visible; it sits between decoration and a readable chart, he says.

He also pointed out an asymmetry in gifts. This came in response to Thudt saying that the making process is more of a reflection than the finished piece.

for you, it serves as a meditation on a person and the topic. … and the other person gets just, just like the end of that.

For the maker it is a long meditation on the person and the subject, while the recipient gets only the end result. Thudt herself, speaking of her father’s bicycle vessel, says she does not know whether he can tell which data corresponds to which pattern.

At the Information is Beautiful Awards 2017, the work received an Honourable Mention in the People, Language & Identity category. I have not found the judges’ comments. That a handcrafted piece of tableware was recognized at a data visualization award can be read as meaningful in itself.

On dataphys.org, the researchers’ collection of examples, it was registered by Pierre Dragicevic and classified as “Passive physical visualization.” Its tags include mementos, personal data, and self-logging, which places it as a work of mementos and self-tracking. The 2025 VISAP Pictorial “Balaton Borders” (Gyeviki, Minkó) cites Thudt and colleagues in a single sentence as a precedent for data craft in ceramics.

The preface to the Handcraft part of Making with Data is written by Carpendale and Lora Oehlberg. Carpendale was Thudt’s advisor and co-authored the paper, so this article does not count it as a third-party assessment.

Takeaways

What we can take from Life in Clay to our own visualization work is deciding in advance, as part of the design, who will look at the chart and when. Thudt decided each vessel’s shape and function from the experience she wanted to evoke and the person who would receive it: the morning bowl, the teapot for two, the bowl like a journey. In every case, the choice of vessel can be read as having already decided when the data is seen and by whom.

For on-screen visualization too, we can ask who will see this chart, when, and where. The answer changes how much of a legend to add and how much detail to draw. By leaving out the legend, Thudt’s vessels limited their audience to people who know the meaning. This can be read as an example of deciding whether to add or cut explanation according to the range of people she wanted to show it to.

Her handling of the material is also instructive. Because the glaze flows, Thudt chose organic patterns from the start. It is an example of deciding the level of detail to draw in advance, to fit the precision of the finished piece.

Thudt herself, however, limits how far this practice can be generalized.

These reflections provide anecdotal evidence from a personal perspective.

The dissertation says that these reflections are anecdotal evidence from a personal perspective, and that whether they apply more broadly is a question for future work.

References

Primary sources

dataphys.org, “2015 – Life in Clay: Sharing Memories through Data Pottery” (includes quotes from the artist’s text and a description of the work): https://dataphys.org/list/life-in-clay-sharing-memories-through-data-pottery/
Alice Thudt, “Life in Clay” (the artist’s official website): https://www.alicethudt.de/projects/life-in-clay
Thudt, A., Hinrichs, U., Carpendale, S. (2017). Data Craft: Integrating Data into Daily Practices and Shared Reflections. CHI 2017 Workshop on Quantified Data & Social Relationships. PDF: https://research-repository.st-andrews.ac.uk/bitstream/handle/10023/10910/Thudt_DataCraft_CHI2017.pdf?sequence=1 / Bibliographic record: https://research-repository.st-andrews.ac.uk/handle/10023/10910?show=full
Thudt, A. (2018). Visualizations for Personal Reflection and Expression. PhD dissertation, University of Calgary, Chapter 9 “Data Craft,” Figure 42, 12.4. doi:10.11575/PRISM/31806 / https://doi.org/10.11575/PRISM/31806 / Record: http://hdl.handle.net/1880/106519
Data Stories #112 “Data Pottery with Alice Thudt” (2018-01-03, interview with the artist): https://datastori.es/112-data-pottery-with-alice-thudt/ / Automatic transcript: https://archive.datastori.es/episode/112
Information is Beautiful Awards, “Life in Clay” (the artist’s entry text): https://www.informationisbeautifulawards.com/showcase/2393-life-in-clay
Thudt, A. “Life in Clay.” Huron, S., Nagel, T., Oehlberg, L., Willett, W. (eds.), Making with Data, CRC Press, 1.2, pp.51–62 (text not read). doi:10.1201/9781003264903-2 / https://doi.org/10.1201/9781003264903-2 / Table of contents: https://www.routledge.com/Making-with-Data-Physical-Design-and-Craft-in-a-Data-Driven-World/Huron-Nagel-Oehlberg-Willett/p/book/9781032182223 / https://makingwithdata.org/press

Secondary sources

Moritz Stefaner’s remarks (Data Stories #112, automatic transcript): https://archive.datastori.es/episode/112
Information is Beautiful Awards 2017, “The Winners: People, Language & Identity”: https://www.informationisbeautifulawards.com/news/262-the-winners-people-language-identity
dataphys.org classification and tags (registered by Pierre Dragicevic): https://dataphys.org/list/life-in-clay-sharing-memories-through-data-pottery/
Gyeviki & Minkó, “Balaton Borders: Data Ceramics for Ecological Reflection” (VISAP'25): https://arxiv.org/pdf/2509.02284

Burning Logs with a Glass Ball to Record Sunshine: Charles Sowers's Solar Totems

Thu, 05 Nov 2026 00:00:00 +0900

Charles Sowers’s Solar Totems is a public sculpture in which a glass ball focuses sunlight to burn logs and record sunshine. As the sun moves across the sky, the burn becomes a single line, and when fog or clouds appear, the line breaks. The lens is advanced a little each day, so each day makes one line. Three logs take one year each.

The work was commissioned by the San Francisco Arts Commission and placed on the plaza in front of the Glen Park Canyon Recreation Center. The city’s announcement is dated June 5, 2017, and the artist’s page gives the dates as “2017-2020.” Below, I follow the artist’s official page and the city’s announcement, and then read the assessments by a researcher and by visitors.

2017: Three logs stand on the recreation center plaza

On June 5, 2017, the city’s arts commission announced that a new work by Sowers had joined the renovated Glen Park Canyon Recreation Center. The announcement introduces Sowers as a local artist who weaves his love of science into his work.

According to the announcement, the material is three reclaimed redwood logs. Each is 30 inches (about 76 cm) in diameter and 12 feet (about 3.7 m) tall, and they were set upright on a south-facing open plaza in front of the building. One of them carries a solar-powered mechanism and a spherical lens.

Three redwood logs with cut faces stand on the plaza in front of the recreation center; the one on the right is covered by a transparent tube

From Charles Sowers / Charles Sowers, “Solar Totems”

The work was not delivered in finished form. Once installed, its record builds up over three years. The photo in the city’s announcement shows Tom DeCaigny (director of cultural affairs), Sowers, Mayor Edwin M. Lee, and Phil Ginsburg (general manager of the Recreation and Park Department). DeCaigny said the work speaks to the natural wonders of this urban park.

The work also has a book chapter. In Making with Data: Physical Design and Craft in a Data-Driven World (edited by Samuel Huron, Till Nagel, Lora Oehlberg, and Wesley Willett; A K Peters/CRC Press, 2022), Part 5, Environment, includes chapter 5.3, “Solar Totems.” Its author is Sowers himself, at pp. 345–358, DOI 10.1201/9781003264903-23. The same part also has a chapter on Perpetual Plastic, and the part’s introduction was written by the researcher Dietmar Offenhuber. I have not been able to check the text of the chapter for this article.

Putting a 19th-century sunshine recorder onto wood

The city’s announcement describes where the idea came from.

Inspired by 19th-century meteorological instruments that used a spherical glass lens to burn a sunshine record on a substrate of wood or paper, Solar Totems is a sculptural data-logger that creates an archive of the interaction of sun, weather and site.

In short, the work was inspired by 19th-century meteorological instruments that focused sunlight with a glass sphere and burned a record of sunshine into wood or paper. The city calls the work a “sculptural data-logger” and says it archives the interaction of sun, weather, and site.

The instrument behind it is known as the Campbell–Stokes recorder. According to Wikipedia, John Francis Campbell devised it in 1853, and George Gabriel Stokes improved it in 1879. Campbell’s first device fitted a large spherical lens to a wooden bowl and let the sun scorch the bowl.

In Stokes’s improved version, the recording medium is a card. Since Solar Totems has the sun write directly on wood, it can also be read as going back to a form that predates the card.

Inside a transparent tube, a glass ball held by a metal arm focuses light onto the face of a log, and the wood glows at the focal point

From Charles Sowers / Charles Sowers, “Solar Totems”

The artist’s page describes the mechanism as follows.

Over the course of 3 years, three reclaimed redwood logs receive the “writing” of the sun as its rays are focused by a spherical lens to lightly burn into the wood. As the sun moves across the sky, the burn becomes a line, preserving a record of sunshine periodically broken by fog or cloudy skies.

Over three years, three reclaimed redwood logs receive the “writing” of the sun. Light focused by a spherical lens lightly burns the wood, and as the sun moves across the sky, the burn becomes a line. The line is broken from time to time by fog or cloudy skies, and that break is itself the record of sunshine.

One line is one day, one log is one year

The artist’s page goes on to say how days and years are marked.

The lens is advanced a small distance each day to create a distinct daily line. After one year the heliograph mechanism is transferred to the next log. In this way, the work is completed on-site and becomes a sculptural archive of the specific atmospheric conditions of the site.

The lens is advanced a little each day, which makes a distinct line for each day. After one year, the heliograph mechanism is moved to the next log. In this way the work is completed on site and becomes a sculptural archive of the atmospheric conditions specific to that place.

The artist’s text does not say in which direction the lens is advanced. The two articles read in a later section both say the lens is advanced “upward.” In the artist’s photo, horizontally running burn lines are stacked in many tiers, one above another.

Under a glass ball, black burn lines running horizontally are stacked in many tiers, and the lines near the bottom are broken into dots. Arrow-shaped tags reading “Oct 1” and “Oct 15” line the left edge

From Charles Sowers / Charles Sowers, “Solar Totems”

In this photo, arrow-shaped tags reading “Oct 1” and “Oct 15” are visible along the edge of the log. They are a scale for reading the order of the lines as dates. The correspondence between the data and the appearance is as follows.

Data	Physical appearance / material	Source
Presence or absence of sunshine	Whether there is a mark burned into the wood where the lens’s focal point fell (a burn line)	Artist’s page
Time of day (the sun’s movement)	The length of the line formed as burn marks connect	Artist’s page (that the lines run horizontally is from the photos)
Breaks in sunshine caused by fog or cloud	Gaps in the line	Artist’s page
Date	One line per day, made by the lens advanced a little each day. The position in the vertical stack	Artist’s page (that the lines are stacked vertically is from the photos; that the direction is upward is from third-party articles)
Year	One log. After one year the mechanism moves to the next log (three logs for three years)	Artist’s page, city announcement
Place	The work is fixed on the plaza. The record can only be taken at that place	Artist’s page, city announcement

There is no dataset received from outside. Sunlight that reaches the plaza is burned into the wood on the spot.

From how the photos look, the face of a log can be read as a grid of “horizontal = time of day, vertical = date,” with the presence or absence of burn showing the value. This way of reading is close to that of a calendar heat map. However, the value is not a continuous quantity; it is almost a binary of “burned or not burned.”

Regarding the Campbell–Stokes recorder, Wikipedia says it measures only sufficiently strong sunlight (bright sunshine), not all visible sunlight. The lines of Solar Totems, which uses the same mechanism, can be read as recording only sunshine above a certain intensity. I have not found any source that says this about the work itself.

A photo looking up at a log from below. Hundreds of black burn lines overlap in arcs on the cut face, and date tags line the edge

From Charles Sowers / Charles Sowers, “Solar Totems”

The city’s announcement also touches on year-to-year comparison. When a year’s record is finished, the sunshine recorder moves to the next log, and park users can compare differences in weather from year to year. Placing the three logs side by side can be read as forming a small multiples display that compares the same season across years.

The artist himself says the following in the city’s announcement.

Taken together, the three transformed logs turn the plaza into a kind of civic solar and atmospheric observatory, artistically expanding our understanding of place and connecting us to our environment through that understanding.

Taken together, the three logs make the plaza a kind of civic observatory of sun and atmosphere. He adds that the work expands our understanding of place and, through that understanding, connects people to their environment.

How did third parties see the work?

In his 2020 paper “What We Talk About When We Talk About Data Physicality,” the researcher Offenhuber divided data physicalization along two axes: epistemological versus ontological, and representational versus relational. Solar Totems appears in Figure 9b as an example of the “ontological and relational” quadrant.

Examples in this last quadrant operate without traditional data sets and don’t use visual languages based on symbolic mappings and representation.

Examples in this quadrant work without traditional datasets and do not use visual languages based on symbolic mapping or representation. In the same paragraph, Offenhuber writes that works in this quadrant sometimes take the form of instruments that reproduce the process of collecting data. He describes Solar Totems as a work that records daily hours of sunshine as burn marks left by a spherical lens, and cites it as belonging to the lineage of Campbell’s sunshine recorder.

Joan Sullivan is a photographer in Quebec, Canada, who photographs renewable energy. On October 16, 2017, she wrote “Solar Totems and Wave Walls” for Artists & Climate Change. She reads one line as a day’s weather pattern, such as sunny, partly cloudy, or cloudy, and writes that a year adds up to “365 solar etchings.” The 365 is Sullivan’s count; the artist does not state a number of lines. Her words of appraisal are these.

This is what climate change artists do best: using metaphor to build awareness, a sense of connection and familiarity to overcome anxiety, fear and resistance.

Using metaphor to build awareness, and creating connection and familiarity to overcome anxiety, fear, and resistance: that, she says, is what artists who work on climate change do best. This article was written before her visit.

On July 14, 2018, Steve Jurvetson posted “Solar Totems in Noe Valley” on his blog. The title names the neighboring district, but what is pictured is the work in Glen Park.

The ball lens slowly advances upward over time… focusing the sunlight to burn daily stripes in a 12′ tall log. Each line records the level of sunlight or shade for the day.

The ball lens advances slowly upward over time, burning daily stripes into a 12-foot log, and each line records that day’s sunlight and shade. In the comments, one reader wrote “brilliantly elegant design.”

Inside a transparent tube are a metal arm laid across and a glass ball, with two small solar panels below. The photographer is reflected on the surface of the tube

From Steve Jurvetson / steve.blog, “Solar Totems in Noe Valley”

Takeaways

The burn lines of Solar Totems are not a chart that someone made after collecting sunshine data. An instrument leaves the record on the spot. I read three things that can be taken back into visualization work.

The first is showing the process of recording. The way burn lines increase by one a day shows how data comes into being, as it is. The instruments that Offenhuber describes as reproducing the process of collecting data can be read as pointing to works like this. In the About page, too, the artist writes that he sometimes makes instrumentation for seeing through phenomena that are normally invisible.

The second is the pattern of the grid and the arrangement. A grid that can be read as horizontal = time of day, vertical = date demands the same reading as a calendar heat map. Dividing the logs by year and placing three side by side allows year-to-year comparison. Even though the material is wood, the pattern in use can be read as the same as that of a chart on paper.

The third is taking on the limits of the instrument as well. A line that borrows the mechanism of the sunshine recorder is thought not to record weak sunlight. When a physical record is used for visualization, one also has to check what is not recorded.

The same artist’s Tidal Memory is permanently installed at the Exploratorium and receives measured tide data. Its water levels in 24 glass columns show the hourly tide levels for one day. Whereas Tidal Memory receives external data and plays it back, Solar Totems records the phenomenon directly, on the spot. It can be read that even within one artist’s work there are two types with different relationships to data.

References

Primary sources

Charles Sowers, “Solar Totems,” artist’s official page: https://www.charlessowers.com/solartotems
Charles Sowers CV: https://www.charlessowers.com/pagecv
Charles Sowers, About Me (approach to his work): https://www.charlessowers.com/about-me
San Francisco Arts Commission, “New Public Art in Glen Park,” 2017-06-05: https://www.sf.gov/news-new-public-art-in-glen-park
Charles Sowers, “Solar Totems,” in Samuel Huron, Till Nagel, Lora Oehlberg, Wesley Willett (eds.), Making with Data: Physical Design and Craft in a Data-Driven World, A K Peters/CRC Press, 2022, Part 5 Environment, 5.3, pp.345–358. doi:10.1201/9781003264903-23
Routledge, Making with Data table of contents: https://www.routledge.com/Making-With-Data-Physical-Design-and-Craft-in-a-Data-Driven-World/Huron-Nagel-Oehlberg-Willett/p/book/9781032182223
makingwithdata.org table of contents: https://makingwithdata.org/blog/structure
dataphys.org, “Tidal Memory” (a different work by the same artist; there is no entry for Solar Totems): https://dataphys.org/list/tidal-memory/

Secondary sources

Dietmar Offenhuber, “What We Talk About When We Talk About Data Physicality,” IEEE Computer Graphics and Applications 40(6): 25–37, 2020. doi:10.1109/MCG.2020.3024146 / arXiv: https://arxiv.org/abs/2006.04631 (PDF: https://arxiv.org/pdf/2006.04631 )
Joan Sullivan, “Solar Totems and Wave Walls,” Artists & Climate Change, 2017-10-16: https://artistsandclimatechange.com/2017/10/16/solar-totems-and-wave-walls/
Steve Jurvetson, “Solar Totems in Noe Valley,” 2018-07-14: https://steve.blog/2018/07/14/solar-totems-in-noe-valley/
Wikipedia, “Campbell–Stokes recorder” (background on the original instrument): https://en.wikipedia.org/wiki/Campbell%E2%80%93Stokes_recorder

90% for Range and Peak, 50% for Order: CairnFORM's Ring Chart

Tue, 03 Nov 2026 00:00:00 +0900

CairnFORM is a cylindrical physical chart that shows the hourly amount of renewable energy as the diameters of stacked rings. It has 10 rings, and each ring stands for one hour from 8:00 to 17:59. In December 2017, a test was held at a public event where 90 visitors read the chart. The question asking for the hours with generation (range) was answered correctly by 93.3%, and the question asking for the time of the peak by 92.2%. The question asking visitors to put the hours in order of quantity was answered correctly by 52.2%.

It was made by Maxime Daniel, Guillaume Rivière, and Nadine Couture of ESTIA and LaBRI (University of Bordeaux) in Bidart, France. It is a prototype from Daniel’s doctoral research (2016–2018), and the dataphys.org list carries it as a 2018 work. Below I read it in the order it was made, following the author version of the main paper (TEI 2019), the long French paper (JIPS, 2020 issue), and the lab’s project page.

A vertical bar chart, rotated around its axis

The starting point was not saving energy but “shifting when electricity is used.” In JIPS, the authors write that they aim to get people at work to use electricity when renewable energy is plentiful. The means is the habit of shifting when to charge a laptop’s battery. When they asked in the lab, 15 of the 18 people who use laptops kept them plugged in almost all the time (JIPS 2.1).

To shift charging, people at work need to know when the day’s renewable peak will be. The authors considered shared spaces as the place to put a display. The TEI 2018 paper says that in a vertical bar chart, the bars in front hide the ones behind, so you cannot read it without walking around it, and goes on:

“train stations, public places, airports, roundabouts are public spaces where people are in a hurry and do not have time to change their point of view.”

The point is that in stations, squares, airports, and intersections people are in a hurry and have no time to change their viewpoint.

So the authors built, as a physical object, a vertical bar chart rotated around its axis. This is how TEI 2019 puts it:

“We designed CairnFORM as an illuminated and dynamic vertical bar chart with cylindrical symmetry that allows 360˚-readability.”

It is a glowing, moving vertical bar chart, and its cylindrical symmetry lets it be read from 360°. The “ring chart” of the title is not a doughnut-shaped pie chart. It is a vertical bar chart in which each bar has become a ring. The name comes from the cairn, a pile of stacked stones (TEI 2018).

The hard part was the ring itself. Electronics sit inside it, it must stay round while expanding, and its diameter must grow to about twice its size. A Hoberman-type mechanism turns into a star shape midway, and the Weather Lamp, which uses six arcs, left gaps between the arcs. The authors went through prototypes with four arms and four arcs, four arms and eight arcs, and eight arms and eight arcs, and added lighting to the last design (TEI 2018, JIPS 5.1).

Photo of four ring-mechanism prototypes, from the contracted state to the expanded state. The one at the right end is the finished form, glowing blue

Source: Maxime Daniel, Guillaume Rivière, Nadine Couture (ESTIA) / tangible.estia.fr, “CairnFORM” project page

The finished device has 10 rings and is 105 cm tall (TEI 2019). Each ring weighs 2.3 kg, and its diameter expands from 35 cm to 62 cm. The arcs are transparent PMMA (acrylic). The cost of materials was about 2,350 euros.

PATSTEC, a catalog of scientific and technical heritage, gives the diameter as “30–60 cm,” which differs from the papers. The later study by Bouzekri & Rivière (2026) gives the height of the chart part of the revised version as 92.5 cm. This can be read as 105 cm being the height of the whole device and 92.5 cm the height of the span where the rings are lined up.

One ring is one hour, and the diameter is the renewable amount

TEI 2019 describes the correspondence between rings and quantity this way:

“a fully retracted ring with light off represents a 0 W availability and a fully expanded ring with full blue brightness represents a max W availability.”

The most contracted, unlit ring stands for 0 W, and the most expanded ring at full blue brightness stands for the upper limit of generation capacity (max W). Diameter and brightness redundantly encode the same quantity.

Data	Physical property	How it is mapped / values
Time (8:00–17:59, one hour at a time)	Ring position in the stack	10 rings = 10 hours. 8:00 at the bottom, 17:00 at the top
Average renewable amount for that hour	Ring diameter	0 W = 35 cm, max W = 62 cm (a width of 1.8 times)
The same amount (redundant encoding)	Brightness	0 W = off, max W = maximum blue brightness
Range shown	Cut-out of time	One daily peak (valley → peak → valley) shown at a time, in sequence
The arrival of the next peak	Motion of the rings expanding and contracting	Animated over 6 seconds
Hue	Does not represent data	Changed from blue to green after the public test
What the data is	Wooden tree model on top and the words “Renewable Energy”	Added as a removable legend

Source: TEI 2019, JIPS (5.2, 6.1, 7.9.1).

There were two reasons for the one-hour step (JIPS 3.4.4). A time window such as “unplug from 15:00 to 16:00” is easier to remember as an instruction. And generation forecasts are more reliable as hourly averages than at minute resolution.

Another rule is not to show a whole day’s peaks at once. In a test with printed paper charts, with 3 groups of 7 people, when all the peaks were drawn together, most people could find only the highest peak (TEI 2019). When the peaks were shown one at a time in sequence, they found all the peaks.

Three photos of a glowing blue ring taken from directly above. The diameter grows from left to right

Source: Maxime Daniel, Guillaume Rivière, Nadine Couture (ESTIA) / tangible.estia.fr, “CairnFORM” project page

December 2017, Biarritz: 90 people read it

On December 1–2, 2017, the authors set up the 10-ring version at a public event, the “24h de l’innovation,” in Biarritz (JIPS chapter 7). They gathered 90 visitors who stopped by and divided them into 3 groups of 30. What they showed was neither measured nor forecast data but synthetic data built so that the smallest difference in diameter was 1 cm (JIPS 7.3). The peak lengths were 3 hours (V3h), 6 hours (V6h), and 9 hours (V9h), one per group.

There were three questions. Name the hours with generation (range). Name the time of the peak (compare). Put the hours in order from smallest to largest quantity (order). The results are in Table 3 of TEI 2019.

Peak shown	Range	Peak	Order
V3h (3 hours)	100%	93.3%	80.0%
V6h (6 hours)	93.3%	93.3%	63.3%
V9h (9 hours)	86.6%	90.0%	13.3%
Average	93.3%	92.2%	52.2%

Accuracy for range and peak was around 90% regardless of the peak length (p = .12, p = .85). Accuracy for order fell as the peak got longer (p = .001). The authors hypothesize that this is due to the limit on how much information people can handle at once (Miller, 1956). It is because one has to compare values many times before and after the peak and keep them in mind.

“physical bar charts and physical ring charts (i.e., physical vertical bar charts with cylindrical symmetry) seem better suited to range tasks and compare tasks, rather than order tasks.”

This is the summary from TEI 2019: physical bar charts, and the ring charts that turn them into cylinders, seem suited to range and compare tasks, not to order tasks.

The test also led to a redesign. One visitor thought it was a clock, and another thought it was a countdown to the end of the event (JIPS 7.8). Some said green would suit renewable energy better than blue. The authors changed the rings to green and put a wooden tree model and the words “Renewable Energy” on top.

Six photo frames. In the three on the left, the color is changing from blue to green and there is no tree model; in the three on the right, a tree model and a “RENEWABLE ENERGY” legend sit on top, with labels from 8H to 17H

Source: Maxime Daniel / dataphys.org, “CairnFORM: a Physical Ring Chart Showing Renewable Energy Data”

The redundant encoding with brightness did not work well. 9 of the 90 people said they could not tell the differences in brightness. Light spilled between the overlapping transparent arcs and made the whole thing too bright (TEI 2019).

Five people also said the differences in diameter were hard to see. The sources disagree on the step value: JIPS gives the smallest diameter difference as 1 cm (30 levels), while TEI 2019 says “the radius granularity was 9 mm.” Since 27 cm divided by 30 is about 0.9 cm, “radius” can be read as meaning the diameter step. JIPS writes that it was later made coarser, to 2 cm (15 levels).

On the UEQ-S, the hedonic quality scored 2.25 and the pragmatic quality 1.67 (on a scale of −3 to 3). The body text of TEI 2019 has the labels “excellent” and “good” the opposite way from JIPS, and judging by the order of the scores, JIPS can be read as correct.

Noticing an expanding ring at the edge of vision

Next is the workplace. The authors tested in the lab whether a person working at a desk could notice the movement of a ring at the edge of their vision (TEI 2019). There were 30 participants and a single ring, placed about 2.6 m away at a 45° angle.

Every movement took 6 seconds, and three ways of varying the speed were compared. The share of people who noticed the movement while concentrating on another task was highest for the exponential type, which starts slowly and accelerates, at 72.2%; constant speed gave 60.5%, and the logarithmic type, which starts fast and decelerates, gave 66.5% (TEI 2019). The results section of JIPS writes 129/167 = 77.2%, and the conclusion gives 72.2%. Either way, the exponential type has the highest value.

After that, a two-month test was run in actual workplaces. However, all I could check is the abstract of a short TEI ‘21 paper.

“Six participants had to shift energy consumption by using energy storage. To do so, they were notified about local energy forecasts. Compared with flat-screen animations, early results show that cylindrical shape-change animations keep a better attractiveness over time.”

Six people used energy storage to shift their consumption and were notified of local energy forecasts. These are early results: compared with flat-screen animations, the cylindrical shape-change animations kept their attractiveness better over time.

The summaries differ between sources. The abstract of the doctoral thesis says “as practical as, and more hedonic than” the flat screen. The project page says that in an 8-week experiment both qualities were above the screen. I have not seen the full text, and I have not checked which is accurate.

How did third parties see this work?

Bae et al. (CHI 2022), in a paper on a design space for physicalization that classifies 47 works, take up CairnFORM.

“Two such examples include Bookly [45] and CairnFORM [13] which both encode data using volume. … studies in visualization show that people have trouble reasoning over changes in volume and are better at reasoning over other visual variables such as surface area”

CairnFORM is given as an example of encoding by volume, and the point is that people are better at reading changes in area than in volume. The authors themselves write “diameter (and brightness),” so there is a mismatch in classification. It can be read that which perceptual findings apply depends on whether the ring is read as the size of a circle or as the swelling of a solid.

Bouzekri & Rivière (2026) compared a revised CairnFORM with PlantFORM, a plant-shaped device made of paper and wood, and with a screen version (28 people). It is a follow-up study by the co-author Rivière, so it is not a fully independent evaluation. Across the six questions asking for charging and discharging times, CairnFORM’s accuracy was highest at 96% (the difference from the screen version CairnSCREEN’s 94% was not significant, p=.187), and 54% of people also chose it as “most innovative.” On the other hand, one participant pointed out a mismatch between the material and the theme.

“We are talking about renewable energy; it affects me enormously, so all this Plexiglass [of CairnFORM] […], if we are in renewable energy, we must go all the way in the product we offer.”

The opinion is that if the subject is renewable energy, the product should go all the way, not be this mass of acrylic.

PATSTEC (a program preserving France’s contemporary scientific and technical heritage) has registered CairnFORM in its catalog as an object worth keeping. It notes that the prototype is now at ESTIA.

Uses

The first is to decide the form from the place where it will stand. The cylinder was a form for places with many people who have no time to walk around, so that the bars do not hide each other. Writing out first from which directions people come and how long they stop can narrow down the candidate forms, it seems.

The second is to choose the display from the question you want answered. Only 13.3% could put the 9-hour peak in order of size. For “when should I charge,” being able to read the peak is enough, and conveying a ranking as well would need something like numeric labels, it seems.

The third is to reduce the amount shown at once. When peaks were shown one at a time, even small peaks were not overlooked. In a physical chart with a limited number of rings, switching in sequence may suit the question better.

The fourth is that the material breaks the encoding. Brightness became hard to tell apart because light spilled over, and a glowing, moving cylinder was mistaken for a clock. For a physical chart, the design needs to include the behavior of the material and the cues for “what the data is.”

References

Primary sources

dataphys.org, “CairnFORM: a Physical Ring Chart Showing Renewable Energy Data” (registered by Maxime Daniel, 2019-06-05): https://dataphys.org/list/cairnform-a-physical-ring-chart-showing-renewable-energy-data/ (tag list: https://dataphys.org/list/tags/self-actuated/ )
Maxime Daniel, Guillaume Rivière, Nadine Couture, “CairnFORM: a Shape-Changing Ring Chart Notifying Renewable Energy Availability in Peripheral Locations,” TEI ‘19, pp. 275–286, 2019. Author version: https://tangible.estia.fr/prototypes/cairnform/pub/papers/cairnform-peripheral_tei2019.pdf / DOI: https://doi.org/10.1145/3294109.3295634 / HAL: https://hal.science/hal-01976793
Maxime Daniel, Guillaume Rivière, Nadine Couture, “Designing an Expandable Illuminated Ring to Build an Actuated Ring Chart,” TEI 2018 Work in Progress, 2018. Author version: https://tangible.estia.fr/prototypes/cairnform/pub/papers/illuminated-ring_tei18_wip.pdf / DOI: https://doi.org/10.1145/3173225.3173294
Maxime Daniel, Guillaume Rivière, Nadine Couture, “CairnFORM,” IHM 2018 Démonstrations, 2018: https://tangible.estia.fr/prototypes/cairnform/pub/papers/cairnform-demonstration_IHM2018.pdf (HAL: https://hal.science/hal-01900064 )
Maxime Daniel, Guillaume Rivière, Nadine Couture, “Un histogramme cylindrique à changement de forme pour afficher des prévisions de disponibilité de l’énergie,” Journal d’Interaction Personne-Système, Vol. 9, No. 1, 2020: https://jips.episciences.org/articles/7102 (PDF: https://jips.episciences.org/en/articles/7102/download / DOI: https://doi.org/10.46298/jips.7102 )
ESTIA, CairnFORM project page: https://tangible.estia.fr/prototypes/cairnform/
Maxime Daniel, CairnFORM (control software, GPL-3.0): https://github.com/maximedaniel/CairnFORM
Maxime Daniel, “CairnFORM: a Shape-Changing Ring Chart Notifying Renewable Energy Availability” (YouTube): https://youtu.be/wZ_3CtDllbA
Checked for the abstract only
- Maxime Daniel, Guillaume Rivière, “Exploring Axisymmetric Shape-Change’s Purposes and Allure for Ambient Display,” TEI ‘21 Work in Progress, 2021: https://hal.science/hal-03040877
- Maxime Daniel, doctoral thesis “Afficheurs cylindriques à changement de forme,” Université de Bordeaux, 2018: https://theses.hal.science/tel-02371620
- IHM 2017 Travaux en cours (CAIRNS): https://hal.science/hal-01578006
- ETIS 2017: https://hal.science/hal-01685710

Secondary sources

S. Sandra Bae et al., “Making Data Tangible: A Cross-disciplinary Design Space for Data Physicalization,” CHI 2022: https://arxiv.org/abs/2202.10520
Élodie Bouzekri, Guillaume Rivière, “Shaping Plant-Like Shape-Changing Interfaces as Vertical Charts,” arXiv, 2026 (follow-up study by co-author Rivière): https://arxiv.org/abs/2604.15902
PATSTEC, CairnFORM (Fiche N° 4104): https://www.patstec.fr/ressources/objets/detail?id=35997
(Mentioned by name only) Zhang, Ilić, Signer, arXiv, 2025: https://arxiv.org/abs/2503.06583

Washing Dirt Off a Wall to Draw Circles, Then Letting Settling Dust Erase Them: Dietmar Offenhuber's Staubmarke

Thu, 05 Nov 2026 00:00:00 +0900

Dietmar Offenhuber’s Staubmarke (pronounced “SHTOWP-mar-keh”; the English title is Dustmark) is a work in which stencils and a pressure washer are used to clean dirt from walls and other surfaces, leaving rows of circles made of dots of different sizes, drawn as a scale. By reading the scale on each mark, viewers can roughly estimate how thick the accumulated layer of dirt is. As dust settles again, the pattern disappears, starting with the smallest circles.

The work was made in October 2018 for Drehmoment, an arts festival held in Stuttgart, Germany. What follows is a reading based on the brochure from the organizer KulturRegion Stuttgart, Offenhuber’s paper “Data by Proxy” (2019), and a press article from Northeastern University. The artist’s official page and the project site dust.zone could not be opened this time.

Washing a Wall in Stuttgart, 2018

Drehmoment was an industrial arts festival organized by KulturRegion Stuttgart, held from October 4 to 28, 2018. Staubmarke was presented together with Michael Saup’s VR work “STAUB” in a single brochure, and the listed partners include the High-Performance Computing Center Stuttgart (HLRS). “STAUB” was shown inside St. Mary’s Church (Marienkirche), while the Staubmarke marks were placed at several locations in the city.

The brochure states the premise shared by both works as follows.

Elektronische Sensoren liefern zwar genaue Messdaten, die Zahlen und Maßeinheiten bleiben für uns jedoch abstrakt.

In other words: electronic sensors do deliver precise measurements, but the numbers and units remain abstract for us.

Offenhuber’s technique is reverse graffiti. Rather than adding paint to a dirty surface, he washes away only part of the dirt to bring out a design. According to a Northeastern University press article (December 2018), he used a pressure washer and stencils to wash patterns into city walls and pavements.

A metal stencil with cut-out circles and letters lies on the ground while a pressure washer sprays water over it

Source: Dietmar Offenhuber / Information is Beautiful Awards, “Dustmarks And Ozone Tattoos: Autographic Displays Of Air Pollution”

The organizer describes the artist’s aim like this.

Dabei geht es ihm darum, einen physischen Referenzpunkt zu den digitalen Messwerten herzustellen.

That is, the aim is to create a physical reference point for the digital measurements.

Circles of Different Dot Sizes Become a Scale

According to the brochure, each mark contains a scale (Skala) uncovered from the layer of dirt.

Jede Staubmarke zeigt eine aus der Staubschicht freigelegte Skala. Anhand dieser können Betrachterinnen und Betrachter ungefähr die Stärke der abgelagerten Patina einschätzen.

In other words: viewers can use this scale to roughly estimate the thickness of the accumulated layer of dirt.

In the photographs, a mark takes the form of a filled circle at the far left, followed by circles made of small dots. The dots get smaller toward the right.

On a dirty concrete wall, a filled circle and three circles whose dots differ slightly in size are lined up, with the address dust.zone washed in below

Source: Work by Dietmar Offenhuber (photographer not credited) / Boston Hassle, “WENT THERE: World Memory, The Art of Data Visualization @ Boston CyberArts Gallery”

Over time, the pattern disappears. The paper “Data by Proxy” says that the finest details go first.

With ongoing pollution, the markers will fade over time, starting with the most delicate textures in the pattern.

The press article likewise explains that as new dust settles on the washed areas, the smallest circles vanish first. This can be read as meaning that the finest step of the scale is the first to disappear, while the coarser steps remain longer.

The correspondence between data and appearance, as far as the sources I examined show, is summarized in the table below.

Data / phenomenon	Appearance / material	What it does	Source
Particulate matter accumulated on city surfaces	Difference in brightness between washed and still-dirty areas	No paint is added; dirt is selectively removed	Paper, §7.3
Thickness of the accumulated layer	A scale uncovered from the layer of dirt	Viewers roughly estimate the thickness of the layer	Brochure
Continued accumulation afterward	The order in which the pattern disappears	Fine elements, the smallest circles, disappear first	Paper, §7.3; press article
Dirt that varies by place	Where the marks are placed	The same type of mark at several locations in the city	Brochure; press article
Measurements from a citizen sensor network	Mark locations overlap with measurement sites	Measured values can be compared with the appearance of the marks	Brochure; paper, §7.3
Substrate	Concrete or stone walls and pedestals	The paper’s figure says “washed into concrete”	Paper, Fig. 9

This can be read as: the more dirt there is, the more clearly the washed circles show, and the more dirt resettles, the more the circles fade. Because the image fades as the quantity grows, it can be read as the reverse of an ordinary chart, which adds ink. The scale can be read as working like a legend for reading the layer’s thickness at a given moment. At the same time, the number of steps that have vanished can be read as a clock that records how much has been deposited since the mark was washed.

Four Locations or Five: Where the Marks Are and What Was Measured

The number of locations where marks were placed differs between sources. The brochure lists five, RG1 to RG5.

No.	Location (as given in the brochure)
RG1	Under the Paulinen Bridge (Tübinger Straße)
RG2	St. Mary’s Church (Tübinger Straße 36)
RG3	Pedestal of a flowerbed on Schillerplatz (in front of the entrance to the Old Palace)
RG4	Wall near the Wagenburg Tunnel (Konrad-Adenauer-Straße)
RG5	Staatsgalerie Stuttgart (Konrad-Adenauer-Straße 30–32)

The press article, on the other hand, writes “His website features four locations.” That is a statement that the project site showed four locations. The brochure says the latest locations are listed on dust.zone, but because I could not open that site, I was not able to confirm which number is final.

On the side of a stone flowerbed pedestal, a filled circle and a circle made of dots are lined up, with the address dust.zone washed in to the right

Source: Dietmar Offenhuber / Information is Beautiful Awards, “Dustmarks And Ozone Tattoos: Autographic Displays Of Air Pollution”

The choice of locations is related to the measurements. The brochure says that references to regional measurements from the citizen measurement network luftdaten.info were added. The paper explains this in a little more detail.

The markers were applied at locations also surveyed by the sensors installed by a citizen sensing initiative, allowing the comparison between data values and physical appearance of the markers.

In other words: the marks were also placed at sites measured by sensors installed by a citizen sensing initiative, so the data values can be compared with the physical appearance of the marks. According to the press article, the project site had links from each location to the citizen group’s data.

As for numerical values on individual marks, they cannot be confirmed from the artist’s side. The exhibition review in Boston Hassle says that each mark also showed the average PM2.5 value for its location. A photograph the artist’s side submitted to the Information is Beautiful Awards also has text under a circle that reads “PM2.5 18 µg/m³.” However, the sources I could read do not say when or in which measurement that value was taken.

On the concrete parapet of a bridge over a road, a row of circles, the address dust.zone, and what appears to be a PM2.5 figure are washed in, with a person walking past in front

Source: Photo by Frank Kleinbach (according to the image file name), work by Dietmar Offenhuber / Information is Beautiful Awards, “Dustmarks And Ozone Tattoos: Autographic Displays Of Air Pollution”

Its Place in “Data by Proxy”

In the paper “Data by Proxy — Material Traces as Autographic Visualizations,” presented at IEEE VIS in 2019, Offenhuber cites this work as an example. An autographic visualization, in the paper’s terms, is a visualization that starts from traces left on physical things rather than from data translated into symbols. Staubmarke appears in Section 7.3, “Sensory Accountability.”

As an example, the project Staubmarke (Fig. 9) visualizes air pollution by applying visual markers on urban surfaces to make accumulations of particulate matter legible.

That is, it is introduced as an example that places markers on urban surfaces so that accumulations of particulate matter can be read.

We use the term sensory accountability to describe approaches that call attention to the complex material qualities of a phenomenon that is often reduced to a single quantitative dimension.

In other words: the paper calls approaches that draw attention to the complex material qualities of a phenomenon often reduced to a single quantity “sensory accountability.” The paper goes on to say that it means asking what a sensor captures, what it ignores, and how the recorded values correspond to the phenomenon as sensed.

The caption of Fig. 9 in the paper classifies this work’s design operations as “aggregating, encoding.” In Table 2 of the same paper, Aggregating means gathering material so that it becomes visible, and Encoding means adding a scale that divides up the phenomenon. Since the work uses a layer of accumulated dirt and washes a scale into it, these two operations can be read as applying to it directly.

The work also has a chapter in the book Making with Data (edited by Huron, Nagel, Oehlberg, and Willett; CRC Press). It is 5.4, “Staubmarke (Dustmark),” in Part 5, “Environment,” written by Offenhuber, pp. 359–369, DOI 10.1201/9781003264903-24. I could not read the chapter text this time. The same Part 5 also includes Perpetual Plastic and Solar Totems. Both are works in which a phenomenon leaves traces in the material at hand. They differ in the direction of the trace: in Solar Totems, burn lines accumulate and remain, whereas in Staubmarke the pattern fades away.

How Third Parties Saw the Work

Gina Lindner wrote about this work in an exhibition review in Boston Hassle (May 2019). The exhibition was “World Memory: The Art of Data Visualization” at the Boston Cyberarts Gallery.

The first and most memorable work I encountered was dust.zone

The review calls it the first work she met at the venue and the most memorable. Lindner writes that on the washed, clean surface a “utopia” of a pollution-free world peeks through. She then points out that it is a time-based work whose patterns disappear within a few months.

dust.zone is also uniquely time-based: these crisp patterns will fade over the coming months as dust reaccumulates.

The same review also reports an incident in Boston. Offenhuber tried reverse graffiti on the gravel outside the gallery, but there was not enough dirt, and no pattern appeared. This is written as something she heard from a gallery staff member.

From the visualization community, an entry that paired it with Ozone Tattoo, “Dustmarks and Ozone Tattoos,” made the Longlist in the Unusual category of the Kantar Information is Beautiful Awards 2019. It was not a winner.

The Northeastern University press article gives as background a conflict between citizen groups and city authorities over how to measure air pollution. Because it is a university press release, it is not an independent critique. I did not find any independent commentary by researchers in this survey.

Takeaways

What we can take away from this work is a design in which disappearing is itself the scale.

In an ordinary chart, marks increase as values increase. In Staubmarke, the more dust settles, the more the image disappears. You draw a graded pattern once and let the phenomenon erase it; by looking at how far it has been erased, you can read how much has accumulated since it was drawn. That is how I read it.

The other is a reference point for checking numbers against the physical thing in the same place. The marks were also placed at the citizen sensors’ measurement sites, so that the values could be compared with the appearance of the wall. It can be read as a way of returning a number on a screen to the place where that number was measured.

The author himself writes about the limits. The paper says that producing an autographic visualization is slow and bound by the materials available. According to Boston Hassle, on the gravel outside the gallery there was too little dirt for a pattern to appear. This method can be read as usable only when the phenomenon leaves enough of a trace on the spot.

References

Primary sources

KulturRegion Stuttgart, “Drehmoment Produktionskunst-Festival 4.–28.10.2018: Michael Saup | Dietmar Offenhuber »STAUB« | »Staubmarke« Mit: HLRS” brochure (PDF, 2018): https://kulturregion-stuttgart.de/fileadmin/user_upload/Projekte/2018_Drehmoment/Kommunen-Projekte/Einzelprogrammhefte/Drehmoment_Stuttgart_Saup_Offenhuber_WEB.pdf
Dietmar Offenhuber, “Data by Proxy — Material Traces as Autographic Visualizations,” arXiv:1907.05454, 2019 (IEEE VIS 2019; Section 7.3, Fig. 9, Table 2): https://arxiv.org/abs/1907.05454
Information is Beautiful Awards, “Dustmarks And Ozone Tattoos: Autographic Displays Of Air Pollution” (entry text and photographs, 2019): https://www.informationisbeautifulawards.com/showcase/3977-dustmarks-and-ozone-tattoos-autographic-displays-of-air-pollution
Dietmar Offenhuber, “Staubmarke (Dustmark),” in Making with Data, pp. 359–369 (Part 5, 5.4): https://doi.org/10.1201/9781003264903-24
Making with Data book site, editors and structure of the parts: https://makingwithdata.org/blog/structure
Table of contents of Making with Data (bookseller’s page): https://preigu.de/buecher/making-with-data/122556908
Pages that could not be opened this time: the artist’s official page https://offenhuber.net/project/staubmarke-dustmark/ / project site http://dust.zone/

Secondary sources

Gina Lindner, “WENT THERE: World Memory, The Art of Data Visualization @ Boston CyberArts Gallery,” Boston Hassle, 2019-05-14: https://bostonhassle.com/went-there-world-memory-the-art-of-data-visualization/
Khalida Sarwari, “He came up with a way to make data tangible and exciting,” Northeastern Global News, 2018-12-18 (university press office): https://news.northeastern.edu/2018/12/18/heres-one-way-to-make-data-tangible-and-exciting
Kantar Information is Beautiful Awards 2019, Longlist (Unusual category): https://www.informationisbeautifulawards.com/showcase/3977-dustmarks-and-ozone-tattoos-autographic-displays-of-air-pollution

The Lines Appear to Be Physical; the Scales Are Text Added Afterward — Steven Pinker's Glowing Line Charts

Tue, 03 Nov 2026 00:00:00 +0900

In a video released in 2018 on Bill Gates’s channel, the line charts in front of Steven Pinker appear, judging from the footage, to be real glowing lines bent and hung in place. The scales, titles, and country names are text added over the video afterward. I have not found any explanation from the production side. The video runs about two and a half minutes and shows three line charts: global life expectancy, life expectancy by age in England, and the share of children who die before age 5.

My entry point is the item “Steven Pinker explains Global Life Expectancy with Physical Line Charts” in dataphys.org, a catalog of data physicalization. The observations and timestamps below are based on the video Gates posted on X (formerly Twitter; about 2 minutes 29 seconds) and its English subtitles. The YouTube version matches in subject, publisher, and timing, but I have not compared the two frame by frame.

February 2018: Two and a Half Minutes on Gates’s Channel

The video is titled “Humanity is fighting back against the Grim Reaper”. It was published on Bill Gates’s YouTube channel on February 13, 2018 (YouTube). The description is a single line.

Author Steven Pinker explains how life spans have skyrocketed in the past two centuries.

In other words, the video presents Pinker explaining how much life spans have grown over the past two centuries. The link in the description redirects to the page for Bill and Melinda Gates’s 2018 Annual Letter, and the URL contains a tracking string, “AnnualLetter2018_Pinker”. I have not been able to read the body of the destination page, but the link suggests the video was made to accompany the Annual Letter.

Eleven days later, on February 24, 2018, Gates posted the video on X.

People today are living longer, healthier, and happier lives than ever before. I asked @sapinker to explain why.

The post says that people today live longer, healthier, and happier lives than ever before, and that Gates asked Pinker to explain why. Because it says “I asked”, the video can be read as one requested from Gates’s side.

Item	Details
Speaker	Steven Pinker. The opening name caption reads “Author / Psychologist / Professor / Harvard University”
Published by	Bill Gates’s YouTube channel (February 13, 2018); X (February 24, 2018)
End screen	The book title “ENLIGHTENMENT NOW” drawn in glowing lines, followed by the “gatesnotes” logo
Production company / director	Not shown in the video or in any text source; cannot be identified

Enlightenment Now is a book Pinker published with Viking in 2018. Its official site describes it as “In 75 jaw-dropping graphs”. I have not been able to confirm which figures in the book correspond to the three graphs in the video.

The dataphys.org entry is titled “Global Life Expectancy”, but the video shows three graphs.

	Title on screen	Time on screen	Lines
A	Global Life Expectancy by Age	about 0:07–0:45	1 red line
B	Life Expectancy by Age in England	about 0:45–1:22	10 lines (at birth, and at ages 1 to 70)
C	Children Dying Before Age 5	about 1:22–2:20	4 lines, then 1 more added for 5

The longest is C (about one minute), which is not in the entry’s title. C is also where a country is added.

The Lines Can Be Held in the Hand

The footage offers four clues that the lines are physical objects.

In the close-up from 0:05 to 0:09, Pinker’s hand grips the end of A’s red line.
In the wide shots at 0:04 and 2:20, the three graphs stand in separate places in the same room. Pinker’s head comes in front of B’s lines and hides them.
In the close-up from 2:04 to 2:06, a hand brings in a yellow line and hangs it on C’s frame. The line sways and settles into place.
In the close-ups, the way the lines look changes with the camera angle.

A close-up in which many glowing lines of England’s life expectancy by age cross the frame, with Pinker behind them bringing his hand toward a line

Source: frame from the Bill Gates (Gates Notes) video / dataphys.org

In the close-up around 1:39, Sweden’s line is visible. It is a thin wire bent into the shape of the data and glowing blue all over. It can be read as an EL wire (a thin cable that glows with electricity) or an LED light tube of some kind. I have not found any source that discloses the material. The Ethiopia line that is brought in bends, so the material is somewhat flexible.

The frame is a metal post standing on the floor. The left and right edges of each graph line up with the posts. The horizontal line at the bottom also looks like a physical crossbar, but that is a reading from appearance.

There are things I could not confirm. dataphys.org writes that the graphs sway slightly when Pinker touches them. In still frames, I could not tell a sway caused by touch apart from camera shake. The line clearly moves when the Ethiopia line is brought in. I have also found no source in which the production side says the lines were built physically, so from the footage alone I cannot fully rule out CG compositing.

The Scales and Country Names Sit on Top of the Footage

Unlike the lines, the text attached to the graphs is not physical. The deciding scene is the final wide shot around 2:20. Only the lines of the three graphs appear in the room, with no scales, titles, or country names. The same lines are on screen without the text, so the text can be identified as something laid over the footage in editing.

A wide shot in which Pinker points left among four glowing line charts, with white text for scales, titles, and country names across the screen

Source: frame from the Bill Gates (Gates Notes) video / dataphys.org

Sorting the correspondence between data and appearance by physical object versus overlaid text gives the following.

Data	Appearance	Physical or overlaid text (judgment from the footage)
Values (years of life expectancy, share dying before 5)	Height of the line	Physical line
Year	Horizontal position of the line	Physical line
Series (age in B, country in C)	Color of the line	Physical line (the lines appear to glow by themselves)
Series names (ages, country names)	White text near the line	Overlaid text
Scales, axis numbers, titles	White text at the left, right, and bottom of the screen	Overlaid text
Frame	Metal posts	Physical
Adding a series	A line is brought by hand and hung	Physical action

The physical object handles the shape of the lines, and the video handles the annotation needed to read that shape as numbers. It can be read as the two together making a graph that can actually be read.

A wide shot in which Pinker points to the upper right of a bundle of glowing lines showing England’s life expectancy by age

Source: frame from the Bill Gates (Gates Notes) video / dataphys.org

Measuring the frames with the overlaid scales as a ruler also reveals how the axes are built. B’s vertical axis has tick intervals that get tighter toward the top, so they are not evenly spaced. The ticks on A’s horizontal axis are 1770, 1820, 1870, 1913, 1950, 1973, 2000, and 2015. The gaps between years vary from 50 years to 15 years, yet the spacing between ticks on screen is nearly equal at about 130–170 px, and the bends in the line sit at those years’ positions. This can be read as a line laid out horizontally by the order of the data points, not by the length of time. As a result, the second half, 102 years from 1913 to 2015 (about 630 px), looks wider than the first half, 143 years from 1770 to 1913 (about 430 px).

The scales are overlaid, but the source of the data is not shown anywhere in the video, nor in the YouTube description. Comparing against the current Our World in Data (OWID) series gives the following.

Graph	Series compared	Points that match	Points that do not match
A World	OWID life expectancy	The years on the scale are a subset of the years in OWID’s world series	From 1913 to 1973, the video’s values are 3–5 years higher
B England	OWID life expectancy at different ages	The age-specific lines start in the 1840s	The video’s value for 1841 is lower
C Under-5 deaths	OWID child mortality	Sweden starts at about 30% in 1751, with a peak in the 1770s	The years where the Canada and Chile lines start differ

For C, Sweden’s starting year and peak position match well. But all of these are comparisons with the current versions, and I cannot tell whether the differences come from the 2018 versions or from reading error. The source cannot be determined.

Tracing and Adding Lines: Narration and Hand Movements

Because the lines are handled as physical objects, as they appear to be, Pinker’s hand movements double as the steps of the explanation. The correspondence between narration and screen is as follows.

Time	Narration	Screen
0:01–0:12	Newborns long ago could expect to live only 30 years	From a wide shot to a close-up of a hand gripping the end of the red line
0:12–0:24	Newborns today can expect 70 years or more, and 80 or more in developed regions	A finger traces A’s line from left to right and points to the high point at the right end
0:45–1:22	Looking at the lines by age, using England as an example	The hand moves up from the lower lines to the upper lines
1:50–1:58	Chile cut its child mortality in 30 years	The hand indicates the steep drop of the red line
2:04–2:07	Looking at an African country such as Ethiopia	A hand brings in the yellow line and hangs it; in the next shot the text “ETHIOPIA” is added

In C, the lines are not all there from the start.

Take an African country like Ethiopia. In 1980, almost a quarter of the children died before their fifth birthday.

The narration says that in Ethiopia in 1980, almost a quarter of children died before their fifth birthday. On this sentence, Pinker brings in the yellow line by hand. The movement of one line being added coincides with the turn in the story toward “progress is happening in developing countries too”. The narration also says, “Each line shows a country representing a different continent.” The five countries can be read as one each for Europe, North America, Asia, South America, and Africa. The subtitles name only Sweden, Chile (“In Latin America”), and Ethiopia (“an African country”).

Because the line is brought and hung by hand, it sways and takes a moment to settle. The addition of a new series is visible as a movement. By choosing physical lines, the video seems to have been able to act out, on screen, the steps of the explanation: tracing with a finger, raising a hand, adding one line.

How Third Parties Saw This Video

The dataphys.org entry was registered by Yvonne Jansen, with information contributed by Cedric Honnet.

At first glance it appears that line charts are digitally overlaid on the video. However, closeups and a light swaying of the graphs when Steven Pinker touches them, as well as the addition of another country towards the end of the video indicate that the graphs visible in this video are actually physical line graphs.

The reading is that, at first glance, the line charts look digitally overlaid on the video, but the close-ups, the sway on touch, and the addition of a country near the end show that they are physical line charts. The category is “Passive physical visualization”, and the tags are “rearrangeable” and “storytelling”. On the lines, this matches the clues in the footage, but the text really is digitally overlaid. As for the sway, I could not confirm it in still frames.

Jansen, Pierre Dragicevic, and Jean-Daniel Fekete presented an evaluation experiment on physical visualizations at CHI 2013 (article in Japanese). The abstract says:

The efficiency of physical visualizations seems to stem from features that are unique to physical objects, such as their ability to be touched and their perfect visual realism.

The conclusion is that the efficiency of physical visualizations seems to come from properties unique to physical objects, such as being touchable and looking perfectly real. The subject was 3D bar charts, not line charts. Even so, being touchable does not reach the video’s viewers; it belongs only to Pinker on screen. Visual realism survives the trip through the video. This video can be read as handing only the latter to its viewers.

In May 2018, Jeremy Lent criticized how the graphs in Enlightenment Now were chosen. While granting that the book’s 75 graphs do show progress, he calls it a “selective view of progress”, that is, a view that looks only at the progress that suits the argument. The way the video adds countries one line at a time can also be read as connecting to the point that the impression changes depending on which countries are chosen.

Takeaways for Use

What this video offers to take away is a way of dividing what is made physical from what is left to the video. Only the shape of the lines, which appear to be physical, was placed in the room; all the text needed for reading was overlaid in editing. The lines can be grabbed by hand, and the text stays legible even when a person stands in front of it.

When filming a physical graph, make physical the things the hand touches in the explanation and the things that get added or removed. Overlay what should be readable from any angle, such as numbers and names. Ending with a wide shot without text lets viewers check how much was physical.

The action of adding a series is also usable. Bringing in one line by hand, as with the Ethiopia line, leaves the moment a new comparison joins as a visible movement. It can be read as marking a break in the story more clearly than drawing in an extra line on screen.

What this video lacks is also a point to take away. The source of the data appears nowhere on screen, and A’s horizontal axis was not proportional to the length of years. Physical objects catch the eye, and for that reason the construction of the axes and the source are easy to overlook. Precisely when showing something physically, the source and the construction of the axes need to be written in the overlaid text or in the description.

References

Primary sources

dataphys.org, “Steven Pinker explains Global Life Expectancy with Physical Line Charts”: https://dataphys.org/list/steven-pinker-explains-global-life-expectancy-with-physical-line-charts/
Bill Gates, “Humanity is fighting back against the Grim Reaper”, YouTube, February 13, 2018: https://www.youtube.com/watch?v=zXJPlwaYKzo
YouTube oEmbed (title and publisher): https://www.youtube.com/oembed?url=https://www.youtube.com/watch?v=zXJPlwaYKzo&format=json
Bill Gates’s post on X, February 24, 2018: https://twitter.com/BillGates/status/967449421238579201
X video streaming playlist (video file and English subtitles): https://video.twimg.com/amplify_video/963248646262284289/pl/2kD1mFd_Szq2Ipmo.m3u8
Link in the YouTube description: http://b-gat.es/2Eac9LN (redirects to https://www.gatesnotes.com/2018-Annual-Letter ; body not checked)
Steven Pinker, Enlightenment Now: The Case for Reason, Science, Humanism, and Progress, Viking, 2018. Introduction on the official site: https://stevenpinker.com/publications/enlightenment-now-case-reason-science-humanism-and-progress
Series used for comparing the data (current versions)
- Our World in Data, “Life expectancy”: https://ourworldindata.org/grapher/life-expectancy (metadata: https://ourworldindata.org/grapher/life-expectancy.metadata.json )
- Our World in Data, “Life expectancy at different ages”: https://ourworldindata.org/grapher/life-expectancy-at-different-ages
- Our World in Data, “Child mortality”: https://ourworldindata.org/grapher/child-mortality (metadata: https://ourworldindata.org/grapher/child-mortality.metadata.json )
- Max Roser, “Twice as long — life expectancy around the world”, Our World in Data, 2018: https://ourworldindata.org/life-expectancy-globally
- James C. Riley, “Estimates of Regional and Global Life Expectancy, 1800–2001,” 2005. doi:10.1111/j.1728-4457.2005.00083.x

Secondary sources

Yvonne Jansen, Pierre Dragicevic, Jean-Daniel Fekete, “Evaluating the Efficiency of Physical Visualizations,” CHI 2013. doi:10.1145/2470654.2481359 / HAL: https://inria.hal.science/hal-00781831
Jeremy Lent, “Steven Pinker’s Ideas About Progress Are Fatally Flawed. These Eight Graphs Show Why.”, Patterns of Meaning, May 17, 2018: https://patternsofmeaning.com/2018/05/17/steven-pinkers-ideas-about-progress-are-fatally-flawed-these-eight-graphs-show-why/
Douglas Fischer’s introduction to Lent’s article, EHN, May 18, 2018: https://www.ehn.org/steven-pinkers-ideas-about-progress-are-fatally-flawed-these-eight-graphs-show-why-resilience
M. Özgür Nevres, “People today are living longer, healthier, and happier lives than ever before”, March 13, 2018: https://ourplnt.com/people-living-longer-healthier-happier

Drawing the Same Flow in Three Forms, Then Laying It Out in Beach Trash: Perpetual Plastic

Sun, 01 Nov 2026 00:00:00 +0900

Perpetual Plastic depicts the flow in which all the plastic ever made divides into four destinations: discarded, in use, incinerated and recycled. The creators show the same flow in three forms: the paper’s flow diagram, a Sankey diagram and a semicircle. For the beach version, they collected trash by color rather than collecting all of it, and arranged it in bands. The author reads the color specification of the bands as being, at the same time, the instruction for which trash to pick up.

The work is a 14 m-diameter installation placed on a beach on Bali, Indonesia, in 2019. It was made by three people: Liina Klauss, Skye Morét and Moritz Stefaner. Below, I read it in the order of the figures, following the official site perpetual-plastic.net, the Truth & Beauty page by Stefaner, and the paper that is the source of the data.

It Starts from the Paper’s Flow Diagram

The starting point is a 2017 paper by Geyer, Jambeck and Law in Science Advances. It estimates the fate of the plastic produced between 1950 and 2015. Its abstract reads as follows.

We estimate that 8300 million metric tons (Mt) as of virgin plastics have been produced to date. As of 2015, approximately 6300 Mt of plastic waste had been generated, around 9% of which had been recycled, 12% was incinerated, and 79% was accumulated in landfills or the natural environment.

In short: the virgin plastic produced to date is 8,300 Mt (million metric tons). The waste generated by 2015 was about 6,300 Mt, of which 9% was recycled, 12% incinerated, and 79% accumulated in landfills or the natural environment.

Fig. 2 of the paper draws these destinations as a flow diagram with arrows and illustrations of buildings. It was drawn by the paper’s authors, not by the creative team. The values are absolute quantities in million metric tons, and the text gives plastic in use as 2,500 Mt (30% of total production).

The four percentages shown in the work can be reproduced by calculation from the paper’s figures. This is not the creators’ explanation but something the author checked. Split the 6,300 Mt of waste into 9%, 12% and 79%, and take the 2,000 Mt left after subtracting 6,300 Mt from the total production of 8,300 Mt as “in use”. Then divide everything by 8,300 Mt.

Destination	Calculation	Amount (Mt)	Share of 8,300 Mt
Recycled	6,300 × 0.09	567	6.8%
Incinerated	6,300 × 0.12	756	9.1%
Discarded (landfill / environment)	6,300 × 0.79	4,977	60.0%
In use	8,300 − 6,300	2,000	24.1%

All four match the work’s figures to one decimal place, and they sum to 8,300 Mt. The match is with the work’s figures, not with the paper’s figure or with the text’s “2,500 Mt (30%) in use”. The paper’s 8,300 Mt includes additives and there is secondary use of recycled material, so the items cannot simply be added and subtracted, which is probably the reason, but the creators give no explanation. It can be read that the work’s four items have been recombined so that they total 100%.

Redrawn as a Sankey Diagram, Bent into a Semicircle

In the “Visualization design” section of the official site, three figures appear in order: the paper’s diagram, “Redesign as Sankey diagram (inspired by Our World in Data)”, and “Semi-circle version”.

Three stages from the left: the paper’s flow diagram, a Sankey diagram flowing from left to right, and the same flow bent into bands of concentric semicircles

Source: Truth & Beauty (Moritz Stefaner), “Perpetual Plastic”. The left figure is based on Geyer, Jambeck & Law (2017) Fig. 2 (CC BY-NC 4.0)

The second is a Sankey diagram. The illustrations of buildings and the arrows are gone, and quantity is expressed only by the thickness of the bands. The Truth & Beauty page states where the idea came from.

The data platform “Our World in Data” created a reworked Sankey diagram, which served as inspiration for our semi-radial redesign.

The explanation is that a Sankey diagram reworked by Our World in Data served as the inspiration for the semi-radial redesign. The wording is “inspiration”, and the intermediate Sankey diagram was redrawn by the team. The author has not been able to check either Our World in Data’s diagram of that time or the figures in this Sankey diagram.

The third is the same flow bent into a semicircle. The thickest band, discarded (gray), is on the outside, the in-use band (blue) is on the inside, and thin green and red bands run between them. The Sankey diagram that stretched long from left to right has become an arc with a center. It can be read as moving closer to a form that people can lay out on a beach and look down on from above.

What the three share is the structure of the flow: a total divides into four destinations. The sources do not allow us to say that even the numbers are the same.

Choosing by Color to Collect the Trash

On screen, the semicircle was painted in a gradient of gray, blue, green and red. In the beach version, the discarded band changes from gray to white. The official site describes how the trash was collected like this.

instead of gathering all waste, the collection is focused on color, shifting perspective away from negative associations of waste to creative action.

Rather than picking up all the trash, they collect it by color. The explanation is that this shifts the perspective from the negative associations of waste to creative action. The discarded band turning white can be read as matching the colors of the trash that can be collected on the beach. Fifty volunteers collected trash from the west coast of Bali in three beach cleanups named “Make Art Not Waste”.

The sketch made during production has the items to pick up written in by color, under the heading “connecting to STORY”. WHITES are white straws, toothbrushes, toothpaste tubes and styrofoam boxes. BLUES are the single-use water cups “Aqua” handed out at ceremonies, and flip-flops. The flip-flops carry the note “3-4 months use. used by everyone.”

A watercolor draft of the semicircular work, with notes listing the items to be used for the white and blue bands

Source: Perpetual Plastic team, from the Perpetual Plastic official site

In the sketch, items are listed under the names of colors. It can be read that the legend’s colors function as a list of things to look for on the beach. The collected trash was sorted into baskets by color.

A basket with a paper reading “GREEN middle” attached, holding green flip-flops and plastic bottles

Source: Perpetual Plastic team, from the Perpetual Plastic official site

Based on the creators’ explanation (an article Klauss posted to designboom) and the annotated photographs, the correspondence between data and material is summarized as follows.

Data	Material / appearance	Value
Total production since 1950	A band of mixed, dark-colored trash	8.3bn tons
Discarded	White band (straws, styrofoam, etc.)	60.0%
Recycled	Green band	6.8%
Incinerated	Red band	9.1%
In use	Blue band (flip-flops, water cups, etc.)	24.1%
Amount for each destination	Band width (designboom explains it as “proportional to the numbers”)	The shares above
Structure of the branching	A semicircle in which the band of the total arcs and divides into four	—
Source of the material	Trash picked up near the same beach	—

The graph-paper sketch retains another calculation. What appear to be band widths read “3.8m white / .43m green / 0.6 red / 2.1 blue”. The semicircle’s area “66.5 m²” has the inside subtracted to give “47 m²”, next to which are “10.5 baskets” and then “At least 22 baskets”. Because it is handwritten, the reading is uncertain.

A semicircular draft on graph paper, with the width of each band, an area calculation and a conversion to a number of baskets written in

Source: Perpetual Plastic team, from the Perpetual Plastic official site

It can be read that they converted area into a number of baskets to estimate the amount of trash to collect. Note that the ratio of the written widths is close to the displayed shares but does not match them, and the exact method of proportion cannot be confirmed.

Laying It Out on the Beach, Racing the Waves

The finished work has a diameter of 14 m and consists of 4,760 pieces of plastic trash. Laying it out took more than 12 hours. The official site tells the story up to the moment the aerial photograph was taken.

Half into being finished, another wave washed over our ordered chaos. Again! … After the 4th attempt we had finally finished. … When a minute later our drone photographer came running towards us: I got it! I got the shot!

Halfway through, another wave destroyed it. On the fourth attempt it was finished, and a minute later the drone photographer came running to say they had got the shot.

An aerial view of the semicircular work on a Bali beach. Shares are written on the white, green, red and blue bands, and the total production of 8.3bn tons on the dark band

Source: Liina Klauss, Skye Morét, Moritz Stefaner, from the Perpetual Plastic official site

The left side of the photograph is the sea, with waves coming in. The work stayed on the beach for only 36 hours and was dismantled within 6 hours (designboom).

The official site also says where the dismantled materials went. The flip-flops were to be used in the next work or recycled in partnership with Indosole. Other recyclable materials were handed to EcoBali. What could not be recycled, such as fishing nets and toothbrushes, went to landfill. It can be read that the work’s own material also split into destinations: continued use, recycling and landfill.

The work won in the Ocean Plastic Innovation Challenge by National Geographic and Sky Ocean Ventures. The official site writes “First prize”, and Truth & Beauty writes “won Gold”. The category and rank come from the creators’ side, and the author has not confirmed them against the organizers’ records.

How Third Parties Saw the Work

The researcher Dietmar Offenhuber takes up this work at the start of his paper on the physicality of data, “What we talk about when we talk about data physicality”. He first writes that applying a measure of efficiency leads to a misreading.

From the conventional logic of visualization, it is not a very efficient display: it uses a lot of material and space to express a handful of numbers, the ratios of the waste stream. But applying such a metric would lead to a complete misreading of the project

By the conventional logic of visualization, it is an inefficient display: it uses a lot of material and space to express a handful of numbers. But applying that measure would lead to a complete misreading of the work, he points out.

The physicalization does not just represent data; it is data.

The physicalization does not merely represent data; it is data. Offenhuber also writes that the Sankey diagram on the beach works mainly as an icon. His reading is that the five numbers are a container, and the content is the specimens of material and the relations generated by the act of making. As examples he cites the degree of wear on the flip-flops and the number of sunscreen containers.

Jenn Schilling, in her review of Making with Data, names this work her favorite from the environment chapter and quotes the creators’ reflection (p.330).

We deliberately chose to create a “data visceralization” from physical, known objects—not only because they were colorful and relatable, but also because they allowed us to encode data variables into a tangible, situated form, at human scale

They chose familiar objects not only because they were colorful and relatable, but because it let them encode data variables into a form that is tangible, situated and at human scale. Schilling comments that a physical representation is larger than a digital one and takes a form that can be touched.

Lisa Wright of Dornob writes that what troubles her most is the black band representing the total since 1950. Note that the designboom article, which describes the encoding most concretely, is a post by the creators.

Takeaways

What we can take away from this work is to keep the order in which the form was changed. The paper’s diagram, the Sankey diagram and the semicircle line up the structure of the same flow in three forms. Shown side by side, others can trace where and what was decided.

Next is to recalculate the shares from the original data. By the author’s calculation, the work’s four numbers can be reproduced from the paper’s abstract, and they also turn out to differ from the paper’s “30% in use”. A physical work is hard to fix once laid out, so it is worth confirming the source of the numbers beforehand.

Third is the relationship between color and material. The creators collected trash by color, and the discarded band, gray on screen, became white on the beach. The color was decided by the materials available, and the legend’s color became the instruction “pick up white straws and styrofoam”. It can be read that in physicalization, a color specification can also serve as a specification for the people who make it.

Finally, the fewer the numbers in a figure, the more room the material has to speak. The official site writes that on Bali, sealed water cups are handed out free and hundreds are left in the environment at a single event. Such circumstances do not appear in the width of a band. There is also a design that minimizes what is said with numbers and leaves the rest to the material and the place.

References

Primary sources

Liina Klauss, Skye Morét, Moritz Stefaner, Perpetual Plastic, 2019. Official site: https://perpetual-plastic.net/
Truth & Beauty, “Perpetual Plastic”: https://truth-and-beauty.net/projects/perpetual-plastic (Markdown version: https://truth-and-beauty.net/projects/perpetual-plastic.md )
Truth & Beauty award article (National Geographic Ocean Plastic Innovation Challenge): https://truth-and-beauty.net/appearances/awards/national-geographic-ocean-plastic-innovation-challenge.md
National Geographic, Ocean Plastic Innovation Challenge winners page (body text not checked): https://www.nationalgeographic.org/funding-opportunities/innovation-challenges/plastic/winners/#datawinner
dataphys.org, “Perpetual Plastic” (registered by Petra Isenberg): https://dataphys.org/list/perpetual-plastic/
designboom (an article posted by Liina Klauss, 2021-08-09): https://www.designboom.com/art/perpetual-plastic-data-sculpture-4760-pieces-of-trash-08-09-2021/
Geyer, R., Jambeck, J. R., Law, K. L. (2017). Production, use, and fate of all plastics ever made. Science Advances 3(7): e1700782. doi:10.1126/sciadv.1700782 / https://advances.sciencemag.org/content/3/7/e1700782 / https://pmc.ncbi.nlm.nih.gov/articles/PMC5517107/
Our World in Data, “Plastic Pollution” (linked by the creators; the Sankey diagram of that time not checked): https://ourworldindata.org/plastic-pollution#global-plastic-production-to-fate
Moritz Stefaner, “Perpetual Plastic” (YouTube): https://www.youtube.com/watch?v=ua9fvPZB68M

Secondary sources

Dietmar Offenhuber, “What we talk about when we talk about data physicality,” arXiv:2006.04631, 2020 (for IEEE Computer Graphics & Applications): https://arxiv.org/abs/2006.04631
Jenn Schilling, “Review: Making With Data…”, Nightingale, 2023-03-29: https://nightingaledvs.com/book-review-making-with-data/
Lisa Wright, “Bali’s ‘Perpetual Plastic’ Sculpture is (Literally) Trash,” Dornob, 2021-08-27: https://dornob.com/balis-perpetual-plastic-sculpture-is-literally-trash/

A 204 m Blue Band and an 8 cm Red Rectangle: The Walkable Bar Chart on Prospect Street

Sun, 01 Nov 2026 00:00:00 +0900

On May 1, 2021, two bars were painted on Prospect Street in New Haven, Connecticut. One was a blue band 670 feet (about 204 m) long, representing Yale University’s $31.2 billion endowment. The other was a red rectangle 3 1/4 inches (about 8 cm) long, representing Yale’s $13 million annual voluntary contribution to the city. Calculating from the reported dimensions, the two bars are drawn at the same scale, and the difference between them is about 2,400 times. This work can be read as taking that difference out of the realm of something compared by eye on paper and turning it into a distance you walk along a street to check.

There are three sources: two local news reports (the New Haven Independent on the day, the Arts Paper two days later) and one entry in a catalog of data physicalization. The reports are Sophie Sonnenfeld’s article in the New Haven Independent (NHI) (May 1, 2021) and Lucy Gellman’s article in the New Haven Arts Paper (May 3). The catalog is “List of Physical Visualizations” by researchers Pierre Dragicevic and Yvonne Jansen, which is where the name “Walkable Bar Chart” came from. Only the Arts Paper gives the dimensions, and the ratios are my own calculation. No report says that participants actually walked along the band to compare the two bars.

May 1, 2021: The street in front of the president’s office

It was a Saturday. In front of the office of Yale President Peter Salovey on Prospect Street, about 150 people gathered, including union members, students, university staff, and residents. The organizers were a group called New Haven Rising and Unite Here Local 34, the union of Yale’s clerical and technical employees.

The words painted on the street were “YALE: RESPECT NEW HAVEN.” According to the Arts Paper, participants painted with brushes and rollers in blue and teal paint, music played, and children joined in.

Participants painting the white letters “RESPECT NEW HAVEN” on a teal ground on the road in front of the president’s office

Source: Sophie Sonnenfeld / New Haven Independent

There were three demands: increase the voluntary contribution to the city, hire local residents, and sign a fair labor contract with Local 34 (NHI). The day was positioned as a lead-up to a rally on May 5. Manuel Camacho, 16, who leads the Ice The Beef Latino Caucus, and others delivered a petition addressed to the president that day.

The Arts Paper also gives the background figures. Yale University and Yale New Haven Health own more than 40% of the city’s tax-exempt property. By Local 34’s estimate, the tax exemption is worth more than $157 million a year. The city projected a $66 million deficit, while Yale reported a $203 million surplus.

Participants with rollers and brushes filling in the area around the letters on the road

Source: Lucy Gellman / New Haven Arts Paper

Measuring the blue band and the red rectangle with the same ruler

The Arts Paper is the only report that links dimensions to amounts.

“Along one side of Prospect Street, a three and one-quarter inch red rectangle depicted Yale’s current $13 million voluntary contribution to the city. Beside it, a 670-foot contiguous blue stripe represented the university’s $31.2 billion endowment. By 10 a.m., it stretched all the way to Trumbull Street like a long, lolling blue tongue.”

Summary: Along one side of the street, a 3 1/4-inch red rectangle represented the $13 million annual voluntary contribution to the city. Next to it, an unbroken 670-foot blue stripe represented the $31.2 billion endowment, and by 10 a.m. it reached Trumbull Street.

The table below summarizes how the data maps to the visual form.

Data	Value	Visual form / material	Dimension
Yale University’s endowment (balance at a point in time)	$31.2 billion (Arts Paper, the university’s official figure)	Length of the blue band painted on the road	670 feet (about 204 m)
Yale’s voluntary contribution to the city (annual amount)	$13 million	Length of the red rectangle painted on the road	3 1/4 inches (about 8 cm)
Distinguishing the two quantities	—	Color (blue and red) and labels on a white ground	—
Common baseline	—	The two bars placed side by side along one side of the street	—

The endowment figure varies slightly between reports. The Arts Paper says $31.2 billion, NHI says $30 billion, and the tweet discussed below says $32 billion. According to the university’s announcement, the endowment was $31.2 billion as of June 30, 2020.

From here on, the calculations are mine. Dividing $31.2 billion by 670 feet gives about $46.6 million per foot. Drawing $13 million at this scale would give about 3.35 inches, which differs from the reported 3.25 inches by about 3%. Taking the endowment as $30 billion gives about 3.48 inches, and $32 billion gives about 3.27 inches, so every value falls between 3 and 3.5 inches. It is reasonable to conclude that the two bars were drawn at the same scale.

The ratio of the amounts is $31.2 billion ÷ $13 million, or about 2,400. The ratio of the lengths is 670 feet (8,040 inches) ÷ 3.25 inches, or about 2,470. The two ratios nearly match.

Photo showing a short red rectangle with the label “YALE’S CONTRIBUTION” in the foreground, and next to it a blue band with the label “YALE’S ENDOWMENT” extending down the street

Source: photo from Davarian L. Baldwin’s tweet (photographer unknown) / List of Physical Visualizations (dataphys.org)

In the photo, the labels “YALE’S CONTRIBUTION” and “YALE’S ENDOWMENT” sit side by side at the starting point of the two bars, and only the blue band continues down the street.

204 m cannot be seen from one spot

In a bar chart drawn to scale, the smaller quantity becomes hard to see. By my calculation, if you draw the same ratio on A4 paper and make the blue bar the full 29.7 cm of the long edge, the red bar is about 0.12 mm, which is almost indistinguishable to the naked eye. NHI describes the red on the street as “a tiny red stripe.”

On the street, this relationship works in the opposite direction. The red rectangle is about 8 cm, a size you can see when you stand at it. In exchange, the blue band becomes about 204 m.

About 204 m is not a length from which you can stand in one spot and compare the two ends. At a walking pace of 80 m per minute, it takes about two and a half minutes to walk from one end to the other (my calculation). The name dataphys.org gave it was “Walkable Bar Chart.” The catalog’s description, however, is just this one sentence:

“A bar chart conveying two quantities, one of which is clearly larger.”

Summary: A bar chart conveying two quantities, one of which is clearly larger.

The “Walkable” in the name can be read as referring to this length. On paper, the difference between the two bars can be compared with a single movement of the eye. On the street, you have to carry your body from beside the red rectangle to the tip of the blue band to check the difference. It can be read as the difference that you compare by sight being replaced by a distance you check on foot.

I want to separate claims carefully here. No report says that participants walked along the band and compared its two ends. What the reports do describe is that the act of painting itself progressed along the street. Anita Dunn and Kuli Baro were painting the section of the band next to the engineering building. Scott Marks, head of New Haven Rising, pointed down the whole street as he explained it to a film crew.

Two people crouching next to the label “ENDOWMENT” at the start of the blue band, painting with brushes in blue paint

Source: Lucy Gellman / New Haven Arts Paper

According to the Arts Paper caption, the photo shows the starting point of the band. From here it continues for 204 m.

There is one more thing to watch for: the kinds of quantities being compared. The endowment is a balance at a point in time, while the contribution is a payment made each year. What the organizers took issue with was the tax exemption. The endowment seems to have been placed there as a gauge of the university’s capacity to pay. This point should be kept in mind separately from the size of the difference.

The placement determines who the graph is addressed to

This graph was painted on a public road in front of the president’s office. NHI also captions a photo as the painting in front of the president’s office. It can be read that the president was imagined as the graph’s first reader.

Camacho states the role of the painting plainly.

“This petition represents a deep consensus from New Haven. If leaders at Yale do not get the message from this petition, they will now have this street painting to remind them that our community demands respect.”

Summary: The petition represents a broad consensus in New Haven. If Yale’s leaders do not get the message from the petition, the street painting will now remind them that the community demands respect.

Of the blue band, Marks said:

“I would want him to look at this line and feel shame,”

Summary: He wants the president to look at this line and feel shame.

The main speaker at the rally was Professor Davarian Baldwin of Trinity College. The next day, Baldwin tweeted with a photo (as quoted by dataphys.org):

“The activists and artists at the Respect New Haven rally yesterday offered this stunning graphic to visualize Yale’s $32 Billion endowment compared to its paltry $13 million contribution to the city of New Haven…a FRACTION of the taxes it would pay if properly assessed.”

Summary: At yesterday’s rally, activists and artists created a stunning graphic comparing Yale’s $32 billion endowment with its paltry $13 million contribution to New Haven, which is only a fraction of the taxes it would pay if properly assessed.

For people who are not on the street, the graph reaches them as a photograph. In the photo, only the near end of the band is visible, and the whole does not fit in one frame.

On the university’s side, spokesperson Karen Peart sent NHI a statement. It described the university’s cooperation with the unions since 2004 and its employment measures during the pandemic. It ended with a line about exploring “creative solutions” ahead of the contract renewal in 2022.

The statement mentions neither the painting nor the comparison between the endowment and the contribution. It can be read that the university did not rebut the bar chart’s comparison with numbers, and instead responded by shifting the topic to labor relations.

Six months later, on November 17, 2021, Yale and the city agreed to increase the voluntary contribution. On top of the $13 million a year, Yale will add $10 million a year for five years and $2 million in the sixth year. The total over the six years is about $135.4 million (Yale News). However, I could not find any source showing that the painting led to the agreement. All that can be said is the order in time.

How did third parties see this painting?

Researchers, art reporters, and readers of political news each look at different things.

First, the researchers’ classification. It was registered in Dragicevic and Jansen’s catalog on May 3, 2021, the day after the tweet. It was recommended by Benjamin Bach, and its category is “Passive physical visualization.”

The walkable tag also includes “Walkable Age Pyramid” (2013), “Walkable Collaboration Network” (2016), and “Planetenweg Uetliberg: The Solar System as a Hiking Path” (1979), among others. It can be read that the catalog classified it not as a protest slogan but as a member of the same family of “read by walking” works as the one that turned the solar system into a mountain trail.

Next, local art reporting. Gellman of the Arts Paper treated it as a mural. The headline is “Mural With A Message Lands On Prospect Street.” Most of the article is about a staff member painting while dancing with her niece and a child painting with his grandfather and father. The emphasis is on the graph having been made together.

Finally, local political reporting and its readers. NHI’s article reported the day within the flow of labor negotiations and the city’s finances, and also carried the university’s statement. In the comments section, criticism of using a public road for a political message stands out. One comment (EngagedCitizen, 2021-05-02) says they are angry that the city allowed it.

“Is New Haven going to let every group graffiti the streets with their own political messaging?” (baker, 2021-05-02)

Summary: Will the city let every group graffiti the streets with its own political messaging?

On the other hand, some comments criticize Yale’s tax breaks and support the painting. Both are readers’ opinions, not the city’s official position. I have also been unable to confirm any official record showing that the city permitted it.

Takeaways

There are four things from this graph that you can take back into visualization work.

Keep the scale, or favor legibility. If you draw a 2,400-fold difference on paper, the smaller bar comes to 0.12 mm. If you compensate with a logarithmic axis or an enlarged inset, the size of the difference becomes harder to convey. This graph can be read as having first fixed the size at which the smaller bar is visible (about 8 cm) and then extended the larger bar out of the field of view. It is one way to solve the problem of making both visible without bending the scale.
Align the starting points and place labels. When a long bar runs out of view, the cue for comparing is the shared starting point. As the photo shows, labels were placed side by side at the starting point of the two bars.
State what kinds of quantities are being compared. When you place quantities of different kinds side by side, such as a balance and an annual amount, only the size of the difference travels on its own. The caption or explanation should record what was compared with what.
Choose the placement as the addressee. By painting in front of the president’s office, they settled who the graph’s first reader would be. Where you place a graph is a choice about whom you want to read it.

Pay attention to how it is recorded as well. The photos show only the near end of the band, and the whole does not fit in one frame. The reason this work’s scale can be checked afterward is that the Arts Paper wrote down the dimensions and amounts in its text.

References

Primary sources

Sophie Sonnenfeld, “Labor Paints Prospect Street With “Respect” Message,” New Haven Independent, 2021-05-01 (text, Yale’s statement, photos). https://www.newhavenindependent.org/2021/05/01/yale_respect_new_haven_paints_/
Lucy Gellman, “Mural With A Message Lands On Prospect Street,” New Haven Arts Paper, 2021-05-03 (correspondence of dimensions and amounts, photos). https://www.newhavenarts.org/arts-paper/articles/mural-with-a-message-lands-on-prospect-street
Tweet by Davarian L. Baldwin (the day after the rally; text as quoted by dataphys.org). https://twitter.com/DavarianBaldwin/status/1388840756895526912
Yale News, “Investment return of 6.8% brings Yale endowment value to $31.2 billion,” 2020-09-24. https://news.yale.edu/2020/09/24/investment-return-68-brings-yale-endowment-value-312-billion
Yale News, “With new $140+ million Yale pledge, Yale, New Haven promote growth, economy,” 2021-11-17. https://news.yale.edu/2021/11/17/new-140-million-yale-pledge-yale-new-haven-promote-growth-economy

Secondary sources

Pierre Dragicevic & Yvonne Jansen, “2021 – Walkable Bar Chart,” List of Physical Visualizations (dataphys.org), 2021-05-03 (photo, description, classification). https://dataphys.org/list/walkable-bar-chart/
The walkable tag in the same catalog. https://dataphys.org/list/tags/walkable/
Comments section of the same New Haven Independent article (EngagedCitizen, baker, and others, 2021-05-02). https://www.newhavenindependent.org/2021/05/01/yale_respect_new_haven_paints_/