Steffen’s, in the great-paper series (Einstein, Watson, Cantor, Hubble, Benjamin): signature and counter. Ideas given, which bind the design: (1) the paper has been reinterpreted since (Einstein 1905, Perrin 1908, and a popular version of what Brown saw and thought); the game sticks to what Brown says, and the later interpretation is discussed in a coda on the final verdict screen; nothing in the game’s play uses the later theory, and the coda is after the verdict and marked as later. (2) Make it fancy, with animated SVG capturing the motions; not the dark palette of Hubble and the Olive; the engine picks the look, and it should be astonishing to look at. (2) overrides the quiet museum-label chassis the engine first wrote (section 13 records the change): the register is now a hand-coloured natural-history plate, outsized and in motion.
Robert Brown, A brief account of microscopical observations made in the months of June, July and August, 1827, on the particles contained in the pollen of plants; and on the general existence of active molecules in organic and inorganic bodies. Privately printed pamphlet, London, dated July 30th, 1828 (“Not published”); reprinted in the Edinburgh New Philosophical Journal 5 (1828): 358–371 and in the Philosophical Magazine 4 (1828): 161–173. With the supplement Additional remarks on active molecules, dated July 28th, 1829 (Edinburgh Journal of Science 1 (1829): 314–319; Philosophical Magazine 6 (1829): 161–166). Both as reprinted in The miscellaneous botanical works of Robert Brown, vol. 1, ed. J. J. Bennett (London: R. Hardwicke, 1866), pp. 463–486. No DOI. Available: the 1866 reprint, scanned (Biodiversity Heritage Library; also the New York Botanical Garden scan the user sent, 24 pages). No data, no figures, no code: the paper is prose. Every number in it is a measurement stated in words (fractions of an inch, minutes, hours, years), read from a scan of typeset text and transcribed by hand; see the config block.
Primary: an estimate, of a descriptive kind: a catalogue of observations with the conditions under which each was made. Particles from the pollen of Clarckia pulchella, oblong, 1/4000 to 1/5000 of an inch long, in motion; smaller spherical “molecules” (his word for the smallest things he could see), 1/20,000 to 1/30,000 of an inch across, in rapid oscillatory motion; the same in every plant examined; the same in plants dried twenty years and a hundred; the same in every part of the plant when bruised; the same in animal tissue, gum, coal, London soot, fossil wood, window glass, granite, a fragment of the Sphinx, lava, obsidian, pumice, meteorites, manganese, nickel, bismuth, antimony, arsenic; the same in wood, linen, paper, cotton, wool, silk, hair and muscle after burning in a candle or blowpipe; short fibrils of two or three molecules in vivid motion, four or five in motion, longer ones at rest; nothing obtained from oil, resin, wax, sulphur, from metals he could not grind, or from bodies that dissolve in water (and, in 1829, sulphur, resin and wax yielding them after all, the earlier failure put down to “defective manipulation”). The 1829 supplement hands over a second set: an open drop lasts under a minute and develops currents from centre to rim, slow at first, then rapid, that “disturb or entirely overcome the proper motion”; drops of water from 1/50 to 1/2000 of an inch shaken into almond oil last more than an hour with the motion “undiminished”; some such drops hold a single particle, which moves; the experiment inverted, minute oil droplets on the surface of a water drop lie at rest while particles inside it move.
Secondary: an argument, by elimination. In 1827: the motion belongs to the particle, not to currents or evaporation. In 1829: it is not animation (he never said so and says so plainly); not currents, not evaporation, not attraction and repulsion between particles, not capillary or hygrometric action at the surface, not air bubbles; the oil-drop experiment answers the most important of these at once; and the cause is one “for which I am unable to account”. The game’s mechanism is a toy built from the catalogue, declared as such.
A counterintuitive result, reached by identification. The motion first looked like life: particles of pollen, the part of a plant that makes new plants, swimming. Brown’s inquiry takes that belief and kills it in stages, until the restless particles have been found in a chip of the Sphinx and in burnt hair, and every physical cause anyone has proposed has been removed from the drop with the motion undiminished. What is left is a property of smallness itself: any matter, ground fine enough and suspended in water, dances, the finest most vividly; and no cause for it. The paper’s discovery is not an explanation; it is the demolition of every explanation on offer, including the author’s own first one. The game lets the player try to stop the dance, with Brown’s own means, and fail as Brown failed.
Look through a microscope at a drop of water with a speck of pollen dust in it, and the specks are dancing: trembling, turning, never still. The obvious guess is that they are alive, because pollen is the living stuff of flowers. So you try to make them stop. You take dust from a plant dried a hundred years ago: it dances. You burn hair in a candle and grind the ash: it dances. You chip a piece off the Sphinx, four thousand years old and never alive, and grind that: it dances. Maybe it is the water drying up and stirring the drop: you seal the drop in oil so nothing can dry; it dances for an hour. Maybe the specks are pulling on each other: you make a drop so small it holds one speck alone; it dances alone. The only thing that ever changes the dance is how small the specks are: big lumps lie still, fine dust dances, the finest dust dances most. Robert Brown spent two summers at this and wrote down, honestly, that he could not say why. Nobody could, for seventy-seven years.
| Game | Position | Format | Feedback device | Chassis |
|---|---|---|---|---|
| Benjamin’s Veil (benjamin) | agent (keeper of one work, a timing decision) | challenge: optimal stopping | the halo and the crowd (face in a frame, halo thinning, crowd in front) | 1930s photomontage poster (Worldly): poster cream, ink, red, halftone grey, gold leaf; Oswald + Libre Franklin; red slabs with an ink rule |
| Hubble’s Twelve Nights (hubble) | experimenter (the paper’s own instruments, a budget of nights) | discovery under a budget | the plate (Figure 1 drawn live), with a reveal | observatory at night (Instrument): sky black, starlight, arc amber, shift red, brass; Jost + IBM Plex Mono; brass-rimmed instrument plates |
| Cantor’s Bargain (cantor) | agent (the constructor of E0, in a contest) | strategic | the strike (rows struck in the Lister’s book) | fairground bill (Loud): big-top red, cream, ink, brass gold, teal; Rye + Alfa Slab One + Work Sans; ticket buttons |
What this game does differently.
Aim, on the start screen, at the top of the play screen, first line of How to play: Dust is dancing in a drop of water. Make it lie still.
Verdict, the end screen’s first lines: The dust is still dancing. Then, in words: what you tried and what it danced through; which of the five suspects you set aside and which you left standing; whether you found the one thing that changes the dance (size) or never looked; the quiet drops you got, if any, were the ones with nothing in them (sulphur, sugar) or no water (the open drop watched to the end). Then what Brown says: that he could not make it stop either, that the smallest specks of any matter, living or never alive, burnt or not, dance in water, and that he could not account for it and did not claim the specks were alive. Then, marked as later and after a rule, the coda.
A. The Sceptic’s Bench (controller; impossible challenge). One drop on the stage, a bench of Brown’s own experiments, seven days. Controls: one tap to choose the day’s experiment, one tap to look. Before pressing, the player is weighing which suspect they believe (life? drying? currents? pull between specks? the drop’s skin?) and which experiment would remove it. After, they see the drop redrawn under the new conditions, dancing or not, and a slip laid over any suspect the experiment answered. Discovery moment: the chip of the Sphinx dances (life is gone), then the sealed drop dances for an hour (drying and currents are gone), then one speck alone dances (pull is gone), and then the coarse grind: lumps still, fine dust dancing, the first and only time the dance changed. Fun: anticipation (will the Sphinx dance?), discovery (it does; then the lumps do not), mastery (the second play goes straight to the oil drop, which answers three suspects in a day, and spends the days on size). Replayability: a second play tests a different belief, or hunts the lever instead of the causes; the verdict differs.
B. Brown’s Summer (experimenter; discovery under a budget). June to August 1827, a cabinet of specimens, a budget of looks; the player chooses what to put under the lens to find out where the motion stops, and at the end states the law (the motion belongs to: pollen / living things / organic matter / everything). The discovery is the widening: plants, dead plants, animals, coal, glass, granite. Deliberation: which specimen is most informative. Fun: anticipation and discovery. Weak point: a sequence of yes, yes, yes; the deliberation is thin once the player sees the pattern, and it repeats Hubble’s seat and shape.
C. The Eye (subject; perception; real time; the reckless one). A play screen with no words: the eyepiece field, a drop evaporating in real time, currents growing from the centre outward, and the player’s one job is to tap the specks that move of themselves and not with the stream; then the same drop sealed in oil, where every speck is its own. Brown’s 1829 distinction (proper motion against currents) under the thumb. Fun: perception, mastery, a clock. Discovery: in the open drop the task gets impossible as the currents grow; in the sealed drop it is easy, because there is nothing but proper motion. Walkthrough written in section 9 for the record. What it gives up: the causes, the Sphinx, the burnt hair, the whole elimination, which is the paper’s spine.
A (the pick). 0:00 Start screen: title in green small capitals; three lines of what the hands will do (grind, drop, seal, look); the aim on its own line; BEGIN. 0:15 The stage: a round drop of pale water, big, on a cream card, with a dozen oblong ink specks turning slowly and a crowd of small round ones trembling fast. One line: this is dust from the pollen of a flower, in water, under Brown’s lens, London, June 1827. The aim at the top. Under the drop: “Dancing.” The shelf: five labels, LIFE, CURRENTS, DRYING, PULL, THE SKIN OF THE DROP, with one line: these are what people have said makes it move. 0:30 The bench: fourteen labels in two columns, DAY 1 OF 7 above. The stranger reads “A chip of the Sphinx” and grins; reads “Burn it first”, “Seal it in almond oil”, “One speck alone”, “Grind coarser”, “Sugar”. They think: it is pollen, it is alive, so stone will not dance. They tap the Sphinx; the line under the bench says what they will do; they tap LOOK. 0:50 The drop is redrawn: grey angular grit, and it dances, the fine grit fastest. The line: “Dancing. Stone, four thousand years old, never alive.” A slip lands on LIFE. The stranger frowns and laughs. 1:10 Day 2. They think: then it is the water, drying up and stirring. They tap “Watch the open drop”; LOOK. The drop sits; then a slow drift from the centre to the rim, then faster, the specks swept outward and the dance lost in the stream; then the rim comes in and the water is gone; grit lies dry on the glass. The line: “The water went. The dust lies dry. Wet it and it dances again.” No slip lands. The stranger thinks: so the currents came after. 1:50 Day 3. “Seal the drop in almond oil.” A small drop inside an amber field, dancing; the clock in the corner runs to an hour; dancing as at the first minute. Slips land on DRYING and CURRENTS. 2:30 Day 4. Pull between specks: “One speck alone.” A tiny drop, one speck, dancing by itself. Slip on PULL. 3:00 Day 5. “Oil on the water’s skin”: the experiment turned round; oil beads on the drop’s edge sit still; inside, dancing. Slip on THE SKIN. All five set aside; the stranger has nothing left to suspect and two days. 3:30 Day 6. “Grind coarser.” Big lumps, still; fine dust between them, dancing. The line: “The lumps lie still. The fine dust dances.” The first time anything changed. 4:00 Day 7. “A fibre of asbestos”: long threads at rest, short bits dancing, the shortest fastest. 4:30 Verdict: The dust is still dancing. You tried seven things; it danced through all of them; you set every suspect aside; you found the one thing that changes it: how small. Brown’s lines. Then, after a rule: what came later.
B (runner-up). 0:00 Start, aim: find out where the motion stops. 0:20 A cabinet of twenty specimens, twelve looks. The stranger taps a fresh flower’s pollen: dancing. Taps a herbarium sheet: dancing. Taps a lump of coal: dancing. 1:30 By the fourth look they know every drawer dances; the remaining looks are spent confirming, and the law at the end is a menu pick. The deliberation died at minute two.
C (reckless). 0:00 A round field of specks, no words, a thumb-sized target ring. Tapping a speck that moves of itself scores; tapping one carried by the stream does not. 0:30 At first every speck moves of itself; easy. 1:00 The drift starts; the stranger cannot tell proper motion from drift and does not know that is the point; “I don’t know what I’m supposed to feel.” 1:30 The sealed drop; easy again; the game ends. Four minutes of perception and nothing of the Sphinx, the burnt hair or the causes. A fine opening for another game; not this one’s spine.
A, The Sceptic’s Bench. The player operates Brown’s own levers (specimen, age, fire, fineness, the open drop, the oil, the single speck, the inversion) at the one target the paper is about, and fails at it the way the paper fails, with the paper’s own experiments and in roughly the paper’s own order if they think as Brown thought. The deliberation is real: which suspect do I hold, and what removes it. The commitment is real: the bench says what you will do, not what you will see. The discovery lands inside the play three times (the Sphinx; the sealed drop; the lumps), not on the end screen. The sentence afterwards has no “guess”: I ground the Sphinx and burnt hair and sealed the drop in oil, and the dust danced through all of it; only the big lumps lay still. Originality: no game in the table has an impossible target, and none has let the player work through an author’s own list of rejected explanations.
Author test. Brown would recognise the Clarckia particles and the molecules, the herbarium specimens, the blowpipe, the Sphinx, the meteorites, the almond oil, the drop of 1/2000 of an inch with one particle in it, the inverted experiment, and the fibrils of asbestos at rest beyond four or five molecules. He would enjoy watching a player sure that the Sphinx will not dance; and he would insist, as he did in 1829, that the game never say the specks are alive, which it does not, and that it end where he ended: unable to account for it. The colleague who knows the answer still has to pick an order for seven days, and still cannot stop it.
Runner-up: B. Switch if the author wants the 1827 summer alone, the widening from pollen to the Sphinx, without the 1829 causes.
What the reckless concept would have given. A wordless real-time field in which the player feels Brown’s distinction between proper motion and drift with their thumb. Kept from it: the open drop watched to the end, drawn live, with the drift growing from the centre and overwhelming the dance, so that the distinction is seen if not played. Given up: the clock, the tapping, the silence.
Seven days, one experiment a day, from a bench of fourteen; then the verdict; about four to five minutes. Screens: start; the stage (the drop, the shelf, the bench, the same screen redrawn each day); the end; How to play; About. No presets: the bench is the paper’s own list of experiments, offered all at once, in no order, so that the order is the player’s. Day 0 is the paper’s opening observation (fresh Clarckia pollen) shown before the first choice. No score: the yardstick is the drop, read by eye, and the verdict is in words. Replay starts a fresh seven days. The budget of seven is ours (section 15): it makes the order a decision; it does not make the discovery, which holds for any order.
Register: a hand-coloured natural-history plate, 1828, outsized and in motion (Austere family, the scientific plate from an old atlas; the user asked for light, animated and astonishing, so the plate is drawn big and everything on it that moved in Brown’s drop moves on the page). It comes from the paper’s own world: Brown’s pamphlet was printed in the same years as the great hand-coloured plates of Curtis’s Botanical Magazine, with engraved line and watercolour washes on cream; and from the pamphlet’s own typography, a Caslon page. The plate is the play screen: the drop is drawn as a lens of water, big, with a rim and a highlight, and the specks in it are engraved and coloured by what they are.
#F6EEDC (the ground); ink #2A2118 (engraved line, text); Prussian blue #1D4E89 with the drop’s wash #CFE4EC and rim #8DB8C9 (water, nothing else); almond gold #E8B84A (oil, nothing else); carmine #B8322A (the day band, chosen labels, the slips); bottle green #2F6B4F (the primary button, moss). Pollen particles ochre #D9A441; the Sphinx’s granite rose-grey #A98B80; soot black; arsenic white; glass pale with a blue edge; asbestos a pale grey-green.The purpose Brown started with (tracing the pollen particles through the style to the ovulum, pp. 463–466, 475–477) and his disagreement with Brongniart (477–479): About only. The families (Onagrariae, Asclepiadeae, Periploceae, Orchideae, Gramineae, Mosses, Equisetum) beyond “moss” and “a leaf”: cut. The change of form of the Clarckia particles (contraction, swelling, turning on the axis): shown as turning only. The “primary combinations” hypothesis Brown advanced in 1827 and withdrew in 1829: About. The uniformity of the molecule’s size, which he was unsure of: the game draws them uniform and the About says he was unsure. The priority discussion (Leeuwenhoek, Gray, Needham, Buffon, Gleichen, Drummond, Bywater): About. The list of substances (travertine, stalactites, lava, obsidian, pumice, volcanic ash, manganese, nickel, plumbago, bismuth, antimony, arsenic, the haddock’s mucus, the fern’s powder): three on the bench (Sphinx, glass, meteorite), the rest named in About. Air bubbles and volatile matter as a suspect: folded into the sealed drop’s result.
Kind: toy, built from the paper’s catalogue of observations (section 2). The About screen says: This game demonstrates that, under its own assumptions, nothing done to the matter, the drop or the specks’ company changes the dance, that the open drop’s streams come after the dance and end with the water, and that the one thing that changes the dance is size. It assumes the toy: a vivacity that depends on size alone, with Brown’s ordering and rest limit and our curve between; currents that grow over the open drop’s minute, his words and our curve; a bench of fourteen looks mapped to five suspects by his own sentences; seven days, ours. The paper establishes the observations as reported and the argument by elimination; it establishes no cause and claims none.
The discovery comes from the paper, not from the rules. The central discovery (every attempt fails; size alone changes the dance) follows from the catalogue, not from the budget: the check B9 runs 300 random seven-day sequences over random seeds and finds the dust dancing in every look with water and specks small enough, and nothing at rest in water outside the two size looks. The budget of seven only decides how many suspects a player happens to set aside; three days suffice for all five (B9c). The heuristics table (section 18) is a property of the bench’s mapping, which is Brown’s; the mapping is listed in the config block with page references (B8).
Robustness. B9d doubles and halves the vivacity exponent and moves the rest limit by a factor of two each way: the ordering (molecule, then the larger pollen particle, then a lump at rest) and the sealed drop’s undiminished dance survive all four.
Later interpretation. Nothing in the mechanism uses Einstein’s or Perrin’s theory; the size dependence is coded from Brown’s ordering (pp. 466–468, 472), not from the later law. The coda on the verdict screen is the only place the later interpretation appears, after a rule, headed “What came later”, and it says which two common claims about Brown are not in his pamphlet.
Headless, in Node, against the paper’s statements (page numbers of the 1866 reprint): sizes (B1); invariance of the dance across stone, glass, meteorite, dead moss, burnt ash and soot (B2); ordering by size and fibril length (B3); the open drop’s currents, its minute, the dry dust (B4); the sealed drop’s hour, no currents, undiminished dance, the single speck (B5); the inverted experiment (B6); sulphur and sugar (B7); the mapping of looks to suspects (B8); independence of the discovery from order, seed, budget and our curve (B9); the heuristics spread (H); no decimals in anything the screen reads (W). Result: calibration-result.txt, 31 passed, 0 failed.
| Week | Suspects set aside | Size lever | Quiet drops |
|---|---|---|---|
| the vitalist (moss, burn, Sphinx, glass, meteorite, soot, sugar) | 1 of 5 (life) | missed | 1 |
| the physicist (open drop, oil, one speck, skin, coarse, asbestos, Sphinx) | 5 of 5 | found | 1 |
| Brown’s own order (moss, burn, Sphinx, asbestos, oil, one speck, skin) | 5 of 5 | found | 0 |
| the cook (sugar, sulphur, open drop, glass, soot, meteorite, moss) | 1 of 5 | missed | 3 |
| the quickest (Sphinx, one speck, skin, coarse, open drop, sugar, sulphur) | 5 of 5 | found | 3 |
The decisions matter: a week spent on the matter alone answers one suspect; a week on the drop answers all five. No rule wins by the rules: the dust dances in every row, which is the sentence the player leaves with. The one simple rule the table shows (the sealed drop with one speck answers three suspects in a day) is Brown’s own 1829 experiment, and the verdict names it to a player who missed it.
mechanism.js, 22.9 kB, the only copy); calibration 31/31.loop.js, loop-log.txt): two weeks printed and read; something visible changes every day, the result is said plainly, no slip lands on the open drop and the line says why.session-test.js, session-log.txt): two weeks at 380 px (the physicist’s, with help opened on day 3 and returned to with the choice kept; then play again and the vitalist’s); no errors, no horizontal scroll, LOOK disabled until a label is tapped, no label pre-chosen, NEXT hidden until the look ends, the verdict reached twice with different text. Session length about four to five minutes is an estimate until a person has measured it.From a brief after the first build. Register and type unchanged.
?, gloss one thing does change it. A size look (Grind coarser, A fibre of asbestos) stamps it in the LOOK green, FOUND, as a double-ruled rubber stamp rather than a red slip; the card then reads Size, only fine dust dances, for the rest of the week. With six cards the shelf became a three-by-two grid so the labels keep their size at 380 px; the red slips moved down onto the gloss line so each suspect’s name stays readable under its slip.shelfState) from the suspects set aside and whether the clue is found: open (the new default line); five aside, no clue (the line says Find it.; the bench head becomes Nothing named is left. What changes the dance?); clue, five not aside (The five on the shelf are still to be answered.); both (Brown kept looking. You may, or stop here.). Any state other than open sets the line in italic ink. The day band and the reading under the plate are unchanged.Checks. node calibration.js: 44 passed, 0 failed. session-test.js, three weeks at 380 px, no errors, no horizontal scroll: week A taps the dimmed label on day 1 (gloss flashes, LOOK stays off, nothing chosen), clears the shelf on day 4, finds the clue on day 5, takes ENOUGH, and gets “In five days” under AFTER FIVE DAYS; week B finds the clue on day 1 and never clears the shelf (ENOUGH never offered); week C clears the shelf by day 4 and never finds the clue (bench head changes on day 5; the unfinished paragraph, alone, on the verdict). New screenshots: screenshot-a-shelf-cleared.png, screenshot-a-clue-found.png, screenshot-b-clue-only.png, screenshot-c-day5-bench.png, screenshot-locked-flash.png, screenshot-a-end.png, screenshot-c-end.png. The per-day shots from the first build (screenshot-a-day*-vignette/mid/result.png, screenshot-b-day*-vignette.png) are kept for the drop and the vignettes, which did not change; their shelf is the five-card one.
One change, after Steffen’s note that the lone speck should be discovered, not unlocked, and that no label on the bench should tell the player what to do first.
lockedGloss are gone; requires and locked() stay in the mechanism, and available() now excludes a locked look.reveals: 'onespeck' and a found text; readLook returns it with the reading. The screen shows it under the reading as a note with a carmine rule and the cap NEW LOOK: Among the sealed drops is one so small it holds a single speck, nothing beside it. That is a new look. The note stays with its reading while the next day’s bench is open, as the reading does, and goes when the next look starts.isNew: the look it refines was the last look taken).Checks. node calibration.js: 48 passed, 0 failed (G1 kept; G2 a fresh bench lists thirteen and never the lone speck, fourteen after the oil; G2b only the oil look announces anything; G2c the new mark is on the day after the oil only; G2d the verdict knows whether a drop was sealed). session-test.js, three weeks at 380 px, no errors, no horizontal scroll: week A checks the lone speck is absent on days 1–2, the note shows after the oil look and still while day 3’s bench is open, the label is marked new on day 3 only; week B never seals a drop, has thirteen labels on every day, and its verdict says the one-speck drop was never found; week C unchanged. Screenshots a-oil-new-look, a-day3-bench-new; locked-flash removed.
Steffen, after playing: once sugar had gone in, “the sugar remains in all the other animations as a large white fleck”. The fleck was the highlight on the water, a hard-edged white ellipse at the drop’s upper left, there since day 0; a drop with nothing in it made it the only thing in the water, and from then on it read as a lump. The highlight is now a soft radial sheen with no edge (gradient ghi, white fading to nothing), so there is nothing in the drop that could be taken for a speck. No other change. Calibration 48/48; session test three weeks, errors none.
Steffen: the clock that counts the minute does not always appear. In Chromium it always did (the clock group’s opacity is 1 for the open drop and the two sealed drops at every probe), so the fault is a phone browser’s: WebKit does not always repaint an SVG group whose opacity attribute flips from 0 to 1. The clock is now shown and hidden by display, with the opacity set alongside. No other change. Calibration 48/48; session test three weeks, errors none.
The clock still did not show on day 1 on Steffen’s MacBook, whatever he picked (Chromium showed it every time). The first time an SVG group inside the animated plate is shown, Safari may not paint it. The clock is no longer part of the SVG: it is a small HTML layer over the plate’s top right corner, toggled by a class, hidden on day 0 and during the vignette, shown for the open drop and the two sealed drops. Same drawing, same place. Calibration 48/48; session test three weeks, errors none.