Paris, 1848. Reconstructed from the record.
Sort mirror-faced sodium ammonium tartrate crystals into two piles, dissolve them, and watch their solutions turn polarized light oppositely. Mix equal amounts and the turn vanishes, revealing what racemic acid concealed.
The objectA tray, two piles, and a polarimeter. The crystal face makes a prediction. The dissolved crystal answers it.
Layer one: the blind sort
Drag a crystal to the pile on the same side as its small dark face. On touch or keyboard, select a crystal and then choose a pile. Dissolve either pile when it holds at least one crystal.
Sodium ammonium paratartrate tray
1Choose by shape alone. The sign badges are hidden.
2Drop the crystal where its small face sits.
Select one crystal. Its optical sign is still hidden.
| pile prediction | sourced sign and token ee | blind result |
|---|---|---|
| No pile dissolved yet. | ||
The solutions
The two ideal stock solutions are kept separate from the face classifier. Their signs come from Pasteur's report. Concentration, mix, path scale, and analyzer angle determine the result live.
α = 18.0° × (0.400 - 0.400) = 0.00°
I / I0 = cos²(0.0° - 0.00°) = 1.0000
The empty and equal settings. Zero solute gives no rotation because there is nothing in the tube. A forced 50:50 mix of equal stocks gives zero by algebra. Neither setting is an independent confirmation; the blind face-to-sign comparison above is.
Layer two: the phase boundary
Below the reported transition, the racemic material forms two homochiral tetrahydrate phases. Above it, one racemic monohydrate phase contains both hands. Temperature does not flip a molecule. It changes how the solid packs.
24.0 °C is below the 27.8 °C review value: two sortable tetrahydrate phases are shown.
one right-handed phase
one left-handed phase
P21212 contains only handedness-preserving operations. Each whole crystal belongs to one population.
Choose a molecule. Inversion sends (x, y, z) to (1-x, 1-y, 1-z) and reaches the opposite hand inside the same crystal.
This is an empirical equilibrium map, not a growth simulation. The primary account says the transition is about 27 °C; a later review gives 27.8 °C. At the cited boundary the page refuses to choose a phase. Hydration, kinetics, and metastability can move what a real crop does.
The check
The geometry table and the optical-sign table are separate inputs. The page cannot make a successful blind sort by drawing the face and sign from one rule.
| anchor | record | what this page checks |
|---|---|---|
| 1848 | Pasteur's memoir | Careful separation, right-faced and left-faced piles with opposite solution rotations, and inactivity for the full crop. |
| 25:29 | Pasteur's footnote | The two raw rotatory-power magnitudes were not exactly equal. Pasteur attributed the mismatch to incomplete separation. |
| about 27 °C | van 't Hoff, 1896 | The primary account's approximate transition wording and the tetrahydrate to monohydrate change. |
| 27.8 °C | Mitchell, 1998 | The later review's value, used as the exact position of the page's boundary control, not claimed as a universal exact threshold. |
| 4 versus 1 water | Kuroda and Mason, 1981 | The formulas parse to four waters for the dextrorotatory tetrahydrate and one for the racemic monohydrate. |
| P21212 / P21/a | Kuroda and Mason, 1981 | The first operation set has no inversion. The second contains inversion, so the page's schematic partner coordinates must sum to (1, 1, 1). |
Live baseline: c_right = 0.400, c_left = 0.400, α = 0.00°, ee = 0.0%, I/I0 = 1.0000. Offline: all 4,096 two-pile assignments and 101 integer mix settings are enumerated.
The story does not get to simplify the record
Pasteur's publication, early biographies, and laboratory notebooks give differing accounts, as H. D. Flack's 2009 review documents. A 2021 critique argues that the heroic eyesight story is overstated and that Auguste Laurent's crystallographic preparation and Jean-Baptiste Biot's optical methods are often minimized. Joseph Gal's 2019 review disputes Gerald Geison's negative reading. None is treated here as the last word.
What survives the disagreement is the operable sequence: classify the whole crystals by a mirror-related face, dissolve the separated populations, and test the solutions independently. Pasteur inferred molecular dissymmetry. The tetrahedral carbon models of Le Bel and van 't Hoff came later.