The calculations

Every rule and every number behind the Hebble lattice results, first in D (as the lab runs it: whole numbers that only move, no zero, no minus, no π), then translated into the old maths, so anyone can check it either way. The code is open: every run can be repeated.

D numerals count in thirteens with digits 1–9, A–D and no zero (how to read them). In the tables, D is what the run prints and old is the same number written the old way.

1. Mass from just light

Code: lang/lab/runs/src/lightmass-src.delta. Run: delta run lang/lab/runs/lightmass.delta 3 79 1B.

The rules, in D

The same rules in the old maths

What was measured, and what the old maths expects

quantityD (measured)old (measured)old maths predicts
the spring’s swing4B steps (counted)63 steps62.8 steps
bright points round the middle one6 at D4–DD tenths of a cell6 at 17.3–18.2 cells6 at 18.1 cells
the next ring out6 at 1AB tenths6 at 31.0 cells6 at 31.4 (√3 × 18.1)
holes spinning one way / the other / not at all23 / 23 / 329 / 29 / 3equal numbers each way
light from two opposite sides (control)stripes: few bright points (B, old 11), most holes with no spin (17 of 2B, old 20 of 37)

What this shows, plainly. D’s counted run and the old maths agree: same swing, same spacing, same balance of spins. The pattern itself (six round one, holes spinning both ways) is what crossing waves do, in either maths. The step from bright light to mass is the one rule put in; that is the step still to be shown.

2. The raft

Code: lang/lab/runs/src/pingfunnel-src.delta. Each step is a quarter of a millisecond; forces in microNewtons, places in nanometres, all kept above a still level so nothing goes below anything.

ruleDold maths
a ball’s pull down the cone (2.7 g on a 2° slope)BD96 × rise of 2° = 562 μNm g sinθ = 0.0027 × 9.81 × 0.0349 = 925 μN
rolling on the tray3 thousandths of its weight on itrolling resistance 0.003 m g cosθ
pushed apart where they overlap(a ball across, less how far apart) × Alinear spring F = k·overlap
the flow’s pull between two balls on the holed platesuction × 64 × the gap still open, ÷ 3F = Δp × open area
gripup to 1A1 thousandths of how hard they pressCoulomb friction, μ = 0.3
warmthmoves toward the air by (air − ball) × h ÷ heat heldNewton cooling, dT/dt = hA(Tair − T)/(mc)
MAGIC: how fast a warm wall flattenssoftness × pull ÷ 1000 × thinness each second(chosen, not measured; the blind test checks it)

Warmth window: D 81 to 98 C close the raft, 9D C melts it; old: 105 to 125 C, 130 C melts.

Hole size and Henry power (seconds until every wall is flat; the Henry’s pull in pascals):

holespower 3power 6power 9power 11
0.3 mm233 s (D 14C), 5,088 Pa (2415)134 s (A4), 9,889 Pa (4669)100 s (79), 14,510 Pa (67B2)87 s (69), 17,515 Pa (7C84)
0.4 mm242 s (158), 4,883 Pa (22B8)140 s (AA), 9,334 Pa (432D)105 s (81), 13,504 Pa (61BA)92 s (71), 16,179 Pa (7497)
0.5 mm256 s (169), 4,615 Pa (213D)150 s (B7), 8,614 Pa (3BC8)113 s (89), 12,255 Pa (5769)100 s (79), 14,519 Pa (67BB)
0.7 mm295 s (199), 3,939 Pa (1A3D)181 s (DC), 6,967 Pa (322C)141 s (AB), 9,456 Pa (43C5)127 s (9A), 10,933 Pa (4C8D)
1 mmnot all in 300 s, 2,774 Pa (1355)271 s (17B), 4,541 Pa (1DB4)222 s (141), 5,645 Pa (2753)204 s (129), 6,242 Pa (2AC2)

Each pull is given the old way, with the run’s own D number beside it.

3. The blind wall test

The wall rule written blind by a helper who was never told the dream or any pattern (files and their stamps: proofs/blind-wall-2026-11-21/). Its own words, in the old maths:

The test: balls scattered, never set in hex; hex judged by numbers (six-fold order round each ball). The pass mark, stamped before any run: mean six-fold order 0.7 or more, and 0.2 above both its start and the cold control. The result goes here when the runs finish, pass or fail.

sealed ballsstart (D / old)end (D / old)six neighbours at end
cold, Henry on3B9 / 0.659433 / 0.7186
120 C, Henry off35D / 0.585487 / 0.7873
120 C, Henry on3B9 / 0.6593B4 / 0.6543

Sealed: FAIL (the third row is below 0.7 and does not rise).

vented ballsstart (D / old)end (D / old)six neighbours at end
cold, Henry on38B / 0.6224B1 / 0.82030
120 C, Henry off3B1 / 0.651435 / 0.7205
120 C, Henry on38B / 0.622461 / 0.75519

Vented: FAIL (it rose 0.133, not 0.2, and stayed below the cold control).

shaken (a hand on the rig)startendsix neighbours at end
sealed, cold, Henry0.6480.91439
sealed, 120 C, Henry0.6480.92545
vented, cold, Henry0.6180.7405
vented, 120 C, Henry0.6180.75711

And a fault under all of these: the lab’s COLD balls pack into hex (up to 30-41 of 100 six round one) where Mike’s real cold rigs never made hex. Every blind-test comparison above was made against those cold controls, so all of them rest on a faulty control. The lab’s cold balls must first behave as real ones do.

Shaken: FAIL as stamped (the heat adds 0.011 and 0.017 over cold). The single shaken runs packed hex, but repeated five times each it does not hold: shaken mean 0.762 (11.6 six round one), still 0.780 (13.8). Withdrawn.

4. How far round a circle is, on the hex: ⬡

round = ⬡ R, and in whole numbers round² = ⬡² R² with ⬡² = 39 (48 in the old numbers). ⬡ is said “hex”: Mike J Hebblethwaite’s symbol for it, 21 November 2026, the shape the number comes from.

Code: lang/lab/runs/src/hexround-src.delta (Doomsday Book d91662d9-63f2-4968-a29e-271635efb87d). Run: delta run lang/lab/runs/hexround.delta 5BC (a radius of 5BC edges).

The rules, in D

What came out

Round ÷ radius: radius A → 7.2; 79 → 6.9; 5BC → 6.93; 46CA → 6.928. As whole numbers: (the walk round)² ÷ (the radius)² → 48 (47.999 at 46CA): that is ⬡² = 39. It never strays more than half an edge from the circle.

In the old maths

⬡ = 4√3 ≈ 6.9282, so round = ⬡R ≈ 6.9282 R, against their 2πR ≈ 6.2832 R. Each hex step crosses a little more than the straight line, by an amount set only by its angle; averaged right round that is 4√3 ÷ 2π, so their π cancels out. Their π is what comes out only if every direction is allowed at once.

The smooth circle, by free tiny steps, counted the same way: a smooth circle in Math D (its walk round over across came out 3.14149, the old pi never put in).