ⅠAREA 51 / CEREBRAL GRAPHIX
Games ▾

Exhibit · exact still plates · RH OPEN

51 Twins.

Two runs obey the same rule for a long time. Then a gate lets only one of them through. Each plate below is drawn from checked data. Nothing on this page is a simulation, and nothing is a claim about the Riemann hypothesis.

I

The pair.

ζ and a function with the same completed reflection symmetry s ↔ 1 − s, but a different Gamma factor and no Euler product, climb the same ladder of positivity tests. The twin’s finite-census scans detect a negative at rung 16,589 near x = 7140.07. Independent interval evidence gives a full-function negative at rung 16,589 throughout x ∈ [7139.83, 7140.28]; the first failing rung is unknown. The ζ track shows the inherited verified-height bound.

The pair: zeta and its Davenport–Heilbronn twinTwo tracks enter through the same reflection F(s) = F(1 − s); the twin's Gamma factor is Γ((s+1)/2), zeta's is Γ(s/2). The finite-census computation, on a finite list of the twin's zeros, reports its first detected negative at W16589; earlier sampled tests are positive. The twin has an off-line zero near 0.8085 + 85.699i; zeta continues through the verified-height bound K_H ≈ 4.7 × 10^12 into an open tip.ζσ = ½twinσ = ½zeros up to height 1000.8085 + 85.699ioff the lineF(s) = F(1 − s)same reflection · other Γ factork = 110³10⁶10⁹10¹²rung index k (log scale)ζ(s)Euler producttwin f(s)no Euler productk ≤ 218W7gate W16589W16589(7140.07) < 0earlier sampled tests: positiveKH ≈ 4.7 × 10¹²guaranteed by verified zerosevery k⇔ RH · OPENtrapped tip: a drawingdevice, not an identityThe pair: zeta and its Davenport–Heilbronn twinTwo tracks enter through the same reflection F(s) = F(1 − s); the twin's Gamma factor is Γ((s+1)/2), zeta's is Γ(s/2). The finite-census computation, on a finite list of the twin's zeros, reports its first detected negative at W16589; earlier sampled tests are positive. The twin has an off-line zero near 0.8085 + 85.699i; zeta continues through the verified-height bound K_H ≈ 4.7 × 10^12 into an open tip.ζσ = ½twinσ = ½zeros up to height 100 · off-line pair in redF(s) = F(1 − s) · same reflection, other Γ factorrung k ↓log scaleζ(s)Eulerproducttwin fno Eulerproductk ≤ 218 · both safeW7gate W16589W16589 < 0at x = 7140.07KH ≈ 4.7 × 10¹² · verified zerosevery k ⇔ RH · OPENtrapped tip: drawing device only
  • PROVEDthe twin must fail some rung
  • EVIDdetected negative at k = 16,589 on the finite zero list (rechecked at 40, 60 and 100 digits)
  • CERTζ passes every k ≤ K_H (verified-height channel)
  • OPENζ passes every k ⇔ RH

Source: TN Postmaster Volume I v5.2, Technical Dossier §§69A–69B. The twin is completed as (5/π)s/2Γ((s+1)/2)f(s): it shares the reflection of ξ, not its Gamma factor. (The textbook odd-character form has (5/π)(s+1)/2. The two differ by the constant (5/π)1/2, which cancels in the reflection and in the logarithmic derivative that every rung uses. The choice is deliberate.) The plate uses the inherited finite zero list. The calibration update certifies positive rungs 1–218 by the PP296 sector argument and 219–227 by a new pair-domination argument, all for every x > 0. A separate interval replay gives W16588(x) > 0 and W16589(x) < 0 throughout x ∈ [7139.83, 7140.28]. These results do not establish a first failing rung or a result for ζ. The shrinking circles are the trapped-tip cascade of Volume I, used only to draw “infinitely many rungs” in finite width.

II

At the table.

In 51 SEDAPS one present can have four legal pasts. They are twins: each one steps to the same present under the same rule. Here the equal 17–17–17 start and elapsed turn count 85 are known, so only a past that traces back to a fair deal in exactly 84 turns is real. Run the tape backward and watch which twins stop.

Twins at the table: four legal pasts of one SEDAPS presentFour legal pasts reach the turn-85 present. Run backward, 01 and 11 stop at once (mod-3 clock), 10 stops at turn 66, and only 00 reaches the 17–17–17 deal. Run forward, the present enters a cycle at turn 356 of period 3,744 and never ends.01734516684turn (backward from the present)00 · the recorded pastruns back to the 17–17–17 deal01 · stops at oncelengths 36·14·1; the clock needs ≡ 2 (mod 3)10 · stops at turn 66all 336 backward states die11 · stops at oncelengths 33·17·1; the clock needs ≡ 2 (mod 3)the present · turn 8535 · 16 · 0 cardsperiod 3,744= 52 × 72never endsturns 85 → 356Twins at the table: four legal pasts of one SEDAPS presentFour legal pasts reach the turn-85 present. Run backward, 01 and 11 stop at once (mod-3 clock), 10 stops at turn 66, and only 00 reaches the 17–17–17 deal. Run forward, the present enters a cycle at turn 356 of period 3,744 and never ends.0173451668400011011the present · turn 8535 · 16 · 0 cardsperiod 3,744= 52 × 72never endsturns 85 → 35600 · the recorded pastruns back to the 17–17–17 deal01 · stops at oncelengths 36·14·1; the clock needs ≡ 2 (mod 3)10 · stops at turn 66all 336 backward states die11 · stops at oncelengths 33·17·1; the clock needs ≡ 2 (mod 3)
Which past reaches the deal?

Pick a past, then run the tape backward from turn 85.

  • PROVED01 and 11 fail the mod-3 clock
  • COMPUTED10 at turn 84: every backward history dies by turn 66
  • CERT10 does come from fair deals, at turns 24, 27, 30 and 33
  • CERT00: the recorded deal replays
  • CERTforward cycle: turn 356, period 3,744

The mod-3 clock: while three queues are live, each loses one card a turn and the winner gains three, so every length is ≡ 17 − t (mod 3). At turn 84 that is 2, and the lengths of 01 and 11 fall in three different classes, so no turn works for them. The turn count matters for 10: fair deals reach it at turns 24, 27, 30 and 33, never at 84. Without the turn count, two twins survive and one bit tells them apart. Details and scripts: SEDAPS research update v0.4.1.

III

Four for four.

The gate does not always pick one. At this certified present, four different fair deals lead to its four different pasts. Every twin is real.

Four for four: a present whose four pasts all come from dealsAt a certified turn-32 present, all four legal pasts are reached from four different 17–17–17 deals. The present then cycles with period 780 = 52 × 15.01731turn (backward from the present)00 · its own deal01 · its own deal10 · its own deal11 · its own dealthe present · turn 3218 · 0 · 33 cardsperiod 780= 52 × 15turns 32 → 1,505Four for four: a present whose four pasts all come from dealsAt a certified turn-32 present, all four legal pasts are reached from four different 17–17–17 deals. The present then cycles with period 780 = 52 × 15.0173100011011the present · turn 3218 · 0 · 33 cardsperiod 780= 52 × 15turns 32 → 1,50500 · its own deal01 · its own deal10 · its own deal11 · its own deal
  • CERTfour deals replayed with the shipped rule
  • CERTforward cycle: period 780 = 52 × 15

Five such presents are certified (turns 32, 35, 41, 50 and 53). So four pasts is the true maximum even among states reached from fair deals.

IV

The periodic remnant.

A SEDAPS game has finitely many states, so a game that never ends must repeat. What is left is a loop. Among 20,000 fixed-seed equal deals, 16,599 (83.0%) entered loops, and every observed loop kept just two queues live. Then the rule is War with the winning card first, and every such loop is a multiple of 52, one more than the 51 cards in play. Loops that keep all three queues live obey no such law. The rigid ones have 29 lengths at 51 cards, only 6 of them multiples of 52, and no equal deal can reach one.

The periodic remnantOf 20,000 seeded 17–17–17 deals, 16,599 never end. Every loop they reach keeps two queues live, and its length is one of 12 values, all multiples of 52. Rigid loops with three live queues have 29 lengths, only 6 of them multiples of 52, and no equal deal reaches them.052 × 1052 × 2052 × 3052 × 4052 × 5052 × 6052 × 7052 × 8016,599 of 20,000 seeded deals never endA game that never ends must repeat. Where can it loop?Two live queues · the loops those deals reach12 lengths 2·lcm(a, 26, 25 − a), all seen; each of the 16,599 loops matches its own a3125206247801,1961,7162,1842,9643,2763,5363,7444,004Three live queues, rigid · loops that no equal deal reaches (proved)29 lengths by formula; only 6 are multiples of 521806309181,0081,0801,8361,9802,8083,672+2020 more rigid lengths lie beyond this scale, up to 64,260; 5 of them are multiples of 52disc area ∝ number of deals · blue: a multiple of 52 · red: notthe turn-85 present cycles at 3,744 = 52 × 72The periodic remnantOf 20,000 seeded 17–17–17 deals, 16,599 never end. Every loop they reach keeps two queues live, and its length is one of 12 values, all multiples of 52. Rigid loops with three live queues have 29 lengths, only 6 of them multiples of 52, and no equal deal reaches them.052 × 1052 × 2052 × 3052 × 4052 × 5052 × 6052 × 7052 × 8016,599 of 20,000 seeded deals loop.Left: the loops they reach. Right: rigid loopswith three live queues, which no equal deal reaches.two live queuesthree live, rigid312 · 2,520 deals520 · 1,242 deals624 · 40 deals780 · 2,079 deals1,196 · 193 deals1,716 · 532 deals2,184 · 761 deals2,964 · 1,294 deals3,276 · 1,778 deals3,536 · 1,730 deals3,744 · 2,213 deals4,004 · 2,217 deals1806309181,0081,0801,8361,9802,8083,672Beyond this scale: 20 more rigid lengths, up to 64,260.Of all 29, only 6 are multiples of 52.Disc area ∝ number of deals.Blue: a multiple of 52. Red: not.
  • CERTseeds 1–2,000 replay exactly with the shipped rule
  • COMPUTEDeach of the 16,599 sampled loops has length 2·lcm(a, 26, 25 − a) for its own a
  • EVID16,599 of 20,000 fixed-seed deals entered loops (83.0%)
  • EVIDno three-live loop was reached in the 20,000 fixed-seed deals
  • PROVEDtwo-queue loops: L = 2·lcm(a, 26, 25 − a), a multiple of 52
  • PROVEDthree-live loops exist at 51 cards; 23 of their 29 rigid periods are not multiples of 52, and no equal deal reaches a rigid one
  • OPENwhether a 17–17–17 deal can reach any three-live loop

Why 52: once one queue is empty, the game is War with the winning card placed first. For odd decks its loops alternate winners (Spivey 2010), and alternation forces the period 2·lcm(a, 26, 25 − a). Those are exactly the 12 lengths drawn here. Complete censuses of every state for 5, 7, 9 and 11 cards find only lengths divisible by n + 1, and no loop with three live queues; 12 odd deck sizes up to 51 were also sampled. Among even decks, 4, 6, 8 and 12 cards never loop. With three queues live, loops do exist: at 12 cards, and by a construction at every multiple of 3 from 12 up, 51 included. One 51-card loop returns after 180 turns, which is not a multiple of 52. The sizes of a rigid loop are never congruent mod 3, while an equal deal keeps them congruent, so no dealt game reaches one. At 51 cards the 29 rigid lengths are 180, 630, 918, 1,008, 1,080, 1,836, 1,980, 2,808, 3,672, 4,590, 4,752, 5,040, 5,148, 6,426, 6,930, 7,560, 9,180, 9,360, 10,098, 11,880, 11,934, 15,120, 16,380, 18,360, 19,656, 20,196, 20,592, 25,704, 64,260 (multiples of 52 in bold). Compare Volume I: the silver cascade leaves an exact rational void, 1/28 of the cone. Same shape of result (messy process, clean remnant); not the same mathematics.

The sample (EVID): the verifier's 2,000 deals, continued to seed 20,000 by evidence/remnant-20k-v1.cjs. Games that end are short: median turn 95, middle half 73–120, and none of the 3,401 ran past turn 288. Endless deals enter their loop between turns 76 and 2,628 (median 635). Balanced loops are the common ones: a = 12, where the two queues are nearly equal, has 2,520 loops; a = 1, where one queue is down to two or three cards, has 40.

V

Order twins.

In Prospect 51, poker reads a hand as a set and ignores the order. The archive lock reads the order. Swap the two red jacks of the archive key and poker cannot tell the difference. The lock can.

Order twins in Prospect 5110S JD JH KS 10D and 10S JH JD KS 10D are both Two pair. Only the first order opens the archive lock.the archive key10♠J♦J♥K♠10♦its order twin10♠J♥J♦K♠10♦pokerorder-blindarchive lockreads the orderTwo pair · 2Two pair · 2=opensstays shutOrder twins in Prospect 5110S JD JH KS 10D and 10S JH JD KS 10D are both Two pair. Only the first order opens the archive lock.the archive key10♠J♦J♥K♠10♦its order twin10♠J♥J♦K♠10♦poker · order-blindkey: Two pair · 2twin: Two pair · 2the same readingarchive lock · reads the orderkey: openstwin: stays shut
  • PROVEDsame poker category, checked by the shipped engine
  • PROVEDonly the key order opens the lock

This has the same logical shape as the orientation firewall in Dossier §69C: a test that is symmetric under swapping cannot see what the swap changes. The shape is shared; the theorems are different.

51 Twins as a game system · proposed, not built

Same rule, one gate.

A “twin” is any pair of runs that one rule cannot separate and another rule can. That gives each table a natural twin mode:

  • SEDAPS · Twin Gate. Show a present and its legal pasts. The player names the past that reaches a fair deal, then runs the tape. The answer is exact and checkable, and some presents admit more than one (plate III).
  • Prospect · Order Twin. When a hand holds the archive key's five cards in the wrong order, show its twin: same poker award, closed lock.
  • The ladder. Twins can agree for a very long time. On the sampled range, the first detected split between ζ and its twin comes at rung 16,589. Where the first split falls is itself the thing to study.

These are proposals for the author. No jbits are staked or paid on this page.

Status ledger

What each plate rests on.

StatementStatusWhere
The Davenport–Heilbronn twin fails some Widder rungPROVEDDossier v5.2 Theorem 69A.1
Finite-census scans detect W16589(7140.07) < 0; earlier sampled tests are positive (finite zero list; 40-, 60- and 100-digit rechecks agree)EVIDDossier §69A; calibration report
Wk(x) > 0 for every x > 0 and integer 1 ≤ k ≤ 227 for the full twinCERTcalibration update; pair-domination extension on the independent 179-zero census
W16588(x) > 0 and W16589(x) < 0 for every x ∈ [7139.83, 7140.28] for the full twinCERTdyadic interval replay and analytic outside bound; first failure unknown
ζ passes every rung k ≤ KH ≈ 4.7 × 1012CERT (verified height)Reading Volume §R13
ζ passes every rungOPEN (⇔ RH)Reading Volume (R.4)
Turn-85 pasts 01 and 11 cannot come from a 17–17–17 dealPROVEDmod-3 clock, SEDAPS update v0.4 §2
Turn-85 past 10 cannot come from a 17–17–17 deal in 84 turnsCOMPUTEDexhaustive search under the proved shape lemma
Past 10 does come from 17–17–17 deals at turns 24, 27, 30 and 33, and at no other turnCERT + COMPUTEDpast10-certificates-v0-4-1.json, verify-past10-v0-4-1.cjs
Turn-85 present cycles from turn 356 with period 3,744; turn-32 present with period 780CERTdeterministic replay, verify-twins-v1.cjs
Four legal pasts all reached from deals (turns 32, 35, 41, 50, 53)CERTdstart4-certificates-v0-4.json
Every cycle reached in the 20,000 fixed-seed 17–17–17 deals had a length divisible by 52EVIDAll 16,599 observed loops had two live queues and matched 2·lcm(a, 26, 25 − a) for their own a (evidence/remnant-20k-v1.json); 12 sampled odd deck sizes from 7 to 51
In the 20,000 fixed-seed deals, 16,599 (83.0%) entered loops; the 3,401 ending games ended by turn 288 (median 95)EVIDReproducible results for seeds 1–20,000; no population confidence interval is claimed; evidence/remnant-20k-v1.cjs
A two-queue loop has length 2·lcm(α/2, (n+1)/2, (n−1−α)/2), a multiple of n + 1PROVED (alternation: Spivey 2010)SEDAPS update v0.4.1 §2; complete censuses n ≤ 11
A rigid three-live loop has period 3·lcm(kA, kB, kC, 17 − kA, 17 − kB, 17 − kC) or 3·lcm(kA, kB, kC, 17, 18) at 51 cards: 29 values, 6 of them multiples of 52, all multiples of 18PROVEDwhole-packet counts k; SEDAPS update v0.4.1 §2
No loop keeps three queues livefalse (CERT, PROVED)CERT at n = 12; PROVED at every multiple of 3 from 12, 51 included (construction); none for n ≤ 11 or n = 13 (complete censuses); SEDAPS update v0.4.1 §2
No such loop is reachable from an equal dealOPEN at 51PROVED for rigid loops at every n; COMPUTED for all loops at n = 9, 12, 15 (complete synchronised-loop searches); the mod-3 clock keeps equal-deal sizes congruent
Key and order twin share a poker category; only the key opens the lockPROVEDProspect 51 engine

Replay the card results: node area-51/twins-51/verify-twins-v1.cjs rebuilds twins-data-v1.js from the shipped SEDAPS rule, certificates and Prospect engine and checks it byte for byte. Its analytic constants are copied from the Dossier evidence; that command does not independently recompute or certify them. The two rung signs at k = 16,588 and 16,589 are recomputed separately by evidence/dh_twin_highprec.py. build-twins-page-v1.cjs redraws this page from that data. Plate IV's sample is rerun by node area-51/twins-51/evidence/remnant-20k-v1.cjs (about 30 seconds): it checks its fast copy of the rule against the shipped one on the first 200 deals, and its first 2,000 deals must reproduce the data file exactly.

Cite

Jeffery Lyn Huckstead, 51 Twins v1: exact still plates, Cerebral Graphix, 2026. Part of Project 51: Reading the Record Across Rules, 10.5281/zenodo.23048482. The TN Postmaster numbers cite Volume I v5.2, 10.5281/zenodo.23004335 (version series 10.5281/zenodo.21968915).

Read further

Twin calibration reportPlay 51 SEDAPSPlay Prospect 51TN PostmasterVolume I · trapped tip