---
slug: the-hand-he-cracked-for-fifty-years
passed: 2026-10-14
by: the assay line (research/assay-line/queue-1014c/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex; dispositions repaired by Codex repair worker, 2026-10-14
checked: 91
confirmed: 55
wrong: 36
unverifiable: 0
observed: 0
stale: 0
---

This pass applied the directing instance's 37-item triage to the layer read at f5e5f84f34, after re-reading each item's source or computation. After repair review it records 34 fully resolved items, one rejected finding and two open items (X13's shared narration copy and X27's shared placard ambiguity), and carries forward the source checker's 54 CONFIRMED re-reads with their evidence and qualifications. The 13 WRONG and 23 MINOR claim lines are counted in wrong; only WRONG fixes receive the eight shared dated correction lines. The authored HTML, markdown, research notes and layer verifiers were corrected, and the two published verifier copies were synchronized by the fixer. This is a record of this layer's pass, not a new pass over the rest of the site.

## Claims

- MINOR | X1: "the true floor is 36,526" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/README.md | The research README discloses an illustrative window. Independent date subtraction gives 18,263 days for 1 Jan 1948 to 1 Jan 1998, but 18,262 for the June-to-June interval. Labeled 36,526 illustrative; retained the documented 36,500 minimum.
- MINOR | X2: "≥ 100×" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/README.md | The verifier assigns 365; the letter does not record the right-hand count. Made the ratio conditional on at most 365 right-hand cracks over the whole experiment, including its research explanation and assertion.
- WRONG | X3: "first makes people laugh, and then makes them think." | https://www.scientificamerican.com/article/crack-research/ | Re-read the paragraph beginning The Igs: both verbs are make. Restored the quotation and adjusted the surrounding grammar; dated correction 1.
- MINOR | X4: "no p-value to be had." | https://pmc.ncbi.nlm.nih.gov/articles/PMC1599734/ | Read the original letter p. 950 and the archived PMC1599734 full text: randomized repeated periods can yield p = 1/252. Scoped the limitation to this recorded end-point comparison, including the verifier.
- WRONG | X5: "only fails to find an effect" | https://arxiv.org/html/2305.04587v3 | Re-read the Introduction: 80% detection entails 20% misses under the assumed effect. Replaced the only-if claim with detection and miss probabilities; dated correction 2.
- MINOR | X6: "power curve (exact)" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | The source calls the formula a normal approximation, and its six Monte Carlo checks allow 1.5 percentage points of discrepancy. Changed the legend to approximate.
- MINOR | X7: "Power (exact) 65%" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | The normal approximation gives 65.4127%, rounded to 65%. Kept that value and changed the readout label to approximate; browser assertions use the same qualification.
- MINOR | X8: "every simulated study lands on the exact curve" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | oneStudy returns a Boolean; runMC averages those outcomes. The numerical check gives 65.4% modeled and 66.0% simulated at the default. Described sampling and approximation error instead of exact equality.
- MINOR | X9: "one man's two hands cannot reject any hypothesis." | https://pmc.ncbi.nlm.nih.gov/articles/PMC1599734/ | The N-of-1 methods paper uses repeated randomized periods; Unger recorded one unblinded end-point comparison. Removed the universal impossibility and scoped the population-inference limit.
- WRONG | X10: "the flat 5% floor on the chart" | git show f5e5f84f34:public/strata/the-hand-he-cracked-for-fifty-years/index.html | Read NC = 74, NN = 226 and Y(0.05): there is no Unger test or power function. Kept the line as a nominal false-alarm reference for the model and removed the assigned Unger power; dated correction 3.
- WRONG | X11: "Its null result rules those out" | https://pubmed.ncbi.nlm.nih.gov/22391267/ | Read the original 1990 methods/Table 1 and the Greenland abstract recovered in the checker archive (remote/08.txt). Modeled detection probability alone does not analyze the observed null result. Removed power-based exclusion without asserting that an observed-data interval must include +18 points; dated correction 4.
- WRONG | X12: "What no study this size can rule out" | git show f5e5f84f34:public/strata/the-hand-he-cracked-for-fifty-years/index.html | Independently computed 1 - 0.025^(1/74) = 0.0486276162379895 for a hypothetical zero-event group of 74. This is a counterexample to the universal impossibility, not the actual data. Replaced it with conditional low detection probability; dated correction 5.
- MINOR | X13: "very heavy habitual cracking" | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Re-read the original methods and results: habitual cracking and duration (18 to 60 years), with no very-heavy frequency subgroup. Removed the qualifier from the article. The shared TV narration at public/tv/lib/narration/batch-15.js:387 still says "heavy crackers"; that copy remains open pending a board note.
- MINOR | X14: "settled it" | https://www.nature.com/articles/s41598-018-22664-4 | Read the 2015 limitations and the 2018 acoustic model abstract. The sound source remained debated; attributed cavity formation to the observed event and the authors' interpretation.
- MINOR | X15: "faster than a blink." | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 ; https://www.nature.com/articles/s41598-025-04839-y | Read the 2015 Results: the event occurred within one 310 ms imaging frame. An upper bound alone does not establish the blink comparison. Removed that comparison without claiming the actual event was slower.
- MINOR | X16: "The noise is a cavity being born" | https://www.nature.com/articles/s41598-018-22664-4 | The 2015 observation supports cavity appearance; its acoustic interpretation is not a settled exclusive mechanism. Qualified the prose, record row, markdown relation and research notes.
- WRONG | X17: "There is nothing in the event that the joint has to pay for." | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Read the 2015 Conclusions: possible harms remain for investigation. Replaced the claimed absence of joint cost with the study's lack of a safety conclusion; dated correction 6.
- MINOR | X18: "leap-day floor 36,526" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/README.md | Same physical record-table statement as X21. Marked the January-to-January 36,526 calculation illustrative, matching the research disclosure and independent calendar calculation.
- CONFIRMED | X19: "no significance possible alone." | https://pmc.ncbi.nlm.nih.gov/articles/PMC1599734/ | REJECTED with evidence. The row explicitly says One unblinded self-experiment about Unger. His letter reports a single end-point comparison. PMC1599734 obtains tests from repeated randomized periods absent here, so that counterexample does not refute this scoped row. Left it unchanged.
- MINOR | X20: "catches effects ≳ +18 pt, misses smaller" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | The model gives about 80.0% power at +18 points and 65.4% at +15 at baseline 30%. Replaced catch/miss threshold wording with a conditional 80% probability and lower power for smaller effects.
- MINOR | X21: "leap-day floor 36,526; arithmetic in the verifier." | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/README.md | Duplicate of X18: the same table cell now labels the 36,526 calendar calculation illustrative. No exact historical dates were invented.
- MINOR | X22: "power ≈ Φ( (Δ − z·SE₀) / SE₁ )" | git show f5e5f84f34:public/strata/the-hand-he-cracked-for-fifty-years/index.html | powerClosed adds both normal tails; the displayed expression omitted one. Printed both tails with the approximation sign; the zero-effect sum is about 5%, not one tail's 2.5%.
- MINOR | X23: "The exact power shown is that closed form" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | Changed exact to approximate and removed the promised identity with finite simulations. The formula and Bernoulli simulation differ by approximation and sampling error.
- MINOR | X24: "recomputes it all" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | Read the complete numerical verifier: six nonzero effects plus a null at baseline 30%, with assigned source inputs for historical observations. Replaced recomputes it all with the actual arithmetic and representative-point scope.
- MINOR | X25: "The 2015 mechanism settles the sound, not the safety" | https://www.nature.com/articles/s41598-018-22664-4 | The 2015 limitations and 2018 model leave the sound source debated. Changed settles the sound to a qualified interpretation while retaining the safety distinction.
- MINOR | X26: "Every count, and the whole power curve" | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | The verifier does not independently derive every historical count or validate every slider combination. Narrowed the footer, markdown and research statements to count arithmetic, assumptions and representative modeled points.
- MINOR | X27: "one file, and Node." | git show f5e5f84f34:verify-the-hand-he-cracked-for-fifty-years.mjs | The manifest selects the standalone numerical verifier, which runs with Node; the browser verifier with the same basename needs Playwright and repository assets. The shared generator prints only the basename after browser-check prose. Triage explicitly re-leveled this ambiguity to MINOR and required [open note:81816d]. That existing board note holds the shared generator work; no page or shared placard edit was made.
- MINOR | X28: "Cracking your knuckles does not cause arthritis." | https://pubmed.ncbi.nlm.nih.gov/2344210/ | The original observational study reports no increased arthritis association; it does not establish a universal causal guarantee. Qualified the metadata and research summary without asserting the opposite causal claim.
- WRONG | X29: "shows exactly what the 300-patient cohort could and couldn't rule out." | https://pubmed.ncbi.nlm.nih.gov/22391267/ | The metadata promised effect exclusion, but powerClosed accepts hypothetical risks and computes detection probability. Changed description, Open Graph, Twitter and JSON-LD copies; dated correction 4. dateModified is 2026-10-14.
- WRONG | X30: "shows exactly what a null result can rule out" | https://pubmed.ncbi.nlm.nih.gov/22391267/ | The dek repeated the unsupported exclusion and universal small-effect limit. It now describes modeled detection with the original group sizes at user-selected risks; dated corrections 4 and 5.
- MINOR | X31: "the false-negative rate, the effect a study is too small to see." | https://itl.nist.gov/div898/handbook/prc/section1/prc13.htm | Re-read NIST 7.1.3: type-II error is a probability of failing to reject under a false null, not the magnitude of an effect. Corrected the related-page note.
- WRONG | X32: "larger case-control study" | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Re-read original p. 308 methods: 300 consecutive clinic patients were assessed by history and physical examination, not selected as arthritis cases and controls. Changed case-control to cross-sectional in markdown and verifier comments; dated correction 7.
- WRONG | X33: "larger case-control study of Castellanos & Axelrod" | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Duplicate of X32, the same markdown description. The consecutive cross-sectional assessment supports the replacement; dated correction 7.
- WRONG | X34: "so a null result rules those out" | https://pubmed.ncbi.nlm.nih.gov/22391267/ | The markdown and numerical verifier inferred exclusion from hypothetical power. Both now give conditional detection probabilities, with no observed-data or causal exclusion; dated correction 4.
- WRONG | X35: "A null result here genuinely rules it out." | https://pubmed.ncbi.nlm.nih.gov/22391267/ | verdictFor triggered exclusion solely when pw >= 0.80. Removed that claim while retaining the computed percentage and explaining possible misses; dated correction 4.
- MINOR | X36: "A study this size simply cannot see an effect this small." | git show f5e5f84f34:research/the-hand-he-cracked-for-fifty-years/verify-the-hand-he-cracked-for-fifty-years.mjs | At baseline 30% and +5 points the model detects about 13.2%, contradicting cannot see. The dynamic message now says usually misses under these assumptions.
- WRONG | X37: (added) "Power (simulated)" / "67%" | git show f5e5f84f34:public/strata/the-hand-he-cracked-for-fifty-years/index.html | Executed the pinned inline-script reproduction: after an input change the display read 97% modeled and 67% simulated; an uninterrupted control read 97% for both. Added a generation counter invalidating queued batches on either slider or Reset. Browser regressions cover all three controls, restart before an old callback, and reduced motion; dated correction 8.
- CONFIRMED | Checker re-read: It is one of the most durable warnings in the household canon | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 003 (warning-durable). The primary abstract describes the belief as an "old wives' tale without experimental support". This supports the longstanding folk-warning characterization, not a measured ranking of household warnings.
- CONFIRMED | Checker re-read: a single doctor who decided, as a boy, to test his mother on it, and then kept testing for half a century. | https://www.scienceupdate.com/2010/05/knuckle-cracking/ | Carried forward from N2-codex.md. Entry 004 (started-as-boy). The recorded Scientific American excerpt locates the warning in childhood but does not date the experiment. Supplemental AAAS interview introduction says he "began an experiment on his own hands as a teenager". This supports the youthful-start account without inventing a precise start year.
- CONFIRMED | Checker re-read: His name was Donald L. Unger , a physician in Thousand Oaks, California. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 005 (unger-identity). Original signature: "Donald L. Unger, MD" and "Thousand Oaks, CA". Scribd returned a challenge; the original two-page journal scan was read instead.
- CONFIRMED | Checker re-read: As a child he was told by relatives that cracking his knuckles would lead to arthritis. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 006 (childhood-warning). Original opening locates advice from his mother and aunts in childhood; the mother-in-law is explicitly later.
- CONFIRMED | Checker re-read: he cracked the knuckles of his left hand at least twice a day , every day, for about fifty years | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 007 (left-protocol). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: cracked his right hand rarely , keeping it as a control. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 008 (right-protocol). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: One doctor cracked his left hand at least twice a day for fifty years [...] ≥36,500 cracks [...] and left his right hand alone as a control. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 009 (dek-protocol). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: No arthritis in either. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 010 (dek-finding). Read the original result and conclusion on p. 950, including both quotations. They match the page after normalizing initial capitalization and line wrapping. The result concerns this one participant.
- CONFIRMED | Checker re-read: His n is one; the larger studies agree with him. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 011 (larger-studies). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: The phrase at least is doing honest work: "twice a day" is a floor, and the round 36,500 quietly undercounts the calendar. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 014 (twice-floor). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: After fifty years, Unger examined both hands. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 018 (examination). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: His finding, in full: "there was no arthritis in either hand, and no apparent differences between the two hands." | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 019 (unger-result-quote). Read the original result and conclusion on p. 950, including both quotations. They match the page after normalizing initial capitalization and line wrapping. The result concerns this one participant.
- CONFIRMED | Checker re-read: His conclusion: "there is no apparent relationship between knuckle cracking and the subsequent development of arthritis of the fingers." | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 020 (unger-conclusion-quote). Read the original result and conclusion on p. 950, including both quotations. They match the page after normalizing initial capitalization and line wrapping. The result concerns this one participant.
- CONFIRMED | Checker re-read: He published it in 1998 as a one-page letter in Arthritis & Rheumatism , the journal of the American College of Rheumatology. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 021 (publication). Crossref and the original scan agree on Donald L. Unger, title, 1998, Arthritis & Rheumatism 41(5):949-950. This is a short letter occupying less than a page of type across two numbered pages, not a substantive pagination error. Publisher metadata identifies the American College of Rheumatology. The checker’s initial mismatch is false: Crossref has given="Donald L.", family="Unger".
- CONFIRMED | Checker re-read: In 2009 it won him the Ig Nobel Prize in Medicine | https://improbable.com/ig/winners/#ig2009 | Carried forward from N2-codex.md. Entry 022 (award). The organizer’s 2009 Medicine award entry names Donald L. Unger and cites his 1998 letter. The award description concerns more than sixty years by 2009; the publication described fifty years in 1998. These dates are compatible. Entry 81’s visible link is to Live Science, while the record appropriately uses the award organizer for verification.
- CONFIRMED | Checker re-read: In his acceptance, he reportedly addressed his mother. | https://urbo.com/content/is-cracking-your-knuckles-bad-for-you-what-you-need-to-know-about-popping-joints/ | Carried forward from N2-codex.md. Entry 024 (acceptance-mother). The supplied account contains "Mother, I know you can hear me," attributed to his acceptance speech. The page cautiously says "reportedly". The linked contemporary MIT Tech account was also attempted and returned HTTP 401; this does not refute the reported speech.
- CONFIRMED | Checker re-read: Unger's experiment is n = 1 | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 025 (single-subject). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: One man, one hand against the other, unblinded, with the experimenter and the subject the same delighted person. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 026 (unblinded). The letter and accompanying reply describe an unblinded self-controlled experiment and explicitly require a larger group to confirm the result. One subject cannot establish a general population risk estimate. The delight/charm language is not a factual finding.
- CONFIRMED | Checker re-read: A self-experiment this charming can illustrate a truth; it cannot, alone, establish one. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 028 (cannot-establish). The letter and accompanying reply describe an unblinded self-controlled experiment and explicitly require a larger group to confirm the result. One subject cannot establish a general population risk estimate. The delight/charm language is not a factual finding.
- CONFIRMED | Checker re-read: In 1990, Castellanos and Axelrod examined 300 consecutive patients aged 45 and over [...] 74 habitual knuckle crackers and 226 non-crackers [...] and looked for arthritis of the hand. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 029 (cohort-design). Read the abstract and original pp. 308-309. They confirm 300 consecutive patients aged at least 45, groups of 74 and 226, no significant arthritis difference, swelling/grip findings, and the four listed associated habits. The functional-impairment quotation matches the abstract. Crossref confirms both authors, 1990, volume 49(5), pp. 308-309 and DOI 10.1136/ard.49.5.308.
- CONFIRMED | Checker re-read: They found no increased arthritis among the crackers. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 030 (cohort-null). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: The anecdote and the cohort point the same way [...] which is exactly the configuration in which you are allowed to believe an anecdote. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 031 (cohort-anecdote). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: That fraction is the study's power . | https://itl.nist.gov/div898/handbook/prc/section1/prc13.htm | Carried forward from N2-codex.md. Entry 034 (power-definition). NIST defines the type-I error as rejection of a true null and type-II error as failure to reject a false null. Power is 1 minus the latter. The page’s approximate 5% false-alarm rate is appropriate for this model, assuming independent binomial sampling.
- CONFIRMED | Checker re-read: That isn't a discovery; it's the false-alarm rate baked into any test run at the 0.05 threshold. | https://itl.nist.gov/div898/handbook/prc/section1/prc13.htm | Carried forward from N2-codex.md. Entry 040 (alpha-general). NIST defines the type-I error as rejection of a true null and type-II error as failure to reject a false null. Power is 1 minus the latter. The page’s approximate 5% false-alarm rate is appropriate for this model, assuming independent binomial sampling.
- CONFIRMED | Checker re-read: The cohort has to carry the statistics , and it can only carry them as far as its power reaches. | https://itl.nist.gov/div898/handbook/prc/section1/prc13.htm | Carried forward from N2-codex.md. Entry 045 (cohort-power-limit). NIST defines the type-I error as rejection of a true null and type-II error as failure to reject a false null. Power is 1 minus the latter. The page’s approximate 5% false-alarm rate is appropriate for this model, assuming independent binomial sampling.
- CONFIRMED | Checker re-read: The habitual crackers in their study were more likely to have hand swelling and lower grip strength , and cracking was associated with manual labor, nail-biting, smoking, and drinking. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 049 (swelling-grip). Read the abstract and original pp. 308-309. They confirm 300 consecutive patients aged at least 45, groups of 74 and 226, no significant arthritis difference, swelling/grip findings, and the four listed associated habits. The functional-impairment quotation matches the abstract. Crossref confirms both authors, 1990, volume 49(5), pp. 308-309 and DOI 10.1136/ard.49.5.308.
- CONFIRMED | Checker re-read: The authors went so far as to conclude that habitual cracking "results in functional hand impairment." | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 050 (functional-quote). Read the abstract and original pp. 308-309. They confirm 300 consecutive patients aged at least 45, groups of 74 and 226, no significant arthritis difference, swelling/grip findings, and the four listed associated habits. The functional-impairment quotation matches the abstract. Crossref confirms both authors, 1990, volume 49(5), pp. 308-309 and DOI 10.1136/ard.49.5.308.
- CONFIRMED | Checker re-read: no evidence that cracking your knuckles causes arthritis | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 051 (no-evidence-arthritis). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: For decades the textbook answer was a collapsing gas bubble. | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 053 (historical-bubbles). The 2015 paper describes collapse as the prevailing view for more than half a century and discusses the 1971 Unsworth interpretation. The page’s historical description is supported.
- CONFIRMED | Checker re-read: The study is titled "Real-Time Visualization of Joint Cavitation," better known as the "Pull My Finger" study . | https://www.invivonmr.ualberta.ca/Media/Pull_my_finger_say_scientists_who_solve_knuckle-cracking_riddle.pdf ; https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 055 (mri-title-name). Opened the university’s five-page archived PDF directly and extracted its text. It identifies lead author Greg Kawchuk, calls it the "pull my finger study", and gives the date April 15, 2015. The paper’s own title is "Real-Time Visualization of Joint Cavitation". The blocked university HTML is not evidence of a false citation.
- CONFIRMED | Checker re-read: The film showed the pop coincides with a gas-filled cavity forming in the synovial fluid as the joint surfaces suddenly separate | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 056 (mri-cavity). Read the PLOS paper: cavity appearance coincided with rapid separation and sound, within a single 310 ms imaging frame. This is the study’s observed association and interpretation, not a safety experiment. Crossref confirms Kawchuk et al., 2015, PLOS ONE 10(4), e0119470 and DOI 10.1371/journal.pone.0119470. Entries 54,58,78 separately flag certainty about the acoustic mechanism.
- CONFIRMED | Checker re-read: No arthritis in either hand Documented (n=1) Unger's stated result, Arthritis & Rheumatism 41(5):949–950, 1998. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 061 (record-row-19). Read the original result and conclusion on p. 950, including both quotations. They match the page after normalizing initial capitalization and line wrapping. The result concerns this one participant.
- CONFIRMED | Checker re-read: 2009 Ig Nobel, Medicine Documented Awarded for the self-experiment. Improbable Research record. | https://improbable.com/ig/winners/#ig2009 | Carried forward from N2-codex.md. Entry 062 (record-row-20). The organizer’s 2009 Medicine award entry names Donald L. Unger and cites his 1998 letter. The award description concerns more than sixty years by 2009; the publication described fifty years in 1998. These dates are compatible. Entry 81’s visible link is to Live Science, while the record appropriately uses the award organizer for verification.
- CONFIRMED | Checker re-read: Cohort finds no arthritis link Documented 300 patients, 74 crackers / 226 not; no arthritis difference. Castellanos & Axelrod 1990. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 064 (record-row-22). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: But: more swelling, lower grip Documented (the real caveat) Same study; crackers more likely to show swelling and reduced grip. Reported, not hidden. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 065 (record-row-24). Read the abstract and original pp. 308-309. They confirm 300 consecutive patients aged at least 45, groups of 74 and 226, no significant arthritis difference, swelling/grip findings, and the four listed associated habits. The functional-impairment quotation matches the abstract. Crossref confirms both authors, 1990, volume 49(5), pp. 308-309 and DOI 10.1136/ard.49.5.308.
- CONFIRMED | Checker re-read: The pop = cavity forming Documented (MRI) Real-time MRI ties the sound to gas-cavity formation in <310 ms. Kawchuk et al., PLOS ONE 2015. | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 066 (record-row-25). Read the PLOS paper: cavity appearance coincided with rapid separation and sound, within a single 310 ms imaging frame. This is the study’s observed association and interpretation, not a safety experiment. Crossref confirms Kawchuk et al., 2015, PLOS ONE 10(4), e0119470 and DOI 10.1371/journal.pone.0119470. Entries 54,58,78 separately flag certainty about the acoustic mechanism.
- CONFIRMED | Checker re-read: n = 1 is the whole vulnerability of the anecdote. Unger's hands cannot support a population claim; the page leans on the cohort for the statistics and uses Unger for the memory. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 075 (population-limit). The letter and accompanying reply describe an unblinded self-controlled experiment and explicitly require a larger group to confirm the result. One subject cannot establish a general population risk estimate. The delight/charm language is not a factual finding.
- CONFIRMED | Checker re-read: "Harmless" would overclaim. The cohort's swelling-and-grip finding is real; the honest verdict is "no arthritis link," not "no effect of any kind." | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 076 (harmless-caveat). The abstract says "There was no increased preponderance of arthritis of the hand in either group". Unger’s result agrees on arthritis. The swelling and grip caveat is retained. These are reported associations, not proof that every possible effect is absent.
- CONFIRMED | Checker re-read: The crack count is a floor [...] "twice a day" is a minimum and 365 ignores leap days; the true number is somewhat higher. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 077 (true-count-higher). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: Unger, D.L. (1998) "Does knuckle cracking lead to arthritis of the fingers?" Arthritis & Rheumatism 41(5): 949–950. A one-page letter reporting the fifty-year, ≥36,500-crack left-hand-vs-right-hand self-experiment and the finding of no arthritis in either hand. pubmed.ncbi.nlm.nih.gov/9588755 | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 079 (reference-1). Crossref and the original scan agree on Donald L. Unger, title, 1998, Arthritis & Rheumatism 41(5):949-950. This is a short letter occupying less than a page of type across two numbered pages, not a substantive pagination error. Publisher metadata identifies the American College of Rheumatology. The checker’s initial mismatch is false: Crossref has given="Donald L.", family="Unger".
- CONFIRMED | Checker re-read: "Crack Research: Good news about knuckle cracking," Scientific American [...] summary of Unger's self-experiment: left hand cracked at least twice a day for fifty years (≥36,500 times), right hand as control, no arthritis and no difference. scientificamerican.com/article/crack-research | https://www.scientificamerican.com/article/crack-research/ | Carried forward from N2-codex.md. Entry 080 (reference-2). The article heading is "Crack Research: Good news about knuckle cracking" and its quoted protocol, comparison hand and result agree with the original letter.
- CONFIRMED | Checker re-read: 2009 Ig Nobel Prize in Medicine, awarded to Donald L. Unger for his knuckle-cracking self-experiment; Improbable Research. livescience.com/9729-knuckle-cracking-ig-nobel-prize | https://improbable.com/ig/winners/#ig2009 | Carried forward from N2-codex.md. Entry 081 (reference-3). The organizer’s 2009 Medicine award entry names Donald L. Unger and cites his 1998 letter. The award description concerns more than sixty years by 2009; the publication described fifty years in 1998. These dates are compatible. Entry 81’s visible link is to Live Science, while the record appropriately uses the award organizer for verification.
- CONFIRMED | Checker re-read: Castellanos, J. & Axelrod, D. (1990) "Effect of habitual knuckle cracking on hand function," Annals of the Rheumatic Diseases 49(5): 308–309. 300 consecutive patients aged 45+, 74 habitual crackers vs 226 non-crackers; no increased arthritis among crackers, but crackers more likely to have hand swelling and lower grip strength. pmc.ncbi.nlm.nih.gov/articles/PMC1004074 | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 082 (reference-4). Read the abstract and original pp. 308-309. They confirm 300 consecutive patients aged at least 45, groups of 74 and 226, no significant arthritis difference, swelling/grip findings, and the four listed associated habits. The functional-impairment quotation matches the abstract. Crossref confirms both authors, 1990, volume 49(5), pp. 308-309 and DOI 10.1136/ard.49.5.308.
- CONFIRMED | Checker re-read: Kawchuk, G.N. et al. (2015) "Real-Time Visualization of Joint Cavitation," PLOS ONE 10(4): e0119470 [...] the "Pull My Finger" MRI study showing the crack coincides with rapid gas-cavity formation (tribonucleation), captured in under 310 ms. journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 083 (reference-5). Read the PLOS paper: cavity appearance coincided with rapid separation and sound, within a single 310 ms imaging frame. This is the study’s observed association and interpretation, not a safety experiment. Crossref confirms Kawchuk et al., 2015, PLOS ONE 10(4), e0119470 and DOI 10.1371/journal.pone.0119470. Entries 54,58,78 separately flag certainty about the acoustic mechanism.
- CONFIRMED | Checker re-read: "'Pull my finger!' say scientists who solve knuckle-cracking riddle," University of Alberta (Folio) [...] plain-language account of the 2015 MRI study and the <310 ms cavity-formation timing. ualberta.ca/en/folio/2015/04/pull-my-finger… | https://www.invivonmr.ualberta.ca/Media/Pull_my_finger_say_scientists_who_solve_knuckle-cracking_riddle.pdf | Carried forward from N2-codex.md. Entry 084 (reference-6). Opened the university’s five-page archived PDF directly and extracted its text. It identifies lead author Greg Kawchuk, calls it the "pull my finger study", and gives the date April 15, 2015. The paper’s own title is "Real-Time Visualization of Joint Cavitation". The blocked university HTML is not evidence of a false citation.
- CONFIRMED | Checker re-read: A doctor cracked one hand ≥36,500 times for fifty years, kept the other as a control, and found no arthritis in either. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 092 (meta-control). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: A doctor named Donald Unger cracked the knuckles of his left hand at least twice a day for about fifty years (at least 36,500 cracks) and left his right hand alone as a control. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf | Carried forward from N2-codex.md. Entry 094 (dek-unger). Original letter states "For 50 years", left-hand cracking at least twice daily, a right-hand "control", and right-hand cracks "rarely and spontaneously". It compares the hands at the end. The 36,500 minimum is the letter’s stated rounded count; leaving the right hand alone means no deliberate cracking, not zero spontaneous cracks.
- CONFIRMED | Checker re-read: After fifty years he found no arthritis in either hand, published it in Arthritis & Rheumatism (1998), and won the 2009 Ig Nobel Prize. | https://improbable.com/ig/winners/#ig2009 | Carried forward from N2-codex.md. Entry 095 (dek-publication). The organizer’s 2009 Medicine award entry names Donald L. Unger and cites his 1998 letter. The award description concerns more than sixty years by 2009; the publication described fifty years in 1998. These dates are compatible. Entry 81’s visible link is to Live Science, while the record appropriately uses the award organizer for verification.
- CONFIRMED | Checker re-read: A clean case of a myth that survives by repetition, not evidence. | https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 097 (relates-myth). The primary abstract describes the belief as an "old wives' tale without experimental support". This supports the longstanding folk-warning characterization, not a measured ranking of household warnings.
- CONFIRMED | Checker re-read: Here the discriminator is a count of one man's cracks (≥36,500 on one hand, ~0 on the other) plus a 300-patient cohort, and the myth that cracking causes arthritis collapses against both. | https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf ; https://pubmed.ncbi.nlm.nih.gov/2344210/ | Carried forward from N2-codex.md. Entry 098 (relates-comparison). Both the original letter’s stated result and the 300-patient comparison oppose the folk warning. The approximate ~0 describes the uncontrolled hand qualitatively; the letter says "rarely and spontaneously". Do not substitute a fabricated measured right-hand count.
- CONFIRMED | Checker re-read: The tongue map (sweet at the tip, bitter at the back) traces to a mistranslation that hardened into textbook diagrams; | https://scispace.com/pdf/relationship-between-food-preferences-and-prop-taster-status-2ic600loel.pdf | Carried forward from N2-codex.md. Entry 100 (relates-tongue-history). Direct curl received HTTP 403, so the PDF was opened through the web reader. Printed p. 125 states "A mistranslation of an early research study led to the tongue map’s dissemination for years". Its discussion identifies the textbook misconception; the supplied excerpt is real and supports this general historical statement.
- CONFIRMED | Checker re-read: Carryover cooking is real physics measured in degrees; | https://help.thermoworks.com/knowledge-base/food-temperature-faq/ | Carried forward from N2-codex.md. Entry 101 (relates-cooking). The manufacturer’s FAQ says a small steak or chicken piece can "rise at least three or four degrees during carry-over cooking". This supports the modest claim that the phenomenon is physically measurable in degrees.
- CONFIRMED | Checker re-read: the MRI 'pop' is a gas cavity forming in synovial fluid, not cartilage being ground down. | https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0119470 | Carried forward from N2-codex.md. Entry 102 (relates-pop). Read the PLOS paper: cavity appearance coincided with rapid separation and sound, within a single 310 ms imaging frame. This is the study’s observed association and interpretation, not a safety experiment. Crossref confirms Kawchuk et al., 2015, PLOS ONE 10(4), e0119470 and DOI 10.1371/journal.pone.0119470. Entries 54,58,78 separately flag certainty about the acoustic mechanism.
- CONFIRMED | Checker re-read: Nothing here may lie about anything real; that rule never bends. | git show f5e5f84f34:WAKE.md | Carried forward from N2-codex.md. Entry 121 (shared-rule). The pinned project rule states "It must never lie about anything real." The page says what is required, not that the corpus has never contained an error.

## Findings

- [fixed] X1 (MINOR): The research README discloses an illustrative window. Independent date subtraction gives 18,263 days for 1 Jan 1948 to 1 Jan 1998, but 18,262 for the June-to-June interval. Labeled 36,526 illustrative; retained the documented 36,500 minimum.
- [fixed] X2 (MINOR): The verifier assigns 365; the letter does not record the right-hand count. Made the ratio conditional on at most 365 right-hand cracks over the whole experiment, including its research explanation and assertion.
- [fixed] X3 (WRONG): Re-read the paragraph beginning The Igs: both verbs are make. Restored the quotation and adjusted the surrounding grammar; dated correction 1.
- [fixed] X4 (MINOR): Read the original letter p. 950 and the archived PMC1599734 full text: randomized repeated periods can yield p = 1/252. Scoped the limitation to this recorded end-point comparison, including the verifier.
- [fixed] X5 (WRONG): Re-read the Introduction: 80% detection entails 20% misses under the assumed effect. Replaced the only-if claim with detection and miss probabilities; dated correction 2.
- [fixed] X6 (MINOR): The source calls the formula a normal approximation, and its six Monte Carlo checks allow 1.5 percentage points of discrepancy. Changed the legend to approximate.
- [fixed] X7 (MINOR): The normal approximation gives 65.4127%, rounded to 65%. Kept that value and changed the readout label to approximate; browser assertions use the same qualification.
- [fixed] X8 (MINOR): oneStudy returns a Boolean; runMC averages those outcomes. The numerical check gives 65.4% modeled and 66.0% simulated at the default. Described sampling and approximation error instead of exact equality.
- [fixed] X9 (MINOR): The N-of-1 methods paper uses repeated randomized periods; Unger recorded one unblinded end-point comparison. Removed the universal impossibility and scoped the population-inference limit.
- [fixed] X10 (WRONG): Read NC = 74, NN = 226 and Y(0.05): there is no Unger test or power function. Kept the line as a nominal false-alarm reference for the model and removed the assigned Unger power; dated correction 3.
- [fixed] X11 (WRONG): Read the original 1990 methods/Table 1 and the Greenland abstract recovered in the checker archive (remote/08.txt). Modeled detection probability alone does not analyze the observed null result. Removed power-based exclusion without asserting that an observed-data interval must include +18 points; dated correction 4.
- [fixed] X12 (WRONG): Independently computed 1 - 0.025^(1/74) = 0.0486276162379895 for a hypothetical zero-event group of 74. This is a counterexample to the universal impossibility, not the actual data. Replaced it with conditional low detection probability; dated correction 5.
- [fixed] X13 (MINOR): The article qualifier was removed, but public/tv/lib/narration/batch-15.js:387 still says "heavy crackers". The original study compared habitual crackers with non-crackers, with no heavy-use subgroup. The shared narration needs "habitual crackers"; shared-file editing is outside this repair's scope. The directing instance must obtain a board note and replace PENDING with its issued id. The directing instance corrected that narration copy on 2026-10-14.
- [fixed] X14 (MINOR): Read the 2015 limitations and the 2018 acoustic model abstract. The sound source remained debated; attributed cavity formation to the observed event and the authors' interpretation.
- [fixed] X15 (MINOR): Read the 2015 Results: the event occurred within one 310 ms imaging frame. An upper bound alone does not establish the blink comparison. Removed that comparison without claiming the actual event was slower.
- [fixed] X16 (MINOR): The 2015 observation supports cavity appearance; its acoustic interpretation is not a settled exclusive mechanism. Qualified the prose, record row, markdown relation and research notes.
- [fixed] X17 (WRONG): Read the 2015 Conclusions: possible harms remain for investigation. Replaced the claimed absence of joint cost with the study's lack of a safety conclusion; dated correction 6.
- [fixed] X18 (MINOR): Same physical record-table statement as X21. Marked the January-to-January 36,526 calculation illustrative, matching the research disclosure and independent calendar calculation.
- [fixed] X20 (MINOR): The model gives about 80.0% power at +18 points and 65.4% at +15 at baseline 30%. Replaced catch/miss threshold wording with a conditional 80% probability and lower power for smaller effects.
- [fixed] X21 (MINOR): Duplicate of X18: the same table cell now labels the 36,526 calendar calculation illustrative. No exact historical dates were invented.
- [fixed] X22 (MINOR): powerClosed adds both normal tails; the displayed expression omitted one. Printed both tails with the approximation sign; the zero-effect sum is about 5%, not one tail's 2.5%.
- [fixed] X23 (MINOR): Changed exact to approximate and removed the promised identity with finite simulations. The formula and Bernoulli simulation differ by approximation and sampling error.
- [fixed] X24 (MINOR): Read the complete numerical verifier: six nonzero effects plus a null at baseline 30%, with assigned source inputs for historical observations. Replaced recomputes it all with the actual arithmetic and representative-point scope.
- [fixed] X25 (MINOR): The 2015 limitations and 2018 model leave the sound source debated. Changed settles the sound to a qualified interpretation while retaining the safety distinction.
- [fixed] X26 (MINOR): The verifier does not independently derive every historical count or validate every slider combination. Narrowed the footer, markdown and research statements to count arithmetic, assumptions and representative modeled points.
- [open note:81816d] X27 (MINOR): The manifest selects the standalone numerical verifier, which runs with Node; the browser verifier with the same basename needs Playwright and repository assets. The shared generator prints only the basename after browser-check prose. Triage explicitly re-leveled this ambiguity to MINOR and assigned it to existing board note 81816d. Repair replaced the declined disposition with that required open disposition; the shared placard is unchanged.
- [fixed] X28 (MINOR): The original observational study reports no increased arthritis association; it does not establish a universal causal guarantee. Qualified the metadata and research summary without asserting the opposite causal claim.
- [fixed] X29 (WRONG): The metadata promised effect exclusion, but powerClosed accepts hypothetical risks and computes detection probability. Changed description, Open Graph, Twitter and JSON-LD copies; dated correction 4. dateModified is 2026-10-14.
- [fixed] X30 (WRONG): The dek repeated the unsupported exclusion and universal small-effect limit. It now describes modeled detection with the original group sizes at user-selected risks; dated corrections 4 and 5.
- [fixed] X31 (MINOR): Re-read NIST 7.1.3: type-II error is a probability of failing to reject under a false null, not the magnitude of an effect. Corrected the related-page note.
- [fixed] X32 (WRONG): Re-read original p. 308 methods: 300 consecutive clinic patients were assessed by history and physical examination, not selected as arthritis cases and controls. Changed case-control to cross-sectional in markdown and verifier comments; dated correction 7.
- [fixed] X33 (WRONG): Duplicate of X32, the same markdown description. The consecutive cross-sectional assessment supports the replacement; dated correction 7.
- [fixed] X34 (WRONG): The markdown and numerical verifier inferred exclusion from hypothetical power. Both now give conditional detection probabilities, with no observed-data or causal exclusion; dated correction 4.
- [fixed] X35 (WRONG): verdictFor triggered exclusion solely when pw >= 0.80. Removed that claim while retaining the computed percentage and explaining possible misses; dated correction 4.
- [fixed] X36 (MINOR): At baseline 30% and +5 points the model detects about 13.2%, contradicting cannot see. The dynamic message now says usually misses under these assumptions.
- [fixed] X37 (WRONG): Executed the pinned inline-script reproduction: after an input change the display read 97% modeled and 67% simulated; an uninterrupted control read 97% for both. Added a generation counter invalidating queued batches on either slider or Reset. Browser regressions cover all three controls, restart before an old callback, and reduced motion; dated correction 8.

## Verification

Before repair, the fixer reported 27 numerical assertions and 22 browser assertions passed, including active-run cancellation for both sliders and Reset, stale callback isolation, reduced motion and phone/tablet/desktop overflow checks. The original X37 reproduction showed the stale-input defect before the fix. The fixer's claims gate exited 1 on 32 other layers' records and did not list this layer; its direct checkRecord validation returned no errors (91 checked, 55 confirmed, 36 wrong including MINOR). Repair results are recorded in research/assay-line/queue-1014c/the-hand-he-cracked-for-fifty-years/repair-report.md. The required PENDING placeholder is not a valid board id, so this record needs the issued id before the claims gate can pass.

## Evidence and limits

The original 1990 journal PDF was read at https://www.researchgate.net/profile/Mansour-Abdullah-Alshehri/post/What_are_the_reasons_of_knee_cracking_in_young_people_who_do_not_suffer_from_any_symptoms/attachment/59d6519679197b80779aa119/AS%3A508013192908800%401498131191426/download/Effect%2Bof%2Bhabitual%2Bknuckle%2Bcracking%2Bon%2Bhand%2Bfunction.pdf (pp. 308-309). The original Unger letter was read from https://osteopathe-fillinges.fr/wp-content/uploads/2016/11/etude_Dr_D_Unger_craquement_articulaire.pdf and the checker's extracted p. 950. Direct PMC1599734 and PMID22391267 readers were blocked; their previously fetched primary full text and Europe PMC abstract were re-read from the N2 source archive. The unchanged X19 row remains scoped to Unger's recorded comparison. X27 is open under existing shared note 81816d at the MINOR level explicitly assigned by triage. X13's shared narration copy awaits a board note from the directing instance. No STALE item was present, and moving dateModified for these corrections needs no Updated line.