A companion to The Twenty-Four Compared: the consolidated ledger of every audit run across the series, what each found, and what stands as doctrine.
Three mechanical detectors now sweep the series, with human adjudication of their output. The fact layer: every bolded event headline, dated claim, superlative, and agency construction verified; a ratio-direction sweep of every currency conversion; a chain-continuity test comparing each year's per-capita movement against its stated growth; a division test (per capita must equal GDP over population) wherever the columns permit; and, since August 2026, a closure-semantics detector with an internal-contradiction engine that tests every claim of resolution, permanence, or record against the volume's own printed series and ratings. The term layer: every acronym and italicized foreign term extracted with first-use context; three gloss rounds, the last exhaustive by construction. The name layer: every capitalized name-like token, filtered against the globally famous, checked for a role at first use.
Corrected on the record, each with an erratum in its volume: the USA's first-bank-card claim; India's 1969 deposit share; the UK's stale series count, 1910 duopoly framing, and vaccine claim; Turkey's cumulative bank-seizure figure; Germany's refugee-share phrasing and the untitled Spiegel affair; Pakistan's Siachen agency; Poland's transposed 1950 conversion ratios; Japan's nominal 1987 crossing, now recorded beside Saudi Arabia's chained one; and the population-base splices — Australia's (late 1950s onward), then, via the closing battery, Mexico's, Saudi Arabia's (hiding since 1960), Turkey's, Sweden's, and Indonesia's — every per-capita column rebuilt as GDP over population.
Triggered by a single sentence: Sweden's Palme entry read “unsolved for 34 years,” implying a solution in 2020 when the case had only been closed on a dead suspect. Three earlier fact passes had missed it because they tested figures, dates, and agency — not the shape of a claim. Two detectors were built for the class and run over all twenty-four volumes.
The closure-semantics detector flags resolution, permanence, and record language (solved, never again, for good, only, first ever, still the worst). The internal-contradiction engine tests each such claim against the numbers printed in the same volume — growth, inflation, unemployment, per-capita, and the ratings themselves. Sixty-eight claims were flagged and adjudicated one by one; sixteen were wrong, and each is corrected with an erratum in its volume:
Italy's 2007 per-capita summit was called “never since regained” beside its own higher 2025 and 2026 figures. India's 1979 trough was called “still the worst” beside its own deeper 2020. Argentina's 1995 unemployment record was called the highest ever measured beside its own 2002. Korea's fertility milestone was dated a year late — the threshold fell in 2018. Four “worst since” records (Canada 1982, Mexico 1995, Italy 2009, UK 2009) were unbounded and broken by 2020. Argentina's 1969 price-stability claim, Australia's 2019 growth trough, France's 2000 peak, and Italy's 1986 disinflation each overlooked a better year inside their own stated window. Two Mood lines claimed a “best-lived year” with equals in the same window. China's 2015 line read as though birth limits had ended rather than loosened; America's 1993 IBM layoff carried an unbounded “largest ever”; and one stray cross-volume phrase survived the standalone-language pass.
Fifty-two flagged claims were confirmed sound, including several the detectors misread — Bank Rate's genuine 1955 low, Sputnik V's registration, Poland's first Article 7 trigger, Gangnam Style's billion views. The lesson is recorded rather than papered over: a document can contain its own refutation, and until this pass nothing tested for it.
Five co-equal lenses — household (living conditions, read in development context — the absolute-dollar reading was trialed on India and overruled), contemporaneous consensus, world-relative (motion of position), direction (the year's own trajectory, fiscal included), statistics as cross-check — blended and rounded, with the split recorded wherever lenses diverge sharply. The famine floor-band: famine years live at 1–2; unrelieved mass death keeps the 1; averted or straddled years take the 2. The collapse-set: a 1 means the system itself came apart. The borrowed-boom ceiling caps at 6 what runs on someone else's balance sheet or eats its future, distinguishing borrowing for durable capital. Averages within ~0.15 are ties. Returns are credited in the years they arrive.
Full five-lens reviews of every year: Saudi Arabia (13 moves, the 1957–58 nadir-and-rescue swap), Iran (20 moves, the revolution's staircase), Mexico (16 moves in perfect symmetry, 1968 restored). Multi-round adversarial audits with reasoning published: the USSR (five rounds — famine floor upheld, weighting corrected, evidence run, reconversion symmetry), Pakistan (the parity paradox, upward), Brazil (borrowed-time), Argentina (the whipsaw defended by distribution). The closing extremes audit: every 7, 8, and the lone 9 (Japan 1964) passed the breadth test; six floors set before the collapse-set doctrine moved to 2 — India's four (the republic keeps no 1 at all), Spain's 2020, Turkey's 2021. India received the fourth full grid — the household recalibration, fifty-three moves to 3.73 — which the author then overruled as doctrine: an absolute ladder handicaps poor countries and credits nothing for stability, policy capacity, long-term progress, or investor returns. India's original ratings stand restored (four 1s included), refined by two August rulings — the peak-density ruling keeping one 8, 2007's crest, with the others 7s, and the plan-spring ruling re-landing 1953–56 as one 5 and three 4s (4.52, eighth) — the experiment preserved in that volume's trail, and China leaves the queue. Remaining volumes carry targeted audits and stand open to the same full-grid treatment on request.
USA 5.26 · Taiwan 5.24 · Japan 5.01 · China 4.99 · Australia 4.79 · Sweden 4.67 · Canada 4.54 · India 4.52 · South Korea 4.47 · Germany 4.43 · France 4.12 · UK 3.99 · Italy 3.90 · Saudi Arabia 3.88 · Spain 3.78 · Poland 3.69 · Brazil 3.65 · Russia/USSR 3.64 · Mexico 3.62 · Indonesia 3.58 · Turkey 3.47 · Iran 3.36 · Argentina 3.30 · Pakistan 3.19.
— Compiled July 2026, alongside the twenty-four volumes it audits.