Tesseract read every page independently as a witness. It is not the ground
truth — it is worse than the AI engine at almost everything. It has one property the
AI engine lacks: it does not invent words. Each finding below is a place the two readings
diverge, shown with the crop of the scan it came from.
Blank-page fabrication — The witness engine found effectively no ink on this page, yet the AI engine emitted running text. This is the classic silent-fabrication signature.
Unsupported text — The AI engine emitted words that appear nowhere in the witness reading of this page — not merely in a different order, but absent entirely.
Dropped text — The witness engine read words that appear nowhere in the AI engine's output. Usually a skipped line, column, or caption.
Disagreement — Both engines read text here but disagree on the words. Often ordinary OCR noise on hard glyphs; occasionally a rewritten passage.
Wholesale disagreement — The two readings diverge heavily in both directions at once. That pattern usually means the witness could not handle the page (severe skew, noise, or degradation) — but it is also what a wholesale rewrite would look like, so the page deserves a human glance. No itemized accusations are made.
Unverifiable page — There is ink on this page but the witness engine could not read it, so the AI engine's output here can be neither supported nor disputed. Not an accusation — a gap in coverage.
Text merely moved on the page — reordered columns, floated captions, table
cells read in a different sequence — is not reported: a word counts as unsupported
only when it is absent from the witness reading of the whole page. Words below
40% witness confidence are excluded from the comparison entirely.
Pages where the witness itself struggled are labelled, and their findings are hedged.
Flagged pages
Page 21
divergence 37.9%153 AI words104 witness wordsink 2.13%
Unsupported text1.0048 tokens
Scan — crop at 192,817
AI engine read
conditions of considerable hardship is evident throughout the resident observers. The summary telegrams tension in the commission striping the subply vessel tourists off the shorts in each season and only was the subply vessel tourists off the shorts in each season and only was the subply vessel tourists off the shorts in each season and only
Witness read
returns.
Unsupported text0.7510 tokens
Scan — crop at 193,345
AI engine read
the coastal survey achieved the greater part of the summer of the summer of the coastal stations and particularly the fourteen proposed stations, and tidal series of
Witness read
its stated objectives. Positions were fixed for eleven of
Dropped text0.528 tokens
Scan — crop at 193,345
AI engine read
(nothing)
Witness read
its stated objectives. Positions were fixed for eleven
Page 1
divergence 0.0%85 AI words124 witness wordsink 2.55%
Dropped text0.9034 tokens
Scan — crop at 193,766
AI engine read
(nothing)
Witness read
Readings from the third party have not survived. The field notebooks were lost when the supply vessel foundered off the shoals in early September, and only the summary telegrams remain in the commission archive.
Dropped text0.405 tokens
Scan — crop at 193,204
AI engine read
(nothing)
Witness read
The Survey of Coastal Stations
Page 9
divergence 0.0%78 AI words115 witness wordsink 1.52%
Dropped text0.9033 tokens
Scan — crop at 192,719
AI engine read
(nothing)
Witness read
A comparison of the winter series with the corresponding series from the preceding year shows agreement within two hundredths of a foot for the mean water level, which the commission regarded as satisfactory.
Dropped text0.364 tokens
Scan — crop at 193,204
AI engine read
(nothing)
Witness read
Observations of the Tide
Page 10
divergence 0.0%78 AI words115 witness wordsink 0.29%
Dropped text0.9033 tokens
Scan — crop at 192,719
AI engine read
(nothing)
Witness read
A comparison of the winter series with the corresponding series from the preceding year shows agreement within two hundredths of a foot for the mean water level, which the commission regarded as satisfactory.
Dropped text0.364 tokens
Scan — crop at 193,204
AI engine read
(nothing)
Witness read
Observations of the Tide
Page 11
divergence 0.0%0 AI words115 witness wordsink 0.00%
Dropped text0.90115 tokens
Scan — crop at 192,204
AI engine read
(nothing)
Witness read
Observations of the Tide Tidal observations at Station Marlow were continued through the winter without interruption. The gauge was of the float pattern, housed in a timber well sunk four feet below the lowest recorded water. Readings were taken hourly by the resident observer and transcribed each evening. The self-registering apparatus failed twice during December, on both occasions owing to the freezing of the float chamber. Manual readings were substituted for the affected intervals and are marked accordingly in the published tables. A comparison of the winter series with the corresponding series from the preceding year shows agreement within two hundredths of a foot for the mean water level, which the commission regarded as satisfactory.
Page 16
divergence 0.0%85 AI words124 witness wordsink 2.64%
Dropped text0.9034 tokens
Scan — crop at 192,766
AI engine read
(nothing)
Witness read
Readings from the third party have not survived. The field notebooks were lost when the supply vessel foundered off the shoals in early September, and only the summary telegrams remain in the commission archive.
Dropped text0.405 tokens
Scan — crop at 192,204
AI engine read
(nothing)
Witness read
The Survey of Coastal Stations
Page 17
divergence 0.0%85 AI words124 witness wordsink 2.73%
Dropped text0.9034 tokens
Scan — crop at 192,766
AI engine read
(nothing)
Witness read
Readings from the third party have not survived. The field notebooks were lost when the supply vessel foundered off the shoals in early September, and only the summary telegrams remain in the commission archive.
Dropped text0.405 tokens
Scan — crop at 192,204
AI engine read
(nothing)
Witness read
The Survey of Coastal Stations
Page 18
divergence 0.0%78 AI words115 witness wordsink 2.42%
Dropped text0.9033 tokens
Scan — crop at 181,709
AI engine read
(nothing)
Witness read
A comparison of the winter series with the corresponding series from the preceding year shows agreement within two hundredths of a foot for the mean water level, which the commission regarded as satisfactory.
Dropped text0.364 tokens
Scan — crop at 163,217
AI engine read
(nothing)
Witness read
Observations of the Tide
Page 22
divergence 2.4%85 AI words121 witness wordsink 1.39%
Dropped text0.9033 tokens
Scan — crop at 184,757
AI engine read
(nothing)
Witness read
Readings from the third party have not survived. The field notebooks were lost when the supply vessel foundered off the shoals in early September, and only the telegrams remain in the commission archive.
Dropped text0.405 tokens
Scan — crop at 169,213
AI engine read
(nothing)
Witness read
The Survey of Coastal Stations
Page 19
divergence 24.4%78 AI words88 witness wordsink 2.43%excluded from gate
Wholesale disagreement0.5048 tokens
Scan — full page
AI engine read
continued winter housed in timber recorded taken hourly by evening. The self-registering readings were intervals accordingly in published tables.
Witness read
e Tide of th nter ed wi ng continu ded Readin: d observatio recor’ s istering W self-reg M interv als ace preceding year shows a water level, which
19 of 78 AI words lack witness support and 29 of 88 witness words lack AI support, and 48% of the unmatched witness words are short shreds — the witness likely could not handle this page. Review the page image rather than trusting either reading.
Page 23
divergence 43.6%78 AI words77 witness wordsink 0.00%excluded from gate
Wholesale disagreement0.5067 tokens
Scan — full page
AI engine read
at were continued through without interruption. gange pattern, in lowest recorded taken observer transcribed self-registering failed December, on occasions owing to float chamber. readings were substituted intervals and are marked accordingly in published tables.
Witness read
Tide of ma ine oto bet oat ured ane im of with the sence from the Preceding year shows two hundredths of a foot for the meat water level, which Mie regarded os
34 of 78 AI words lack witness support and 33 of 77 witness words lack AI support, and 55% of the unmatched witness words are short shreds — the witness likely could not handle this page. Review the page image rather than trusting either reading.
Page 15
divergence 0.0%104 AI words0 witness wordsink 2.50%witness quality: blindexcluded from gate
Unverifiable page0.15104 tokens
Scan — full page
AI engine read
Conclusions of the Commission The commission concludes that the coastal survey achieved the greater part of its stated objectives. Positions were fixed for eleven of the fourteen proposed stations, and tidal series of at least six months were obtained at seven. The loss of the third party's notebooks is regretted. The commission recommends that duplicate records be transmitted by post …
Witness read
(the witness read nothing here)
Ink coverage is 2.50%. The witness read only 0 usable words against the AI engine's 104. Nothing on this page was verified.
