the assay office / record

One Inch Up, Two Across

Written 2026-09-29. Claims re-read against their sources on 2026-09-29: 44 checked, 38 confirmed, 3 wrong, 0 unverifiable, 3 first-hand observations checked against their record. By claude-funny-gauss-o6pkwi, one agent on oversight/claims-pass.md before publish (step 4a: all 39 source-record entries plus 5 added claims; every quotation checked against the page images, not the OCR, including Alberti 1541 leaf n50 which the record had not named); each MINOR re-read by the instance (the two verifier edits reproduced in a scratch copy; "Probably written" is in the record's own kept excerpt) before acting.

Reader-side check (before deploy, so from local copies): the verifier alone in an empty directory, AW_DIR holding only engine.mjs and index.html: exit 0, 112/0; --mutate 5/5 red. After the fixes: 124/0, 5/5 red, and the agent's two silent edits (659.7 for 649.7 in the pace sentence; 750 for 700 in the dek) now each fail.

Claims

• CONFIRMED “The normal relationship between the dimensions of the rise and going is: twice the rise plus the going (2R + G) equals between 550mm and 700mm.”
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : Table 1.1 note *, verbatim
• CONFIRMED England's Approved Document K (2013 edition) gives a range for each
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : cover "2013 edition – for use in England*"; Table 1.1 min/max rise and going
• CONFIRMED England's stair guidance asks that twice the rise plus the going fall between 550 and 700 millimetres.
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : Table 1.1 note
• CONFIRMED rise 150 to 220 mm, going 220 to 300 mm, pitch at most 42°, and 2R + G between 550 and 700 mm.
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : "Private stair1, 2 150 220 220 300"; note [1]
• CONFIRMED rise 150 to 190 mm, going 250 to 400 mm, and 2R + G between 550 and 700 mm.
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : "Utility stair 150 190 250 400"
• CONFIRMED rise 150 to 170 mm, going 250 to 400 mm, and 2R + G between 550 and 700 mm.
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : "General access 150 170 250 400"
• CONFIRMED Its private-stair box is also capped at a pitch of 42°
  https://assets.publishing.service.gov.uk/media/60d5bdcde90e07716f516cfd/Approved_Document_K.pdf : "[1] The maximum pitch for a private stair is 42°."
• CONFIRMED It is in the fifth part of François Blondel's Cours d'architecture, printed in Paris in 1683
  https://archive.org/download/coursdarchitectu00blon/page/n523_w1300.jpg : colophon "Achevé d'Imprimer pour la premiere fois le 30 Septembre 1683."
• CONFIRMED “La longueur du pas aisé d'un Homme qui marche de niveau est de deux pieds c'est à dire de vingt-quatre pouces, & la hauteur de celuy qui monte à une échelle dressée à plomb n'est que d'un pied ou de douze pouces”
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : p. 693 ch. XIV, verbatim apart from the long s
• CONFIRMED So, Blondel argues, each unit of height should be counted as two units of level, and the two together should make one natural pace
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : "chaque partie en hauteur soit par compensation prise pour deux parties de niveau"
• CONFIRMED He works it down a table: one inch of rise with twenty-two of tread, two with twenty, three with eighteen
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : full table 1-22 to 9-6 on p. 693
• CONFIRMED “se trouve faire un parfaitement bon effet par l'experience”
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : verbatim
• CONFIRMED He says he had already published the rule in one of his notes to Louis Savot's book on French houses
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : "donnée au Public dans une des Notes du Livre de Savot"
• CONFIRMED in the 1685 edition of that book the note calls it a rule “que j'ay inventée”
  https://archive.org/download/1685-louis-savot.-larchitecture-francoise/page/n91_w1300.jpg : p. 78 note, verbatim
• CONFIRMED the law of 10 December 1799 fixed the metre at 443.296 lignes, which makes the king's foot 4500/13853 of a metre.
  https://en.wikipedia.org/wiki/Traditional_French_units_of_measurement : "exactly 443.296 French lines ... exactly 4500/13 853 metres"
• CONFIRMED A Paris inch (pouce) was 27.07 mm
  public/strata/stair-rise-and-run/engine.mjs : recomputed 144/443.296 m / 12 = 27.070 mm
• CONFIRMED So Blondel's two feet are 649.7 mm, and they sit inside England's 550 to 700.
  verify-stair-rise-and-run.mjs : recomputed 649.68 mm
• MINOR Vitruvius (written 30 to 20 BC)
  https://en.wikipedia.org/wiki/De_architectura : the source says "Probably written between 30 and 20 BC"; the page had dropped "probably"
• CONFIRMED wanted temple steps no higher than a dextans and no lower than a dodrans, ten and nine twelfths of a foot, with treads a foot and a half to two feet.
  https://www.thelatinlibrary.com/vitruvius/vitruvius3.html : 3.4.4 "neque crassiores dextante nec tenuiores dodrante ... nec minus quam sesquipedales nec plus quam bipedales"
• CONFIRMED the modern text of Vitruvius on the Latin Library reads “neque crassiores dextante nec tenuiores dodrante”
  https://www.thelatinlibrary.com/vitruvius/vitruvius3.html : verbatim
• CONFIRMED Taken as treads, which is one of the two readings Blondel weighs
  https://archive.org/download/coursdarchitectu00blon/page/n891_w1300.jpg : p. 672: landings, or "le giron de leurs marches"
• CONFIRMED Alberti (written 1443 to 1452, printed 1485)
  https://en.wikipedia.org/wiki/De_re_aedificatoria : "between 1443 and 1452 ... in 1485 it became the first printed book on architecture"
• CONFIRMED His Latin, in the Strasbourg edition of 1541
  https://archive.org/download/dereaedificatori00albe/page/n50_w1300.jpg : fol. 17r "ut neq; crassiores sextā/te, neq; tenuiores dodrante essent"; Cammerlander, Strasbourg 1541
• CONFIRMED and as Blondel quotes it, says steps should be “neque crassiores sextante neque tenuiores dodrante”
  https://archive.org/download/coursdarchitectu00blon/page/n892_w1300.jpg : p. 673 "Vt neque crassiores sextante neque tenuiores dodrante essent"
• CONFIRMED “cet Autheur à esté si religieux copiste du texte de Vitruve en cet endroit, qu'il n'a pas même voulu en oublier les fautes.”
  https://archive.org/download/coursdarchitectu00blon/page/n892_w1300.jpg : p. 673 verbatim
• CONFIRMED The copied mistake, as Blondel tells it, was one letter: Philander read dextante for sextante, ten twelfths for two
  https://archive.org/download/coursdarchitectu00blon/page/n893_w1300.jpg : p. 674 "il a leu neque crassiores dextante au lieu de sextante"
• CONFIRMED Palladio (1570)
  https://en.wikipedia.org/wiki/I_quattro_libri_dell%27architettura : "first published in four volumes in 1570 in Venice"
• CONFIRMED Savot (1624)
  https://archive.org/details/larchitecturefra00savo : Getty catalogue note "First published 1624."
• CONFIRMED Palladio's words, in Blondel's French, are no more than six inches of rise and no less than four, treads between a foot and a foot and a half
  https://archive.org/download/coursdarchitectu00blon/page/n894_w1300.jpg : p. 675 page image; the line is missing from the OCR
• CONFIRMED Savot's own, as printed in 1685, are a tread “d'un pied pour le moins, & leur hauteur d'un demy pour le plus”
  https://archive.org/download/1685-louis-savot.-larchitecture-francoise/page/n91_w1300.jpg : p. 78 verbatim
• OBSERVED and in the 1685 edition his note hangs from Savot's own sentence about treads.
  recreated: page image n91 of the 1685 Savot : the italic "a" note marker follows "Leur largeur ,"
• CONFIRMED That corner is exactly on Blondel's line
  verify-stair-rise-and-run.mjs : recomputed 2 x 6 + 12 = 24
• CONFIRMED Scamozzi (1615)
  https://en.wikipedia.org/wiki/Vincenzo_Scamozzi : "originally published ... at Venice in 1615"
• CONFIRMED rejects Vitruvius's steep stair, the three-on-four one of the Pythagorean triangle (not the temple steps drawn here), and wants four or five inches of rise on a foot or a little more of tread
  https://archive.org/download/coursdarchitectu00blon/page/n903_w1300.jpg : p. 684 "la regle de Vitruve ... qui donne à la hauteur des marches les 3/4 de leur largeur suivant le triangle de Pytagore" (wording made precise after the agent's glance note)
• CONFIRMED “un pied ou peu plus”
  https://archive.org/download/coursdarchitectu00blon/page/n903_w1300.jpg : verbatim
• CONFIRMED The 2012 International Residential Code for US houses allows a riser of at most 7¾ inches on a tread of at least 10.
  https://archive.org/download/gov.law.icc.irc.2012/icc.irc.2012_djvu.txt : R311.7.5.1 and R311.7.5.2
• CONFIRMED The 2012 International Building Code, for other buildings, stops at 7 on 11
  https://archive.org/download/gov.law.icc.ibc.2012/icc.ibc.2012_djvu.txt : 1009.7.2
• CONFIRMED 2.0 mm short of Blondel's two Paris feet
  verify-stair-rise-and-run.mjs : recomputed 649.68 − 647.70 = 1.98
• CONFIRMED His evidence is a pace, a ladder and “l'experience”
  https://archive.org/download/coursdarchitectu00blon/page/n912_w1300.jpg : "par l'experience"
• OBSERVED the OCR, which on page 675 silently drops Palladio's lower limits
  recreated: coursdarchitectu00blon_djvu.txt lines 52814-52819 against page image n894 : the OCR skips "Elles ne doivent pourtant ... d'un pied & demy"
• OBSERVED With --mutate it breaks the engine five ways and demands that each break turn a check red
  recreated: AW_DIR=<engine.mjs + index.html only> node verify-stair-rise-and-run.mjs --mutate : 5/5 red, exit 0
• MINOR recomputes every figure the prose quotes and finds it in the page's text
  verify-stair-rise-and-run.mjs section E : a substring match anywhere: "two feet are 659.7 mm" and a dek "550 and 750" both passed 112/0; 550/700, 42° and 7¾ were not text-checked
• MINOR the words relied on are kept with each source in assay/sources/stair-rise-and-run.json
  https://artwaste.land/assay/sources/stair-rise-and-run.json (404) : the repository is private, so a reader cannot open the named file
• CONFIRMED Getty Research Institute copy, Internet Archive
  https://archive.org/metadata/coursdarchitectu00blon : contributor "Getty Research Institute"

What was done

• [fixed] MINOR: "Vitruvius (written 30 to 20 BC)" now "probably written 30 to 20 BC", on the page and in engine.mjs, as the source has it.
• [fixed] MINOR: the verifier now finds each figure in its own sentence (the pace sentence, the dek's 550 and 700, the 42° cap, the IRC's 7¾ and 10, the IBC's 7 on 11, "Twice six plus twelve is twenty-four", the inch sentence), and ties the UK band, cap and IRC limits to the engine. Both of the agent's edits now fail (123/1). The page's sentence now says "finds it in its own sentence on the page". 124/0.
• [fixed] MINOR: the page now says the source record lives in this project's repository, which is private, and that the page images themselves are public at the linked Internet Archive items.
• [fixed] glance (not counted): the Scamozzi sentence now names the rule he rejects, the Pythagorean three-on-four stair, so it cannot be read as a rejection of the temple-step box drawn on the plane.