MIGNE.APP

Cruces

In visionem Ezechielis

Richard of Saint Victor — read the work →

These are the edition's working notes, published as they were written. They record, work by work, where Migne's plate is defective or the reading uncertain: the printed form, what the English does with it, and why. They are not a groomed apparatus. Where a note argues with itself, withdraws an earlier decision, or leaves a question open, that is the record of how the reading was actually reached, and it is left standing — a conjecture that is visible can be checked, and one that has been tidied away cannot.

The English never silently repairs the page. A construable but wrong reading is rendered literally and the conjecture is recorded here, never applied to the text. How the English is produced and checked →

Migne, PL 196, cols 0527A–0545A (Prologue and Capp. I–IV). Printed readings are rendered literally in the translation per translation-style.md Pattern 7; conjectures live here only.

Carried type (Pattern 10 + 12 — [sic: …] in the English)

Readings rendered literally though they will not construe

Scripture divergences from the received text

Pattern 9 items (no English exponent — logged, English unchanged)

Punctuation carried per Pattern 8

Negation fidelity

Every printed non / nec / neque / nisi / nunquam / nullus / nihil / haud in cols 0527A–0545A is carried into the English, and none is added. Two paragraphs are negation-dense and were read against the Latin word by word: col 0529B ("Revertentes ergo non revertebantur … non retro vertebantur, quia non convertebantur" — four negations, all present) and col 0541A–B ("neminem esse tam improbum … quem … contraire non pudeat, qui … reprehendere non erubescat … si … non subsannaverit" — a coherent double-negative period, not a defect). No confirmed dropped- or intrusive-negative site was found in chunks 0000–0004. The nearest approach is the haud dubium ante quod of col 0537B above, where the negation itself is intact and the corruption lies in the adjacent word.

Numerals — all checked, all internally consistent

Logged as verified rather than as defects, since this volume has a documented history of corrupt measurements:

Translation-convention notes (not cruces)

Richard of St Victor, PL 196, cols 0545B–0559C. Rendered under translation-style.md Pattern 7 / 7a: the printed reading stands in the English, the conjecture lives here.

0546A (chunk 0005) — utrumque for utrimque in the Ezekiel lemma

Migne prints, quoting Ezek. 40:12: Et mensus est marginem ante thalamos cubiti unius, et cubitus unus finis utrumque. The Vulgate has utrimque, "on both sides," and Richard's own argument requires it (he immediately glosses the verse as meaning the one-cubit margin lay "not only in front, but also on either side"). Decisive internal evidence: at 0553B the same lemma is re-quoted and there Migne prints utrinque. So the plate contradicts itself between two printings of one verse, twelve columns apart.

Rendered as printed ("and one cubit the boundary, both"), with a [var: Vulg. utrimque, "on both sides"] beside it. No [sic: …] — utrumque is a real Latin word, merely the wrong one (Pattern 12's boundary), so what stands on the page is our English, not his type.

0549A (chunk 0006) — liber where the syntax requires libet

…et quae de thalamis seu liminibus hoc loco dicenda videbantur ordine prosecuti sumus, liber hoc loco ad majorem dictorum dicendorumque evidentiam … colligere, ut… The infinitive colligere has no governing verb; libet ("it pleases [me] … to gather") supplies one exactly and is a one-letter difference. Rendered with the printed word taken as the adjective it is — "I am free in this place … to gather" — which construes with the infinitive and keeps liber visible. No marker: the word is real and translatable.

0549A / 0550A / 0549 (chunk 0006) — anchor sequence, NOT reported as a defect

Recorded only so a later pass does not re-open it. The run reads [0549A] … [0550A] … [0549], i.e. out of order and including a bare band-less anchor. Per the standing ruling this is normal positional transcription; no text breaks mid-word or mid-construction across any of them. Reproduced verbatim, no [ed: …], no patch. (The bare anchors [0549] here and elsewhere in this work are likewise carried as printed.)

0551C (chunk 0007) — ispius for ipsius

…cujus limen prius, et postmodum ispius vestibuli mensuram describit… Metathesis; not a word. Carried into the English untranslated in italics under Patterns 7/10 and marked [sic: ispius]. Conjecture: ipsius, certain.

0552B (chunk 0007) — quinquagina for quinquaginta

…duo collateralia triplicis porticus vestibula hinc inde quinquagina cubitos longitudinis… Not a word. Carried untranslated in italics, marked [sic: quinquagina]. Conjecture: quinquaginta, certain — fifty is the figure the whole chapter runs on, and Migne prints quinquaginta correctly at 0553D and 0555C.

0553A–B (chunk 0007) — ut in superioribus where the argument demands non ut

This is the negation site. Migne prints:

Ecce limen hoc ut superiora portae limen dicitur, et tamen hic, ut in superioribus, nihil determinationis adjungitur, et ex eo ipso innuitur quod vere et proprie portae dicatur.

Richard has spent the two preceding columns proving the opposite: at 0551B and again at 0551D–0552A he insists that in the earlier cases a determination (juxta vestibulum portae intrinsecus) was appended, and that its presence is precisely what shows those thresholds were not properly the gate's. Here he argues that nothing is appended, and that the absence is what proves this threshold IS properly the gate's. The clause as printed therefore asserts the reverse of the sentence it sits inside, and of the chapter around it. Conjecture: non ut in superioribus (or ut non in superioribus) — a dropped non, the PL 196 defect class.

Rendered exactly as printed ("and yet here, as in the foregoing, nothing of determination is added"), so the English contradicts itself where the plate does. No repair, no marker — there is nothing carried, only a printed word doing the wrong work.

0553D (chunk 0007) — vestibularum for vestibulorum

Nullum ergo hujusmodi vestibularum limen X cubitorum capere potuit. Feminine genitive plural of a neuter noun. Pattern 9: a declension/gender mismatch has no English exponent — there is no emendation to make or undo. Translated normally ("of porches of this kind"); no marker, no churn on a later sweep.

0556D (chunk 0008) — sentence ends with no terminal stop

…Nam interior porticus convertebat faciem suam ad exteriora, et exterior nihilominus convertebat suam ad interiora — the plate simply stops; the next paragraph opens Notandum sane…. Under Pattern 8 no mark is supplied: the English paragraph ends unpunctuated as the Latin does.

0559A (chunk 0009) — quadraginta cubitos where the work elsewhere prints sexaginta

The measurement defect. Migne prints, in Richard's back-reference to the first description: de longitudine, et fecit frontes per quadraginta cubitos. Every other occurrence of that lemma gives sixty — 0554C (Et fecit frontes per sexaginta cubitos), 0555C (et fecit frontes per LX cubitos), and Richard's own arithmetic at 0554A–B (XXV + XXV + X = LX). Forty also breaks the very inference he draws from it three lines later: two porches of twenty-five plus a third of ten fill sixty, not forty.

Rendered as printed ("and he made fronts of forty cubits"), so the figure and the arithmetic disagree on the page, which is the evidence. Conjecture: sexaginta, effectively certain.

0559B (chunk 0009) — mid-sentence interpunct

…quod quidem legitur et apud secundam portam · Et vestibulum, inquit… Migne sets a raised point where a full stop or comma would be expected. Reproduced verbatim (Pattern 8); not normalized.

0559C (chunk 0009) — longitudo where the sense requires latitudo

Notandum quod porticum per portam intelligit, et quod pavimenti solam longitudinem describit: nam ejus longitudo ex alio melius patebit. The second noun must be the dimension NOT yet described; having just said that only the length is given, "for its length will appear better from elsewhere" is self-cancelling. The parallel passage at 0558B–C settles it: Notandum quod solam longitudinem centum cubitorum describit. Nam latitudo pavimenti quanta fuerit … melius ex sequentibus patebit. Conjecture: latitudo.

Rendered as printed ("only the length … for its length will appear better"). No marker.

Conventions applied, beyond the work-wide list

Chunks found otherwise clean

0005 (cols 0545B–0546D) carries no defect but the utrumque lemma above; its arithmetic (3 × 8 = 24, + 5 + 5 = 34; 34 > 25 breadth, < 50 length) is internally sound. 0008 (cols 0554C–0557B) is clean apart from the missing terminal stop at 0556D — all its figures (25 + 25 + 10 = 60; 15 + 15 = 30 chambers; 100 cubits gate to gate) agree with the rest of the work.

Blind polarity read, 2026-07-30 (unit 11553-u2) — one item left UNRESOLVED

0551A — Quam autem vir ille … metitur … Scriptura innuit quod …: our clause binding may invert Richard's inference. NOT changed; recorded for a ruling.

Migne prints Quam autem vir ille qui portam ingreditur hoc limen primo loco, et ante thalamos metitur, patenter, ut arbitror, Scriptura innuit quod in primo praedictorum VII vestibulorum invenitur.

Our English makes the measuring-order the thing Scripture hints, and the location the reason for it ("…Scripture hints that that man … measures this threshold in the first place, and before the little chambers, in that it is found in the first of the aforesaid VII porches").

The reading the argument seems to want is the reverse: because he measures it first and before the chambers, Scripture plainly hints that it is found in the first porch — i.e. the quod clause is the content of innuit, not a causal tail. Two things support that: the location is what this stretch is out to establish, and Richard states it as settled at 0551C (Primo de limine dicit quod erat in primo vestibulo).

Left as printed, deliberately. Quam is unaccounted for on either parse, and cruces-0014.md records a comparable unresolvable Quam autem … at 0571D where the English supplies a bracketed word — so this is likely the same defective construction rather than a mis-parse of a sound one. Re-binding the clause would assert a reading of a sentence that does not fully construe, which is the emendation Pattern 7 forbids. No claim is reversed to its opposite — both propositions are Richard's — so this is not a polarity defect; it inverts the direction of an inference. Flagged for Wilson.

0536D — oportet totum hoc et australem plagam habere vestibulum (chunk 0002, found by unit 11553-u1). Our English makes totum hoc a co-subject rather than the object of habere. The resulting claim is the same (the south side has all of this too), so it is not a polarity defect and was not changed — recorded because it is a genuine mis-parse and was in no cruces file before.

Richard of St Victor, PL 196, cols 0559C–0570D. Rendered under translation-style.md Pattern 7 / 7a: the printed reading stands in the English, the conjecture lives here.

Written as an audit of a translation stint that finished the four English chunks and stopped before writing its apparatus. All four Latin chunks were read against their English twins sentence by sentence; what was found, what was changed, and what was deliberately left alone are all recorded below, including the candidates examined and rejected.

Carried type (Patterns 7 / 10 / 12 — [sic: …] in the English)

0559D–0560A (chunk 0010) — habenta for habentia

Et mensus est fenestras ejus, et vestibulum ejus in circuitu habenta L cubitos longitudinis… Not a word. Decisive internal evidence: Migne prints the correct habentia two lines earlier in the very sentence that sets up this re-quotation (in sequentibus, habentia, vel aliquid ejusmodi subaudiendum est), so the plate supplies its own conjecture. Carried untranslated in italics, [sic: habenta]. Conjecture: habentia, certain.

0563A (chunk 0011) — demidii for dimidii

…longitudine cubiti unius et demidii, et latitudine cubiti unius et dimidii… The same word is printed both ways inside one lemma, four words apart. Carried untranslated in italics, [sic: demidii].

Pattern 9 was considered and rejected here. demidius could be argued as an e/i orthographic variant and so translated normally; but the plate's own correct dimidii in the same breath makes this a compositor's slip on this plate, not a period convention — which is precisely the boundary Pattern 9 draws ("reserve carry-through for forms unique to this plate"). Carried.

0564C (chunk 0011) — in fallor for ni fallor

…ipsumque, in fallor, in his verbis describitur… The parenthesis is Richard's standing ni fallor, "unless I am mistaken"; in fallor will not construe at all (in + a first-person passive verb). Per Pattern 10 the whole run is carried, including the real word in, since the defect is the pair and not either token: [sic: in fallor]. Conjecture: ni fallor, certain.

0565A (chunk 0011) — contracuntem for contra euntem

…sicut australem viam illa quae ab Occidente erant contracuntem habebant. Run-together type of the kind Pattern 10 names, with a letter lost at the join. Carried untranslated in italics, [sic: contracuntem]. Conjecture: contra euntem ("running over against them"), which is what the immediately preceding clause has been arguing with contra three times (non contra viam Aquilonis … et non contra ea).

0566A (chunk 0012) — totitem for totidem

…ejusdemque longitudinis atrium includebant, totitemque gradibus assurgebant… Not a word. Carried untranslated in italics, [sic: totitem]. Conjecture: totidem, certain — Richard's point is that the eastern porticoes rose by as many steps as the others, the eight steps of 0560C.

0566B (chunk 0012) — Occiente for Occidente

…quid tamen vel quantum de illo medio centenario ab Occiente … tumor montis detraxerit… A dropped d; not a word. Carried untranslated in italics, [sic: Occiente]. Conjecture: Occidente, certain — the plate prints Occidente correctly twice within twenty lines (0566B versus Occidentem, 0568C ab Occidente in Orientem).

0566D (chunk 0013) — Scriturae for Scripturae

Libet hic, sicut et supra, … Scriturae testimonia in unum colligere… A dropped p; not a word. Carried untranslated in italics, [sic: Scriturae]. Conjecture: Scripturae, certain — the word is printed correctly some forty times in these four chunks.

0569D (chunk 0013) — darietemque for parietemque

Post quem parietem si porticus aquilonaris domus incoepit, darietemque post parietem habuit… Not a word; d/p type substitution. Carried untranslated in italics, [sic: darietemque]. Conjecture: parietemque, certain on the sentence's own evidence — parietem stands correctly twice in the same clause, before and after it.

The numeral that propagates — 0560B (chunk 0010)

This is the finding the range was opened for. cruces-0005.md records that at 0559A — three lines before this chunk begins — Migne prints et fecit frontes per quadraginta cubitos where the same lemma is printed sexaginta at 0554C and LX at 0555C, and where Richard's own arithmetic (XXV + XXV + X = LX, 0554A–B) requires sixty.

The corrupt figure recurs at 0560B, inside Richard's own prose:

Per istud vestibulum viginti quinque cubitorum longitudinis facile probari poterit, quod mensura illa frontium quadraginta cubitorum intra parietes sumpta fuerit.

The inference is that the twenty-five-cubit porch proves the fronts-measure to be an interior measure. It works only with sixty: the front is composed of two porches of twenty-five and a middle span of ten, so 25 + 25 + 10 = 60, and the twenty-five being an inside figure makes the sixty an inside figure. With forty the composition is impossible — two porches of twenty-five will not fit inside forty at all, and the sentence destroys the argument it is offered as proving.

So the defect is not isolated to 0559A: it appears twice within two columns, once in a lemma and once in the commentary that reasons from it. That is worth recording, because a single occurrence looks like a compositor's slip and two consecutive ones raise the possibility of a corrupt exemplar behind this stint of the plate.

Rendered as printed — "that measure of the fronts of forty cubits was taken within the walls" — so that the figure and the arithmetic disagree on the page. No marker: quadraginta is a real word doing the wrong work (Pattern 12's boundary). Conjecture: sexaginta, effectively certain.

Readings rendered literally though they will not construe

0565C (chunk 0012) — utrum utrobique

…et idcirco duobus porticibus interioris atrii hinc inde utrum utrobique simile spatium in coquinarum officium relinquimus. Two adjacent forms of uter with nothing to govern utrum, in a sentence that already carries hinc inde. The sense wanted is "to each of the two, on both sides." Rendered as closely as the printed words allow — "on this side and on that, to either on both sides" — which is redundant English because the Latin is redundant Latin. Both words are real, so no [sic: …]. No confident conjecture is offered (utrique utrobique is the likeliest, but the 9604 boundary applies: the run is corrupt past secure restoration).

0566D (chunk 0013) — the chapter head, De quantitate vel aedificiorum

CAPUT XIII. De quantitate vel aedificiorum quae templo adjacebant. The vel has no second member: a noun has dropped out after it (vel qualitate is the obvious supplement, and Richard's chapter does treat both size and disposition). Rendered as printed — "On the size or of the buildings which lay adjacent to the temple" — which is broken English because the head is broken Latin. No marker; the words are real. This is a head, so it is also a sacred marker: the ## line stands 1:1 with the Latin twin and was not repaired.

0563C–0564B (chunk 0011) — extra portam interiorem re-quoted as contra portam interiorem

Migne prints the lemma (Ezek. 40:44) at 0563C as Et extra portam interiorem gazophylacia cantorum in atrio interiori, and Richard re-quotes his own lemma at 0564B as et contra portam interiorem gazophylacia cantorum in atrio interiori. The plate contradicts itself between a lemma and its own re-quotation half a column apart — the same shape of defect as 0584A–B (satius / latius) in cruces-0014.md, though here nothing in the argument turns on it, since Richard's whole discussion is about in atrio interiori and not about the preposition. Both are rendered as printed ("outside the inner gate" / "over against the inner gate"). No [var: …]: the divergence is internal to Migne, not a divergence from the received text, and Pattern 14 exists for the latter.

0567D (chunk 0013) — a terrae cubitis quinquaginta — the existing [var: …] endorsed

The stint left a [var: Vulg. a terra cubitis quinquaginta, "from the ground, fifty cubits"] against Migne's eminebant de inferioribus et de mediis aedificii a terrae cubitis quinquaginta (Ezek. 42:6), rendering the printed genitive literally as "by fifty cubits of the earth." Checked and kept. The rendering is faithful to what is printed, the note names its witness as Pattern 14 requires, and its content is not a verbatim substring of the Latin twin (which prints terrae, not terra), so the verifier's mirror test is satisfied. It is the only [var: …] in the four chunks.

Punctuation carried per Pattern 8

No sentence-boundary mark was supplied or deleted anywhere in the four chunks — every terminal stop, question mark and mid-sentence full stop in cols 0559C–0570D stands where Migne sets it. (Checked explicitly, because the merging defect logged at 11542 col. 1041D and 11436 col. 0349A is exactly what an unaudited stint produces.)

Numerals — every sum verified, and stated as results

This work has a confirmed corrupt measurement at 0559A and a second occurrence of it at 0560B (above), so every figure in these four chunks was checked against Richard's own reckoning rather than spot-checked. Apart from the 0560B quadraginta, all of them close, and several close exactly.

The one arithmetical tension worth recording

Every figure is rendered as printed and none is altered. Recorded because it is the only place in these four chunks where Richard's own figures do not close, and because it is not reducible to the 0560B defect.

Negation fidelity (Pattern 7a) — read clause by clause

Every printed non / nec / nisi / ne / nihil / nullus / nunquam / nusquam / nondum / necdum / neminem / nemo in cols 0559C–0570D is carried into the English, and none is added. The negation-dense passages, which is where the PL 196 defect class lives:

No dropped negative, no intrusive negative, and no comparative reversal was found in chunks 0010–0013. The 0553A–B site recorded in cruces-0005.md has no counterpart in this range; the two places where the plate defeats its own argument here (0560B quadraginta, 0566D Scriturae) are a numeral and broken type, not polarity.

Also checked for the other three mechanisms of Pattern 7a:

The u/n sweep

qnod for quod is confirmed three times in this work, once at 0589D (chunk 0018). Cols 0559C–0570D were swept for it and its relatives (nou, euim, antem, uon, Domiuus, qni, qnae, snnt, iu), and every hapax token in the four Latin chunks was listed and read individually rather than pattern-matched.

No u/n inversion occurs anywhere in chunks 0010–0013. The eight broken forms this range does carry (habenta, demidii, in fallor, contracuntem, totitem, Occiente, Scriturae, darietemque) are all other mechanisms: dropped letters, a d/p substitution, and one run-together. Recorded as a negative result so a later pass does not repeat the sweep.

Pattern 9 items (no English exponent — logged, English unchanged, do not churn)

Cross-confirmation of a peer's conjecture — 0566D (chunk 0013)

cruces-0005.md records at 0549A that Migne prints __liber__ hoc loco … colligere, where the infinitive has no governing verb, and conjectures libet. Chunk 0013 confirms it. At 0566D the plate prints the same sentence-frame, seventeen columns later and correctly:

Libet hic, sicut et supra, in descriptione porticuum hinc inde Scriturae testimonia in unum colligere…

— libet + colligere, with Richard's own back-reference sicut et supra pointing at the 0549A passage. The peer's conjecture is now near-certain on the plate's own evidence. No edit is made to chunk 0006 (that is a peer's file), but the corroboration is recorded here so the conjecture can be upgraded when the work's cruces are next consolidated.

Conventions reconciled — two chunks were edited

Both edits are terminology, both bring a chunk into line with settled work-wide vocabulary, and both left every column anchor, [n: …], [sic: …] and ## head untouched. node scripts/verify-english.mjs 11553 passes after the edits.

Flagged, not fixed: chunks 0019 (×2) and 0020 (×1) still keep the italic Latin, per the decision recorded in cruces-0018.md. With chunk 0013 conformed, the work now stands at six "enclosure" against three italic Latin, so the remaining three are the ones to sweep — but they are a peer's files and the change is not one this audit should make unilaterally. Recorded here and in the return so the sweep can be done in one pass.

Conventions applied without change (checked against the work-wide list)

porticus portico · vestibulum porch · atrium court · gazophylacium chamber · thalamus little chamber · cubitus cubit · mensa table · pavimentum pavement · altare holocausti altar of holocaust · aedificium building · coenaculum upper chamber · frons front · determinatio / determinare determination / determine · deambulatio walk (0569B, matching the peer convention at 0583C). All were verified present and consistent in all four chunks; none required an edit.

Anchor sequence — recorded, NOT reported

Per the standing ruling, logged only so a later pass does not re-open it. Chunk 0013 opens on a bare, band-less [0566D] standing alone on its own line between the head and the first paragraph, exactly as the Latin twin does. Every other anchor in the four chunks is banded and in order (0559D–0562B, 0562C–0565B, 0565C–0566C, 0566D–0570D), and each chunk's first anchor follows its predecessor's last without a gap.

No text breaks mid-word across any anchor in these four chunks — each of the twenty-eight joins was checked. Several break mid-construction (0560B superius [0560B] nusquam descripserat; 0567D quinquaginta [0567D] cubitis; 0568D octoginta cubitorum [0568D] exstitit), which is normal positional transcription and not a defect. Reproduced verbatim; no [ed: …], no patch.

Candidates examined and rejected

Recorded so this audit can be told from one that did not read.

  1. 0562D lavantur where the argument has been about immolari — Si ad solam orientalem victimae lavantur in diebus non festis…, in a paragraph whose premise is ut super eas holocaustum immolari debeat. Not a defect: chunk 0010 has just established at 0561C–D that washing belongs to the gazophylacia and sacrificing to the mensae, and Richard is deliberately running the dilemma across both offices. Rendered as printed, no crux.
  2. 0563D quas tamen … reponebant with no expressed subject — the subject sacerdotes stands in the preceding clause. Ordinary Latin ellipsis, not a defect.
  3. 0564A Sic thalamus interior aula dici solet — read for a lost word (aulae?), rejected: the sentence construes as printed, and the following non quod intra ipsam sit fixes aula as the thing the chamber is called. Rendered normally.
  4. 0567B in superioribus humiliora in the Ezek. 42:5 lemma — considered for a [var: …], and not marked. The divergence from the received text, if any, is one of word-order and does not change the sense Richard draws (that the upper chambers were lower because they carried the porticoes), and Pattern 14 explicitly requires the sense to turn on it. Rendered as printed.
  5. 0568C versumque ad viam respicientem ad mare — the -que has nothing to co-ordinate inside the lemma as printed. Read as loose lemma-stitching rather than broken type; rendered as printed, no marker.
  6. Chunk 0010's [sic: habenta] sitting inside a partly-italic lemma — checked that no nested italics were produced and that the marker's content is a verbatim substring of the Latin twin. It is. Same check run on all eight [sic: …] markers in the range; all pass.
  7. Anchor [0566D] alone on its own line in chunk 0013 — checked against the Latin twin, which does the same. Not an [ed: …] case; nothing is lost.
  8. heads: frontmatter still reading CAPUT X. etc. while the body reads CHAPTER X. — left alone. The frontmatter mirrors the Latin twin's field and the verifier aligns on it; changing it is a pipeline decision, not a translation one.

Chunks found wholly clean

Richard of St Victor, PL 196, cols 0571A–0585. Rendered under translation-style.md Pattern 7 / 7a: the printed reading stands in the English, the conjecture lives here.

Carried type (Patterns 10 + 12 — [sic: …] in the English)

The same sentence carries a second defect which is not marked, because it is a real word doing the wrong work: "inter ista et orientale et pavimentum" will not construe. The sense demands inter ista et orientale pavimentum ("between these and the eastern pavement"), and the whole paragraph is a comparison of the two lateral pavements with the eastern one. The intruding second et is rendered as printed ("between these and the eastern and the pavement"), which is unconstruable English because the Latin is unconstruable Latin.

Scripture divergence from the received text ([var: …])

Both are rendered as printed — "the temple was preferable in the upper parts" in the lemma, "the temple was wider in the upper parts" in the gloss — with [var: Vulg. latius, "wider"] beside the lemma. No [sic: …]: satius is a real Latin word, merely the wrong one (Pattern 12's boundary). Conjecture: latius, effectively certain, on the strength of Richard's own re-quotation.

Readings rendered literally though they will not construe

Numerals — the arithmetic, checked figure by figure

This volume has a documented history of corrupt measurements (cf. 0559A in chunk 0009), so every figure in these four chunks was checked against Richard's own reckoning. All of them agree, and one of them agrees remarkably:

The one numerical tension worth recording

Pattern 9 items (no English exponent — logged, English unchanged)

Punctuation carried per Pattern 8

Negation fidelity (Pattern 7a)

Every printed non / nec / nisi / nihil / nullus / nunquam / ne in cols 0571A–0585 is carried into the English, and none is added. The negation-dense passages were read against the Latin word by word:

No confirmed dropped- or intrusive-negative site was found in chunks 0014–0017. The 0553A–B site recorded in cruces-0005.md has no counterpart here; the two places where the printed text contradicts itself in this batch (0574D et orientale et pavimentum, 0584A–B satius / latius) are lexical, not polarity, defects.

Conventions decided here, beyond the work-wide list

Anchor sequence — recorded, NOT reported

Per the standing ruling, logged only so that a later pass does not re-open it. Chunk 0015 ends [0577] [0579], both bare and band-less, with [0578] never appearing; chunk 0016 opens at the bare [0580]; chunk 0017 runs [0585A] … [0586A] … [0585], i.e. out of order and ending on a bare anchor, and the next chunk opens at [0587A]. All are normal positional transcription. No text breaks mid-word or mid-construction across any anchor in these four chunks — each was checked at the join. Reproduced verbatim; no [ed: …], no patch.

Chunks found otherwise clean

0014 (cols 0571A–0573C) carries no broken type at all and no numerical defect; its only entries above are the Quam autem vestibulum ellipsis at 0571D and the pro amore suo of 0572A, both real words rendered as printed. Its arithmetic is sound at every point.

Richard of St Victor, PL 196, cols 0587A–0600C — the closing four chunks of the work. Rendered under translation-style.md Pattern 7 / 7a: the printed reading stands in the English, the conjecture lives here.

Carried type (Patterns 7 / 10 / 12 — [sic: …] in the English)

0588B (chunk 0018) — Ante solaria for antesolaria

__Ante solaria__ quaedam intelligimus esse istas ethecas, quas, ut Hieronymus in his verbis testatur, Romae vocant solaria de coenaculorum parietibus eminentia…

Read as printed, ante solaria is "before the lofts," which strands the neuter plural quaedam with no noun to modify (ethecas is feminine). The word required is antesolaria, and Migne himself prints it as one word twelve lines later, at 0588D: per limina solariorum situs __antesolariorum__ exprimuntur. So the plate divides at 0588B a compound it sets whole at 0588D. Split type per Pattern 10; the whole run is carried, including the real word Ante, and marked [sic: Ante solaria].

Conjecture: antesolaria, certain on the plate's own evidence.

0589D (chunk 0018) — qnod for quod

Angulos itaque vocat, eo __qnod__ angulos tenerent ligna illa… u/n inversion, the standard Migne compositor slip already logged twice in this work (0535C latitudiuis, 0544C qninquaginta). Not a word; carried untranslated in italics, [sic: qnod]. Conjecture: quod, certain.

0590D (chunk 0018) — sta for ista

…quae ab eo loco incipiebat ubi __sta__ porticus porticui aquilonaris domus juncta erat… A dropped initial letter; sta is not a Latin word in this position. Carried untranslated in italics, [sic: sta]. Conjecture: ista ("that portico"), near-certain — Richard uses ista porticus of this same triple portico repeatedly in Capp. XIII–XVII.

0598A (chunk 0021) — altars for altaris

Erat itaque suprema extremitas __altars__, vel coronae, ubi altare terminabatur… A dropped i; not a word. Carried untranslated in italics, [sic: altars]. Conjecture: altaris, certain — the genitive is required by extremitas, and the plate prints altaris correctly in the same sentence's neighbours (0597D, 0598B, 0598C).

Readings rendered literally though they will not construe

0587B (chunk 0018) — hunc inde for hinc inde

…unde constat inter templum et utramque domum __hunc inde__ quadraginta cubitos fuisse… The phrase Richard uses throughout the work for "on this side and on that" is hinc inde (0588B in this very chunk, 0590D, and passim). hunc is a real Latin word (masculine accusative of hic) and therefore takes no [sic: …] under Pattern 12's boundary, but it governs nothing here and leaves inde stranded. Rendered as closely as the printed words allow — "there were, this one thence, forty cubits" — rather than smoothed to the idiom. Conjecture: hinc inde.

0591B (chunk 0018) — Nondum sane where the argument requires Notandum sane

Ostium unum ad Aquilonem. __Nondum sane__, quia quod prius quasi singulariter posuit, postea pluraliter designavit.

Nondum ("not yet") answers nothing in the preceding sentence and leaves the quia clause without a main verb. Notandum sane — "it must indeed be noted" — is Richard's standing formula and supplies exactly what is wanted; he uses it at 0588A in this chunk, at 0594B in chunk 0019, and at 0558B, 0558C, 0558D, 0559C earlier in the work. The corruption is t→n, a→u, d dropped.

This is a negation site and the printed negation is carried: the English reads "Not yet indeed, because what he first set down as if in the singular, he afterward marked out in the plural," which does not follow, exactly as the Latin does not follow. No repair, no marker.

0591D–0592A (chunk 0019) — de interioribus aedificii where the lemma reads de inferioribus

__Inferior__ porticus de __interioribus__ aedificii eminebat, quia supra gazophylacia sita erat, quae __inferioribus__ aedificiis respondebant. __Superior__ porticus de __mediis__ aedificii eminebat…

The sentence pair is built as a strict parallel on the scripture Richard is glossing, which he quotes twice in these two chunks: Eminebant de __inferioribus__, et de __mediis__ aedificii (0592C, and again at 0595D). The lower portico must therefore project de inferioribus, not de interioribus — and the same sentence's own relative clause (quae inferioribus aedificiis respondebant) prints inferioribus correctly.

Rendered as printed ("The lower portico projected from the inner parts of the building"), which breaks the parallel on the page as the plate breaks it. interioribus is a real word, so no [sic: …]. Conjecture: inferioribus, effectively certain.

0597C (chunk 0020) — ut inimicitias vitaret

Forte ergo ut __inimicitias__ vitaret, imo absque dubio certi causa mysterii, ut plenum mysticumque numerum exprimeret, minores hoc loco calamos posuit…

"Enmities" has no antecedent: the whole passage is about why the prophet reckoned in the lesser reed rather than the greater, and nothing hostile is in view. Something in the class of inaequalitates / ambiguitates is what the argument wants. The word is real and translatable, so it is rendered as printed ("that he might avoid enmities") and no conjecture is offered here with any confidence — the reading is corrupt past secure restoration, and is marked rather than adjudicated (the 9604 boundary).

Punctuation carried per Pattern 8

Migne's abbreviations in the Cap. XX glosses (0599B–0599C) — decision recorded

The italic gloss paragraphs opening Cap. XX are set very tight and Migne abbreviates inside them: sanctific. twice, and sanctuar, (with a comma where the period belongs). These are translated, not carried. Reasoning: an abbreviation is not broken type — it is legible, unambiguous (the unabbreviated sanctificatum / sanctuarium stand in the same sentence and in the scripture blocks immediately below), and it has a full English exponent. Marking them [sic: …] would blame Migne for a defect where he merely compressed, and would bury the four genuine carried runs above in noise, which is the argument Pattern 9 makes. The stray comma after sanctuar is reproduced as a comma, per Pattern 8.

Numerals — checked, and internally consistent throughout

Logged as verified results, since this volume has a documented history of corrupt measurements and this work has already produced one (0559A).

Negation fidelity (Pattern 7a)

Every printed non / nec / neque / nisi / nunquam / nullus / nihil / nondum in cols 0587A–0600C is carried into the English, and none is added. The negation-dense passages were read against the Latin clause by clause:

No confirmed dropped- or intrusive-negative site was found in chunks 0018–0021. The two sites where the plate defeats its own argument (0591B Nondum sane, 0591D interioribus) are corruptions of adjacent words, not of a negation, and both are rendered as printed.

Conventions decided here, beyond the work-wide list

A divergence from chunks 0005–0013 that needs a sweep, not a silent choice

These four chunks translate the chapter head-word: CAPUT XVII. → CHAPTER XVII., per the settled CAPUT PRIMUM → "CHAPTER THE FIRST." convention recorded in cruces-0000.md and per the standing rule that ## head text is translated.

The work is currently inconsistent about this. Chunks 0000–0004 translate it (CHAPTER THE FIRST, CHAPTER II, CHAPTER III, CHAPTER IV); chunks 0005–0013 leave it as CAPUT V. … CAPUT XII. This is flagged rather than resolved unilaterally: the fix is a one-line sweep of the English heads in 0005–0013 (and of 0014–0017 if the same choice was made there), not an edit any single agent should make to a peer's files.