Agreement needs an explanation.

Original editorial teaching draft. The two Getty photographs have been inspected; object identifications, attribution, dating, and reported repainting are credited to the institutional records. This workshop supplies a method and a working notebook, rather than a completed critical edition.

First ask what the second witness is

Reading an inscription twice can improve a transcription. It does not necessarily give you two independent witnesses. If you read a photograph and then a catalogue caption based on that photograph, the second reading may help you understand the first while adding no independent evidence of the original lettering. The relation between the sources matters as much as the number of sources in your bibliography.

The default pair on the comparison bench comes from the Hours of Simon de Varie, Getty Ms. 7 (85.ML.27). The museum identifies the first opening as Simon de Varie Kneeling in Prayer, folio 2, accession 85.ML.27.2. It identifies the second as Coat of Arms Held by a Woman and a Greyhound, folio 2v, accession 85.ML.27.2v. Both records give Jean Fouquet and 1455. These are the museum’s attribution and dating, preserved as catalogue statements. [1] [2]

They are different openings in one manuscript. Their inscriptions can reveal how words and visual arrangements recur within that book. Agreement between them does not by itself establish that an unrelated copy preserved the same wording. The comparison therefore begins with a limited question: how does a selected inscription appear in these two located displays?

Keep the catalogue’s folio notation. The first record supplies “fol. 2”; the second supplies a verso designation. Our notebook does not silently add a recto designation to the first. You can investigate the fuller sequence, but a revised locator should cite the evidence for that revision. A pair of consecutive-looking numbers is also insufficient evidence that the leaves have always occupied their present positions.

The Portuguese dossier offers another kind of pairing. Its source is the Torre do Tombo’s presentation of Livro do Armeiro-Mor, book 19, folio 87, reproduced on PDF page 23. Comparing it with a French devotional manuscript can test editorial method, the mediation of captions, or the relation of lettering to an armorial image. Their texts are not copies of one common passage. An English translation making both readable does not turn them into textual variants. [3]

Locate the correspondence before aligning the words

Open the two Getty images together. On folio 2, the central miniature contains the kneeling figure and a woman supporting a shield. Lettered bands frame the scene. On folio 2v, a woman stands to the left of the armorial display and a white greyhound to the right. The animal’s collar carries lettering, while further lettering occupies scrolls above and below. Those placements can be described from the photographs. Identification of the person, attribution to an artist, and the interpretation of repainting require their separately cited sources.

Start with the lower scroll in each opening. Give your selection a precise description: the whole visible inscription, a specified group of letters, or a doubtful joined form. Then consider the upper scrolls and the collar. A word repeated at two positions on one leaf must receive two locations. Otherwise the apparatus may report a difference without letting the next reader find which occurrence you meant.

Getty’s account of folio 2v identifies the desire motto on the scrolls and the collar, and the other personal motto on the lower scroll. That account gives a useful route into the image. It is a modern reading and interpretation, not your own transcription. Record what you can resolve at the available scale; cite the catalogue when it supplies a reading you could not establish from the photograph alone. [4]

A collar curves around a body. A scroll bends through folds. Both may interrupt the visible sequence. The order in which a modern editor presents the words is already a decision about that arrangement. Describe the path you followed through the lettering. If a portion cannot be read, mark the gap under your notation policy instead of completing it from the more legible occurrence.

An exact difference is only a starting point

Bring a saved single-witness draft into each column, or enter a short reading directly. The bench makes a separate copy. Imported editor details, notes, and original apparatus entries travel with that copy into the exports, even though only scope, diplomatic transcription, normalized reading, and translation are editable in the columns. Return to the desk to revise the original apparatus, then preview the revised draft before replacing the comparison’s copy.

Choose one text layer and run the comparison. The preview marks text supplied only on the left with a strike-through and text supplied only on the right with an underline. Shared text remains unmarked. These marks concern your entered strings. They do not describe paint strokes, establish which source came first, or judge which reading is correct.

The comparison preserves case, punctuation, accents, and whitespace exactly. A supplied line break can therefore produce a difference. An accented character stored as one character can differ from a letter followed by a combining accent, even when both look alike on screen. The bench applies no automatic Unicode normalization. Before interpreting an apparent difference, ask whether it comes from the witness, your notation policy, or the way the text was entered.

The preview accepts up to 600 tokens per side, counting word runs and whitespace runs separately. A space between two words is one token of its own. Longer readings remain in the notebook and exports, but require shorter excerpts for highlighting. The tool declines that preview instead of quietly discarding the end of a passage. Repeated words can also admit more than one alignment. The displayed sequence is a convenient textual comparison, not a historical demonstration of correspondence.

Now compare the normalized layer. Suppose you regularized case and supplied ordinary spaces in both readings. Some diplomatic differences may disappear. That result can be useful: it shows what your normalization does. It cannot show that the original inscriptions were identical. Write the normalization policy alongside the result so the reader can recover the distinction.

A translation comparison asks a further question. Different English phrases can convey the same established French reading; identical English phrases can conceal a difference in the source wording. Explain the grammatical or interpretive choice instead of using the size of the highlighted passage as a score of translation quality. Return to the source-language layer wherever the historical argument depends on a particular word.

Explain differences at the place where they occur

The located comparison entries connect two loci, two readings or visible features, a difference category, a review status, and an explanation. Add them yourself. The text preview cannot know whether a changed character represents an uncertain stroke, an abbreviation expansion, or a typographical error in your own draft. That judgment belongs in the apparatus.

For example, a student may find a joined form clear on one scroll and doubtful on another. A provisional entry should identify both locations, describe the doubtful form, give the proposed resolution, and say whether the proposal depends on the parallel. It should also name a check that could disprove it: a higher-resolution view, another occurrence, or a competent reader’s inspection. Calling the entry “checked” records that a named check took place. It does not confer the museum’s approval.

A material difference requires another kind of explanation. Getty reports that the shield on folio 2v was repainted in the seventeenth century with fleurs-de-lis over the earlier display. This is an institutional catalogue claim about an intervention. The photograph permits a description of the present surface; the project has not conducted technical imaging or reconstructed the earlier arms. The notebook’s material-change category preserves that distinction. [2]

A changed shield and a recurring motto need not have the same material history. Do not date an inscription by attaching the manuscript’s catalogue date to every visible mark. Likewise, repetition of wording cannot establish that two inscriptions were executed at the same moment. Separate the observation, the catalogue’s account, and any further chronological inference. When the evidence runs out, state which additional source would be needed.

Keep the comparison reproducible when it changes

The bench saves a notebook for each ordered pair. Left folio 2 with right folio 2v has its own record. Swapping brings the readings and located entries across if the reverse pair has no saved notebook; otherwise it restores that existing notebook. Directional prose in an argument needs your review. A sentence referring to “the left occurrence” does not become correct merely because its accompanying text moved.

Import offers a preview before replacement and an export of the destination notebook. A complete comparison import may target a different pair from the one currently displayed, so the backup belongs to that destination. A single-witness import must match its selected column. Neither operation overwrites the single-witness desk. If storage is unavailable, the open page retains the work for export; clearing or closing it can still lose that temporary state.

The source records accompany the portable JSON and readable Markdown. Imported URLs and rights statements cannot substitute for the bench’s known source records. Preserve dated exports when an editorial change alters the argument. The shared pair link contains the selected witness identifiers, with no readings or notes. A colleague receiving the link sees the sources; send the export separately when you intend to share the analysis.

Image rights also belong to the witness inventory. Folio 2 uses Getty’s CC0 IIIF image. The local folio 2v photograph derives from the Google Art Project reproduction on Wikimedia Commons, whose file record gives Public Domain Mark / PD-Art. That is a different rights record and must remain identified as such. Both local files reproduce the full photographed leaf in compressed form. The Portuguese dossier remains linked to the archive while reproduction clearance is unresolved. [5] [6]

A comparison whose limits improve its answer

Assignment: Submit a 900–1,200-word report and a comparison export. Select one precisely located passage or visual feature from each Getty opening. State the question, establish the relationship between the witnesses, provide diplomatic readings under a shared policy, and explain your normalization and translation. Include at least three located comparison entries. An entry may document a demonstrated agreement, provided it explains what that agreement can establish.

Show one place where an editorial decision changes the apparent comparison. Then distinguish a source observation from a catalogue interpretation and identify a missing check that could revise your conclusion. If you use the Portuguese dossier instead, frame the assignment as a comparison of source mediation or method. Explain why a textual-descent claim would require different evidence.

Criterion	Points	Strong evidence
Witness relationship	25	Exact locations, access forms, source dependence, appropriate question
Reading policy	25	Separate layers and disclosed interventions, with no concealed gap filling
Located apparatus	30	Reproducible observations, reasoned alternatives, defined review checks
Argument and revision	20	Proportionate conclusion, preserved sources and rights, a concrete next check

What would a careful conclusion sound like?

“The two selected displays occur in one manuscript. Under my stated normalization policy their entered readings agree, while the diplomatic records preserve a difference in arrangement. One uncertain form was resolved provisionally from the parallel occurrence and remains labeled as dependent on it. This comparison documents recurrence within the book; it does not establish an independent line of transmission or the date of each painted inscription.”

A useful next step names a location or a form of evidence. “Do more research” leaves the problem where it began. “Check the doubtful collar group against a higher-resolution source image, then record whether the parallel still supplies the reading” tells another researcher how to continue.

Open the comparison bench ↗

Source trail

1. Getty, Simon de Varie Kneeling in Prayer, Ms. 7, fol. 2; accession 85.ML.27.2.
2. Getty, Coat of Arms Held by a Woman and a Greyhound, Ms. 7, fol. 2v; accession 85.ML.27.2v. Institutional metadata and reported repainting.
3. Torre do Tombo presentation, PDF p. 23, liv. 19, f. 87. Modern caption verified; original-hand reading remains open.
4. Getty’s Google Arts & Culture account of folio 2v, modern reading and interpretation; not an independent historical witness.
5. Getty IIIF manifest, fol. 2 canvas and CC0 image declaration.
6. Wikimedia Commons folio 2v file and rights record, Google Art Project image, Public Domain Mark / PD-Art. The downloaded original’s SHA-1 matches the file record; the local image is a resized JPEG.

Sources consulted 9 October 2026. Continue with the single-witness workshop, translation seminar, or documentary capstone.