On page 171 of the catalogue for Kyoto National Museum's 2018 sword exhibition, a blade appeared under the Momoyama period with a length of 70.7 centimetres. Consult the museum's correction sheet for entry 130 and the period becomes Edo, the length 58.9 centimetres. A reader could have copied the first figure faithfully, converted it accurately into inches, and built an entirely mistaken comparison around it. The error would have survived every calculation. Its remedy was another piece of paper.
The sheet may be only a page long, but it is a particularly good introduction to reading an exhibition catalogue. It corrects dates, measurements, names, credits and even the inclusion of an enlarged photograph. Each kind of correction answers a different historical question. Follow them through the book and a sword catalogue begins to look less like a list of settled facts and more like a set of carefully indexed claims, each attached to a particular object, component, language and edition.
A book, a room and a timetable
Swords of Kyoto ran from September 29 to November 25, 2018. The National Museum's exhibition archive identifies the venue, the sections and a selection of highlighted works. Its checklist serves a different purpose. It ties numbered works to their display schedules. A catalogue can bring the whole exhibition between two covers even when no individual visitor could have seen the same arrangement. Two readers might recall different objects from the same room and both be right.
Consider the paired screens showing the Gion Festival. The museum lists the right screen for the first part of the exhibition and the left for the second. The published entry lets a historian discuss the conception of the pair. A claim about what appeared on the wall on October 10 needs the timetable as well. Nor does the catalogue's juxtaposition of a painting and a blade establish that a photographer looking along a particular gallery wall could have captured both. A chapter is an arrangement of ideas. The installation had its own arrangement in space and time.
The distinction can be turned into a simple exercise. Imagine captioning a photograph lent by someone who visited the show but remembers only that it was autumn. An identifiable screen might narrow the date considerably. A catalogue number alone would establish a link with the exhibition but might leave the date open. The two kinds of evidence fit together neatly if the caption preserves the distinction between an identification and an inference. Without that distinction a helpful clue becomes the source of an invented date.
Six different dates can fit on one card
A catalogue may give the date inscribed on a tang, a broader attribution to a period, a later date for the mount, the date an institution received the object, an exhibition date and the date of the catalogue itself. There is no competition among these dates once each has its subject. Trouble begins in a database with a single column labelled year. What would go in that column for a signed blade surrounded by decorative work produced centuries later?

Masazane's blade in New York, accession 2001.574, makes a good comparative exercise outside the Kyoto exhibition. The Met records an inscription naming the maker and another giving a day, month and year, which it presents as a date in 1526. Its acquisition belongs to the twenty-first century. One date helps locate the smith's work. The other helps locate the object in the museum's institutional history. The accession number happens to start with the year of acquisition in this case. That does not make the first digits of every institution's identifier a dependable source for dating metalwork.

The Met's accession 91.2.54, associated with Kuniyoshi, makes the exercise harder and more interesting. It brings a blade dated 1525 together with nineteenth-century mounts. Look only at the beautifully finished exterior and it is easy to forget that the older inscription refers to a different surface. A catalogue designer could reasonably place this work in an exhibition on Muromachi makers or on later lacquer and metalwork. The same number could then appear in two chapter headings separated by centuries. The object would not have changed. The question put to it would.
Reading a Japanese era in translation
An era name in a transcribed inscription is a historical expression. A year in an English caption is an editor's way of making that expression legible to another audience. It is important to record which form is actually visible on the object and which comes from the museum's interpretation. A Gregorian year added by the editor should not reappear in quotation marks as if those four Arabic numerals had been cut into the steel.
The Kyoto exhibition includes an inscription attributed to Awataguchi Ikkanshi Tadatsuna. In the museum's English presentation the date includes an auspicious day in the eighth month of 1709. A reader looking for a day-by-day sequence of the artist's work might wish for the missing day number. The catalogue has not supplied one. Auspicious language is part of the information conveyed by the inscription, and turning it into a precise calendar day would discard the actual wording in favour of a modern researcher's preference.
Where timing matters, it helps to keep a transcription and an English date in separate fields, accompanied by the institution's identification of the component. This is especially important when joining two catalogues in different languages. The second writer may be translating the first, independently reading the original, or summarising an earlier published transcription. Three identical printed years might descend from one reading. Counting the printed dates gives a false impression of three independent witnesses unless the chain of transmission is clear.
A correction can change the people in the story
Some corrections change almost nothing about the physical object and a great deal about its biography. For entry 168 the Kyoto notice changes a reference to Toyotomi Hideyoshi to one to Toyotomi Hideyori. An English writer unfamiliar with the names could easily take the difference for a transliteration variant. It is a change of person. A provenance diagram based on the uncorrected passage would need a different node, with potentially different dates and connections.
The correction tells us which name the museum intended at that place in the text. It does not supply a full account of how the mistaken name was introduced. It is tempting to blame a copyist or to invent a confusion between two documents. Neither explanation follows from the notice. The editorial history that can actually be reconstructed is modest and valuable. A particular printed statement was replaced by another under the museum's authority. A later biographical claim should refer to the amended statement and distinguish it from a newly discovered historical document.
Another line moves an entire entry from one period to another. That kind of change has consequences well beyond the label. Imagine a researcher selecting all Momoyama works for a small exhibition of reproductions. The old checklist would introduce an Edo work before the researcher had even started comparing form or decoration. Correcting the dimension while leaving the period untouched would fix one field and preserve the underlying selection error. A notice naming several fields for one entry deserves to be read as a whole.
When the lender moves one line
Near the end of the Kyoto sheet three consecutive entries illustrate a different editorial problem. A blank becomes a Tokyo National Museum credit. The next entry receives a Kyoto National Museum donor credit. The following credit becomes blank. The notice does not ask a reader to decide where a line might once have slipped on a proof. It names the affected entries individually. That precision matters to anyone trying to locate an object years after the exhibition has closed.
A researcher working from a photograph of a single page could otherwise end up writing to the wrong institution. More subtly, a historian tracing patterns of collecting could count the wrong gift in the wrong collection. Lender, donor and present owner answer related but distinct questions. A donor credit records a transfer named by the receiving institution. A lender credit records an exhibition relationship. A blank gives no positive identification at all. Filling it from the line above might make a spreadsheet look tidy while fabricating a relationship.

For comparison, the Met's peidao with Japanese blade and Chinese mounting, 32.75.301a, b, records Giovanni P. Morosini's collection and presentation by his daughter Giulia. Even in a concise credit there are two different roles. Replace them with a single generic owner field and the history becomes poorer. Replace them with the name of a museum that later borrowed the work and the account is positively misleading. The credit is often the shortest part of a label and the part most in need of careful grammar when reproduced.
What the enlarged image can establish
For Kyoto entry 26 the errata sheet instructs readers to delete an enlarged image. That small instruction has serious implications for visual research. The printed image still exists in copies of the book. A reader can describe its presence as a fact about that edition. The museum has withdrawn its use in that entry. A researcher looking for a detail of the exhibited work therefore cannot treat the image as an authorised illustration of that detail without further evidence.
Images travel farther and faster than their captions. A cropped detail might be saved under the object number, copied into teaching slides and remembered years later by someone who never saw the book. Recording the page, figure, caption and any correction at the time the image is selected gives later readers a way back. This practice also protects honest disagreement. Two readers may interpret a verified photograph differently. An image whose relation to the object has been withdrawn presents an earlier problem that interpretation alone cannot cure.

The four comparative photographs in this essay are from New York records, with their own accession numbers. They are here to show what a component, alternate view or ensemble credit can look like in a second collection. They are not photographs of the numbered Kyoto exhibits discussed above. Setting two museum records alongside each other is a legitimate comparison. Quietly substituting a similar-looking blade for a missing view of a named Kyoto object would turn that comparison into a false illustration.
Building a citation another reader can use
Suppose you want to cite the amended length for entry 130. Include the exhibition title, institution, catalogue entry and page; then identify the museum's correction sheet and the corrected value. If the original error matters to your argument, record the superseded figure separately and label it as such. A reader then has two straightforward tasks. They can find the printed statement, and they can find the authority for departing from it. Neither task requires guessing which version you happened to have open.
For bilingual publications there can be another step. The Kyoto sheet includes corrections to the English entries as well. A historian working from the English list should follow the entry number across languages and check whether the change pertains to the same component. Overall length and blade length can both be correctly expressed in centimetres while referring to different things. A bilingual list with slightly different abbreviations is a poor place to resolve the distinction from memory.
A correction can also preserve historical spelling. In several entries the Kyoto notice amends the written form of Osaka. A modern English map can help a reader find the city. It cannot decide which historical characters should appear in a transcription of an inscription. A historian interested in regional names might care intensely about that difference, while a visitor planning a journey might barely notice it. An accurate catalogue leaves information usable for both questions by preserving the original writing alongside a searchable modern place name where appropriate.
From page proof to historical argument
In a reading room, try constructing two labels for the same object. Give the first only enough space for an institution, title and date. Let the second explain what was dated, the basis of the dating and any correction affecting it. The first label may be quite adequate for a gallery visitor. The second permits a scholar to compare an inscription with a stylistic attribution or to distinguish a recent gift from an old holding. The short version works because a longer record exists somewhere behind it.
The pleasure of the Kyoto errata is that one can watch that longer record being repaired. A person returns to the right sentence. A donor returns to the right row. An enlarged image ceases to stand for a detail it cannot reliably document. Read in that light, a book's loose correction sheet belongs with its plates and essays. Removing it might make a handsome copy easier to shelve. Keeping it with the book leaves future readers a better exhibition.
Matching a detail to the object
In the first comparative photograph the dark end of Masazane's blade occupies a narrow strip of the frame. The photograph is useful precisely because a curator chose to isolate that area. The third illustration reverses the choice. The peidao accession includes a view of a dressed figure, which places the work in a larger visual setting. A catalogue reader tracing dimensions must return from that image to the component record before assigning a length to anything in the picture. A difference in framing changes the question an image can answer even when its underlying accession is secure.
Questions from the reading room
Which version of entry 130 should I use?
For a statement about the exhibited work, use the corrected Edo attribution and the amended 58.9-centimetre measurement. Identify the entry and page and cite the Kyoto museum notice. If the purpose is to study the history of the catalogue itself, the superseded Momoyama attribution and earlier measurement are also evidence, provided that their status is clear.
Does a catalogue number identify a work outside that exhibition?
An exhibition number normally identifies an entry within one exhibition or publication. For a cross-collection comparison, also record the lender or holding institution, the object title, and its accession number if available. Preserve the exhibition number so that someone working from the printed book can retrieve the same entry and its corrections.
Why might two visitors remember different screens?
The Kyoto presentation rotated works. For the Gion Festival screens, the museum gives different periods for the right and left screens. A recollection of one visit and an illustration of the full pair can both be accurate. A particular date of attendance should be checked against the published schedule before reconstructing the installation.
How should I quote a Japanese inscription containing an era date?
Separate the original wording or its transcription from the museum’s interpretation and calendar conversion. If the record gives a month and an auspicious day without a day number, retain that degree of precision. Keeping these fields distinct helps show what comes from the object and what the cataloguer supplied for readers.
Does removing a photograph from an entry mean that the object was withdrawn?
Those are different editorial acts. The Kyoto notice identifies an enlarged image for removal under entry 26. That instruction limits what the image may be used to illustrate. It does not by itself establish that the exhibited work was withdrawn or that every other photograph of it was rejected.
What is the easiest correction to overlook in a provenance study?
A credit line moved among adjacent entries can be easy to miss because every institutional name may still look plausible. Check the exact entry, component and page. The revisions around numbers 195 to 197 show why both a seemingly empty cell and a familiar museum name need to be checked against the publisher’s notice.
Museum films
The Met — The Japanese Sword and the Japanese Aesthetic
The British Museum — How the samurai sword evolved over 1000 years | Curator's Corner S10 Ep11
