Small Objects in Translation: Kogatana and Kozuka in European Catalogues

A search for kozuka in the Ashmolean Museum’s collection turns up a copper guard decorated to resemble strips of bamboo. The entry is not a mistake. The word occurs in a description of the rim of an opening in the guard. No little handle appears among the objects registered under that number, and the museum counts one item. This is an excellent place to begin a history of translation. The same word can identify the object being catalogued, one part of an assembly, a neighbouring part that is no longer present, or the object for which an opening was made.

A kozuka is the handle of a small knife associated with a Japanese sword mounting. The word kogatana names the small blade or knife, with the exact extent often left to the accompanying description. English-language cataloguers have variously led with knife, handle, grip, fitting or a Japanese term. Their choices tell us how the object was being considered at that moment. Four pieces in Cleveland, two photographed as handles and two as blades with handles, let us put the titles to a visual test. An exhibition catalogue printed in London and a research catalogue from Oxford tell us how the words reached readers on quite different occasions.

The handle that gets its own name

A dark elongated handle with golden chickens in a low landscape
Cleveland 1940.673, knife handle with a family of chickens. The collection records shakudō and gold. Cleveland Museum of Art. CC0.

The Cleveland Museum of Art’s handle with a family of chickens appears as a long, low rectangle. A bird’s tail makes a fine arch and smaller forms are spaced against a dark ground. The photographed face invites the close looking one might give a small painting. The collection records copper-gold alloy and gold, gives nine centimetres as the overall extent, and identifies the piece as a bequest of James Parmelee. Nothing in that title requires the presence of an attached blade. The object can be catalogued and illustrated coherently as a work of decorative metalwork in its own right.

The title makes a delicate distinction between the object and its subject. The chickens belong to the design; the handle is the physical article; the accession number, 1940.673, identifies the museum holding. A search combining “knife” with “chickens” could recover the record, but it would make a peculiar label if the missing word “handle” were not restored. Historical catalogues often compress medium, motif, name and object class into one line. The reader has to expand that line without making a larger claim than its author intended.

In the catalogue of Japanese lacquer and metalwork exhibited at the Burlington Fine Arts Club in 1894, the explanation of sword furniture directs the English reader’s attention to the decorated outer face of the small knife handle. It identifies that handle with the Japanese term kozuka. The choice of face is significant in a catalogue of decorative work. A reader interested in metal artists could examine an ornament and compare schools or subjects without asking the display to include a complete mounting. A missing component in a modern image does not on its own establish loss or breakage after accession.

The name of the whole depends on the question

A pale narrow blade extends left from a dark handle with gold-coloured tortoises
Cleveland 1992.374, blade with handle decorated with a family of tortoises. Cleveland Museum of Art. CC0.

Cleveland 1992.374 has the family of tortoises named in its title, but its photograph begins far to the left of the decorated area. A complete narrow blade extends from the handle. The catalogue describes it as a blade with handle and adds Kozuka in parentheses. Here the Japanese word functions as an access point for a museum visitor searching for the decorated fitting, while the English wording identifies the extent of the illustrated assembly. Translating the parenthesis as if it were a strict name for every part of the photograph would undo the precision of the English phrase.

Compare this object with the chickens. The long blade takes up much of the digital frame. At first glance the decoration looks smaller and less elaborate. The composition explains part of that effect: the picture has to accommodate an assembly appreciably longer than its decorated portion. The two photographs are not at a common scale merely because they appear at the same display width. A close comparison of the subjects would ideally include details of equivalent extent as well as the complete museum views. Without both, one can easily mistake the proportion of the page for the proportion of a work.

The tortoise example belongs to the bequest of Maud Eells Corning. The chickens and tortoises thus came into the Cleveland collection under different accessions from different sources. A curator can put them together for a discussion of animal imagery or part names. That new association has no automatic bearing on where they were made or the company they kept before entering the museum. The distinction is especially useful in the case of fittings. Objects designed to occupy small places on a larger ensemble often acquire rich, entirely independent collecting histories.

A blade and decorated handle photographed as one long assembly
Cleveland 1992.375, blade with handle. The material field records metal. Cleveland Museum of Art. CC0.

Another Corning piece, 1992.375, immediately follows the tortoise assembly in the accession sequence. The catalogue again uses the English description of a blade with handle. Its material field simply says metal. On the screen the handle has darker and lighter accents, and the blade varies in tone along its length. Those observations do not authorize the more precise material description supplied for the chicken handle. The donor and proximity of registration numbers connect these two assembled examples in museum history. They cannot supply every material or attribute that the shorter records leave unspecified.

A London exhibition plate with its own numbering system

An illustrated catalogue associated with the Japan-British Exhibition in London brings a different editorial order into view. A plate preserved in the New York Public Library is indexed under sword guards and kozuka. The original exhibition took place in 1910, while the library dates the illustrated item to 1911. Its source was compiled by the office of the Japanese government commission to the exhibition. These facts locate the plate in a particular act of representation for an international audience. They do not make its numbers accession numbers of the library that now provides the image.

The library identifies four entries on the plate as knife grips: 389, 398, 401 and 407. Other numbers on the same plate identify guards, smaller ornaments and a cap-and-ring set. This is a particularly helpful example because all these pieces can appear under a general heading about sword furniture. A text search recovers the page. It takes the number key and attention to the individual form to determine which item the English term modifies. A catalogue image can carry the right keyword and still be the wrong picture for a particular claim.

The plate also records a historical choice about what could be shown together. Small, independently legible pieces can be grouped within a rectangular field more efficiently than they can be arranged in a narrative account. A visitor equipped with a numbered list can move between image and description without losing the place. The disadvantage becomes apparent a century later if a picture editor detaches a single illustrated fitting and carries over the heading of the whole page. Number, catalogue edition and institutional source then do more to preserve identification than another enthusiastic adjective about the workmanship.

A curious translation history has developed around the English word “furniture.” To someone unfamiliar with the specialist use, a catalogue of sword furniture can sound like a catalogue of stands or storage cabinets. In these publications it gathers the fittings associated with a mounting. A kozuka might be among those fittings whether shown alone or attached to its small blade. That is why the general section heading is a poor substitute for the individual entry. A reader looking for a sword stand and one looking for a decorated knife handle could otherwise leave with exactly the wrong idea about the same page.

The Oxford object that is not a handle

The Ashmolean record that opened this essay has accession number EAX.10945. Albert James Koop’s catalogue of the A. H. Church collection gives it number 945. The museum identifies the piece as a guard with bamboo, records a bequest from Sir Arthur H. Church in 1915, and gives copper and a dark alloy at the rim of the accessory opening. Koop’s notes describe the bamboo pattern on each face and name an attributed or inscribed maker in the entry. The work itself is classified as a tsuba. The kozuka occurs in a compound describing an opening.

The last point is more than a grammatical nicety. A full-text search for kozuka will index the description of that opening, irrespective of the object class. Someone counting search results to estimate how many handles a collection contains could therefore count the Oxford guard as one. The institution’s own object-type field, its item count and the accompanying photograph offer three ways of catching the error. A catalogue is an argument constructed from fields that qualify one another. It is not simply a bag of keywords into which the object has disappeared.

The dimensions reinforce the point. The recorded height and width of the Ashmolean piece are both around seven centimetres, with a much smaller depth. The fencing of vertical bamboo strips, the multiple openings and the shape of the guard account for the visual character of a broad, flat object. In contrast, the Cleveland handle with chickens has a markedly elongated form. Shape alone should not be made the sole test of identification, but it is a useful check against an isolated search term. When the shapes and the title seem to disagree, the first thing to inspect is what the term is doing in its sentence.

The bibliography introduces a further pair of numbers. In an online publication the object can occupy a particular place among more than a thousand results. That position belongs to the order in which the digital interface happens to present it. Number 945 refers to the catalogue entry and EAX.10945 to the accession. They need not agree in their full form. The reader who notes only the visible position in a search list can lose the object entirely after a new filter or a change in the interface. The accession and the explicit catalogue reference make a much more durable cross-reference.

What an apparently short measurement asks of us

Dark inlaid bronze handle with a compact ornamental grouping and smaller accents
Cleveland 1947.230, inlaid bronze handle, gift of Mather College Library. Cleveland Museum of Art. CC0.

A fourth Cleveland piece, 1947.230, has a dark elongated face with a compact ornamental group and smaller motifs toward its end. The museum identifies inlaid bronze, an approximate nineteenth-century date and the gift of Mather College Library. The displayed dimensional statement supplies 1.4 centimetres under an overall heading. In the absence of a named axis for that isolated figure it would be unwise to publish it as the length of the entire handle. The image shows its proportions clearly. The webpage does not give a complete set of dimensions from which all three physical extents can be inferred.

This is a useful kind of friction in a museum record. The field has not failed simply because it gives us less information than we want. It supplies a figure attached to an identified accession, which can be checked against a fuller register or a subsequent measurement. A modern writer who silently substitutes a more plausible-looking length may leave readers worse off, since later users cannot tell where the figure came from. It is often enough to omit that measurement from a comparison and retain the more secure information about the object’s classification, appearance and documented source.

Notice how different the problem is from the one posed by the tortoise assembly. There the immediate question was whether the measurement, if supplied, would refer to the handle alone or the entire blade and handle. Here the object class is clear but the axis requires clarification. “What was measured?” has at least two parts: which components were included, and in which direction did the measurement run? Very short museum captions can answer both. Very long descriptions can leave both open if they reproduce the numerical value without attending to its field.

Changing custodians and changing words

The European catalogues discussed here represent different sorts of institution. The Burlington Fine Arts Club brought works into an exhibition and supplied terminology to a viewing public. The Japanese government commission presented works in a London exhibition setting and issued an illustrated record. The Ashmolean preserves a collector’s bequest and makes a scholar’s catalogue accessible alongside an object database. A word encountered in all three contexts has passed through different hands. Apparent agreement across publications can reflect the durability of a useful term as easily as it can a new and independent examination of an individual object.

The Cleveland photographs make the distinction tangible. An exhibition curator could group the examples by animal subject, a metals specialist by the medium recorded for each, a registrar by the donors and a translator by the extent of the object denoted in the title. All four groupings are historically meaningful. The danger comes when a fact established under one system is allowed to migrate into another. A common donor is not a common maker. A shared Japanese term is not proof of an identical assembly. A comparable size on the website is not a common physical measurement.

Some older authors use romanized words with different spacing, punctuation or treatment of long vowels. A transcription intended for identification should preserve the wording used in the particular catalogue and, if helpful, add the normalized modern term separately. That way a reader can find the historical entry under its original spelling while understanding which component the present writer means. Modernization becomes a problem only when it erases evidence of a difference in classification or gives an old author a precision that was actually supplied by a later museum editor.

An honest translation can be quite short

Consider three possible labels for the examples above. For Cleveland 1940.673, “knife handle with chickens” expresses both the extent and the subject. For 1992.374, “small blade with its decorated handle” prepares a reader to see an assembly and allows the parenthetical Japanese term to specify the fitting. For Oxford EAX.10945, “guard with a lined accessory opening” captures what the keyword search was actually matching. Each is economical. None requires pretending that one English word can do every job performed by kozuka across the historical records.

Once those differences are visible, the decorative surfaces can be enjoyed without the constant risk of losing track of the object. Chickens on a dark rectangle, tortoises on a handle at the end of a long blade, and a pattern of bamboo around an empty opening no longer compete to supply the one correct illustration of a dictionary term. Each tells us something about the habits of a collector, a cataloguer or a museum photographer. The small scale of the pieces has made their words travel particularly far. Their accession numbers, complete views and historical catalogue entries allow the words to find their way home.

A final advantage of keeping the complete views is that disagreement becomes possible without confusion. A scholar may propose a more exact material for an object catalogued merely as metal, a different reading of an inscription, or another translation for the handle in a nineteenth-century list. With the entry number, accession and extent of the pictured assembly on record, the reader can test that new claim on the same object. That is a more durable achievement for a catalogue than making every historical term sound perfectly consistent in English.

Questions about catalogue translations

What is the difference between a kozuka and a kogatana in the Ashmolean glossary?

The glossary identifies the kozuka as the handle holding the small knife called the kogatana. The surrounding entry establishes which component is actually catalogued.

Why does a search for kozuka retrieve the bamboo guard in Oxford?

The entry describes the rim of an accessory opening. The registered object, EAX.10945, is a guard, and the museum counts one item.

Which Cleveland photograph shows a handle with chickens?

Accession 1940.673 is catalogued as a knife handle with a family of chickens. Its record identifies shakudō and gold.

Which Cleveland objects are explicitly titled blades with handles?

Accessions 1992.374 and 1992.375 are described as blades with handles. The former has a family of tortoises in its title.

Which London plate numbers does the New York Public Library identify as knife grips?

The library’s note identifies 389, 398, 401 and 407. They are exhibition catalogue item numbers.

Does 1.4 centimetres securely establish the length of Cleveland 1947.230?

The online record gives a single figure under an overall heading without a named axis. That is insufficient to identify the length of the entire handle.

Museum films

JASA: Japanese Art Society of America — Samurai Splendor: Sword Fittings from Edo Japan (Lecture by Markus Sesko)

The Met — The Japanese Sword and the Japanese Aesthetic

Collection records and further reading

Related Articles