Habaki and Seppa in the Museum Drawer: Cataloguing Loose Fittings

Two entries in the Birmingham Museum of Art’s collection look remarkably alike until you reach the measurements. Both concern Japanese metalwork dated around 1800. Both give the same medium, mercury-gilded steel. Both record that something from the Ray Deadman collection went on loan to the museum in 1994 and was purchased in 2017. One object, however, is a sleeve with a depth of about a centimetre. The other is a spacer whose published dimensions describe two much shorter spans. Their accession numbers end in five and six. If you encountered them in a drawer with their labels turned over, what would tell you that you were looking at two different kinds of evidence?

The museum calls the first a habaki and the second a seppa. These Japanese terms survive the move into English better than one might expect. A translation such as collar gives a fair idea of the first form, but another collar on another part of a Japanese mounting can also be called a collar in English. Washer evokes the relative thinness of the second form while saying almost nothing about the care that an artist might have taken over its surface. The Japanese words help a reader return to the same pieces across catalogues whose English titles take different approaches.

What fits on the front of a card

Birmingham’s habaki bears the number 2017.14.5. Its entry gives three dimensions, approximately 3.8 by 2.5 by 1 centimetre. The English title calls it a scabbard sleeve. A researcher who knows the term blade collar might initially search under a different noun and miss the record altogether. The two phrases approach the same small object from different directions. One emphasizes its relation to the scabbard, the other its location on the blade. Keeping the Japanese term alongside the descriptive English title is a practical way to recover both searches without requiring either institution to give up its vocabulary.

The seppa, 2017.14.6, is catalogued as a metal spacer. The reported dimensions are 1.9 by 2.2 centimetres. No third measurement appears in the public entry. In a spreadsheet it is surprisingly easy for a missing third measurement to become zero when a column is converted to numbers. That would have the effect of making a real piece of metal physically disappear in one dimension. A blank is the right result here. One can compare the lengths actually recorded, and say which card gives a depth, without pretending the published figures answer a question about the thickness of the smaller object.

Gold-colored metal collar with dark inner opening, photographed on a textured surface
Comparative photograph of a habaki. It illustrates the depth of a collar; the separately catalogued Birmingham example has its own material description and accession. MathieuMD, CC BY-SA 4.0. WebP reproduction.

A photograph of a separate gold-colored habaki illustrates what that third dimension contributes. The camera looks across the opening of the collar. It would give a different impression if the same piece were photographed straight down with no indication of depth. This is a comparative image by an identified photographer, not a photograph of Birmingham’s accession. Its role is to make a general question about shape visible. The material description of the Birmingham object still comes from Birmingham. A warm reflection on a screen cannot tell the reader the composition of an object held in a different collection.

The two Birmingham entries distinguish unsigned work, the absence of recorded marks and the absence of an inscription. That may look like an elaborate way to leave three boxes empty. In fact it lets a future researcher ask three separate questions. A number or sign can appear without naming a maker. A short dedication might identify an owner without identifying the person who made the piece. A signature is one particular kind of writing with a special claim about authorship. The card that separates these possibilities has already done something important for the historical study of anonymous work.

What a guard can tell us about a missing spacer

An openwork Japanese guard now in the Cooper Hewitt collection provides a useful change of scale. The eye first notices the negative spaces in the design. Look again and the comparatively plain central area becomes a subject in its own right. The term seppa-dai refers to an area of the guard associated with the spacers. A visitor who remembers only the first half of the term might count this flat area as a separate seppa. The photograph makes that mistake easy to understand and equally easy to correct. The area belongs to the object being photographed.

Circular openwork iron guard with a flat central area around a shaped opening
Seventeenth-century pierced guard in the Cooper Hewitt collection. The flat area at its center belongs to the guard itself. Cooper Hewitt, Smithsonian Design Museum, Public domain. WebP reproduction.

It is worth thinking through what the distinction means for a museum inventory. If a researcher wanted to count all loose spacers in a department, guards whose descriptions mention the central seat would not automatically qualify. A search for the string seppa might return both kinds of entry. If the researcher instead wished to study the area left relatively plain on openwork guards, many apparently irrelevant search results might become valuable. The usefulness of a word changes with the question asked. It helps to record the component named in the title before using search counts as evidence for a collection’s historical preferences.

A photograph can mislead in the opposite direction as well. The absence of a separate spacer from a picture of a guard does not show that no spacer ever belonged to the historical mounting. The photographer may have been documenting the guard alone; a museum may have acquired one fitting after the rest had already dispersed. It is also possible that companion objects survive under other numbers or at another location. These are distinct hypotheses. To move from one to the next requires a link in an inventory, an early photograph, a description of a group, or some other record of association.

A readily available drawing can supply nomenclature without solving provenance. Near the end of the nineteenth century the Burlington Fine Arts Club exhibited Japanese lacquer and metalwork and published a catalogue that discussed the smaller fittings. Its English vocabulary includes a washer or thin plate for the seppa and a separate description for the habaki. The publication is evidence of what an English-speaking collecting audience was taught to call these things at that time. It cannot settle which terms a modern Japanese curator prefers, still less determine the earlier history of a particular Birmingham or New York accession.

When a number belongs to a set

The Saint Louis Art Museum takes a different approach in its record for a nineteenth-century sword set. One museum number, 394:1923, covers a group described in gold and enamels. The dimensions are divided into lines for the kashira, fuchi, habaki and kojiri. One can therefore cite a measurement for the habaki without claiming that it has a stand-alone accession like Birmingham’s. The number describes the collection’s chosen unit. The separate measurement lines describe what the cataloguer actually measured. Both scales of description matter, especially when comparing data from more than one institution.

There is a small grammatical warning in this example. The medium appears at the level of the set. If one wishes to state exactly which material occurs in a particular place, one needs a more detailed account than a medium field shared by several pieces. It would be unsafe scholarship to divide the named materials evenly among the four and call the result a component analysis. The useful observation is simpler. The set can be identified under one number, while different parts remain individually visible in its written record. A researcher may need to refer a reader to a specific line rather than just to the headline.

Which museum owns more fittings? Comparing the two Birmingham object numbers with the one Saint Louis number would be a particularly poor way to find out. A single museum record can cover several physical pieces. A component within a set can be measured separately even when the online interface has no separate page for it. Different public databases also choose different levels of detail for related works. To make a meaningful comparison one would need to declare a counting unit. An accession, a physical object, a named component and a photograph are four plausible choices, with markedly different results.

A drawer imagined from two acquisitions

Let us return to Birmingham’s dates. The two pieces were on loan in 1994. Both were purchased from the Deadman collection in 2017, with credits commemorating William M. Spencer III. These matching details make a strong argument for telling their modern collection stories together. They give no name for a patron in Edo Japan. Close accession numbers show the museum recording acquisitions in relation to one another, but a prefix and two adjacent suffixes are not documentary proof that the same historical owner commissioned a fitted pair around 1800.

One way to see the value of that limit is to try writing the label for a loan exhibition. A defensible caption would say that the sleeve and spacer share modern collection history and were both catalogued with the same approximate date and medium. A much bolder caption would call them surviving parts of a single named sword. That would be a claim about a time long before the documented loan. It would require an older record identifying both pieces in relation to that object. The difference between the two captions is small on the printed page and large in its effect on everything a visitor learns.

The geographical label needs care too. The Birmingham in these records is in Alabama. Without that detail an English reader could readily start looking in the collections of a different city. A complete reference need only name the institution, its location and the object number. The maker remains anonymous, the country is given as Japan and the earlier period is identified as Edo. Following the works from Japan into a collection in the American South raises interesting questions about collecting. It supplies no justification for an invented account of the journeys taken in between.

What a drawing preserves

A very different sort of drawer is represented by Ranzan Tsuneyuki’s personal album of designs for metal carving. The Metropolitan Museum dates it to the middle of the nineteenth century. On the illustrated leaf two designs occupy a page that also bears writing. A drawing can preserve an arrangement even when no completed example has been located. An investigator studying a loose fitting might use the album to compare an outline, the arrangement of an opening or the relation of ornament to an unadorned central space. The drawing is evidence for a design, which makes it valuable even if a matching piece of metal never turns up.

Page of Ranzan Tsuneyuki’s album with two round designs and notes
A leaf from Ranzan Tsuneyuki’s design album, accession 36.120.699. Designs in a sketchbook and detached metal objects enter inventories differently. Image courtesy of The Metropolitan Museum of Art, Open Access.

Two round shapes on a page also expose a difficulty with the word pair. Designs can be drawn together as alternatives. One can record the opposite faces of one object. Another page might bring together two related but separate objects. Before deciding which relation a particular artist intended, a reader should check the accompanying text, the arrangement of neighboring leaves and the museum’s account. Simply counting circles gives the wrong kind of answer. The album is especially instructive in a study of small metalwork because the relationship among two-dimensional representations is not identical to the relationship among objects in a drawer.

On Gotō Ichijō’s guard in the Metropolitan collection, the museum’s second photograph provides a second view of the same work. The surface and openings become legible in a different way while the accession remains 43.120.728. A collection search might offer both images independently, leaving the impression of a larger group. The handwritten design, the two-sided metal object and the two digital views each involve a different kind of multiplicity. The visitor who follows the accession through the gallery can bring these forms of evidence together without doubling the number of historical objects.

Dark guard with an oblong opening and gold highlights on the reverse
Second museum view of the Gotō Ichijō guard, 43.120.728. A second view adds evidence without adding another guard to the collection. Image courtesy of The Metropolitan Museum of Art, Open Access.

The drawer label in another language

Imagine lending just the Birmingham spacer to an exhibition abroad. Its registration record needs a short title that can travel with the object. A label in another language might quite reasonably retain seppa, give the receiving museum’s preferred translation in parentheses and name the lender. The condition photograph, however, needs the individual number. The group history belongs in an accompanying note. Giving every document exactly the same heading would make the labels tidy while increasing the chance of mixing up the sleeve and the spacer when the exhibition returns. These are administrative distinctions with consequences for the history one can write later.

Now imagine looking for the item in the printed catalogue a generation after that exhibition closed. The visitor might remember a gold-colored object without remembering whether the label described gold or gilding. Another might remember seeing a collar when the loan was actually the spacer. A stable accession attached to the printed photograph would let both readers recover the contemporary description. It would also give a later scholar a way to document any genuine change in the interpretation of the material. A number looks prosaic beside a good photograph of Japanese metalwork. It becomes rather more eloquent when it allows an argument to survive translation and the return of a loan.

A modern collar with a documented commission

A commission by the Metropolitan Museum adds one unusually well-documented comparison. Miyairi Norihiro made a reconstruction of an early double-edged work in the collection. The commission is dated 2013, and the catalogue explicitly identifies the collar as sterling silver. The museum explains why a new reconstruction could be informative beside an excavated historical original. For our purposes the important point is the separation of original, modern study object and named component. A metal specified for the modern commission is excellent evidence for that work. It cannot automatically identify what was present on the archaeological original centuries earlier.

Here the historian enjoys details that are missing from the Birmingham pieces. A maker is identified; a museum explains its purpose in commissioning the object; one component has a precise material identification. Yet the modern work does not supersede the anonymous ones as historical evidence. The Birmingham pair answers questions about older workmanship and modern collecting that a commissioned reconstruction was never intended to answer. Collections are strongest when different degrees of documentation remain visible. Otherwise it is too easy to mistake the best-documented work for the most typical historical example.

A photograph made for the next reader

In February 2025 a team worked at the British Museum to extend photographic documentation of Japanese sword ornaments. A report published by the Sainsbury Institute describes close and multi-angle approaches. In the latter, eight views of an object helped record the sides that a frontal photograph can lose. The workshop is useful as a concrete example of why a museum might publish visual material beyond the familiar face of a guard. A caption for a spacer presents much the same challenge. The smallest change of viewpoint can turn an apparently flat shape into an object with discernible depth.

The report also records limits of the available equipment and suggestions for future work. Such details belong to the history of museum knowledge. A researcher seeing a particular sequence of images may otherwise assume that every possible orientation was considered and that omissions reflect decisions about artistic importance. In practice choices can reflect the time available, the stability of a support and the difficulty of keeping a small object in focus. A photographic absence deserves to be recorded honestly. It very seldom warrants a categorical claim about how a historical fitting looked on the side the camera never reached.

Our two Birmingham cards can now be read as compact histories rather than near-duplicates. Each preserves an approximate date for manufacture and a firm date for a modern purchase. Each records a particular combination of form, material and name. The seppa prompts a question about the central field on a guard. The habaki leads to a silver collar on a documented modern commission. Both lead to a collection in Alabama where a loan preceded acquisition by more than two decades. The differences fit into a few lines in a drawer. Following them out into the world takes rather longer.

Questions about the fittings and their records

Why do Birmingham’s habaki and seppa have separate numbers?

The museum describes the scabbard sleeve under 2017.14.5 and the metal spacer under 2017.14.6. Separate records preserve their different forms and measurements. The two entries record common collection and acquisition history. They do not identify a historic blade to which both can independently be traced.

What is the difference between a seppa and a seppa-dai in a photograph?

Seppa names a separate spacer. Seppa-dai describes the central seat or area on the guard associated with the spacers. A clear view of a guard can show that area even when the photograph contains no detached spacer. The caption should identify which object actually appears.

Are both Birmingham objects described as copper?

The Birmingham Museum of Art lists mercury-gilded steel for both. The identification belongs to those two accessions. The Met documents a sterling-silver habaki on a modern museum commission. A general description based on another collection or on the apparent color of a photograph should not displace the object-specific medium.

When did the Deadman pieces become museum purchases?

Both records give a loan in 1994 and purchase in 2017. They date the pieces to around 1800. These entries distinguish at least three events: the approximate production date, the arrival on loan and the subsequent acquisition in memory of William M. Spencer III.

Can four separately measured pieces share one museum number?

Yes. The Saint Louis Art Museum describes a nineteenth-century sword set under 394:1923 and lists dimensions for its kashira, fuchi, habaki and kojiri. The materials and acquisition credit appear at set level. Any claim about an individual component should respect how that particular museum has divided the record.

What does a design album contribute to the history of loose fittings?

Ranzan Tsuneyuki’s album preserves designs in a named maker’s own working collection. It offers evidence for artistic choices and for the collecting of drawings. To match one drawn motif to one surviving fitting would require a closer documented correspondence than their shared category or similar outline.

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