Shosoin Straight Swords and Warabitedo: Nara Treasury Blades, Inlaid Mountings, and Early Curvature

A curl at one end of an eighth-century sword

In the Tokyo National Museum there is an iron sword whose most eloquent feature occupies very little of its length. The end of the grip turns back like the tip of a young fern. The museum identifies the piece by an excavation locality in Hokkaido and by a number, J-34274. Far to the south, an eighth-century document at Todaiji describes an imperial sword by the colors of its straps, the lining of its bag, and the disposition of gold and silver. It even distinguishes the shape of the point. Each description can tell us something that the other cannot. Taken together they change the question we ask about early Japanese swords. Before trying to find the first ancestor of the curved tachi, we can ask who kept these objects, what they saw in them, and why anything survives for us to see today.

The fern is an especially good place to start because it is so easy to misread. A warabite sword takes its name from a feature of its handle. The word does not prescribe how much its blade must curve. A straight-looking blade and a markedly curled pommel can belong to the same object without contradiction. Conversely, a curved silhouette photographed without its grip cannot establish membership in this family. English spellings vary, including warabitetō and warabitedo. The Japanese characters for bracken and hand point back to the same sprouting shape. In the Tokyo museum's English catalogue the plain phrase “curled pommel” does more explanatory work than an elaborate theory of the origins of the katana.

Curled-pommel sword from Moizari No. 11, Hokkaido, J-34274
Curled-pommel sword from Moizari No. 11, Hokkaido, J-34274. Tokyo National Museum, ColBase.

What the Todaiji clerks actually recorded

The great donation associated with Empress Komyo after the death of Emperor Shomu in 756 gave Todaiji things of very different kinds. A written account traveled with the gift. Its list of swords is available in transcription through the Tokyo National Research Institute for Cultural Properties. It opens with a heading for a hundred swords. The items beneath that heading range from individual objects to groups whose shared characteristics could be described once. The opening number belongs to a documentary act in the eighth century. It cannot be used as a current count of complete swords in the Shosoin, still less as the number of identical swords that once equipped a company of guards. An inventory has a purpose, a date, and a grammar of its own.

Consider the entry for the sword known by the formidable name Kingindenso no kara-tachi. The clerk gives the blade length as two shaku, six sun and four bu. The point has edges on both sides. The handle covering, mountain-shaped forms and vine ornament are specified. White leather and purple leather perform different roles in the suspension. The list ends with the appearance of the bag, a red-ground patterned fabric with a pale green lining. Read aloud, the sequence moves from steel through skin and precious metal to fabric. In the donor's accounting all these materials belonged to the identity of the gift. A modern steel-only display would suppress some of the very details by which the recipient could recognize it.

The entries to either side show why that specificity mattered. Another decorated sword is assigned a slightly different blade length and a different combination of suspension and bag. Several have comparable points. The fact that an object has a decorated scabbard and two edges near its tip would leave more than one candidate on the page. A combination including a particular textile and color narrows it down. Today an institution might use an accession number and photographs for that job. The clerk had dimensions, counted units, colors and positions of ornaments. A reader who treats the fabric notes as incidental luxury misses part of the original system for telling things apart.

The length is a useful trap for a modern interpreter. A table that silently applies a familiar modern value for the shaku will produce a neat decimal figure, complete with a spurious impression of precision. Historical rulers, units and conventions need to be established for the context in question. A measurement of a surviving blade also needs its endpoints recorded. Overall length with mountings, blade length alone, and the length of a modern replica are three perfectly legitimate numbers that answer different questions. Keeping the original units in the transcription preserves information that can be tested against other evidence. Conversion is a second operation with assumptions of its own.

Red-stained ivory ruler, N-83, Tokyo National Museum, via ColBase
Red-stained ivory ruler, N-83, Tokyo National Museum, via ColBase. Tokyo National Museum, via ColBase.

Why colors and linings are historical evidence

Imagine checking a consignment delivered without electric lighting or photographs. Little about the visual distinction between a fine red bag and a pale lining would have been trivial. Opening the bag revealed differences that remained invisible when it was closed. A worn strap could be replaced while a distinctive metal fitting stayed with the sword. Two pieces that looked much alike in storage could have quite different histories of repair. The inventory records a moment when particular parts belonged together. It does not promise that every part would remain attached throughout the next twelve centuries. Scholars gain more by preserving that distinction than by treating every modern photograph as an illustration of the ensemble in exactly its original state.

The record also preserves information about the donor's community. Someone could distinguish kinds of weave and expected the recipient or the recipient's staff to understand those distinctions. Leather of similar appearance could be called out separately depending on where it belonged. Textiles could be important enough to describe after an expensive blade had already been counted. These observations bring several largely anonymous professions into view. We do not know the names of the people who cut each strap, prepared the bags or received the bundles, but their expertise is implicit in the careful differentiation of accessories. A sword described in this way was already the work of more than one craft community before it entered the treasury.

There is a comparable problem with the word Tang. It matters whether a writer used a name for a style, an appearance or a particular object, and whether a modern translator has added a claim about where the object originated. The register contains compounds that do not all use identical wording. Objects, craftsmen, designs, words and raw materials did travel around East Asia, sometimes by different routes. Evidence that a motif was admired at the Nara court is excellent evidence of taste there. Assigning the metal, the maker or the entire assembly to a particular foreign workshop requires a different kind of demonstration. An object can bear several histories of movement at the same time.

The photograph, the treasure and the replica

One striking feature of a modern search for a Shosoin sword is the number of credible museum results that lead to reproductions. They deserve attention in their own right. Nara National Museum catalogues a group of reproductions of treasury swords under collection number 253. Its record for a silver- and gold-decorated example distinguishes the work in the collection from the older model. The Japanese exhibition on recreating Shosoin treasures dates this project to the early Meiji era and identifies the Nara exhibition company behind the 1875 copy. The date is part of the object's historical value. Meiji observers recorded what they thought essential about a much earlier object and undertook to translate that observation into a new work.

The decoration of the model for that copy includes clouds, running animals and plant scrolls distributed across the scabbard, with metal fittings punctuating its length. Decoration changes scale as the eye moves from the whole outline to an individual motif. Seen from across a room the fittings break up a long surface. Seen close to, the scrolling forms draw neighboring details together again. The museum reproduction makes these two ways of seeing available even when the original is elsewhere. Its existence does not entitle us to place a photograph of the copy beside an ancient date without a caption. Giving the two dates can turn an apparent problem into a miniature history of the study of craft.

A particularly good comparison is an eighth-century red-stained ivory ruler in Tokyo, from the Horyuji treasures. It belongs to a different collection and has a different function. The tiny decoration on the surface and the intervals along the edge bring together ornament and the articulation of space. It shows the sort of material culture within which an inventory could record divisions smaller than a finger's breadth. It does not supply the missing calibration for every number in the Todaiji list. Precisely because its origin and collection history are identifiable, the ruler can be used as a contemporary comparison without pretending that it once sat on the clerk's desk.

The Shosoin kara-tachi with gold and silver decoration, photographed in the 1932 Imperial Museum catalogue
The Shosoin kara-tachi with gold and silver decoration, photographed in the 1932 Imperial Museum catalogue. Imperial Museum, Shosoin Gomotsu Zuroku, volume four, 1932.

What a treasury selects

The familiar exterior view of the Shosoin encourages a visitor to imagine that time simply stopped inside the building. Yet the existence of the donation record, supplementary annotations, modern replicas and published technical studies points in the opposite direction. People selected objects, recorded what they saw, kept track of movement, inspected materials and found ways of showing fragile things. Keeping something for over a millennium is a succession of decisions. A picture of the repository establishes a setting for that work. It cannot establish uninterrupted residence in one compartment for every item that appeared on one page of a donation list.

This matters when the word survival enters the story. A precious court mounting that remains recognizable to a modern researcher benefited from an institution able to preserve organic materials as well as metal. A burial environment may leave us iron while changing its outline or removing an attached organic covering. Neither object represents the average state of an eighth-century sword in use. The different histories of preservation have sorted the evidence before the historian ever encounters it. The temptation is to contrast a court that loved brilliant colors with a northern society that preferred bare iron. That contrast would quietly turn a difference in preservation into a difference in human preference.

The hunting ornament on a Shosoin silver vessel provides a useful control on a different inference. Precious metal, line and animal motifs could articulate a curved surface far removed from a sword scabbard. Designers working in different media faced related visual problems. There is no need to imagine the same artisan making both pieces, or to assign a specific animal motif to a single kingdom. Following ornament from metal vessel to scabbard opens a wider question about the circulation of designs. It also disciplines the comparison. The evidence is an observable treatment of surface, not an undocumented journey by a particular craftsman.

Tamura Sokichi’s replica of the decorated Shosoin kara-tachi, Tokyo National Museum F-19823
Tamura Sokichi’s replica of the decorated Shosoin kara-tachi, Tokyo National Museum F-19823. Tokyo National Museum, ColBase; work by Tamura Sokichi. License.

Five entries in a Tokyo exhibition

The Tokyo National Museum's exhibition on warabite swords makes a small but revealing exercise in reading a checklist. It lists five entries. The Hokkaido sword comes from the Kashiwagi Higashi site, specifically the Moizari Number Eleven burial, at Eniwa. The museum assigns it to the eighth century and uses the local archaeological designation Satsumon. A second object comes from Tokushima on Shikoku. Two more refer to Nagano locations, one accompanied by a qualification about its reported findspot. The last item is a twentieth-century reproduction whose ancient model lacks a known findspot. Counting all five dots on a map as five equivalent ancient discoveries would distort an exhibition that took care to explain their differences.

Notice how even the period names shift as the eye travels down the list. Satsumon is appropriate to the Hokkaido find. Asuka and Nara appear as chronological terms for works elsewhere in the archipelago. A common century allows comparison across this geography without forcing the northern archaeological sequence to adopt the administrative vocabulary of the capital. This is why both a name and a date belong in a good caption. The same date can describe societies with different burial customs and different relations to a court. The same word on two labels need not imply that the objects arrived at the museum along the same route.

The museum describes the prominence of such finds in northeastern graves and the existence of finds much farther west. Among the explanations discussed for some western examples is the displacement of people known in court records as Emishi. The wording matters. An explanation for a pattern of distribution is something different from a biographical identification of the owner of a particular unsigned sword. Objects could travel with people, pass between them, or remain in a family across more than one generation. A findspot establishes where an object was deposited or discovered. It does not, all by itself, tell us what the person who made it called themselves or where they learned their craft.

Is this the beginning of curvature?

A textbook silhouette can give early sword history a deceptively satisfying direction. A straight line occupies the beginning of the page and a deeply curved tachi the end. In between, it is tempting to place a northern sword and draw arrows. The actual pieces resist that arrangement. Curling at the pommel, an angle where grip meets blade, and a continuous curve along the back are separate observations. Each can look stronger or weaker as a photograph is rotated or foreshortened. Archaeological surfaces also carry the consequences of damage and burial. A claim about the history of blade curvature has to specify the part being compared and the condition in which that part has survived.

Consider what even a convincing sequence of dated objects would leave open. A form could appear in more than one place, remain in use after another form appeared, or disappear from surviving deposits because the custom of burial changed. A well-preserved court gift could be older than some apparently simple grave finds. Nor would establishing a sequence of dates automatically establish a line of instruction between workshops. Continuity is a historical claim requiring intermediaries, geographical context and comparison of several characteristics. It is more interesting to ask which characteristics travel together than to confer the title of first katana on a striking early piece.

The metal itself offers a final lesson in the limits of the photograph. An excavation image can reveal an overall silhouette, an interrupted contour or the outline of a surviving metal grip. An inventory can report a named surface material that is difficult to see today. Neither source on its own provides a chemical composition, a hardness figure or the internal structure of every part. Conservation studies can ask material questions of a specific object. Their answers carry the identification of the specimen and the limits of the method used. Generalizing results from a modern replica to all of the original treasury swords would collapse precisely the distinctions the collection labels preserve.

Two different ways to meet an eighth-century person

The most vivid encounter in the Todaiji document may be with a person who noticed a pale green lining and judged that it belonged in a permanent account. On the northern side it may be with people who placed a recognizable object in a particular burial at Eniwa. Neither encounter gives us a personal name. One rests on written choices and the other on a documented archaeological context. Between them stand modern excavators, registrars and conservators whose choices make both encounters possible. Their labels repay the same attentive reading as the objects. A reported findspot, a date assigned to a copy or a qualifier before a place name can alter the whole story we tell.

To bring these early swords back into focus, put three lines under each image in an imaginary exhibition. Use the first to say what survives, including the particular feature that prompted the identification. Use the second for the earliest securely associated record or the recorded place of excavation. Reserve the third for what remains unknown. For the Nara treasure those lines will speak of a donation, precious materials and the limits of identification across time. For the sword from Hokkaido they will speak of a curled pommel, an eighth-century burial and the limits of what that find can say about later swords. The reward is two sharply drawn histories, each more revealing than a single arrow from straight to curved.

Frequently asked questions

What is the Shosoin sword inventory?

The Todaiji donation list of 756 records groups and individual objects, including swords, mountings, straps and bags. It describes a gift at a particular date.

What makes a warabite sword recognizable?

Its name refers to the curl at the end of the grip, reminiscent of an opening bracken frond. Blade curvature is a separate feature.

Where was the Tokyo National Museum warabite J-34274 found?

It comes from the Moizari No. 11 burial at the Kashiwagi Higashi site in Eniwa, Hokkaido. The museum assigns it to the eighth-century Satsumon period.

Were all warabite swords found in northern Japan?

Northern finds are prominent, but the Tokyo exhibition also documents finds in Nagano and Tokushima. The quality of findspot information varies by object.

Are all Shosoin sword photographs images of ancient originals?

Some show modern reproductions. Nara National Museum identifies its replica group under collection number 253, so the date of the photographed work matters.

Does the curled pommel prove direct descent to the curved tachi?

The pommel and the curvature of the blade describe different parts. A historical relationship would require dated intermediate evidence beyond resemblance.

Films and talks

The Met — Art of the Samurai—An Introduction

The Met — Conserving Japanese Works of Art in Foreign Collections

Collection records and further reading

Related Articles