Reading the Bansenshūkai · 03
Did the Bansenshūkai Say Ninja Walked on Water?
The famous mizugumo is often imagined as footwear for crossing a pond. The manual’s own entry begins somewhere less spectacular and more useful: with the geometry of a measured ring.
A fresh translation
Explanation of the mizugumo diagram.
The mizugumo figure measures two shaku, one sun, and eight bu across.
Its inner opening is one shaku, one sun, and eight bu, cut as a circle;
the outer band is five sun wide on each side.
Original text
水蜘蛛之圖説
一水蜘蛛圖、身ノ指渡シ、二尺一寸八分、内一
尺一寸八分、中ヲ円クトルナリ、外側幅、一
方ニテ、五寸ツゝナリ、
Words and concepts
- 水蜘蛛 (mizugumo)
Literally “water spider.” The name identifies the device in this entry, but it does not by itself tell us how a person used it. - 圖説 (zusetsu)
An explanation tied to a diagram. The heading directs attention to shape and construction rather than presenting a narrative of an operation. - 指渡シ (sashi-watashi)
A measurement across the object—its overall span or diameter in this circular design. - 尺・寸・分 (shaku, sun, bu)
Successive units of length. The article preserves the source’s units instead of imposing a modern metric conversion on an early-modern manuscript.
Begin with what the entry measures
The first safeguard in reading a famous object is almost embarrassingly simple: describe the evidence before naming the feat. These lines do not show a shinobi stepping onto water. They do not report a successful crossing, name a river, or describe a person’s posture. They give a heading and three related dimensions.
The whole figure is measured across. A circular opening occupies the center. The band surrounding that opening has a stated width on either side. In other words, the passage begins as a compact specification for an annular object. The noun mizugumo may prompt an image, but the verbs and numbers direct the reader toward a measured drawing.
This difference matters because a diagram and an account of use answer different questions. A diagram can preserve intended proportions, parts, and relations. An operational account might explain how weight was placed, how balance was maintained, what depth of water was suitable, or how the device moved. The quoted lines supply the first kind of evidence. They do not supply the second.
The arithmetic forms a quiet check
The measurements are not an arbitrary cluster. The stated outside span is two shaku, one sun, and eight bu. The inner opening is one shaku, one sun, and eight bu. Their difference is one shaku. Divided across the two sides of a ring, that leaves five sun for each side—the width given in the final line.
That internal agreement tells us something modest but important. The transcription describes a coherent geometric plan: outside diameter, inside diameter, and radial band width fit one another. It does not prove that every surviving drawing was executed perfectly, that a full-size object was built from this witness, or that the design performed as later reconstructions propose. It does show that the compiler or copyist was recording more than a picturesque name.
Preserving the Japanese units also keeps the claim proportional to the evidence. Historical standards and the practical interpretation of manuscript measurements can require explanation beyond a short excerpt. Converting the figures into a single modern decimal would look precise while hiding those questions. For this close reading, the relationship among the measurements is firmer than an unqualified metric equivalent.
A construction record is not yet a use manual
The surrounding entry continues beyond the four quoted lines. It lists the thickness of a board and describes fittings, leather, cords, and other details associated with the illustrated assembly. Several points are marked for oral transmission. The public manuscript image also carries drawn views of the object and its parts. Together, these features make the section richer as a construction record while leaving the decisive performance question open.
This is a common difficulty when reading technical manuscripts. A written page may function as a reminder for someone who already received demonstration or oral instruction. It may preserve dimensions that are hard to remember while omitting movements that were easier to show than to write. It may also restrict a detail deliberately. “Not written here” can therefore be evidence about the form of transmission without becoming evidence for whatever later reader wishes to supply.
The Iga City catalog makes this boundary unusually visible. Its description notes that the manuscript gives materials and dimensions but does not state the method of use, treating that use as part of the oral element. A separate museum record for the mizugumo offers a later interpretive tradition in which a person sits astride the center rather than walking on paired devices. That interpretation is useful to identify as interpretation; it should not be inserted into the quoted lines as though the lines themselves said it.
Why the name exerts so much pressure
“Water spider” is a remarkably productive name. In English as in Japanese, it encourages the mind to complete a scene: light feet, a water surface, perhaps a figure moving where ordinary people cannot. Once that scene is familiar, every ring-shaped drawing can seem to illustrate it. The source is then read backward from the image already in the reader’s head.
The remedy is not to dismiss the object. It is to slow down the sequence of inference. First comes the manuscript heading. Next come the proportions and listed components. Then comes the diagram. Only after those are established can a reader compare proposed uses, reconstructions, or museum explanations. Each later step needs its own evidence.
This order also prevents an opposite mistake. Because the passage does not teach water-walking, it would be rash to declare the entry meaningless or fraudulent. The dimensions are systematic. The construction vocabulary is specific. The oral markers show that the written page did not claim to be self-sufficient. A responsible reading can acknowledge the technical character of the record without pretending that the record proves a spectacular performance.
What the four lines do not prove
They do not prove that a shinobi wore one ring on each foot. The quoted Japanese contains no instruction to stand, step, or walk. It does not prove that the object could support a person on open water, because the passage gives no test, load, depth, or account of success. It does not establish how frequently an object of this design was made or used.
The lines also do not prove the later straddling interpretation. That proposal answers some questions raised by the ring and central opening, but it belongs to the history of interpreting and reconstructing the object. The distinction is not pedantic. It lets readers see where the early-modern witness ends and where later explanation begins.
Finally, the passage does not represent every ninja tool or every copy of the Bansenshūkai. It is one labeled entry in Scroll Nineteen of the public Iga City witness. Manuscript diagrams can differ in clarity, annotations can enter through copying, and a catalog can add modern description. Naming the witness and location is therefore part of the argument, not a formality added at the end.
How to read a tool diagram
A useful reading method has four stages. Identify what the object is called. Separate direct measurements from inferred purpose. Note what the surrounding text assigns to oral transmission. Then compare later interpretations without merging them into the manuscript. The mizugumo entry rewards exactly this discipline.
Read that way, the diagram becomes more interesting than a yes-or-no test of legend. It shows an early-modern compiler preserving geometry, materials, and assembly while leaving practice partly outside the page. It also shows how easily a memorable label can outrun the evidence attached to it.
The strongest conclusion is therefore deliberately narrow. The Bansenshūkai records a measured, ring-shaped device under the name mizugumo. These four lines make its basic geometry recoverable. They do not tell us that ninja walked on water—and recognizing that limit is not a failure to solve the mystery. It is the point at which careful reading begins.
Source
Bansenshūkai (萬川集海), Scroll Nineteen, “Ninja Implements II” (巻之十九 忍器二), section “Explanation of the Mizugumo Diagram” (「水蜘蛛之圖説」). The quoted wording and diagram context were checked against the Iga City Digital Museum public transcription and the corresponding manuscript viewer, image 6. The museum’s catalog record for Scroll Nineteen identifies the Edo-period witness and summarizes the distinction between recorded construction and unwritten use; its separate mizugumo object record explicitly presents the straddling explanation as a transmitted interpretation.