An old karate master sits on a mountainside above the Motobu peninsula of Okinawa and tells a young Englishman a story. Two gods, he says, once fought each other hand to hand — Ka-shi-ma and Ka-chu-ri — and from the way they moved came the art. The young man, who is Dukes, has read almost the same story in a manuscript named the Ta-ka-no-kabi, preserved in Japanese Ju Jitsu schools and in the Ryukyuan Kempo Hishu, his own temple's chain.1 The master can recite the names but cannot say what they mean; the meaning, Dukes believes, fell out of the story centuries ago when it crossed from Sanskrit into Chinese into Japanese, leaving only the sounds. This page is about a legend that survived as accurate noise — and about Dukes's attempt to reverse-engineer the sense back out of the noise, an attempt that runs ahead of any evidence but his ear.
The Ta-ka-no-kabi is the name Japanese Ju Jitsu schools give to a legendary manuscript account — brought, the tradition says, from India via a Tibetan monastery into China — recording a hand-to-hand combat held between two deities whose names are given phonetically as Ka-shi-ma and Ka-chu-ri.2 The account describes their movements and says they used these techniques "to control and restrain their followers."3 Dukes encountered the same story orally on Okinawa, told by a karate master who held the names but not their sense.4 He treats it as a textbook case of a living tradition "passed down from generation to generation among people who don't really understand its constituents, but who nevertheless still retain accurate elements of an earlier Chinese tradition."5
Then Dukes does the thing the master could not: he reconstructs the meaning. The page records both halves — the legend as received, which is the firm part, and Dukes's phonetic reconstruction, which is the soft part and is flagged throughout.
Here is Dukes's chain. Taka no Kabi, he says, literally means "the giving and receiving of the high(er) places" (Chinese: Kao Cha Li), and this represents the Sanskrit Devaloka dana adana — "the heavenly realm of those who give and those who receive" — "a meaning almost the same."6 The deity names Kashima and Kachuri he reads as Chinese transliterations of the Sanskrit Buddhist term Ksatre(ya) ksetra — "the place, or land, of the Ksatreya," both a synonym for India and a name for "a place where warriors train and exercise control."7 The whole name, he concludes, likely represents a lost Sarvastivada source, something like Devaloka danadana Ksatreya ksetra, which — "if the tradition is accurate" — passed from the Vikramasila monastery of India, the main destination for Tibetans coming to be taught.8 He adds one more, openly tentative: the region of India holding the most Sarvastivada/Mahasanghika monasteries was named Danakataka, "a word which can be translated fancifully as the 'gift of the closed hand.'"9
[CONTESTED — author's own reconstruction, no independent corroboration.] This is Dukes's ear at work, not a documented philology. Each leg of the chain — Taka no Kabi → Devaloka dana adana, Kashima/Kachuri → Ksatreya ksetra, the lost Danadana Ksatreya ksetra source, the Danakataka "closed hand" — is a phonetic resemblance that Dukes himself hedges ("probably," "seems likely," "if the tradition is accurate," "translated fancifully"). The reconstruction is internally elegant and may even be right in spirit, but it rests on sound-similarity across three languages and two scripts, with no surviving manuscript, no inscription, and no second scholar to check it. Filed as the author's own — documented in full because it is the point of the passage, endorsed as fact by no one, including, in his careful hedging, Dukes.
Strip the etymology away and a sturdier argument remains, and it is the reason the page is worth keeping. Dukes's real claim is not "I have decoded the names." It is "the names survived intact across centuries of people who didn't understand them, which tells you something about how oral traditions actually carry information." The karate master could recite Ka-shi-ma and Ka-chu-ri precisely while having no idea they might once have been Ksatreya ksetra. The sounds were preserved because they were treated as sacred fixed tokens rather than as meaningful words that could be paraphrased and drifted. Meaningless-to-the-carrier is, paradoxically, a preservative: you can't improve or modernize what you don't understand, so you transmit it verbatim.
That is a genuine and testable claim about transmission, and it stands whether or not the specific Sanskrit decoding holds. A tradition can be a high-fidelity carrier of sound and a zero-fidelity carrier of sense at the same time. The fossil keeps its shape; the soft tissue is long gone. Dukes's etymology is an attempt to reconstruct the soft tissue from the shape — and the honest reading is that the shape is solid evidence and the reconstruction is informed guesswork.
One phrase in the legend repays close reading: the two deities used their combat techniques "to control and restrain their followers."3 Notice what it does not say. It does not say the gods fought to kill, or to win, or to conquer each other. The combat is framed as a governance technique — a way of managing and disciplining a following. That detail dovetails with Dukes's reading of Ksatreya ksetra as "a place where warriors train and exercise control," and it is the part of the reconstruction that needs no Sanskrit to support it: the legend's own internal logic already presents the martial art as an instrument of ordered authority rather than of destruction. Even if every etymological leg fails, the surviving narrative content independently encodes the book's master theme — that the warrior arts are tools of restraint and control, the Bodhisattva's holding-back, not the conqueror's release. The legend says it in plain story before Dukes says it in reconstructed Sanskrit. That convergence is the strongest thing on the page, because the two registers were transmitted separately and still point the same way.
You have the manuscript open and the temptation is to follow Dukes straight down the etymology — it is beautiful, and beautiful chains are seductive. Stop. First you sort the firm from the soft. Firm: a named legend (Ta-ka-no-kabi) exists in multiple Japanese schools and in the temple chain; an Okinawan master told the same story orally; the deities' names are recited precisely; the narrative says they fought to "control and restrain their followers." You can build on those. Soft: every Sanskrit decoding. You mark each with the author's own hedges — "probably," "seems likely," "fancifully" — and you do not let any of them become a premise for anything else. Then you ask the load-bearing question the firm half actually supports: how does a tradition keep a sound perfectly while losing its meaning entirely? You watch the karate master answer it without knowing he has — reciting flawlessly, explaining nothing. What survives your sort is smaller than Dukes's reconstruction and stronger: a real demonstration of high-fidelity, low-comprehension oral transmission, with the etymology bracketed as a hypothesis no one can currently test.
Evidence base: the legend is attested, Dukes says, in the Ryukyuan Kempo Hishu, the Itosushi, and "other Japanese manuscripts," plus his own oral encounter on Okinawa.2 The Kempo Hishu is one of his temple-chain sources — privately held, uncheckable by outsiders — so the textual attestation carries the standing Kongoryuji caveat. The oral encounter is first-person and vivid but singular.
Tension, preserved: the firm legend and the soft reconstruction sit in the same passage and Dukes does not visibly separate their epistemic weight — he moves from "the manuscript records a combat between two deities" (reportable) to "Kashima probably represents Ksatreya ksetra" (his conjecture) in adjacent sentences, with only soft modal verbs marking the seam. This page makes the seam explicit; Dukes leaves it implicit.
Open seam: the legend says the manuscript came "from India, via a Tibetan monastery, into China," and Dukes routes it specifically through Vikramasila — but Vikramasila is his inference from "Tibetans mainly came here to be taught," not a detail in the legend itself. The legend gives a vague Tibetan way-station; Dukes supplies a named monastery. Whether that naming is recovery or invention is unresolved.
This page is Dukes at his most characteristic and most exposed. The Kongoryuji manuscript chain is his deepest well and his least checkable one, and here it feeds a story whose meaning he has to reconstruct because the carriers lost it. The result is a double layer of insider warrant: the source is private, and the decoding is personal. Against the vault's Samuel/White critical-historical corpus the contrast is sharp — Samuel would treat the surviving sounds as data and the Sanskrit reconstruction as exactly the kind of speculative etymologizing his method is built to refuse, since sound-similarity across languages is the classic generator of false cognates. Yet Dukes is doing something the critical historians sometimes can't: taking a living oral tradition seriously as a carrier in its own right, attending to what the karate master actually preserved rather than dismissing him for not understanding it. The convergence is on the fidelity point — even a hostile reader has to grant that the names were transmitted precisely. The divergence is on the decoding — where Dukes hears a recoverable Sanskrit original, the critical method hears only the very thing it warns against. Both readings are in the passage; this page keeps both.
Karma and Samskaras: The Memory of Consciousness — The samskara model describes how a mind carries forward the shape of past activity as a fixed imprint that runs without being understood — a groove cut once and replayed indefinitely, the doer long gone but the pattern intact. The Taka-no-Kabi legend is that same structure operating at the scale of a culture rather than a single mind. The karate master reciting Ka-shi-ma and Ka-chu-ri without comprehension is a community-level samskara: an imprint transmitted with high fidelity precisely because no one re-examines it. Put the two pages together and an insight emerges that neither holds alone — the property that makes samskaras durable (they run below comprehension, so they aren't edited) is the same property that makes oral traditions faithful carriers of meaningless-seeming tokens. Fidelity and unconsciousness are not opposites; in both the individual psyche and the transmission line, not understanding is what protects the imprint from drift. The decoding problem Dukes faces with the legend is the same problem a practitioner faces with a samskara: the pattern is perfectly preserved and its original meaning is exactly what's missing.
Bodhidharma: The Revisionist Biography — Both pages are Dukes reading legend as compressed information rather than as noise to be discarded. On the Bodhidharma page his method is decompression — Hakuin's gory verse resolves, line by line, into geography and doctrine. Here he attempts the same decompression on the deity-combat legend, resolving the sounds into a lost Sarvastivada title. The pages share a method and differ in evidentiary fate, and that difference is the insight. Bodhidharma's verse decompresses against a cited external commentary — Dukes can show his work. The Taka-no-Kabi decompression has no commentary, no second source, nothing but Dukes's ear; it is the same operation run without a check. Setting them side by side calibrates the method itself: legend-as-compressed-information is a powerful reading when there is an independent key to decompress against, and a speculative one when the only key is the reader's own reconstruction. The Bodhidharma page shows the method's best case; this page shows its limit.
Sharpest implication: an oral tradition can be a perfect carrier of sound and a total loss of sense simultaneously — which means the fidelity of a transmission tells you nothing about whether its meaning survived, and a community can guard a treasure for centuries without holding the key to it.
Generative questions: