You start with a list of holy men. Just names, like a phone book. Then, generations later, someone arranges them: this one taught that one, this lineage came from that founder. The bare list has become a family tree. Then a writer takes the family tree and spins it into stories — Nāgārjuna meeting the goddess on a mountain, Gorakh rescuing his teacher from a kingdom of women. The family tree has become a legend.
White borrows this three-step idea from the French scholar Marcel Detienne, and uses it to run the process backward. From the legends we have, back through the family trees, down toward the original lists — peeling off layers to rebuild a history. He calls it a "hermeneutics of retrieval." In plainer words: digging.
Detienne's argument, from his 1981 book: "raw myth," if it ever existed, started as lists of names. Then spoken retelling turned the lists into family trees. Then written interpretation — once writing existed to freeze and rework things — turned the trees into legends.1
His radical punchline: the myth, "fragile and evanescent, is metamorphosed into something else as soon as it is recounted." So for all practical purposes the original myth doesn't exist. What scholars actually study is always the legend — the retelling, the interpretation — whether they sit inside or outside the tradition.1
White flips this into a tool. The lists of Siddhas, he says, sit in the gray zone between the raw list and the first layer of interpretation.2 By reading between the lines of the legends — the tales of the eighty-four great adepts, the many conflicting lists of the Nine Nāths — he tries to peel back the layers and rebuild a history.2 The method is like archaeology: scrape off the legend to find the family tree, scrape off the tree to glimpse the list, and treat the list as the closest thing to bedrock.
Why does the three-step model work in reverse as a reconstruction tool? Because each step adds a known kind of stuff — so running it backward means subtracting known kinds of distortion.
A bare list is almost interpretation-free — just names (though even a list has an order, and order smuggles in priority). The jump from list to family tree adds relationship claims: who taught whom, who descends from whom. Those are later impositions; the list didn't contain them. The jump from family tree to legend adds story: settings, dialogues, miracles, the goddess on the mountain.
So when White faces a rich legend, he can ask: which parts are story-dressing (subtract), which parts are family-tree claims (note as later interpretation), and which parts are list-residue (names that were probably there at the start)? The bare names survive as the most reliable layer, precisely because they carry the least interpretation.
But the tool has a hard limit White inherits and never escapes. You can never reach the original myth, because it was already gone the instant it was first told. The list is "bedrock," but even the list is something — an ordering, a selection — and behind it there's no pristine original to recover. So the digging doesn't deliver the truth. It delivers a less-interpreted layer, which is the best a text-historian can do. That's why White's reconstructions are always phrased as "elements to aid in rebuilding," not as recovered facts.
This page is the engine room behind White's detective work. It's the method that lets the Alchemical Ghost be diagnosed (the ghost is a legend — a late, name-fused construct — mistaken for an origin). It underwrites how The Alchemists and Yogis Were One Crowd reads the rosters: shared names are list-residue, the teacher-student links are later family-tree, the rescue-legends are story. And it pairs with White's symbol-web method to form his full read-from-outside toolkit.
The clearest demonstration is the lists of the Nine Nāths. There's no single agreed roster — the "Nine Nāths" show up in several conflicting lists, with a thick layer of legend grown around them.2
Run Detienne backward on it. The legend layer: the elaborate tales of each Nāth's miracles and meetings. The family-tree layer: the teacher-student chains linking them — including the impossible claim that Gorakh was Matsyendra's disciple when they lived three centuries apart. The list layer: the bare names — and here's where it pays off, because the lists don't agree on who the Nine were.
That disagreement, which looks like a mess of bad data, is the evidence. If the names varied while the number stayed fixed at nine, then the number was the real structure and the names were just fillers. So the "Nine Nāths" was a category — nine slots, like nine openings of the body in one scheme — that different traditions filled with different names, rather than a historical brotherhood of nine men.3
That's the method paying off, and showing its limit at the same time. It retrieves that "Nine Nāths" was a category. It cannot retrieve who the "original" Nine actually were — because there was never a pristine original beneath the category.
You're handed a tradition's founding story — rich, detailed, full of named founders and their deeds — and you want the history inside it. Don't ask "is this true?" Ask Detienne's three questions, peeling from the outside in.
Start at the surface and name the legend: the stories, settings, dialogues, miracles. Set them aside — not as false, but as the latest, most-worked layer. Underneath, find the family tree: the claims about who taught whom, who founded what. Note these as later impositions — real data about how the tradition organized itself, but not bedrock. At the bottom you reach the lists: the bare names, the numbers. Treat these as the least-worked layer and read them for structure — do the lists agree? does a number stay fixed while names vary? is the order doing work?
Don't expect to reach a pristine original; Detienne warned you there isn't one. What you get is a layered map: this part is story, this part is lineage-claim, this part is name-residue. That map is the history — not a recovered truth, but an honest cross-section of how much interpretation sits at each level. Reading any origin-story this way keeps you from doing what Ray did with the Alchemical Ghost: mistaking the top, most-worked layer for the bottom, foundational one.
The method fails when its user forgets there's no bedrock and treats the retrieved list-layer as "what really happened." You spot it when a scholar peels back the legend, reaches the names, and then treats them as recovered historical persons with biographies — re-mythologizing from the bottom up. The tell is confident reconstruction: "the original Nine Nāths were actually these nine men, and here's what they did." Detienne's whole point is that you can't get there; the myth was gone before writing.
The opposite failure is paralysis — deciding that since you can't reach the original, nothing can be said, and giving up. White walks between: the lists give you less-interpreted material and structural evidence (the category-not-persons finding), which is real and usable, without pretending it's the vanished original. When a reading either claims to have recovered the pristine founders or throws up its hands, it has lost Detienne's middle path.
The method's value shows in what it yields: the Nine Nāths as category-not-brotherhood, the impossible Gorakh-Matsyendra chronology exposed as a family-tree imposition, the Alchemical Ghost diagnosed as a late construct mistaken for an origin.23
The deep tension is Detienne's own radicalism. If "the myth does not exist" and dies the instant it's told, then all the layers are interpretation, and the "list" bedrock is only relatively less worked, not truly foundational. White accepts this — his reconstructions are deliberately tentative.
An open question: does the method privilege names too much? A bare name-list is low-interpretation but also low-information. The richest historical content often lives in the family-tree and legend layers White is trained to subtract. The method is conservative by design, and its conservatism may throw out real signal along with the noise.
White is importing Detienne — a classicist working on Greek myth — into the Indian field, and the borrowing is itself the interesting move. The fit is good: Detienne's list-to-tree-to-legend model, built for Greek material, maps cleanly onto the Siddha rosters. That suggests the layering process is general to how spoken name-traditions become written legends — not specific to one culture.
The productive tension is with the living-practitioner sources the vault holds. A lineage tradition experiences its family tree (this guru taught that disciple) as living, handed-down fact. Detienne-via-White treats the family tree as a layer of later interpretation to be peeled away. So what the insider receives as transmission-history, the textual outsider reads as a later layer over a barer list. Neither cancels the other. The lineage's family tree may be both a real transmission and a later construction, and Detienne's method can flag the construction without disproving the transmission. The honest position holds the two readings side by side, rather than letting the digging declare the lineage's self-understanding "merely" legend.
Plain version: every origin-story you inherit has been retold so many times that the original is gone — but the retellings come in layers, and you can read the layers to tell recent dressing from old bones.
The sharpest version. Detienne's claim that "the myth does not exist" — that it's gone the instant it's first told — applies to every founding story you've ever trusted. Your field's. Your family's. Your own story about yourself. The vivid origin you carry is the legend layer. Underneath is a family tree of impositions. Underneath that, a barer list. And beneath even that, nothing pristine to recover. Take this seriously and it dissolves the comfort of "what really happened." There's no recoverable original — only layers of retelling — and the work is never to reach the truth but to honestly date the layers. The most you can ever say of any origin is which parts are recent dressing and which are old bones. And the satisfying, story-shaped parts are always the dressing.
Questions worth chasing: