For most of the twentieth century, every history of Indian alchemy opened with the same foundational text: the Rasaratnākara, the "Jewel-Mine of Mercury," credited to the great Nāgārjuna — India's first and primal alchemy classic.
There was one problem. It was a ghost. The "Rasaratnākara of Nāgārjuna" was never a single book. It was three separate texts that a scholar in 1903 accidentally fused into one — and everyone after him cited the fusion as if it were real.1
White tells this story for two reasons. One, it matters for dating his sources. Two, it's a confession: he repeated the ghost himself, in his own early work, before he learned better.2 The Alchemical Ghost is a lesson in how one citation error, made by an authority, can haunt a whole field for eighty years.
The error has a clean anatomy. In 1903, Prafulla Candra Ray — the founding figure of Indian chemistry-history — identified a work he called the Rasaratnākara, credited to Nāgārjuna, as India's "primal and primordial alchemical classic."3 His authority was so great that the claim got "amplified by Sylvain Lévi, Jean Filliozat, and Mircea Eliade" — a chain of the field's biggest names, each citing the one before.3
Then in 1983, Dominik Wujastyk published an article called "An Alchemical Ghost" and proved the so-called Rasaratnākara of Nāgārjuna was a splice of three texts:4
Three works, two or three authors, centuries apart, welded into one phantom "classic."
And here's the delicious part: the error was catchable from day one. In 1903 — the same year Ray published — another scholar, Palmyr Cordier, correctly described the very same text under its proper name and noted it was "quite identical to the Rasaratnākara recently used by Prof. P. C. Ray."5 The correction existed at the same moment as the error. It just didn't have Ray's authority behind it, so it lost.
Two reasons, and White's account shows both.
First, the prestige cascade. Ray was the authority. Lévi, Filliozat, and Eliade amplified him. Once three giants have repeated a claim, people stop checking it and start inheriting it. Each scholar cited the consensus, not the manuscript. The error spread not despite the field's expertise but through it — the more eminent the repeaters, the more unquestionable the ghost.
Second, the error made a better story. A field wants a primal classic — one foundational book by one towering name to anchor its origin tale. The ghost handed Indian-alchemy studies exactly that: one book, the Rasaratnākara, by one legend, Nāgārjuna, at the dawn of the tradition. That tidiness is itself a spreading force. A clean origin story gets cited, taught, and repeated more than a messy one about three separate texts of shaky date. The error didn't just survive despite being convenient. It survived because it was convenient, out-competing the truer but messier account.
The splice was extra-seductive because the Rasendramaṅgala invites it. The text dresses its late alchemist up as the ancient philosopher: it stages a scene atop a sacred mountain with "the Mādhyamika philosopher Nāgārjuna," a Buddhist goddess, a king, and a crowd of Buddhist luminaries, with "a smattering of Buddhist terminology" to complete the costume.6 So the late book wears the mask of the thousand-year-older sage. The ghost wasn't only a clerical slip. It was a misreading the text itself encouraged. (This is the Nāgārjuna problem in miniature — many figures, one borrowed name.)
This page is the area's exhibit on epistemic hygiene. It shows White's method correcting itself in public — and his read-the-symbol-web method is only trustworthy if its user will name his own past mistakes. Here he does. It also supplies the dating discipline behind every chronological claim in the area: once the "primal classic" turned out to be a late splice, the whole timeline of Indian alchemy had to be rebuilt.
The sharpest thing about White's treatment is that he doesn't tell this from a safe distance. He flags, in his own notes, that he repeated the Alchemical Ghost himself — and that Wujastyk had to nudge him toward the fix.2 White's first article on Indian alchemy, the one Eliade helped him publish, sat inside the very consensus Wujastyk later blew up. So when White recounts the ghost, he's documenting his own teachers' error and his own part in it.
This matters as a study in what self-correction looks like in a field with no lab. There's no experiment to re-run, no replication to fail. The only way to catch the ghost was for a scholar to go back to the actual manuscripts and count — and find that the "one text" was three. And the only way to spread the fix was for the field's people, White included, to publicly retract. White models the retraction: he names Ray's error, names the eminent amplifiers, names Wujastyk's fix, and names his own complicity.
Then watch the consequences, because the confession isn't just good manners — it's load-bearing. The whole timeline had been anchored to the ghost as the "primal classic." Pull the ghost out and the timeline floats free and has to be re-pinned. White re-pins it to the Rasārṇava, which he dates to the eleventh century on independent grounds.6 A reader who didn't know the ghost had been exorcised would have no way to judge whether White's new timeline was sound or just a rival guess. The confession is what makes the rebuilt timeline checkable: it tells you exactly which foundation was rotten and what replaced it.
There's a final twist that deepens the lesson. White notes that Wujastyk himself errs in one detail.7 The corrector is also fallible. Ghost-hunting doesn't deliver a final, error-free text — only a less-wrong one, waiting for the next person's recount. The integrity of the field isn't a state it reaches. It's a thing it keeps doing: recount after recount, confession after confession.
You're building on a "well-established" foundational source that everyone in your area cites. Before you add your brick, run the ghost-hunt.
Trace the citation backward — past the recent works, past the mid-century names, to the first assertion, the Ray-equivalent. Ask: did the amplifiers each check the primary source, or did they cite each other? (The tell of a cascade is that the secondary citations multiply while primary-source checks stay at zero.) Then do the Wujastyk move: go to the actual document, or the closest thing to it, and count. Is this "one text" really one text? Is this "single author" really one person? Is the famous old name attached to the famous old work — or has a late book borrowed an ancient name to borrow its status?
And if you find a ghost, do the hard part: check whether you've cited it before, and if you have, say so. The workflow isn't just doubting other people's foundations. It's being willing to find your own name in the cascade and announce it. That last step is what separates a field that self-corrects from one that quietly updates and pretends it always knew.
The failure this page is about is the prestige-laundered error — a mistake that survives because authoritative repetition has replaced verification. You spot it by a specific pattern: a claim cited everywhere, traceable to one originating authority, amplified by big names, with a thinning trail of primary-source checks behind the secondary citations. Everyone "knows" the foundational fact, but nobody in living memory has gone back to the document.
A second tell, present in the ghost, is narrative convenience: the error survives partly because it makes a better story (the late alchemist is the ancient philosopher; the field has a primal classic). When an inherited "fact" is both unchecked-in-decades and suspiciously story-shaped, suspect a ghost. The cure is always the same unglamorous act Wujastyk performed: go count.
The evidence is documentary and decisive: Wujastyk's 1983 proof that the "Rasaratnākara of Nāgārjuna" is three texts, Cordier's 1903 simultaneous correct reading, and the citation chain from Ray through Lévi, Filliozat, and Eliade.345
The standing tension is about authority itself. The same prestige that let Ray's error spread is the prestige a field needs to function — you can't re-verify every foundation yourself. So the ghost exposes a trade-off with no clean fix: trust in authority is both the field's engine and its weak spot.
An open question: how many other ghosts still haunt the Indian-alchemy record, given that its chronology rests on so much attribution-by-legend and manuscript-splicing? White's whole detective apparatus is, in a sense, an admission that we don't yet know.
The drama here is a four-way relationship. Ray and Eliade are the authorities whose prestige spread the ghost — and Eliade is also White's acknowledged intellectual mentor, the man who helped him publish his first, ghost-repeating article.2 Wujastyk is the corrector. White is the inheritor who converts.
The tension is generational and almost filial. White owes Eliade his entry into the field, and he has to overturn the error Eliade amplified — including in his own early work. So how does a field actually progress? Not by the authorities correcting themselves, but by a later, less famous figure (Wujastyk) going back to the manuscripts, and by the authorities' students (White) choosing the manuscript over the master. White's handling — crediting Eliade warmly while dismantling the chronology Eliade helped freeze in place — is a model of how to honor a teacher and correct him at once. The tension never resolves; it's held. Gratitude and correction live side by side. That's exactly the posture a healthy field demands of its inheritors.
Plain version: a confident mistake from a trusted source can outlive everyone who could have caught it, because people cite the reputation instead of rechecking the thing — and the fix always comes from someone willing to go back and count.
The sharpest version. The most authoritative, most-cited foundation in any field you trust is — by that very fact — the one least likely to have been recently re-checked, and therefore the prime candidate to be a ghost. Prestige doesn't just fail to protect against error. It actively shields error from the one thing that kills it: someone unglamorous going back to count. Take the Alchemical Ghost seriously and it flips your instinct. The claim everyone "knows," from the source no one questions, isn't your safest brick. It's your most suspect one. And the further, more personal implication, the one White lives out: if you've built on a ghost, you've probably spread it — and the health of your field depends on your willingness to find your own name in the cascade.
Questions worth chasing: