Every language has words that mean two things. Usually it does not matter — context decides, nobody notices, and the ambiguity dies unremarked.
Sometimes it matters enormously, and nobody notices then either.
This volume has at least three places where a single word with two meanings is holding up something the tradition treats as settled: a woman's consent to her own marriage, the kinship between two gods who are quarrelling about exactly that, and the entire doctrine that makes a living human being an object of worship.
In all three cases Debroy flags the fork in a footnote and declines to choose. In all three the tradition kept one reading and forgot there had been a second.
The pattern is worth a page because it changes what kind of thing you are reading. A doctrine that rests on a pun is not false. It is contingent — it could have gone the other way, and the record of it going one way rather than the other is a footnote rather than an argument.
Satī has completed a twelve-month vow. Śiva appears. He offers her a boon and asks her to name it, twice, because she cannot get the words out — "she was embarrassed and was incapable of stating what was in her heart… she was again enveloped in shame."1
Finally she manages a formula that names nothing:
"O one who bestows boons! Please grant me the boon I desire, without any impediments."
And Śiva answers before she finishes: "Be my wife."
Debroy's note on that exact word:
"The word vara means boon, as well as groom. Hence, implicitly, Satī asked for the groom of her choice."2
Read the sentence with the second meaning loaded and it stops being an evasion. Grant me the vara I desire, without impediments is, on one reading, a woman naming her husband — and it is the only moment in the Satī Khaṇḍa where she chooses anything.
She never has to say it. The word says it for her, and she keeps her shame intact, which the text has just spent a paragraph establishing she cannot part with.
So her entire agency in her own marriage sits inside a double meaning, and the tradition remembers the episode as Śiva granted her a boon.
The second fork is smaller and lands somewhere much worse.
In chapter 77 Śiva is consecrating Viṣṇu, and he turns to Brahmā and says "O son!" Debroy notes:
"Since Rudra was born from Brahmā, this can be translated as father too. The word used is tata."3
Tata runs both directions. The same word, in the same sentence, addressed by Śiva to Brahmā, is grammatically either "my son" or "my father."
Now recall what the surrounding forty chapters are about. Viṣṇu's central rebuke to Brahmā is: "you think of Parameśvara Rudra as your own son."4 The volume's founding myth is a quarrel over who came out of whose navel.5 Its theology has Rudra emerging from Brahmā's eyebrows and Brahmā emerging from Śiva's right side, in the same Saṃhitā.6
The kinship direction between these two gods is the disputed thing. And at one point in the text the word that states it is a word that means both.
The dispute is not resolved by the grammar. It is reproduced by the grammar — which is either a very deft piece of writing or an accident, and there is no way to tell which.
The third is the one with institutional consequences, and this build logged it two Saṃhitās ago.
Debroy's endnote 322 gives guru two derivations: plainly "large", in a list of physical descriptors — and guṇa + ruddha, "one who obstructs the guṇas."7
Chapter 25 builds the whole discipleship apparatus on the second: the guru as the being who arrests the operation of the guṇas in you, and is therefore worth worshipping the way you would worship a god.
On the first reading he is a big man.
The tradition took the second, and an entire structure of authority followed — the guru as an object of pūjā, the transmission that cannot happen without him, the elaborate offences against him. Debroy flags the fork and declines to adjudicate, and the whole weight sits on the side he did not choose for us.
Set them out and the pattern is uncomfortable.
| Word | Reading A | Reading B | What was kept |
|---|---|---|---|
| vara (ch 69) | boon | groom | Both, quietly — the episode reads as a boon and functions as a choice |
| tata (ch 77) | son | father | Undecided, in a text whose main quarrel is this |
| guru (ch 25) | large | obstructor of guṇas | B, and a doctrine of worship built on it |
Add the volume's own licence for the method. It derives siṃha, lion, from hiṃsā, violence — and it derives Śiva by splitting the syllables, śi + va.8 The text does etymology as argument, openly, all the time.
Which means the three forks are not accidents of a sloppy manuscript. They sit inside a tradition that treats word-analysis as a source of doctrine. In a system where a derivation counts as a reason, a word with two derivations is a live disagreement carried inside a single term — and it can travel for a thousand years without anyone having to notice they are holding two positions.
The three also differ in one respect worth naming. Vara is almost certainly deliberate — a pun that lets a woman speak without speaking is a writer's move. Guru is almost certainly not. And tata is genuinely unreadable. A method that cannot distinguish craft from accident is not going to be able to tell you which of the three you are looking at, and this build should stop pretending otherwise.
Here is the transferable version, and it is not really about Sanskrit.
Somewhere in the founding documents of anything you belong to — a company's charter, a contract, a constitution, a religion — there is a word that could have been read two ways. It was read one way. The reading hardened, everything downstream was built on it, and now the reading is invisible because it is simply what the word means.
The tell is that the justification for the practice is the word itself. We do it this way because that's what "guru" means. When a practice's whole warrant is a definition, go find out whether the definition was ever contested — because if the word had two senses at the time, somebody chose, and choosing is not the same as discovering.
Three questions, cheap to ask:
Is the term doing more work than a term should? A word that settles a question all by itself is suspicious. Real questions are not usually settled by vocabulary.
Who benefits from this reading rather than the other one? Guru as "large" builds nothing. Guru as "obstructor of the guṇas" builds an institution. That asymmetry is information.
Is the alternative reading still recoverable? Sometimes it is sitting in a footnote, as here. More often it is gone, and then the honest position is that you do not know what the word meant and are choosing to continue.
None of this means the reading is wrong. It means it is a decision wearing the clothes of a fact, which is a different thing to be careful about.
Debroy's consistency here is the thing that makes the finding possible, and it is worth stating plainly.
His stated method is word-for-word translation with interpretation relegated to the notes.9 So when a word forks, he prints one reading in the body — he has to, English will not carry both — and puts the other in a footnote. He does not argue for either.
That looks like modesty and it is also a method choice with a cost. Every fork he flags is a place where the English reader is holding a decision Debroy made silently in the line and disowned in the note.
But the alternative is worse, and this build has now seen it. At chapter 73 Debroy records: "the text repeatedly says nadi (river). We have corrected it to Nandi."10 That is a fork he closed rather than flagged — and it is the one case where the seam is invisible from the English.
So the three puns are not a weakness in the translation. They are the visible part of a process that is happening everywhere and is usually silent. Wherever Debroy declines to choose, you can see the choice. Everywhere else he chose.
The last observation, and it is the one that keeps this from being a debunking exercise.
The tradition does not experience these as problems.
A Sanskrit nirvacana is not an etymology in the modern sense — it is not a claim about historical descent. It is a claim that the sound of a word discloses something true about the thing. On that understanding, siṃha from hiṃsā and siṃha from anywhere else are not competitors. Both are disclosures. Two derivations mean the thing has two properties.
Which is why the forks never had to be resolved. Nobody was holding an inconsistency, because in the operative theory of language there was no inconsistency to hold.
That is the honest frame, and it cuts both ways. It means the tradition is not being sloppy. And it means the tradition had no mechanism for noticing when a doctrine of worship got built on one branch of a pun, because it did not have the category of branch. Additive meanings are wonderful for a poet and catastrophic for an audit.
Strongest support is now four flagged forks in the translator's own notes — 777 (vara = boon/groom, explicitly connected to Satī's implicit choice), 822 (tata = son/father), 322 (guru = large / obstructor of guṇas), and 871 (avadāna = a heroic act / to cut into pieces) — plus the volume's open use of nirvacana as argument.
The fourth, added 2026-08-10 from the endnote pass of 855–905, is the cruellest and the most structurally interesting. At chapter 88, Vīrabhadra addresses the gods who have come to Dakṣa's sacrifice expecting a share:
"Expecting to receive a gift, all of you arrived here. I will now give you a return gift… I will grant you avadāna and satisfy you now."11
Debroy: *"Dāna is a gift. The gods came to the sacrifice, expecting dāna. Vīrabhadra promised them avadāna — the word means a heroic act. But it also means to cut into pieces."*11
So the threat is delivered in the vocabulary of hospitality, and the pun is doing the whole of the work: a sacrifice is an economy of shares and gifts, the gods came for theirs, and what they are offered is a homophone. The other three puns carry doctrine; this one carries a joke, and the joke is about the ritual economy the entire Khaṇḍa turns on.
It also sharpens the page's own claim. Three flagged double meanings could be a translator noticing lexical range. Four, one of which is a deliberate wordplay inside a character's speech rather than an ambiguity in the text's own voice, is harder to explain that way — the fourth is the volume using the fork rather than merely containing one.
Tension one, and it is the page. Satī's only act of choice in her own marriage is carried by a double meaning, not by a sentence. Remove the pun and she asks for nothing.
Tension two. The kinship direction between Śiva and Brahmā is the volume's central genealogical dispute, and at one point the word stating it means both things.34
Tension three. The most institutionally consequential item in the volume's classification — the guru as an object of worship — rests on the branch Debroy declines to endorse.7
Tension four. Debroy flags forks and closes them elsewhere without flagging (endnote 787). The visible forks are a sample of an invisible population, and there is no way to estimate its size.
Open question. How many forks are there? Three surfaced across roughly 800 notes read to date, which is a rate, not a total. A dedicated pass hunting only for notes containing "also means," "can be translated as," or two glosses joined by "or" would produce a real count in one sweep — and if the rate holds across chapters 96–150 the volume has five or six, which is enough to ask which doctrines are load-bearing on them.
The Derivation of Śiva holds the volume's etymological method at full strength — fourteen derivations, several mutually incompatible, clustered on the words a person repeats daily. That page shows etymology installing doctrine; this one shows what happens when the installation had two options and the record of the choice survives only in an apparatus the tradition never read.
He Answered Before She Could Ask is the same scene from the other side. Read without the pun it is a woman too ashamed to speak being spoken over; read with it, she named her husband and kept her shame. Both readings are in the text and the difference between them is one footnote.
The Quarrel Over Who Is Whose Father holds the paternity dispute that tata reproduces. The volume argues about this for forty chapters and then uses a word that declines to take a side.
Brahmana Not Caste but Knower of Brahman is the vault's other case of a philological fork with social consequences, arriving from outside this tradition's own apparatus. There the correction came from a modern philologist; here it comes from the translator's footnote — and in both the tradition itself had no way to raise the question.
Chinese Buddhist Translation Pitfalls (history). A monk renders a Japanese surname into Chinese characters chosen for sound; the characters also carry meaning, and the teacher spends years walking into temples under a name that reads "wild, fierce, liberated." The original was Kalyāṇa Mitra, "spiritual good friend." A phrase about gentleness came out the other end meaning its opposite, and nobody could catch it because each step was locally correct.
The Śiva Purāṇa's forks are the same machinery running inside one language, which is why they are harder to see.
No script change is required. Guru does not have to cross a linguistic boundary to acquire a second sense — it arrives with one. So there is no moment of translation at which anyone is on alert, no bilingual reader to catch it, and no obviously wrong output to trigger a check.
What the pair gives that neither gives alone: cross-language corruption is self-announcing and intra-language corruption is not, and we have built all our defences against the first. The Otomo error is visible precisely because a brawler's name on a gentle monk is absurd. A doctrine of guru-worship built on one branch of a native pun produces nothing absurd — it produces a functioning institution. Which suggests the general rule: the more competent a tradition is in its own language, the less equipped it is to notice its own semantic drift, because every intermediate step reads perfectly.
The Power of Metaphor and Hidden Commands (behavioral-mechanics). The operator's device is a sentence that delivers two messages at once — a surface message the listener processes consciously and an embedded one that lands without being examined, because attention is committed to the surface.
Satī's vara is that structure occurring in a woman's mouth, in her own favour, and it is the only place in the vault where the technique is used defensively.
The surface message is a formula. Grant me the boon I desire. Unobjectionable, modest, and it names nothing. The embedded message is a proposal. And because the surface reading is complete and sufficient, she is never exposed — if he had answered the other reading, nothing would have been said.
The insight neither produces alone: ambiguity is the only rhetorical device that protects the speaker rather than advantaging them, and it is therefore the characteristic instrument of whoever cannot afford to be on record. The manuals treat double meaning as an operator's tool for smuggling something past a listener. Here it is the opposite — the listener has all the power, and the double meaning exists so the speaker can be refused without ever having asked. Which reframes a great deal of indirect speech: look at who would bear the cost of the plain version, and you will usually find the reason for the ambiguity there rather than in any intent to manipulate.
Sharpest implication. Three times in this volume a single word with two meanings is holding up something the tradition treats as settled, and each time Debroy flags the fork and refuses to adjudicate. Vara (en. 777) means boon and groom — so Satī's shame-choked formula, "grant me the boon I desire, without any impediments," is on one reading a woman naming her husband, and it is her only act of choice in the whole Khaṇḍa. Tata (en. 822) means son and father, and it appears in Śiva's mouth addressed to Brahmā — in a volume whose central quarrel is precisely who is whose son. Guru (en. 322) means "large" or "one who obstructs the guṇas," and an entire doctrine making a living person an object of worship is built on the second. The volume licenses the method itself — deriving siṃha from hiṃsā, Śiva from śi + va — which means these are not scribal accidents but live disagreements carried inside single terms, invisible because in nirvacana two derivations are additive rather than competing. And the three visible forks are a sample: at ch 73 Debroy closed one silently — "the text repeatedly says nadi… we have corrected it to Nandi" — so wherever he declines to choose you can see the choice, and everywhere else he chose.
Generative questions.