Home › What the text actually says
What an interpolation is, and how anyone can tell
The popular claimIf a line is in the book, it was always in the book.
A manuscript is a copy of a copy of a copy. Somewhere in that chain a scribe writes a marginal note and the next scribe, finding it in the margin, takes it for a line that was dropped and puts it back into the body. Somewhere else a reciter adds two verses that make a passage clearer, and they are clearer, and they stay.
This is not corruption in any sinister sense and it is not peculiar to Indian literature. It happened to Homer, to the Bible, to Shakespeare's quartos. It is simply what copying does over centuries, and the study of it is a mature discipline with results you can check.
The tradition knew. Sanskrit commentators have a word for it — प्रक्षिप्त, prakṣipta, "thrown in" — and another for a variant reading, पाठान्तर, pāṭhāntara, which they record when they meet one. This page is not importing a foreign suspicion into the literature. It is describing an apparatus the literature already had.
The strongest evidence: it is missing elsewhere
Everything else on this page is inference. This one is close to observation.
Manuscripts of a large work fall into families — groups that descend from a common ancestor and share the same distinctive errors. If a passage appears in one family and is absent from another that separated centuries earlier, the simplest explanation is that it entered after the split.
The more independent the families, and the wider apart geographically, the stronger the case. A passage present in Kashmir and absent in Kerala is in a different position from one absent in a single late copy.
Metre
Sanskrit verse is countable. The anuṣṭubh śloka is four quarters of eight syllables with defined constraints on the heavy and light positions; triṣṭubh is eleven, jagatī twelve. The rules are strict enough that a competent reciter hears a violation the way a musician hears a wrong note.
So a line that will not scan, sitting inside a passage that scans perfectly, is a flag. Not a verdict — poets take licence, and a defective line can simply be a copying error rather than an addition — but a flag.
The wrong century's language
Sanskrit changed. Vedic Sanskrit has grammatical machinery — the subjunctive, the injunctive, a far richer set of infinitives — that classical Sanskrit lost. Vocabulary arrives and departs; some words are simply not attested before a certain period.
A passage carrying late forms inside an early text is doing something that needs explaining. This is the same reasoning that dates any document anywhere, and it is unusually reliable for Sanskrit because the grammatical tradition documented the language so precisely.
The sentence that closes without it
The most satisfying test, and the easiest to explain.
Sometimes a narrative or an argument runs around a passage. A question is asked; then comes a block on a different subject; then the answer arrives, matched to the question as though nothing intervened. Lift the block out and the two halves join without a seam.
A text that reads better with a passage removed is telling you something. Combined with manuscript evidence it is close to decisive; alone it is a strong suspicion, because a digression is also a thing authors do on purpose.
Knowing what it should not know
A passage that presupposes an institution, a doctrine, a coin, a title or a polemic that did not yet exist when the surrounding text was composed has dated itself. This criterion does the most work and needs the most care, because it is the easiest to apply badly — it requires knowing when the thing referred to actually appeared, which is itself a scholarly question.
Why śruti is the hard case
Everything above assumes copying. The Vedic corpus was not primarily copied; it was recited under the interlocking pattern system described on the companion page — jaṭā, ghana and the rest — whose entire purpose is to make an inserted or altered word break several recitation sequences at once and be caught in the room.
So the two classes of text are not equally vulnerable, and that is not a claim of faith but of mechanism. Interpolation is a live question for the epics, the Purāṇas and the dharma literature. It is a much smaller question for the Saṁhitās.
The abuse of the word
This has to be said, because the page would be dishonest without it.
"That verse is an interpolation" is a claim, and it requires the same evidence as any other claim. Used without evidence it is simply a way of deleting a line one finds embarrassing, and it is used that way constantly — by defenders of a tradition wanting an ugly passage gone, and by critics wanting a passage that complicates their argument gone.
The test is simple and we hold ourselves to it here: which manuscripts lack it, which criterion fires, and who established that. If those three cannot be answered, the correct word is not "interpolation" but "I would rather this were not here."
What this page does not say
It does not say that an interpolated passage is worthless or that it did no work in the world. A verse added in the twelfth century and recited ever since has shaped practice for eight hundred years, and its lateness does not undo that. Dating a line and dismissing it are different acts.
It does not say these methods deliver certainty. They deliver degrees of confidence, and the honest form of a finding is this passage is absent from the Kashmiri and Southern recensions, breaks metre, and uses a term unattested before the Gupta period — three independent criteria converging — not scholars say it is fake.
And it does not hand you a licence. A reader who takes from this page only the sentence "so anything inconvenient can be called an interpolation" has taken the opposite of what is here.
Source — The standard apparatus of textual criticism as applied to Sanskrit literature; the Bhandarkar Oriental Research Institute's Mahabharata apparatus; the traditional vocabulary of prakshipta (interpolated) and pathantara (variant reading) used by Sanskrit commentators themselves
What is genuinely debated
Which specific passages are interpolated is argued, often sharply, and a criterion failing to fire is not proof that a passage is original. The word is also used polemically — by defenders to discard a line they dislike and by critics to discard a line that complicates their case — and a reader should be as suspicious of a convenient "that is an interpolation" as of a convenient quotation.
Every claim on this site names its source, and every calculator shows its arithmetic.