वार्त्तिकानि — where Kātyāyana corrects Pāṇini

What a vārttika is

परिभाषा

Not a gloss, and not a translation. A vārttika is a piece of criticism — terse to the point of being cryptic on its own, and meant to be read sitting right beside the sūtra it examines, the way a reviewer's margin note is meaningless without the manuscript.

व्याख्यानं वार्त्तिकम् The tradition's own job description

Four words, and they are the whole brief: व्याख्यानं वार्त्तिकम्, "a vārttika is an explanation." Not a restatement, not a summary — an explanation, which presupposes something in the original left open enough to need one. And what a vārttika actually checks, in the tradition's own three-part formula:

उक्तानुक्तदुरुक्तचिन्ता वार्त्तिकम्

"Consideration of what is said, what is unsaid, and what is badly said — that is a vārttika's business." (rendering mine) Three verbs, three different repair jobs, and this page is organised around exactly that split — below.

मुनित्रयम् — the three sages, and why order matters. Pāṇini states the grammar. Kātyāyana tests it. Patañjali, a century or two later still, adjudicates Kātyāyana's tests in the महाभाष्यम् — accepting some vārttikas, refining others, occasionally overruling them. The tradition states its own precedence rule for this chain in four words: यथोत्तरं मुनीनां प्रामाण्यम्, "the later sage's authority [prevails]." That does not make Pāṇini's text provisional — it makes a vārttika the opening move in an argument that is still being adjudicated by the time Patañjali writes, not a corrigendum filed against a closed book. Page A9 is about how Patañjali actually argues; this page stops at what Kātyāyana hands him.

Three kinds of note, three real examples

उक्तानुक्तदुरुक्तम्

Each of the three is illustrated below with one vārttika whose own wording carries the marker unambiguously — a rare thing to be able to promise about terse Śāstric Sanskrit, and the reason these three were chosen over louder-looking candidates elsewhere in the 921.

अनुक्तम् What the sūtra left unsaid

The commonest kind, by a wide margin

gives the affix कटच् — roughly, "a collection of" — after a short, specific list of prefixed roots. Kātyāyana's second vārttika on it:

Rendering mine. "An addition should be made: अलाबू (bottle-gourd), तिल (sesame) and उमा (flax), in the sense रजस्, husk." Three more bases the sūtra's own list did not reach, each licensing a real word — अलाबूकटः, तिलकटः, उमाकटः, husk of gourd, sesame, flax. उपसंख्यानम्, "an addition," is this whole category's own name for itself: of all 921 vārttikas contain that word, or वाच्यम्/वक्तव्यम्, "should be said" — a floor, not the true share, since most additions are terser still and just name the missing case outright.

दुरुक्तम् What the sūtra said badly

lists eight environments where a grammatical fiction — treating an elided or replaced sound as though it were still there — is switched off. Kātyāyana's fourth vārttika on it uses a word this page has not needed yet:

Rendering mine. "There is a दोषः — a fault — in it, regarding संयोगादिलोप (cluster-initial elision), लत्व (the change to ल्) and णत्व (cerebral ण्). Three more environments the list should have excluded and did not — not a gap Kātyāyana is filling politely, but an error he is naming as one." This is the rarer move: of the 921, only a handful use दोषः on themselves this directly. Most of what looks like correction elsewhere in the corpus turns out, on inspection, to be another addition — see the count in the next section.

उक्तम् Checking what the sūtra already covers

The rarest of the three is not an addition or a fix at all — it is Kātyāyana raising a case, checking whether the grammar as it stands already reaches it, and answering his own question. deals with elision after certain divine-name compounds. An objector raises षडङ्गुलिदत्तः, "given by six-fingered [X]" — does the derived adjective षडिकः actually come out right? Kātyāyana's second vārttika:

Rendering mine. "[It] is सिद्धम् — already accomplished — from the separate statement that ठक् follows षष् before a vowel-initial [affix]." No addition, no correction: a different rule, already on the books, already covers it, so nothing here needs to change. This is उक्तचिन्ता doing its actual job — not rubber-stamping the sūtra, but checking a live objection against the whole grammar before conceding that anything is missing.

What a vārttika actually changes

पूर्वापरम्

कटच् The rule's family, before and after

Left alone, gives exactly one affix, कटच्, after exactly the bases the sūtra names. Its nine vārttikas do not just add bases to that one affix — three of them introduce entirely separate affixes for senses the sūtra never mentioned, each real in the bhāṣya's own examples:

Before — 5.2.29 alone	After — with its vārttikas
कटच् — a collectionon the sūtra's own bases only	कटच् — extended to अलाबू, तिल, उमा, भङ्गाअलाबूकटः · तिलकटः · उमाकटः · भङ्गाकटः
—	तैलच् — oilइङ्गुदतैलम् · सर्षपतैलम्
—	पटच् — a spreadअविपटः · उष्ट्रपटः
—	गोयुगच् — a yoked pairउष्ट्रगोयुगम् · खरगोयुगम्

Every word in that table is quoted, not composed. They are the Kāśikā-bhāṣya's own illustrations of these vārttikas, read straight from C.layer('bhashya','5.2.29') — none built for this page. A reader who stops at the sūtra sees one affix; a reader who reads the vārttikas too sees the shape of an entire small paradigm the sūtra by itself does not show.

1.1.72 Absorbed into the standing apparatus

Some vārttikas stop being "notes on a sūtra" altogether and become rules in their own right. This is checkable, not a received opinion: every one of the 921 vārttikas was compared, string for string, against the 133 Paribhāṣenduśekhara metarules this site's own corpus carries — the tradition's separate, later compilation of governing principles. came back an exact match. Three of them sit on a single sūtra, — itself one of the eight most-argued-over rules in the whole text:

Vārttika, on its sūtra	Paribhāṣā number

What "absorbed" means here, precisely. The corpus does not tag downstream adoption by name, so this page is not going to claim which later manuals cite these as sūtras outright. What it can show is stronger than a claim: the exact text of a Kātyāyana vārttika is, word for word, one of the 133 principles later grammarians reach for constantly, filed and numbered on its own, no longer presented as a footnote to 1.1.72. That is what absorption looks like when it is checkable instead of asserted.

13% drew comment. Why these, and not the other 87%?

वितरणम्

That is not a coverage gap to apologise for — a vārttika is expensive, close scrutiny by a named grammarian, and most rules simply do not need it. The interesting question is where the 516 cluster, and the corpus answers it two ways.

By rule type

Every sūtra in the corpus carries a type — संज्ञा (naming), परिभाषा (interpretive), विधि (operative), अतिदेश (extension by analogy) or अधिकार (governing span). Comparing the rate at which each type draws a vārttika — not the raw count, which just tracks how many sūtras of that type exist — shows a real skew:

Type	Sūtras	Drew a vārttika	Rate

अतिदेश — extension rules — draw comment at roughly one and a half times the corpus average rate, the highest of the five. An अतिदेश rule's entire content is "treat X as if it were Y for the purposes of some other rule." That is exactly the kind of claim Kātyāyana's method exists to test: how far does the analogy actually hold, and does it silently misfire somewhere the sūtra didn't anticipate? A rule that simply assigns a name, by contrast, has almost nothing left to argue about once it is stated.

By adhyāya

Adhyāya	Sūtras	Drew a vārttika	Rate

The first two adhyāyas — definitions and case-assignment — draw comment at roughly twice the rate of adhyāya 7. That is the tripādī's leading pāda, the start of the strictly ordered, mutually-blind zone page A4 covers: dense with rules, but each one narrowly mechanical once its inputs are fixed. Adhyāyas 1–2 are the opposite — a handful of definitions (सवर्ण, अव्यय, the six kārakas) that every other rule in the text leans on, so an error or a gap there has the largest possible blast radius. Scrutiny follows leverage, not density.

Even the loudest sūtras are mostly being finished, not fixed

Where the printed edition tags a vārttika's own rhetorical role — it does this for some sūtras and not others, a real gap in the source rather than a smoothed-over one — sūtras ( vārttikas) have every one of their notes tagged. The single largest tag, by a wide margin: . Even — tied for the single most-argued-over sūtra in the text, ten vārttikas — has all ten tagged उपसंख्यानवार्तिकम्, "addition." The most contested rule in the Aṣṭādhyāyī was not accused of being wrong ten times. It was found incomplete ten times.

All 516, browsable

Interactive

Filter by adhyāya or by how contested a sūtra was, then click one to read its vārttikas — and, for the more heavily argued sūtras, Patañjali's bhāṣya on them, shown together the way the tradition actually reads them.

अध्यायः

वार्त्तिकानि

सूत्रम्	अ.	वार्त्तिकानि

Click a row above to read it.

Where this comes from. Every row is C.vartika(ref) in full — no vārttika is shortened. Bhāṣya text is inlined for the 52 sūtras that drew four or more (the contested tail this page keeps returning to); below that line, the detail panel links into the full reading view instead of inlining prose most readers will not open, to keep this file under its size budget.

Check yourself

अभ्यासः

Six questions. Pick an answer to see which sūtra decides it.