क्त्वा · ल्यप् · तुमुन्

क्त्वा — one sūtra, two conditions

3.4.21

English "having gone" places no restriction at all on who did the going. Sanskrit's gerund does — and the whole rule is one sūtra, stating both restrictions at once.

क्त्वाThe rule that licenses it

3.4.21 — after a verb whose agent is the same [as the main verb's], in prior time. Two conditions, both in that one compound word: समानकर्तृक (same agent) and पूर्वकाल (prior time, not simultaneous). Violate either and क्त्वा is simply the wrong word — not stylistically weak, not unusual, wrong.

समानकर्तृकयोःthe two actions share one agent पूर्वकालेthis one happens first

Same agent — the ordinary case

"Having eaten, [he] goes" — the Kāśikā's own example on this sūtra. One person, Devadatta, both eats and goes; the eating is finished before the going starts. Both conditions hold, so क्त्वा is exactly right.

Same agent, violated — the trap

"Having gone home, his mother fed him" cannot be क्त्वा. The one who went and the one who fed are different people. Forcing क्त्वा onto that sentence does not produce a wrong-sounding sentence — it produces a different sentence, because क्त्वा always attaches its prior action to whoever is doing the main verb. The Kāśikā gives the same trap with its own pair, asking समानकर्तृकयोः इति किम्? — "why does the sūtra say same agent?":

Attempting "the Brāhmaṇa having eaten, Devadatta goes" — but as written this can only mean "Devadatta, having eaten, goes." The Brāhmaṇa's meal has no way to attach.

"The Brāhmaṇa having eaten (locative absolute), Devadatta goes." Two different agents need a locative absolute — taught below — not क्त्वा.

Prior time, violated — the other trap

Same agent is not enough by itself. The two actions must also be sequential. — "he goes and talks [at the same time]" — is one agent doing two simultaneous things, so क्त्वा cannot be used here either; Sanskrit falls back to a plain च. The Kāśikā notes one further exception to its own exception — आस्यं व्यादाय स्वपिति, "he sleeps with his mouth open" — where a genuinely simultaneous posture is still idiomatically allowed without क्त्वा, which is exactly the kind of edge a rule this short has to have.

ल्यप् — the same word, wearing a prefix

7.1.37

क्त्वा and ल्यप् are not two affixes to memorise separately. They are one affix with two spellings, and which spelling shows up is decided by exactly one fact: is there an उपसर्ग in front of the root?

ल्यप्The replacement rule

7.1.37 — in a compound whose first member is not नञ्, क्त्वा is replaced by ल्यप्. "Compound" here means a root fused with a preceding उपसर्ग; ल्यप्'s own इत् letters (ल्, प्) vanish, so what actually shows on the page is just -य.

धातुः	bare क्त्वा — no उपसर्ग	with उपसर्ग → ल्यप्

The Kāśikā's own contrast on this sūtra: — but stripped of its उपसर्ग, plain . Same root, same meaning "having done" / "having carried" — the affix changes shape only because a prefix is now attached in front of it.

Two sandhi effects the prefix drags in

Because ल्यप् surfaces as a bare -य rather than -त्वा, it exposes the root's final sound directly to whatever follows — and two of this series' own sandhi rules fire as a result. Both are the same mechanism page 03 teaches; this is where they show up inside a kṛt affix rather than between two words.

तुक् ण-त्व

— 6.1.71 . A root ending in a short vowel, before a पित् affix (one whose इत् includes प्; ल्यप् is one, bare त्वा is not), inserts the augment तुक् — surfacing as त्. कृ ends in short ऋ, so with an उपसर्ग: . Without one, bare त्वा is not पित् and नो तुक् is needed at all — कृत्वा, not *कृत्त्वा.

— 8.4.14 , a targeted case of the general , reaching past an intervening vowel by . नम् is cited in the Dhātupāṭha as णमँ — its ण् already surfaces as न् by a general rule — and an उपसर्ग ending in र् or ष् (प्र does) restores exactly that ण्. गत्वा but आगत्य; नत्वा but प्रणम्य — the last pair is walked step by step below. See page 03's consonant sandhi for the same न्→ण् shift working between words.

तुमुन् — purpose
3.3.10

Where क्त्वा looks backward ("having done"), तुमुन् looks forward: the infinitive of purpose, "in order to do" or simply "to do".

तुमुन्The purpose infinitive

3.3.10 — तुमुन् and ण्वुल् come after a root, in a sense of future time, when [that root's] action is the purpose of another action. तुमुन्'s own इत् is न्, so what shows is -तुम्. The Kāśikā's example: — "goes in order to eat".

The family resemblance

तुमुन् is built exactly like the gerundive तव्यत् you met earlier in this series (3.1.96 ) — the same guṇa grade of the root, the same indifference to the इट् augment — it is simply indeclinable where तव्यत् inflects. Learn कृ's shape once and all three below share it:

कर्तुम्to do — तुमुन्, indeclinable कर्तव्यम्to-be-done — तव्यत्, a gerundive that declines the doer — कृ + तृच्, an ordinary noun

All three share the guṇa-graded stem कर्-. कृ itself never shows guṇa on its own — कृत्वा, कृतम् keep the bare root — but every one of these three forward/obligation/agent affixes pulls it to कर्- by the same rule, 7.3.84 , and none of them adds the इट् augment a शब्दरूप-reader might expect (कर्तुम्, not *करितुम्).

कर्तृ, declined. The corpus's own ऋकारान्त paradigm — not retyped, pulled the same way every noun table in this series is.

विभक्तिः एकवचनम् द्विवचनम् बहुवचनम्

The verbs that ask for it

तुमुन् is the standard complement of a closed set of governing verbs — "able to", "begin to", "wish to". The Kāśikā names them by listing the roots directly in the sūtra text:

आ changes the meaning, not just the direction. Bare रभ् means ("to be glad"); with आ it idiomatically means to begin — the same kind of उपसर्ग-driven meaning shift this series' upasarga page catalogues.

When क्त्वा can't — the two absolutes
2.3.37 · 2.3.38

क्त्वा's same-agent condition is not a limitation to work around quietly. Sanskrit has two dedicated constructions for exactly the case क्त्वा refuses — a different agent in the subordinate clause — and both turn on a participle placed in an oblique case with no verb at all.

सति सप्तमीThe locative absolute

2.3.37 — by whose action the time of another action is marked, that [word] takes the seventh [locative]. Put the differing agent and its participle both in the locative, and the whole phrase means "while X was doing Y" or "when X had done Y" — with no finite verb needed at all.

"While the cows were being milked, he went; and came back once they had been milked" — the Kāśikā's own pair, two locative absolutes back to back.

षष्ठी चानादरेThe genitive of disregard

2.3.38 — and the sixth [genitive], in disregard. Same shape as the locative absolute, but with a pointed extra sense: the main action happens despite, or heedless of, the participle's action.

"In spite of her weeping, he went forth [as an ascetic]" — either case is grammatical here; the genitive sharpens it into disregard.

The thief fled while I looked on. The Kāśikā's commentary on this very sūtra reaches for almost exactly that sentence:

"The Nandas were slaughtered like cattle, the Rākṣasa looking on" — a genitive absolute of pure helplessness. A subordinate clause with a different agent from the main clause is exactly the case क्त्वा was built to exclude, and exactly the case these two sūtras exist to cover.

Same agent?
Interactive

Pick which relationship the subordinate clause has to the main clause. One choice keeps क्त्वा; the other breaks it, and the sūtra says exactly why.

अन्वयः — the relationship

Watch नम् become ल्यप्
Interactive

The same root, three ways: bare, prefixed, and — for contrast — a short-vowel root taking the other sandhi effect. Every step below is a rule the live vidyut-prakriya engine actually cites when deriving these forms, read off it once, not composed by hand.

The engine also derives an alternate प्रणत्य for प्र + नम् — 6.1.71's तुक् fires there too, because 6.4.38 makes नम्'s internal nasal loss optional before ल्यप्. This page follows प्रणम्य, the form actually found across the literature; both are grammatical.

All three, three roots
Reference

Every form below was checked against the same engine the प्रकृया player runs, not typed from memory.

धातुः क्त्वा (bare) ल्यप् (+ उपसर्ग) तुमुन् तव्यत्

Where every form on this page comes from. क्त्वा, ल्यप् and तुमुन् produce indeclinables, which the Aṣṭādhyāयी site's noun paradigms do not carry — so, exactly as page 12's four-root walkthrough did, every form here was read once from the live vidyut-prakriyā engine (the same one krdanta.html runs) and hardcoded with the rule history that produced it. The Sanskrit example sentences are the corpus's own Kāśikā and Vāsu commentary text on each sūtra, quoted — not paraphrased — and checked at build time to still say what this page claims; the English translations are written for this series.

Check yourself
अभ्यासः

Six questions. Every one is answered by a sūtra, not by feel.