Rule. After the roots kaṇḍū and the rest, the affix yak comes obligatorily, in their own sense. Of yak only ya is added; the k is a marker. Bhaṭṭoji's gloss: "let yak be added obligatorily, in their own sense, after these dhātus".
"After a dhātu" (dhātoḥ) is carried forward from SK 2629. Bhaṭṭoji's commentators both say so. They add that the word vā "optionally" of SK 2629 is not carried forward. That is why yak is obligatory. If vā were carried on, both say, a form without yak would also arise. One of them gives kaṇḍavati.
Example.
root 1944 kaṇḍūñ "in rubbing the limbs" (gātravigharṣaṇe); the ñ is a marker
→ this rule: kaṇḍū + yak → kaṇḍū-ya
→ yak is marked with k, so the ū takes no guṇa → kaṇḍūya
→ kaṇḍūya ends in yak, so it is a dhātu (SK 2304)
→ laṭ, the ending tip, the class-sign śap → kaṇḍūya + a + ti
→ the two *a*s give one a (SK 191) → kaṇḍūyati.
The root is marked with ñ. So when the fruit goes to the agent, the ātmanepada comes (SK 2158) → kaṇḍūyate.
Applies when / blocked by. Why does the gloss say "after dhātus"? Bhaṭṭoji answers: "so that it may not come after prātipadikas". For kaṇḍū and the rest are of two kinds (dvidhā). Each is a dhātu, and each is also a prātipadika, a noun stem. Yak comes after the dhātu only. The noun stem kaṇḍū stays a noun; both of his commentators give kaṇḍūḥ. One of them, in his notes on the next chapter, says these items have been shown not to be nāmadhātus. That is why Bhaṭṭoji finishes the nāmadhātus of chapter 58 first and states kaṇḍū and the rest separately.
How do we know there are two kinds? Both of Bhaṭṭoji's commentators point to the Bhāṣya for this. One of them lays out the proof.
- Yak carries the marker k, whose work is to block guṇa. That work has a purpose only after a dhātu. So the items are dhātus.
- The item is taught with a long ū: kaṇḍūñ. For a dhātu alone, a short kaṇḍu would do. SK 2298 would lengthen it before yak and still give
kaṇḍūyati. So the long ū would be pointless. - The long ū earns its place in the noun
kaṇḍūḥ. This noun is used as it stands, not analysed out of a phrase (avigṛhīta). Both commentators name it. So the items are prātipadikas too.
By this proof he also rejects an older author's analysis kaṇḍūṃ karoti "he makes kaṇḍū" → kaṇḍūyati. The items are not noun stems only.
Because yak comes after a dhātu, it is an ārdhadhātuka affix. That has a visible result in group D. A stem's final a is lost before it (SK 2308), as in sukhyati. If yak came after the noun stem, one commentator says, it would not be ārdhadhātuka, and the a would stay. The other makes the same point: then the a could not be lost, and sukhyati could not be formed.
What does svārthe "in their own sense" exclude? One of Bhaṭṭoji's commentators explains. An older author said that yak comes in the sense of kṛñ, "making". Svārthe rejects that: yak adds no meaning. The same commentator rejects the claim that karaṇe "in the sense of making" is carried here from SK 2673. Nothing supports that anuvṛtti, he says.
Where does nityam come from? Some carry it forward from SK 2634, nityaṃ kauṭilye gatau. One of Bhaṭṭoji's commentators rejects this, for two reasons. Nitya there has the force of eva, "only". And the vā of SK 2629 continues there too. So nityam here is simply the result of vā not being carried.
Watch out. The markers ñ and ṅ on the items decide the endings. An item marked with ñ takes both sets of endings (SK 2158): kaṇḍūñ, irañ, asūñ. An item marked with ṅ takes ātmanepada only (SK 2157): mahīṅ, hṛṇīṅ. Every other item takes parasmaipada (SK 2159). Do not give mantūyate to mantu unless you follow Candra, who marks it with ñ (group B).
Let yak be added obligatorily (nityam), in their own sense (svārthe), after these dhātus [— 'dhātoḥ' carried forward (anuvṛtti) from SK 2629]. Why [say] "after dhātus"? So that it may not come after prātipadikas. For kaṇḍū and the rest are of two kinds: dhātus and prātipadikas. (1944) kaṇḍūñ — in the sense of rubbing the limbs (gātravigharṣaṇa): kaṇḍūyati, kaṇḍūyate. (1945) mantu — in the sense of offence (aparādha); in the sense of anger (roṣa), according to some: mantūyati. Candra, however, states it as ñ-marked (ñit): mantūyate. (1946) valgu — in the sense of honouring (pūjā) and of sweetness (mādhurya): valgūyati. (1947) asu — in the sense of tormenting (upatāpa). Some [read it as] asū, [or as] asūñ: asyati; asūyati, asūyate. (1948) leṭ, (1949) loṭ — in the sense of roguery (dhaurtya), of priority (pūrvabhāva) and of dreaming; in the sense of shining, according to some: leṭyati, leṭitā; loṭyati, loṭitā. (1950) lelā — in the sense of shining. (1951) iras, (1952) iraj, (1953) irañ — in the sense of envy: irasyati, irajyati. [For irañ,] lengthening by "hali ca" (SK 354): īryati, īryate. (1954) uṣas — in the sense of becoming dawn (prabhātībhāva). (1955) veda — in the sense of roguery and of dreaming. (1956) medhā — in the sense of quick grasping (āśugrahaṇa): medhāyati. (1957) kuṣubha — in the sense of throwing: kuṣubhyati. (1958) magadha — in the sense of surrounding (pariveṣṭana); in the sense of being a low servant (nīcadāsya), according to others. (1959) tantas, (1960) pampas — in the sense of pain. (1961) sukha, (1962) duḥkha — in the sense of that action (tatkriyā): sukhyati, duḥkhyati — the meaning is "he experiences happiness", "he experiences pain". (1963) sapara — in the sense of honouring (pūjā). (1964) arara — in the sense of an action [done] with an ārā (a goad) (ārākarman). (1965) bhiṣajū — in the sense of healing (cikitsā). (1966) bhiṣṇaj — in the sense of attending (upasevā). (1967) iṣudha — in the sense of bearing arrows. (1968) caraṇa, (1969) varaṇa — in the sense of going. (1970) curaṇa — in the sense of stealing. (1971) turaṇa — in the sense of hastening. (1972) bhuraṇa — in the sense of holding and nourishing (dhāraṇa, poṣaṇa). (1973) gadgada — in the sense of faltering speech. (1974) elā, (1975) kelā, (1976) khelā — in the sense of sporting (vilāsa); (1977) ilā, according to others. (1978) lekhā — in the sense of slipping (skhalana) as well. Others say that this ends in a (adanta): lekhyati. (1979) liṭa — in the sense of smallness and of censure: liṭyati. (1980) lāṭa — in the sense of living. (1981) hṛṇīṅ — in the sense of anger and of shame. (1982) mahīṅ — in the sense of honouring: mahīyate — the meaning is "he receives honour". (1983) rekhā — in the sense of praising and of reaching (āsādana). (1984) dravas — in the sense of torment and of attending (paricaraṇa). (1985) tiras — in the sense of disappearing (antardhi). (1986) agada — in the sense of being free from disease. (1987) uras — having the sense of strength: urasyati — the meaning is "he becomes strong". (1988) taraṇa — in the sense of going. (1989) payas — in the sense of prasṛti (a particular quantity). (1990) saṃbhūyas — in the sense of being abundant (prabhūtabhāva). (1991) ambara, (1992) saṃvara — in the sense of gathering (saṃbharaṇa). This [kaṇḍvādi list] is an ākṛtigaṇa (a class whose membership is known from usage, not closed by the list). [Here end kaṇḍū and the rest.]
- commentary TB (sk.2678.t1, opening; re-filed from the tail of sk.2677.t1 on 2026-10-05; NSP p. 453): from 'dhātor ekācaḥ…' [SK 2629] 'dhātoḥ' continues; 'vā' has ceased, otherwise a form without yak would also result (OCR 'kaṇḍvayati'), hence 'nityam'. TB rejects the view of some that 'nityam' is carried forward from 'nityaṃ kauṭilye gatau' [SK 2634]: 'nitya' there stands in the sense of eva, to show that the takra-kauṇḍinya maxim is not absolute; and 'vā' continues there too. TB (sk.2678.t1) on 'svārthe': it is meant against the predecessor's (prāc) statement that yak comes in the sense of kṛñ ('making'); it later rejects the claim that 'karaṇe' continues here from 'śabdavaira…' [SK 2673]. Followed: the anuvṛtti bracket, 'obligatorily', 'in their own sense'.
- commentary BM: 'dhātubhyaḥ' because it continues (anuvṛtti) from 'dhātor ekācaḥ…'; 'nityam' because 'vā' has ceased, otherwise kaṇḍavati and the like would also arise. 'dvidhā hi' is made clear in the Bhāṣya; hence forms without analysis (avigṛhīta) such as kaṇḍūḥ. Followed: the anuvṛtti bracket and 'obligatorily'.
- commentary TB (sk.2678.t1) on 'dvidhā': yak being k-marked (kit, to block guṇa) shows they are dhātus; the long-vowelled reading kaṇḍūñ shows they are also prātipadikas; it cites the Bhāṣya verse 'dhātuprakaraṇād dhātuḥ…'. If they were only dhātus, the long-vowelled reading would be pointless; with two kinds, kit is meaningful for the dhātus and the long reading for the unanalysed forms such as kaṇḍūḥ. It also rejects the predecessor's analysis 'kaṇḍūṃ karoti kaṇḍūyati'. Nor, it says, should it be said that even granting the two kinds yak comes only after the prātipadika: then an undesired form would result from the dhātu in laṭ and the rest, and after prātipadikas such as sukha, duḥkha the deletion of a would be impossible (allopāsaṃbhava), so sukhyati and the like would not be formed. Recorded.
- commentary BM: leṭitā has y deleted by 'yasya halaḥ' [SK 2631]. medhā: āśugrahaṇa is understanding quickly; a variant reads 'sugrahaṇe'. sukhyati: before yak, a is deleted by 'ato lopaḥ' [SK 2308]; from the prātipadikas yak would not be ārdhadhātuka and a would not be deleted; the curādi sukha, duḥkha give sukhayati, duḥkhayati. arara: ārā = goad (pratoda), ārākarma = an act having it as instrument. Followed: the gloss of ārākarman.
- commentary TB (sk.2678.t1): leṭitā has y deleted by 'yasya halaḥ' [SK 2631]; arara: ārā = goad (pratoda), ārākarma = an act having it as instrument; sukhyati: in the curādi the forms sukhayati etc. are given. Followed: the gloss of ārākarman.
- commentary BM on lekhā: lekhāyati [with ā kept], and lekhyati, a being deleted by 'ato lopaḥ' before yak after the adanta [stem]. Recorded: the kaumudī gives only lekhyati, for the adanta view.
- commentary TB (sk.2678.t1) on 'adanto 'yam': lekhyati; in the ā-final alternative lekhāyati. Recorded: the kaumudī gives only lekhyati, for the adanta view.
- commentary TB (sk.2678.t1): 'pūrvabhāva' = pūrvatva (priority); prasṛti is a particular quantity (parimāṇaviśeṣa); prabhūtabhāva = bāhulya (abundance), citing Amara; it adds saṃbhūyasyati, asaṃbhūyasīt, with saṃ not separated [for the augment], since separation applies only where an affix producing the dhātu-saṃjñā is prescribed after a prātipadika. Followed: 'priority', 'a particular quantity', 'being abundant'.
- commentary BM: 'ākṛtigaṇo 'yam' — that is, kaṇḍvādi; hence duvas, 'saṃdīpane', and the like are included: 'samidhāgniṃ duvasyata'. Followed: '[kaṇḍvādi list]'.
- commentary TB (sk.2678.t1) on 'ākṛtigaṇaḥ': hence duvas, 'saṃdīpane', and the like are established, and 'samidhāgniṃ duvasyata' is in use (OCR 'duvatsandīpana', 'samidhāmi duvasyate'). Recorded.
- commentary BM reads the lemma as 'bhiṣaj cikitsāyām' and says it ends in j (jānto 'yam): bhiṣajyati. Recorded: the English keeps the transmitted 'bhiṣajū'.
- textual Every dhātu number in our list is one lower than in the 1915 print (ours 1944–1992, NSP 1945–1993; NSP pp. 453–454, scans page-0463/0464 read). The numbers are given as transmitted. Ours 'bhiṣajū' (1965) matches the print, though BM and the usual dhātupāṭha read bhiṣaj.
- textual Ours reads 'veda' (1955), 'gadgada' (1973) and 'agada' (1986); the print has halanta stems: ved, gadgad, agad (NSP p. 454, scan). Ours 'dhāraṇapopaṇayoḥ' (1972) is a slip for dhāraṇapoṣaṇayoḥ, translated so. Ours lacks the closing colophon 'iti kaṇḍvādayaḥ ||' that the print has (NSP p. 454, scan); it is supplied in square brackets.