तिङन्तणिच्प्रकरणम्

Covers SK 2575–2607 (33 sūtras): all present. Every unit has its own kaumudī. SK 2579 and SK 2601 share the wording ṇau ca saṃścaṅoḥ. Each is outlined under its OWN sūtra, as in the 1915 print and the English: 2579 = 6.1.31, saṃprasāraṇa for śvi; 2601 = 2.4.51, gāṅ for iṅ. Our source.json carries the other sūtra's kaumudī (and BM) under each, so the Devanāgarī there does not match (see "Still open for Ajai"). The spine is not an Aṣṭādhyāyī run. The SK gathers the rules that bear on the causative ṇic (hetumaṇṇic) from AS books 1, 2, 3, 6, 7 and 8: 1.4.55 → 3.1.26 → 7.4.80–81 → 6.1.31 → 8.3.116 → 7.1.63–64 → 7.4.96 → 6.1.18 → 7.3.37 → 6.1.32 → 7.4.4–6 → 7.3.38–39 → 1.3.70 → 6.1.56 → 1.3.68 → 7.3.40 → 6.1.57 → 7.3.41–43 → 6.1.48 → 2.4.51 → 6.1.49 → 6.1.55 → 6.4.90–91 → 8.4.21 → 2.4.46. There are three short runs: the 7.3.37–43 augments and final substitutes (split by the ātmanepada and ā rules), the 7.4.4–6 caṅ rules, and the 6.1.48–57 ā-for-ec rules (scattered). Every AS-ref below is checked against references/ashtadhyayi/sutraani-data.json. That file numbers 6.1.31 as skn 2601 and 2.4.51 as skn 2579, which is the same swap as our kaumudī. Our sūtra fields, the print and the English all have 2579 = 6.1.31.

• Anuvṛtti from later in the SK. Several rules here carry words from sūtras the SK teaches only later:
• dhātor ekāco halādeḥ kriyāsamabhihāre yaṅ (3.1.22, SK 2629) gives dhātoḥ at 2576;
• atra lopo 'bhyāsasya (7.4.58, SK 2620) gives abhyāsasya at 2577 and 2583;
• iṅaś ca (2.4.48, SK 2616) gives iṅaḥ at 2601;
• ṇer aṇau… (1.3.67, SK 2738) gives ṇeḥ at 2592 and 2594, and gṛdhivañcyoḥ pralambhane (1.3.69, SK 2739) gives pralambhana through the ca of 2592;
• na raparasṛpi… (8.3.110, SK 3168) gives na at 2580, and the nipātana bhujanyubjau… (7.3.61, SK 2877) supplies b for ubj.
• The outline names them where BM does.
• Anuvṛtti from p53. The curādi prakaraṇa taught the rules this one leans on most:
• artihrī… puk ṇau (7.3.36, SK 2570) gives ṇau at 2585, 2590, 2591, 2595 and 2597–2599;
• cisphuror ṇau (6.1.54, SK 2569) gives ṇau at 2593, 2596 and 2603;
• ṇau caṅy upadhāyā hrasvaḥ (7.4.1, SK 2314) gives "caṅpare ṇau" and upadhāyāḥ at 2587–2589;
• at smṛdṝtvara… (7.4.95, SK 2566) gives at at 2583.

tiṅantaṇicprakaraṇam — outline (p54, SK 2575–2607, all 33 sūtras)

p53 (tiṅantacurādiprakaraṇam) closed at 2574 with the gaṇasūtra pucchādiṣu… and the print's colophon "iti curādayaḥ".

• Where the prakaraṇa begins. The print heads it "tiṅante ṇyantaprakriyā 51" (NSP p. 428, scan page-0438). Our name and number, tiṅantaṇicprakaraṇam 54, follow Wikisource style and our own numbering. sk2575 opens with the saṃjñā tatprayojako hetuś ca (1.4.55), and sk2576 with the rule hetumati ca (3.1.26).
• Where it ends. At 2607 ṇau gamir abodhane, with the vārttika on īrṣy and the remark on a dhātu "from which the prompting has ceased" (prārthayanti). The print closes "iti ṇyantaprakriyā" (NSP p. 433, scan page-0443). Ours lacks it, so the English supplies it in brackets. The print then heads "tiṅante sannantaprakriyā 52", and p55 (tiṅantasanprakaraṇam) opens with dhātoḥ karmaṇaḥ samānakartṛkād icchāyāṃ vā (SK 2608).
• Contents in outline.
• 2575–2576: the causative. The prompter is hetu and kartṛ, and ṇic comes when its activity is meant: bhāvayati / bhāvayate.
• 2577–2578: the caṅ abhyāsa's u → i. abībhavat, apīpavat; optional for sru … cyu: asisravat / asusravat. These are recalled with them: nāglopi… (aśaśāsat, acīcakāsat) and ṇic on a ṇic-final (corayati, acūcurat).
• 2579, 2584, 2586: saṃprasāraṇa before ṇi + caṅ/san for śvi, svāpi and hvā: aśūśavat / aśiśvayat, asūṣupat, ajūhavat / ajuhāvat.
• 2580–2583: the caṅ-aorist of particular stems.
• No upasarga-ṣ for stambh, siv and sah: avātastambhat.
• The vowel-initial roots: āṭiṭat, mā bhavān ididhat, aubjijat.
• num for rabh and labh: ararambhat.
• Optional a in the abhyāsa of veṣṭ and ceṣṭ: avaveṣṭat / aviveṣṭat. The kāṇyādi option: acīkaṇat / acakāṇat.
• 2585, 2590, 2591: augments other than puk. yuk (śāyayati, hvāyayati), juk (vājayati), nuk and luk (vilīnayati, vilālayati). luk for pā (pālayati) is a vārttika under 2589.
• 2587–2589: pā, sthā and ghrā in the caṅ-aorist. apīpyat, atiṣṭhipat, ajighripat / ajighrapat. The puk forms (arpayati … sthāpayati) and ur ṛt (acīkṛtat / acikīrtat) are recalled here.
• 2592–2596: lī, bhī and smi. Ātmanepada even when the fruit is not the agent's (lāpayate, bhāpayate). ā when the fear or wonder comes from the prompter (bhāpayate, vismāpayate), and ṣuk (bhīṣayate). With an instrument none of these apply: kuñcikayainaṃ bhāyayati.
• 2597–2603: root-specific substitutes before ṇi.
• Final v, t or p: sphāvayati, śātayati, ropayati / rohayati.
• ā for the ec: krāpayati, adhyāpayati, jāpayati, sādhayati, vāpayati / vāyayati.
• gāṅ for iṅ: adhyajīgapat / adhyāpipat.
• 2604–2607: the close.
• ū for duṣ (dūṣayati; optional in cittavirāga), with recalls of mit, rañj, ci and sphur.
• ṇ for both n's of an: prāṇiṇat.
• gam for iṇ: gamayati, but pratyāyayati. ik is like iṇ (adhigamayati), plus ghātayati and the īrṣy vārttika (airṣyiyat / airṣiṣyat).

Threads that run through the prakaraṇa:

• Two ṇics. The curādi ṇic of p53 is svārthika, in the root's own sense. The ṇic of 2576 is hetumaṇṇic: it comes when the prompter's activity (preṣaṇa, adhyeṣaṇa, anumati, upadeśa …; BM, TB) is to be expressed. Both form a ṇic-final dhātu (sanādyantā dhātavaḥ, 3.1.32, SK 2304).
• When hetumaṇṇic follows a ṇic-final (cori + ṇic), the first ṇi is deleted by ṇer aniṭi (6.4.51, SK 2313), and that deletion sets aside its vṛddhi by aco ñṇiti (7.2.115, SK 254). The reason is pūrvavipratiṣedha or apavāda-hood (2578, BM). Result: corayati, not *corāyayati.
• In the caṅ-aorist the "ṇi" of ṇau caṅi is a class (ākṛti), so the shortening and the lengthening still apply: acūcurat (2578).
• Pada. ṇicaś ca (1.3.74, SK 2564) gives ātmanepada when the fruit goes to the agent: bhāvayate. 2592 (lī) and 2594 (bhī, smi) give it even when it does not (akartrabhiprāya, BM). This holds only for a ṇic-final whose base is the root itself. A second ṇi takes parasmaipada: vismāpayan (2596, BM). The anuprayukta bhū takes no ātmanepada: bhāvayāṃ babhūva. That follows our BM and the print. The WS kaumudī and BM read bhāvayāṃ babhūve, so the BM transmission is split.
• The caṅ-aorist frame (taught in p53) has these steps:
• caṅ by ṇiśridrusrubhyaḥ kartari caṅ (3.1.48, SK 2312);
• ṇi deleted by SK 2313;
• the penultimate shortened by SK 2314;
• doubling;
• sanvadbhāva when a light syllable follows the abhyāsa (sanvallaghuni caṅpare 'naglope, 7.4.93, SK 2316), giving i by sanyataḥ (7.4.79, SK 2317);
• lengthening by dīrgho laghoḥ (7.4.94, SK 2318).

This prakaraṇa adjusts the frame at each point:

• the abhyāsa vowel: i for u (2577–2578); a, optionally (2583, and at smṛ… under 2582); ī after penultimate loss (2587);
• the stem's penultimate: i for sthā and ghrā (2588–2589); num makes rabh and labh heavy (2581–2582);
• what is doubled: saṃprasāraṇa comes first (2579, 2584, 2586), and gāṅ comes first (2601).
• "No vowel-substitute before ṇi while doubling is due" (dvitve kārye ṇāv ajādeśo na). It makes bhū double before vṛddhi and āv (abībhavat, 2577), and sphur double with its u (apusphurat, 2605). BM and TB limit it to cases where the part after the abhyāsa has an a. So gāṅ may come first (adhyajīgapat, 2601), and so may vṛddhi of iṇ (pratyāyiyat, BM/TB under 2607).
• A conjunct after the abhyāsa blocks the lengthening. asusravat (2578), ararambhat / alalambhat (2582), abibhrajat (2583), aśiśvayat (2579). The audit notes a tension under 2578: BM there denies sanvadbhāva "because two consonants intervene", but under 2582 BM says laghuparatva holds even across several consonants. The outline states only the undisputed point: the abhyāsa is heavy before the conjunct, so there is no lengthening.
• The puk default and its displacers.
• artihrī… puk ṇau (SK 2570) gives puk to ṛ, hrī, vlī, rī, knūyī, kṣmāyī and the ā-final roots: arpayati … sthāpayati (recalled under 2587). It also follows every ā made for an ec: krāpayati, bhāpayate, vāpayati, sādhayati.
• Displacing it: yuk (2585), luk for pā (vārttika under 2589), juk (2590), nuk and luk (2591), ṣuk (2595), and the final substitutes v, t and p (2597–2599).
• ā for an ec before ṇi. It is carried as āt and ecaḥ from ādeca upadeśe 'śiti (6.1.45, SK 2370), at different strengths:
• optional for bhī from the prompter (2593);
• invariable for smi (2596);
• for krī, iṅ and ji (2600);
• for sidh in a this-worldly sense (2602);
• optional for vī in conceiving (2603).
• Recalled with them: vibhāṣā līyateḥ (6.1.51, SK 2509, under 2590), the vārttika making lī's ā invariable in three senses (2591), and cisphuror ṇau (SK 2569, under 2605).
• "From the hetu" (hetubhaye). 2593–2596 need fear or wonder that comes from the prompter. When the cause is an instrument there is no ā, no ṣuk and no ātmanepada: kuñcikayainaṃ bhāyayati, vismāyayati. The Raghuvaṃśa's "vismāpayan" is defended either as a scribal slip for vismāyayan, or as a second ṇi on smāpi (2596).
• Root citations that select a root.
• tiṣṭhateḥ (with śtip) excludes the yaṅluk stem (TB, 2588).
• līyateḥ (with yak) covers both lī and līṅ (2590–2591).
• sidhyateḥ (with śyan) excludes the bhvādi sidh (2602).
• vīyateḥ is the adādi vī. Whether vyeñ is also covered is disputed: TB sets out "some" in favour, and BM rejects it (2603).
• doṣaḥ cites duṣ in its guṇa form, so the rule bites only where guṇa is due (2604).
• The luk-vikaraṇa paribhāṣā excludes pā rakṣaṇe (2585) and the adādi vā (2590).
• TB adds that the lakṣaṇapratipadokta paribhāṣā does not operate in this section: dhyāpayati, kāpayati (2585, 2590).
• Recalled rules (smāryate). Bhaṭṭoji re-cites earlier rules where the ṇic forms need them:
• nāglopi… (SK 2572) and ṇyantāṇ ṇic under 2578;
• her acaṅi (SK 2531) and at smṛdṝ… (SK 2566) under 2582;
• bhrājabhāsa… (SK 2565) under 2583;
• artihrī… under 2587;
• ur ṛt (SK 2567) under 2589;
• vibhāṣā līyateḥ under 2590;
• ūd upadhāyā gohaḥ (SK 2364) under 2603;
• mitāṃ hrasvaḥ (SK 2568), janījṝṣ… and cisphuror ṇau under 2605;
• iṇvad ikaḥ and hanas to 'ciṇṇaloḥ (SK 2574) under 2607.
• The kaumudī lists forms bare. Lakāra labels and derivation stages are the translator's, following BM and TB as recorded in the unit notes.
• Where the commentary lies. Only the kaumudī blocks are translated (SK-TRANSLATE §25).
• The BM swap. Our sk.2579.b1 is BM on 2601 and 2602. BM on 2579 runs from the tail of sk.2578.b1 into sk.2601.b1, which ends with the lead-in to 2580. TB is not swapped.
• BM missing from our source, read in Wikisource uttara-2:
• 2580 (0100–0101);
• 2586 (0103). sk.2586.b1 is BM on SK 2417 abhyastasya ca (p44).
• 2602 has no BM block; its BM is in sk.2579.b1.
• TB printed under a neighbouring unit:
• 2575 under sk.2574.t1;
• 2578, 2579 and 2580 under sk.2577.t1;
• 2585 under sk.2584.t1;
• 2587 under sk.2586.t1;
• 2589 at the tail of sk.2588.t1;
• 2591 under sk.2590.t1;
• 2601 at the tail of sk.2600.t1;
• 2603 split: it breaks off in sk.2603.t1 and ends at the head of the editorial note sk.2594.n1;
• 2604–2605 in sk.2594.n1;
• 2605's apusphurat at the tail of sk.2603.t1.
• There is no TB on 2597–2599.
• The tail of sk.2607.t1 is TB on SK 2609, and TB on 2608 is missing. That is p55's matter.
• BM garbled in ours (Wikisource has the full text): 2591 (the vilālayati gloss and the vārttika), 2596 (the 'pāṭhaḥ' sentence), 2603 (the "kecit" view), 2605 (ghātako for dhāvato; the pakṣiṇaḥ gloss and the cāpayati half are lost).

The p54 audit (.work/sk-run/p54-audit/sk2575-2585.md, sk2586-2596.md, sk2597-2607.md) found no MUST-FIX. Where its MINOR items bear on a reading, this outline follows the audit. A fixer is applying them to the notes in parallel.

• 2575: the kartṛ designation is added by the ca (samuccaya, BM). TB says the ca sets aside the one-designation rule. Neither calls it anuvṛtti.
• 2578: "aglopi [in that rule] will be exemplified in the subdhātu section". One example, mā bhavān ūnanat, has already been given (BM).
• 2580: "where the condition stands in an upasarga" is the vārttika upasarganimittasya pratiṣedhaḥ (BM), not the sūtra. The abhyāsa-conditioned ṣ still comes in both paryasīṣivat and nyasīṣahat.
• 2582: BM's "after num the a is followed by a conjunct, so no sanvadbhāva" is said of ararambhat. It holds equally for alalambhat.
• 2583: "by bhrājabhāsa… (SK 2565)" names one rule. There is no "and so on".
• 2585: BM says only that ṇau completes the sūtra (śeṣa). Taking it from SK 2570 is the translator's inference.
• 2590: TB: "Vāmana too agrees here" (apy atra), not "elsewhere".
• 2596: TB's "some" read prayojyakartrī as a genitive compound, and BM reads it as a bahuvrīhi. They agree only in result.
• 2600: that the ec replaced is the vṛddhi vowel ai is the translator's interpretation. The commentators are silent.
• 2602: the print spells the counter-example siddhyati.
• 2603: WS BM also reads prayojana where TB's print reads prayojano 'rthaḥ. "prajana" is an inference.
• 2604: ūt does run on from 6.4.89. Ours reads "ūdupadhāyā ity anuvartate", and WS BM only "upadhāyā".
• 2605: the expansion of janījṝṣ… is the translator's. TB's print reads apit, which is probably a misprint for aṣit (BM).
• 2607: the WS kaumudī reads airṣyyat with BM. Ours, the print and TB's lemma read airṣyiyat.

Print readings (NSP, scan-confirmed) that the outline follows:

• 2578: dhātvakṣare, nāglopi and subdhātu-.
• 2580: nimittād and mā bhavān ididhat. Also b for the transmitted "vaḥ" (aubjijat).
• 2589: avavartat.
• 2592: syād akartṛge 'pi and vartikām.
• 2593: ātvaṃ vā.
• 2607: tṛtīyaikāca iti, halādiḥśeṣāt, and the colophon "iti ṇyantaprakriyā".
• 2591: ours and the print read "agamau". The English follows WS āgamau ("augments").

Still open for Ajai (mūla):

• 2579 / 2601: the kaumudī and the BM are swapped in ours.
• Our sk.2579.k1 is the gāṅ text (2.4.51), and our sk.2601.k1 is the śvi text (6.1.31).
• The print (scans page-0439, page-0442), WS, the sūtra numbers and the commentators' lead-ins all support the English. Each unit's English follows its own sūtra.
• source.json has not been edited. Until it is fixed, the reader shows mismatched Devanāgarī.
• 2583: ours and the print read "vāṇiloṭhī apy anyatra". The BM lemma and the WS kaumudī read cāṇiloṭī (caṇa dāne, luṭa steye). vāṇi is already among the Nyāsa's four, so the count of twelve requires cāṇiloṭī. The English keeps the transmitted names.
• 2580: ours and the print read adidrapat. BM treats adidravat (from dru) as the intended form, and adidrapat as a reading from drā with puk.
• 2591: "agamau" (ours and the print) against "āgamau" (WS).
• 2607: the colophon is missing in ours.
• The other mūla corrections above (2578, 2580, 2589, 2592, 2593, 2607) are for Ajai's printed copy.

Ordering: grouped by outcome, which departs from SK order in three places:

• 2584 and 2586 are pulled forward into group C, with 2579.
• 2585 is moved to group E, with the other non-puk augments 2590–2591.
• Group E (2585, 2590–2591) is placed before group F (2587–2589), by its first sūtra.

Everything else is in SK order.

A. The causative: the prompter is hetu, and ṇic expresses its activity — running example: bhavantaṃ prerayati: bhū + ṇic → bhau-i → bhāv-i → bhāvi + śap + ti → bhāvayati; ātmanepada bhāvayate

• 2575 tatprayojako hetuś ca 1.4.55 — the prompter of the agent is called hetu, and also kartṛ. tat refers back to the kartṛ of svatantraḥ kartā (1.4.54, SK 559). The ca adds the kartṛ designation of that rule (samuccaya, BM). TB says the ca sets aside the rule that a thing takes only one designation.
• In bhāvayati the one who impels is both hetu and kartṛ. Being kartṛ, he is what the lakāra expresses (laḥ karmaṇi ca…, 3.4.69, SK 2152; TB). The one who becomes is the prompted agent.
• Uses of "hetu": bhīsmyor hetubhaye (2594) and bhiyo hetubhaye ṣuk (2595) (TB).
• 2576 hetumati ca 3.1.26 — ṇic comes after a dhātu when the activity that has the hetu as its locus (hetumat: preṣaṇa and the like) is to be expressed. ṇic is carried from satyāpa… curādibhyo ṇic (3.1.25, SK 2563), and dhātoḥ from dhātor ekāco… (3.1.22, SK 2629) (BM).
• bhū + ṇic: vṛddhi gives bhau, and āv gives bhāvi. With the lakāra: bhāvayati, "he impels one who is becoming".
• bhāvayate: ātmanepada by ṇicaś ca (1.3.74, SK 2564) when the fruit goes to the agent.
• bhāvayāṃ babhūva: the auxiliary bhū takes no ātmanepada, since bhū is not an ātmanepadin root. This is our BM and the print. WS reads bhāvayāṃ babhūve, and its BM argues for it.
• "And the like" covers adhyeṣaṇā (a request to an equal or superior), anumati (a king's consent) and upadeśa (BM, TB). TB: ṇic expresses the prompter's activity together with the prompted one's activity, and loṭ expresses the prompter's activity without it.

B. The caṅ-aorist: i for the u of the abhyāsa — running example: bhāv-i + caṅ → (bhū doubled first) bu-bhāv-i-a-t → (shortening) bu-bhav-a-t → (u → i, 2577, via sanvadbhāva) bi-bhav → (lengthening) abībhavat

• 2577 oḥ puyaṇjy apare 7.4.80 — i replaces the u of an abhyāsa before san when a labial, a yaṇ or j followed by a comes next. sani is carried from sanyataḥ (7.4.79, SK 2317), abhyāsasya from atra lopo 'bhyāsasya (7.4.58, SK 2620), and it from bhṛñām it (7.4.76, SK 2496), under the aṅgasya adhikāra (BM, TB).
• abībhavat:
• A vowel-substitute before ṇi is not made while doubling is due, so bhū is doubled before vṛddhi and āv.
• After the shortening the abhyāsa has no a, so sanyataḥ cannot give i.
• Through sanvadbhāva this rule does: the u of bu stands before bh, a labial, followed by a.
• Then dīrgho laghoḥ applies.
• Sthānivadbhāva of the shortening is off "by the very undertaking of the rule" (BM, TB).
• The same with other labials: apīpavat (pūṅ), amīmavat (mūṅ). With a yaṇ: ayīyavat (yu), arīravat (ru), alīlavat (lūñ). With j: ajījavat (ju, a sautra root).
• Counter-examples (TB): apare kim? bubhūṣati. puyaṇji kim? ūrṇunaviṣati.
• 2578 sravatiśṛṇotidravatipravatiplavaticyavatīnāṃ vā 7.4.81 — for sru, śru, dru, pru, plu and cyu, the i is optional when a syllable of the dhātu followed by a comes next (dhātvakṣare, print). apare is carried, but not puyaṇji (BM).
• sru-i + caṅ: asisravat / asusravat. The abhyāsa stands before the conjunct sr, so it is heavy and there is no lengthening. On BM's further claim of "no sanvadbhāva", see the Threads.
• TB: aśiśravat / aśuśravat, adidravat / adudravat, apipravat / apupravat, acicyavat / acucyavat. Counter-example: śuśrūṣati.
• Recalled: nāglopiśāsvṛditām (7.4.2, SK 2572, print "nāglopi") blocks the shortening:
• aśaśāsat (śās);
• aḍuḍhaukat (ṛ-marked ḍhaukṛ);
• acīcakāsat (cakāsṛ), on the view that the abhyāsa need only be followed by a light syllable. On the other view: acacakāsat.
• aglopi will be exemplified in the subdhātu section (print). BM: mā bhavān ūnanat has already been given; atitirāyat and others will follow.
• Recalled: ṇic on a ṇic-final. In cori + ṇic, the deletion of ṇi sets aside its vṛddhi, by pūrvavipratiṣedha ("ṇyallopāv iyaṅ…") or as an apavāda: corayati. In luṅ, shortening (SK 2314) and lengthening (SK 2318) still apply, because ṇi in ṇau caṅi is a class (ākṛti): acūcurat. Otherwise the stem's aglopin status would block both, giving *acucorat (TB).

C. Saṃprasāraṇa before a ṇi followed by caṅ (or san): śvi, svāpi, hvā — running example: śvi-i-a-t → (saṃprasāraṇa + pūrvarūpa first) śu-i-a-t → śu-śu → śu-śāv-i → śu-śav → aśūśavat; the other option aśiśvayat (2584 and 2586 pulled forward)

• 2579 ṇau ca saṃścaṅoḥ 6.1.31 — before a ṇi followed by san or caṅ, saṃprasāraṇa optionally replaces the v of śvi (śvayati). vibhāṣā śveḥ (6.1.30, SK 2420) is carried whole. BM takes saṃprasāraṇam from ṣyaṅaḥ saṃprasāraṇam… (6.1.13, SK 1003). TB names "hvaḥ saṃpra—", which in fact follows this rule.
• The text is the print's (NSP p. 429, scan page-0439). Ours has it under 2601. See above.
• paribhāṣā saṃprasāraṇaṃ tadāśrayaṃ ca kāryaṃ balavat: saṃprasāraṇa and pūrvarūpa come before the antaraṅga vṛddhi and āy. In śvi-i-a-t the v becomes u, giving śu. śu is doubled, the second member takes vṛddhi and āv, the penultimate is shortened, and sanvadbhāva lengthens the abhyāsa: aśūśavat. TB: the paribhāṣā is drawn from "ubhayeṣām" in liṭy abhyāsasyobhayeṣām.
• In the other option, the abhyāsa śi stands before the conjunct śv. It is not light, so there is no lengthening: aśiśvayat.
• 2584 svāpeś caṅi 6.1.18 — saṃprasāraṇa replaces the v of svāpi (the ṇic-final of svap) when caṅ follows. saṃprasāraṇam is carried from SK 1003 (BM).
• svāpi + caṅ: saṃprasāraṇa comes first (by the paribhāṣā of 2579), then doubling, pūrvarūpa, sanvadbhāva with lengthening, and ṣ: asūṣupat.
• caṅi kim? svāpayati (TB).
• 2586 hvaḥ saṃprasāraṇam 6.1.32 — saṃprasāraṇa replaces the v of hvā, which is hveñ with ā already made, before a ṇi followed by san or caṅ. ṇau ca saṃścaṅoḥ (2579) continues (TB).
• hvā-i-a-t: saṃprasāraṇa and pūrvarūpa come first, giving hu-i-a-t. No hvā is left for yuk (2585). hu is doubled, the second member takes vṛddhi and āv, the penultimate is shortened, and sanvadbhāva with lengthening gives ajūhavat (BM from Wikisource).
• ajuhāvat: the shortening is optional by the vārttika kāṇyādīnāṃ vā (under 2583), whose Nyāsa list includes hvāyi (TB).
• Contrast laṭ: hvāyayati, with yuk (2585).

D. The caṅ-aorist of particular stems: no upasarga-ṣ, the vowel-initial roots, num, and a in the abhyāsa — running example: rabh-i + caṅ → rambh-i (num, 2581) → ra-rambh-a-t → ararambhat (heavy, so no sanvadbhāva); veṣṭ-i + caṅ → avaveṣṭat / aviveṣṭat

• 2580 stambhusivusahāṃ caṅi 8.3.116 — ṣ does not replace the s of stambh, siv and sah when caṅ follows. Under the vārttika upasarganimittasya pratiṣedhaḥ, the bar covers only a ṣ whose condition stands in an upasarga (BM).
• Carried in: na from na raparasṛpi… (8.3.110, SK 3168), and saḥ from saheḥ sāḍaḥ saḥ (8.3.56, SK 335), under the mūrdhanya adhikāra.
• The upasarga-ṣ would come from stambheḥ (8.3.67, SK 2272) or avāc cālambana… (8.3.68, SK 2273) for stambh, and from parinivibhyaḥ… (8.3.70, SK 2275) for the other two.
• Forms: avātastambhat. paryasīṣivat and nyasīṣahat: the abhyāsa-conditioned ṣ still comes in both (BM).
• Vowel-initial roots (the rest of the kaumudī):
• āṭiṭat (aṭ) and āśiśat (aś): ṭi and śi are doubled together with the ṇic, so there is no sanvadbhāva (TB).
• The shortening comes before doubling, though it is bahiraṅga. The ṛ-marking of oṇṛ shows this: it would be pointless if ṇi were doubled first.
• So edh-i: shortening gives idh, then dhi is doubled: mā bhavān ididhat (print).
• mā bhavān predidhat takes no vṛddhi, because etyedhatyūṭhsu (6.1.89, SK 73) needs edh to begin with ec.
• na ndrāḥ saṃyogādayaḥ (6.1.3, SK 2446) stops n, d and r from being doubled: aundidat (und, di doubled), āḍḍiḍat (aḍḍ), ārcicat (arc).
• ubj has d as its penultimate in upadeśa. na ndrāḥ blocks the d, so only ji is doubled. b for d by the nipātana bhujanyubjau… (7.3.61, SK 2877) waits until after doubling: aubjijat. Ours and the print read "dasya vaḥ", and the English gives b, as the form requires.
• na ndrāḥ applies only after an initial vowel (ajādeḥ, from ajāder dvitīyasya, 6.1.2, SK 2176). So it does not apply in adidrapat. BM prefers adidravat (from dru).
• 2581 rabher aśabliṭoḥ 7.1.63 — the augment num is added to rabh before a vowel-initial affix, but not before śap or liṭ. num is carried from idito num dhātoḥ (7.1.58, SK 2262), and aci from radhijabhor aci (7.1.61, SK 2302) (BM, TB).
• Counter-examples (TB): ārabdham (not aci); rabhate, rebhe (śap, liṭ).
• Its form is given under 2582: ararambhat.
• 2582 labheś ca 7.1.64 — labh likewise takes num. TB says the rule is split off (yogavibhāga) so that only labh continues into āṅo yi.
• ararambhat: after num the a is followed by a conjunct and is not light, so there is no sanvadbhāva (BM). The same holds for alalambhat.
• Recalled:
• her acaṅi (7.3.56, SK 2531) withholds kutva only "not before caṅ", so before caṅ the h of hi is not velarised: ajīhayat.
• at smṛdṝtvara… (7.4.95, SK 2566) gives a for the abhyāsa: asasmarat, adadarat. By the t-marking of at there is no lengthening.
• 2583 vibhāṣā veṣṭiceṣṭyoḥ 7.4.96 — a optionally replaces the abhyāsa vowel of veṣṭ and ceṣṭ before a ṇi followed by caṅ. Carried in: abhyāsasya (SK 2620), at (SK 2566), and caṅpare from SK 2316; ṇau is understood from the sense (BM).
• Forms: avaveṣṭat / aviveṣṭat, acaceṣṭat / aciceṣṭat.
• By bhrājabhāsa… (7.4.3, SK 2565) the shortening is optional:
• abibhrajat, with shortening: i by sanvadbhāva, but no lengthening before the conjunct;
• ababhrājat, without.
• vārttika (sk.2583.k2) kāṇyādīnāṃ vā: for kāṇi and the rest, the shortening of SK 2314 is optional. The roots:
• the Bhāṣya's six ṇic-final roots: kaṇa, raṇa, bhaṇa, śraṇa, lupa and heṭha;
• the Nyāsa's four: hvāyi, vāṇi, loṭi and lopi;
• "vāṇi and loṭhi too, elsewhere" (ours and the print; BM and WS read cāṇiloṭī);
• twelve in all.
• Forms: acīkaṇat / acakāṇat.

E. Augments that displace puk before ṇi: yuk, juk, nuk and luk (and luk for pā) — running example: śā + ṇi → śā-y-i (yuk, not puk) → śāyayati; vā → vāj-i → vājayati; lī → līn-i → vilīnayati (2585 moved here, with 2590–2591)

• 2585 śācchāsāhvāvyāvepāṃ yuk 7.3.37 — the augment yuk is added before ṇi to śā, chā, sā, hvā, vyā, ve and pā. These are śo, cho, ṣo, hveñ and vyeñ cited with ā made, plus veñ and pā pāne (BM). It is an apavāda of puk (SK 2570), which the ā-final roots would otherwise take. "When ṇi follows" is the kaumudī's own śeṣa. Taking it from SK 2570 is an inference (audit).
• śāyayati, hvāyayati (ours and the print; WS adds more forms).
• TB's luṅ forms: aśīśayat, acicchayat, asīṣayat, [a]vivyayat, avīvayat, apīpayat.
• pā rakṣaṇe is not taken, being a luk-vikaraṇa root. Its form pālayati comes under 2589.
• The citation with ā made shows that the lakṣaṇapratipadokta paribhāṣā is off here: dhyāpayati, kāpayati (TB).
• 2590 vo vidhūnane juk 7.3.38 — juk is added before ṇi to vā in the sense of shaking. ṇau is carried from SK 2570 (BM).
• vājayati.
• Why "shaking"? keśān vāpayati ("makes the hair fragrant", BM), with puk.
• BM and TB take vā as ovai śoṣaṇe with ā made. The adādi vā is excluded as a luk-vikaraṇa root. Vāmana too agrees (TB). vājayati would follow from vaja gatau anyway, so the augment's only result is to exclude puk.
• Recalled: vibhāṣā līyateḥ (6.1.51, SK 2509). It cites the root with yak, so it covers both lī and līṅ. It leads into 2591.
• 2591 līlor nuglukāv anyatarasyāṃ snehavipātane 7.3.39 — before ṇi, the augments nuk (for lī) and luk (for lā) are optional in the sense of liquefying a fat. ṇau is carried from SK 2570. "Augments" follows the WS reading āgamau.
• The forms on ghṛtam:
• vilīnayati: lī or līṅ ending in ī, with nuk;
• vilāyayati: the same without nuk, with vṛddhi and āy;
• vilālayati: lā (adādi lā, or lī with ā made), with luk;
• vilāpayati: without luk, with puk.
• The sūtra fuses an ī ("lī ī"), so there is no nuk in the ā option.
• Why "liquefying a fat"? lohaṃ vilāpayati, vilāyayati: no nuk or luk. With ā there is puk; without ā, vṛddhi and āy.
• vārttika (sk.2591.k2) pralambhanābhibhavapūjāsu liyo nityam ātvam aśiti: lī takes ā invariably before a non-śit affix in the senses of deceiving, overpowering and honouring. This feeds 2592. TB: why "not śit"? līyate, lināti.
• k3 is only the editorial tag and is rendered "—".
• For luk for pā (vārttika pāter ṇau lug vaktavyaḥ, an apavāda of puk: pālayati), see 2589.

F. pā, sthā, ghrā in the caṅ-aorist: penultimate deleted with ī, or i for the penultimate — running example: pāy-i + caṅ → a-pa-pāy-a-t → (penultimate ā deleted) a-pa-py-a-t → (ī for the abhyāsa final) apīpyat; sthāp-i → a-ti-ṣṭhap-a-t → atiṣṭhipat

• 2587 lopaḥ pibater īc cābhyāsasya 7.4.4 — before a ṇi followed by caṅ, the penultimate of pā (pibati) is deleted, and ī replaces the final of the abhyāsa. "caṅpare ṇau" and upadhāyāḥ are carried from ṇau caṅy upadhāyā hrasvaḥ (7.4.1, SK 2314) (BM).
• pā + yuk (2585) → pāyi. The doubled stem stands as apīpay-a-t; the ā of pāy is deleted.
• ī replaces the final of the abhyāsa. nānarthake 'lo 'ntyavidhiḥ does not hold for a change made to an abhyāsa.
• The stem is now aglopin with a heavy penultimate, so sanvadbhāva's i and its lengthening are not available. That is why ī is enjoined: apīpyat (BM, TB).
• Recalled: puk by artihrī… (SK 2570), illustrated in order:
• arpayati (ṛ: puk, being nitya, sets aside vṛddhi, then guṇa);
• hrepayati, vlepayati, repayati;
• knopayati (knūyī: y deleted by lopo vyor vali, 6.1.66, SK 873; guṇa because the stem ends in puk);
• kṣmāpayati (kṣmāyī);
• sthāpayati (an ā-final root).
• 2588 tiṣṭhater it 7.4.5 — i replaces the penultimate of sthā before a ṇi followed by caṅ. This is carried from SK 2314 (BM).
• From sthāpayati, the caṅ form stands as atiṣṭhap-a-t, and a → i: atiṣṭhipat.
• The śtip citation (tiṣṭhateḥ) excludes the yaṅluk stem: atāsthāpat (TB).
• 2589 jighrater vā 7.4.6 — for ghrā (jighrati) the i for the penultimate is optional. This is carried as in 2588, as BM's gloss supplies it.
• ajighrap-a-t → ajighripat / ajighrapat.
• Recalled: ur ṛt (7.4.7, SK 2567), optional ṛ for a penultimate ṛ-vowel before a ṇi followed by caṅ:
• acīkṛtat (kṝt with ṛ: kṛt doubled, a for the abhyāsa ṛ, cutva, i by sanvadbhāva, lengthening) / acikīrtat (without ṛ: i with r by upadhāyāś ca, 7.1.101, SK 2571; kirt doubled; lengthening by upadhāyāṃ ca, 8.2.78, SK 2265);
• avīvṛtat / avavartat (print). Without ṛ, laghūpadhaguṇa leaves no light syllable after the abhyāsa, so there is neither sanvadbhāva nor lengthening.
• amīmṛjat / amamārjat (vṛddhi by mṛjer vṛddhiḥ, 7.2.114, SK 2473).
• vārttika (sk.2589.k2) pāter ṇau lug vaktavyaḥ: luk as an augment for pā rakṣaṇe. It is the apavāda of puk: pālayati. pālayati would also follow from pāla rakṣaṇe, so the only result is that puk is excluded (BM, TB).

G. lī, bhī, smi: ātmanepada without the agent's fruit; ā, ṣuk when fear or wonder comes from the prompter — running example: bhī + ṇic (fear from the prompter) → bhai-i → bhā-i (2593) → bhāp-i (puk) → muṇḍo bhāpayate (ātmanepada, 2594); or bhī-ṣ-i (ṣuk, 2595) → bhīṣayate

• 2592 liyaḥ saṃmānanaśālīnīkaraṇayoś ca 1.3.70 — ātmanepada comes after lī and līṅ ending in ṇi in the senses of honouring and overpowering. By ca it also comes in the sense of deceiving, from gṛdhivañcyoḥ pralambhane (1.3.69, SK 2739). It applies even when the fruit does not go to the agent, since ṇicaś ca already covers the case where it does (BM).
• Carried in: ātmanepada from anudāttaṅita ātmanepadam (1.3.12, SK 2157), and ṇeḥ from ṇer aṇau… (1.3.67, SK 2738).
• With ā by the vārttika under 2591, and puk:
• jaṭābhir lāpayate (he gains honour by his matted locks);
• śyeno vartikām ullāpayate (the hawk overpowers the quail; print "vartikām");
• bālam ullāpayate (he deceives the child).
• 2593 bibheter hetubhaye 6.1.56 — ā optionally replaces the ec of bhī before ṇi when the fear comes from the hetu (the prompter). Carried in: ecaḥ and āt from SK 2370, vibhāṣā from SK 2509, and ṇau from cisphuror ṇau (6.1.54, SK 2569) (BM, TB). "ātvaṃ vā" is the print's reading.
• Its forms appear under 2594 and 2595: bhāpayate (ā, puk) and bhīṣayate (no ā, ṣuk).
• 2594 bhīsmyor hetubhaye 1.3.68 — ātmanepada comes after bhī and smi ending in ṇi when fear (or wonder) comes from the hetu. bhaya stands for the sense of each root (upalakṣaṇa), so it covers smaya for smi. The sixth case is used for the fifth (vyatyaya, TB). Carried in: SK 2157 and SK 2738. The rule is for the case where the fruit is not the agent's (BM).
• muṇḍo bhāpayate, with ā (2593), puk and ātmanepada.
• 2595 bhiyo hetubhaye ṣuk 7.3.40 — the augment ṣuk is added to bhī ending in ī before ṇi when the fear comes from the hetu. The sūtra fuses an ī ("bhī ī"), so there is no ṣuk in the ā option (BM). ṇau is carried from SK 2570.
• bhīṣayate.
• TB's print reads "in the ā option only bhāyayate". The translator takes this as probably a misprint, since it conflicts with bhāpayate.
• 2596 nityaṃ smayateḥ 6.1.57 — ā invariably replaces the ec of smi before ṇi when the wonder comes from the hetu. nitya stops the vibhāṣā. hetubhaye is carried from 2593, standing here for wonder alone (BM).
• jaṭilo vismāpayate (the ascetic with matted locks astonishes).
• With an instrument, not the hetu, there is no ā, no ṣuk and no ātmanepada: kuñcikayainaṃ bhāyayati, vismāyayati ("with a kuñcikā", a hair-knot).
• Objection: how then can the Raghuvaṃśa have "vismāpayan vismitam ātmavṛttau"? The wonder there comes from an instrument, human speech. Otherwise śānac too would come.
• Reply:
• The traditional scholars read vismāyayan.
• Or else: human speech, the prayojya-kartrī, astonishes the king (vismāpayate), and the lion does so through it. That makes vismāpayan a śatṛ after a second ṇi on smāpi, which is parasmaipada because 2594 does not reach a doubly ṇic-final stem (BM).
• TB's first view of prayojyakartrī (the second ṇic) differs from BM's (a bahuvrīhi). The English stays neutral.

H. Root-specific substitutes before ṇi: final v / t / p, ā for an ec, gāṅ for iṅ — running example: adhi + iṅ + ṇic → adhi-ai-i → (ā, 2600) adhi-ā-i → (puk) adhyāpayati; luṅ adhyajīgapat (gāṅ, 2601) / adhyāpipat

• 2597 sphāyo vaḥ 7.3.41 — v replaces the final y of sphāy before ṇi. ṇau is carried from SK 2570 (BM). "The final" is the translator's addition, by alo 'ntyasya (1.1.52, SK 42).
• sphāvayati.
• 2598 śader agatau taḥ 7.3.42 — t replaces the final of śad before ṇi, except in the sense of going. The a of "taḥ" is for pronunciation only (BM).
• śātayati.
• In the sense of going: gāḥ śādayati govindaḥ (makes the cows go).
• 2599 ruhaḥ po 'nyatarasyām 7.3.43 — p optionally replaces the final of ruh before ṇi (ṇau śeṣa, as in 2597).
• ropayati / rohayati.
• 2600 krīṅjīnāṃ ṇau 6.1.48 — ā replaces the ec of krī, iṅ and ji before ṇi. ecaḥ and āt are carried from SK 2370 (BM). That the ec is the vṛddhi vowel is the translator's reading. Then puk.
• krāpayati, adhyāpayati (adhi + iṅ), jāpayati.
• The commentators' luṅ forms: acikrapat, ajījapat.
• BM's lead-in to 2601: "adhi i i a t iti sthite".
• 2601 ṇau ca saṃścaṅoḥ 2.4.51 — when a ṇi followed by san or caṅ is intended, gāṅ optionally replaces iṅ. The locative is one of domain (viṣayasaptamī, BM): "when ṇi is intended". Carried in: iṅaḥ from iṅaś ca (2.4.48, SK 2616), gāṅ from gāṅ liṭi (2.4.49, SK 2459), and vibhāṣā from vibhāṣā luṅlṛṅoḥ (2.4.50, SK 2460).
• The text is the print's (NSP p. 432, scan page-0442). Ours has it under 2579.
• adhyajīgapat, with gāṅ: gāṅ before ṇi, then puk and shortening give adhi gap-i-a-t. gap is doubled, and halādiḥ śeṣaḥ, shortening, i by sanvadbhāva, lengthening and cutva follow.
• The bar on substitutes before ṇi does not stop gāṅ. It holds only when the part after the abhyāsa has an a, and with gāṅ withheld only the ṇic would be doubled (BM, TB).
• adhyāpipat, without gāṅ: ā (2600) and puk give adhi āp-i-a-t, and pi is doubled.
• 2602 sidhyater apāralaukike 6.1.49 — ā replaces the ec of sidh (sidhyati) before ṇi when the sense does not concern the other world. ecaḥ and āt are carried from SK 2370, and ṇau from SK 2600 (BM). The śyan citation excludes the bhvādi root (TB).
• annaṃ sādhayati (produces the food), with ā and puk.
• Why "not the other world"? tāpasaḥ siddhyati (print spelling) means he ascertains the truth. Hence sedhayati tāpasaṃ tapaḥ, with guṇa and no ā.
• 2603 prajane vīyateḥ 6.1.55 — ā optionally replaces the ec of vī before ṇi in the sense of conceiving. Carried in: ecaḥ and āt from SK 2370, ṇau from SK 2569, and vibhāṣā from SK 2509 (BM).
• vāpayati / vāyayati gāḥ purovātaḥ (the east wind makes the cows conceive).
• vīyateḥ: the adādi vī. Whether the yak citation also takes in vyeñ is disputed: TB's "some" say yes, BM says no. The English stays neutral.
• Recalled: ūd upadhāyā gohaḥ (6.4.89, SK 2364): gūhayati. The ū is an apavāda of laghūpadhaguṇa, and it leads into 2604.

I. The close: ū for duṣ, ṇ for both n's of an, gam for iṇ; the end of the ṇyanta — running example: duṣ + ṇic → dūṣ-i (ū, not guṇa) → dūṣayati; pra + an-i + caṅ → ('ni' doubled) → prāṇiṇat; iṇ + ṇic → gam-i → gamayati

• 2604 doṣo ṇau 6.4.90 — ū replaces the penultimate of duṣ (duṣyati) before ṇi. ūt and upadhāyāḥ run on from SK 2364. duṣ is cited as doṣ, its guṇa form, so the rule applies only where guṇa is due. "duṣaḥ" would have been easy to say (BM).
• dūṣayati.
• ṇau kim? doṣaḥ (TB).
• 2605 vā cittavirāge 6.4.91 — the ū is optional when the mind is estranged (virāga = aprītatā). It is carried from SK 2364 and 2604.
• cittaṃ dūṣayati doṣayati vā kāmaḥ (desire estranges the mind, turning it from the daily rites; BM, TB).
• Recalled:
• mitāṃ hrasvaḥ (6.4.92, SK 2568): the mit roots of the bhvādi and curādi: ghaṭayati.
• The gaṇasūtra janījṝṣ[knasurañjo 'mantāś ca] (expansion the translator's): janayati, jarayati. But jṛṇāti, which is not ṣit, gives jārayati.
• vārttika (sk.2605.k2) rañjer ṇau mṛgaramaṇe nalopaḥ: for rañj before ṇi, the n is lost in the sense of hunting.
• rajayati mṛgān.
• Not deer: rañjayati pakṣiṇaḥ.
• Not hunting: rañjayati mṛgāṃs tṛṇadānena.
• The curādi ciñ is jñapādi and mit (cisphuror ṇau, SK 2569): capayati, cayayati. The svādi ci is not mit: cāpayati, cāyayati.
• sphur: sphārayati, sphorayati. In the caṅ-aorist, sphu is doubled with its u, then optional ā: apusphurat / apuspharat (shortened).
• 2606 ubhau sābhyāsasya 8.4.21 — both n's of an, together with its abhyāsa, become ṇ when the condition, an r in an upasarga, is present. Carried in: aniteḥ (8.4.19, SK 2478) and upasargāt (8.4.14, SK 2287), under the adhikāra of raṣābhyāṃ no ṇaḥ samānapade (8.4.1, SK 235) (BM).
• pra an-i-a-t: ṇatva by aniteḥ is asiddha, so ni is doubled. The later n is then separated from the r by the abhyāsa, and this rule gives ṇ to both: prāṇiṇat.
• The rule shows that pūrvatrāsiddhīyam advirvacane is not universal (anitya), hence ūrṇunāva (BM, TB).
• TB: sābhyāsasya kim? prāṇ namati. "ubhau" secures both at once.
• 2607 ṇau gamir abodhane 2.4.46 — gam replaces iṇ before ṇi, except in the sense of making known. iṇaḥ is carried from iṇo gā luṅi (2.4.45, SK 2458). The i of "gami" is for pronunciation (BM).
• gamayati (makes reach).
• Making known: pratyāyayati. BM and TB give its luṅ as pratyāyiyat: vṛddhi and āy, being para, set aside iṇo yaṇ (6.4.81, SK 2455), and the ṇic-bearing yi is doubled.
• vārttika (sk.2607.k2) iṇvad ikaḥ: ik smaraṇe is like iṇ: adhigamayati (reminds).
• Recalled: hanas to 'ciṇṇaloḥ (7.3.32, SK 2574) gives t for the n of han before ṇi, and ho hanteḥ… (7.3.54, SK 358) velarises the h: ghātayati (luṅ ajīghatat, BM). Then īrṣyayati.
• vārttika (sk.2607.k4) īrṣyates tṛtīyasyeti vaktavyam: for īrṣy, doubling "of the third". There are two readings:
• the third consonant. This stops the ṣ from being doubled (r is barred by na ndrāḥ): airṣyiyat (ours, the print, TB; BM and WS read airṣyyat).
• the third ekāc. This works in the sannanta as an apavāda of ajāder dvitīyasya (6.1.2, SK 2176): īrṣyiyiṣati (TB). With caṅ there is no third ekāc, so ṣyi, the second, is doubled, and halādiḥ śeṣaḥ (7.4.60, SK 2179) leaves only ṣ in the abhyāsa: airṣiṣyat.
• nivṛttapreṣaṇa: when the prompting has ceased, a dhātu with hetumaṇṇic keeps the plain dhātu's meaning. This accounts for Māgha's "prārthayanti śayanotthitaṃ priyāḥ" (parasmaipada, though arth is āgarvīya). "The same is to be worked out for all transitive dhātus."
• Close: "[iti ṇyantaprakriyā]" (Here ends the ṇyanta-prakriyā), supplied from the print. TB closes alike. p55 opens with the sannanta, SK 2608.