तिङन्तदिवादिप्रकरणम्

Covers SK 2505–2522 (18 sūtras): all present. SK 2521 has no kaumudī of its own, in ours or in the print. The print sets 2521 and 2522 one after the other with a single gloss ("śyan vā syāt") and the examples after 2522, so both are placed in group I. Like p46, the spine is not an Aṣṭādhyāyī run. The SK follows the Dhātupāṭha order of the divādi roots (1107 divu … 1246 gṛdhu) and teaches each rule at the root where it first bites, drawing on AS books 3, 6, 7 and 8 (3.1.69 → 7.2.57 → 6.4.63 → 6.1.50 … 3.1.72). Every AS-ref below is checked against references/ashtadhyayi/sutraani-data.json.

• Where the anuvṛtti comes from. Many rules here carry words from sūtras the SK teaches elsewhere, earlier and later. Examples:
• sārvadhātuke yak (3.1.67, SK 2756) gives sārvadhātuke at 2505, and yak is how 2509 cites līyati.
• udito vā (7.2.56, SK 3328) gives vā at 2506.
• ghor lopo leṭi vā (7.3.70, SK 3584) gives lopaḥ at 2510.
• nodāttopadeśasya… (7.3.34, SK 2763) and āto yuk ciṇkṛtoḥ (7.3.33, SK 2761) give na and ciṇkṛtoḥ at 2512.
• na bhābhūpū… (8.4.34, SK 2840) gives na at 2518.
• The outline names them where BM does.

tiṅantadivādiprakaraṇam — outline (p47, SK 2505–2522, all 18 sūtras)

p46 (tiṅantajuhotyādiprakaraṇam) closed at 2504 with 1106 gā and "iti juhotyādayaḥ".

• Where the prakaraṇa begins. The print heads it "tiṅante divādayaḥ 44" (NSP p. 401, scan page-0411). Our prakaraṇa number 47 differs systemically, but the name agrees. sk2505 opens with the Dhātupāṭha entry 1107 divu and then gives divādibhyaḥ śyan (3.1.69). "jhṛṣantāḥ parasmaipadinaḥ": the roots up to jhṝṣ (1107–1131) are parasmaipadin.
• Where it ends. At 2522 with 1246 gṛdhu and "vṛt". Then comes the closing sentence: "the puṣādi and the divādi are finished". Some take the divādi as an ākṛtigaṇa, which accounts for kṣīyate and mṛgyati. The colophon "iti divādayaḥ" is in ours and in the print (NSP p. 406, scan page-0416), so no bracket is needed. The print then heads "tiṅante svādayaḥ 45", and p48 opens with ṣuñ (SK 2523).
• Contents in outline.
• 2505: div. śyan replaces śap, and śyan is ṅit, so there is no guṇa: dīvyati, dideva, adevīt; ṣivu … nṛtī.
• 2506: nṛt. Optional iṭ before an s-initial ārdhadhātuka other than sic: nartiṣyati / nartsyati, anartīt; trasī … ṣūṅ, dūṅ, dīṅ.
• 2507–2509: dīṅ, mī/mi, lī/līṅ. yuṭ before a vowel-initial kṅit (didīye), and ā in the domain of an ec and before lyap: dātā, adāsta; optionally for lī: letā / lātā. Then the gaṇasūtra svādaya oditaḥ.
• 2510: śo, cho, ṣo, do. o is lost before śyan: śyati, chyati, syati, dyati.
• 2511–2513: janī … liśa (the ātmanepada block). jā before a śit affix (jāyate). No vṛddhi of jan before ciṇ (ajani). ciṇ after pad before ta (praṇyapādi). Also tapa/vāvṛtu, ṇah, budh, yudh, sṛj and rādh.
• 2514: śliṣ. ksa only in the sense of embracing: aślikṣat, but samaśliṣat (aṅ) and passive aśleṣi / aślikṣata.
• 2515–2518: the radh group. Optional iṭ (rarandhitha / raraddha), no num before a non-liṭ iṭ (radhitā), num before a jhal for naś (nanaṃṣṭha, naṃṣṭā), and no ṇ for a ṣ-final naś (pranaṃṣṭā). Also tṛp, dṛp, druh, muh, ṣṇuh, ṣṇih.
• 2519: the śam group. A long vowel before śyan: śāmyati, tāmyati, kṣāmyati, mādyati.
• 2520–2522: as and yas. thuk before aṅ (āsthat), and optional śyan for yas without an upasarga or after sam: yasyati / yasati, saṃyasyati / saṃyasati. The run jasu … gṛdhu follows.

Threads that run through the prakaraṇa:

• śyan is the class-mark. divādibhyaḥ śyan (2505) replaces śap. śyan is not pit, so it counts as ṅit (sārvadhātukam apit, 1.2.4, SK 2234; BM). As a result:
• There is no guṇa of the root: dīvyati, nṛtyati, puṣyati, śuṣyati, kṣubhyati.
• A nasal penultimate is lost before it (aniditām…, 6.4.24, SK 415): rajyati, bhraśyati.
• It triggers saṃprasāraṇa (grahijyā…, 6.1.16, SK 2412): vidhyati.
• Four rules act on the root before śyan specifically: lengthening of div by hali ca (8.2.77, SK 354); o-loss (2510); jā for jan (2511); the long vowel of the śam group (2519).
• For a few roots śyan is optional, alternating with śap: trasī, bhramu, klamu by vā bhrāśa… (3.1.70, SK 2321), and yas (2521–2522).
• The pada sections of the gaṇa, as the kaumudī announces them and BM fixes their ends:
• 1107–1131, divu … jhṝṣ, parasmaipadin ("jhṛṣantāḥ"; BM printed under sk.2574.b1);
• 1132–1144, ṣūṅ … prīṅ, ātmanepadin (ṅit);
• "four parasmaipadin", śo … do (1145–1148);
• "fifteen ātmanepadin", janī … vāśṛ (1149–1163);
• "five svaritet", mṛṣa … śapa (1164–1168), ubhayapadin;
• "eleven anudāttet", pada … liśa (1169–1179);
• "parasmaipadin to the end of the gaṇa", from rādh (1180).
• The gaṇasūtra svādaya oditaḥ (under 2509) makes ṣūṅ … vrīṅ (1132–1140) odit, so the niṣṭhā t becomes n (oditaś ca, 8.2.45, SK 3019): prasūnam, ḍīnaḥ.
• Three sub-classes inside the gaṇa:
• puṣādi, from 1182 puṣa to the end of the gaṇa (the "siddhānta" of 2518). It takes aṅ in the parasmaipada luṅ (puṣādidyutādy-ḷditaḥ parasmaipadeṣu, 3.1.55, SK 2343): apuṣat, aśuṣat, aradhat, anaśat, adruhat, aśamat, āsthat, avyuṣat, agṛdhat.
• radhādi, the eight roots radh, naś, tṛp, dṛp, druh, muh, ṣṇuh and ṣṇih (1193–1200, closed by "vṛt"). They are veṭ (2515).
• śamādi, the eight roots śam, tam, dam, śram, bhram, kṣam, klam and mad (1201–1208). They take the long vowel before śyan (2519).
• Which luṅ a root takes. The prakaraṇa sorts four luṅ affixes:
• aṅ for the puṣādi, in parasmaipada only. That is why 2520 needs asyativakti… (SK 2438) for aṅ in the ātmanepada, as in paryāsthata.
• ksa by śala igupadhād aniṭaḥ ksaḥ (3.1.45, SK 2336) for an aniṭ root that ends in a sibilant with an ik penultimate: alikṣata. For śliṣ, 2514 restricts it to the sense of embracing.
• ciṇ, optional in the ta of dīp, jan, budh and pūr by dīpajanabudha… (3.1.61, SK 2328): ajani / ajaniṣṭa, abodhi / abuddha. It is obligatory for pad by 2513: praṇyapādi.
• sic elsewhere: anartīt, avyoṣīt, and sic by the vārttika spṛśamṛśa… for tṛp and dṛp (atārpsīt).
• ā for an ec-final or ec-producing root (ādeca upadeśe 'śiti, 6.1.45, SK 2370):
• śo, cho, ṣo and do → ā outside the śit domain: śaśau, śātā, dadau, praṇidātā, adāt.
• 2508 extends ā to mī, mi and dīṅ in the domain of an ec and before lyap: dātā, adāsta.
• 2509 makes it optional for lī and līṅ: letā / lātā.
• dīṅ with ā is not ghu for sthāghvor ic ca: adāsta, not *adita (BM, citing a Bhāṣya vārttika).
• The kaumudī lists forms bare. Lakāra and person labels are the translator's, given in Pāṇinian terms. Derivation stages follow BM and TB as recorded in the unit notes.
• Dhātu numbering. In our source the numbers from asu onward run one lower than the print: ours 1209 asu = print 1210, and ours 1246 gṛdhu = print 1247. The English gives the numbers as transmitted. The print misnumbers jhṝṣ as 2131 (for 1131).
• Where the commentary lies. Only the kaumudī blocks are translated (SK-TRANSLATE §25).
• TB on the divādi opening (divu, ṣivu, nṛtī; NSP pp. 401–402, scan pages 0411–0412) is in no source.json. The exception is its last word, 'siddham', misfiled at the tail of sk.2504.t1. It is transcribed from the scan in sk2505's notes.
• BM on the divādi opening (jhṛṣantāḥ = up to jṝṣ/jhṝṣ; dīvu is udit) is misfiled at the tail of sk.2574.b1, after "atha divādayaḥ".
• TB printed under a neighbouring unit:
• 2507 at the tail of sk.2506.t1;
• 2509 and 2510 inside sk.2508.t1;
• 2512 in sk.2511.t1, from "—ñiti ṇiti";
• 2514 at the tail of sk.2513.t1, where the first line lies under the scan watermark;
• 2515 in sk.2514.t1;
• 2517 at the tail of sk.2516.t1;
• 2518 under sk.2517.t1;
• 2519's dhātu lines under sk.2518.t1.
• There is no t1 for 2505, 2507, 2509 or 2510. The print has no TB on 2520. The tail of sk.2522.t1 (ṣuñ, sunva, sunavāni) is TB on SK 2523 (p48).
• BM on 2521 is printed under sk.2522.b1; sk.2521.b1 carries only the pratīka.
• BM lead-ins to the next sūtra close these blocks: sk.2511.b1 (→ 2512), sk.2513.b1 (→ 2514), sk.2515.b1 (→ 2516) and sk.2516.b1 (→ 2517, "nanaś-tha").

Where the p47 audit (.work/sk-run/p47-53-audit-fixes.md and the reports .work/sk-run/p47-53-audit/sk2505-2513.md and sk2514-2522.md) corrects a reading, this outline follows the corrected reading. The fixer is applying these corrections to en.yml in parallel.

• 2514 (MUST-FIX): aniṭaḥ qualifies the root, not cli.
• aniṭaḥ is the ablative carried from śala igupadhād aniṭaḥ ksaḥ (SK 2336). The rule reads: "after this [root], when it is aniṭ, ksa replaces cli".
• TB (under sk.2513.t1): "aniṭaḥ kim? — so that it does not apply to the seṭ bhvādi śliṣu dāhe: aśleṣīt".
• cli never takes iṭ, so "an aniṭ cli" says nothing.
• 2505: BM states that "ending with jhṝṣ" covers 1107–1131 (under sk.2574.b1). It is not the translator's inference.
• 2508: the English should use one construal of ejnimitte in both places, the neutral "before / in the domain of an aśit affix that occasions an ec". BM glosses it as the locative of what follows (parasaptamī, "pare"). TB insists on a locative of domain (viṣayasaptamī), and that matters for upadāyaḥ: ā first, then ghañ and yuk; if erac came first, the result would be *upadā.
• 2510: the print's number after 'yamarama' reads clearly as 2337 on a crop. That is a misprint for 2377 (yamaramanamātāṃ sak ca); SK 2337 is ksasyāci. In TB's verse, 'rāghava' ends in the saṃbuddhi ending.
• 2512: Bhaṭṭoji writes only "tato vāvṛtyamānā … iti bhaṭṭiḥ", with no "hence". So: "Bhaṭṭi [has]: 'tato vāvṛtyamānā…'".
• 2516: ours and the print read "aṅi num", without the "tu" found in Wikisource and BM's pratīka. The outline claims no contrast.
• 2518: "ṇatvaṃ na syāt" is Bhaṭṭoji's own wording. It spells out the anuvṛtti of ṇaḥ (SK 235) and na (SK 2840), so it is not bracketed supply.
• 2520: the "upasargād asyatyūhyoḥ" that the kaumudī says "will be stated" is the vārttika upasargād asyatyūhyor veti vācyam, under SK 2701 (sk.2701.k4, p61).

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

• 2505: deviṣyati, sasnāsa, "pluṣa ca" and krīḍā. The print shares the slip kriḍā, and krīḍā is the Dhātupāṭha reading. For nyaṣevīt the print glyph is undecidable, but BM, TB and Wikisource read nyaṣevīt.
• 2506: svaratisūti.
• 2509: lāsyate and niṣṭhānatvam.
• 2510: the missing clause "iti sico vā luk | lugabhāve yamarama–", chedane / chyati, and prādurbhāve.
• 2512: vṛtudhātoḥ, nahyate, śapyate and anudāttetaḥ.
• 2513: taśabde, praṇyapādi, apatsata, 'yam ity eke, rādho and pupoṣa.
• 2514: asmāt, kṣudhita and saṃrāddhiḥ.
• 2516: liḍvarje.
• 2518: pranaṃṣṭā, vaicityam and puṣādayas.
• 2519: aṣit.
• 2520: aṅi pare.
• 2522: ucyati, tuṣṭau, krodhe, avṛtkṛtatvāt and kryāditvāt.

Still open for Ajai (mūla):

• 2506: ours and the print read jīryeta. jṝṣ is parasmaipadin, and Wikisource reads jīryet. The English gives jīryet.
• 2513: after "virādhyantaṃ kṣameta kaḥ" the print has three sentences that ours lacks: "druhyantam ity arthaḥ | rādhyaty odanaḥ | sidhyatīty arthaḥ". The English supplies them in brackets.
• 2514:
• Ours and the print read "samaśliṣaj jatu kāṣṭham | aṅ"; BM and Wikisource read "samāśliṣaj … āṅ". The English follows the print, and BM's reading is recorded.
• Ours and the print read aśliḍhvam; BM, TB and Wikisource read aśliḍḍhvam. It may be an optional form.
• 2517: ours and the print read nanaṣṭa; Wikisource and BM's pratīka read nanaṃṣṭha. Only nanaṃṣṭha shows the rule, and the English gives it.
• 2522:
• bisa / busa: the print is internally mixed (visa, visyati, abisat, vusa). The English follows ours.
• pyus / yus: this follows the print. BM's text evidently had vyus / vus.

Ordering: SK order throughout.

A. div: śyan for śap, and no guṇa before it — running example: div + śap + ti → div + śyan + ti → div + ya + ti → dīv + ya + ti → dīvyati

• 2505 divādibhyaḥ śyan 3.1.69 — after div and the rest, śyan comes. It is an apavāda of śap, from kartari śap (3.1.68, SK 2167). kartari is carried from that rule and sārvadhātuke from sārvadhātuke yak (3.1.67, SK 2756) (BM). The ś and n of śyan are it.
• div + ya + ti: śyan, being apit, is ṅit, so there is no guṇa (BM). The i is lengthened before the consonant by hali ca (SK 354): dīvyati.
• Outside the śit domain the root is plain and seṭ, with guṇa: dideva, devitā, deviṣyati (print), adevīt, adeviṣyat. dīvyatu, adīvyat, dīvyet, dīvyāt.
• Root entry 1107 divu, with ten senses. Its kānti means desire (icchā), since shining is listed separately (TB). The u-marking gives optional iṭ before ktvā and none in the niṣṭhā (udito vā, SK 3328): devitvā / dyūtvā, dyūtam (TB, scan only; BM under sk.2574.b1).
• 1108 ṣivu: pariṣīvyati and pariṣiṣeva, with ṣ by parinivibhyaḥ… (SK 2275). The restriction of SK 2277 is off here because it carries "prāk sitāt". nyaṣevīt / nyasevīt is optional by sivādīnāṃ vā… (SK 2359).
• 1109–1115:
• srivu and ṣṭhivu. "Some do not read" ṣṭhivu.
• ṣṇusu: snusyati, suṣṇosa.
• ṣṇasu: sasnāsa (print).
• knasu: caknāsa.
• vyuṣa and pluṣa: vuvyoṣa, and "pluṣa ca" (print). Both are read again in the puṣādi at 2522.
• 1116 nṛtī gātravikṣepe: nṛtyati, nanarta. The ī-marking denies iṭ in the niṣṭhā (śvīdito niṣṭhāyām, SK 3039). TB says it shows yasya vibhāṣā (SK 3025) to be anitya. This leads into 2506.

B. nṛt and four others: optional iṭ before an s-initial non-sic ārdhadhātuka — running example: nṛt + sya + ti → nart + i + ṣya + ti → nartiṣyati / nart + sya + ti → nartsyati; but luṅ a-nart + iṭ + sīt → anartīt (iṭ invariable)

• 2506 se 'sici kṛtacṛtacchṛdatṛdanṛtaḥ 7.2.57 — after kṛt, cṛt, chṛd, tṛd and nṛt, iṭ is optional before an s-initial ārdhadhātuka other than sic.
• iṭ and ārdhadhātukasya come from ārdhadhātukasyeḍ valādeḥ (7.2.35, SK 2184), and vā from udito vā (7.2.56, SK 3328). "se" and "asici" are locatives standing for sixth cases. The rule makes optional an iṭ that would otherwise be obligatory (BM).
• nartiṣyati / nartsyati.
• anartīt: because the rule says "asici", iṭ before sic stays obligatory (BM, TB).
• nṛtyet, nṛtyāt. TB adds ninartiṣati / ninṛtsati and the kṛt forms.
• 1117 trasī: śyan is optional by vā bhrāśa… (SK 2321), giving trasyati / trasati. The e with abhyāsa-loss is optional by vā jṝbhramutrasām (SK 2356), giving tresatuḥ / tatrasatuḥ.
• 1118–1129: kutha, putha, gudha, kṣipa (kṣeptā), puṣpa (puṣpyati, pupuṣpa), tima / ṣṭima / ṣṭīma, vrīḍa, iṣa, and ṣaha / ṣuha. "The sense of cak is satisfaction (tṛpti)."
• kutha pūtībhāve: Bhaṭṭoji glosses it daurgandhya (stinking). TB takes it as pavitrībhavana (becoming pure), and BM gives both. The English keeps Bhaṭṭoji's gloss.
• 1130–1131 jṝṣ, jhṝṣ vayohānau:
• jīryati: i by ṝta id dhātoḥ (SK 2390), then hali ca.
• jajaratuḥ / jeratuḥ: guṇa by ṛcchaty ṝtām (SK 2383), then the option of SK 2356. na śasadada… keeps SK 2260 off (TB).
• jaritā / jarītā by vṝto vā (SK 2391).
• jīryet: ours and the print read jīryeta (mūla, for Ajai).
• luṅ ajarat: aṅ is optional by jṝstambhu… (SK 2291), with guṇa by ṛdṛśo 'ṅi guṇaḥ (SK 2406). Otherwise ajāriṣṭām.
• From jhṝṣ: jhīryati, ajhārīt.
• 1132 ṣūṅ prāṇiprasave: sūyate, suṣuve.
• suṣuviṣe: svaratisūti… (SK 2279, print) would make iṭ optional. The prohibition śryukaḥ kiti (SK 2381) sets that option aside, even though SK 2279 is the later rule, because the prohibition section was begun earlier (BM, TB). The krādi-restriction (SK 2293) then makes iṭ obligatory.
• sotā / savitā.
• 1133 dūṅ: dūyate. 1134 dīṅ kṣaye: dīyate, which leads into 2507.

C. dīṅ, mī/mi, lī/līṅ: yuṭ before a vowel-initial kṅit; ā in the domain of an ec — running example: dī-dī + e → didī + yuṭ + e → didīye; dī + tā → (e due) dā + tā → dātā; lī + tā → letā / lātā

• 2507 dīṅo yuḍ aci kṅiti 6.4.63 — after dīṅ, an ac-initial kit or ṅit ārdhadhātuka takes the augment yuṭ.
• ārdhadhātuke is the adhikāra of SK 2307, qualified by "ac" through tadādividhi. dīṅaḥ is a fifth case, and the seventh "aci kṅiti" stands for a sixth (BM; TB under sk.2506.t1).
• didī + e: yaṇ by er anekācaḥ… (SK 272) is the later rule, but yuṭ is nitya and applies first: didīye (BM, TB).
• vārttika (sk.2507.k2) vugyuṭāv uvaṅyaṇoḥ siddhau vaktavyau: vuk and yuṭ count as valid against uvaṅ and yaṇ. yuṭ falls in the ābhīya section and would otherwise be asiddha, which would let yaṇ give *didyye (BM, TB).
• TB's counter-examples: dedīyate (not aci) and upadānam (not kṅit).
• 2508 mīnātiminotidīṅāṃ lyapi ca 6.1.50 — ā replaces mī (mīnāti), mi (minoti) and dīṅ before lyap. By ca it also replaces them in the domain of an aśit affix that occasions an ec.
• āt comes from ādeca upadeśe 'śiti (6.1.45, SK 2370). ca gathers in ecaḥ and aśiti from the same rule, with nimitte supplied (BM).
• Construal: the English should read the same neutral "in the domain of / before" in sutra_en and k1 (audit). BM reads a locative of what follows. TB reads a locative of domain, so that in upadāyaḥ ā precedes ghañ and yuk (āto yuk ciṇkṛtoḥ, SK 2761); if erac (SK 3231) came first, the result would be *upadā. TB's lyap forms: pramāya, upadāya.
• dī + tā: tā would bring guṇa (e, an ec), so that is the domain for ā: dātā, dāsyate.
• adāsta, adāsthāḥ: once ā comes, dīṅ would be ghu, but sthāghvor ic ca (SK 2389) is barred by the vārttika "sthāghvor ittve dīṅaḥ pratiṣedhaḥ" (BM). TB also reports a view that dīṅ is never ghu.
• 1135 ḍīṅ: ḍīyate, ḍiḍye. It is svādi-odit, so ḍīnaḥ (TB).
• 1136 dhīṅ: dhīyate, didhye, dhetā.
• 1137 mīṅ hiṃsāyām: mīyate. "hiṃsā here is separation from life."
• 1138 rīṅ: rīyate. 1139 līṅ, which leads into 2509.
• 2509 vibhāṣā līyateḥ 6.1.51 — for lī and līṅ, ā is optional in the domain of an ec and before lyap.
• "līyateḥ" cites the root with yak (sārvadhātuke yak, SK 2756), not with śyan. So it covers both the divādi līṅ and the kryādi lī śleṣaṇe (BM, TB). āt, lyapi and the ec condition are carried from SK 2370 and 2508.
• lī + tā → letā / lātā; leṣyate / lāsyate (print). TB's lyap forms: vilāya / vilīya.
• Why "in the domain of an ec"? līyate, lilye.
• 1140 vrīṅ vṛṇotyarthe: vrīyate, vivriye.
• gaṇasūtra svādaya oditaḥ: ṣūṅ … vrīṅ (1132–1140) are odit, "its effect is n for the niṣṭhā" (niṣṭhānatvam, print; oditaś ca, SK 3019).
• 1141–1144: pīṅ (pīyate); māṅ (māyate, where ghumāsthā… (SK 2462) does not act, being ārdhadhātuka-bound, TB); īṅ (ayāṃcakre); prīṅ (prīyate, "sakarmaka"). BM: that is, when prīti is gratifying (tarpaṇa).
• "Now four parasmaipadin roots": 1145 śo tanūkaraṇe, which leads into 2510.

D. śo, cho, ṣo, do: o lost before śyan — running example: śo + śyan + ti → ś + ya + ti → śyati; outside śyan śo → śā: śaśau, śātā

• 2510 otaḥ śyani 7.3.71 — the final o of the root is lost before śyan. lopaḥ is carried from ghor lopo leṭi vā (7.3.70, SK 3584) (BM).
• śyati, śyataḥ, śyanti.
• Outside śyan, ā comes by SK 2370, with au for ṇal (āta au ṇalaḥ, SK 2371): śaśau, śaśatuḥ, śātā, śāsyati.
• luṅ: sic is optionally lost by vibhāṣā ghrādheṭ… (2.4.78, SK 2376), giving aśāt, aśātām, aśuḥ (jus by ātaḥ, SK 2227). Otherwise iṭ and sak come by yamaramanamātāṃ sak ca (SK 2377), giving aśāsīt, aśāsiṣṭām. The clause "iti sico vā luk | lugabhāve yamarama–" is missing from ours and is supplied from the print. The print itself misnumbers 2377 as 2337 (audit).
• 1146 cho chedane: chyati (print; required by this rule).
• 1147 ṣo antakarmaṇi: syati, sasau. abhiṣyati, abhyaṣyat (ṣ by SK 2270 and 2276). abhisasau takes no ṣ, by sthādiṣv… (SK 2277).
• 1148 do avakhaṇḍane: dyati, dadau. praṇidātā has ṇ by ner gada… (SK 2285). deyāt (er liṅi, SK 2374); adāt (gātisthā…, SK 2223).
• "Now fifteen ātmanepadin roots": 1149 janī prādurbhāve (print). Its ī-marking is for śvīditaḥ. This leads into 2511.

E. janī … liśa (the ātmanepada block): jā before a śit affix; no vṛddhi of jan before ciṇ; ciṇ after pad before ta — running example: jan + śyan + te → jā + ya + te → jāyate; a-jan + ciṇ + ta → a-jan-i-(ta) → ajani (not *ajāni); pra-ni-a-pad + ciṇ + ta → praṇyapādi

• 2511 jñājanor jā 7.3.79 — jā replaces jñā and jan before a śit affix. śiti comes from ṣṭhivuklamucamāṃ śiti (7.3.75, SK 2320) (BM).
• jāyate. For jñā, which takes śnā: jānāti.
• Why not a short ja, which ato dīrgho yañi (SK 2170) would lengthen anyway? The long jā shows that an aṅga operation is not repeated once an aṅga operation has been done. BM and TB both call it a jñāpaka.
• Outside the śit domain there is no jā. In jajñe, jajñāte, jajñire, the penultimate is lost by gamahana… (SK 2363) and n → ñ by ścutva (BM). janitā, janiṣyate.
• "ciṇ optionally" by dīpajana… (3.1.61, SK 2328). With a-jan-i-ta standing, vṛddhi of the penultimate would be due (BM lead-in), which leads into 2512.
• 2512 janivadhyoś ca 7.3.35 — for jan and vadh, there is no vṛddhi of the penultimate before ciṇ, or before a ñit or ṇit kṛt.
• Five anuvṛttis (BM): upadhāyāḥ from ata upadhāyāḥ (SK 2282); vṛddhiḥ from mṛjer vṛddhiḥ (SK 2473); na from nodāttopadeśasya… (SK 2763); ciṇkṛtoḥ from āto yuk ciṇkṛtoḥ (SK 2761); ñṇiti from aco ñṇiti (SK 254).
• ajani (ciṇ; the ta is lost by ciṇo luk), or ajaniṣṭa (sic). TB's kṛt forms: janaḥ (ghañ), janakaḥ.
• 1150–1158: dīpī (adīpi / adīpiṣṭa, by the same option of SK 2328), pūrī (apūri / apūriṣṭa), tūrī, dhūrī, gūrī, ghūrī, jūrī, śūrī, cūrī. All are ī-marked, so no iṭ in the niṣṭhā: dīptaḥ, pūrṇaḥ (TB).
• 1159 tapa aiśvarye vā: in the sense of sovereignty it optionally takes śyan and taṅ, giving tapyate ("is master"); otherwise tapati.
• "Some" instead read the vā as the head of the next root, vāvṛtu. The variant reading pata (with the t and the p transposed) is supported by the ṛc "patyamānaḥ".
• 1160 vṛtu varaṇe: vṛtyate. On the other view it is vāvṛtyate. Bhaṭṭi [has] "tato vāvṛtyamānā sā rāmaśālāṃ nyavikṣata" (audit: no "hence"). vṛtudhātoḥ is the print's reading.
• 1161–1163: kliśa (kliśitā), kāśṛ, vāśṛ (vavāśe).
• "Five svaritet":
• 1164 mṛṣa: mṛṣyati / mṛṣyate.
• 1165 īśucir: aśucat (aṅ, being ir-marked) / aśocīt, and ātmanepada aśociṣṭa.
• 1166 ṇaha: nahyati / nahyate (print), nanāha. nanaddha / nehitha: iṭ in thal is optional by Bhāradvāja's restriction (SK 2296). Without iṭ, h → dh by naho dhaḥ (SK 440). With iṭ, e and loss of the abhyāsa by thali ca seṭi (SK 2261). Also naddhā, anātsīt.
• 1167 rañja: rajyati, with the n lost by aniditām… (SK 415).
• 1168 śapa: śapyate (print).
• "Eleven anudāttet" (anudāttetaḥ, print): 1169 pada gatau: padyate, pede, pattā, patsīṣṭa. This leads into 2513.
• 2513 ciṇ te padaḥ 3.1.60 — after pad, ciṇ replaces cli when ta follows. cleḥ comes from cleḥ sic (3.1.44, SK 2222). taśabde (print) means the ātmanepada prathama singular (BM, TB; counter-example apatthāḥ).
• pra-ni-a-pad + ciṇ + ta: vṛddhi of the penultimate, then ta lost by ciṇo luk (6.4.104, SK 2329), and ṇ by ner gada… (SK 2285): praṇyapādi (print).
• apatsātām, apatsata (print), with sic.
• 1170 khida: akhitta. 1171 vida sattāyām: vettā (aniṭ).
• 1172 budha: abodhi / abuddha (ciṇ optional by SK 2328), bhutsīṣṭa (kit by liṅsicau), abhutsātām.
• 1173 yudha: yuddhā, ayuddha. How can the parasmaipada yudhyati be explained? Bhaṭṭoji gives two answers:
• It is kyac from the noun yudh, "desires battle".
• Or ātmanepada marked by an anudātta it is anitya. BM rejects this second answer as not in the Bhāṣya.
• The English renders both answers and endorses neither.
• 1174 anor rudha kāme: anurudhyate. As a divādi root it takes śyan in place of rudhādi śnam (TB).
• 1175 aṇa: āṇe, aṇitā. "Some say it ends in a dental" ('yam, print).
• 1176 mana: manyate, mantā.
• 1177 yuja samādhau: intransitive, since the stilling of the mind is included in the root's meaning (BM).
• 1178 sṛja visarge:
• sṛkṣīṣṭa, asṛṣṭa: the affixes are kit by liṅsicāv ātmanepadeṣu (1.2.11, SK 2300). So there is no guṇa and no am, because sṛjidṛśor jhaly am akiti (6.1.58, SK 2405) is only for an akit affix.
• sasṛjiṣe (iṭ by the krādi-restriction), sraṣṭā, srakṣyate.
• 1179 liśa: likṣīṣṭa. alikṣata takes ksa by śala igupadhād aniṭaḥ ksaḥ (3.1.45, SK 2336).
• "Parasmaipadin to the end of the gaṇa": 1180 rādho 'karmakād vṛddhāv eva (rādho, print).
• The eva is displaced (bhinnakrama): śyan comes only after intransitive rādh. "vṛddhau" gives one example of an intransitive sense; it is not a parisaṅkhyā (BM, TB).
• aparādhyati; virādhyantam. Then [druhyantam … rādhyaty odanaḥ … sidhyati], three print sentences that ours lacks (mūla).
• kṛṣṇāya rādhyati.
• rādh "injure" is transitive, so rādho hiṃsāyām (SK 2532) does not apply: rarādhatuḥ.
• 1181 vyadha: vidhyati, with saṃprasāraṇa by grahijyā… (SK 2412) before the ṅit śyan. Also vivyādha, vividhatuḥ, avyātsīt.
• 1182 puṣa is the head of the puṣādi: apuṣat, with aṅ by puṣādi… (3.1.55, SK 2343). Also pupoṣa (print).
• 1183–1185: śuṣa (aśuṣat), tuṣa, duṣa. 1186 śliṣa: śliṣyati, śiśleṣa. BM's lead-in, on the luṅ of śliṣ, leads into 2514.

F. śliṣ: ksa, and only in the sense of embracing — running example: a-śliṣ + ksa + t → a-śliṣ-sa-t → aślikṣat (ṣ → k before s, s → ṣ) "embraced"; not embracing: sam-a-śliṣ + aṅ + t → samaśliṣat

• 2514 śliṣa āliṅgane 3.1.46 — read as two yogas (yogavibhāga).
• śliṣaḥ: after śliṣ, when the root is aniṭ, ksa replaces cli. cleḥ comes from SK 2222, and aniṭaḥ … ksaḥ from śala igupadhād aniṭaḥ ksaḥ (3.1.45, SK 2336).
• aniṭaḥ qualifies the root (audit MUST-FIX). It excludes the seṭ bhvādi śliṣu dāhe: aśleṣīt (TB).
• This ksa is an apavāda of the puṣādi aṅ (SK 2343), not of ciṇ, by the maxim of the preceding exception (purastād apavāda-nyāya). SK 2336 would already give ksa, so the point of re-prescribing it is to set aside that aṅ (TB).
• āliṅgane: ksa comes only in the sense of embracing. By the force of the split (yogavibhāga) this restriction binds the ksa of SK 2336 as well (BM, TB).
• aślikṣat kanyāṃ devadattaḥ.
• samaśliṣaj jatu kāṣṭham ("lac and wood stuck together"): the sense is contact, so the puṣādi aṅ applies. This is the print's reading; BM reads samāśliṣat … āṅ (mūla, for Ajai).
• Passive, not embracing: only sic, since the puṣādi aṅ is parasmaipada-only (BM).
• Singular with ciṇ: aśleṣi.
• aślikṣātām, where sic and ksa give the same form (ksasyāci).
• Where the two differ: aślikṣata (sic, at for jha), aśliṣṭhāḥ, aśliḍhvam. Ours and the print read aśliḍhvam; BM, TB and Wikisource read aśliḍḍhvam.
• 1187 śaka vibhāṣito marṣaṇe:
• It is ubhayapadin: śakyati / śakyate, "is able".
• śekitha / śaśaktha; aśakat / aśakta.
• "Some say it is seṭ", and the ḷ-marked śaki of the aniṭ-kārikās follows them: śakitā.
• 1188 ṣvidā: siṣveda, asvidat. It is ñīt according to the Nyāsakāra and others, but not according to Haradatta. On the ñīt view: svinnaḥ, svinnam / sveditam (TB).
• 1189 krudha: kroddhā.
• 1190 kṣudha: kṣoddhā.
• How can kṣudhita (print) be explained? Mādhava derives it with itac from the kvip noun kṣudh.
• "In truth", iṭ comes by vasatikṣudhor iṭ (7.2.52, SK 3046).
• 1191 śudha: śuśodha.
• 1192 ṣidhu saṃrāddhau (saṃrāddhiḥ, print): the ū-marked reading is an error. BM grounds this on authority, TB on the anudātta list. Forms: sidhyati, asidhat.
• 1193 radha is the head of the radh group. rarandhatuḥ takes num by radhijabhor aci (7.1.61, SK 2302). This leads into 2515.

G. The radh group (radh, naś, tṛp, dṛp, druh, muh, ṣṇuh, ṣṇih): optional iṭ; num governed for radh and naś; no ṇ for a ṣ-final naś — running example: ra-radh + tha → rarandhitha / raraddha; radh + i + tā → radhitā (no num); na-naś + tha → na-naṃś + tha → nanaṃṣṭha; pra + naṃṣ + tā → pranaṃṣṭā (no ṇ)

• 2515 radhādibhyaś ca 7.2.45 — after radh and the rest, a val-initial ārdhadhātuka takes iṭ optionally.
• ārdhadhātukasyeḍ valādeḥ comes from SK 2184, and vā from svaratisūti… (7.2.44, SK 2279) (BM). This is an option where iṭ would otherwise be invariable.
• The eight roots are radh, naś, tṛp, dṛp, druh, muh, ṣṇuh and ṣṇih (TB).
• liṭ thal: rarandhitha / raraddha; vas: rarandhiva / redhva.
• TB: because the roots are designated as a gaṇa, yaṅ-luk gives only rāradhitā.
• BM's lead-in: in luṭ, before iṭ, num by radhijabhor aci would be due. This leads into 2516.
• 2516 neṭy aliṭi radheḥ 7.1.62 — radh takes no num before an iṭ that is not of liṭ. num comes from idito num dhātoḥ (7.1.58, SK 2262), and Bhaṭṭoji's gloss is liḍvarje (print) (BM).
• radhitā, raddhā; radhiṣyati, ratsyati.
• The liṭ iṭ is exempt: rarandhiva. The affix follows a conjunct, so it is not kit, and the n stays (TB).
• "aṅi num" (ours = print, without tu): before aṅ, num comes by SK 2302, and the n is lost by aniditām… (SK 415): aradhat.
• 1194 ṇaśa: naśyati, nanāśa, neśatuḥ. neśitha takes iṭ (veṭ by 2515), with e and loss of the abhyāsa by thali ca seṭi. BM's lead-in, with nanaś-tha standing and no iṭ, leads into 2517.
• 2517 masjinaśor jhali 7.1.60 — for masj and naś, num comes before a jhal. num comes from SK 2262 (BM).
• nanaś + tha → nanaṃś + tha, then ś → ṣ by vraśca… and ṣṭutva: nanaṃṣṭha. Wikisource and BM read this; ours and the print read nanaṣṭa (mūla, for Ajai).
• neśiva / neśva, neśima / neśma.
• naśitā / naṃṣṭā. TB: the ku of naśer vā is pada-final only, so it does not act here. naśiṣyati / naṅkṣyati.
• naśyet, naśyāt, anaśat (puṣādi aṅ). praṇaśyati, which leads into 2518.
• 2518 naśeḥ ṣāntasya 8.4.36 — the n of naś does not become ṇ when naś ends in ṣ.
• Bhaṭṭoji's own gloss "ṇatvaṃ na syāt" spells out ṇaḥ from raṣābhyāṃ no ṇaḥ samānapade (8.4.1, SK 235) and na from na bhābhūpū… (8.4.34, SK 2840) (BM, TB). It is not bracketed supply (audit).
• Why "ṣāntasya"? praṇaśyati.
• pranaṃṣṭā (print).
• "anta" is there so that the rule covers a root that was ṣ-final (bhūtapūrva). In pranaṅkṣyati the ṣ has become k, yet the n stays. TB: "anta" is otherwise superfluous and is a jñāpaka.
• naśiṣyati.
• 1195 tṛp prīṇane: prīṇana covers both satisfaction and satisfying another (nāgnis tṛpyati kāṣṭhānām; pitṝn atārpsīt).
• liṭ thal: tatarpitha (veṭ, as radhādi); tatraptha (am by anudāttasya cardupadhasyānyatarasyām, 6.1.59, SK 2402); tatarptha.
• luṭ: tarpitā / tarptā / traptā.
• luṅ: sic optionally, by the vārttika spṛśamṛśakṛṣa… (under SK 2407; print vārttika sign). Hence atārpsīt (vṛddhi by SK 2267), atrāpsīt, atarpīt; otherwise the puṣādi aṅ: atṛpat.
• 1196 dṛp: veṭ as radhādi. "Their being anudātta is for am": its listing among the aniṭ roots serves am, not the absence of iṭ (BM).
• 1197 druh: gh or ḍh optionally by vā druhamuhaṣṇuhaṣṇihām (8.2.33, SK 327).
• dudrogdha / dudroḍha / dudrohitha; drohitā / drogdhā / droḍhā.
• dhrokṣyati is the same form whether gh or ḍh is substituted.
• adruhat.
• 1198 muh vaicitye (vaicityam, print): mumohitha / mumogdha / mumoḍha, amuhat.
• 1199 ṣṇuh: suṣṇoha, suṣṇogdha / suṣṇoḍha, snokṣyati.
• 1200 ṣṇih: siṣṇeha. Then "vṛt": the radh group ends. "The puṣādi (puṣādayas, print), however, run to the end of the gaṇa; this is the siddhānta."
• 1201 śamu upaśame, which leads into 2519.

H. The śam group (śam, tam, dam, śram, bhram, kṣam, klam, mad): a long vowel before śyan — running example: pra-ni + śam + śyan + ti → pra-ni + śām + ya + ti → praṇiśāmyati

• 2519 śamām aṣṭānāṃ dīrghaḥ śyani 7.3.74 — the eight roots śam and the rest get a long vowel before śyan. The plural means "śam and the rest" (BM, TB, who list the eight).
• praṇiśāmyati: the ṇ comes because śāmyati is named in ner gada… (SK 2285).
• Outside śyan the vowel stays short: śematuḥ, śemitha, śamitā; aśamat (puṣādi aṅ).
• 1202 tamu: tāmyati, atamat.
• 1203 damu upaśame: here upaśama is made from ṇic-śam ("making calm"), so the root is transitive, unlike śam. dāmyati = śamayati; adamat.
• 1204 śramu: śrāmyati, aśramat.
• 1205 bhramu: śyan is optional by vā bhrāśa… (SK 2321). Where śyan comes, the long vowel follows by this rule: bhrāmyati; abhramat. "The rest as in bhvādi."
• 1206 kṣamū: kṣāmyati.
• As an ūdit root it is veṭ: cakṣamitha / cakṣantha.
• cakṣamiva / cakṣaṇva: m → n by mvoś ca (8.2.65, SK 2309), then ṇ.
• kṣamitā / kṣantā.
• "This one is not ṣit" (aṣit, print). So kṣāntiḥ comes from the divādi root, and kṣamā from the bhvādi kṣamūṣ.
• 1207 klamu: klāmyati / klāmati. The long vowel before śyan already comes by ṣṭhivuklamu… (SK 2320), so klam is read in the śam group for the sake of ghinuṇ (śamity aṣṭābhyo ghinuṇ, 3.2.141, SK 3121); aklamat.
• 1208 madī: mādyati, amadat. "The eight, śam and the rest, are finished."
• 1209 asu kṣepaṇe (print 1210): asyati, āsa, asitā. This leads into 2520.

I. as and yas: thuk before aṅ; optional śyan for yas — running example: as + aṅ + t → as-thuk + a + t → (āṭ) āsthat; yas + śyan/śap + ti → yasyati / yasati; sam + yas → saṃyasyati / saṃyasati; but pra + yas → prayasyati only

• 2520 asyates thuk 7.4.17 — as takes the augment thuk before aṅ. aṅi pare (print) is carried from ṛdṛśo 'ṅi guṇaḥ (7.4.16, SK 2406) (BM). thuk is kit, so it becomes the final of the root (BM).
• āsthat.
• as is already puṣādi and gets aṅ that way. It is named in asyativaktikhyātibhyo 'ṅ (3.1.52, SK 2438) for aṅ in the ātmanepada, where the puṣādi aṅ cannot reach.
• The taṅ itself "will be stated" by the vārttika upasargād asyatyūhyor vā (under SK 2701, sk.2701.k4; audit): paryāsthata.
• 1210 yasu prayatne, which leads into 2521.
• 2521 yaso 'nupasargāt 3.1.71 — after yas without an upasarga, śyan comes optionally. śyan comes from SK 2505 and vā from vā bhrāśa… (SK 2321).
• There is no kaumudī of its own. BM (under sk.2522.b1) calls 2521 and 2522 "two sūtras" (sūtradvayam) with one gloss.
• yasyati / yasati.
• Because "anupasargāt" is a paryudāsa, śyan is obligatory after any upasarga: prayasyati.
• TB reports a view that "yasaḥ" alone plus "samaḥ" as a niyama would suffice.
• 2522 saṃyasaś ca 3.1.72 — after yas preceded by sam, śyan is also optional: saṃyasyati / saṃyasati. Without this rule it would be obligatory, by the paryudāsa of 2521 (BM).
• 1211 jasu: jasyati. 1212–1213 tasu, dasu: atasat, adasat.
• 1214 vasu: vavāsa, vavasatuḥ, since na śasadada… (6.4.126, SK 2263) blocks e and abhyāsa-loss (SK 2260). On the view that the root is b-initial (baśādi): besatuḥ, besuḥ.
• 1215 vyuṣa vibhāge: read again, with a different sense, for the puṣādi aṅ: avyuṣat. The earlier vyuṣa dāhe (1114) has sic only: avyoṣīt (BM, TB).
• "Others say pyus; still others yus." That is the print's reading. BM's text evidently had vyus / vus.
• 1216 pluṣa: apluṣat. The earlier reading (1115) is for sic, "they say". The kaumudī's own rejoinder: the bhvādi reading already gives that (āhuḥ = asvarasa, BM).
• 1217–1222: bisa (bisyati, abisat; the print is mixed, visa / abisat / vusa), kusa, busa, musa, masī ("some read samī"), luṭha.
• 1223 uca: ucyati (print), uvoca, ūcatuḥ, mā bhavān ucat.
• 1224–1225 bhṛśu, bhraṃśu: babharśa, abhṛśat. bhraśyati loses its n by aniditām… (SK 415).
• 1226–1229: vṛśa, kṛśa, ñitṛṣā, hṛṣa (tuṣṭau, print). hṛṣa differs from the bhvādi root by śyan and aṅ.
• 1230–1231 ruṣa, riṣa: veṭ by tīṣasahalubharuṣariṣaḥ (7.2.48, SK 2340): roṣitā / roṣṭā.
• 1232–1237: ḍipa, kupa (krodhe, print), gupa, yupa, rupa, lupa. lupyati is seṭ (lopitā): the aniṭ-kārikās take only the tudādi lup, which is paired with lip.
• 1238 lubha: veṭ by SK 2340: lobhitā / lobdhā; alubhat. lobhati too, since the bhvādi is not closed off by vṛt (avṛtkṛtatvāt, print).
• 1239–1241 kṣubha, ṇabha, tubha: these are also dyutādi and kryādi (kryāditvāt, print). They are read here only for śyan.
• 1242 klidū: veṭ as ūdit: cikleditha / ciklettha, kleditā / klettā.
• 1243 ñimidā: medyati, with guṇa by mider guṇaḥ (7.3.82, SK 2346); amidat.
• The puṣādi reading keeps out amedīt.
• Bhaṭṭoji concedes that a reading among the ātmanepada roots outside the dyutādi "would be proper" (iṣṭāpatti, BM).
• 1244 ñikṣvidā.
• 1245 ṛdhu: ānardha, ārdhat.
• 1246 gṛdhu: agṛdhat. Then "vṛt".
• Close: "the puṣādi and the divādi are finished". Some make the vṛt mark only the end of the puṣādi, with divādi an ākṛtigaṇa (kṣīyate, mṛgyati). Then "iti divādayaḥ" (Here end the divādi roots), which is in ours and the print. p48 opens with svādi, ṣuñ (SK 2523).