तद्धिताधिकारे चातुरर्थिकप्रकरणम्

Covers SK 1202–1311 (110 sūtras): all present. The spine is Aṣṭādhyāyī 4.2.1–91. The SK splices in the rules from other pādas that its derivations need, at the point where it reads them: 7.3.51 (ka for ṭha), 7.4.27 (rīṅ), 7.3.21–23 (vṛddhi in deity-dvandvas), 6.4.164–165 (in kept before aṇ), 7.1.1 (yu/vu → ana/aka), 1.4.16 (pada before a sit affix), 6.3.71 (mum before pāta), 2.4.58 (luk of yuvan aṇ/iñ), 1.2.51–57 (yuktavadbhāva and the aśiṣya set) and 6.4.153 (luk of cha).

taddhitādhikāre cāturarthikaprakaraṇam — outline (p28, SK 1202–1311, all 110 sūtras)

This prakaraṇa has two parts. The commentators treat SK 1202–1278 (A 4.2.1–66) as the raktādyarthaka section: a chain of single senses, each set up by a heading sūtra that names a case and a meaning. The affix is the prāgdīvyatīya aṇ etc. (prāg dīvyato 'ṇ, 4.1.83), and special affixes override it for particular bases. The senses run in order:

• tena raktam ("dyed with that"), nakṣatreṇa yuktaḥ kālaḥ ("a time joined with a nakṣatra"), dṛṣṭaṃ sāma ("a sāman seen by"), parivṛto rathaḥ ("a chariot covered with").
• tatroddhṛtam ("taken out into"), saṃskṛtaṃ bhakṣāḥ ("food prepared in"), sāsmin paurṇamāsī ("the month whose full-moon is that").
• sāsya devatā ("whose deity is that"), tasya samūhaḥ ("a collection of that"), viṣayo deśe ("the region that is the domain of").
• the pragātha, battle, game and action senses, tad adhīte tad veda ("he studies / knows that"), and the luk rules for that sense.

BM marks two internal closes: iti raktādhikāraḥ after 1203, and iti yuktādyarthakāḥ after 1225. From SK 1279 (tad asminn asti), four place-name senses are set up (1279–1282): "that is in it", "founded by", "dwelling of" and "not far from". The ca of 1282 makes every later affix cāturarthika, that is, valid in all four senses (1283–1311). Inside this second part, the lup of the affix for janapadas (1293) brings in the yuktavadbhāva rules (1294, 1300), where gender and number follow the base. It also brings in Pāṇini's rejection of the older teachers' definitions (1295–1299). The prakaraṇa ends with matup and its relatives for rivers and places (1304–1311).

Only the kaumudī blocks are translated (SK-TRANSLATE §25). The readings cited as BM (Bālamanoramā) and TB (Tattvabodhinī) are the ones recorded in the unit notes. Where the p28 audit corrects a reading, the corrected reading is followed. This applies to 1205, 1206, 1215, 1216, 1217, 1226, 1227, 1241, 1242, 1244, 1247, 1251, 1262, 1265, 1274, 1276, 1277, 1279, 1289, 1292, 1295, 1306 and 1311. Ordering: SK order throughout; nothing is reordered. The SK itself reads 1.2.52 (1300) after 1.2.57, and this outline keeps that order.

Affix shapes used below (7.1.2, 7.3.50): ṭh → ik, ḍh → ey, ch → īy, gh → iy, kh → īn, ph → āyan. vu → aka and yu → ana (7.1.1, SK 1247). A ñ-, ṇ- or k-marker brings ādivṛddhi (7.2.117–118). A stem-final a/ā/i/ī is deleted before the affix by yasyeti ca (6.4.148, SK 311). A ḍ-marker deletes the ṭi.

A. "Dyed with that" (tena raktaṃ rāgāt) — running example: kaṣāyeṇa raktaṃ vastram → kāṣāyam

• 1202 tena raktaṃ rāgāt 4.2.1 — after a third-case word that names a dye (rāga, "that by which a thing is dyed"), aṇ etc. in the sense "dyed with that". kaṣāya (a mineral dye) + aṇ → ādivṛddhi turns the first a into ā, and the final a is deleted: kāṣāyam. The source's māñjikam is probably an error for māñjiṣṭham (BM). Counter-example: devadattena raktaṃ vastram takes no affix, because Devadatta is not a dye.
• 1203 lākṣārocanāṭ ṭhak 4.2.2 — ṭhak (ṭh → ik) replaces aṇ after lākṣā and rocanā: lākṣikaḥ, raucanikaḥ. Four vārttikas extend it:
• śakalakardamābhyām upasaṃkhyānam gives śākalikaḥ, kārdamikaḥ. The Vṛttikāra also allows aṇ here (śākalaḥ, kārdamaḥ), which BM and TB say is not in the Bhāṣya.
• nīlyā an gives nīlam.
• pītāt kan gives pītakam.
• haridrāmahārajanābhyām añ gives hāridram, māhārajanam (añ differs from aṇ only in accent).

B. "A time joined with a nakṣatra", and its lup — running example: puṣyeṇa yuktam ahaḥ → pauṣam ahaḥ, but adya puṣyaḥ (with lup)

• 1204 nakṣatreṇa yuktaḥ kālaḥ 4.2.3 — after a third-case nakṣatra word, aṇ etc. in the sense "a time joined with it", meaning joined with the moon that stands in that nakṣatra (BM, TB). puṣya + aṇ → vṛddhi; the y is deleted by the vārttika tiṣyapuṣyayor nakṣatrāṇi yalopaḥ (under 6.4.149, per BM): pauṣam ahaḥ. The feminine takes ṅīp: pauṣī rātriḥ.
• 1205 lub aviśeṣe 4.2.4 — lup of that affix when no internal division of the sixty-daṇḍa day (such as day or night) is understood: adya puṣyaḥ. Objection: then why is the full-moon day called pauṣī and not puṣya? Response: the forms phālgunī, kārttikī and caitrī in 1225 are a jñāpaka that this lup does not reach the full-moon day. For śravaṇa, however, lup applies by this very nipātana śravaṇā (in 1225), and so does the absence of yuktavadbhāva (BM). Since nipātanas are also non-blocking (abādhakāny api nipātanāni), śrāvaṇī is correct as well.
• 1206 saṃjñāyāṃ śravaṇāśvatthābhyām 4.2.5 — after śravaṇa and aśvattha, lup when a name is meant. The rule is for the case where a particular division of time is understood, which 1205 does not cover. The words "lup, from 1205" are the translator's inference; BM says only "lup syāt". Examples: śravaṇā (a night), aśvatthaḥ (a muhūrta). With no name meant there is no lup: śrāvaṇī, āśvatthī.
• 1207 dvandvāc chaḥ 4.2.6 — after a dvandva of nakṣatra names, cha (→ īy), whether or not a division of time is meant. tiṣyapunarvasu + īya; the final u takes guṇa (or guṇaḥ 6.4.146): tiṣyapunarvasavīyam ahaḥ. Also rādhānurādhīyā rātriḥ. TB: lup (1205) does not remove this cha, because as an apavāda standing between the rules it applies only to the earlier aṇ.

C. "Seen" (of a sāman), "covered" (of a chariot), and the nipātana kaumāra — running example: vastraiḥ parivṛto rathaḥ → vāstro rathaḥ

• 1208 dṛṣṭaṃ sāma 4.2.7 — tena is still in force: aṇ etc. in the sense "seen by", when what is seen is a sāman. The source gives the phrase vasiṣṭhena dṛṣṭaṃ sāma without a derived form. Vārttika aṇ ḍid vā: here aṇ is optionally ḍit. uśanas + aṇ as ḍit loses the ṭi as, giving auśanam; without the ḍ, auśanasam.
• 1209 kaler ḍhak 4.2.8 — ḍhak (→ ey) after kali; the i is deleted: kāleyaṃ sāma. BM calls this a vārttika; the spine numbers it as a sūtra.
• 1210 vāmadevāḍ ḍyaḍḍyau 4.2.9 — ḍyat and ḍya after vāmadeva. The ḍ deletes the ṭi a: vāmadevyam. A Bhāṣya verse raises the question why ḍit is needed when yasyeti ca alone would give this form. The answer: as ḍit, the affixes are not covered by the nañ-accent rule yayatoś cātadarthe (6.2.156), so avāmadevyam does not become end-acute. This works through the paribhāṣās niranubandhakagrahaṇe na sānubandhakasya and tadanubandhaka…, which this rule itself makes known (BM; TB agrees).
• 1211 parivṛto rathaḥ 4.2.10 — aṇ etc. in the sense "covered with that", when a chariot is covered. vastra + aṇ → vāstro rathaḥ. parivṛta means enveloped on all sides, so chātraiḥ parivṛto rathaḥ ("surrounded by pupils") takes no affix. For a body, vastreṇa parivṛtaḥ kāyaḥ, there is no affix either.
• 1212 pāṇḍukambalād iniḥ 4.2.11 — ini after pāṇḍukambala (a coloured royal coverlet): pāṇḍukambalī. The matvarthīya ini (5.2.115) would already give this form; the rule is there to exclude aṇ.
• 1213 dvaipavaiyāghrād añ 4.2.12 — añ (differing from aṇ in accent) after dvaipa ("a panther's hide", itself from dvīpin + añ) and vaiyāghra: dvaipo rathaḥ, vaiyāghraḥ. In vaiyāghra, the aic augment comes by na yvābhyām (7.3.3; BM).
• 1214 kaumārāpūrvavacane 4.2.13 — the nipātana kaumāra, used when the woman has had no previous husband (apūrvatva); tena no longer applies. kumārī + aṇ gives kaumāraḥ patiḥ (the man who married her) or, in its own sense with ṅīp, kaumārī bhāryā.

D. Locative senses: "taken out into", "lying on (as a vow)", "prepared in" — running example: udaśviti saṃskṛtaḥ → audaśvitkaḥ / audaśvitaḥ

• 1215 tatroddhṛtam amatrebhyaḥ 4.2.14 — after a locative word naming a vessel, aṇ etc. in the sense "taken out [and set down] in it". Here uddharati means setting down after taking out, which is why the case is locative. śarāva + aṇ → śārāva odanaḥ. tatra is carried as far as 4.2.21. TB: the Vṛttikāra takes uddhṛta as "left over after eating" (bhuktocchiṣṭa).
• 1216 sthaṇḍilāc chayitari vrate 4.2.15 — after sthaṇḍila in the locative, aṇ etc. as prescribed, in the sense "one who lies there", when the lying is under a vow. TB: vrata is a restraint enjoined by śāstra, and the vow is conveyed by the word as a whole (samudāyena). Example: sthāṇḍilo bhikṣuḥ.
• 1217 saṃskṛtaṃ bhakṣāḥ 4.2.16 — after a locative word, aṇ in the sense "prepared in it", when the thing is food: bhrāṣṭra + aṇ → bhrāṣṭrā yavāḥ. In aṣṭasu kapāleṣu saṃskṛtaḥ, the aṇ is the cause of a taddhitārtha dvigu, so it is elided by dvigor lug anapatye (SK 1080; TB) and no vṛddhi appears: aṣṭakapālaḥ puroḍāśaḥ (BM and TB read the pratīka as aṣṭākapāla).
• 1218 śūlokhād yat 4.2.17 — yat, overriding aṇ, after śūla ("spit") and ukhā ("pot"); the final a/ā is deleted: śūlyaṃ māṃsam, ukhyam.
• 1219 dadhnaṣ ṭhak 4.2.18 — ṭhak after dadhi (curd is only the locus); the i is deleted: dādhikam.
• 1220 udaśvito 'nyatarasyām 4.2.19 — after udaśvit ("buttermilk"), ṭhak optionally, otherwise aṇ. On the running form: with ṭhak, udaśvit + ṭha waits for 1221; with aṇ, audaśvitaḥ.
• 1221 isusuktāntāt kaḥ 7.3.51 — the ṭha after a stem ending in is, us, uk or t is replaced by ka (not ika). On the running form: the stem ends in t, so ṭha → ka: audaśvitkaḥ. Only the is and us taught by name (pratipadokta, the uṇādi affixes) count. So there is no ka in āśiṣikaḥ (āśis from śās) or auṣikaḥ. Vārttika doṣa upasaṃkhyānam: dauṣkaḥ.
• 1222 kṣīrāḍ ḍhañ 4.2.20 — ḍhañ (→ ey) after kṣīra. Only saṃskṛtam carries over from 1217, not bhakṣāḥ, so the rule also covers gruel: kṣaireyī yavāgūḥ.

E. The month named after its full-moon (sāsmin paurṇamāsīti) — running example: pauṣī paurṇamāsy asmin → pauṣo māsaḥ

• 1223 sāsmin paurṇamāsīti 4.2.21 — after a first-case word naming a full-moon day, aṇ etc. in the sense "in which that is [the full-moon]". The iti restricts this to names (saṃjñā). pauṣī (from 1204) + aṇ; its ī is deleted by yasyeti ca: pauṣo māsaḥ. BM's counter-example, with no affix, is the fifteen-night period.
• 1224 āgrahāyaṇyaśvatthāṭ ṭhak 4.2.22 — ṭhak after āgrahāyaṇī and aśvattha. āgrahāyaṇī is a prajñādi aṇ + ṅīp, with ṇ by pūrvapadāt saṃjñāyām: āgrahāyaṇikaḥ. Also āśvatthikaḥ; the sūtra's own form aśvattha shows, by nipātana, that lup applies here even for a full-moon.
• 1225 vibhāṣā phālgunīśravaṇākārttikīcaitrībhyaḥ 4.2.23 — ṭhak optionally, otherwise aṇ: phālgunikaḥ / phālgunaḥ, śrāvaṇikaḥ / śrāvaṇaḥ, kārtikikaḥ / kārtikaḥ, caitrikaḥ / caitraḥ. This is the jñāpaka that 1205 relies on. BM closes the section here: iti yuktādyarthakāḥ.

F. "Whose deity is that" (sāsya devatā) — running example: indro devatāsya → aindraṃ haviḥ; with a stem rule, pitaro devatā asya → pitryam

• 1226 sāsya devatā 4.2.24 — after a first-case word, aṇ etc. in the sense "that is its deity". The deity is either the particular recipient intended when a substance is given up, or the one praised by a mantra. indra + aṇ gives aindraṃ haviḥ and aindro mantraḥ; also pāśupatam, bārhaspatyam. āgneyo vai brāhmaṇo devatayā takes its affix in the śaiṣika sense: ḍhak by the vārttika sarvatrāgni[kalibhyām] (BM on 1236). TB: the doubt over this usage should not be raised at all.
• 1227 kasyet 4.2.25 — ka ("Prajāpati") takes the substitute i together with the affix: ki + a. Ādivṛddhi prevails over the elision by yasyeti ca (SK 311), because it is the later rule (paratvāt). The i of ki takes ādivṛddhi → kai, and ai → āy before the a: kāyaṃ haviḥ. The same reasoning explains śrāyam, where no i-rule applies.
• 1228 śukrād ghan 4.2.26 — ghan (→ iya) after śukra: śukriyam.
• 1229 aponaptrapānnaptṛbhyāṃ ghaḥ 4.2.27 — gha after aponaptṛ and apānnaptṛ: aponaptriyam, apānnaptriyam. The -naptṛ forms are a nipātana that holds only together with the affix, so the ritual direction keeps the base forms: aponapāte 'nubrūhi.
• 1230 cha ca 4.2.28 — cha after the same two: aponaptrīyam, apānnaptrīyam. The rule is split from 1229 so that the two affixes do not pair with the two bases in respective order (yathāsaṃkhya). Vārttika śatarudrād ghaś ca: "the hundred (= countless) Rudras" gives śatarudriyam / śatarudrīyam. Because gha and cha are taught explicitly, they escape the dvigu-luk of 1080.
• 1231 mahendrād ghāṇau ca 4.2.29 — gha, aṇ, and (by ca) cha after mahendra: mahendriyam, māhendram, mahendrīyam.
• 1232 somāṭ ṭyaṇ 4.2.30 — ṭyaṇ after soma: saumyam. Because ṭ is an it-marker, the feminine takes ṅīp, and the y is deleted by halas taddhitasya (BM): saumī ṛk.
• 1233 vāyvṛtupitruṣaso yat 4.2.31 — yat after vāyu, ṛtu, pitṛ and uṣas. The final u takes guṇa, then o → av (vānto yi pratyaye): vāyavyam, ṛtavyam.
• 1234 rīṅ ṛtaḥ 7.4.27 — a ṛ-final aṅga takes rīṅ before a ya that is neither kṛt nor sārvadhātuka, and before cvi. On the running form: pitṛ + ya → pitrī + ya; the ī is deleted by yasyeti ca: pitryam. The long ī exists for the kyac form pitrīyati. The other form of 1233 is uṣasyam (bha, so no ru).
• 1235 dyāvāpṛthivīśunāsīramarutvadagnīṣomavāstoṣpatigṛhamedhāc cha ca 4.2.32 — cha, and (by ca) yat, after these deity-compounds: dyāvāpṛthivīyam / dyāvāpṛthivyam, śunāsīrīyam / śunāsīryam (the source prints śūnā-).
• 1236 agner ḍhak 4.2.33 — ḍhak after agni: āgneyam. By sarvatrāgnikalibhyām it holds in every prāgdīvyatīya sense (BM).
• 1237 kālebhyo bhavavat 4.2.34 — atideśa: after words for times, the affixes are the same as in the bhava sense, and only the affix prescribed for that base. The ṭhañ of kālāṭ ṭhañ gives māsikam; the eṇya of prāvṛṣa eṇyaḥ gives prāvṛṣeṇyam.
• 1238 mahārājaproṣṭhapadāṭ ṭhañ 4.2.35 — ṭhañ after mahārāja (Vaiśravaṇa) and proṣṭhapada: māhārājikam, prauṣṭhapadikam.

G. Vṛddhi in deity-dvandvas, two return-to-topic vārttikas, and the kinship nipātanas — running example: agnāmarutau devatā asya → āgnimārutam

• 1239 devatādvandve ca 7.3.21 — this is taken up incidentally (BM). In a dvandva of deity names, both members take ādi-vṛddhi before a ñit, ṇit or kit affix. agniś ca marut ca gives agnāmarutau (ānaṅ, 6.3.26). Before the vṛddhi-affix, id vṛddhau (6.3.28) gives i in place of the ānaṅ ā: agni-marut + aṇ → the a of agni → ā, and the a of marut → ā: āgnimārutam.
• 1240 nendrasya parasya 7.3.22 — no vṛddhi for indra as the final member: saumendraḥ [caruḥ]. As the first member it does take vṛddhi: aindrāgnaḥ.
• 1241 dīrghāc ca varuṇasya 7.3.23 — no vṛddhi for varuṇa after a long vowel. indrāvaruṇa + aṇ: varuṇa follows the long ā and keeps its u: aindrāvaruṇam. In āgnivāruṇī, id vṛddhau has made the vowel before varuṇa a short i, so both members take vṛddhi. Then two vārttikas return to the main topic (atha prakṛtam):
• tad asmin vartata iti navayajñādibhya upasaṃkhyānam, with ṭhañ as BM records: nāvayajñikaḥ kālaḥ, pākayajñikaḥ.
• pūrṇamāsād aṇ vaktavyaḥ: paurṇamāsī (the tithi). TB reports Haradatta and others: taking paurṇamāsī as a bahuvrīhi with the prajñādi aṇ makes this vārttika unnecessary.
• 1242 pitṛvyamātulamātāmahapitāmahāḥ 4.2.36 — the nipātanas pitṛvya, mātula, mātāmaha and pitāmaha, filled out by vārttikas:
• pitur bhrātari vyat: pitṛvyaḥ.
• mātur ḍulac (ṭi deleted): mātulaḥ.
• mātṛpitṛbhyāṃ pitari ḍāmahac: mātāmahaḥ, pitāmahaḥ. TB notes that the Bhāṣya instead has ānaṅ + mahac.
• mātari ṣic ca (ṣ-marker, so ṅīṣ): mātāmahī, pitāmahī.
• aver dugdhe soḍhadūsamarīsaco vaktavyaḥ: because they are read with s, there is no ṣ: avisoḍham, avidūsam, avimarīsam. TB records others who limit this argument to avisoḍham.
• tilān niṣphalāt piñjapejau ("barren sesame"): tilapiñjaḥ, tilapejaḥ.
• piñjaś chandasi ḍic ca: tilpiñjaḥ.

H. "A collection of that" (tasya samūhaḥ), I: aṇ, vuñ, yañ, ṭhañ, yan, tal, ṇas and their stem rules — running example: kākānāṃ samūhaḥ → kākam; with a stem rule, garbhiṇīnāṃ samūhaḥ → gārbhiṇam

• 1243 tasya samūhaḥ 4.2.37 — after a sixth-case word, aṇ etc. in the sense "a collection of that". The sense runs as far as 4.2.51, and the derivatives are neuter by usage. kāka + aṇ → kākam; baka → bākam. These examples were chosen because none of the apavāda rules below applies to them (TB).
• 1244 bhikṣādibhyo 'ṇ 4.2.38 — aṇ after the bhikṣā list:
• For bhikṣā it overrides the ṭhak of acitta words (1256): bhaikṣam.
• For garbhiṇī it overrides the añ of anudātta-initial words (1253). bhasyāḍhe removes the ṅīp, leaving garbhin + a. The ṭi-elision by nas taddhite (SK 679) then becomes due, and 1245 blocks it.
• 1245 inaṇy anapatye 6.4.164 — a stem ending in in keeps its original form before a non-apatya aṇ, so the ṭi is not deleted. On the running form: garbhin + aṇ keeps in: gārbhiṇam. Likewise yuvati → puṃvadbhāva → yuvan + aṇ: yauvanam. From yuvat (śatṛ, anudātta-initial) with añ, where this rule does not apply: yauvatam.
• 1246 gotrokṣoṣṭrorabhrarājarājanyarājaputravatsamanuṣyājād vuñ 4.2.39 — vuñ after a gotra word and after ukṣan, uṣṭra, urabhra, rājan, rājanya, rājaputra, vatsa, manuṣya and aja. Here gotra is the ordinary (laukika) gotra, meaning any offspring.
• 1247 yuvor anākau 7.1.1 — nasalised yu → ana and vu → aka. On vuñ: glucukāyanī + vuñ → vu → aka, with ādivṛddhi and the ī deleted: glaucukāyanakam; ukṣan → aukṣakam. In rājanyakam and mānuṣyakam, the kaumudī says the y-elision of āpatyasya ca (SK 1082) would be due, and the vārttika prakṛtyā 'ke rājanyamanuṣyayuvānaḥ keeps the y: rājanyakam, mānuṣyakam. TB dissents (printed under 1244.t1): rājanya and manuṣya are rūḍhi words, so the y-elision never reaches them; naming them in the vārttika is superfluous, and naming them in 4.2.39 is meaningful. The kaumudī's "yalope prāpte" is translated as it stands. Vārttika vṛddhāc ceti vaktavyam (vṛddha here is the word itself): vārdhakam.
• 1248 kedārād yañ ca 4.2.40 — yañ, and (by ca) vuñ, after kedāra: kaidāryam, kaidārakam. Vārttika: yañ only (no vuñ) after gaṇikā: gāṇikyam.
• 1249 ṭhañ kavacinaś ca 4.2.41 — ṭhañ after kavacin, and (by ca) after kedāra. kavacin + ik: the ṭi in is deleted: kāvacikam, kaidārikam.
• 1250 brāhmaṇamāṇavavāḍavād yan 4.2.42 — yan after brāhmaṇa, māṇava and vāḍava: brāhmaṇyam, māṇavyam, vāḍavyam. Vārttika pṛṣṭhād upasaṃkhyānam: pṛṣṭhyam.
• 1251 grāmajanabandhubhyas tal 4.2.43 — tal (→ -tā, feminine) after grāma, jana and bandhu: grāmatā, janatā, bandhutā. Three vārttikas follow:
• gajasahāyābhyāṃ ca: gajatā, sahāyatā.
• ahnaḥ khaḥ kratau: kha → īn after ahan, with the ṭi deleted: ahīnaḥ, a kratu whose soma-pressings (sutyā) take a series of days. Where no kratu is meant, the khaṇḍikādi añ applies (1254). The ṭi stays because of the restriction ahnaṣ ṭakhor eva (SK 789): āhnaḥ.
• parśvā ṇas vaktavyaḥ: feeds 1252.
• 1252 siti ca 1.4.16 — before an affix with s as it-marker, the preceding stem is pada, so it is not bha, and the guṇa of or guṇaḥ (6.4.146) is blocked. parśu + ṇas: the u does not take guṇa, and yaṇ applies: pārśvam.

I. "A collection of that", II: añ, the dharmavat atideśa, ṭhak, and the ya / ini / tra / kaṭyac collectives — running example: saktūnāṃ samūhaḥ → sāktukam (ṭhak with ka by 1221)

• 1253 anudāttāder añ 4.2.44 — añ after anudātta-initial words: kapota and mayūra are madhyodātta, giving kāpotam, māyūram. The examples are animate because for an inanimate base the later ṭhak (1256) prevails (TB). TB also rejects śakunīnāṃ ca laghu pūrvam as a ground for treating these words as ādyudātta.
• 1254 khaṇḍikādibhyaś ca 4.2.45 — añ after khaṇḍikā etc. (so that the accent falls on the initial; TB adds that it blocks ṭhak): khāṇḍikam.
• 1255 caraṇebhyo dharmavat 4.2.46 — atideśa: after names of Vedic schools, the collective affixes are the same as for the sense dharma. The vuñ of gotracaraṇād vuñ gives kāṭhakam; the ñya of chandogaukthika… gives chāndogyam.
• 1256 acittahastidhenoṣ ṭhak 4.2.47 — ṭhak after inanimate words and after hastin and dhenu. On the running form: saktu ends in u, so ṭha → ka (1221): sāktukam; likewise dhainukam. For hastin, ṭh → ik and the ṭi is deleted: hāstikam.
• 1257 keśāśvābhyāṃ yañchāv anyatarasyām 4.2.48 — optionally yañ after keśa (otherwise ṭhak) and cha after aśva (otherwise aṇ): kaiśyam / kaiśikam, aśvīyam / āśvam.
• 1258 pāśādibhyo yaḥ 4.2.49 — ya (feminine by usage) after pāśa etc.: pāśyā, tṛṇyā, dhūmyā, vanyā, vātyā.
• 1259 khalagorathāt 4.2.50 — ya after khala, go and ratha: khalyā, gavyā (o → av), rathyā. They are listed separately so that they alone carry into 1260.
• 1260 initrakaṭyacaś ca 4.2.51 — ini, tra and kaṭyac, in order, after khala, go and ratha: khalinī, gotrā, rathakaṭyā. Vārttika khalādibhya inir vaktavyaḥ (an ākṛtigaṇa): ḍākinī, kuṭumbinī.

J. Remaining single senses: domain-region, pragātha, battle, game, action — running example: śibīnāṃ viṣayo deśaḥ → śaibaḥ

• 1261 viṣayo deśe 4.2.52 — samūha stops here and tasya continues. After a sixth-case word, aṇ etc. in the sense "the thoroughly frequented domain", when that is a region: śibi + aṇ → ādivṛddhi turns the i into ai, and the final i is deleted: śaibaḥ. devadattasya viṣayo 'nuvākaḥ takes no affix, because an anuvāka is not a region.
• 1262 rājanyādibhyo vuñ 4.2.53 — vuñ (→ aka) after rājanya etc.: rājanyakaḥ, "the region that is the domain of the Rājanyas". BM's text transmits viśeṣo where viṣayo is meant.
• 1263 bhaurikyādyaiṣukāryādibhyo vidhalbhaktalau 4.2.54 — vidhal after the bhaurikyādi list and bhaktal after the aiṣukāryādi list, in the domain-region sense: bhaurikividham, bhaulikividham; aiṣukāribhaktam, sārasāyanabhaktam (neuter by usage).
• 1264 so 'syādir iti cchandasaḥ pragātheṣu 4.2.55 — after a metre name, aṇ in the sense "that is the beginning of this [pragātha]": paṅkti + aṇ → pāṅktaḥ pragāthaḥ. Vārttika svārtha upasaṃkhyānam (aṇ in the sense of the base itself): traiṣṭubham.
• 1265 saṃgrāme prayojanayoddhṛbhyaḥ 4.2.56 — so 'sya is carried forward: after first-case words for a purpose or for warriors, aṇ etc. in the sense "that is [the purpose / are the warriors] of this", when "this" is a battle: saubhadraḥ (Subhadrā the purpose), bhārataḥ (the Bharatas the warriors).
• 1266 tad asyāṃ praharaṇam iti krīḍāyāṃ ṇaḥ 4.2.57 — ṇa in the sense "that is the weapon in this [game]". ṇa is taught rather than aṇ, which would bring ṅīp, so the feminine takes ṭāp: dāṇḍā, mauṣṭā.
• 1267 ghañaḥ sā 'syāṃ kriyeti ñaḥ 4.2.58 — after a ghañ-final word naming an action, ña in the sense "that action is in this [feminine thing]". Because ghañ is a kṛt, a compound with a gati or kāraka before the ghañ-word is also covered. This feeds 1268.
• 1268 śyenatilasya pāte ñe 6.3.71 — the augment mum is added to śyena and tila before pāta + ña. śyenapāta + ña → mum after śyena, with vṛddhi on the compound: śyainaṃpātā mṛgayā; tailaṃpātā svadhā. For other bases there is no mum: dāṇḍapātā tithiḥ.

K. "He studies that / he knows that" (tad adhīte tad veda) — running example: vyākaraṇam adhīte veda vā → vaiyākaraṇaḥ

• 1269 tad adhīte tad veda 4.2.59 — after a second-case word, aṇ etc. in the two senses "he studies that" and "he knows that" (tad is repeated to keep them apart). vyākaraṇa + aṇ; the aic augment comes by na yvābhyām (7.3.3), so the prefix is vai-: vaiyākaraṇaḥ.
• 1270 kratūkthādisūtrāntāṭ ṭhak 4.2.60 — ṭhak after particular kratu names, the ukthādi list, and words ending in sūtra:
• kratu names: āgniṣṭomikaḥ, vājapeyikaḥ (the knower sense from the primary meaning; the student sense when the word names the expounding text).
• uktha: aukthikaḥ, but no ṭhak or aṇ when uktha means the sāman itself.
• the rest of ukthādi: naiyāyikaḥ, vārttikaḥ, laukāyatikaḥ.

A string of vārttikas follows:

• sūtra-final words not beginning with kalpa: sāṃgrahasūtrikaḥ, but kālpasūtraḥ.
• also after vidyā, lakṣaṇa and kalpa: vāyasavidyikaḥ, āśvalakṣaṇikaḥ, pārāśarakalpikaḥ.
• not after vidyā with aṅga, kṣatra, dharma or tri before it: āṅgavidyaḥ, traividyaḥ (from trividhā vidyā; the BM/TB trividyā discussion is on this example).
• ākhyāna and the like: yāvakrītikaḥ; vāsavadattikaḥ (the cha of 1467 undergoes lup by lub ākhyāyikābhyo bahulam, then ṭhak); aitihāsikaḥ, paurāṇikaḥ.
• luk after sarva- and sa-: sarvavedaḥ, savārttikaḥ; with the dvigu-luk, dvitantraḥ.
• ikan after pada-final words and ṣikan after śata/ṣaṣṭi-pathin: pūrvapadikaḥ, śatapathikaḥ / -ī.
• 1271 kramādibhyo vun 4.2.61 — vun (→ aka) after krama, pada, śikṣā and mīmāṃsā: kramakaḥ.
• 1272 anubrāhmaṇād iniḥ 4.2.62 — ini after anubrāhmaṇa, to block aṇ (matvarthīya ini would give the form anyway): anubrāhmaṇī.
• 1273 vasantādibhyaṣ ṭhak 4.2.63 — ṭhak after vasanta etc.: vāsantikaḥ. In ātharvaṇikaḥ, the ṭi is not deleted, because the form is a nipātana in dāṇḍināyana… (SK 1145).

L. Luk of the student-affix, and the derivation of pāṇini / pāṇinīya — running example: paṇ → paṇa → paṇī → pāṇinaḥ → pāṇiniḥ → pāṇinīyam → pāṇinīyaḥ (commentators close the raktādyarthaka section at 1278)

• 1274 proktāl luk 4.2.64 — the student/knower affix that follows an affix in the sense "proclaimed" undergoes luk. The kaumudī then begins the pāṇini chain: paṇana → paṇaḥ (ka by the vārttika ghañarthe kavidhānam, on 3.3.58); + possessive ini → paṇī; + gotra aṇ → pāṇinaḥ. Because this aṇ is apatya, 1245 does not apply, and the ṭi-elision would be due.
• 1275 gāthividathikeśigaṇipaṇinaś ca 6.4.165 — gāthin, vidathin, keśin, gaṇin and paṇin keep their form before aṇ, even an apatya aṇ. On the running form, paṇin + aṇ keeps in: pāṇinaḥ. Then the yuvan iñ (1095) comes after the gotra form: pāṇiniḥ.
• 1276 ṇyakṣatriyārṣañito yūni lug aṇiñoḥ 2.4.58 — a yuvan aṇ or iñ is dropped after a gotra-affix-final word in one of four classes. Each pair gives father and son the same form:
• ending in ṇya: kauravyaḥ
• ending in a kṣatriya gotra-affix: śvāphalkaḥ
• ending in a ṛṣi-affix (ārṣa): vāsiṣṭhaḥ
• ending in a ñit affix: taikāyaniḥ

Counter-examples:

• śivādi aṇ is none of these: kauhaṅaḥ → kauhaṅiḥ.
• not a yuvan affix: vāmarathāḥ.

On the running form, the iñ of pāṇini is not dropped. ārṣa takes in only the ṛṣi-aṇ taught by name (pratipadokta), and the aṇ of pāṇina is the general aṇ. BM holds this reason conflicts with the Bhāṣya on ātreya and should be set aside; TB reports that objection as others' view (ity āhuḥ) and itself grounds pāṇini on the Bhāṣya usage dākṣīputrasya pāṇineḥ. Next, pāṇini + cha (vṛddhāc chaḥ, SK 1337) → pāṇinīyam. The aṇ of iñaś ca (SK 1333) does not come, because that rule needs a gotra iñ. Then the student aṇ undergoes luk by 1274. This affects the accent and the feminine: pāṇinīyaḥ, pāṇinīyā.

• 1277 sūtrāc ca kopadhāt 4.2.65 — luk of the student affix after a word that names a sūtra and has k as its penultimate, even where the sense is not prokta. aṣṭaka is kan in the parimāṇa sense (Pāṇini's eight chapters); its students are aṣṭakāḥ. BM and TB gloss pratīkas that the transmitted kaumudī lacks: saṅkhyāprakṛtikāt (a restriction to numeral-based words), with the counter-examples māhāvārttikaḥ and kālāpakaḥ. A restricting statement has probably dropped out of the source. This is recorded in the notes and not translated.
• 1278 chandobrāhmaṇāni ca tadviṣayāṇi 4.2.66 — niyama: prokta-affixed names of chandas and brāhmaṇa texts are bound to the student affix and never used without it. kaṭha + ṇini (a pupil of Vaiśampāyana), with luk by kaṭhacarakāl luk (SK 1487); + the student aṇ, with luk by 1274: kaṭhāḥ.

M. The four senses themselves (deśe tannāmni) — running example: udumbarāḥ santy asmin deśe → audumbaraḥ

• 1279 tad asminn astīti deśe tannāmni 4.2.67 — after a first-case word, the affix as prescribed in the sense "that exists in this", when the derivative is the name of a country (an apavāda of matup). On the running form: udumbara + aṇ → ādivṛddhi turns the u into au, and the a is deleted: audumbaraḥ. TB: deśe tannāmni carries into the next three sūtras.
• 1280 tena nirvṛttam 4.2.68 — after an instrumental word, "brought into being (founded) by that": kuśāmba + aṇ + ṅīp → kauśāmbī nagarī.
• 1281 tasya nivāsaḥ 4.2.69 — after a sixth-case word, "the dwelling of that": śaibaḥ. BM reports that a nivāsa is mere residence, while a viṣaya implies ownership. Linking this to viṣayo deśe (1261) is the translator's step.
• 1282 adūrabhavaś ca 4.2.70 — "not far from that": vaidiśam (the town near the Vidiśā). The ca brings the first three senses together. From here on, every affix is cāturarthika (caturarthī + ṭhañ as an adhyātmādi word, TB).

N. Cāturarthika affixes by base class: añ, aṇ, and the seventeen-fold rule — running example: kakṣatur asmin asti → kākṣatavam

• 1283 orañ 4.2.71 — after a u-final stem, añ (the añ-adhikāra runs to 4.2.77). On the running form: kakṣatu + añ → ādivṛddhi turns the first a into ā, and the final u takes guṇa → av: kākṣatavam. For a river, matup (1304) prevails because it is the later rule: ikṣumatī.
• 1284 matoś ca bahvajaṅgāt 4.2.72 — añ after a matup-final word whose aṅga has many vowels: saidhrakāvatam. ahi has two vowels, so aṇ: āhimatam. aṅga is stated so that "many vowels" qualifies the base, not the matup-final word.
• 1285 bahvacaḥ kūpeṣu 4.2.73 — añ after a many-voweled word when a well is named: dairghavaratraḥ kūpaḥ.
• 1286 udak ca vipāśaḥ 4.2.74 — añ for wells on the north bank of the Vipāś, even after a word with few vowels: dāttaḥ. On the south bank the affix is aṇ.
• 1287 saṃkalādibhyaś ca 4.2.75 — añ after saṅkala etc.; kūpeṣu has ceased: sāṅkalam, pauṣkalam.
• 1288 strīṣu sauvīrasālvaprākṣu 4.2.76 — añ for feminine place-names in Sauvīra, Sālva and the East (then ṅīp): dāttāmitrī nagarī, vaidhūmāgnī, mākandī.
• 1289 suvāstvādibhyo 'ṇ 4.2.77 — aṇ, overriding añ, after suvāstu etc.: sauvāstavam, vārṇavam (the source reads varmu, for varṇu). Naming aṇ explicitly also blocks matup for a river: sauvāstavī (a river so named after the Suvāstu).
• 1290 roṇī 4.2.78 — aṇ after roṇī and words ending in it, overriding the well-añ of 1285: rauṇaḥ, ājakaroṇaḥ.
• 1291 kopadhāc ca 4.2.79 — aṇ, overriding añ, after a stem with k as penultimate: kārṇacchidrakaḥ kūpaḥ, kārkavākavam, traiśaṅkavam.
• 1292 vuñchaṇ…kumudādibhyaḥ 4.2.80 — seventeen affixes after seventeen gaṇas, in order, in the four senses (caturarthī):

1. arīhaṇādi + vuñ: ārīhaṇakam

2. kṛśāśvādi + chaṇ: kārśāśvīyam

3. ṛśyādi + ka: ṛśyakam

4. kumudādi + ṭhac: kumudikam

5. kāśādi + ila: kāśilaḥ

6. tṛṇādi + sa: tṛṇasam

7. prekṣādi + ini: prekṣī

8. aśmādi + ra: aśmaraḥ

9. sakhyādi + ḍhañ: sākheyam

10. saṅkāśādi + ṇya: sāṅkāśyam

11. balādi + ya: balyam

12. pakṣādi + phak: pākṣāyaṇaḥ; the gaṇasūtra pathaḥ pantha ca inside pakṣādi gives pānthāyanaḥ (patho 'dūrabhavaḥ)

13. karṇādi + phiñ: kārṇāyaniḥ

14. sutaṅgamādi + iñ: sautaṅgamiḥ

15. pragadi[n]ādi + ñya: the stem is pragadin, giving prāgadyaḥ

16. varāhādi + kak: vārīhakaḥ (as printed; vārāhakaḥ is expected)

17. kumudādi + ṭhak: kaumudīkaḥ (as printed)

BM and TB (printed under 1291.b1 and 1290.t1):

• The compound is two dvandvas of seventeen members each.
• kumuda is not reduced by ekaśeṣa, so that the affixes pair in order.
• pragadin is kept with its n, so that it is not taken as i-final.
• ādi goes with each member, so this is effectively seventeen rules.
• The four senses apply as each case allows.
• prekṣī means "made by a prekṣa (one who looks)" or "by prekṣā".

O. Lup of the cāturarthika affix, yuktavadbhāva, and Pāṇini's aśiṣya set — running example: pañcālānāṃ nivāso janapadaḥ → pañcālāḥ (lup, plural as of the base)

• 1293 janapade lup 4.2.81 — when a janapada is named, lup of the cāturarthika affix. The aṇ of 1281 after pañcāla undergoes lup, leaving pañcāla.
• 1294 lupi yuktavad vyaktivacane 1.2.51 — under lup, gender and number follow the base (yukta), not the word janapada. So the form is plural: pañcālāḥ; likewise kuravaḥ, aṅgāḥ, vaṅgāḥ, kaliṅgāḥ.
• 1295 tad aśiṣyaṃ saṃjñāpramāṇatvāt 1.2.53 — the yuktavat rule (1294) need not be taught, because names (saṃjñā) are authoritative; BM glosses this as "worldly usage". BM: this is an earlier teachers' sūtra that Pāṇini rejects. TB: it was framed in deference to them.
• 1296 lub yogāprakhyānāt 1.2.54 — lup need not be taught either (TB: that is, 1293 and 1301), because no derivational connection is perceived. pañcālāḥ is not understood as "connected with the Pañcāla kings"; it is simply a rūḍha name.
• 1297 yogapramāṇe ca tadabhāve 'darśanaṃ syāt 1.2.55 — if the connection were authoritative, the word would not be used once the connection is gone. But pañcāla is still used under a non-Pañcāla ruler, so it is conventional.
• 1298 pradhānapratyayārthavacanam arthasyānyapramāṇatvāt 1.2.56 — the earlier teachers' definition "the affix-meaning is principal (the qualificand)" also need not be taught, since meaning is known from usage.
• 1299 kālopasarjane ca tulyam 1.2.57 — the same holds for the older definitions of time, such as adyatana ("today"), and of upasarjana ("the qualifier is upasarjana", and the like). These are not to be taught either, because usage establishes them.
• 1300 viśeṣaṇānāṃ cājāteḥ 1.2.52 — (read by the SK after 1.2.57) qualifiers of a lup-sense also take the base's gender and number, except a jāti word. On the running form: pañcālā ramaṇīyāḥ and godau ramaṇīyau, but pañcālā janapadaḥ and godau grāmaḥ. Three vārttikas:
• harītakyādiṣu vyaktiḥ (only gender): harītakyaḥ.
• khalatikādiṣu vacanam (only number): khalatikaṃ vanāni.
• manuṣyalupi pratiṣedhaḥ (not for qualifiers when a man is meant): cañcā abhirūpaḥ.
• 1301 varaṇādibhyaś ca 4.2.82 — lup after varaṇā etc., even where no janapada is meant: varaṇānām adūrabhavaṃ nagaram → varaṇāḥ (feminine plural, following the base).
• 1302 śarkarāyā vā 4.2.83 — after śarkarā, the lup is optional. The forms are given under 1303.
• 1303 ṭhakchau ca 4.2.84 — ṭhak and cha after śarkarā. Its membership in kumudādi and varāhādi (1292) also gives ṭhac and kak, and the vā of 1302 brings back the general aṇ with optional lup. Six forms result:
• śarkarā (lup of aṇ)
• śārkaram (aṇ)
• śārkarikam (ṭhak)
• śarkarīyam (cha)
• śarkarikam (ṭhac)
• śārkarakam (kak)

P. matup and its relatives for rivers and places; kuk and the luk of cha — running example: ikṣavaḥ santy asyām → ikṣumatī; naḍa → naḍvān / naḍvalaḥ / naḍakīyam

• 1304 nadyāṃ matup 4.2.85 — matup (-mat, u and p are it-markers) when a river is named: ikṣu + mat + ṅīp → ikṣumatī. It prevails over añ (1283) because it is the later rule.
• 1305 madhvādibhyaś ca 4.2.86 — matup after madhu etc., for a place other than a river: madhumān.
• 1306 kumudanaḍavetasebhyo ḍmatup 4.2.87 — ḍmatup after kumuda, naḍa and vetasa (BM, with the Bhāṣya and Kaiyaṭa: a matvartha affix). The ḍ deletes the final a, and the m of mat → v:
• by jhayaḥ (SK 1898): kumudvān, naḍvān. BM holds that mād upadhāyāḥ applies to naḍvān too; TB holds that jhayaḥ applies there by paratva.
• by mād upadhāyāḥ (SK 1897): vetasvān.

Vārttika mahiṣāc ca: mahiṣmān (a country name).

• 1307 naḍaśādāḍ ḍvalac 4.2.88 — ḍvalac after naḍa and śāda ("mud / grass"); the ḍ deletes the ṭi: naḍvalaḥ, śādvalaḥ.
• 1308 śikhāyā valac 4.2.89 — valac after śikhā, blocking aṇ for a place name: śikhāvalam.
• 1309 utkarādibhyaś chaḥ 4.2.90 — cha (→ īy) after utkara etc.: utkarīyaḥ (in any of the four senses).
• 1310 naḍādīnāṃ kuk ca 4.2.91 — cha, with the augment kuk on the base, after naḍa etc.: naḍa + k + īya → naḍakīyam. Two gaṇasūtras:
• kruñcā hrasvatvaṃ ca (the ā is shortened): kruñcakīyaḥ.
• takṣan nalopaś ca (the n is deleted): takṣakīyaḥ.
• 1311 bilvakādibhyaś chasya luk 6.4.153 — after bilvaka etc. (naḍādi members with kuk already added), cha undergoes luk before a further taddhita. bilva + kuk + cha → bilvakīyā; + the bhava aṇ, before which the whole īya undergoes luk → bailvakāḥ. Also vetrakīyāḥ → vaitrakāḥ; BM's pratīka supports the transmitted plural. The rule says chasya ("of cha") so that the kuk survives; otherwise saṃniyogaśiṣṭānām… would remove it along with cha. It says luk, not lopa, so that the whole īya goes, and not only the y.