After this chapter you will be able to recognise and derive the aluk compounds: the compounds in which the first member keeps its case-ending instead of losing it. In every compound you have met so far the case-ending inside the compound is dropped: rājñaḥ puruṣaḥ → rājapuruṣaḥ. This chapter gives the closed set of rules that withhold that dropping. You will know which case survives after which first member, and before which last member: the fifth case in stokānmuktaḥ, the third in ojasākṛtam and ātmanāpañcamaḥ, the fourth in the grammarians' own terms ātmanepadam and parasmaipadam, the seventh in names such as tvacisāraḥ, kaṇṭhekālaḥ and yudhiṣṭhiraḥ and optionally in hastebandhaḥ, kheśayaḥ and pūrvāhṇetare, and the sixth in the abusive caurasyakulam and in hotuḥputraḥ. You will also see the places where the dropping is expressly restored (sthaṇḍilaśāyī, samasthaḥ), and three rules on the sound s that act on this chapter's compounds: SK 967 and SK 983 on forms that keep the ending (gaviṣṭhiraḥ, mātuḥṣvasā), and SK 984 on the luk alternative of SK 982 (mātṛṣvasā). Every one of the twenty-seven rules is shown on a concrete word.
rājñaḥ puruṣaḥ "the king's man" becomes rājapuruṣaḥ, with the ending of rājñaḥ gone. Aluk means "no luk": the ending stays. Chapter 18 gave the compound rules stokāntikadūrārthakṛcchrāṇi ktena (SK 701), kartṛkaraṇe kṛtā bahulam (SK 694), caturthī tadarthārthabalihitasukharakṣitaiḥ (SK 698) and saṃjñāyām (SK 721), which this chapter cites as the rules that make the compound in the first place.rājñaḥ puruṣaḥ → rājapuruṣaḥ (luk, the default) against stokāt muktaḥ → stokānmuktaḥ (aluk, under this heading)The chapter opens with a two-word heading. It does nothing on its own; it supplies the words "aluk" and "when an uttarapada follows" to every rule that comes after it. The running example sets the default beside the exception: rājapuruṣaḥ drops its ending as usual, while stokānmuktaḥ keeps the ending -āt because a rule under this heading says so.
Rule. This sūtra is an adhikāra, a heading. Its two words are read into the rules that follow: aluk, "the luk of the case-ending by SK 650 does not take place", and uttarapade, "when the last member of a compound follows". The two words do not run equally far. Aluk governs up to, but not including, the rule ānaṅ ṛtaḥ (A 6.3.25, SK 921). Uttarapade governs to the end of this quarter (pāda) of the Aṣṭādhyāyī, A 6.3; Bhaṭṭoji's commentators note that the Bhāṣya, the ancient commentary, is the authority for that extent. Every rule below is therefore to be read as "the *n*th case-ending of the first member is not elided when the last member X follows".
Example.
rājñaḥ puruṣaḥ "the king's man" → compound rājan + as + puruṣa → the ending as is elided by luk (SK 650), the default → rājapuruṣaḥ.
stokāt muktaḥ "freed from a little" → compound stoka + āt + mukta → this heading, together with SK 959, withholds the luk → stokānmuktaḥ (Section B).
Watch out. The heading is not a vidhi, an operative rule that does something by itself. Bhaṭṭoji's commentators point out that if it were, it would keep the case-ending in rājapuruṣa and every other compound, and SK 959 would then have nothing left to do. It only holds two words over for the rules that follow.
stokāt muktaḥ → stokānmuktaḥOne rule keeps the fifth case. The compound stokāt muktaḥ was formed in chapter 18 by SK 701, which compounds stoka "a little", words meaning antika "near", words meaning dūra "far" and kṛcchra "hardship" with a word ending in kta. Here the ending -āt of stokāt is kept, and the result is stokānmuktaḥ.
Rule. After stoka and the rest — that is, stoka, words meaning antika, words meaning dūra, and kṛcchra, the list of SK 701 — the pañcamī is not elided when an uttarapada follows.
Example.
stokāt muktaḥ "freed from a little" → compound by SK 701 → stoka + āt + mukta → the pañcamī -āt is the ending this rule protects; mukta is the uttarapada → the luk does not happen → stokāt-mukta → the final t becomes n before m → stokānmuktaḥ.
Bhaṭṭoji's commentators fill in the rest of the list: antikād āgataḥ, abhyāśād āgataḥ "come from near", dūrād āgataḥ, viprakṛṣṭād āgataḥ "come from far", kṛcchrād āgataḥ "come out of hardship", and alpān muktaḥ beside stokānmuktaḥ.
Applies when / blocked by. Only when an uttarapada follows. In niṣkrāntaḥ stokāt → nistokaḥ "gone out from a little", stoka is itself the last member and nothing follows it, so its ending is elided as usual. Bhaṭṭoji's commentators add that the dual and plural, stokān muktau, stokān muktāḥ, are not compounded at all, simply because usage does not have them.
A vārttika adds one more form: brāhmaṇācchaṃsina upasaṃkhyānam, "brāhmaṇācchaṃsin is to be added". An upasaṃkhyāna is an addition to a rule. brāhmaṇāt + śaṃsin → brāhmaṇācchaṃsī, the name of a particular officiating priest (ṛtvij). The śastras, recitations prescribed in the Brāhmaṇa, are themselves figuratively called brāhmaṇāni. He who recites (śaṃsati) them is the brāhmaṇācchaṃsī. The word is properly an object, so the case should be the dvitīyā; that the pañcamī stands in its place follows from the very fact of the addition.
ojasā kṛtam → ojasākṛtam (also manasā guptā → manasāguptā, ātmanā pañcamaḥ → ātmanāpañcamaḥ)Four rules keep the third case. The compound ojasā kṛtam "done by strength" was formed by SK 694, which compounds an instrumental with a kṛt-word. The ending -ā of ojasā stays: ojasākṛtam. The same happens after manas in a name (manasāguptā) and before ājñāyin, and after ātman before an ordinal (ātmanāpañcamaḥ).
Rule. After ojas "strength", sahas "might", ambhas "water" and tamas "darkness", the tṛtīyā is not elided when an uttarapada follows. The sūtra's first word packs the four nouns into one group-dvandva (samāhāra-dvandva), as Bhaṭṭoji's commentators note.
Example.
ojasā kṛtam "done by strength" → compound by SK 694 → ojas + ā + kṛta → the tṛtīyā -ā is kept before the uttarapada kṛta → ojasākṛtam.
Bhaṭṭoji writes "and so on"; his commentators spell it out: sahasākṛtam, ambhasākṛtam, tamasāvṛtam "covered by darkness". They add that tamovṛtam, with the ending dropped, is simply incorrect; a poet's use of a form like it is defended in another way, and scholars discuss this.
Applies when / blocked by. Two vārttikas widen the list. Añjasa upasaṃkhyānam: after añjas too — añjasā kṛtam → añjasākṛtam "done straightforwardly", añjas meaning straightness (ārjava). Puṃsānujo januṣāndha iti ca: two set forms — puṃsānujaḥ "the younger, because his elder is a male (pumān)", with the tṛtīyā puṃsā kept, and januṣāndhaḥ "blind by birth (januṣā)", that is, blind from birth on.
Rule. After manas "mind", the tṛtīyā is not elided when an uttarapada follows and the compound is a name. The word tṛtīyāyāḥ is carried forward from SK 960.
Example.
manasā guptā "guarded by the mind" → compound manas + ā + guptā → the ending -ā is kept because the compound is a name → manasāguptā, the name of some woman.
Applies when / blocked by. Only in a name. When the same words do not form a name, the ending is elided: manoguptā.
Rule. And after manas, the tṛtīyā is not elided when ājñāyin follows. Manasaḥ is carried forward from SK 961. Bhaṭṭoji's commentators take this rule as serving the case that is not a name, which SK 961 does not cover.
Example.
manasā ājñātuṃ śīlam asya "whose habit is to direct by the mind" → manas + ā + ājñāyin → the ending -ā is kept → manasājñāyī. Bhaṭṭoji's commentators derive ājñāyin with the affix ṇini (A 3.2.78) and the augment yuk (A 7.3.33). An augment (āgama) is an element inserted into an affix or stem, replacing nothing; yuk supplies the y of ājñāyin.
Rule. After ātman "self", the tṛtīyā is not elided when an uttarapada follows. A vārttika, pūraṇa iti vaktavyam, restricts this: only when the uttarapada ends in a pūraṇa affix, the affix that makes an ordinal number ("fifth").
Example.
ātmanā pañcamaḥ "himself the fifth" → ātman + ā + pañcama → pañcama ends in a pūraṇa affix, so the ending -ā is kept → ātmanāpañcamaḥ.
Applies when / blocked by. Only before an ordinal. Otherwise the ending is elided: ātmakṛtam "done by oneself". In the phrase janārdanas tvām ātmacaturtha eva, the word ātmacaturtha is a bahuvrīhi, not a case of this rule, so its ending is dropped. Scholars discuss how the tṛtīyā in ātmanāpañcamaḥ is to be construed, and whether pūraṇe belongs to the sūtra itself or to the vārttika, as the Kaumudī has it.
ātmane padam → ātmanepadam (also parasmai padam → parasmaipadam)Two rules keep the fourth case, and only in the technical vocabulary of grammarians themselves. Ātmanepada "the ending for oneself" and parasmaipada "the ending for another" are the names of the two sets of verb-endings. Both keep the dative of their first member.
Rule. After ātman (carried forward from SK 963), the caturthī is not elided in a technical designation of the grammarians (vaiyākaraṇī ākhyā). The caturthī here means "for the purpose of" (tādarthya): the ending is "for oneself".
Example.
ātmane padam "the ending for oneself" → ātman + e + pada → the caturthī -e is kept because this is a grammarians' term → ātmanepadam.
Likewise ātmanebhāṣāḥ, "roots conjugated for oneself"; Bhaṭṭoji's commentators note that this term is not in the Aṣṭādhyāyī but occurs in the Dhātupāṭha.
Applies when / blocked by. The compound itself needs a licence. Bhaṭṭoji's commentators note that a compound might be doubted here. The members show no relation of material and product (prakṛti–vikṛti), the relation on which SK 698 compounds a dative. Bhaṭṭoji obtains the compound by yogavibhāga: splitting the rule SK 698 so that its first word caturthī works on its own. Bhaṭṭoji's commentators add that the very provision of aluk would also justify the compound, and that vaiyākaraṇī is formed with the affix aṇ (A 4.3.73).
Rule. And after para "another", the caturthī is not elided in a grammarians' designation. Both caturthyāḥ and vaiyākaraṇākhyāyām are carried forward from SK 964.
Example.
parasmai padam "the ending for another" → the caturthī of para is parasmai → parasmai + pada → the ending is kept → parasmaipadam.
Likewise parasmaibhāṣāḥ, roots conjugated with those endings.
tvaci sāraḥ → tvacisāraḥ (also kaṇṭhe kālaḥ → kaṇṭhekālaḥ, stambe ramate → stamberamaḥ)The seventh case has the longest section of the chapter. This group holds the rules that keep it without option. The general rule is SK 966: after a first member ending in a consonant or in a, the locative ending stays when the compound is a name, as in tvacisāraḥ "bamboo". The rules after it either restrict that (tax-names among the easterners), extend it to non-names (madhyeguruḥ), or apply it to particular lists: body-part words (kaṇṭhekālaḥ), tatpuruṣas ending in a kṛt (stamberamaḥ), and four words before ja.
Rule. After a first member ending in a hal (a consonant) or in a, the saptamī is not elided when an uttarapada follows and the compound is a name.
Example.
tvaci sāraḥ "pith in the bark" → tvac ends in a consonant → tvac + i + sāra → the saptamī -i is kept because the compound is a name → tvacisāraḥ, a name of bamboo. Bhaṭṭoji's commentators note that this is a bahuvrīhi whose two members are not in the same case (a vyadhikaraṇa bahuvrīhi), and that this very form is the jñāpaka that licenses such a compound.
The second consonant-final example, yudhiṣṭhiraḥ, and the a-final example, araṇyetilakāḥ, are given under SK 967 (Section I), where the ṣatva rule that they also need is taught.
Applies when / blocked by. Why "ending in a hal or in a"? A first member ending in ī or i gets no aluk: nadīdohaḥ (SK 968) and guptibandhaḥ (SK 971).
Rule. In a name of a tax (kāra, the king's share) current among the easterners (prāñcaḥ), the saptamī after a consonant-final or a-final first member is not elided when a consonant-initial uttarapada follows.
Example.
mukuṭe kārṣāpaṇam "a kārṣāpaṇa coin on the crown" → mukuṭa + e + kārṣāpaṇa → kārṣāpaṇa begins with a consonant; the compound is an eastern tax-name → the saptamī -e is kept → mukuṭekārṣāpaṇam.
dṛṣadi māṣakaḥ "a māṣaka on the grindstone" → dṛṣad + i + māṣaka → dṛṣadimāṣakaḥ.
Applies when / blocked by. These forms are names after a consonant- or a-final stem, so SK 966 would already give them. The sūtra therefore works as a niyama, a restriction, and Bhaṭṭoji reads it three ways: only in a tax-name, only among the easterners, only before a consonant-initial member. The counter-examples show each limit. abhyāhitapaśuḥ is an eastern name, but of something other than a tax: no aluk. yūthapaśuḥ is not eastern: no aluk. avikaṭoraṇaḥ has a vowel-initial last member (uraṇa "a ram"; avikaṭa "a flock of sheep" is formed with the affix kaṭac): no aluk. And nadyāṃ dohaḥ → nadīdohaḥ, the tax levied at a river-crossing, has an ī-final first member: no aluk.
Rule. After madhya "middle", the saptamī is not elided when guru "heavy" follows. Saptamyāḥ is carried forward. Bhaṭṭoji's commentators take this rule as serving the non-name case, since SK 966 has already covered names.
Example.
madhye guruḥ "heavy in the middle" → madhya + e + guru → the saptamī -e is kept → madhyeguruḥ.
A vārttika, antāc ca, adds anta "end": ante guruḥ → anteguruḥ "heavy at the end".
Rule. After a word for a part of the body (svāṅga) other than mūrdhan and mastaka, both meaning "head", the saptamī is not elided when an uttarapada follows, provided that uttarapada is not kāma "desire". Bhaṭṭoji's commentators read saṃjñāyām "when a name" into this rule as well.
Example.
kaṇṭhe kālaḥ "black in the throat" → kaṇṭha + e + kāla → the saptamī -e is kept → kaṇṭhekālaḥ, a name of Śiva.
urasi lomā "hair on the chest" → uras + i + loman → urasilomā, someone's name.
Both are bahuvrīhis whose members are not in the same case, and Bhaṭṭoji's commentators take this rule as the jñāpaka that licenses them.
Applies when / blocked by. Why "other than mūrdhan and mastaka"? mūrdhaśikhaḥ, mastakaśikhaḥ "with a tuft on the head": the ending is elided. Why "not before kāma"? mukhe kāmo 'sya → mukhakāmaḥ "whose desire is in the mouth": elided.
Rule. In a tatpuruṣa whose uttarapada ends in a kṛt affix, the saptamī is variously (bahulam) not elided. Bhaṭṭoji's commentators carry saptamyāḥ and saṃjñāyām forward, so the rule is for names.
Example.
stambe ramate "delights in a clump of grass" → the kṛt affix ac is added to the root by stambakarṇayo ramijapoḥ (A 3.2.13, SK 2927), and stambe is its upapada → upapada compound stamba + e + rama → the saptamī -e is kept → stamberamaḥ "the elephant".
karṇe japati "whispers in the ear" → the same affix ac → karṇa + e + japa → karṇejapaḥ "the informer".
Applies when / blocked by. "Variously" means that in some cases the ending is elided after all. kuruṣu carati "who moves among the Kurus" → affix ṭa by careṣ ṭaḥ (A 3.2.16) → kurucaraḥ, with luk. Bhaṭṭoji's commentators observe that this absence of aluk is secured by bahulam alone, so haladantāt need not be carried into this rule.
Rule. After prāvṛṣ "the rains", śarad "autumn", kāla "time" and div "sky", the saptamī is not elided when ja follows.
Example.
prāvṛṣi + ja → the saptamī -i is kept → prāvṛṣijaḥ.
Likewise śaradijaḥ, kālejaḥ, divijaḥ.
Applies when / blocked by. All four first members end in a consonant or in a, and the compounds are names, so SK 966 already gives these forms. Bhaṭṭoji calls this sūtra a prapañca, an elaboration of the preceding rule that adds nothing new.
haste bandhaḥ → hastebandhaḥ / hastabandhaḥ (also pūrvāhṇe tare → pūrvāhṇetare / pūrvāhṇatare, khe śayaḥ → kheśayaḥ / khaśayaḥ)Four rules make the seventh case optional before particular last members. Each therefore yields two correct forms, one with the ending and one without: hastebandhaḥ / hastabandhaḥ, varṣejaḥ / varṣajaḥ, pūrvāhṇetare / pūrvāhṇatare, kheśayaḥ / khaśayaḥ. Two of these rules stand between the obligatory rules in the Kaumudī's own order (SK 971 and SK 974); they are gathered here so that the options are seen together.
Rule. And when bandha "binding" follows, the saptamī after a consonant-final or a-final first member is optionally not elided. Bhaṭṭoji supplies haladantāt from SK 966; his commentators take bandha as the noun formed with the affix ghañ.
Example.
haste bandhaḥ "a binding on the hand" → hasta + e + bandha → the option: keep the saptamī → hastebandhaḥ; or let it be elided → hastabandhaḥ.
Applies when / blocked by. Why "after a consonant-final or a-final"? After the i-final gupti only guptibandhaḥ is formed. Bhaṭṭoji names no compound type for hastebandhaḥ; scholars discuss whether it is a tatpuruṣa made by saṃjñāyām (SK 721) or a bahuvrīhi, given the bar in SK 977.
Rule. After varṣa, kṣara, śara and vara, the saptamī is optionally not elided when ja follows. Bhaṭṭoji supplies what the sūtra leaves out: "after these, the saptamī, when ja follows" — saptamyāḥ by anuvṛtti and je from SK 973.
Example.
varṣe + ja → keep the saptamī → varṣejaḥ; or elide it → varṣajaḥ.
Likewise kṣarejaḥ / kṣarajaḥ, śarejaḥ / śarajaḥ, varejaḥ / varajaḥ.
Rule. After a name of a time, the saptamī is optionally not elided when gha, kāla or tana follows. Gha is the common name of the two comparison affixes tarap "more" and tamap "most" (A 1.1.22). Tana is the affix that makes words like "belonging to the forenoon".
Example. The time-name is pūrvāhṇa "forenoon", and its saptamī is pūrvāhṇe.
Before gha: pūrvāhṇe + tara → pūrvāhṇetare / pūrvāhṇatare; pūrvāhṇe + tama → pūrvāhṇetame / pūrvāhṇatame.
Before kāla: pūrvāhṇe + kāla → pūrvāhṇekāle / pūrvāhṇakāle.
Before tana: pūrvāhṇe + tana → pūrvāhṇetane / pūrvāhṇatane.
Applies when / blocked by. Bhaṭṭoji's commentators note that here tarap and tamap are added to the word already carrying its saptamī. They add that pūrvāhṇetane has the affix ṭyu with the augment tuṭ, which supplies the t of tana, by vibhāṣā pūrvāhṇāparāhṇābhyām (A 4.3.24, SK 1392).
Rule. After a first member that is not a time-word, the saptamī is optionally not elided when śaya "lying", vāsa "dwelling" or vāsin "dweller" follows. The option is carried forward from SK 974, as the paired forms show.
Example.
khe śayaḥ "lying in the sky" → kha + e + śaya → kheśayaḥ / khaśayaḥ.
grāme vāsaḥ "dwelling in the village" → grāmevāsaḥ / grāmavāsaḥ.
grāme vāsī "a dweller in the village" → grāmevāsī / grāmavāsī.
Applies when / blocked by. Haladantāt is still carried: after the i-final bhūmi "ground" only bhūmiśayaḥ is formed. A vārttika, apo yoniyanmatuṣu, adds the word ap "water": its saptamī apsu is kept before yoni "origin", before the taddhita affix yat, and before the affix matup "possessing".
apsu yonir utpattir yasya "whose origin is in the waters" → apsuyoniḥ.
apsu bhavaḥ "produced in the waters" → apsu + yat → apsavyaḥ; Bhaṭṭoji's commentators add that yat comes by the digādi list (A 4.3.54), and that the u takes guṇa (A 6.4.146), that is, is replaced by o, one of the guṇa vowels a, e, o, and the o then becomes av (A 6.1.79).
apsu + matup → apsumantau, "the two containing apsu", the two ājyabhāga offerings whose mantras have that word.
sthaṇḍile śāyī → sthaṇḍilaśāyī (also same sthaḥ → samasthaḥ)Two rules switch the aluk off again. Before a last member ending in in, or siddha, or a form of the root bandh, and before stha in ordinary speech, the seventh case is elided as in any other compound.
Rule. And there is no aluk of the saptamī when the uttarapada ends in the affix in, or is siddha "accomplished", or is a form of badhnāti "binds". The ending is elided as usual.
Example.
sthaṇḍile śāyī "one who lies on the bare ground" → sthaṇḍila + e + śāyin → śāyin ends in in → the saptamī is elided → sthaṇḍilaśāyī.
sāṃkāśye siddhaḥ "accomplished at Sāṃkāśya" → sāṃkāśyasiddhaḥ.
cakre baddhaḥ "bound to the wheel" → cakrabaddhaḥ; Bhaṭṭoji's commentators note that this compound of a saptamī with a kta-word is made by the vārttika sādhanaṃ kṛtā.
Applies when / blocked by. Scholars discuss whether tatpuruṣe is carried into this rule, so that the bar holds in a tatpuruṣa and the option of SK 971 survives in a bahuvrīhi.
Rule. And there is no aluk of the saptamī when stha "standing" follows, in the spoken language (bhāṣā) as opposed to the Veda.
Example.
same sthaḥ "standing on level ground" → sama + e + stha → the saptamī is elided → samasthaḥ.
Applies when / blocked by. Why "in the spoken language"? In the Vedic phrase kṛṣṇo 'syākhareṣṭhaḥ the saptamī of ākhara is kept before stha, and the s then becomes ṣ. Bhaṭṭoji's commentators hold that this sūtra negates SK 972 only, not SK 970, by the maxim that a negation bears on the rule immediately before it. Their evidence is that the Bhāṣya accepts kaṇṭhestha with the saptamī kept.
caurasya kulam → caurasyakulam (also dāsyāḥ putraḥ → dāsyāḥputraḥ / dāsīputraḥ, hotuḥ putraḥ → hotuḥputraḥ)Four rules keep the sixth case. The first keeps it whenever the compound is meant as abuse (caurasyakulam "a thief's brood") and carries five vārttikas of fixed forms. The second makes it optional before putra in reproach. The last two concern first members ending in ṛ that name a relation of teaching or of birth, such as hotṛ and pitṛ: the genitive hotuḥ, pituḥ is kept, obligatorily before most words and optionally before svasṛ "sister" and pati "husband".
Rule. The ṣaṣṭhī is not elided when an uttarapada follows and abuse (ākrośa, reviling) is meant. Aluk and uttarapade are supplied from the heading.
Example.
caurasya kulam "the family of a thief", said as abuse → caura + asya + kula → the ṣaṣṭhī -asya is kept → caurasyakulam.
Applies when / blocked by. Why "when abuse is meant"? brāhmaṇakulam "a brāhmaṇa's family" is no insult, so the ending is elided. Five vārttikas then lay down fixed forms in which the ṣaṣṭhī is kept:
vācaḥ + yukti → vācoyuktiḥ "the use of words"; diśaḥ + daṇḍa → diśodaṇḍaḥ; paśyataḥ + hara → paśyatoharaḥ "one who steals while the owner looks on", the ṣaṣṭhī of paśyat coming from ṣaṣṭhī cānādare (A 2.3.38), the rule for a person disregarded.amuṣya apatyam "so-and-so's offspring" → the affix phak, because amuṣya is in the naḍādi list (A 4.1.99, SK 1101) → āmuṣyāyaṇaḥ. amuṣyaputrasya bhāvaḥ "the state of being so-and-so's son" → the affix vuñ, because amuṣyaputra is in the manojñādi list (A 5.1.133, SK 1798) → āmuṣyaputrikā. Likewise āmuṣyakulikā. Bhaṭṭoji's commentators call all three nipātana, ready-made forms.devānām + priya → devānāṃpriyaḥ, "dear to the gods", only when it means a fool; otherwise devapriyaḥ.śunaḥ, before śepa, puccha and lāṅgūla "tail" — śunaḥśepaḥ, śunaḥpucchaḥ, śunolāṅgūlaḥ; Bhaṭṭoji's commentators take them as bahuvrīhis that are names of particular sages.divaḥ + dāsa → divodāsaḥ, the name of a royal sage. Scholars discuss the exact wording of this and the preceding vārttika as printed.Rule. When putra "son" follows, the ṣaṣṭhī is optionally not elided, in reproach (nindā). The condition of reproach comes from ākrośe carried forward from SK 979.
Example.
dāsyāḥ putraḥ "a slave-girl's son", as a taunt → dāsī + āḥ + putra → keep the ṣaṣṭhī → dāsyāḥputraḥ; or elide it → dāsīputraḥ.
Applies when / blocked by. Why "in reproach"? brāhmaṇīputraḥ "a brāhmaṇa woman's son" is not a taunt: only the form with luk.
Rule. After a first member ending in ṛ that denotes a relation through learning (vidyā-sambandha) or a relation through birth (yoni-sambandha), the ṣaṣṭhī is not elided when an uttarapada follows. Bhaṭṭoji's commentators note that the sūtra's plural stands for a singular.
Example.
hotuḥ antevāsī "the pupil of a hotṛ" → hotṛ, the priest who performs a particular Ṛgvedic rite, is a word of teaching-relation → hotṛ + uḥ + antevāsin → the ṣaṣṭhī -uḥ is kept → hoturantevāsī (the visarga, the final breath-sound ḥ, becomes r before the vowel).
hotuḥ putraḥ "a hotṛ's son" → hotuḥputraḥ.
pituḥ antevāsī "a father's pupil" → pitṛ is a word of birth-relation → piturantevāsī; pituḥ putraḥ → pituḥputraḥ.
Applies when / blocked by. A vārttika, vidyāyonisambandhebhyas tatpūrvottarapadagrahaṇam, requires that both members, the prior and the last, denote such a relation. So not in hotṛdhanam "a hotṛ's wealth": dhana denotes neither relation, and the ending is elided. Either relation suffices for each member, so hotuḥputraḥ, teaching-word plus birth-word, stands.
Rule. After a first member ending in ṛ, the ṣaṣṭhī is optionally not elided when svasṛ "sister" or pati "husband" follows. Bhaṭṭoji's commentators carry the restriction of SK 981 with ṛdantāt: the first member must denote a relation of learning or of birth.
Example.
mātuḥ svasā "mother's sister" → mātṛ + uḥ + svasṛ → keep the ṣaṣṭhī → mātuḥsvasā; or elide it → mātṛsvasā.
Section I shows what the ṣatva rules then do to the s of svasṛ in each of these two forms.
Applies when / blocked by. Because the restriction is carried, bhoktṛ "an eater", which names no such relation, gets no aluk: bhoktṛsvasā only.
gavi sthiraḥ → gaviṣṭhiraḥ (also mātuḥ svasā → mātuḥṣvasā / mātuḥsvasā, mātṛṣvasā)Three rules from the last book of the Aṣṭādhyāyī (A 8.3) replace the s of a last member by ṣ. The Kaumudī reads them here because the forms they act on come from this chapter. SK 967 and SK 983 act on aluk forms: the sthira of gaviṣṭhiraḥ and yudhiṣṭhiraḥ stands after a kept seventh case, and the svasṛ of mātuḥṣvasā after a kept sixth (SK 982). SK 984 acts on the luk alternative of SK 982, the svasṛ of mātṛsvasā. All three are governed by apadāntasya mūrdhanyaḥ (A 8.3.55): a retroflex (mūrdhanya) sound replaces an s that is not at the end of a word. In the Kaumudī's own order SK 967 comes right after SK 966; it is placed here with its two companions.
Rule. After gavi and yudhi, the s of sthira "firm" becomes ṣ. Gavi and yudhi are the seventh-case forms of go "cow, ox" and yudh "battle", quoted in the sūtra as they sound (an anukaraṇa, an imitation). The genitive saḥ "of s" is carried forward from saheḥ sāḍaḥ saḥ (A 8.3.56), and Bhaṭṭoji's commentators read the nominative sthiraḥ as a genitive: "the s of sthira".
Example.
gavi sthiraḥ "firm on the ox" → go + i → the o gives way to the substitute av → gavi, the saptamī kept → gavi + sthira → the s of sthira stands after i and is not word-final → it becomes ṣ → gaviṣthira → the th after ṣ becomes the retroflex ṭh (this retroflexing of a dental after ṣ is ṣṭutva) → gaviṣṭhiraḥ.
yudhi sthiraḥ "firm in battle" → yudh is a consonant-final stem, so SK 966 keeps its saptamī -i in a name → yudhi + sthira → s becomes ṣ → yudhiṣṭhiraḥ, the name of Dharmaputra. Bhaṭṭoji's commentators add that yudh here is the root with the affix kvip in the sense of the action, and that sthira is made from sthā with the affix kirac of the Uṇādi list.
araṇye tilakāḥ → araṇya is a-final, so SK 966 keeps its saptamī in a name → araṇyetilakāḥ. The locative compound rests on saṃjñāyām (SK 721), since tilaka is not in the śauṇḍādi list.
Applies when / blocked by. Two things are special about gaviṣṭhiraḥ. First, the keeping of the ending rests on nothing but the sūtra's own wording gavi: Bhaṭṭoji says the aluk "follows from the very word gavi". His commentators explain why no ordinary rule could give it. The luk of the ending is an outer (bahiraṅga) operation, and by a maxim it beats even inner (antaraṅga) ones. So it would apply before the av of gavi ever arose, and there would be no consonant-final stem for SK 966 to work on. Second, the ṣatva itself needs a rule of its own, because the general ṣatva rule for an s that is a substitute or part of an affix, ādeśapratyayayoḥ (A 8.3.59), is blocked here by sāt padādyoḥ (A 8.3.111).
A vārttika, hṛddyubhyāṃ ca, adds two non-names: after hṛd "heart" and div "sky" the saptamī is not elided. SK 970 cannot give hṛdispṛk from the body-part word hṛd, because it carries saṃjñāyām.
hṛdi + spṛś "touching the heart" → hṛdispṛk; the stem hṛd replaces hṛdaya before the ending by A 6.1.63.
divi + spṛś "touching the sky" → divispṛk.
Bhaṭṭoji's commentators note, from the Bhāṣya, that in both the saptamī stands for the object by the very force of the aluk provision.
Rule. After mātuḥ and pituḥ — the sixth-case forms kept by SK 982 — the s of svasṛ optionally becomes ṣ, in a compound. Three words are carried forward: svasā from SK 984 (read as a genitive, "of svasṛ"), the genitive saḥ from A 8.3.56, and samāse "in a compound" from samāse 'ṅguleḥ saṅgaḥ (A 8.3.80).
Example.
mātuḥ + svasā, the aluk form of SK 982 → the option: s becomes ṣ → mātuḥṣvasā; or it stays → mātuḥsvasā.
Likewise pituḥṣvasā / pituḥsvasā.
Applies when / blocked by. Bhaṭṭoji closes with "but in the luk alternative —", meaning that for the form with the ending elided a special rule follows: SK 984.
Rule. After mātṛ and pitṛ, the s of svasṛ becomes ṣ, in a compound. This is the luk alternative of SK 982, and here the ṣatva is obligatory. The Kaumudī reads this rule after A 8.3.85, in reverse of the Aṣṭādhyāyī's order.
Example.
mātuḥ svasā "mother's sister" → the ending is elided by SK 982's luk option → mātṛ + svasā → the s becomes ṣ → mātṛṣvasā.
Likewise pitṛṣvasā.
Applies when / blocked by. Only in a compound. In a sentence there is not even the option: mātuḥ svasā, pituḥ svasā, with s unchanged. Bhaṭṭoji's commentators point out that this is the fruit of carrying samāse into the rule. They add why a special rule is needed at all: this s is neither a substitute nor an affix, so no general ṣatva rule reaches it. With this rule the Kaumudī's section on the aluk compounds ends.
| SK | Sūtra | What it does | Example |
|---|---|---|---|
| 958 | aluguttarapade | Heading: "aluk" (to A 6.3.25) and "when an uttarapada follows" (to the end of A 6.3). | rājapuruṣaḥ (luk) vs stokānmuktaḥ (aluk) |
| 959 | pañcamyāḥ stokādibhyaḥ | Pañcamī kept after stoka, near-words, far-words, kṛcchra. | stokāt muktaḥ → stokānmuktaḥ |
| 960 | ojaḥsaho'mbhastamasas tṛtīyāyāḥ | Tṛtīyā kept after ojas, sahas, ambhas, tamas (vārttikas: añjas; puṃsānuja, januṣāndha). | ojasā kṛtam → ojasākṛtam |
| 961 | manasaḥ saṃjñāyām | Tṛtīyā kept after manas in a name. | manasāguptā (name) vs manoguptā |
| 962 | ājñāyini ca | Tṛtīyā kept after manas before ājñāyin. | manasājñāyī |
| 963 | ātmanaś ca | Tṛtīyā kept after ātman, before an ordinal (vārttika). | ātmanāpañcamaḥ vs ātmakṛtam |
| 964 | vaiyākaraṇākhyāyāṃ caturthyāḥ | Caturthī kept after ātman in a grammarians' term. | ātmane padam → ātmanepadam |
| 965 | parasya ca | Caturthī kept after para in a grammarians' term. | parasmai padam → parasmaipadam |
| 966 | haladantāt saptamyāḥ saṃjñāyām | Saptamī kept after a consonant- or a-final stem in a name. | tvaci sāraḥ → tvacisāraḥ |
| 968 | kāranāmni ca prācāṃ halādau | Restriction: in eastern tax-names, only before a consonant. | mukuṭekārṣāpaṇam; not yūthapaśuḥ |
| 969 | madhyād gurau | Saptamī kept after madhya (and anta) before guru, non-names. | madhyeguruḥ, anteguruḥ |
| 970 | amūrdhamastakāt svāṅgād akāme | Saptamī kept after a body-part word (not mūrdhan, mastaka), not before kāma. | kaṇṭhekālaḥ; not mukhakāmaḥ |
| 972 | tatpuruṣe kṛti bahulam | Saptamī variously kept in a tatpuruṣa ending in a kṛt. | stamberamaḥ, karṇejapaḥ; but kurucaraḥ |
| 973 | prāvṛṭśaratkāladivāṃ je | Saptamī kept after prāvṛṣ, śarad, kāla, div before ja (elaboration of 966). | prāvṛṣijaḥ, divijaḥ |
| 971 | bandhe ca vibhāṣā | Saptamī optionally kept before bandha. | hastebandhaḥ / hastabandhaḥ |
| 974 | vibhāṣā varṣakṣaraśaravarāt | Saptamī optionally kept after varṣa, kṣara, śara, vara before ja. | varṣejaḥ / varṣajaḥ |
| 975 | ghakālataneṣu kālanāmnaḥ | Saptamī optionally kept after a time-name before tarap/tamap, kāla, tana. | pūrvāhṇetare / pūrvāhṇatare |
| 976 | śayavāsavāsiṣv akālāt | Saptamī optionally kept after a non-time-word before śaya, vāsa, vāsin (vārttika: apsu). | kheśayaḥ / khaśayaḥ; apsuyoniḥ |
| 977 | nensiddhabadhnātiṣu ca | No aluk before an in-final word, siddha, baddha. | sthaṇḍilaśāyī, cakrabaddhaḥ |
| 978 | sthe ca bhāṣāyām | No aluk before stha in the spoken language. | samasthaḥ; Vedic ākhareṣṭhaḥ |
| 979 | ṣaṣṭhyā ākrośe | Ṣaṣṭhī kept in abuse (five vārttikas of fixed forms). | caurasyakulam; devānāṃpriyaḥ |
| 980 | putre 'nyatarasyām | Ṣaṣṭhī optionally kept before putra in reproach. | dāsyāḥputraḥ / dāsīputraḥ |
| 981 | ṛto vidyāyonisambandhebhyaḥ | Ṣaṣṭhī kept after an ṛ-final relation word (both members relation words). | hotuḥputraḥ; not hotṛdhanam |
| 982 | vibhāṣā svasṛpatyoḥ | Ṣaṣṭhī optionally kept before svasṛ, pati. | mātuḥsvasā / mātṛsvasā → finally mātuḥṣvasā or mātuḥsvasā (SK 983) / mātṛṣvasā (ṣ obligatory, SK 984) |
| 967 | gaviyudhibhyāṃ sthiraḥ | s of sthira → ṣ after gavi, yudhi (vārttika: hṛdi, divi kept). | gaviṣṭhiraḥ, yudhiṣṭhiraḥ; hṛdispṛk |
| 983 | mātuḥpiturbhyām anyatarasyām | s of svasṛ optionally → ṣ after mātuḥ, pituḥ, in a compound. | mātuḥṣvasā / mātuḥsvasā |
| 984 | mātṛpitṛbhyāṃ svasā | s of svasṛ → ṣ after mātṛ, pitṛ, in a compound. | mātṛṣvasā; sentence mātuḥ svasā |