10. Ajanta-napuṃsakaliṅga-prakaraṇa — Neuter nouns whose stem ends in a vowel

After this chapter you will be able to decline any neuter noun whose stem ends in a vowel: jñāna "knowledge", vāri "water", madhu "honey", dadhi "curd", dhātṛ "the sustainer", and the rarer stems in ā, ī, ū, o, ai and au. You will see that the neuter borrows almost everything from the masculine declension of chapter 8 and changes only a few things: the nominative and accusative endings are dropped or recast, a nasal augment num is slipped in before many endings, long stem-vowels are first shortened, and a handful of pronouns and irregular stems get special substitutes. Fifteen rules do all of this, and each is shown at work on a concrete word.

Before you start

• Chapter 1 (Saṃjñā-prakaraṇa) is assumed: you can read a marker letter (it), unpack a pratyāhāra such as ac, ik, ec, hal and jhal, and you know what guṇa, vṛddhi, hrasva and dīrgha mean.
• Chapter 2 (Paribhāṣā-prakaraṇa) is assumed: a substitute replaces a sthānin (the original element); a marker on a substitute or affix tells the grammar how it is applied.
• Chapter 3 (Ac-sandhi) is assumed for the vowel joinings used inside the derivations, above all a + ī → e and the single-vowel substitute called pūrvarūpa.
• Chapter 8 (Ajanta-puṃliṅga-prakaraṇa, the masculine declension) is assumed above all. From it you know the twenty-one case-endings su, au, jas; am, auṭ, śas; ṭā, bhyām, bhis; ṅe, bhyām, bhyas; ṅasi, bhyām, bhyas; ṅas, os, ām; ṅi, os, sup; the paradigms of rāma, hari, kroṣṭṛ and rai; the names aṅga, pada, bha, sarvanāmasthāna and ghi; the lengthening rules supi ca and nāmi; and the change of n to ṇ (ṇatva). Every "the rest like rāma" or "like hari" in this chapter sends you back there.
• Chapter 9 (the feminine declension) is not needed, except that the ending auṄ (the nominative and accusative dual) was there replaced by śī after a feminine ā-stem; the same replacement returns here for neuters.

The Sanskrit word for "neuter" is napuṃsaka; Bhaṭṭoji also says klība. A stem "ending in a vowel" is ajanta (ac "vowel" + anta "end").

A. The nominative and accusative singular: su and am vanish by default, become am after an a-stem — running example: vāri, jñāna → jñānam

In the masculine, rāma + su gave rāmaḥ and rāma + am gave rāmam. The neuter treats these two endings differently. The general rule (SK 319) simply deletes them, so that vāri "water" is both nominative and accusative singular. The special rule for a-stems (SK 309) replaces both endings by am, so that jñāna gives jñānam in both cases. Bhaṭṭoji prints SK 319 later, after the ā-stems, but it is taught here first, because SK 309 is exactly the exception to it.

SK 319 · स्वमोर्नपुंसकात् · svamor napuṃsakāt (A 7.1.23)

Rule. After a neuter aṅga (a stem that an ending is being attached to), the endings su (nominative singular) and am (accusative singular) undergo luk. Luk is the deletion of a whole affix, leaving nothing; the word luk is carried into this rule by anuvṛtti from the earlier rule ṣaḍbhyo luk (A 7.1.22).

Example.

vāri + su → su is deleted by luk → vāri (nominative singular).

vāri + am → am is deleted by luk → vāri (accusative singular).

Applies when / blocked by. Blocked after an a-stem by SK 309, and after the five pronouns of SK 315. Bhaṭṭoji's commentators make one point explicit: luk deletes the whole of am, not just its first sound. The paribhāṣā ādeḥ parasya (A 1.1.54), which normally lets a substitute replace only the first sound of a following element, does not apply, because luk is by definition the elision of the entire affix; it is a sarvādeśa, a substitute for the whole.

SK 309 · अतोऽम् · ato'm (A 7.1.24)

Rule. After a neuter aṅga ending in short a, the endings su and am are both replaced by am. This is an apavāda (an exception) to SK 319: where SK 319 would delete the ending, SK 309 keeps an am instead. Replacing am by am is not idle; it is exactly what saves the accusative ending from luk.

Example.

jñāna + su → su → am → jñāna + am

jñāna + am → a + a → the single vowel a by ami pūrvaḥ (SK 194; the pūrvarūpa, "the shape of the former vowel") → jñānam (nominative singular).

jñāna + am (accusative) → the same steps → jñānam (accusative singular).

Vocative singular: he jñāna + su → su → am → pūrvarūpa → he jñānam → by eṅ hrasvāt (SK 193) the consonant of the vocative ending is dropped → he jñāna.

Applies when / blocked by. Only after a stem in short a. "After" is read by tadanta-vidhi: "after a" means "after a stem ending in a", because the heading aṅgasya is here turned into an ablative and qualified by ataḥ.

Watch out. Why is the substitute am and not just m? Bhaṭṭoji's commentators explain: if a bare m replaced su, that m would count as a consonant-initial case-ending, and supi ca (A 7.3.102) would wrongly lengthen the stem to jñānām. So the substitute is the full syllable am, which then merges with the stem-final a by pūrvarūpa. In the vocative, what eṅ hrasvāt removes is only the m left over after pūrvarūpa; scholars discuss exactly how that residue still counts as the vocative ending.

B. The nominative and accusative dual: auṄ becomes śī (that is, ī) — running example: jñāna → jñāne

The dual ending is auṄ (the marker Ṅ is dropped; the ending is au). After a neuter it is replaced by śī, whose marker ś is dropped, leaving ī. Because ī is a vowel-initial ending and not a sarvanāmasthāna, the stem before it is a bha. That status would normally trigger a deletion of the stem-final a (SK 311), but a vārttika stops it, and a + ī simply join as e: jñāne.

SK 310 · नपुंसकाच्च · napuṃsakāc ca (A 7.1.19)

Rule. After a neuter stem too, the ending auṄ is replaced by śī. Both words, auṄ (the sthānin) and śī (the substitute), are carried in by anuvṛtti from the preceding rules; the ca "also" extends to neuters the replacement that the feminine ā-stems already have. auṄ is the grammar's name for the au-ending of the nominative and accusative dual.

Example.

jñāna + auṄ → auṄ → śī; the markers Ṅ and ś are dropped → jñāna + ī.

The stem is now a bha: ī is a vowel-initial ending and not a sarvanāmasthāna. What happens next is decided by SK 311.

SK 311 · यस्येति च · yasyeti ca (A 6.4.148)

Rule. The final i-vowel or a-vowel of a bha stem is deleted when an ī or a taddhita affix (a secondary affix that forms a noun from a noun) follows. The word ya in the sūtra is a compound of i and a; bhasya "of a bha" is the governing heading (adhikāra); taddhite is carried in by anuvṛtti. Read in the present derivation, this rule threatens to delete the a of jñāna before ī.

Example.

jñāna + ī → SK 311 would delete the final a of the bha stem → (blocked; see below)

Vārttika: "a prohibition should be stated before the śī that replaces auṅ". So the a stays.

jñāna + ī → a + ī → e by ād guṇaḥ (A 6.1.87) → jñāne (nominative and accusative dual).

Applies when / blocked by. The deletion is blocked by the vārttika auṅaḥ śyāṃ pratiṣedhaḥ whenever the ī is the śī substituted for auṅ. Bhaṭṭoji's commentators note that the vārttika is worded in the feminine (śyām), showing that the ending śī is treated as invariably feminine; scholars discuss whether the word auṅ in it is strictly needed.

C. The nominative and accusative plural: jas and śas become śi, and num enters — running example: jñāna → jñānāni

In the masculine, jas gave rāmāḥ and śas gave rāmān. The neuter replaces both plural endings by śi (that is, i), declares that śi counts as a sarvanāmasthāna, and then, because a sarvanāmasthāna follows, inserts the augment num into the stem and lengthens the vowel before it. The result is the familiar jñānāni.

SK 312 · जश्शसोः शिः · jaśśasoḥ śiḥ (A 7.1.20)

Rule. After a neuter stem, the endings jas (nominative plural) and śas (accusative plural) are both replaced by śi. The marker ś is dropped; what remains is the short vowel i. The word klībāt "after a neuter" is carried in by anuvṛtti.

Example.

jñāna + jas → jas → śi → jñāna + i.

jñāna + śas → śas → śi → jñāna + i.

What the stem now does before this i depends on the next two rules.

SK 313 · शि सर्वनामस्थानम् · śi sarvanāmasthānam (A 1.1.42)

Rule. The affix śi has the name sarvanāmasthāna. This is a saṃjñā rule (a naming rule). You met the sarvanāmasthāna in chapter 8: the first five endings su, au, jas, am, auṭ, before which a stem is not a bha and is not a pada, and before which several strengthening rules fire. That earlier naming rule, suḍ anapuṃsakasya (A 1.1.43), explicitly excludes neuters, so without SK 313 the neuter śi would not have the name at all.

Example.

jñāna + i (the i is śi) → śi is a sarvanāmasthāna → the stem is not a bha, and rules conditioned by "a following sarvanāmasthāna" may now apply (SK 314, SK 317).

SK 314 · नपुंसकस्य झलचः · napuṃsakasya jhalacaḥ (A 7.1.72)

Rule. A neuter aṅga ending in a jhal consonant or in a vowel (ac) takes the augment num when a sarvanāmasthāna follows. An augment (āgama) is a sound added to a form, not a substitute for anything. Num is n with the markers u and m; the word num is carried in by anuvṛtti from idito num (A 7.1.58). Because it is marked with m (it is mit), num is inserted after the last vowel of the stem. Jhalac is a compound of jhal and ac; by tadanta-vidhi the rule applies to a stem ending in either.

Example.

jñāna + i (śi, a sarvanāmasthāna by SK 313; the stem ends in a vowel) → num is inserted after the last vowel → jñānan + i

jñānan + i → the penultimate vowel (upadhā, the vowel before the final consonant n) is lengthened by sarvanāmasthāne ca (A 6.4.8) → jñānān + i → jñānāni (nominative plural).

The accusative plural is the same: jñānāni.

Applies when / blocked by. Only before a sarvanāmasthāna; before other vowel-initial endings SK 320 takes over for ik-stems (i, u, ṛ).

The paradigm of jñāna is now complete in its neuter-specific part. "The rest is like rāma", Bhaṭṭoji says: the instrumental onward are jñānena, jñānābhyām and so on, exactly as in chapter 8. So too dhana "wealth", vana "forest", phala "fruit".

D. The five pronominal neuters: su and am become ad, and the stem loses its ṭi — running example: katara → katarat

Five pronouns — katara "which of two", katama "which of many", anya "other", anyatara "either", itara "the other" — do not take am in the neuter nominative and accusative singular. They take a substitute adḍ, marked with ḍ, and that marker causes the stem's final vowel to be cut off: katara + ad → katar + ad → katarat.

SK 315 · अद्ड् डतरादिभ्यः पञ्चभ्यः · adḍ ḍatarādibhyaḥ pañcabhyaḥ (A 7.1.25)

Rule. After the five neuter pronouns ḍatara and the rest, the endings su and am are replaced by adḍ. The substitute is ad; the ḍ is a marker, so the substitute is ḍit, "marked with ḍ", and that marker matters for SK 316. The five are ḍatara, ḍatama, anya, anyatara, itara, read from the sarvādi list (the list of pronoun stems beginning with sarva); ḍatara and ḍatama are really the affixes ḍatarac and ḍatamac, so words ending in them (such as katara, katama) are included by tadanta-vidhi. The words su and am come by anuvṛtti from SK 319.

Example.

katara + su → su → adḍ; the marker ḍ is dropped → katara + ad.

katara + am → likewise → katara + ad.

Applies when / blocked by. The vārttika ekatarāt pratiṣedhaḥ (given under SK 316) excludes ekatara "one of two".

Watch out. The sūtra writes adḍ exactly as it stands, though two ordinary changes are due. The d before the marker ḍ would become ḍ (ṣṭutva, the change of a dental to the matching retroflex), and that final ḍ would then be dropped as the last consonant of a cluster (saṃyogānta-lopa). Bhaṭṭoji's commentators, following Kaiyaṭa, note that both are deliberately left undone, so that the substitute ad and its marker ḍ can be read off.

SK 316 · टेः · ṭeḥ (A 6.4.143)

Rule. Before a ḍit affix, the ṭi of a bha stem is deleted. The ṭi of a form is its last vowel together with whatever follows that vowel; in katara the ṭi is the final a. The stem is a bha because ad is a vowel-initial ending that is not a sarvanāmasthāna. The final d is optionally written t at the end of an utterance.

Example.

katara + ad (ad is ḍit) → the ṭi, the final a, is deleted → katar + ad → katarad, or at a pause katarat (nominative and accusative singular).

Dual: katare; plural: katarāṇi — these follow SK 310 and SK 312–314 like jñāna.

Vocative: he katarat. Because the ṭi is gone, neither the lengthening of prathamayoḥ (SK 164) nor the vocative deletion of eṅ hrasvāt (SK 193) has anything to act on.

The accusative is the same; the rest of the paradigm is like the masculine.

Applies when / blocked by. Only for a bha stem: the counter-example is pañcamaḥ, whose stem is not a bha before its ending and so keeps its ṭi. Likewise katamat, anyat, anyatarat, itarat. But anyatama is not one of the five, so its neuter is simply anyatamam.

Vārttika (SK 316.k2): "a prohibition should be stated after ekatara." So ekatara takes the ordinary am of SK 309: ekataram.

Watch out. Here Bhaṭṭoji begins the story of ajara "unageing", a neuter a-stem that can take the substitute jaras (from chapter 8). In the singular, once su has become am by SK 309, the jaras substitution does not happen: by the saṃnipāta-paribhāṣā (an operation that depends on a given combination may not destroy that combination), the am that arose in dependence on the a-final stem cannot now trigger a change of that stem. So ajaram. In the dual, jaras wins by paratva (the later rule prevails): ajarasī beside ajare, and since ajaras now ends in a jhal, it takes num by SK 314 before the plural śi — which is where the next rule comes in.

E. Lengthening before the n of an s-final cluster, and the irregular substitutes hṛd, udan, āsan, māṃs — running example: ajaras → ajarāṃsi

Once num has entered ajaras the stem ends in the cluster ns. A special rule lengthens the vowel before that n, giving ajarāṃsi. The same rule serves mahat "great". Bhaṭṭoji then closes the a-stems by listing four neuter nouns that swap in a shorter stem from the accusative plural onward.

SK 317 · सान्तमहतः संयोगस्य · sāntamahataḥ saṃyogasya (A 6.4.10)

Rule. For the n that belongs to a consonant cluster ending in s, or to mahat, the vowel immediately before that n is lengthened when a sarvanāmasthāna follows, unless that sarvanāmasthāna is the vocative singular. Three words are carried in by anuvṛtti: na "of the n", dīrgha "long" (from ḍhralope pūrvasya, A 6.3.111) and the condition sarvanāmasthāne cāsaṃbuddhau (A 6.4.8). Saṃyogasya and mahataḥ are partitive genitives: "the n that is part of the cluster / of mahat."

Example.

ajara + śi → jaras replaces the stem-final → ajaras + i → num by SK 314 (the stem ends in a jhal) → ajarans + i

ajarans + i → the vowel before the n of the s-final cluster is lengthened → ajarāns + i

ajarāns + i → n → anusvāra by naś cāpadāntasya (A 8.3.24) → ajarāṃsi (nominative and accusative plural); beside it, without jaras, ajarāṇi.

Accusative singular: ajara + am → luk is due by SK 319; its exception am by SK 309 is due; but jaras is later than both and wins by paratva → ajaras + am → ajarasam. And now the am is not deleted: jaras arose in dependence on the combination with am, so by the saṃnipāta-paribhāṣā it cannot cause the luk of that very am. Beside it, without jaras, ajaram.

The full paradigm, each cell "with jaras / without":

The oblique cases are "like the masculine" of chapter 8.

Watch out. The word upadhā in this rule means simply "the vowel just before", Bhaṭṭoji's commentators observe, not the technical penultimate-of-the-word.

The irregular neuters. By paddanno… (SK 228) four neuter a-stems replace their stem, from the accusative plural (śas) onward, with a shorter form: hṛdaya "heart" → hṛd, udaka "water" → udan, āsya "mouth" → āsan, and by the vārttika on that rule māṃsa "flesh" → māṃs. In the nominative series (su to auṭ) each is regular, like jñāna; the substitution is optional, and without it the whole word is like jñāna.

How the forms arise, as Bhaṭṭoji's commentators trace them:

hṛdaya + śas → śi → hṛd substituted → num after the ṛ by SK 314 → n → anusvāra by naś cāpadāntasya, then assimilated to the following d → hṛndi.

udaka + śi → udan → the vowel before the final n is lengthened by sarvanāmasthāne ca → udāni; here allopo'naḥ (SK 234, the deletion of the a of a final an in a bha) does not apply, because śi is a sarvanāmasthāna.

udaka + ā (instrumental) → udan → the a of an is deleted by SK 234 → udnā.

udaka + bhyām → udan → before bhyām the stem is a pada, so its final n is deleted by na lopaḥ prātipadikāntasya (A 8.2.7) → udabhyām.

Locative singular udani / udni: the a-deletion is optional before ṅi (by vibhāṣā ṅiśyoḥ, A 6.4.136).

māṃsa + bhyām → māṃs → the s is dropped as the last member of a cluster (saṃyogānta-lopa) → the cause of the anusvāra is gone, so the nasal reverts to n → mānbhyām; the n is not deleted, because the cluster-final deletion is asiddha (treated as not having happened) for that rule.

Watch out. Bhaṭṭoji adds that the list in paddanno… is really "of this type", so that hṛd and the others may be used even in the nominative series, as in the Bhāṣya's māṃspacanī "a pot for cooking flesh". Scholars discuss how far these substitutes belong to the classical language and not only to the Veda; Bhaṭṭoji's conclusion is that, following usage, pad and the rest may be employed in the classical language too. This closes the a-stems: ity adantāḥ.

F. Long-vowel stems are shortened first — running example: śrīpā → śrīpam

A neuter whose stem ends in a long vowel does not get a separate paradigm. One rule shortens the vowel, and the stem then declines like the short-vowel neuter it has become: śrīpā "protector of prosperity" becomes śrīpa and goes like jñāna; later sudhī becomes sudhi and sulū becomes sulu, both going like vāri.

SK 318 · ह्रस्वो नपुंसके प्रातिपदिकस्य · hrasvo napuṃsake prātipadikasya (A 1.2.47)

Rule. In the neuter, a prātipadika (a nominal stem: a meaningful item that is neither a root nor an affix) ending in a vowel is made short (hrasva). Only a vowel can be shortened, so the rule is read of a stem ending in a vowel (tadanta-vidhi).

Example.

śrīpā (neuter; formed from śrī + the root pā "protect" with a zero-affix, viṭ or kvip — an affix that adds no sound of its own — like viśvapā) → final ā → a → śrīpa

śrīpa + su → am by SK 309 → pūrvarūpa → śrīpam (nominative and accusative singular).

Plural: śrīpa + śi → num, lengthening → śrīpāṇi (the n becomes ṇ after the r of śrī, even across the two members, by ekāj uttarapade ṇaḥ, A 8.4.12).

Instrumental: śrīpeṇa. Dative: śrīpa + ṅe → ya for ṅe → a lengthened by supi ca (A 7.3.102) → śrīpāya.

The rest is like jñāna.

Applies when / blocked by. Bhaṭṭoji's commentators note that the rule does not apply to something that is a prātipadika only by courtesy, such as kul- + e in kule, where the final e is a sandhi vowel counted as the stem's end; the repetition of the word prātipadika in the sūtra signals this.

Watch out. In śrīpāya the stem ends in ā again, and pā is still a root, so āto dhātoḥ (SK 240), which deletes the final ā of a root-stem before a vowel-initial ending, seems due. It does not apply, by the saṃnipāta-paribhāṣā: the substitute ya arose in dependence on the a-final stem and cannot be the cause of deleting that very vowel. Scholars discuss the exact reasoning; the form śrīpāya is not in doubt. This closes the ā-stems: ity ādantāḥ.

G. Stems in i, ī, u, ū, ṛ: num before every vowel-initial ending, the masculine option, and the anaṅ stems — running example: vāri, dadhi

The stem vāri "water" shows the whole pattern. Its nominative and accusative singular is bare vāri (SK 319). Before every vowel-initial ending, num is inserted (SK 320): vāriṇī, vārīṇi, vāriṇā, vāriṇe. Before consonant-initial endings it is simply like hari. Two further rules refine this. Stems that mean the same thing in the masculine may optionally take the masculine forms instead (SK 321): anādaye beside anādine. And four stems — asthi, dadhi, sakthi, akṣi — replace their final i by an instead of taking num (SK 322): dadhnā. The same rules then run sudhi, madhu, sulu and dhātṛ.

SK 320 · इकोऽचि विभक्तौ · iko'ci vibhaktau (A 7.1.73)

Rule. A neuter aṅga ending in an ik vowel (i, u, ṛ, ḷ) takes the augment num when a vowel-initial case-ending (vibhakti) follows. Num is carried in by anuvṛtti from idito num dhātoḥ (A 7.1.58), and napuṃsakasya from SK 314. Being mit, num goes after the last vowel of the stem.

Example.

vāri + auṄ → śī by SK 310 → vāri + ī → num → vārin + ī → n → ṇ after the r (the change carries across the num, by aṭkupvāṅnumvyavāye 'pi) → vāriṇī (nominative and accusative dual).

vāri + jas → śi by SK 312 → vāri + i → num → vārin + i → the vowel before n lengthened by sarvanāmasthāne ca → ṇatva → vārīṇi (nominative and accusative plural).

Instrumental: vāri + ṭā → ṭā → nā by āṅo nā (SK 244) → num → ṇatva → vāriṇā.

Dative: vāri + ṅe → num → vāriṇe; ablative and genitive vāriṇaḥ; genitive and locative dual vāriṇoḥ; locative vāriṇi.

Genitive plural: vāri + ām → not num but the augment nuṭ (the n that begins the ending, giving nām) → the stem-vowel is lengthened before nām by nāmi (SK 209) → vārīṇām.

Before a consonant-initial ending: like hari (vāribhyām, vāribhiḥ, vāriṣu …).

Applies when / blocked by. Only before a vibhakti, a case-ending. Bhaṭṭoji's commentators give the reason: madhu "wine" forms mādhava "belonging to wine" with the taddhita affix aṇ, which is vowel-initial; had the rule said merely "before a vowel", num would enter there and give a wrong form.

Vārttika (SK 320.k2): "num prevails over vṛddhi, the substitution of au (auttva), the treatment as a tṛ-stem (tṛjvadbhāva) and guṇa, by pūrvavipratiṣedha." Ordinarily, when two rules conflict, the later one in the Aṣṭādhyāyī wins (paratva). Pūrvavipratiṣedha is the reverse: the earlier rule is made to win. So in the dative vāri + ṅe, the guṇa of gher ṅiti (SK 245) is later and would give vāraye; but num is made to prevail → vāriṇe. Likewise in the ablative, genitive and dual against guṇa and yaṇ, and in the locative against the au of acca gheḥ. Bhaṭṭoji's commentators list where each loser still has its own field: vṛddhi in gāvau, auttva in harau, tṛjvadbhāva in kroṣṭrā, guṇa in haraye.

Watch out. Two points on the paradigm.

First, the genitive plural. Num is applicable here too, and is later than nuṭ; but the form taken is with nuṭ. The difference: num becomes part of the stem, so a stem in -in would not be "vowel-final" and nāmi could not lengthen it; with nuṭ the stem stays vāri and is lengthened to vārī. Hence vārīṇām.

Second, the vocative. he vāri + su → su deleted by SK 319. Ordinarily a deleted (luk) affix leaves no trace for stem-operations — this is the prohibition na lumatāṅgasya (SK 263). But Bhaṭṭoji treats that prohibition as anitya (not binding in every case), so in one alternative the vocative guṇa of hrasvasya guṇaḥ (A 7.3.108) does apply: he vāre beside he vāri. The word ac in the present sūtra is taken as the hint (jñāpaka) that the prohibition is not binding.

SK 321 · तृतीयादिषु भाषितपुंस्कं पुंवद्गालवस्य · tṛtīyādiṣu bhāṣitapuṃskaṃ puṃvad gālavasya (A 7.1.74)

Rule. From the instrumental (the third case, tṛtīyā) onward, before a vowel-initial ending, a neuter ik-stem that is bhāṣitapuṃska behaves as masculine (puṃvat) — in the opinion of Gālava, which makes the option free. A word is bhāṣitapuṃska when it is also used in the masculine with the same pravṛtti-nimitta, the ground on which the word is applied to a thing (the property, such as pot-ness for ghaṭa, that makes the word fit). "Behaving as masculine" means: no shortening by SK 318 and no num by SK 320. Iko'ci and napuṃsakasya come by anuvṛtti from SK 320 and SK 314.

Example.

anādi "beginningless" (masculine, feminine or neuter alike, always on the ground of having no origin — so it is bhāṣitapuṃska):

dative, masculine option: anādi + ṅe → no num → guṇa by gher ṅiti (SK 245) → anādaye.

dative, neuter option: anādi + ṅe → num → anādine.

So too ablative and genitive anādeḥ / anādinaḥ; genitive and locative dual anādyoḥ / anādinoḥ. In the genitive plural only anādīnām, the two options giving the same form. The nominative and accusative, and all consonant-initial endings, are like vāri.

Applies when / blocked by. Only with identity of ground. Bhaṭṭoji's own contrast: pīlu the tree is masculine; its fruit, also called pīlu (the affix aṇ that would form the fruit-name is deleted by phale luk, A 4.3.163), is neuter. The tree-word is applied on the ground of being that kind of tree, the fruit-word on the ground of being that tree's fruit. The grounds differ, so the fruit takes no masculine forms: dative pīlune, never pīlave. A verse Bhaṭṭoji's commentators quote sums it up: pīlu the tree, pīlu the fruit; pīlune, not pīlave.

SK 322 · अस्थिदधिसक्थ्यक्ष्णामनङुदात्तः · asthidadhisakthyakṣṇām anaṅ udāttaḥ (A 7.1.75)

Rule. For asthi "bone", dadhi "curd", sakthi "thigh" and akṣi "eye", the substitute anaṅ replaces the final vowel before a vowel-initial ending from the instrumental onward, and it is udātta. In anaṅ the ṅ is a marker and the last a is for pronunciation; the substitute is an. Because it is ṅit, it replaces only the final sound of the stem (by ṅic ca, A 1.1.53). This is an apavāda to the num of SK 320. Once the stem ends in an and is a bha, allopo'naḥ (SK 234) deletes the a of an.

Example.

dadhi + ṭā → final i → an → dadhan + ā → the a of an deleted by SK 234 → dadhnā (instrumental singular).

Dative dadhne; ablative dadhnaḥ; genitive dadhnaḥ; genitive and locative dual dadhnoḥ; locative singular dadhni or dadhani, the a-deletion being optional before ṅi by vibhāṣā ṅiśyoḥ (A 6.4.136). The rest is like vāri: dadhi, dadhinī, dadhīni. So too asthi, sakthi, akṣi.

Applies when / blocked by. Also for a compound ending in one of the four (tadanta), because the whole compound is the aṅga: atidadhi "that which has surpassed curd" → instrumental atidadhnā, even when the compound is masculine (atidadhnā brāhmaṇena). But a masculine dadhi (root dhā + affix ki) takes no anaṅ, because the rule's napuṃsakasya requires asthi and the rest to be neuter; its instrumental is dadhinā (nā for ṭā, SK 244).

Watch out. Anaṅ does not touch the nominative and accusative, which are not "from the instrumental onward".

This closes the i-stems. Bhaṭṭoji now runs the remaining vowel-stems through the same three rules — SK 320 (num), SK 321 (masculine option), SK 318 (shortening) — and the tṛjvadbhāva of chapter 8.

ī-stems. sudhī "of good mind" is shortened by SK 318 to sudhi and goes like vāri: sudhi, sudhinī (num wins over the iyaṅ substitute by paratva), sudhīni, vocative he sudhe / he sudhi. It is bhāṣitapuṃska (the ground, "thinking well" or "having good understanding", is the same in every gender), so instrumental sudhiyā (masculine option) or sudhinā (num). pradhī likewise: pradhyā (masculine option; here yaṇ applies by er anekācaḥ, A 6.4.82, since the exception na bhūsudhiyoḥ names only bhū and sudhī) or pradhinā.

u-stems. madhu "honey": madhu, madhunī, madhūni, vocative he madho / he madhu. So ambu "water" and the like. madhu has no masculine option: as a neuter it means liquor or flower-nectar, as a masculine the spring or the month Caitra — different grounds; and madhu "wine of the grape" is neuter only. sānu "ridge" optionally takes the substitute snu from śas on (by the vārttika māṃsapṛtanāsānūnām… on SK 228): plural snūni / sānūni; being masculine-or-neuter with one ground, it does have the masculine option.

priyakroṣṭu "whose jackal is dear" (from kroṣṭu "jackal", which in chapter 8 may be treated like the tṛ-stem kroṣṭṛ, tṛjvadbhāva, before a sarvanāmasthāna): priyakroṣṭu, dual priyakroṣṭunī. Plural: before śi, a sarvanāmasthāna, tṛjvadbhāva is due and is later; but by the vārttika of SK 320 num prevails by pūrvavipratiṣedha → priyakroṣṭūni. From the instrumental on, three forms can arise:

In the masculine-plus-tṛ column there is no num, because the word is being treated as non-neuter. Genitive plural: priyakroṣṭūnām, with nuṭ. Before consonant-initial endings: like madhu. Bhaṭṭoji's commentators add why pūrvavipratiṣedha, and not merely the obligatory nature of num, is invoked: what is set aside in such a conflict stays set aside, so tṛjvadbhāva cannot return once num has won.

ū-stems. sulū "one who cuts well" (root lū + kvip) is shortened to sulu: sulu, sulunī (num wins by paratva over the yaṇ of oḥ supi, A 6.4.83), sulūni. Instrumental: masculine option sulvā (no shortening, no num, so the ū becomes v by oḥ supi) or neuter sulunā. Dative sulve / sulune, and so on — two forms in each vowel-initial case.

ṛ-stems. dhātṛ "the sustainer": dhātṛ, dual dhātṛṇī (num, then n → ṇ after ṛ), plural dhātṝṇi (num, lengthening by sarvanāmasthāne ca, ṇatva). The anaṅ of the masculine dhātā does not appear, because su has been deleted by luk and na lumatāṅgasya blocks it; yet, that prohibition being anitya, the vocative has he dhātaḥ (with guṇa) beside he dhātṛ. It is bhāṣitapuṃska (the ground, "being one who sustains", is the same), so:

Genitive plural dhātṝṇām with nuṭ. So too jñātṛ, kartṛ and the rest.

H. Stems in e, o, ai, au: the shortening gives i or u, never a — running example: pradyo → pradyu

A neuter stem ending in an ec vowel is shortened by SK 318, but e, o, ai, au have no short forms of their own. One paribhāṣā decides what "short o" means: the ik vowel nearest it, u. So pradyo becomes pradyu, prarai becomes prari, sunau becomes sunu, and all three then decline like vāri.

SK 323 · एच इग्घ्रस्वादेशे · eca ig ghrasvādeśe (A 1.1.48)

Rule. When a short vowel is prescribed as substitute for an ec vowel (e, o, ai, au), the substitute is an ik vowel (i or u), and never a. This is a paribhāṣā of the restrictive kind (niyama): it does not prescribe a new substitute but narrows the choice. Bhaṭṭoji's commentators explain why a was a live danger: the first part of each ec vowel resembles a and the second part resembles i or u, so without this rule a would be as close a match as i or u. With a excluded, the pairing follows by itself: e and ai → i; o and au → u. No rule of one-to-one correspondence is needed.

Example.

pradyo "whose sky is exalted" (neuter) → SK 318 demands a short vowel for o → by SK 323 the substitute is u → pradyu

pradyu then goes like vāri: pradyu, dual pradyunī, plural pradyūni, instrumental pradyunā, and so on.

prarai "whose wealth is exalted" (neuter) → ai → i → prari; prariṇī, prarīṇi, prariṇā.

sunau "having a good ship" (neuter) → au → u → sunu; sununī, sunūni, sununā, dative sunune.

For an e-stem Bhaṭṭoji's commentators suggest smṛte or susmṛti, to be worked out the same way.

Applies when / blocked by. No masculine option (SK 321) for pradyu: the masculine word is pradyo, ending in o, while the neuter word is pradyu, ending in u; since the ik-stem is a different word, it is not bhāṣitapuṃska. Scholars discuss this, some holding that a stem altered in one sound is still the same word and so does have the option. The same applies to prari and sunu.

Watch out. prari keeps one feature of rai. Before a consonant-initial ending, rāyo hali (from chapter 8) turns the final of rai into ā, and it applies here too, because a thing altered in one part is treated as the same thing (ekadeśavikṛtam ananyavat): prarābhyām, prarābhiḥ. In the nominative singular there is no consonant-initial ending, su having been deleted, so no ā: prari. In the genitive plural the augment is nuṭ (prevailing over num by pūrvavipratiṣedha), and then not the ā of rāyo hali but the lengthening of nāmi (SK 209): prarīṇām. The reason is the saṃnipāta-paribhāṣā: nuṭ came in because the stem ended in a short vowel and cannot be the cause of a change that removes that short ending; nāmi, on the other hand, was framed precisely to lengthen before nām, and that purpose overrides the paribhāṣā, as was said under rāma in chapter 8. Scholars discuss this; another view gives prarāṇām.

This closes the chapter: ity ajantā napuṃsakaliṅgāḥ, "so far the vowel-stem neuters".

Terms used in this chapter

• napuṃsaka / klība — neuter; the gender this chapter declines.
• ajanta — ending in a vowel (ac + anta).
• aṅga — the stem to which an affix is being attached; the base of every rule here.
• su, am, auṄ, jas, śas — the nominative singular, accusative singular, nominative-accusative dual, nominative plural and accusative plural endings; the markers Ṅ and the like are dropped.
• vibhakti — a case-ending; the twenty-one sup endings of chapter 8.
• luk — deletion of an entire affix, leaving nothing behind (SK 319).
• anuvṛtti — carrying a word forward from an earlier sūtra into a later one that does not repeat it.
• sarvādeśa — a substitute for the whole of an element, not just its first or last sound.
• apavāda — an exception: a narrower rule that overrides a wider one in its own field (SK 309 to SK 319).
• pūrvarūpa — the single vowel, shaped like the former one, that replaces a + a before am (SK 194).
• saṃbuddhi — the vocative singular ending.
• hal-mātra-lopa — deletion of the bare consonant of the vocative ending (SK 193).
• tadanta-vidhi — the reading by which "after a" means "after a stem ending in a".
• sthānin / ādeśa — the original element / the substitute that replaces it.
• śī, śi — the substitutes ī (for the dual auṄ) and i (for the plurals jas and śas); their ś is a marker.
• bha — the name a stem bears before a vowel-initial or y-initial ending that is not a sarvanāmasthāna; several deletions apply only to a bha.
• sarvanāmasthāna — the name of the endings su, au, jas, am, auṭ (in the non-neuter) and of śi (SK 313); before them a stem is strengthened, not weakened.
• taddhita — a secondary affix forming a noun from a noun (mentioned in SK 311 and SK 320).
• adhikāra — a heading rule whose wording is read into the rules that follow it.
• vārttika — a supplementary statement on a sūtra, correcting or completing it.
• saṃjñā — a technical name given by a rule (SK 313 gives one).
• āgama — an augment: a sound added to a form without replacing anything.
• num — the augment n (markers u, m), inserted after the last vowel of the stem (SK 314, SK 320).
• mit — marked with m; an augment so marked goes after the last vowel.
• jhal — the pratyāhāra of all stops and spirants (the non-nasal, non-semivowel consonants).
• upadhā — the vowel immediately before a final consonant; the penultimate sound.
• ḍit — marked with ḍ; a ḍit affix causes the deletion of the preceding ṭi (SK 315–316).
• sarvādi — the list of pronoun stems beginning with sarva, from which the five of SK 315 are read.
• ṣṭutva — the change of a dental to the matching retroflex, as d → ḍ before the marker ḍ of adḍ (SK 315).
• ṭi — the last vowel of a form together with whatever follows it.
• avasāna — a pause; the end of an utterance.
• paratva — the principle that, of two conflicting rules, the one later in the Aṣṭādhyāyī prevails.
• saṃnipāta-paribhāṣā — an operation that came about through a given combination may not then destroy that combination.
• jaras — the optional substitute for the stem jarā/jara before vowel-initial endings (chapter 8).
• saṃyoga — a cluster of consonants with no vowel between.
• anusvāra — the pure nasal written ṃ, here replacing n before s.
• saṃyogānta-lopa — deletion of the last consonant of a word-final cluster.
• asiddha — treated as not having taken effect with respect to an earlier rule.
• pada — a finished word; a stem is a pada before endings beginning with bhyām and the like.
• prātipadika — a nominal stem: a meaningful item that is neither a root nor an affix (SK 318).
• viṭ / kvip — zero-affixes: affixes that add no sound of their own, after which a root such as pā stands as a noun-stem (śrīpā, viśvapā, sulū; SK 318).
• hrasva / dīrgha — short / long vowel.
• ṇatva — the change of n to ṇ after r, ṛ or ṣ, even across num.
• ik — the pratyāhāra i, u, ṛ, ḷ.
• nuṭ — the augment n that begins the genitive plural ending, giving nām.
• pratyayalakṣaṇa — treating a deleted affix as still present for the sake of a rule; na lumatāṅgasya (SK 263) forbids this after luk.
• anitya — not binding in every case; said of a rule that is sometimes ignored.
• jñāpaka — a hint: a word in a sūtra whose only purpose is to reveal some principle.
• pūrvavipratiṣedha — the reverse of paratva: the earlier of two conflicting rules is made to prevail (SK 320.k2).
• auttva — the substitution of au for a final i or u before the locative ṅi.
• tṛjvadbhāva — the treatment of kroṣṭu as if it were the tṛ-stem kroṣṭṛ (chapter 8).
• yaṇ — the semivowels y, v, r, l as substitutes for i, u, ṛ, ḷ before a vowel.
• iyaṅ — the substitute iy for a final ī before a vowel-initial ending.
• bhāṣitapuṃska — a word that is also used in the masculine with the same ground of application (SK 321).
• pravṛtti-nimitta — the ground of application: the property that makes a word fit the thing it names.
• puṃvat — "as if masculine": no shortening, no num.
• ghi — the name of a short i- or u-stem (chapter 8), before whose ṅit endings guṇa applies.
• anaṅ — the substitute an (marker ṅ) for the final i of asthi, dadhi, sakthi, akṣi (SK 322).
• ṅit — marked with ṅ; a ṅit substitute replaces only the final sound of the sthānin.
• udātta — a vowel produced high in its place of articulation.
• allopa — deletion of the a of a stem-final an in a bha (SK 234).
• ec — the pratyāhāra e, o, ai, au.
• paribhāṣā — a rule about how other rules are to be read.
• niyama — a rule that restricts a wider rule rather than adding anything new (SK 323).
• ekadeśavikṛtam ananyavat — the maxim that a thing altered in one part is treated as the same thing.

Summary

Stems run through the same rules: sudhī → sudhi (sudhiyā / sudhinā), madhu (madhunī, madhūni), sānu (snūni / sānūni), priyakroṣṭu (priyakroṣṭūni; priyakroṣṭrā / priyakroṣṭunā), sulū → sulu (sulvā / sulunā), dhātṛ (dhātṛṇī, dhātṝṇi; dhātrā / dhātṛṇā).