After this chapter you will be able to take a masculine noun whose stem ends in a vowel — rāma, sarva, hari, sakhi, pati, kroṣṭṛ, pitṛ, go, rai — and derive all twenty-one of its case forms by Pāṇini's own steps. You will meet the twenty-one case-endings (the sup affixes) for the first time, the names prātipadika, pratyaya, vibhakti, aṅga, sarvanāman, sarvanāmasthāna, pada, bha, ghi and nadī, and the small set of operations — a substitute for an ending, a lengthening of the stem, an augment, an elision — from which every paradigm is built. The rāma paradigm is worked out in full first; every later stem-type is then a matter of noting where it departs from rāma. Each of the 109 rules below is shown at work on a concrete word, and every paradigm form that Bhaṭṭoji gives is set out in a small table.
su … sup to rāmaBefore a single case form can be derived, the grammar must say what a noun stem is, what an affix is, on which side the affix goes, after what kind of base the case-endings come, and what the case-endings are. This group gives those nine rules. Read together they say: rāma is a prātipadika; the twenty-one items su, au, jas … are affixes; they are placed after rāma; they are called vibhakti; and they fall into seven triads of singular, dual and plural.
Rule. A meaningful word-form that is not a root (dhātu), not an affix (pratyaya), and does not end in an affix, is named prātipadika. A prātipadika is what English grammars call a noun stem: the part of a noun to which the case-endings are attached.
Example.
rāma → it has a meaning (the name Rāma) → it is not a root, not an affix, and does not end in an affix → rāma is a prātipadika.
bhū (the root "be") → it is a dhātu → not a prātipadika.
su (the first case-ending) → it is an affix → not a prātipadika.
Applies when / blocked by. The word "meaningful" (arthavat) matters: a stretch of sound with no meaning of its own is not a prātipadika. Forms that do end in an affix are covered by the next rule, SK 179, which admits some of them.
Rule. A form ending in a kṛt affix (an affix added to a root, making a noun from a verb), a form ending in a taddhita affix (an affix added to a noun, making a new noun), and a compound (samāsa) are also prātipadika. These three would be shut out by SK 178, which excludes anything ending in an affix.
Example.
rājapuruṣa ("king's man") → it is a compound → it is a prātipadika, and so it can take the case-endings just like rāma.
Applies when / blocked by. Bhaṭṭoji points out that a compound is meaningful and so was already a prātipadika by SK 178. The mention of "compound" here is therefore a niyama, a restriction: where an aggregate of words has a finished word (pada) as its first part, it can be a prātipadika only if it is a compound. A sentence, which also has a pada as its first part, is thereby ruled out.
Rule. "Affix." This is an adhikāra, a heading: everything taught from here to the end of the fifth book of the Aṣṭādhyāyī is an affix (pratyaya). The case-endings of SK 183, taught at A 4.1.2, fall inside this stretch, so each of them is a pratyaya.
Example.
su, au, jas … (taught at A 4.1.2) → A 4.1.2 lies under the heading pratyayaḥ → each of them is an affix.
Watch out. Bhaṭṭoji's commentators observe that Pāṇini chose the long, meaningful name pratyaya when a one-syllable label would have done. They read this as a hint (jñāpaka) that substitutes for an affix and augments added to an affix are not themselves affixes. This will matter when an ending is replaced (for example ṭā → ina in SK 201).
Rule. "And after." A second heading with the same reach as SK 180: the affix comes after the base from which it is taught. That is why the case-ending follows the stem — rāma + su, never su + rāma.
Example.
rāma + the affix su → the affix is placed after its base → rāma-su, not su-rāma.
Rule. A heading that runs to the end of the fifth book: the affixes taught from here on come after a base ending in ṅī, after a base ending in āp, or after a prātipadika. Ṅī and āp are the feminine affixes (the ī of gaurī, the ā of ramā); a stem ending in one of them is called ṅī-ending or āp-ending. This heading opens the slot into which the case-endings of the next rule are inserted.
Example.
rāma → it is a prātipadika (SK 178) → the affixes taught under this heading, including su … sup, may follow it.
Applies when / blocked by. A paribhāṣā — a maxim that tells how other rules are read — already says that "prātipadika" includes a stem qualified by gender, so a ṅī- or āp-ending stem would count anyway. Bhaṭṭoji's commentators explain why ṅī and āp are still named: so that a taddhita affix arises after the complete feminine stem, and not before the feminine affix is added.
Rule. After a ṅī-ending stem, an āp-ending stem or a prātipadika come the affixes su, au, jas; am, auṭ, śas; ṭā, bhyām, bhis; ṅe, bhyām, bhyas; ṅasi, bhyām, bhyas; ṅas, os, ām; ṅi, os, sup. These twenty-one are the sup affixes, the case-endings. The name sup is a pratyāhāra: from su, the first, to the p that ends the last.
Several of the sounds in these names are markers (it) and never appear in a word. Bhaṭṭoji lists them: the u of su, the final vowel of ṅasi, and the j of jas, the ś of śas, the ṭ of ṭā and auṭ, the ṅ of ṅe, ṅasi, ṅas, ṅi, and the p of sup. Why each is a marker is taught by SK 189, 190 and 195 below. The table gives the endings and the sound that actually gets attached:
| case | singular | dual | plural |
|---|---|---|---|
| 1 (prathamā, nominative) | su → s | au → au | jas → as |
| 2 (dvitīyā, accusative) | am → am | auṭ → au | śas → as |
| 3 (tṛtīyā, instrumental) | ṭā → ā | bhyām | bhis |
| 4 (caturthī, dative) | ṅe → e | bhyām | bhyas |
| 5 (pañcamī, ablative) | ṅasi → as | bhyām | bhyas |
| 6 (ṣaṣṭhī, genitive) | ṅas → as | os | ām |
| 7 (saptamī, locative) | ṅi → i | os | sup → su |
Example.
rāma + su → rāma-s (the u of su is a marker) → the nominative singular, finished at SK 187 as rāmaḥ.
Watch out. The markers are not idle. A rule can name an ending by its full form (jas, ṅe, ṅasi) and so pick out exactly one of several endings that sound alike: jas and śas both attach as as, and ṅasi and ṅas both attach as as. A ṅ-marked ending is called ṅit, and several later rules act only before a ṅit ending.
Rule. The sup affixes, and likewise the verb-endings (tiṅ), are also named vibhakti. The seven triads su au jas, am auṭ śas and so on are known by the names of the older grammarians, prathamā ("first") through saptamī ("seventh"), and Pāṇini's grammar uses those names too.
Example.
rāma + su → su is a sup affix → it is a vibhakti of the first triad, prathamā.
Watch out. The name vibhakti has its own effects: SK 190 below keeps certain sounds from being markers when they stand in a vibhakti.
Rule. Of the sup affixes, each successive group of three is named, one by one, ekavacana (singular), dvivacana (dual) and bahuvacana (plural). So su is singular, au dual, jas plural; am singular, auṭ dual, śas plural; and so on through ṅi, os, sup.
Example.
su, au, jas → first, second and third of a triad → ekavacana, dvivacana, bahuvacana respectively.
Rule. When two things are meant, the dual endings are used; when one thing is meant, the singular endings are used.
Example.
one Rāma → rāma + su → rāmaḥ.
two Rāmas → rāma + au → rāmau (SK 188).
Watch out. The rule for the plural, "many", is SK 187, the first rule of the next group, where the paradigm of rāma begins.
rāma, nom. sg. rāmaḥ → loc. pl. rāmeṣuThis is the model paradigm of the chapter. Twenty-six rules take rāma from the nominative singular to the locative plural. Along the way you meet: how a marker in an ending is spotted (SK 189, 190, 195); what happens when the stem's a meets the vowel of an ending (SK 191, 194, 196); the vocative (SK 192, 193); the name aṅga (SK 199, 200); the substitutes that the a-stem imposes on its endings (SK 201, 203, 204); the lengthening and e-substitution of the stem's a (SK 202, 205, 207, 209); the augment of the genitive plural (SK 208); and the two retroflexions, n → ṇ (SK 197, 198) and s → ṣ (SK 210–212). The finished paradigm is given after SK 212.
Rule. When many things are meant, the plural endings are used. With this the three numbers are placed, and Bhaṭṭoji derives the first form.
Example.
rāma + su → the u of su is a marker → rāma + s
→ the final s becomes ru (sasajuṣo ruḥ, SK 162) → rāma + r
→ at a pause the r becomes visarga (Chapter 6) → rāmaḥ (nominative singular).
Watch out. Nothing in this section changes the stem; every step falls on the ending. That is the normal pattern: the stem is touched only when a rule names it.
Rule. When several words of the same form stand in one and the same case-ending, one alone is kept. This "one remaining" is called ekaśeṣa. It is how "two Rāmas" comes to be a single stem with a dual ending, rather than rāma said twice.
Example.
rāma + rāma + au (two Rāmas) → the two like-formed words share one ending → only one rāma remains → rāma + au
→ prathamayoḥ pūrvasavarṇaḥ (SK 164) would merge a + au into one long vowel → but nādici (SK 165) forbids it before ic (every vowel except a and ā) → vṛddhir eci (SK 72) applies instead: a + au → au → rāmau (nominative dual).
Applies when / blocked by. Only for words of like form, only in one vibhakti. Bhaṭṭoji's commentators mention that a doubled name such as rājarāja is not an ekaśeṣa but a compound.
Rule. A palatal (cu: c, ch, j, jh, ñ) or a retroflex (ṭu: ṭ, ṭh, ḍ, ḍh, ṇ) standing at the beginning of an affix is a marker (it). This is why the j of jas and the ṭ of ṭā are markers, and why ṭā attaches as ā.
Example.
jas (nominative plural) → its initial j is a palatal → j is it → the ending attaches as as.
Watch out. Bhaṭṭoji's commentators explain why the j is still written if it is only a marker: it lets a later rule say "jas" and mean the nominative plural alone, distinct from śas, which also attaches as as. You will see this at SK 214, jasaḥ śī.
Rule. A dental (tu: t, th, d, dh, n), an s or an m is not a marker when it stands in a vibhakti. Chapter 1 made a final consonant of any upadeśa a marker (SK 1); this rule withdraws that for the case- and verb-endings, so that the s of jas, the m of am, the t of a verb-ending and so on are kept.
Example.
jas → its final s would be a marker by SK 1 → but s stands in a vibhakti → the s is not a marker → as is what attaches.
Applies when / blocked by. Only in a vibhakti. In an affix that is not a vibhakti, a final s, m or dental is still a marker.
Rule. After a short a that is not at the end of a pada, when a guṇa vowel (a, e, o) follows, the single substitute for both is the form of the following vowel. This is called pararūpa, "the form of the latter". In the nominative plural of rāma this rule would apply — but it is beaten.
Example.
rāma + jas → rāma + as (the j is a marker, the s stays) → a + a: pararūpa by this rule would give rāmas
→ but prathamayoḥ pūrvasavarṇaḥ (SK 164) also applies, giving one long ā; being the later rule it wins (paratva) → rāmās
→ s → ru → visarga → rāmāḥ (nominative plural).
Applies when / blocked by. Why does SK 164 win, when SK 191 looks like the more specific rule? Bhaṭṭoji answers with a paribhāṣā: an exception stated earlier blocks only the rules that immediately follow it, not later ones. ato guṇe stands just before akaḥ savarṇe dīrghaḥ (SK 85) and is an exception only to that rule, not to prathamayoḥ pūrvasavarṇaḥ, which comes later. So the two compete as equals and the later one, SK 164, prevails.
Rule. The singular of the first case (su), when used in address, is named sambuddhi. In English terms, sambuddhi is the vocative singular. There is no separate set of vocative endings: the vocative dual and plural are simply the nominative dual and plural used in address.
Example.
he rāma + su ("O Rāma!") → the nominative singular used in address → this su is the sambuddhi.
Rule. After a stem (aṅga) ending in eṅ (e or o) or in a short vowel, the consonant belonging to the sambuddhi is dropped. Since the sambuddhi is su and its u is a marker, this drops the whole ending.
Example.
he rāma + su → the stem ends in short a → the s of the sambuddhi is dropped → he rāma (vocative singular).
Vocative dual and plural: he rāmau, he rāmāḥ — the nominative forms, since only the singular has a special vocative.
Applies when / blocked by. Why does the rule mention eṅ at all, when a vowel-stem is short anyway? For hari and viṣṇu. In their vocative, SK 242 first turns the final short vowel into a guṇa vowel — that rule applies first because it is later and obligatory — and the stem then ends in e or o, no longer in a short vowel. Without "eṅ" the s would survive; with it, he hare, he viṣṇo.
Rule. When the vowel of the ending am follows an ak vowel (a, i, u, ṛ, ḷ, short or long), the single substitute for both vowels is the form of the prior one. This is pūrvarūpa, "the form of the former": the stem's vowel survives and the a of am vanishes.
Example.
rāma + am → a + a: the single substitute is the prior a → rāmam (accusative singular).
rāma + auṭ → the ṭ is a marker (SK 189) → rāma + au → vṛddhi as at SK 188 → rāmau (accusative dual).
Watch out. Bhaṭṭoji's commentators stress that the rule is about the vowel of am, not am as a whole; that is why the m remains and the result is rāmam, not rāma.
Rule. An l, an ś or a velar (ku: k, kh, g, gh, ṅ) at the beginning of an affix is a marker, unless the affix is a taddhita. So the ś of śas is a marker and the ending attaches as as.
Example.
śas (accusative plural) → its initial ś is at the start of a non-taddhita affix → ś is it → attaches as as.
Rule. In the masculine, the s of śas becomes n when it stands after the long vowel produced by prathamayoḥ pūrvasavarṇaḥ. "That" (tasmāt) in the sūtra refers to that long vowel.
Example.
rāma + śas → rāma + as → a + a merge into the long vowel of the stem's kind (SK 164) → rāmās
→ the s stands after that long vowel, and the noun is masculine → s → n → rāmān (accusative plural).
Applies when / blocked by. Only in the masculine; the rule is stated for the masculine only.
Rule. Within one pada, an n that follows r or ṣ becomes ṇ even when something intervenes between them, provided the intervening sounds are only these: an aṭ (any vowel, or h, y, v, r), a velar (ku), a labial (pu: p, ph, b, bh, m), the prefix āṅ, or the augment num — singly or in any combination. The basic rule that n → ṇ after r or ṣ is SK 235 (raṣābhyāṃ no ṇaḥ), printed later in this chapter; this rule extends it across intervening sounds, and it is what gives rāma its two retroflex forms.
Example.
rāmena (SK 201) → r … n with ā (an aṭ) and m (a labial) between → n → ṇ → rāmeṇa (instrumental singular).
rāmānām (SK 209) → r … n with ā, m, ā between → rāmāṇām (genitive plural).
Applies when / blocked by. Bhaṭṭoji explains two of the items. āṅ is listed so that this rule can override the prohibition "even when a pada intervenes" (SK 1057). num is listed to stand for the anusvāra, and could even be omitted, since the anusvāra and its kin are already taught inside aṭ. The next rule gives the one place the retroflexion must not reach.
Rule. An n at the end of a pada is not made retroflex. This protects the accusative plural.
Example.
rāmān → r … n with only ā, m, ā between, so SK 197 would apply → but this n ends the pada → no change → rāmān.
Rule. The form that begins with the base after which an affix is prescribed is named aṅga when that affix follows. So when su follows rāma, rāma is the aṅga; and once a form has grown — say, a root plus a class-sign — the whole of it, from the base onward, is the aṅga before the next affix.
Example.
rāma + su → su is prescribed after a prātipadika, and rāma begins with that base → rāma is the aṅga before su.
bhava + mi (in bhavāmi, "I am") → the affix mi is prescribed after the root bhū → the form beginning with the root, bhava, including the class-sign a, is the aṅga. This is the point of "beginning with that" (tadādi).
Applies when / blocked by. Bhaṭṭoji gives three cautions. "Beginning with that" (tadādi): the aṅga is the whole form from the base onward, so in bhavāmi it is bhava, class-sign included, as the Example shows. "Prescribed" (vidhi): in strī and iyatī the item concerned is not one after which the affix was prescribed, so it is not an aṅga. "When the affix follows": the name belongs to the form before the affix, not to the form together with the affix, nor to anything larger.
Rule. "Of the aṅga." A heading in the genitive, read into the rules that follow: each of them is understood as an operation on an aṅga. The lengthenings and vowel changes of the stem in this chapter — SK 202, 205, 207, 209 and the rest — are all "of the aṅga".
Example.
supi ca (SK 202) → read with this heading: "the a of the aṅga becomes long before a sup beginning with yañ".
Rule. After an aṅga ending in a, the endings ṭā, ṅasi and ṅas are replaced, in order, by ina, āt and sya. This is a substitute for the ending, not for the stem; and being of more than one sound, each replaces the whole ending.
Example.
rāma + ṭā → ṭā → ina → rāma + ina → a + i → e (ād guṇaḥ, SK 69) → rāmena → n → ṇ (SK 197) → rāmeṇa (instrumental singular).
rāma + ṅasi → ṅasi → āt → rāma + āt → a + ā → ā (SK 85) → rāmāt (ablative singular; its final consonant is treated at SK 206).
rāma + ṅas → ṅas → sya → rāmasya (genitive singular).
Watch out. The order in the sūtra is the order of the substitutes: ṭā gets ina, ṅasi gets āt, ṅas gets sya. Matching one list to another in order is the standard way a sūtra pairs items.
Rule. The final a of an aṅga becomes long before a sup affix beginning with a yañ consonant (y, v, r, l, the nasals, and the voiced stops jh, bh, gh, ḍh, dh, j, b, g, ḍ, d). Of the case-endings, bhyām, bhis and bhyas begin with bh, a yañ.
Example.
rāma + bhyām → the stem's a stands before a yañ-initial sup → a → ā → rāmābhyām (instrumental, dative and ablative dual).
Applies when / blocked by. bhis and bhyas also begin with a yañ, but they are plural and are claimed first by SK 203 and SK 205.
Rule. After an aṅga ending in a, the ending bhis is replaced by ais. Because ais has more than one sound, it replaces the whole of bhis (anekāl śit sarvasya), not just its last sound.
Example.
rāma + bhis → bhis → ais → rāma + ais → a + ai → ai (vṛddhir eci, SK 72) → rāmais → s → ru → visarga → rāmaiḥ (instrumental plural).
Rule. After an aṅga ending in a, the dative singular ending ṅe is replaced by ya.
Example.
rāma + ṅe → ṅe → ya → rāma + ya
→ ya stands in place of a sup and so counts as one (sthānivadbhāva); it begins with y, a yañ → the stem's a is lengthened by SK 202 → rāmāya (dative singular).
Dative dual: rāma + bhyām → rāmābhyām (SK 202).
Applies when / blocked by. One might object that a paribhāṣā forbids an operation caused by a combination from destroying that very combination (the sannipāta-paribhāṣā): here the substitute ya arose because the stem ended in short a, and the lengthening then removes that short a. Bhaṭṭoji replies that this paribhāṣā is not invariable, as the form kaṣṭāya in the sūtra kaṣṭāya kramaṇe (SK 2670) shows, and so it does not operate here.
Rule. The final a of an aṅga becomes e before a plural sup affix beginning with a jhal (the stops, the sibilants and h — every consonant except the nasals and the semivowels). Three plural endings begin with a jhal: bhis, bhyas and sup (attached as su). bhis is taken by SK 203 first, so this rule reaches bhyas and sup.
Example.
rāma + bhyas → plural, jhal-initial → a → e → rāmebhyas → s → ru → visarga → rāmebhyaḥ (dative and ablative plural).
rāma + su (locative plural) → a → e → rāme + su, finished at SK 212 as rāmeṣu.
Applies when / blocked by. Bhaṭṭoji tests each word of the sūtra. "Plural": in the singular there is no e — rāmaḥ, rāmasya. "Jhal": before the vowel-initial ām of the genitive plural there is no e — rāmāṇām. "Sup": before a verb-ending beginning with a jhal, such as the dhvam of pacadhvam, there is no e.
Watch out. This rule, not SK 202, governs bhyas: the e of rāmebhyaḥ is taught here, while the ā of rāmābhyām comes from SK 202, because bhyām is dual.
Rule. At a pause (avasāna), a jhal optionally becomes the corresponding car (the first, voiceless unaspirated stop of its class, or a sibilant). This settles the last sound of the ablative singular rāmāt, whose t, being pada-final, has already become d by jaśtva (jhalāṃ jaśo 'nte, Chapter 5).
Example.
rāma + ṅasi → rāmāt (SK 201) → at a pause the final t becomes d by jaśtva → rāmād → optionally d → t by this rule → rāmāt or rāmād.
With the optional doubling of a consonant after a vowel (anaci ca), four forms result: rāmātt, rāmāt, rāmādd, rāmād.
The remaining forms of the fifth and sixth cases: rāmābhyām (ablative dual, SK 202), rāmebhyaḥ (ablative plural, SK 205), rāmasya (genitive singular, SK 201).
Watch out. If the s of rāmasya is doubled, one might expect khari ca (SK 121) to turn the first s into a t, the car of the dental class. Bhaṭṭoji says no: by the principle of closest likeness (āntaratamya) the substitute for s is s itself, since t differs from s in effort (it is unaspirated). That a separate rule is needed to turn s into t before s in verbs shows the same thing.
Rule. The final a of an aṅga also becomes e before the ending os (genitive and locative dual).
Example.
rāma + os → a → e → rāme + os → e → ay before a vowel (eco 'yavāyāvaḥ) → rāmayos → s → ru → visarga → rāmayoḥ (genitive and locative dual).
Rule. The genitive plural ending ām receives the augment nuṭ when it follows an aṅga ending in a short vowel, or in a nadī stem (a feminine in long ī or ū, SK 266), or in āp. An augment (āgama) is a sound inserted into a form; the ṭ of nuṭ is a marker meaning "goes at the front" (Chapter 2), and the u is only there for pronunciation. So ām becomes nām.
Example.
rāma + ām → the stem ends in short a → ām takes nuṭ → rāma + nām, finished at SK 209.
Rule. An aṅga ending in a vowel becomes long before nām, the genitive plural ending with its augment.
Example.
rāma + nām → the stem's a becomes ā → rāmānām → n → ṇ by SK 197 (r … ṇ with ā, m, ā between) → rāmāṇām (genitive plural).
Locative singular: rāma + ṅi → rāma + i → a + i → e (SK 69) → rāme.
Locative dual: rāmayoḥ, as at SK 207.
Applies when / blocked by. Why is this rule needed, when SK 202 already lengthens a before a yañ-initial sup, and nām begins with n, a yañ? Because SK 202 is blocked here by the sannipāta-paribhāṣā: the augment nuṭ came in because the stem ended in a short vowel, and SK 202 would destroy that condition. But the very fact that Pāṇini took the trouble to state nāmi shows that it is meant to apply; by this "force of the undertaking" (ārambha-sāmarthya) it sets the paribhāṣā aside.
Rule. A heading that runs to the end of the third quarter of the eighth book: in the rules that follow, a dental sound that is not at the end of a pada is replaced by the retroflex (mūrdhanya) sound. The operation this chapter draws from it is ṣatva, s → ṣ.
Example.
rāme + su → the s of the ending is not pada-final → it is a candidate for ṣ, provided SK 211 and 212 are also satisfied.
Rule. A further heading: the retroflexion just announced operates on a sound that stands after iṇ or after ku. iṇ here is the pratyāhāra read up to the later ṇ of the Śivasūtras (the one in la ṇ), so it names i, u, ṛ, ḷ, e, o, ai, au, h, y, v, r, l — every vowel except a and ā, plus h, y, v, r, l. ku is the velar class.
Example.
rāme + su → the s stands after e, an iṇ sound → the condition is met.
rāmasya → the s stands after a, which is not an iṇ sound → no retroflexion.
Watch out. Bhaṭṭoji's commentators note that the sūtra's form iṇkoḥ (a genitive singular) is irregular for a pair of items; scholars discuss this.
Rule. An s that is a substitute, or is part of an affix, and that stands after iṇ or a velar and is not pada-final, is replaced by ṣ. The word "s" itself is carried into this rule from an earlier sūtra (saheḥ sāḍaḥ saḥ, SK 335). The ṣ that replaces it is, like the s, open and voiceless.
Example.
rāma + sup → the p is a marker → rāma + su → the stem's a becomes e before a plural jhal-initial sup (SK 205) → rāme + su → the s is part of an affix, stands after e, and is not pada-final → s → ṣ → rāmeṣu (locative plural).
Applies when / blocked by. Bhaṭṭoji tests each condition. "After iṇ or a velar": rāmasya keeps its s, which follows a. "A substitute or part of an affix": in supīḥ, supisau, supisaḥ the s belongs to the root and stays. "Not pada-final": in haris tatra the s ends the pada and stays.
The paradigm of rāma. Kṛṣṇa, Mukunda and every other masculine a-stem decline in the same way.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | rāmaḥ | rāmau | rāmāḥ |
| vocative | he rāma | he rāmau | he rāmāḥ |
| 2 accusative | rāmam | rāmau | rāmān |
| 3 instrumental | rāmeṇa | rāmābhyām | rāmaiḥ |
| 4 dative | rāmāya | rāmābhyām | rāmebhyaḥ |
| 5 ablative | rāmāt / rāmād | rāmābhyām | rāmebhyaḥ |
| 6 genitive | rāmasya | rāmayoḥ | rāmāṇām |
| 7 locative | rāme | rāmayoḥ | rāmeṣu |
sarva, nom. pl. sarve, gen. pl. sarveṣāmA pronoun such as sarva ("all") is an a-stem and declines like rāma — except in five places, where a special substitute or augment steps in: the nominative plural, the dative singular, the ablative singular, the locative singular and the genitive plural. The grammar handles this by giving a list of words the name sarvanāman (SK 213) and then writing four rules that act only on a sarvanāman (SK 214–217). The remaining rules of the group (SK 218–226) say when a word is on the list only in a certain sense, only optionally, or not at all.
Rule. The word-forms sarva and the others of its list (the sarvādi group, given in full under SK 217) are named sarvanāman. The name also belongs to a form that ends in one of these words (the tadanta reading, Chapter 2).
Example.
sarva → it heads the sarvādi list → it is a sarvanāman → the four rules SK 214–217 apply to it.
paramasarva ("the highest all") → it ends in sarva → it too is a sarvanāman → the affixes tral and akac that attach to a sarvanāman give paramasarvatra and paramasarvakān.
Applies when / blocked by. How do we know that a form ending in a sarvādi word counts? From SK 224, dvandve ca, which withholds the name from a dvandva ending in such a word — a prohibition that would be pointless unless such forms were otherwise included. A rule that reveals something by presupposing it is called a jñāpaka.
Rule. After a sarvanāman ending in a, the nominative plural ending jas is replaced by śī. Since śī has more than one sound, it replaces the whole of jas; then its ś is a marker (SK 195), so what attaches is ī.
Example.
sarva + jas → jas → śī → sarva + śī → the ś is a marker → sarva + ī → a + ī → e (ād guṇaḥ, SK 69) → sarve (nominative plural).
Applies when / blocked by. One might object: a maxim says that having more than one sound does not count when the extra sound is only a marker, so śī should be treated as the single sound ī and replace only the s of jas. Bhaṭṭoji answers that whether a substitute replaces the whole is judged before its marker is dropped; at that stage śī has two sounds and the maxim does not apply.
Rule. After a sarvanāman ending in a, the dative singular ending ṅe is replaced by smai. Compare rāma, where ṅe became ya (SK 204).
Example.
sarva + ṅe → ṅe → smai → sarvasmai (dative singular).
Rule. After a sarvanāman ending in a, the ablative singular ṅasi is replaced by smāt and the locative singular ṅi by smin.
Example.
sarva + ṅasi → ṅasi → smāt → sarvasmāt (ablative singular).
sarva + ṅi → ṅi → smin → sarvasmin (locative singular).
Rule. The genitive plural ending ām, when it is prescribed after a sarvanāman, receives the augment suṭ — that is, an s at its front — provided the aṅga it follows ends in a or ā. In sarva + ām that aṅga is the sarvanāman sarva itself.
Example.
sarva + ām → ām takes the augment s → sarva + sām → sām is a plural sup beginning with a jhal, so the stem's a becomes e (SK 205) → sarve + sām → the s is part of an affix, follows e and is not pada-final → s → ṣ (SK 212) → sarveṣām (genitive plural).
The rest of sarva is like rāma:
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | sarvaḥ | sarvau | sarve |
| 2 accusative | sarvam | sarvau | sarvān |
| 3 instrumental | sarveṇa | sarvābhyām | sarvaiḥ |
| 4 dative | sarvasmai | sarvābhyām | sarvebhyaḥ |
| 5 ablative | sarvasmāt | sarvābhyām | sarvebhyaḥ |
| 6 genitive | sarvasya | sarvayoḥ | sarveṣām |
| 7 locative | sarvasmin | sarvayoḥ | sarveṣu |
viśva and the other a-ending words of the list decline the same way.
The sarvādi list. Bhaṭṭoji counts thirty-five words: sarva, viśva, ubha, ubhaya, ḍatara, ḍatama, anya, anyatara, itara, tvat, tva, nema, sama, sima; then, with conditions stated in the list itself, pūrva, para, avara, dakṣiṇa, uttara, apara, adhara (when they mark relative position and are not a proper name), sva (when it does not mean a kinsman or wealth), antara (in the senses "outer" and "undergarment"); and finally tyad, tad, yad, etad, idam, adas, eka, dvi, yuṣmad, asmad, bhavatu, kim.
Bhaṭṭoji adds notes on several of these. ubha means a pair and is always dual; it is on the list only so that the affix akac can be added to it (ubhakau). ubhaya is used where no dual ending follows (a vārttika says so), and its nominative plural is ubhaye; whether ubhaya has a dual at all, scholars discuss. ḍatara and ḍatama are affixes, so words ending in them (katara, katama) are meant. anyatara and anyatama are underived words for choosing between two and among many; anyatama is not on the list and is not a sarvanāman. tvat and tva both mean "other". nema means "half". sama is a sarvanāman when it means "all", not when it means "equal" — Pāṇini himself writes the plain genitive samānām, not sameṣām, in a sūtra. A vārttika adds that antara is not a sarvanāman when it qualifies a city: antarāyāṃ puri, not antarasyām.
Rule. pūrva ("prior, eastern"), para, avara, dakṣiṇa ("southern, right"), uttara, apara and adhara are sarvanāman when they express a fixed relative position (vyavasthā) and are not used as a proper name. Because they are already on the list, the name would be invariable; this sūtra, read inside the list, makes it optional before jas.
Example.
pūrva + jas → as a sarvanāman, jas → śī → pūrve; or, the name not being taken, jas stays → pūrvāḥ (nominative plural, both correct).
Applies when / blocked by. Bhaṭṭoji defines vyavasthā as a fixed limit depending on what the word itself denotes. "Relative position": in dakṣiṇā gāthakāḥ, "skilful singers", dakṣiṇa means "skilful", not "southern", so only dakṣiṇāḥ, never dakṣiṇe. "Not a proper name": uttarāḥ kuravaḥ, "the Northern Kurus", is a name, so uttarāḥ, never uttare.
Rule. sva is a sarvanāman — optionally before jas — when it does not denote a kinsman or wealth.
Example.
sva + jas meaning "one's own" or "selves" → sve or svāḥ (nominative plural).
sva + jas meaning "kinsmen" or "possessions" → only svāḥ.
Rule. antara is a sarvanāman — optionally before jas — in the sense "outer" (bahiryoga) and in the sense "undergarment" (upasaṃvyāna).
Example.
antare gṛhāḥ or antarā gṛhāḥ ("the outer houses"; the second form is antarāḥ with its visarga lost before the voiced g).
antare śāṭakāḥ or antarā śāṭakāḥ ("the undergarments").
Rule. After the nine words pūrva … antara (the seven of SK 218, sva and antara), smāt and smin replace ṅasi and ṅi only optionally.
Example.
pūrva + ṅasi → pūrvasmāt or pūrvāt (ablative singular).
pūrva + ṅi → pūrvasmin or pūrve (locative singular).
para and the rest behave alike; everything else is as for sarva.
Watch out. Bhaṭṭoji adds that eka, when it means the number one, is always singular.
Rule. When a bahuvrīhi (a possessive compound, "whose X is Y") is being formed, a sarvādi word does not get the name sarvanāman.
Example.
tvatkapitṛka ("he whose father is you"), matkapitṛka ("he whose father is I") → in the derivation of the compound the name is withheld → the pronominal affix akac cannot enter → the compound shows no akac.
Applies when / blocked by. Bhaṭṭoji explains that the prohibition acts in the derivational sentence before the compound is formed; otherwise akac would appear there, as it does in the ordinary analytic sentence, and would then be heard in the compound too. Scholars discuss this, since the Mahābhāṣya accepts the forms with akac and rejects the sūtra. A vārttika adds that sarvādi words used as a proper name or as a subordinate member of a compound are not sarvanāman: sarvāya (to a man named Sarva), atisarvāya, atikataram, atitat.
Rule. In a compound whose first member stands in the instrumental sense (a tṛtīyā-tatpuruṣa), a sarvādi word is not a sarvanāman. The "not" is carried down from SK 222.
Example.
māsapūrva ("earlier by a month") + ṅe → pūrva is not a sarvanāman here → no smai → māsapūrvāya (dative singular).
Likewise in the sentence that means the same as the compound: māsena pūrvāya, not pūrvasmai.
Rule. In a dvandva (a compound meaning "X and Y and Z"), the sarvanāman name is withheld from the whole compound.
Example.
varṇāśrametara ("caste, stage of life and the rest") + ām → the compound is not a sarvanāman → no augment suṭ → varṇāśrametarāṇām (genitive plural), not varṇāśrametareṣām.
Applies when / blocked by. The prohibition is of the aggregate, not of its members; itara by itself remains a sarvanāman. Nor can suṭ sneak back in by the tadanta reading ("ending in a sarvanāman"), because SK 217 was explained as applying to an ām prescribed after a sarvanāman, and this ām is prescribed after the compound.
Rule. In a dvandva the name comes back optionally before jas — that is, only for the śī-substitution.
Example.
varṇāśrametara + jas → varṇāśrametare or varṇāśrametarāḥ (nominative plural).
Applies when / blocked by. Because the option concerns only the śī-substitution, the pronominal affix akac does not follow from it; the ordinary affix kap is used instead: varṇāśrametarakāḥ.
Rule. prathama ("first"), carama ("last"), words ending in the affix taya, alpa, ardha, katipaya and nema are optionally sarvanāman before jas.
Example.
prathama + jas → prathame or prathamāḥ; the rest of prathama is like rāma.
dvitaya + jas → dvitaye or dvitayāḥ; the rest like rāma.
nema + jas → neme or nemāḥ; the rest like sarva, since nema is on the list.
Applies when / blocked by. Since taya is an affix, the sūtra means words ending in it. A vārttika adds that a word ending in tīya is optionally a sarvanāman before the ṅit endings: dvitīyasmai or dvitīyāya; likewise tṛtīya. This does not extend to paṭujātīyāya, because the tīya inside jātīya is not a meaningful element on its own.
jarā/nirjarā, pāda, yūṣan, ahan, viśvapāThe a-stems end with a handful of words that swap in a different stem before some endings: jarā becomes jaras, pāda becomes pad, yūṣa becomes yūṣan. To say before which endings, Pāṇini divides the twenty-one endings into the first five (sarvanāmasthāna, SK 229) and the rest, and names the stem pada before some of the rest and bha before others (SK 230–232). Rules that say "of a bha" then govern the replacement stems. The group ends with the ā-stems (viśvapā), whose special feature is the dropping of the root's ā in the bha positions.
Rule. For the word jarā ("old age"), jaras is optionally the substitute before a vibhakti beginning with a vowel. The example is the masculine nirjara (nir + jarā, "free of old age").
Example.
nirjara + au → jaras replaces the jara part → nirjarasau (nominative dual); nirjarasaḥ (nominative plural).
nirjara + ṭā → jaras wins over the a-stem substitute ina (SK 201) because it is the later rule → nirjarasā (instrumental singular); nirjarase (dative singular); nirjarasaḥ (ablative and genitive singular).
In the other option, and before every consonant-initial ending, like rāma: nirjaraḥ, nirjarau, nirjarābhyām.
Applies when / blocked by. Two paribhāṣās are at work. In the domain of pada and aṅga, a rule for a word applies also to what ends in it — hence to nirjarā. And a substitute replaces only what is actually pointed out; so, although jaras has more than one sound, it does not replace the whole compound but only the jarā part — and jarā is not a different word for having lost its length in nirjara. Bhaṭṭoji records that the author of the Vṛtti derived nirjarasena, nirjarasāt and nirjarasaiḥ by applying the a-stem substitutes first, and that the Vṛtti's followers accepted nirjarasya for the genitive singular; scholars discuss this, and Bhaṭṭoji sides with the Mahābhāṣya.
Rule. For pāda ("foot"), danta ("tooth"), nāsikā ("nose"), māsa ("month"), hṛdaya ("heart"), niśā ("night"), asṛj ("blood"), yūṣa ("broth"), doṣā ("arm"), yakṛt ("liver"), śakṛt ("dung"), udaka ("water") and āsya ("mouth"), the substitutes pad, dat, nas, mās, hṛt, niś, asan, yūṣan, doṣan, yakan, śakan, udan and āsan optionally appear before śas and the endings after it.
Example.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | pādaḥ | pādau | pādāḥ |
| 2 accusative | pādam | pādau | padaḥ / pādān |
| 3 instrumental | padā / pādena |
Before su … auṭ only pāda appears; from śas on, either stem.
Applies when / blocked by. Bhaṭṭoji rejects the Kāśikā's statement that āsan is the substitute for a word āsan; the substitute is for āsya. The word "and the rest" (prabhṛti) in the sūtra is discussed at SK 237.
Rule. The five endings named by the pratyāhāra suṭ — su, au, jas, am, auṭ, from su to the ṭ of auṭ — are named sarvanāmasthāna for a stem that is not neuter. These are the nominative singular, dual and plural and the accusative singular and dual, the "strong" cases.
Example.
pāda + au → au is one of the first five endings, and pāda is masculine → this au is a sarvanāmasthāna.
pāda + śas → śas is the sixth ending → not a sarvanāmasthāna.
Rule. Before an affix of the su-series that is not a sarvanāmasthāna — the series is counted from su up to the affix kap — what precedes is named pada. This is a second way of being a pada, beside the definition in Chapter 1 (a form ending in sup or tiṅ): here the bare stem, before a consonant-initial ending such as bhyām, is a pada, and pada rules like jaśtva and n-elision can act on it.
Example.
dat + bhyām → bhyām is not a sarvanāmasthāna, and begins with a consonant → dat is a pada → jaśtva (Chapter 5) turns its final t to d → dadbhyām.
Rule. Before such an affix (of the su-series up to kap, not a sarvanāmasthāna) that begins with y or with a vowel, what precedes is named bha.
Example.
dat + śas → śas is not a sarvanāmasthāna and begins with a vowel → dat is a bha.
dat + ṭā → dat + ā → dat is a bha.
Watch out. Before śas or ṭā the conditions of SK 230 are also met: these endings are not sarvanāmasthāna. So the stem seems to be both pada and bha. The next rule decides.
Rule. A heading: from here up to the sūtra kaḍārāḥ karmadhāraye (A 2.2.38), one item receives only one name — the name given by the later rule, or the name that would otherwise have no scope. bha (SK 231) is later than pada (SK 230); so before a vowel-initial ending like śas the stem is a bha and not a pada, and pada operations such as jaśtva do not touch it.
Example.
dat + śas → dat is a bha only, not a pada → no jaśtva → dataḥ (accusative plural); likewise datā (instrumental singular).
dat + bhyām → here dat is a pada → jaśtva → dadbhyām.
mās + śas → māsaḥ (accusative plural); māsā (instrumental singular).
mās + bhyām → mās is a pada → its final s becomes ru (SK 162), the r becomes y before the voiced bh (SK 167), and the y is dropped (SK 171) → mābhyām; likewise mābhiḥ.
Rule. "Of a bha." A heading: the operations that follow are of a bha — they apply to an aṅga that bears that name.
Example.
allopo 'naḥ (SK 234) → read under this heading: "of a bha, the a of an is dropped".
Rule. The a of a stem-final an is dropped (lopa) when the stem is a bha — that is, before a non-sarvanāmasthāna ending of the su-series beginning with y or a vowel. The running example is yūṣan, the optional replacement of yūṣa from śas onward (SK 228).
Example.
yūṣan + śas → yūṣan + as → the stem is a bha → its a before n is dropped → yūṣn + as, finished at SK 235 as yūṣṇaḥ.
Rule. An n standing after r or ṣ becomes ṇ, when both sounds are in the same pada. This is the base rule of retroflexion (ṇatva); SK 197, met in the rāma paradigm, extended it across intervening sounds.
Example.
yūṣn + as → ṣ directly before n, in one pada → n → ṇ → yūṣṇaḥ (accusative plural).
yūṣan + ṭā → yūṣn + ā → yūṣṇā (instrumental singular).
Applies when / blocked by. Bhaṭṭoji notes a second route: if the dropped a is held to count as still present for an operation on what precedes it (sthānivadbhāva), then ṣ and n are separated by that a, and the retroflexion comes instead by SK 197, which allows an aṭ to intervene. Two vārttikas say that the maxim "no sthānivadbhāva in the Tripādī" does not hold for retroflexion; scholars discuss this.
Rule. The final n of a pada that is the end of a prātipadika is dropped. Before bhyām, bhis and bhyas the stem yūṣan is a pada (SK 230), so its n goes.
Example.
yūṣan + bhyām → yūṣan is a pada → its final n is dropped → yūṣabhyām; likewise yūṣabhiḥ, yūṣabhyaḥ.
Applies when / blocked by. This rule lies in the Tripādī and is asiddha for the earlier rules (SK 12). So although the stem now ends in a, the lengthening of SK 202, the e of SK 205 and the ais of SK 203 do not see that a and do not apply: yūṣabhyām, not yūṣābhyām; yūṣabhiḥ, not yūṣaiḥ; yūṣabhyaḥ, not yūṣebhyaḥ.
Rule. The a of a stem-final an is dropped only optionally before ṅi (locative singular) or śī.
Example.
yūṣan + ṅi → yūṣṇi (with the elision) or yūṣaṇi (without it; the ṇ by SK 197 across the a). In the option where yūṣan is not substituted at all, like rāma: yūṣe.
Applies when / blocked by. Bhaṭṭoji explains that "and the rest" in SK 228 marks a type, not a closed list. So the Mahābhāṣya's kakuddoṣaṇī shows doṣan even before the śī that replaces a dual ending; lexical forms like padbhiḥ and hṛt are covered; and doṣ ("arm") is found in the neuter (dakṣiṇaṃ doḥ) and, by association with masculine words, in the masculine. dvyahna means "occurring in two days".
Rule. For ahan ("day") preceded by a number, by vi or by sāya, the substitute ahan optionally appears before ṅi. The compounds concerned end in ahna. Bhaṭṭoji gives three locative forms for each: the plain a-stem form, and two that show the restored ahan.
Example.
dvyahna ("a two-day period") + ṅi → dvyahni, dvyahani, dvyahne (locative singular; the last is the form without the substitute).
vyahna ("one whose day is past") → vyahni, vyahani, vyahne.
sāyāhna ("evening of the day") → sāyāhni, sāyāhani, sāyāhne.
With this the a-stems end. Next, the ā-stem viśvapā ("protector of all"), whose nominative singular is viśvapāḥ.
Rule. After a long vowel, the merging of the first two endings into a long vowel of the stem's kind (prathamayoḥ pūrvasavarṇaḥ, SK 164) does not take place before jas, nor before an ic vowel.
Example.
viśvapā + au → no pūrvasavarṇa → vṛddhir eci (SK 72) → viśvapau (nominative dual).
viśvapā + jas → viśvapā + as → no pūrvasavarṇa → akaḥ savarṇe dīrghaḥ (SK 85) → viśvapāḥ (nominative plural).
Applies when / blocked by. Bhaṭṭoji admits that for viśvapā the rule changes nothing: before au the blocker nādici (SK 165) already suffices, and before jas both routes give viśvapāḥ. He presents it here because it is decisive for the feminine forms gauryau and gauryaḥ (Chapter 9).
Rule. When a bha-stem ends in a root ending in ā, its final ā is dropped. By alo 'ntyasya (SK 42) only the last sound goes.
Example.
viśvapā + śas → the stem is a bha and ends in the root pā → the ā is dropped → viśvap + as → viśvapaḥ (accusative plural).
viśvapā + bhyām → not a bha → viśvapābhyām. śaṅkhadhmā ("conch-blower") and the like decline alike.
Applies when / blocked by. "Of a root": hāhā (the name of a gandharva) ends in ā but not in a root, so its ā stays and ordinary sandhi applies:
| case | form |
|---|---|
| 2 accusative pl. | hāhān |
| 3 instrumental sg. | hāhā (ā + ā → ā, SK 85) |
| 4 dative sg. | hāhai (ā + e → ai, SK 72) |
| 5/6 ablative, genitive sg. | hāhāḥ (ā + a → ā) |
| 6/7 genitive, locative du. | hāhauḥ (ā + o → au) |
| 7 locative sg. | hāhe (ā + i → e, SK 69) |
The rest is like viśvapā. Bhaṭṭoji adds that by splitting the sūtra at "ātaḥ" the ā of a non-root is dropped in a few cases, such as ktvaḥ and śnaḥ. With this the ā-stems end; the i-stems begin with hariḥ (nominative singular, like rāmaḥ) and harī (nominative dual, i + au merged into long ī by SK 164).
hari (u-stems śambhu, guru decline alike)A stem in short i or u differs from rāma in one main way: its final vowel is strengthened to a guṇa vowel (e, o) before some endings, and it takes the instrumental nā and the locative au. To govern this, the grammar gives the i- and u-stems a name, ghi (SK 243), and writes four rules for a ghi (SK 244–247). Whatever hari does, śambhu, guru, viṣṇu and every other short-u masculine does with o for e.
Rule. An aṅga ending in a short vowel takes guṇa before jas. This extends the guṇa of SK 242 (there, only in the vocative) to the nominative plural.
Example.
hari + jas → hari + as → the final i takes guṇa → hare + as → e → ay before a vowel → harayas → s → ru → visarga → harayaḥ (nominative plural).
Rule. In the sambuddhi (vocative singular), a final short vowel takes guṇa. Then, the stem now ending in e or o, the s of the sambuddhi is dropped by SK 193.
Example.
he hari + su → i → e → he hare + s → s dropped after eṅ (SK 193) → he hare.
Accusative: hari + am → pūrvarūpa (SK 194) → harim; hari + auṭ → i + au merged into long ī (SK 164) → harī; hari + śas → harīs → s → n (SK 196) → harīn.
Rule. A stem ending in short i or u that is not named nadī is named ghi — except sakhi ("friend"). "The remainder" (śeṣa) means: those short-i/u stems left over once the nadī name (SK 266) has taken its share.
Example.
hari → ends in short i, is not nadī, is not sakhi → it is a ghi → SK 244–247 apply to it.
sakhi → excluded by name → not a ghi → declined at SK 248–256.
Applies when / blocked by. Bhaṭṭoji tests the wording. "The remainder": matyai (dative of the feminine mati taken as nadī) shows a short-i stem that is nadī, not ghi; he adds that the heading "one name only" (SK 232) would secure this anyway, so the word is there for clarity. "Short": vātapramye (long ī) is not ghi. "i or u": mātre (ṛ) is not ghi.
Rule. After a ghi, the instrumental singular ending ṭā — called āṅ by the older grammarians — is replaced by nā, provided the stem is not feminine.
Example.
hari + ṭā → ṭā → nā → hari + nā → n → ṇ after r with i (an aṭ) between (SK 197) → hariṇā (instrumental singular).
Feminine mati + ṭā → no nā → matyā.
Rule. A ghi takes guṇa before a sup affix marked with ṅ (a ṅit ending: ṅe, ṅasi, ṅas, ṅi).
Example.
hari + ṅe → hari + e → guṇa → hare + e → e → ay before a vowel → haraye (dative singular).
Applies when / blocked by. "Of a ghi": sakhi is not a ghi, so sakhye, no guṇa. "ṅit": bhyām is not ṅit, so haribhyām. "A sup": the feminine affix ṅīṣ is ṅit but not a sup, so paṭvī, no guṇa.
Rule. When the a of ṅasi or ṅas follows an eṅ (e or o), the single substitute for both vowels is the form of the prior one (pūrvarūpa). The guṇa of SK 245 has just turned hari into hare before these ṅit endings; this rule then swallows the a of the ending.
Example.
hari + ṅasi → guṇa (SK 245) → hare + as → e + a → e → hares → hareḥ (ablative singular); identically hareḥ (genitive singular).
hari + os → i → y before a vowel (iko yaṇ aci) → haryoḥ (genitive and locative dual).
hari + ām → nuṭ after a short vowel (SK 208) → hari + nām → lengthening before nām (SK 209) → harīnām → n → ṇ (SK 197) → harīṇām (genitive plural).
Rule. After a ghi in i or u, the locative singular ṅi is replaced by au (this part is the rule aut, SK 256, whose wording is carried into this sūtra), and the final vowel of the ghi is replaced by short a.
Example.
hari + ṅi → ṅi → au and i → a → hara + au → a + au → au (SK 72) → harau (locative singular).
hari + os → haryoḥ (locative dual); hari + sup → s → ṣ after i (SK 212) → hariṣu (locative plural).
The paradigm of hari. śrīpati, agni, ravi, kavi and the rest follow it; śambhu, viṣṇu, vāyu, bhānu follow it with u for i and o for e.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | hariḥ | harī | harayaḥ |
| vocative | he hare | he harī | he harayaḥ |
| 2 accusative | harim | harī | harīn |
| 3 instrumental | hariṇā | haribhyām | haribhiḥ |
| 4 dative | haraye | haribhyām | haribhyaḥ |
| 5 ablative | hareḥ | haribhyām | haribhyaḥ |
| 6 genitive | hareḥ | haryoḥ | harīṇām |
| 7 locative | harau | haryoḥ | hariṣu |
sakhi and pati — the ghi exceptions, with the vṛddhi/anaṅ/su-lopa apparatus — running examples: sakhi, patisakhi ("friend") was expressly excluded from the name ghi. Instead of guṇa it takes vṛddhi in the strong cases (sakhāyau), and its nominative singular is sakhā, built by replacing the final i with an, lengthening, and dropping both the ending and the n. Six of the ten rules here are general machinery — the names upadhā and apṛkta, the lengthening of an n-stem, the dropping of a lone consonant ending — that every consonant-stem chapter will use again. pati ("lord") is a ghi only inside a compound.
Rule. For the aṅga sakhi, the substitute anaṅ appears when su follows and the form is not a vocative. anaṅ is an with a marker ṅ; being ṅit, it replaces only the final sound of the stem (ṅic ca, SK 43).
Example.
sakhi + su → i → an → sakhan + s, carried on at SK 250 and 252.
Rule. The sound immediately before the last sound of a form is named upadhā, the penult.
Example.
sakhan → its last sound is n → the a before it is the upadhā.
Rule. The upadhā of an aṅga ending in n is lengthened before a sarvanāmasthāna that is not the sambuddhi.
Example.
sakhan + s → su is a sarvanāmasthāna, and not vocative → the upadhā a becomes ā → sakhān + s, carried on at SK 252.
Rule. An affix consisting of a single sound is named apṛkta, "unjoined". The ending su, once its marker u is discarded, is the single sound s, and so is apṛkta.
Example.
su → attaches as s, one sound → apṛkta.
Rule. After a stem ending in a consonant, or in a long ṅī or āp, the affix su, ti or si is dropped when it is apṛkta and a consonant. This su-lopa is what produces the nominative singular of every consonant-stem.
Example.
sakhān + s → the stem ends in a consonant, and s is a single-consonant affix → s is dropped → sakhān → the final n of the prātipadika-pada is dropped (SK 236) → sakhā (nominative singular).
Vocative: he sakhi + su → the substitute anaṅ is for the non-vocative only → guṇa (SK 242) and loss of the s (SK 193) → he sakhe.
Applies when / blocked by. Bhaṭṭoji tests every word. "Consonant, ṅī, āp": grāmaṇīḥ ends in none of these (its ī is not the feminine affix), so its s stays. "Long": niṣkauśāmbiḥ, atikhaṭvaḥ, whose feminine ī and ā have been shortened in the compound, keep the s. "su, ti, si": in abhaitsīt the s is the affix sic, not one of these; si means the verb-ending sip that goes with tip. "Apṛkta": in bibharti the ending ti has two sounds. "Consonant": in bibheda the ending is the vowel a. And why "consonant" first in the list, when the name hal is repeated? For rājā: the n of rājan could not be dropped otherwise, because the dropping of a cluster's last consonant (saṃyogāntasya lopaḥ) is asiddha.
Rule. After the aṅga sakhi, a sarvanāmasthāna other than the sambuddhi is treated as if it were marked with ṇ (ṇidvat). A ṇ-marked affix causes vṛddhi by the next rule.
Example.
sakhi + au → au is a sarvanāmasthāna, not vocative → it counts as ṇit → vṛddhi by SK 254.
Rule. A vowel-final aṅga takes vṛddhi before an affix marked with ñ or ṇ. For sakhi the strong endings count as ṇit (SK 253), so its i becomes ai, which is āy before a vowel.
Example.
sakhi + au → i → ai → sakhai + au → ai → āy before a vowel (eco 'yavāyāvaḥ) → sakhāyau (nominative and accusative dual).
sakhi + jas → sakhāyaḥ (nominative plural). sakhi + am → sakhāyam (accusative singular).
Since sakhi is not a ghi, no ghi operation applies: sakhyā (instrumental singular, i → y before ā), sakhye (dative singular).
Rule. The a of ṅasi or ṅas that follows khy or ty — that is, the words khi, khī, ti, tī once their vowel has become y — is replaced by u.
Example.
sakhi + ṅasi → sakhi + as → i → y before a vowel → sakhy + as → the a after khy becomes u → sakhyus → sakhyuḥ (ablative singular); identically sakhyuḥ (genitive singular).
Rule. After a ghi in i or u, the locative singular ṅi is replaced by au. The u in the sūtra's "ut" is carried forward for the sake of the next rule (SK 247, which was placed earlier in this chapter).
Example.
sakhi + ṅi → ṅi → au → i → y before the vowel → sakhyau (locative singular). The rest of sakhi is like hari.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | sakhā | sakhāyau | sakhāyaḥ |
| vocative | he sakhe | ||
| 2 accusative | sakhāyam | sakhāyau | sakhīn |
| 3 instrumental | sakhyā | sakhibhyām | sakhibhiḥ |
| 4 dative | sakhye | sakhibhyām | sakhibhyaḥ |
| 5 ablative | sakhyuḥ | sakhibhyām | sakhibhyaḥ |
| 6 genitive | sakhyuḥ | sakhyoḥ | sakhīnām |
| 7 locative | sakhyau | sakhyoḥ | sakhiṣu |
Applies when / blocked by. Compounds ending in sakhi behave in two ways. susakhi ("good friend"): the anaṅ and ṇit treatments are aṅga operations and reach a form ending in sakhi, so susakhā, susakhāyau, susakhāyaḥ; but the whole susakhi is not the word sakhi, so the exclusion "not sakhi" in SK 243 does not bite and susakhi is a ghi: susakhinā, susakhaye, susakheḥ (no u by SK 255, since guṇa rather than y has been applied), susakhau. atisakhi ("eminent friend") and the bahuvrīhi paramasakhi go the same way: atisakhā, paramasakhā, paramasakhāyau. But atisakhi in the sense "one who has surpassed a female friend (sakhī)" declines like hari: its sakhi is a by-product of shortening sakhī in the compound (SK 656), and a form produced by a rule does not count as the word directly named in anaṅ sau.
Rule. The word pati ("lord, husband") is a ghi only in a compound. On its own it is not, and so takes none of the ghi operations.
Example.
Uncompounded: patyā (instrumental singular), patye (dative singular), patyuḥ (ablative singular, by SK 255), patyuḥ (genitive singular), patyau (locative singular); the rest like hari.
In a compound: bhūpati + ṭā → ghi → nā → bhūpatinā; bhūpati + ṅe → guṇa → bhūpataye.
Watch out. pati takes no anaṅ and no vṛddhi: its nominative singular is patiḥ, its dual patī, not patā, patāyau. Bhaṭṭoji ends the section by noting that kati ("how many") is always plural, which opens the next group.
kati, tri, dvikati is a short-i stem, yet its nominative and accusative plural are simply kati, with no ending and no guṇa. To get this, the grammar names kati a ṣaṭ (SK 258–259), drops its plural endings by luk (SK 260–261), and then explains why the dropped ending does not still cause guṇa (SK 262–263). The numerals tri and dvi close the i-stems.
Rule. bahu ("many"), gaṇa ("group"), a stem ending in the affix vatu and a stem ending in the affix ḍati (such as kati) are named saṃkhyā, "number-word".
Example.
kati → ends in the affix ḍati → it is a saṃkhyā.
Rule. A saṃkhyā ending in ḍati is also named ṣaṭ. The name ṣaṭ is what the next rules act on.
Example.
kati → a saṃkhyā in ḍati → it is a ṣaṭ.
Rule. The disappearance of an affix, when it is brought about by the words luk, ślu or lup, is named luk, ślu or lup respectively. These are three named kinds of elision, distinct from plain lopa.
Example.
SK 261 says "luk" of jas and śas → the disappearance of those endings after kati is a luk.
Rule. After a stem named ṣaṭ, the endings jas and śas undergo luk.
Example.
kati + jas → luk of jas → kati (nominative plural).
kati + śas → luk of śas → kati (accusative plural).
Rule. Even when an affix has been elided, an operation that depends on it still applies, as if the affix were present. This is pratyaya-lakṣaṇa, "the mark of the affix".
Example.
kati + jas → jas has disappeared → by this rule jasi ca (SK 241) would still give guṇa, kate → but see the next rule.
Rule. When the affix has been elided by a word containing lu (luk, ślu, lup), the operation on the aṅga that the affix would have caused does not take place. The pratyaya-lakṣaṇa of SK 262 is thus shut off for aṅga operations after a luk.
Example.
kati + jas → jas dropped by luk → no guṇa of the aṅga → kati (nominative plural).
| plural | |
|---|---|
| 1 nominative | kati |
| 2 accusative | kati |
| 3 instrumental | katibhiḥ |
| 4 dative | katibhyaḥ |
| 5 ablative | katibhyaḥ |
| 6 genitive | katīnām |
| 7 locative | katiṣu |
Bhaṭṭoji adds that asmad, yuṣmad and the ṣaṭ words have one form for all three genders, and that tri ("three") is always plural: trayaḥ (nominative, with guṇa by SK 241), trīn (accusative), tribhiḥ, tribhyaḥ.
Rule. Before ām, the stem tri is replaced by traya.
Example.
tri + ām → tri → traya → traya + ām → nuṭ after a short vowel (SK 208) → traya + nām → lengthening (SK 209) → trayānām → n → ṇ after r (SK 197) → trayāṇām (genitive plural).
paramatrayāṇām ("of the highest three"). Locative plural: triṣu.
Applies when / blocked by. Some hold that traya does not replace a subordinate tri, giving priyatrīṇām; Bhaṭṭoji's own view is priyatrayāṇām. Scholars discuss this. He closes with the note that dvi ("two") is always dual.
Rule. For the pronominal stems tyad and those after it in the sarvādi list, the final sound is replaced by a when a vibhakti follows. A vārttika restricts this to the words up to dvi, so that yuṣmad, asmad, bhavatu and kim are not affected.
Example.
dvi + au → i → a → dva + au → a + au → au (SK 72) → dvau (nominative and accusative dual).
dvi + bhyām → dva + bhyām → lengthening before a yañ-initial sup (SK 202) → dvābhyām (instrumental, dative, ablative dual).
dvi + os → dva + os → a → e before os (SK 207) → dvayoḥ (genitive and locative dual).
Applies when / blocked by. "Up to dvi": bhavat keeps its t, giving bhavān, bhavantau, bhavantaḥ. As a proper name or as a subordinate member the substitution fails, because it belongs to the inner sub-list of sarvādi: a man named Dvi is dviḥ, dvī; atidvi ("who has passed the two") declines like hari; but where dvi is the main member, paramadvau. Bhaṭṭoji also notes auḍulomiḥ, auḍulomī, whose plural is uḍulomāḥ by a vārttika giving the affix a for descendants of loman in the plural: auḍulomim, auḍulomī, uḍulomān.
With this the short-i stems end. Bhaṭṭoji then runs three long-ī masculines that are not feminine and not roots, and so behave like hari wherever length allows: vātapramī ("wind-measurer", the root mā with an ī affix from the Uṇādi list), yayī ("road") and papī ("sun"). Because the stem ends in a long vowel, the nominative plural does not merge (SK 239), the genitive plural takes no nuṭ (SK 208 needs a short vowel), and the ī becomes y before au, jas, ṭā, ṅe, ṅasi, ṅas, os and ām — not before am, śas or ṅi, where the table shows vātapramīm, vātapramīn and vātapramī.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | vātapramīḥ | vātapramyau | vātapramyaḥ |
| vocative | he vātapramīḥ | ||
| 2 accusative | vātapramīm | vātapramyau | vātapramīn |
| 3 instrumental | vātapramyā | vātapramībhyām | vātapramībhiḥ |
| 4 dative | vātapramye | vātapramībhyām | vātapramībhyaḥ |
| 5 ablative | vātapramyaḥ | vātapramībhyām | vātapramībhyaḥ |
| 6 genitive | vātapramyaḥ | vātapramyoḥ | vātapramyām |
| 7 locative | vātapramī | vātapramyoḥ | vātapramīṣu |
The locative singular vātapramī is ī + i merged by akaḥ savarṇe dīrghaḥ. When vātapramī is instead formed with the affix kvip, it is a root-stem and takes y also before am, śas and ṅi (by SK 272, like pradhī): vātapramyam, vātapramyaḥ, vātapramyi. The next group opens with bahuśreyasī ("one who has many better women"), a masculine bahuvrīhi ending in the feminine ṅī; because its stem ends in a long ṅī, its su is dropped by SK 252: bahuśreyasī.
bahuśreyasī (nadī), kumārī (dhātu-ī), sudhī, pradhīThree kinds of masculine end in long ī (or ū), and they part company before the vowel-initial endings. A stem whose ī is the feminine affix — like the bahuvrīhi bahuśreyasī — is a nadī and takes the nadī operations: a short vocative, the augment āṭ before ṅit endings, nuṭ in the genitive plural and ām for the locative singular (SK 266–270). A stem that ends in a root — kumārī in the sense "one who behaves like a maiden", pradhī, unnī — turns its ī into y before a vowel (SK 272), unless it is monosyllabic (nī: iy, SK 271) or is sudhī (SK 273).
Rule. A word ending in long ī or long ū that is by nature feminine is named nadī. The sūtra's yū stands for ī and ū combined.
Example.
śreyasī ("better woman") → ends in long ī and is always feminine → it is a nadī.
bahuśreyasī (masculine, "he who has many better women") → ends in the nadī śreyasī → the nadī operations apply to it, by the vārttika below.
Applies when / blocked by. A vārttika adds "the prior gender is taken": a word that was feminine keeps its nadī status even after it has become a subordinate member of a compound, as śreyasī has in bahuśreyasī. Bhaṭṭoji's commentators note that the form yū itself does not follow the ordinary rules; scholars discuss this.
Rule. A stem meaning "mother" (ambā and its synonyms) or ending in a nadī is shortened in the sambuddhi.
Example.
he bahuśreyasī + su → the final ī becomes i → the s of the sambuddhi is dropped after a short vowel (SK 193) → he bahuśreyasi (vocative singular).
Accusative plural: bahuśreyasī + śas → bahuśreyasīs → s → n (SK 196) → bahuśreyasīn.
Rule. After a stem ending in a nadī, the ṅit endings (ṅe, ṅasi, ṅas, ṅi) receive the augment āṭ — an ā at their front, the ṭ being the marker that places it there.
Example.
bahuśreyasī + ṅe → bahuśreyasī + ā + e, finished at SK 269.
Rule. When a vowel follows the augment āṭ, the single substitute for both is the vṛddhi vowel.
Example.
bahuśreyasī + ā + e → ā + e → ai → ī → y before the vowel (iko yaṇ aci) → bahuśreyasyai (dative singular).
bahuśreyasī + ā + as → ā + a → ā → bahuśreyasyāḥ (ablative and genitive singular).
bahuśreyasī + ām → nuṭ, because the stem ends in a nadī (SK 208) → bahuśreyasīnām (genitive plural).
Rule. After a stem ending in a nadī, in āp, or in the word nī, the locative singular ṅi is replaced by ām.
Example.
bahuśreyasī + ṅi → ṅi → ām → the augment āṭ (SK 268) rather than nuṭ — āṭ wins as the later rule → bahuśreyasī + ā + ām → ā + ā → ā, ī → y → bahuśreyasyām (locative singular).
The rest of bahuśreyasī is like vātapramī (the table under SK 265). atilakṣmī ("who has surpassed Lakṣmī") differs in one point: its ī is an Uṇādi affix, not ṅī, so its su is not dropped — atilakṣmīḥ; the rest like bahuśreyasī.
Applies when / blocked by. Bhaṭṭoji now introduces a masculine kumārī: "a brāhmaṇa who desires a maiden for himself, or who behaves like a maiden". It is formed from the denominative base in kyac ("desire X") or in the ācāra-kvip ("behave like X"), with the agent affix kvip; it ends in a long ṅī, so its su is dropped by SK 252: kumārī. Because this stem ends in a root, the next three rules decide its vowel-initial forms.
Rule. Before a vowel-initial affix, the i-vowel or u-vowel of an aṅga that ends in the class-sign śnu, or is a root ending in i/ī or u/ū, or is the word bhrū ("brow"), is replaced by iyaṅ or uvaṅ respectively. Being ṅit, each replaces only the final vowel; by closest likeness, iy goes for the i-vowel and uv for the u-vowel.
Example.
kumārī + au → the stem ends in a root in ī, a vowel follows → iyaṅ would apply: kumāriyau — but the next rule steps in first.
nī + au ("leader") → monosyllabic, so SK 272 does not reach it → niyau (nominative dual).
Rule. For the i-vowel of a polysyllabic aṅga ending in a root, when that vowel is not preceded by a consonant cluster belonging to the root, the substitute before a vowel-initial affix is yaṇ (y), not iyaṅ.
Example.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | kumārī | kumāryau | kumāryaḥ |
| vocative | he kumāri | ||
| 2 accusative | kumāryam | kumāryau | kumāryaḥ |
| 4 dative | kumāryai | ||
| 5/6 ablative, genitive | kumāryāḥ | kumārīṇām | |
| 7 locative | kumāryām |
pradhī ("of superior mind"): pradhīḥ, pradhyau, pradhyaḥ; pradhyam, pradhyaḥ.
unnī ("one who leads up"): unnyau, unnyaḥ; he unnīḥ; unnyam; locative unnyām (by SK 270, nī being named there). Here the cluster nn does not block y, because only a cluster inside the root counts, and nn straddles prefix and root. Likewise grāmaṇīḥ.
Applies when / blocked by. "Polysyllabic": the monosyllabic nī keeps iyaṅ: nīḥ, niyau, niyaḥ; niyam, niyaḥ (before am and śas the later iyaṅ beats pūrvarūpa); niyām. "Not preceded by a root-internal cluster": suśriyau, yavakriyau (śr, kr belong to the roots). A vārttika adds that y is not wanted when the first member is neither a gati (a prefix construed with the action) nor a kāraka (a participant in the action): śuddhadhiyau, paramadhiyau. Bhaṭṭoji then explains durdhiyaḥ and vṛścikabhiyā: in durdhī the dur is not a gati with respect to dhī, which names no action; and in vṛścikabhī the scorpion's role as source of fear is not treated as a kāraka, or else the compound is analysed as "the fear relating to a scorpion", vṛścika-sambandhinī bhī, with the latter member of vṛścika-sambandhinī dropped (uttarapada-lopa).
Rule. For bhū and sudhī, the y/v substitution of SK 272 and SK 281 does not take place before a vowel-initial sup; iyaṅ and uvaṅ apply instead.
Example.
sudhī + au → no y → iyaṅ → sudhiyau (nominative dual); sudhiyaḥ (nominative plural), and so on.
Applies when / blocked by. Bhaṭṭoji now takes sakhī built from the denominative sakhīyati ("desires a friend") with kvip: the a and y of the denominative affix are dropped. If the dropped a counted as present, y would follow; a vārttika says that what is dropped before kvip does not count as present. Since a word changed in one part is still the same word, the anaṅ and ṇit treatments of sakhi apply to it: sakhā, sakhāyau, sakhāyaḥ; he sakhe; sakhāyam (SK 253 being later than both pūrvarūpa and y), sakhāyau; but sakhyaḥ (accusative plural, with y). The related kvip-words "who desires sakha, sukha, suta" give sakhīḥ, sukhīḥ, sutīḥ; dual sakhyau, sukhyau, sutyau; and, since SK 255 covers long ī too, sakhyuḥ, sukhyuḥ, sutyuḥ. "Who desires lūna, kṣāma, prastīma" give lūnīḥ, kṣāmīḥ, prastīmīḥ, with lūnyuḥ, kṣāmyuḥ, prastīmyuḥ. śuṣkī ("who desires śuṣka") instead takes iyaṅ: śuṣkīḥ, śuṣkiyau, śuṣkiyaḥ, ablative and genitive śuṣkiyaḥ. With this the i-stems end; śambhu declines like hari, and so do viṣṇu, vāyu, bhānu.
kroṣṭṛ, dhātṛ, pitṛ, nṛ; khalapū, varṣābhūAn ṛ-stem has three shapes: ār or ar in the strong cases (dhātāraḥ, pitaraḥ), ṛ or r elsewhere, and ā in the nominative singular (dhātā), built exactly as sakhā was. Bhaṭṭoji introduces the type through kroṣṭu ("jackal"), a u-stem that is made to behave as the ṛ-stem kroṣṭṛ in the strong cases, and optionally before the vowel-initial endings. The u-stems then close with the long-ū roots khalapū and varṣābhū, whose ū becomes v before a vowel.
Rule. Before a sarvanāmasthāna that is not the vocative, the word kroṣṭu behaves like a word ending in the agent affix tṛc — that is, the word kroṣṭṛ is used in its place. A rule that makes one thing behave like another is called an atideśa, an extension; here it is an extension of form.
Example.
kroṣṭu + au → au is a non-vocative sarvanāmasthāna → kroṣṭṛ + au, carried on at SK 277.
Rule. An aṅga ending in ṛ takes guṇa before ṅi and before a sarvanāmasthāna. The guṇa of ṛ is a, and it is followed by r (uraṇ raparaḥ), giving ar.
Example.
kroṣṭṛ + au → guṇa → kroṣṭar + au, carried on at SK 277.
kroṣṭṛ + ṅi → kroṣṭar + i → kroṣṭari (locative singular, SK 280).
Applies when / blocked by. Before su this guṇa is set aside by the next rule.
Rule. For a stem ending in ṛ, and for uśanas, purudaṃsas and anehas, the substitute anaṅ appears before a non-vocative su — the same substitute that sakhi took at SK 248.
Example.
kroṣṭṛ + su → ṛ → an → kroṣṭan + s → the upadhā is lengthened before a non-vocative sarvanāmasthāna (SK 250) → kroṣṭān + s → the s is dropped after a consonant (SK 252) → the final n is dropped (SK 236) → kroṣṭā (nominative singular).
Rule. For ap ("water"), for a stem in the affix tṛn or tṛc, and for svasṛ, naptṛ, neṣṭṛ, tvaṣṭṛ, kṣattṛ, hotṛ, potṛ and praśāstṛ, the upadhā is lengthened before a non-vocative sarvanāmasthāna. After guṇa the upadhā is the a of ar, so ar becomes ār.
Example.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | kroṣṭā | kroṣṭārau | kroṣṭāraḥ |
| 2 accusative | kroṣṭāram | kroṣṭārau | kroṣṭūn |
The accusative plural kroṣṭūn is the u-stem again: śas is not a sarvanāmasthāna, so kroṣṭu stays, and u + as merge into long ū (SK 164) with s → n (SK 196).
Applies when / blocked by. Why list naptṛ and the rest, when they end in tṛn or tṛc anyway? Bhaṭṭoji answers that, on the view that these words are derived, the list is a restriction: of Uṇādi-derived words in tṛn/tṛc, only these seven are lengthened. Hence pitṛ ("father") and bhrātṛ ("brother") are not: pitaraḥ, not pitāraḥ. udgātṛ is lengthened, because the Mahābhāṣya uses udgātāraḥ.
Rule. Before the vowel-initial endings from the instrumental onward, kroṣṭu is optionally treated as kroṣṭṛ.
Example.
kroṣṭṛ + ṭā → ṛ → r before the vowel → kroṣṭrā; in the other option, like śambhu: kroṣṭunā (instrumental singular).
kroṣṭṛ + ṅe → kroṣṭre; or kroṣṭave (dative singular).
Rule. After a stem ending in ṛ, when the a of ṅasi or ṅas follows, the single substitute for both vowels is short u, followed by r since it stands for an ṛ.
Example.
kroṣṭṛ + ṅasi → kroṣṭṛ + as → ṛ + a → ur → kroṣṭur + s, finished at SK 280.
pitṛ + ṅas → pituḥ (genitive singular); dhātṛ + ṅasi → dhātuḥ (ablative singular).
Watch out. The t in "ut" restricts the substitute to the short vowel; scholars discuss whether this is needed, and Bhaṭṭoji's commentators cite the epic form savitūḥ with a long vowel.
Rule. After r, in a pada-final cluster, only the s is dropped, not any other consonant. This restricts saṃyogāntasya lopaḥ (Chapter 5). The r then becomes visarga.
Example.
kroṣṭur + s → of the cluster rs only the s is dropped → kroṣṭur → r → visarga → kroṣṭuḥ (ablative and genitive singular, in the option where kroṣṭu is treated as kroṣṭṛ).
Genitive plural: kroṣṭu + ām → nuṭ after a short vowel → kroṣṭu + nām → lengthening (SK 209) → kroṣṭūnām.
Locative singular: kroṣṭṛ + ṅi → guṇa (SK 275) → kroṣṭari. Genitive and locative dual: kroṣṭṛ + os → ṛ → r → kroṣṭroḥ. In the other option, and before every consonant-initial ending, like śambhu: kroṣṭoḥ, kroṣṭubhyām, kroṣṭuṣu.
Applies when / blocked by. In the genitive plural the kroṣṭṛ treatment, being the later rule, would beat nuṭ and give kroṣṭṝṇām. A vārttika rules that nuṭ takes precedence over num, over the r of aci ra, and over the kroṣṭṛ treatment, by "prior counter-restriction" (pūrva-vipratiṣedha). Hence kroṣṭūnām.
With this the u-stems end. Bhaṭṭoji then gives the long-ū stems. hūhū (a gandharva), an underived word: hūhūḥ, hūhvau, hūhvaḥ; hūhūm, hūhvau, hūhūn, and so on, its ū becoming v before a vowel. aticamū ("who has surpassed an army"), ending in the nadī camū, adds the nadī operations: he aticamu, aticamvai, aticamvāḥ (ablative and genitive), aticamūnām, aticamvām. Then khalapū ("threshing-floor sweeper", the root pū).
Rule. For the u-vowel of a polysyllabic aṅga ending in a root, when that vowel is not preceded by a cluster belonging to the root, the substitute before a vowel-initial sup is v (yaṇ), not uvaṅ. This is the u-counterpart of SK 272.
Example.
khalapū + au → ū → v → khalapvau (nominative dual); khalapvaḥ (nominative plural). sulū and the like decline alike.
Applies when / blocked by. "Polysyllabic": the one-syllable lū takes uvaṅ — lūḥ, luvau, luvaḥ. "Cluster belonging to the root": in ullū the cluster ll straddles prefix and root, so v applies — ullūḥ, ullvau, ullvaḥ. "Not preceded by a cluster": in kaṭapru the pr is inside the root, so uvaṅ — kaṭapruvau, kaṭapruvaḥ. The vārttika of SK 272 applies here too: parama is neither gati nor kāraka, so paramaluvau. "A sup": the verb form luluvatuḥ keeps uv. And by SK 273, bhū takes no v: svabhūḥ, svabhuvau, svabhuvaḥ; likewise svayambhū.
Rule. The ū of varṣābhū ("frog", literally "born in the rains") becomes v before a vowel-initial sup, although bhū was excluded by SK 273.
Example.
varṣābhū + au → varṣābhvau (nominative and accusative dual); varṣābhvaḥ (nominative plural).
Applies when / blocked by. A vārttika extends the v to bhū after dṛn, kara and punar: dṛnbhvau, dṛnbhvaḥ (like khalapū); karabhvau, karabhvaḥ — or, in the reading kāra, kārabhvau, kārabhvaḥ; punarbhvau. dṛmbhū (an Uṇādi word, Uṇādi-sūtra 93) is not bhū: dṛmbhvau, dṛmbhvaḥ; dṛmbhūm, dṛmbhvau, dṛmbhūn, the rest like hūhū. dṛgbhū and kārābhū decline like svayambhū. With this the ū-stems end.
Bhaṭṭoji now returns to the ṛ-stems proper. dhātṛ ("creator", in tṛc):
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | dhātā | dhātārau | dhātāraḥ |
| vocative | he dhātaḥ | ||
| 6 genitive | dhātṝṇām |
The vocative he dhātaḥ has guṇa (SK 275) and the r as visarga. The genitive plural needs a vārttika: n becomes ṇ after an ṛ-vowel as well as after r — hence dhātṝṇām, with the lengthening of SK 209. naptṛ and the rest of the seven follow dhātṛ: udgātārau. pitṛ has no lengthening (SK 277): pitā, pitarau, pitaraḥ; pitaram, pitarau; the rest like dhātṛ. So too jāmātṛ, bhrātṛ. nṛ ("man"): nā, narau, naraḥ; he naḥ.
Rule. For nṛ, the lengthening before nām (SK 209) is optional.
Example.
nṛ + nām → nṝṇām or nṛṇām (genitive plural). With this the ṛ-stems end.
Applies when / blocked by. Bhaṭṭoji closes the vowel-stems short of o with three rare types. When the roots kṝ and tṝ are quoted as words, a paribhāṣā lets the quotation optionally behave like its original: with the root's i-substitution and r, kīḥ, kirau, kiraḥ; tīḥ, tirau, tiraḥ, declined like gīḥ (a consonant-stem); without it, since SK 275 and 276 are limited by their t to short ṛ, neither anaṅ nor guṇa applies — kṝḥ, krau, kraḥ; kṝm, krau, kṝn; krā, kre. The roots gamḷ and śakḷ quoted as words take anaṅ: gamā, śakā; where guṇa applies, the a is followed by l — gamalau, gamalaḥ, gamalam, gamalau; gamlā, gamle; and in the ablative and genitive singular, u followed by l (SK 279) with the last consonant of the cluster dropped — gamul, śakul. Stems in e: seḥ, sayau, sayaḥ; smṛteḥ, smṛtayau, smṛtayaḥ.
go, rai, glauThe chapter ends with the three stems that end in o, ai and au. go ("cow, bull") takes vṛddhi in the strong cases (gauḥ, gāvaḥ) and a bare ā before am and śas (gām, gāḥ). rai ("wealth") turns its ai into ā before a consonant (rābhyām) and into āy before a vowel (rāyaḥ). glau ("moon") needs no rule of its own.
Rule. A sarvanāmasthāna affix after the word go is treated as ṇit, and so causes vṛddhi of the stem's o (SK 254): o → au, which is āv before a vowel.
Example.
go + su → o → au → gaus → gauḥ (nominative singular).
go + au → gau + au → au → āv before a vowel → gāvau (nominative dual); go + jas → gāvaḥ (nominative plural).
Rule. The sūtra is split "ā, otaḥ": when am or śas follows a stem-final o, the single substitute for the o and the vowel of the ending is ā. Because am is named beside śas, it means only the case-ending am, not the verb-ending am.
Example.
go + am → o + a → ā → gām (accusative singular). go + auṭ → gāvau (accusative dual, by SK 284). go + śas → gās → gāḥ (accusative plural).
Then, with o → av before a vowel: gavā (instrumental singular), gave (dative singular); and, with pūrvarūpa (SK 246), goḥ (ablative singular), goḥ (genitive singular), and so on.
Applies when / blocked by. The verb forms acinavam, asunavam show o + am untouched, since their am is not a sup. A vārttika generalises SK 284: a sarvanāmasthāna prescribed after any o-final stem is ṇit — hence sudyauḥ, sudyāvau, sudyāvaḥ. The word "prescribed" keeps the vocatives he bhāno, he bhānavaḥ safe, since their o arises only by guṇa. The bahuvrīhi smṛtau ("by whom Śambhu, the u, is remembered") is the example for an au-stem declined with these rules: smṛtauḥ, smṛtāvau, smṛtāvaḥ; smṛtām, smṛtāvau, smṛtāḥ. With this the o-stems end.
Rule. For the word rai, the final sound is replaced by ā before a vibhakti beginning with a consonant. Before a vowel the ordinary ai → āy applies.
Example.
| singular | dual | plural | |
|---|---|---|---|
| 1 nominative | rāḥ | rāyau | rāyaḥ |
| 2 accusative | rāyam | rāyau | rāyaḥ |
| 3 instrumental | rāyā | rābhyām |
rai + su → su begins with a consonant → ai → ā → rās → rāḥ. rai + bhyām → rābhyām. rai + au → ai → āy → rāyau. With this the ai-stems end.
The au-stem glau needs only sandhi: glauḥ, glāvau, glāvaḥ; glāvam, glāvau, glāvaḥ, and so on, with au → āv before a vowel.
Applies when / blocked by. One might ask whether SK 285 gives glām for the accusative singular, since o and au are both back vowels. Bhaṭṭoji answers no: the Śivasūtra ai au c groups au with ai, not with o, and thereby shows that o and au are not savarṇa; a rule for o does not reach au. With this the au-stems, and the masculine vowel-stems, end.
| SK | what it does | example |
|---|---|---|
| 178 | meaningful non-root, non-affix form = prātipadika | rāma |
| 179 | kṛt-, taddhita-ending forms and compounds too | rājapuruṣa |
| 180 | heading: what follows is an affix | su … sup |
| 181 | the affix comes after the base | rāma-su |
| 182 | affixes follow a ṅī-, āp-ending or prātipadika base | rāma + su |
| 183 | the twenty-one sup endings | su, au, jas … |
| 184 | sup and tiṅ are vibhakti; prathamā … saptamī | |
| 185 | each triad: singular, dual, plural | |
| 186 | dual for two, singular for one | rāmaḥ, rāmau |
| 187 | plural for many; nom. sg. | rāmaḥ |
| 188 | one of like words remains | rāmau |
| 189 | initial cu/ṭu of an affix is a marker | jas → as |
| 190 | tu, s, m are not markers in a vibhakti | jas keeps s |
| 191 | a + guṇa vowel → pararūpa (beaten here) | rāmāḥ |
| 192 | nom. sg. in address = sambuddhi | he rāma |
| 193 | sambuddhi's s dropped after eṅ / short vowel | he rāma, he hare |
| 194 | ak + vowel of am → pūrvarūpa | rāmam |
| 195 | initial l, ś, ku of a non-taddhita affix is a marker | śas → as |
| 196 | s of śas → n after the long vowel, masculine | rāmān |
| 197 | ṇatva across aṭ, ku, pu, āṅ, num | rāmeṇa |
| 198 | no ṇatva for pada-final n | rāmān |
| 199 | the form before its affix = aṅga | rāma before su |
| 200 | heading: "of the aṅga" | |
| 201 | ṭā, ṅasi, ṅas → ina, āt, sya after a | rāmeṇa, rāmāt, rāmasya |
| 202 | a → ā before yañ-initial sup | rāmābhyām |
| 203 | bhis → ais after a | rāmaiḥ |
| 204 | ṅe → ya after a | rāmāya |
| 205 | a → e before jhal-initial plural sup | rāmebhyaḥ |
| 206 | jhal → car optionally at a pause | rāmāt / rāmād |
| 207 | a → e before os | rāmayoḥ |
| 208 | ām takes nuṭ after short vowel, nadī, āp | rāma + nām |
| 209 | vowel-final aṅga lengthened before nām | rāmāṇām |
| 210 | heading: non-final dental → retroflex | |
| 211 | heading: after iṇ or ku | |
| 212 | affix-s / substitute-s → ṣ | rāmeṣu |
| 213 | sarva etc. = sarvanāman | sarva |
| 214 | jas → śī after an a-pronoun | sarve |
| 215 | ṅe → smai | sarvasmai |
| 216 | ṅasi, ṅi → smāt, smin | sarvasmāt, sarvasmin |
| 217 | ām takes suṭ after a pronoun | sarveṣām |
| 218 | pūrva etc. pronouns in relative sense; optional before jas | pūrve / pūrvāḥ |
| 219 | sva, not kinsman/wealth | sve / svāḥ |
| 220 | antara, "outer" / "undergarment" | antare / antarāḥ |
| 221 | after the nine, smāt/smin optional | pūrvasmāt / pūrvāt |
| 222 | no sarvanāman in a bahuvrīhi | tvatkapitṛka |
| 223 | nor in a tṛtīyā compound | māsapūrvāya |
| 224 | nor in a dvandva | varṇāśrametarāṇām |
| 225 | but optionally before jas | varṇāśrametare / -āḥ |
| 226 | prathama, carama, -taya, alpa, ardha, katipaya, nema: optional before jas | prathame / prathamāḥ |
| 227 | jarā → jaras optionally before a vowel | nirjarasau |
| 228 | pāda → pad etc. optionally from śas on | padaḥ / pādān |
| 229 | su, au, jas, am, auṭ = sarvanāmasthāna | |
| 230 | stem before other endings = pada | dadbhyām |
| 231 | before y-/vowel-initial ones = bha | dataḥ |
| 232 | one name only: bha wins over pada | dataḥ, mābhyām |
| 233 | heading: "of a bha" | |
| 234 | a of final an dropped in a bha | yūṣṇaḥ |
| 235 | n → ṇ after r, ṣ in one pada | yūṣṇā |
| 236 | final n of a prātipadika-pada dropped; asiddha | yūṣabhyām |
| 237 | that a optionally dropped before ṅi, śī | yūṣṇi / yūṣaṇi |
| 238 | ahan restored optionally before ṅi | dvyahni / dvyahani / dvyahne |
| 239 | no pūrvasavarṇa after a long vowel before jas, ic | viśvapau, viśvapāḥ |
| 240 | root-final ā dropped in a bha | viśvapaḥ |
| 241 | guṇa of a short-vowel stem before jas | harayaḥ |
| 242 | guṇa in the sambuddhi | he hare |
| 243 | short i/u stem, not nadī, not sakhi = ghi | hari |
| 244 | ṭā → nā after a ghi, non-feminine | hariṇā |
| 245 | guṇa of a ghi before a ṅit sup | haraye |
| 246 | eṅ + a of ṅasi/ṅas → pūrvarūpa | hareḥ |
| 247 | ṅi → au and ghi-final → a | harau |
| 248 | sakhi: final → an before su | sakhan- |
| 249 | the sound before the last = upadhā | |
| 250 | upadhā of an n-stem lengthened before strong endings | sakhān- |
| 251 | single-sound affix = apṛkta | s |
| 252 | apṛkta consonant su/ti/si dropped after consonant, long ṅī/āp | sakhā, rājā |
| 253 | sakhi's strong endings are ṇidvat | |
| 254 | vṛddhi before ñit/ṇit | sakhāyau |
| 255 | a of ṅasi/ṅas → u after khy, ty | sakhyuḥ |
| 256 | ṅi → au after i/u | sakhyau |
| 257 | pati is ghi only in a compound | patyā, bhūpatinā |
| 258 | bahu, gaṇa, -vatu, -ḍati = saṃkhyā | kati |
| 259 | -ḍati numeral = ṣaṭ | kati |
| 260 | luk, ślu, lup named | |
| 261 | luk of jas, śas after a ṣaṭ | kati |
| 262 | elided affix still conditions operations | |
| 263 | not after luk, for aṅga operations | kati (no guṇa) |
| 264 | tri → traya before ām | trayāṇām |
| 265 | tyad … dvi: final → a before a vibhakti | dvau, dvābhyām |
| 266 | feminine word in ī/ū = nadī | śreyasī |
| 267 | nadī / "mother" stem shortened in the sambuddhi | he bahuśreyasi |
| 268 | ṅit endings take āṭ after a nadī | |
| 269 | āṭ + vowel → vṛddhi | bahuśreyasyai |
| 270 | ṅi → ām after nadī, āp, nī | bahuśreyasyām |
| 271 | i/u-vowel → iy/uv before a vowel (roots, śnu, bhrū) | niyau |
| 272 | polysyllabic root-stem: ī → y instead | kumāryau, pradhyau |
| 273 | not for bhū, sudhī | sudhiyau, svabhuvau |
| 274 | kroṣṭu = kroṣṭṛ before strong endings | kroṣṭārau |
| 275 | guṇa of ṛ before ṅi and strong endings | kroṣṭari, pitaraḥ |
| 276 | ṛ-stem, uśanas etc.: final → an before su | kroṣṭā, dhātā |
| 277 | upadhā of ap, tṛn/tṛc, svasṛ, naptṛ … lengthened | kroṣṭāraḥ, dhātāraḥ |
| 278 | kroṣṭṛ optional before vowel-initial oblique endings | kroṣṭrā / kroṣṭunā |
| 279 | ṛ + a of ṅasi/ṅas → ur | pituḥ, kroṣṭuḥ |
| 280 | after r only the s of a final cluster drops | kroṣṭuḥ |
| 281 | polysyllabic root-stem: ū → v before a vowel-initial sup | khalapvau |
| 282 | varṣābhū: ū → v | varṣābhvau |
| 283 | nṛ: lengthening before nām optional | nṝṇām / nṛṇām |
| 284 | strong endings after go are ṇidvat | gauḥ, gāvaḥ |
| 285 | o + am/śas → ā | gām, gāḥ |
| 286 | rai: ai → ā before a consonant | rāḥ, rābhyām; glauḥ |