Chapter 3 taught you how two vowels that meet in close utterance combine. This chapter teaches the exceptions: the vowels that keep their own form (prakṛti) and refuse to combine. After it you will be able to recognise a pluta vowel (a vowel protracted to three mātrās) and a pragṛhya vowel (a vowel that Pāṇini names exempt from sandhi), to say which of the sandhi rules of chapter 3 each of them blocks, and to give the alternative forms where the exemption is optional. You will also learn where a pluta is made in the first place, when a pluta is told to behave as if it were not one, and two small rules that supply a substitute or a nasal vowel at the edge of this territory.
harī etau, ehi kṛṣṇa3 atraOne sūtra governs the whole chapter. It says that two kinds of vowel, the pluta and the pragṛhya, never combine with a vowel that follows. Everything after it either defines those two kinds or adjusts the exemption. In harī etau ("these two Haris") the final ī of the dual harī stays as it is instead of becoming y. In ehi kṛṣṇa3 atra gauś carati ("Come, Kṛṣṇa! Here a cow is grazing") the protracted a3 of kṛṣṇa3 does not merge with the a of atra.
Rule. A pluta vowel (three mātrās, written 3 after the vowel) and a pragṛhya vowel (a vowel given that name by SK 100–109) invariably (nityam) keep their own form when any vowel (ac) follows. Keeping one's own form is prakṛtibhāva: no sandhi of any kind takes place. The word prakṛtyā, "in its own form", is not written in this sūtra; it comes in by anuvṛtti from an earlier rule. Bhaṭṭoji says the two kinds of vowel "will be defined later": the pluta by SK 93–97, the pragṛhya by SK 100–109.
Example.
ehi kṛṣṇa3 atra → the final a of kṛṣṇa is pluta (SK 95, calling from afar) → a3 + a would take savarṇadīrgha → pluta blocks it → ehi kṛṣṇa3 atra.
harī + etau → the ī of the dual harī is pragṛhya (SK 100) → ī + e would take yaṇ → pragṛhya blocks it → harī etau.
Applies when / blocked by. Only when a vowel follows; before a consonant there is no vowel sandhi to block. Why nityam, "invariably"? Because SK 91 is a rival: it also keeps a pada-final ī in its own form, but shortens it as well. SK 91 is the later sūtra, so by paratva it would win, and harī etau would lose its long ī. The word nityam says that the bare prakṛtibhāva of this rule — no shortening — is the one that must apply. Bhaṭṭoji's commentators put it this way: only the bare prakṛtibhāva of SK 90 is fit to apply here, and nityam is what makes it apply.
Watch out. prakṛtibhāva is not a change; it is the absence of one. When you derive a form, you still write the step, because it records that a rule of chapter 3 was set aside and why.
cakri atra, brahma ṛṣiḥThese two rules are the optional cousins of SK 90. A vowel at the end of a pada may, in Śākalya's view, stay uncombined — but it is shortened at the same time. Each rule therefore gives two correct forms: the kept-and-shortened one, and the ordinary sandhi form. In cakrī atra ("the discus-bearer here") the ī may stay and shorten, cakri atra, or take yaṇ, cakryatra. In brahmā ṛṣiḥ ("Brahmā, the seer") the ā may stay and shorten, brahma ṛṣiḥ, or take guṇa, brahmarṣiḥ.
Rule. In Śākalya's view, an ik vowel (i u ṛ ḷ, long or short) at the end of a pada, before a dissimilar (asavarṇa) vowel, keeps its own form and becomes short (hrasva). Because a teacher is named, the rule is optional: the other alternative is ordinary yaṇ.
Example.
cakrī + atra → ī is pada-final ik, a is a dissimilar vowel → Śākalya: ī stays and is shortened to i → cakri atra.
cakrī + atra → the other alternative: iko yaṇ aci → ī becomes y → cakryatra.
Applies when / blocked by. The vowel must end a pada. In gauryau ("two fair ones") the ī of gaurī stands before the ending au, inside the word, so yaṇ is obligatory. Two vārttikas narrow the rule further. na samāse: not in a compound (samāsa) — vāpyām aśvaḥ ("a horse in the tank") compounds to vāpyaśvaḥ, with yaṇ and no option. siti ca: not before a sit affix, one whose marker (it) is s — in pārśvam ("the ribs"), formed from parśu with the sit affix ṆaS, the first vowel of parśu takes vṛddhi (ā) and the final u takes yaṇ (v), giving pārśvam; Śākalya's option does not arise. Bhaṭṭoji's commentators note that these two are one vārttika in the original, which Bhaṭṭoji explains as two.
Watch out. The sūtra prescribes only shortening; it does not say "and no yaṇ". Bhaṭṭoji, following the Bhāṣya, explains that shortening by itself implies keeping the vowel: if Pāṇini had wanted y here he would simply have prescribed yaṇ. So the ca need not be read as carrying prakṛtyā forward; the result is the same.
Rule. When ṛ follows, a pada-final ak vowel is treated as in SK 91: optionally shortened and kept in its own form. ak is the pratyāhāra from a to the marker k of the second Śivasūtra, so it adds a and ā to the ik vowels. The words hrasva, ca (bringing prakṛtyā with it) and padānta ("pada-final") are carried forward; asavarṇe is not.
Example.
brahmā + ṛṣiḥ → ā is pada-final ak, ṛ follows → ā is shortened and kept → brahma ṛṣiḥ.
brahmā + ṛṣiḥ → the other alternative: ād guṇaḥ gives ar (the r following the guṇa a) → brahmarṣiḥ.
Applies when / blocked by. SK 91 could not reach brahmā, because ā is not an ik; this rule exists for exactly such cases. Only a pada-final vowel qualifies: in ā ṛcchat ("he went"), the ā is the augment āṭ prefixed to the verb, not the end of a pada, so vṛddhi runs and gives ārcchat. Unlike SK 91, this rule applies even inside a compound; the vārttika na samāse does not extend here: sapta + ṛṣīṇām gives saptaṛṣīṇām (kept) beside saptarṣīṇām (guṇa).
devadatta3SK 90 assumed that some vowels are pluta. This group is where Bhaṭṭoji makes them. A heading rule (SK 93) says what is protracted — the ṭi of a sentence — and the four rules after it say when: in a return-greeting, in calling from afar, on the particles hai and he, and on the heavy vowels of a name. The running example is the name devadatta, called out with its last a stretched to devadatta3.
Rule. The ṭi of a sentence (vākya) becomes pluta and udātta. The ṭi of a word is its last vowel together with whatever follows that vowel. This sūtra is an adhikāra: it prescribes nothing by itself but is read into SK 94–97, each of which names an occasion. The word padasya, "of a pada", is itself carried into it from earlier, so the ṭi in question is that of a pada.
Example.
saktūn piba devadatta3 → the last pada of the sentence is devadatta → its ṭi is the final a → pluta and udātta: devadatta3 (the occasion is supplied by SK 95).
grāmaṃ gacchāgnici3t → the last pada ends in a consonant → its ṭi is it, the i plus the t → the i is pluta: agnici3t.
Applies when / blocked by. Two words earn their place. vākyasya: without it, "the ṭi of a pada" would protract every word of the sentence; with it, only the last word qualifies. ṭeḥ: without it, the rule would read "the vowel at the end of the sentence", and a sentence ending in a consonant, like gacchāgnicit, would get no pluta; naming the ṭi reaches the vowel inside it.
Rule. In a return-greeting (pratyabhivāda) addressed to a non-śūdra — a twice-born, dvija — the ṭi of the sentence is pluta and udātta. A pratyabhivāda is the benediction pronounced, according to rule, in answer to a greeting made according to rule.
Example.
Greeting: abhivādaye devadatto 'ham ("I, Devadatta, greet you").
Reply: bho āyuṣmān edhi devadatta → the ṭi of the last pada is the a of devadatta → bho āyuṣmān edhi devadatta3 ("Long life to you, Devadatta!").
Applies when / blocked by. Two vārttikas adjust the rule, and Bhaṭṭoji adds a remark of his own. striyāṃ na: not when the reply is to a woman — abhivādaye gārgy aham is answered bho āyuṣmatī bhava gārgi, with no pluta. Bhaṭṭoji adds that the pluta is wanted only where a personal name or a family name (gotra) ends the reply; in the bare āyuṣmān edhi the i of edhi is not protracted, being neither. bho-rājanya-viśāṃ vā: for the word bhoḥ, and for the names of a kṣatriya or a vaiśya, the pluta is optional — āyuṣmān edhi bhoḥ3, āyuṣmān edhīndravarma3n, āyuṣmān edhīndrapālita3, each also correct without the 3. For bhoḥ the rule would not otherwise apply at all; for the two names it would apply invariably; vā makes both optional. Why "non-śūdra"? The Bhāṣya's example of a reply to a śūdra is kuśaly asi tuṣajaka ("Are you well, Tuṣajaka?"), with no pluta.
Watch out. In indravarma3n the pluta falls on the a before the final n. That is SK 93 at work: the ṭi is an, and the vowel inside it is protracted. The smṛtis that regulate the greeting and the reply are cited by Bhaṭṭoji's commentators; scholars discuss how far they and the Bhāṣya agree.
Rule. Also when someone is addressed (sambodhana) from afar, the ṭi of the sentence is pluta and udātta. "Afar" means a distance at which the hearer cannot catch ordinary speech and extra effort of voice is needed.
Example.
saktūn piba devadatta ("Drink the gruel, Devadatta!") → said from afar → the ṭi of devadatta is its final a → saktūn piba devadatta3.
Applies when / blocked by. Bhaṭṭoji reads hūta, "calling", as sambodhana, address in general, not as the act of summoning. If it meant summoning, the rule would need an explicit word of calling, as in ehi devadatta ("come, Devadatta"); read as address, it applies to any sentence spoken to someone far off, as the example shows.
Rule. When the particles hai or he are used in a sentence of calling from afar, the pluta falls on hai or he alone. hai and he are two indeclinables (avyaya) that signal address. This is a niyama, a restricting rule: the pluta was already available by SK 95 and SK 97, and this sūtra fixes where it may go.
Example.
he rāma → he is used, calling from afar → the pluta goes to he → he3 rāma.
rāma hai → likewise → rāma hai3.
pāhi hai rāma → only hai is protracted, not the a of rāma → pāhi hai3 rāma.
Applies when / blocked by. Without this rule, SK 97 would allow the final vowel of another word to be protracted too; haihayoḥ, "of hai and he", shuts that out. The word prayoga, "employment", is also necessary. Without it, the heading vākyasya ṭeḥ would confine the pluta to a sentence-final hai, as in rāma hai3; with it, hai and he are protracted even when they are not final, as in he3 rāma.
Rule. In the view of the Eastern grammarians (prācām), in a sentence of calling from afar, each single heavy (guru) vowel of the word addressed, other than ṛ, may be protracted even when it is not the final vowel — one vowel at a time. Because a school is named, the rule is optional.
Example.
ehi devadatta (the ehi, "come", must be understood, so that there is a sentence) → the heavy e is protracted → de3vadatta.
ehi devadatta → the heavy a before the cluster tt is protracted → devad3atta.
ehi devadatta → the final a is protracted (by SK 93 and 95, whether heavy or not) → devadatta3.
Applies when / blocked by. Why guru? So that the light a after v in devadatta is never protracted. Why "other than ṛ"? In kṛṣṇa3 only the final a can be pluta; the heavy ṛ is excluded. Why ekaika, "each single one"? For alternation: one heavy vowel is protracted in each utterance, never all of them at once. The final vowel gets its pluta from the heading ṭeḥ of SK 93; this rule adds the non-final heavy ones.
Watch out. Bhaṭṭoji splits the sūtra in two (yoga-vibhāga): the first rule ends at ekaikasya, and prācām stands as a rule on its own into which only pluta is carried. Its meaning is then "pluta is in the opinion of the Easterners", and the effect is that every pluta in the grammar becomes optional. That is why no separate vārttika is needed to make all pluta optional.
suśloketi, cinuhītiA pluta vowel blocks sandhi by SK 90. Two rules now say that in a particular position it stops blocking: sandhi runs as if the vowel were an ordinary one. The protraction is cancelled only where sandhi actually runs; where no sandhi applies, the vowel is still heard as pluta. The trigger is the word iti that the makers of the word-by-word recitation (padapāṭha) add after a Vedic word. So the Vedic suśloka3 iti becomes suśloketi, and cinuhī3 iti may become cinuhīti.
Rule. When an upasthita follows, a pluta behaves like a non-pluta (aplutavat). The upasthita is the non-Vedic word iti — the iti that the padakāras, the makers of the padapāṭha, insert after a word at the word-separation. "Like a non-pluta" means: it undergoes yaṇ and the other sandhi operations.
Example.
suśloka3 iti → iti here is non-Vedic, inserted by the padakāra → the pluta a3 behaves like a → ād guṇaḥ: a + i gives e → suśloketi.
Applies when / blocked by. The Taittirīya recension reads the mantra with the pluta suśloka3; the padapāṭha adds iti, and there the sandhi form suśloketi is found in the Bhāṣya. Bhaṭṭoji's commentators add that padakāras who instead separate suśloka3 iti suśloka3 iti are not wrong, since the sandhi rules apply only in saṃhitā.
Watch out. Why vat, "like"? If the sūtra said simply "a non-pluta replaces it", the protraction itself would be removed. Then in agnī3 iti — the imitation of a vocative dual, whose ī is pragṛhya by SK 100 — the three-mātrā ī would no longer be heard. With vat, only the pluta's operation, its prakṛtibhāva by SK 90, is cancelled. The ī stays apart from iti by virtue of being pragṛhya, and it is still pronounced pluta. Scholars discuss the wording of Bhaṭṭoji's gloss here, which is transmitted in two readings.
Rule. In Cākravarmaṇa's view, a pluta ī3 before any vowel behaves like a non-pluta. Because a teacher is named, the rule is optional. aplutavat is carried forward from SK 98, and aci, "when a vowel follows", from iko yaṇ aci; upasthite is not carried forward, so the following word need not be iti.
Example.
cinuhī3 iti → ī3 before a vowel → Cākravarmaṇa: behaves like ī → yaṇ → cinuhīti; or the pluta keeps its own form → cinuhī3 iti.
cinuhī3 idam → the same option before a vowel other than iti → cinu hīdam or cinuhī3 idam.
Applies when / blocked by. The option holds in both settings. Before iti, SK 98 would have made the non-pluta behaviour obligatory; before any other vowel, SK 98 would not have applied at all. This sūtra, framed for both, makes both optional. Where does the pluta of hī come from? cinu hi ("gather, indeed") is the reply to a question, and the hi in a reply to a question is optionally pluta by a rule outside this chapter (A 8.2.93); idam begins a separate sentence, so hi is sentence-final.
harī etau, viṣṇū imauSK 90 assumed the name pragṛhya. This group gives it: nine sūtras of the first quarter of the Aṣṭādhyāyī (A 1.1.11–19), each a saṃjñā rule naming some class of vowels pragṛhya. None of them changes anything; the work is always done afterwards by SK 90, which keeps a pragṛhya vowel apart from a following vowel. The running examples are the duals harī etau and viṣṇū imau, in which the ī and the ū do not become y and v.
Rule. A dual form (dvivacana) ending in ī, ū or e is named pragṛhya. The compound īdūdet qualifies dvivacana, so the tadanta reading applies: "ending in ī/ū/e", not "being ī/ū/e".
Example.
harī + etau → harī is a dual ending in ī → pragṛhya → SK 90 blocks yaṇ → harī etau.
viṣṇū + imau → a dual ending in ū → pragṛhya → no yaṇ → viṣṇū imau.
gaṅge + amū → a dual ending in e → pragṛhya → no ay substitution → gaṅge amū.
pacete + imau → the verb's dual ending āte has become ete, which ends in e → pragṛhya → pacete imau.
Applies when / blocked by. The last example shows why the tadanta reading matters. The dual ending of pacete is not itself the vowel e; it merely ends in e. Read as "a dual that is e", the rule would miss it. Bhaṭṭoji's commentators note that the word dvivacana itself does not take the tadanta reading, because in a saṃjñā rule the mention of an affix does not extend to what ends in it.
Watch out. In the epic phrase maṇīvoṣṭrasya one might think maṇī + iva and ask why the ī did not stay apart as a pragṛhya. Bhaṭṭoji answers that the word here is the particle va (or vā), meaning "like", so the form is maṇī + va and no vowel sandhi is in question at all.
Rule. The ī and ū of the pronoun adas ("that") that follow its m are pragṛhya. īd-ūt and pragṛhyam are carried forward from SK 100; et and dvivacana are not, because an e after the m of adas never occurs, and the ī after that m is a plural.
Example.
amī + īśāḥ → amī is the nominative plural of adas: ade becomes amī by two rules of the Tripādī (A 8.2.81, e → ī; A 8.2.80, d → m) → the ī follows the m → pragṛhya → no savarṇadīrgha → amī īśāḥ ("those lords").
rāmakṛṣṇāv amū + āsāte → amū is the nominative dual: adau becomes amū by A 8.2.80 → the ū follows the m → pragṛhya → rāmakṛṣṇāv amū āsāte ("those two, Rāma and Kṛṣṇa, sit").
Applies when / blocked by. SK 100 could not reach either form. amī is a plural, not a dual. amū is a dual, but its ū and m arise in the Tripādī, which is asiddha for SK 100 (chapter 1, SK 12); to SK 100 the word is still adau, which ends in au. This sūtra is different: its own word māt, "after the m", presupposes that the m has been made, so for it the Tripādī changes count. Why māt? Consider amuke 'tra, with the infix aka inside the pronoun: here the e does not follow the m, and it is not pragṛhya. Without māt, the e of SK 100 would have been carried forward along with ī and ū, and the e of amuke would wrongly have been named.
Watch out. Bhaṭṭoji writes rāmakṛṣṇāv amū to make clear that the masculine is meant. The feminine and neuter duals are also amū, but there a dual e is present before the Tripādī changes, and SK 100 already names them.
Rule. A form ending in the affix śe is pragṛhya. The affix śe is met in Vedic forms such as asme, "to us".
Example.
asme + indrābṛhaspatī → asme ends in śe → pragṛhya → no ay substitution → asme indrābṛhaspatī ("Indra and Bṛhaspati, to us …").
Rule. A particle (nipāta) consisting of a single vowel (ekāc), except āṅ, is pragṛhya. āṅ is the ā that carries the marker ṅ; the marker is never heard, so the rule tells you how to know it by sense.
Example.
i + indraḥ → i, a particle of astonishment, is a single vowel → pragṛhya → no savarṇadīrgha → i indraḥ.
u + umeśaḥ → u, a particle of deliberation → pragṛhya → u umeśaḥ.
ā + evaṃ nu manyase → ā of a resumed sentence ("Ah, so now you think this"), not ṅit → pragṛhya → ā evaṃ nu manyase; likewise ā evaṃ kila tat (ā of recollection).
ā + uṣṇam → ā in the sense "a little" is āṅ, marked ṅ → not pragṛhya → ād guṇaḥ → oṣṇam ("slightly hot").
Applies when / blocked by. Bhaṭṭoji quotes the verse that sorts the two *ā*s: the ā meaning "a little", the ā joined to an action, and the ā of limit or of inclusion are ṅit; the ā that changes the drift of a sentence already begun, and the ā of recollection, are not ṅit (aṅit). Only the latter two are pragṛhya.
Watch out. ekāc must be read as "one, and a vowel" — the particle is a single vowel. If it were read "having one vowel", a particle like pra would qualify, and pra + idam could not become predam.
Rule. A particle (nipāta) ending in o is pragṛhya. nipāta is carried forward from SK 103.
Example.
aho + īśāḥ → aho is a particle ending in o → pragṛhya → no av substitution → aho īśāḥ ("Ah, the lords!").
Rule. In Śākalya's view, an o that owes its existence to the vocative singular ending (sambuddhi) is pragṛhya before a non-Vedic iti. The rule is optional, since a teacher is named. ārṣa means "found in the Veda"; anārṣe, "in non-Vedic usage".
Example.
viṣṇo + iti → the o of viṣṇo arose in the vocative (by guṇa of the short final, A 7.3.108) → Śākalya: pragṛhya → viṣṇo iti.
viṣṇo + iti → the other alternative: the o is not pragṛhya → the av substitution → viṣṇav iti.
Applies when / blocked by. viṣṇo ends in o but is not a particle, so SK 104 cannot name it; this rule is needed. Why anārṣe? In the Vedic brahmabandhav ity abravīt ("he said, 'O kinsman of brāhmaṇas'") the o is not pragṛhya, and av appears. Bhaṭṭoji lists a third form, viṣṇa iti, without saying which alternative it belongs to; his commentators do not spell out its derivation.
Rule. The particle uñ — the u whose marker is ñ — is optionally pragṛhya before iti. Bhaṭṭoji says "as stated before": the words śākalyasya and itau of SK 105 are carried forward, and Śākalya's name brings the option.
Example.
u + iti → uñ before iti → pragṛhya → u iti.
u + iti → the other alternative: not pragṛhya → yaṇ → v iti.
Applies when / blocked by. By SK 103, uñ, a single-vowel particle, is already pragṛhya invariably. This rule turns that into an option before iti, and that is what makes the sandhi form v iti possible at all.
Watch out. Bhaṭṭoji's commentators treat the received text uñ ūm̐ as one sūtra that is explained as two, SK 106 and SK 107. Read as a single sūtra it would yield only ūm̐ iti and u iti; the split is what admits the third form v iti.
Rule. For uñ before iti, the substitute ūm̐ — long, nasal (anunāsika), and itself pragṛhya — optionally occurs.
Example.
u + iti → uñ before iti → the substitute ūm̐ → being pragṛhya, it stays apart → ūm̐ iti.
Applies when / blocked by. With SK 106 this gives three correct forms for the particle before iti: ūm̐ iti (substitute made), u iti (no substitute, pragṛhya by SK 106), and v iti (no substitute, not pragṛhya, yaṇ).
Rule. A pada ending in ī or ū that stands in the sense of the locative (saptamī) — its locative ending having been dropped — is pragṛhya. The dropping (luk) of the case-ending is by supāṃ suluk (SK 3561). pragṛhyam is carried forward and, since the sūtra speaks of two vowels, is read as a dual.
Example.
somo gaurī + adhiśritaḥ → gaurī stands for gauryām, "in the gaurī", its locative ending dropped → pragṛhya → no yaṇ → somo gaurī adhiśritaḥ ("the soma is set in the gaurī").
māmakī tanū + iti → māmakī tanū stands for māmakyāṃ tanvām, "in my body" → pragṛhya → māmakī tanū iti; in the padapāṭha māmakī iti, tanū iti.
Applies when / blocked by. Why artha, "in the sense of", rather than simply "in the locative"? To exclude a word that has passed into a different meaning inside a compound (vṛtti). vāpyām aśvaḥ compounds to vāpyaśvaḥ: the first member vāpī has lost its locative ending, but it no longer rests in the bare locative sense — it now bears on the horse as well — so it is not pragṛhya, and yaṇ runs. A form that keeps its locative ending, such as vātapramī, is not an example either, because the base itself does not then carry the locative sense.
kimvuktam, dadhiṃTwo rules from the Tripādī close the chapter. The first supplies a v for the particle u in a place where SK 103 would otherwise forbid any sandhi: kimu uktam may become kimvuktam. The second lets a short vowel that is not pragṛhya be pronounced through the nose at a pause: dadhi may be said dadhiṃ. With it Bhaṭṭoji's section on vowel sandhi ends.
Rule. After a maY consonant, the particle uñ optionally becomes va when a vowel follows. maY is the pratyāhāra read off the Śivasūtras from ma to the marker y; the m of kim is one. aci, "when a vowel follows", is carried forward from an earlier sūtra.
Example.
kimu + uktam → the u is uñ, after the m of kim, before a vowel → optionally va → kimvuktam ("what was said?"); otherwise kimu uktam.
Applies when / blocked by. The rule is needed because uñ is invariably pragṛhya by SK 103, so yaṇ could never give the v. Why is it placed here, in the Tripādī, with the extra word va, rather than next to iko yaṇ aci as a simple option of yaṇ? Because of asiddha. Placed early, the v would be visible to the rule mo 'nusvāraḥ, which turns m into anusvāra when v follows, and the m of kim would change. Placed in the Tripādī, after that anusvāra rule, the va substitution is asiddha for it, and the m stays.
Rule. An aṇ vowel that is not pragṛhya optionally becomes anunāsika — nasal — at a pause (avasāna). The words vā, "optionally", and avasāne are carried forward. Here aṇ is formed with the first ṇ of the Śivasūtras, so it means a, i, u only, on the strength of the traditional exposition.
Example.
dadhi at a pause → its final i is an aṇ and not pragṛhya → optionally nasal → dadhiṃ (the ṃ marks the nasal i) or dadhi.
agnī at a pause → the ī of the dual is pragṛhya (SK 100) → not nasalised → agnī.
Applies when / blocked by. Why "not pragṛhya"? So that a pragṛhya vowel such as the ī of agnī stays as it is; the exemption of this chapter is mirrored here. Because aṇ is read as a i u, the final ṛ of kartṛ is not nasalised.
Watch out. This sūtra closes Bhaṭṭoji's section on vowel sandhi (ac-sandhi). Bhaṭṭoji's commentators remark that the very name ac-sandhi shows no sandhi between ac and sandhi; the grammar allows a technical word to be left uncombined to avoid ambiguity.
| SK | Sūtra | What it does | Example |
|---|---|---|---|
| 90 | plutapragṛhyā aci nityam | A pluta or pragṛhya vowel invariably keeps its form before a vowel. | harī etau; ehi kṛṣṇa3 atra |
| 91 | iko 'savarṇe śākalyasya hrasvaś ca | Pada-final ik before a dissimilar vowel optionally stays and shortens; not in a compound, not before a sit affix. | cakri atra / cakryatra |
| 92 | ṛtyakaḥ | Pada-final ak before ṛ optionally stays and shortens, even in a compound. | brahma ṛṣiḥ / brahmarṣiḥ |
| 93 | vākyasya ṭeḥ pluta udāttaḥ | Heading: the ṭi of a sentence is pluta and udātta. | gacchāgnici3t |
| 94 | pratyabhivāde 'śūdre | …in a return-greeting to a non-śūdra; not to a woman; optional for bhoḥ and kṣatriya/vaiśya names. | āyuṣmān edhi devadatta3 |
| 95 | dūrād dhūte ca | …in address from afar. | saktūn piba devadatta3 |
| 96 | haiheprayoge haihayoḥ | When hai/he are used, only they are pluta. | he3 rāma; rāma hai3 |
| 97 | guror anṛto 'nantyasyāpy ekaikasya prācām | Each heavy vowel other than ṛ, even non-final, optionally pluta, one at a time. | de3vadatta / devad3atta / devadatta3 |
| 98 | aplutavad upasthite | Before non-Vedic iti a pluta behaves as a non-pluta. | suśloka3 iti → suśloketi |
| 99 | ī3 cākravarmaṇasya | A pluta ī3 before any vowel optionally behaves as a non-pluta. | cinuhī3 iti / cinuhīti |
| 100 | īdūded dvivacanaṃ pragṛhyam | A dual ending in ī/ū/e is pragṛhya. | harī etau; gaṅge amū; pacete imau |
| 101 | adaso māt | The ī/ū of adas after its m are pragṛhya. | amī īśāḥ; amū āsāte |
| 102 | śe | A form ending in the affix śe is pragṛhya. | asme indrābṛhaspatī |
| 103 | nipāta ekāj anāṅ | A single-vowel particle, except āṅ, is pragṛhya. | i indraḥ; ā evaṃ; but oṣṇam |
| 104 | ot | A particle ending in o is pragṛhya. | aho īśāḥ |
| 105 | sambuddhau śākalyasyetāvanārṣe | A vocative o before non-Vedic iti is optionally pragṛhya. | viṣṇo iti / viṣṇav iti |
| 106 | uñaḥ | The particle uñ is optionally pragṛhya before iti. | u iti / v iti |
| 107 | ūm̐ | uñ before iti optionally becomes the pragṛhya ūm̐. | ūm̐ iti |
| 109 | īdūtau ca saptamyarthe | A pada in ī/ū standing in locative sense (ending dropped) is pragṛhya; not inside a compound. | somo gaurī adhiśritaḥ; but vāpyaśvaḥ |
| 108 | maya uño vo vā | uñ after a maY consonant optionally becomes va before a vowel; asiddha, so no anusvāra. | kimu uktam → kimvuktam |
| 110 | aṇo 'pragṛhyasyānunāsikaḥ | A non-pragṛhya a/i/u at a pause is optionally nasal. | dadhiṃ / dadhi; but agnī |