Every earlier chapter built the sounds of a word. This chapter teaches its accent (svara): which vowel is pronounced high, which low, and which falls from high to low. Accent is heard above all in the recitation of the Veda, so most examples are mantras. The chapter has 416 sūtras: SK 3650 to SK 3978, with the 87 phiṭ-sūtras of Śāntanava read after SK 3707. This first stretch, SK 3650–3701, has three blocks, each with its own heading in the print. First come the "common accents" (sādhāraṇa-svara, SK 3650–3670), which act inside a word and across words. Next come the root accents (dhātu-svara, SK 3671–3679). Last come the first word accents (SK 3680–3701), under the print's heading atha pratyayasvarāḥ, "now the accents of affixes". After this stretch you will be able to say why gopāyátaṃ naḥ has its one udātta on ya. You will know why a vocative is ágne at the head of a line but wholly low in imáṃ me gaṅge. You will see why kva + avaram gives kvā́varam, and why a monotone replaces the accents in a rite. You will also place the udātta in root forms such as dádati and juhóti, and in words such as pākáḥ, pañcábhiḥ, paúṃsyā, dáttaḥ, máma and kā́myā.
á, ā́. An independent svarita, one that a rule prescribes in its own right, is marked with a grave: à, ā̀.dáttaḥ. An antodātta word has it on its last vowel: pākáḥ.o3.kvà. SK 3730 makes certain verb-endings anudātta. SK 3935 (tiṅṅatiṅaḥ) makes a finite verb wholly anudātta after a non-verb.gopāya + tam → gopāyátaṃ naḥ (udātta on ya, everything else anudātta); deva + ṅīp → (udātta a elided) devī́m (udātta on vī)A word is put together from pieces, and many pieces bring an udātta of their own. Which one survives? SK 3650 answers: the one a rule prescribes, and all the other vowels go low. SK 3651 and SK 3652 then say what happens when an udātta vowel is elided. The running examples are gopāyátaṃ naḥ "[you two] guard us", with its udātta on ya, and devī́m, with its udātta on vī.
Rule. This is a paribhāṣā, a rule that tells us how other rules work. Its domain is every rule that prescribes an accent (svaravidhi). When such a rule makes one vowel of a pada udātta or svarita, all the other vowels of that pada are anudātta.
Example.
gup + āya → gopāya is a dhātu by SK 2304
→ the dhātu accent (SK 3671) makes the final a of ya udātta
→ this rule makes every other vowel anudātta
→ gopāyátaṃ naḥ, udātta on ya.
So the root gup loses its own udātta, and āya loses the udātta of its affix accent.
Applies when / blocked by. A vārttika, a supplementary rule that Bhaṭṭoji quotes, adds that an accent taught when another is already present (satiśiṣṭa) is the stronger. That is why the dhātu accent of gopāya wins here. But the vārttika excepts the vikaraṇas, the affixes placed between root and ending. In yajñáṃ-yajñam abhí vṛdhé gṛṇītáḥ the vikaraṇa śnā comes later than the ending tas. Still, it does not win: tas keeps its udātta. So gṛṇītáḥ has its udātta on the final taḥ. The other udāttas of the line are on jñaṃ, bhi and dhe.
Watch out. The Subodhinī says this is a paribhāṣā and not a heading (adhikāra), because the sūtra is not marked with a svarita, the sign of a heading. It also says "one" (eka) stands for whatever a rule prescribes. So SK 3688 can leave two udāttas in one word. And the paribhāṣā does not reach the Tripādī. So the svarita that SK 3660 later gives to taṃ in gopāyátaṃ survives.
Rule. When an udātta vowel is elided before an anudātta, that anudātta becomes udātta. The accent so produced is called udāttanivṛtti-svara, "the accent from the loss of an udātta".
Example.
deva + ṅīp → deva is antodātta, udātta on its final a (SK 3710, as the Subodhinī explains); ṅīp is anudātta (SK 3709)
→ the final a of deva is elided before ī (SK 311): that a was the udātta
→ this rule: the anudātta ī becomes udātta
→ devī́ṃ vā́cam "divine speech", udātta on vī (and on vā of vācam).
Watch out. The Subodhinī defends the word anudāttasya. In mā hi dhukṣātām the a of the aorist affix ksa is udātta and is elided. Because of this word, the first vowel of the ending ātām becomes udātta, not its last.
Rule. When añc follows with its a elided (so that only c is left), the final vowel of the preceding member is udātta. This is an exception (apavāda) to SK 3651.
Example.
deva + añc + ṅīp → adri replaces the last part of deva (SK 418): devadri + añc + ī
→ the a of añc is elided (SK 416) and the vowel lengthened (SK 417): devadrīcī
→ the elided a was udātta; SK 3651 would make the next vowel udātta
→ this rule instead makes the final vowel of the prior member udātta
→ devadrī́cīṃ nayata devayántaḥ "lead the one turned toward the gods, you who seek the gods", udātta on drī (and on yan).
The ī after drī is a dependent svarita by SK 3660.
Applies when / blocked by. A vārttika: not before a taddhita (ataddhite). In dādhīcaḥ and mādhūcaḥ only the accent of the affix applies. The Subodhinī explains. In dadhi + añc + aṇ the affix aṇ is udātta, and this rule, though taught later, does not displace it. So the udātta stays on the vowel of aṇ, the final syllable: dādhīcáḥ.
ágna índra váruṇa mítra dévāḥ (each vocative udātta on its first syllable) against imáṃ me gaṅge yamune sarasvati (vocatives after a pada: wholly anudātta)A vocative (āmantrita) has two accents, from two books of the Aṣṭādhyāyī. The rule of book 6 (ṣāṣṭha) makes its first vowel udātta. The rule of book 8 (āṣṭamika) makes it wholly anudātta after a pada. Which one applies depends on what stands before the vocative, and on whether that earlier word counts. The running examples are a string of vocatives at the head of a line, ágna índra váruṇa mítra dévāḥ, each udātta on its first syllable. Against them stands imáṃ me gaṅge yamune sarasvati, where three vocatives follow me and have no udātta at all.
Rule. The first vowel of a vocative is udātta.
Example.
agne, indra, varuṇa, mitra, devāḥ → each is a vocative (SK 411)
→ this rule: the first vowel of each is udātta
→ ágna índra váruṇa mítra dévāḥ "O Agni, Indra, Varuṇa, Mitra, gods", udātta on a, i, va, mi and de.
Applies when / blocked by. This is the ṣāṣṭha rule. The next sūtra, the āṣṭamika rule, overrides it after a pada. Here indra and the rest are not "after a pada", because of SK 412 (see SK 3654).
Rule. A vocative that stands after a pada, and not at the head of a pāda, is wholly anudātta. Bhaṭṭoji calls this āṣṭamika rule an apavāda of the ṣāṣṭha rule, SK 3653.
Example.
imam me gaṅge yamune sarasvati → gaṅge, yamune, sarasvati stand after the pada me
→ this rule: all three are wholly anudātta
→ imáṃ me gaṅge yamune sarasvati "this [praise] of mine, O Gaṅgā, Yamunā, Sarasvatī". The only udātta is on maṃ.
Applies when / blocked by.
śútudri stómam, udātta on śu of śutudri (and on sto).ágna índra the word agne counts as absent. So indra is not after a pada, and it keeps its udātta on i.ágne tejasvin "O Agni, brilliant one" and ágne trātaḥ "O Agni, protector", tejasvin and trātaḥ are wholly anudātta.ághnye dévi sárasvati each earlier word counts as absent. So each word escapes this rule and is udātta on its first syllable: a, de, sa.Watch out. The 1915 print marks the accent of the last example only on aghnye. The rule being illustrated gives dévi and sárasvati their udātta as well.
The conditions "after a pada" and "not at the head of a pāda" come from Bhaṭṭoji's gloss. They match SK 402 and SK 403. That they are carried from there is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. As Pāṇini words it: a general term is optionally treated as non-existent when a particular term follows. Bhaṭṭoji reports how the Bhāṣya, the great commentary on the sūtras, recasts it. The Bhāṣya adds "plural" (bahuvacanam) and moves "general term" into SK 413. So the rule reads: when a vocative qualifier follows, a preceding vocative in the plural is optionally treated as non-existent.
Example.
devīḥ ṣaṭ urvīḥ → ṣaṭ "six" qualifies the goddesses
→ of the two options, the one where devīḥ is not treated as non-existent: it counts as present, so ṣaḷ urvīḥ stands after a pada
→ SK 3654: ṣaḷ urvīḥ is wholly anudātta
→ dévīḥ ṣaḷ urvīr urú naḥ kṛṇota "O goddesses, six wide ones, make wide room for us", udātta on de of devīḥ (and on ru of uru).
Applies when / blocked by. In devāḥ śaraṇyāḥ "O gods, protectors" the second word may or may not lose its accent. The Subodhinī says the word "plural" divides the work between SK 413 and this rule.
Rule. When a vocative follows, a word ending in a case-ending is treated as a member of that vocative, for accent.
Example.
dravatpāṇī śubhaḥ patī → śubhaḥ, the sixth case of śubh (from śubh with kvip), joins the vocative patī
→ the whole śubhaspatī is one vocative: SK 3653 makes its first vowel udātta
→ dravatpāṇī counts as absent (SK 412), so śubhaspatī stands at the head of the pāda and SK 3654 does not apply
→ drávatpāṇī śúbhaspatī "you two of swift hands, lords of splendour", udātta on dra and śu.
In yát te divo duhitar marta bhójanam the word divaḥ joins duhitar. Both stand after te, so both are wholly anudātta by SK 3654. The udāttas are on yat and bho. paraśunā vṛścan "O you cutting with an axe" is a further example.
Applies when / blocked by.
ayám agne jaritā́ and eténāgne bráhmaṇā. There ayam and etena stay separate words, with their own udātta on yam and te. Bhaṭṭoji adds that the same limit follows from carrying samartha forward (SK 647).ā́ te pitar marutām and práti tvā duhitar divaḥ, marutām and divaḥ join the vocative before them and share its loss of accent. The udāttas are on ā and pra.uccair adhīyāna "O you reciting aloud", uccaiḥ does not join the vocative.upāgny adhīyāna "O you studying near the fire", upāgni joins the vocative. By SK 3653 the whole takes the udātta on its first syllable, u: úpāgny adhīyāna.Watch out. "Member" (aṅga) here means a part of the following word, as the Subodhinī explains. When the vocative is named, the joined word is included.
agním īḷe (udātta gnim, the following ī svarita); kva + avaram → kvā́varam (merged ā udātta)When two vowels meet, their accents meet too. Five rules of the Tripādī settle the result. A vowel after a yaṇ may become svarita. A single vowel that replaces two, one of them udātta, is udātta, or sometimes svarita. And an anudātta just after an udātta becomes svarita, unless an udātta or svarita follows it. The svaritas these rules create (SK 3657, SK 3660) are Tripādī, so they count as not applied for SK 3650, and the rest of the word does not go low again. The running examples are agním īḷe "I praise Agni", udātta on gnim with the next ī svarita, and kvā́varam, where the merged ā is udātta.
Rule. A yaṇ (y, v, r, l) may stand in place of an udātta vowel or of a svarita vowel. An anudātta right after such a yaṇ becomes svarita.
Example.
abhi abhi hi → abhi is antodātta (by a phiṭ-sūtra, says the Subodhinī); the second abhi is doubled (SK 2140) and anudātta (SK 83, SK 3670)
→ the udātta i of the first abhi becomes y before the vowel (SK 47)
→ this rule: the anudātta a after that y becomes svarita
→ abhy abhi hi, with the a after y svarita.
The second example shows the yaṇ of a svarita: khalapvyāśā. The Subodhinī explains it in two steps.
khalapū + ṅi → khalapū is antodātta; its ū becomes v before the ending: an udātta-yaṇ
→ this rule: the anudātta i of the ending becomes svarita
→ that svarita i becomes y before āśā: now a svarita-yaṇ
→ this rule again: the anudātta ā at the start of āśā becomes svarita
→ khalapvyāśā, with the ā after y svarita.
Applies when / blocked by. This svarita belongs to the Tripādī and so is asiddha for SK 3650. So the rest of the word is not made anudātta (śeṣanighāta).
Watch out. Bhaṭṭoji's text gives no accented copy of these two examples. The svarita is placed here as the rule and the Subodhinī place it. The Subodhinī, following the Bhāṣya, splits the rule in two. The first part works after the yaṇ of an udātta. The second works only after the yaṇ of a svarita made by the first part.
Rule. A single substitute (ekādeśa), one vowel that replaces two, is udātta when one of the two was udātta.
Example.
vas + aśvāḥ → vas is anudātta (SK 405, under SK 403); aśva is udātta on its first a
→ vas becomes vo, and o + a merge into o (SK 86)
→ this rule: that single o is udātta
→ vó 'śvāḥ "your horses", udātta on vo.
kva + avaram → kva is svarita (SK 3729); the first a of avara is udātta
→ the two a's merge into ā
→ this rule: the merged ā is udātta
→ kvā́varaṃ marutaḥ, udātta on kvā.
Watch out. The Subodhinī says why the rule is needed. If both vowels were udātta, the closest-likeness rule (āntaratamya) would already give an udātta. The rule matters when the other vowel is anudātta or svarita. Then closest likeness would give a svarita. Bhaṭṭoji's two examples show one of each.
Rule. Suppose an anudātta stands at the start of a pada, and the single substitute of SK 3658 includes it. Then that single substitute is optionally svarita. In the other option it is udātta by SK 3658.
Example. The option is fixed in its distribution (vyavasthita-vibhāṣā).
vi + idam → vī̀daṃ jyótir hṛ́daye "… this light in the heart". vī̀ is an independent svarita; the udāttas are on jyo and hṛ.divi + īyate → asyá ślóko divī́yate "his fame goes to heaven", udātta on vī (and on sya and ślo).te + avadan → tè 'vadan "they spoke"; saḥ + ayam → sò 'yám ā́gāt "he, this one, has come". tè and sò are independent svaritas; the udāttas are on yam and ā.Bhaṭṭoji quotes the Prātiśākhya. It names three junctions with a svarita: praśleṣa, kṣaipra and abhinihita. The Subodhinī explains them. Praśleṣa is two short i's merging, as in vī̀dam. Kṣaipra is a yaṇ in place of an udātta or svarita, as in abhy abhi (SK 3657). Abhinihita is the merger of SK 86, as in tè 'vadan.
Watch out. The Subodhinī says this distribution is the Ṛgvedins' (Bahvṛcas). Other schools learn theirs from their own Prātiśākhyas. The print writes a numeral after vī and so. It is a sign of the recitation notation, not a pluta, and it is left out here.
Rule. In continuous recitation (saṃhitā), an anudātta right after an udātta becomes svarita. This is the dependent svarita, left unmarked in this chapter.
Example.
agnim īḷe → agni is antodātta and the ending am anudātta; their merged vowel is udātta (SK 3658): gnim
→ īḷe is a finite verb after a non-verb, so it is wholly anudātta (SK 3935)
→ this rule: the anudātta ī after gnim becomes svarita
→ agním īḷe "I praise Agni", udātta on gnim, and ī svarita.
In tám īśānā́saḥ "him, the masters" the udāttas are on tam and nā. The ī and the saḥ after them are svarita.
Applies when / blocked by. The Subodhinī says saṃhitāyām is carried from SK 3627. So in the word-by-word recitation (padapāṭha) the syllable stays anudātta. It also notes that the m of agnim does not stand in the way, because a consonant does not count in an accent rule.
Watch out. This svarita too is asiddha for SK 3650. So the rest of the word does not go low again, and both the udātta and the svarita are heard.
Rule. An anudātta does not become svarita by SK 3660 when an udātta or a svarita follows it. Gārgya, Kāśyapa and Gālava hold that it does.
Example.
prá yá ārúḥ → the ā of āruḥ comes after the udātta yá
→ SK 3660 would make it svarita, but the udātta rú follows it
→ this rule: it stays anudātta
→ prá yá ārúḥ, udātta on pra, ya and ru. That ā is also made "lower" by SK 3669.
vó 'śvāḥ kvā̀ bhī́śavaḥ → the ā of śvāḥ comes after the udātta vó, but the independent svarita kvā̀ follows
→ this rule: it stays anudātta. The udāttas are on vo and bhī.
Applies when / blocked by. The Subodhinī says udaya is a Prātiśākhya word for "following" (para). That "anudātta" and "svarita" are carried here from SK 3660 is not stated by Bhaṭṭoji or his commentators; it is inferred.
āgaccha bho māṇavaka (called from afar: the whole sentence ekaśruti); subrahmaṇya + om → subrahmaṇyó3m (svarita merged o → udātta o3)Speech normally has the three accents (traisvarya). Six sūtras, 1.2.33–38 of the Aṣṭādhyāyī, say where a sentence is spoken on one level pitch instead. This monotone is ekaśruti. The Subodhinī defines it as the state in which the accents stand undistinguished. It comes when calling from afar, in the rite, and optionally in parts of the Veda. In one formula, the Subrahmaṇyā, the monotone is barred, and a svarita becomes udātta. The running examples are āgaccha bho māṇavaka "come here, boy!", called from afar and so monotone, and subrahmaṇyó3m, where a merged svarita o becomes udātta.
Rule. When one calls someone from a distance, the sentence is ekaśruti. This is an exception to the three accents.
Example.
āgaccha bho māṇavaka → a call from a distance
→ this rule: the sentence is spoken on one pitch
→ āgaccha bho māṇavaka "come here, boy!", with no udātta, anudātta or svarita distinguished.
Applies when / blocked by. The Subodhinī takes sambuddhi in its plain sense, "calling", not as the technical name of the vocative singular. So a plural call is covered too: āgacchata devabrāhmaṇāḥ. "Distance" means the person called does not hear an ordinary call.
Watch out. The Subodhinī adds two points. In the example the last vowel of the sentence is pluta and udātta by SK 95. And called from near by, the sentence keeps three accents. ā is udātta as an upasarga, gaccha is unaccented as a verb, bho is udātta as a particle, and māṇavaka is unaccented as a vocative: ā́ gaccha bhó māṇavaka.
Rule. In a sacrificial rite a mantra is ekaśruti. Japa, the nyūṅkhas and the sāmans are excepted.
Example.
agnír mūrdhā́ diváḥ kakút "Agni is the head, the summit of heaven" → in private recitation (svādhyāya) it keeps three accents: udātta on gni, dhā, vaḥ and kut
→ used in the rite: this rule
→ the same words on one pitch.
Applies when / blocked by. Japa keeps its accent: mámāgne várco vihavéṣv astu "may splendour be mine, Agni, at the invocations", udātta on ma, va and ve. Bhaṭṭoji defines the three exceptions. Japa is low-voiced utterance, like the speech of one under water. The nyūṅkhas are the sixteen o-sounds. Sāman is the name for the chants.
Watch out. The Subodhinī gets the word "mantra" from the form of the exception. "Not japa" means something like japa, that is, a mantra.
Rule. In the rite, the word vauṣaṭ is optionally uccaistarām, "higher". Otherwise it is ekaśruti by SK 3663.
Example. Bhaṭṭoji gives none.
Applies when / blocked by. The Subodhinī says vaṣaṭ in the sūtra stands for vauṣaṭ, which has the same meaning. It takes "higher" as udāttatara, "higher than udātta". SK 3610 makes the first syllable of vauṣaṭ pluta and udātta. This rule then makes the final syllable higher still. It notes that others take the comparative ending as adding no comparison.
Rule. In the Veda (chandas), ekaśruti is optional.
Example. The option is fixed in its distribution:
Watch out. The Subodhinī says the sūtra's vibhāṣā also extends the option to ordinary speech. Bhaṭṭoji speaks only of the Veda.
Rule. The Subrahmaṇyā is a particular prose formula (nigada). In it, the ekaśruti of SK 3663 and SK 3665 does not apply. Instead, an udātta replaces a svarita.
Example.
subrahman + yat (SK 1650, "fit for su-brahman") → subrahmaṇya; yat is svarita by SK 3729
→ with the feminine ā, then with om (SK 80), the merged o is svarita, as the Subodhinī explains
→ this rule: that svarita becomes udātta
→ subrahmaṇyó3m, udātta on the pluta o3.
Applies when / blocked by. Four vārttikas follow, each for this same formula.
gārgyó yajate, udātta on gyo. This sets aside the first-syllable udātta that the ñ-marked affix yañ would give (SK 3686).dākṣéḥ pitā yajate, udātta on kṣeḥ.gārgyásyá pitā yajate, udātta on gya and sya.devadattasyá or devadattásyá. That the final stays udātta in a name is not stated by Bhaṭṭoji or his commentators; it is inferred.Watch out. A bracketed sentence of the print asks: does SK 3658 not already make the merged vowel udātta? It answers that "anudātta" is carried into SK 3658, so that rule does not cover a merger with a svarita. The Subodhinī says the opposite at SK 3668. Scholars discuss this.
Rule. In the Subrahmaṇyā, the svarita in deva and brahman becomes anudātta, not udātta as SK 3666 would make it.
Example.
devāḥ brahmāṇaḥ āgacchata → each vocative is udātta on its first syllable; vā and hmā would be svarita after them (SK 3660)
→ SK 3666 would make those svaritas udātta
→ this rule: they become anudātta
→ dévā bráhmāṇa āgacchata "gods, brahmins, come!", udātta on de and bra only.
Applies when / blocked by. The Subodhinī gives this reading for the two words taken separately. If they refer to one group, brahmāṇaḥ may lose its own accent by SK 3655. Then the rule would be needed only for deva.
imám me gaṅge yamune sarasvati (udātta mám; me dependent svarita; every syllable after it ekaśruti); divé-dive (udātta vé; the repeat unaccented)Three last common accents shape a line in continuous recitation. After a svarita, a run of anudāttas is spoken on one pitch, called pracaya. Just before an udātta or svarita, an anudātta sinks lower still, called sannatara. And the second copy of a doubled word is anudātta. The running examples are imám me gaṅge yamune sarasvati, with the udātta on mam, me a dependent svarita, and everything after it on one pitch. The second is divé-dive "day after day", with the udātta on ve and the repeat unaccented.
Rule. In continuous recitation, the anudāttas that follow a svarita become ekaśruti. This one-pitch run is also called pracaya.
Example.
imam me gaṅge yamune sarasvati → idam + ending gives the udātta mám (SK 3658)
→ me is anudātta (SK 403) and becomes a dependent svarita after mám (SK 3660)
→ gaṅge yamune sarasvati are wholly anudātta as vocatives after a pada (SK 3654)
→ this rule: those anudāttas after the svarita me become ekaśruti
→ imám me gaṅge yamune sarasvati, udātta on mam; me svarita; the rest on one pitch.
Applies when / blocked by. The Subodhinī says the plural "anudāttas" names the class. So one, two or many syllables are covered: one in pacati, two in agnim īḷe purohitam, many here.
Watch out. The Subodhinī asks how a Tripādī svarita (SK 3660) can feed this rule, which stands in book 1. Its answer: the Bhāṣya and the Vārttika move the nine sūtras from SK 8 (A 1.2.32–40) to the end of the grammar. So they see the Tripādī's results.
Rule. An anudātta that has an udātta or a svarita after it becomes sannatara, "lower": anudāttatara, lower than an anudātta.
Example.
sarasvati śutudri → sarasvati lost its accent after me; śutudri begins a pāda and has its udātta on śu (SK 3653)
→ this rule: the final i of sarasvati, just before śú, becomes sannatara
→ sarasvati śútudri.
vyàcakṣayat svàḥ → svar is svarita by a phiṭ-sūtra (phiṭ 74); vya, where the upasarga vi met the a of acakṣayat, is svarita too
→ this rule: the ya just before svàḥ becomes sannatara.
Watch out. The Subodhinī says sanna means "low". "Lower" is measured against another anudātta. The unit closes by quoting SK 83 (tasya param āmreḍitam). That is the frame for the next sūtra, not an example of this one.
Rule. The second copy of a doubled word, the āmreḍita of SK 83, is anudātta.
Example.
dive + dive → doubled; the second copy is the āmreḍita (SK 83)
→ this rule: the second dive is anudātta
→ divé-dive "day after day", udātta on ve of the first word.
In the repeat, di is a dependent svarita (SK 3660) and ve is ekaśruti (SK 3668).
Applies when / blocked by. Bhaṭṭoji quotes SK 83 at the end of SK 3669 as the frame. Reading "the second copy of a doubled word" as carried from it is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. The block ends "thus the common accents" (iti sādhāraṇasvarāḥ). The Subodhinī says they are "common" because they act both within one pada and across two padas. Rules such as SK 3686 act only inside one pada.
gopāya (dhātu) → gopāyátam (udātta on the final ya); dā + jhi → (at for jhi) dadati → dádati (initial da udātta)The print now opens a new block, "now the root accents" (atha dhātusvarāḥ). The Subodhinī explains the order. A base (prakṛti) is either a root or a noun stem, and the root comes first. The general rule makes the root's last vowel udātta. Then come roots that take the udātta on their first vowel before certain endings: svap and its group, hiṃs, and the reduplicated stems. The running examples are gopāyátam, udātta on ya, the root's final, and dádati "they give", udātta on the first da.
Several rules here speak of a lasārvadhātuka. This is a sārvadhātuka ending that stands in place of a la-affix, such as the present ending anti. The word is carried from SK 3730, later in this chapter.
Rule. The final vowel of a dhātu is udātta.
Example.
gopāya + tam → gopāya is a dhātu (SK 2304)
→ this rule: its final vowel, the a of ya, is udātta
→ gopāyátaṃ naḥ "[you two] guard us", udātta on ya.
taṃ is a dependent svarita (SK 3660) and naḥ is ekaśruti (SK 3668).
as + sip → one s is dropped: asi
→ this rule: the root's own vowel a is udātta
→ ási satyáḥ, udātta on a (and on tya).
si is a dependent svarita, and the sa of satya is sannatara before tyá (SK 3669).
Applies when / blocked by. "The final is udātta" (anta udāttaḥ) is carried from SK 3680, which the Siddhānta-kaumudī gives later. The Subodhinī names that source.
Rule. Take svap and its group, and hiṃs. Before a lasārvadhātuka that begins with a vowel and has no iṭ, their first vowel is optionally udātta. The svap group is a sub-group inside the ad class. The Subodhinī says it runs from svap to the end of that class.
Example.
svap + anti → a vowel-initial lasārvadhātuka without iṭ
→ this rule (one option): the first vowel of the root is udātta
→ svápanti "they sleep", udātta on sva.
So also śvásanti (udātta on śva) and híṃsanti (udātta on hiṃ).
In the other option the word is madhyodātta, with the udātta in the middle, by the accent of the affix. That puts the udātta on the a of the ending: svapánti. This form is not spelled out by Bhaṭṭoji or his commentators; it is inferred.
Applies when / blocked by. "Lasārvadhātuka" comes from SK 3730, as the Subodhinī says. "The first vowel … optionally" comes from SK 3731. Why "without iṭ"? The Subodhinī gives svapitaḥ, which has iṭ by SK 2474.
Watch out. Bhaṭṭoji, following the Vṛtti, a running commentary on the sūtras, wants the rule only before a k- or ṅ-marked ending. So it does not apply in svapāni or hinasāni, whose ending āni is pit. The Subodhinī notes that this limit is not found in the Bhāṣya. Scholars discuss this.
Rule. Before a lasārvadhātuka that begins with a vowel and has no iṭ, the first vowel of a reduplicated stem (abhyasta) is udātta. This time the rule is obligatory.
Example.
dā + jhi → reduplicated stem dadā; at replaces jhi (SK 2479); the ā is lost (SK 2483)
→ dadati: a vowel-initial ending without iṭ
→ this rule: the first vowel of the stem is udātta
→ yé dádati priyā́ vásu "who give dear goods", udātta on the first da (and on ye, yā, va).
Applies when / blocked by. This rule comes later in the Aṣṭādhyāyī (6.1.189) than the rule of SK 3710 (6.1.163), so it overrides the accent that SK 3710 gives a c-marked affix. The participle ending śānac is c-marked. Yet dádhānā índre has its udātta on the first da of dadhānā (and on i of indre).
Watch out. The Subodhinī explains why ādiḥ is said again, though it already runs from SK 3731. Restating it cuts off the "optionally" that came with it. So this rule is obligatory. Why "before a vowel"? dadyāt. Why "without iṭ"? jakṣitha.
Rule. The first vowel of a reduplicated stem is also udātta before a lasārvadhātuka that has no udātta at all.
Example.
dadhā + si → the ending si is pit and anudātta: it has no udātta
→ this rule: the first vowel of the stem is udātta
→ dádhāsi rátnaṃ dráviṇaṃ ca dāśúṣe "you give treasure and wealth to the worshipper", udātta on the first da of dadhāsi (and on ra, dra, śu).
Watch out. The Subodhinī reads anudātte as "that in which no udātta exists". It does not mean an ending that is anudātta by name. That reading gives the right form in mā hi sma dadyāt.
hu + ślu + tip → juhóti (udātta ho, the vowel before the pit ending); cikīrṣa + ṇvul (aka) → cikī́rṣakaḥ (udātta kī)Five more root rules put the udātta on the vowel just before a suffix, or offer the first vowel as an option. Nine reduplicated roots take it before a pit ending, and every base takes it before an l-marked suffix. Before ṇamul, before the yak of a self-acting verb and in a caṅ form, the udātta is optional. The running examples are juhóti "he offers", udātta on ho, and cikī́rṣakaḥ "one who wishes to do", udātta on kī.
Rule. Take nine roots as reduplicated stems: bhī, hrī, bhṛ, hu, mad, jan, dhan, daridrā and jāgṛ. Before a pit lasārvadhātuka, the vowel just before the suffix is udātta. This is an exception to SK 3673.
Example.
hu + ślu + tip → reduplicated stem juho; the ending ti is pit
→ this rule: the vowel before the suffix, o, is udātta
→ yò 'gnihotráṃ juhóti "who offers the agnihotra", udātta on ho (and on traṃ).
yò is an independent svarita by SK 3659.
More examples:
mamáttu naḥ párijmā, udātta on the second ma (and on pa).mātā́ yád vīráṃ dadhánat, udātta on dha (and on tā, yad, raṃ).jāgárṣi tvám, udātta on ga (and on tvam).Applies when / blocked by. The Subodhinī explains the forms. mamattu is mad in the imperative, with ślu by SK 3400. dadhanat is dhan in the Vedic subjunctive. Why "pit"? daridrati.
Watch out. The Subodhinī reads the word pūrvam "before" as a hint (jñāpaka). In accent rules a seventh-case word means "a form ending in that item". It applies this to SK 3733 and SK 3679, and expressly not to SK 3682.
Rule. Before a suffix marked with l (lit), the vowel just before the suffix is udātta.
Example.
cikīrṣa + ṇvul → ṇvul has the markers ṇ and l, so it is lit; vu becomes aka
→ the final a of the desiderative stem is elided: cikīrṣ + aka
→ this rule: the vowel before the suffix, the ī of kī, is udātta
→ cikī́rṣakaḥ, udātta on kī.
Applies when / blocked by. "Before the suffix" is carried from SK 3675. The Subodhinī says this keeps the accent off the whole lit-final form.
Watch out. The elided a does not count as still present. The Subodhinī says the elision of a vowel does not count as that vowel in an accent rule (unlike the r of SK 3682).
Rule. Before ṇamul, the first vowel is optionally udātta. Bhaṭṭoji says "of reduplicated stems".
Example.
lolūya + ṇamul → lolūyam, then doubled (SK 2140)
→ this rule: the first vowel is udātta
→ the second copy is anudātta (SK 3670)
→ lólūyaṃlolūyam, udātta on the first lo.
Applies when / blocked by. In the other option, the accent of SK 3676 applies, since ṇamul is lit. That puts the udātta on the vowel before ṇamul. Bhaṭṭoji does not give that form.
Watch out. The Subodhinī accepts "of reduplicated stems" as stating what holds in effect.
Rule. Take a root that ends in a vowel as first taught (upadeśa). Before the kartṛyak, its first vowel is optionally udātta. The kartṛyak is the yak used when the ending denotes the agent. Here the object of the action acts by itself as its own agent.
Example.
lū + yak + te → lū ends in a vowel; yak is used because the field "reaps itself"
→ this rule (one option): the first vowel is udātta
→ lū́yate kedāraḥ svayam eva "the field gets reaped by itself", udātta on lū.
Applies when / blocked by. In the other option yak is udātta: lūyáte, udātta on ya. The Subodhinī gives this form. "In upadeśa" is carried from SK 3730, as the Subodhinī says. "The first vowel … optionally" is carried from SK 3677; this is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. In a root that ends in caṅ, the penultimate vowel is optionally udātta.
Example.
kṛ + ṇic + caṅ → the caṅ-final stem cīkara: syllables cī-ka-ra
→ this rule (one option): the penultimate, ka, holding the root's a, is udātta
→ mā hi cīkáratām, udātta on ka.
In the other option caṅ is udātta: cīkarátām, udātta on ra.
Applies when / blocked by. "Penultimate" (upottamam) is carried from SK 3733, as the Subodhinī says. caṅi means "in a caṅ-final form", by the hint of SK 3675. The rule needs three syllables. So it does not apply in mā hi dadhat, where the stem dadha has only two.
Watch out. The block ends "thus the root accents" (iti dhātusvarāḥ).
pac + ghañ → (ku, SK 2863) pākáḥ (udātta on the final kaḥ); pañcan + bhis → pañcábhiḥ (udātta on the penultimate ca)The print opens its third block here: "now the accents of affixes" (atha pratyayasvarāḥ). These rules fix the udātta of whole words. The first ones put it on the last vowel, or on the penultimate of certain numerals. The running examples are pākáḥ "cooking", udātta on kaḥ, and pañcábhiḥ "by five", udātta on ca.
Rule. A word ending in ghañ has its final vowel udātta when it comes from karṣ (karṣati), or when the ghañ-final word itself contains ā (ātvat, "having ā"). This is an exception to SK 3686, which would make a ñ-marked form udātta on its first vowel.
Example.
pac + ghañ → vṛddhi, and k for c (SK 2863): pāka
→ the ghañ-final word pāka contains ā (the root pac itself has none)
→ this rule: the final vowel is udātta
→ pākáḥ, udātta on kaḥ.
kṛṣ + ghañ from karṣati → karṣáḥ "drawing", udātta on ṣaḥ.
Applies when / blocked by. The sūtra cites karṣ with its present stem, from karṣati. So kṛṣ of the tud class (kṛṣati) is not covered. It keeps the first-vowel udātta of SK 3686: kárṣaḥ, udātta on kar.
Rule. The words of the uñcha list also have their final vowel udātta.
Example.
vaiśvānaráḥ kuśikébhir yugé-yuge, udātta on ge of the first yuge (and on ra, ke).yóge-yoge tavástaram, udātta on yo (and on va).gā́vaḥ sómasya prathamásya bhakṣáḥ, udātta on kṣaḥ (and on gā, so, ma).úd uttamáṃ varuṇa, udātta on maṃ; śaśvattamám īḷate, udātta on mam.Applies when / blocked by. "The final is udātta" is carried from SK 3680. That source is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. In yugé-yuge the second copy is anudātta by SK 3670.
Rule. Before the ending śas, the final vowel of catur is udātta. catur is cat + uran (uṇādi 736). Being n-marked, it would otherwise be udātta on its first vowel (SK 3686).
Example.
catur + śas → the ending śas
→ this rule: the last vowel of catur, the u of tu, is udātta
→ catúraḥ kalpayantaḥ, udātta on tu.
Applies when / blocked by. Not in cátasraḥ paśya. In catasṛ + as the r replaces ṛ (SK 299). That substitute counts as the original ṛ for an operation on what stands before it. So ta is not the final vowel, and the word keeps its first-vowel udātta on ca.
Watch out. The udātta falls on catur itself, not on the ending. The Subodhinī says the express mention of śas keeps the operation off the ending.
Rule. After ṣaṭ, tri or catur, take a word ending in a case-ending that begins with a jhal. Its penultimate vowel is udātta. By SK 369, from chapter 11, the name ṣaṭ also covers the numerals ending in ṣ or n, such as pañcan. This is an exception to SK 3725, which would make the ending udātta.
Example.
pañcan + bhis → bhis begins with the jhal bh; the n is lost (SK 236): pañcabhiḥ
→ this rule: the penultimate vowel, ca, is udātta
→ adhvaryúbhiḥ pañcábhiḥ "by five adhvaryu priests", udātta on ca.
So also navábhir vā́jair navatī́ ca (udātta va), saptábhyo jā́yamānaḥ "being born from seven" (udātta pta) and ā́ daśábhir vivásvataḥ (udātta śa).
Applies when / blocked by.
ā́ ṣaḍbhír hūyámānaḥ and devaís tribhíḥ. Here SK 3725 makes the ending udātta: bhíḥ. These words have two syllables, so they have no penultimate in the sense of this rule. This reason is not stated by Bhaṭṭoji or his commentators; it is inferred.navānā́ṃ navatīnā́m "of nine nineties". The ending nām does not begin with a jhal, so this rule does not apply, and the ending carries the udātta, nā́.Watch out. "After ṣaṭ, tri, catur" and "case-ending" come from SK 3725 and SK 3714, later in this chapter. Bhaṭṭoji supplies the words but does not name the rules; the sources are inferred.
Rule. In ordinary usage (bhāṣā), SK 3683 is optional. Bhaṭṭoji says only "in the domain just stated".
Example. The Subodhinī gives the forms.
pañcan + bhis → this rule: the penultimate is optionally udātta
→ pañcábhiḥ, udātta on ca;
→ or, by SK 3725, the ending is udātta: pañcabhíḥ, udātta on bhiḥ.
So also saptabhiḥ, tisṛbhiḥ and caturbhiḥ.
Watch out. The choice belongs to ordinary speech. The Vedic examples of SK 3683 all have the penultimate udātta.
puṃs + ṣyañ (ñit; brāhmaṇādi) → paúṃsya (udātta on the first syllable pau): yásmin víśvāni paúṃsyāThese rules put the udātta on the first vowel of a word. The central one is SK 3686: a word made with a ñ- or n-marked affix is ādyudātta. Around it stand single words and short lists: sarva, pathin and mathin, the tavai form, kṣaya, jaya, the vṛṣa list and certain names. The running example is paúṃsyā "manly powers", from puṃs with the ñ-marked ṣyañ. Its udātta is on the first syllable, pau.
Rule. When a case-ending follows, the first vowel of sarva "all" is udātta.
Example.
sarva + jas → a case-ending follows
→ this rule: the first vowel is udātta
→ sárve nandanti yaśásā "all rejoice in glory", udātta on sa (and on śa of yaśasā).
Applies when / blocked by. "First vowel" (ādiḥ) is carried from SK 3673, and "udātta" from SK 3680. Bhaṭṭoji supplies both words without naming the rules; the sources are inferred.
Watch out. sarva also stands in the uñcha list of SK 3681, as antodātta. The Subodhinī explains why. That listing is there only so that sārva "made of sarva" can take the affix añ (SK 1520). What is actually heard is always the first-vowel udātta, sárva-.
Rule. A word ending in an affix marked with ñ, or with n, always has its first vowel udātta.
Example.
puṃs + ṣyañ → ṣyañ in the sense of action, since puṃs is in the brāhmaṇa list; ṣyañ is ñ-marked
→ this rule: the first vowel is udātta
→ yásmin víśvāni paúṃsyā "in whom all manly powers …", udātta on pau (and on ya, vi).
cāy + asun → the uṇādi affix asun, with the augment nuṭ and a short vowel (uṇādi 639): canas; asun is n-marked
→ this rule: the first vowel is udātta
→ suté dadhiṣva naś cánaḥ "at the pressing, take our food", udātta on ca (and on te of sute).
Applies when / blocked by. The Subodhinī reads ñniti as "of a form ending in a ñ- or n-marked affix", by the hint of SK 3675. "Udātta" is carried from SK 3680; this is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. SK 3680 is the exception: a ghañ-final word from karṣ, or one that contains ā, is antodātta, pākáḥ.
Rule. Before a sarvanāmasthāna (a strong case-ending), pathin "path" and mathin have their first vowel udātta. Both are formed with the affix ini and would otherwise be antodātta, as the Subodhinī says.
Example.
pathin + su → a strong ending; panthāḥ by SK 365
→ this rule: the first vowel is udātta
→ ayáṃ pánthāḥ "this path", udātta on pa (and on yaṃ).
Applies when / blocked by. Why "before a sarvanāmasthāna"? jyótiṣmataḥ pathó rakṣa "guard the luminous paths". Here pathin is a bha stem, and its last part in, which held the udātta, is elided. By SK 3651 the following ending becomes udātta: thó.
Rule. A word ending in the affix tavai has its first and its last vowels udātta, both at once.
Example.
dā + tavai → by SK 3441: dātavai
→ before u, ai becomes āy, and the y is lost (SK 67): dātavā u
→ this rule: the first vowel and the last vowel are both udātta
→ harṣase dā́tavā́ u, udātta on dā and on vā.
Applies when / blocked by. "First" (ādiḥ) is carried from SK 3686. This is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. The Subodhinī says yugapat "at once" rules out taking the two in turn. Without it, SK 3650 ("all but one") would forbid two udāttas in one word.
Rule. kṣaya, when it means "a dwelling", has its first vowel udātta.
Example.
kṣi + gha → gha in the sense of the place, "that in which they dwell" (SK 3296): kṣaya
→ this rule: the first vowel is udātta
→ své kṣáye śucivrata "in your own dwelling, O you of bright vows", udātta on kṣa (and on své).
Applies when / blocked by. The Subodhinī gives the counter-example vyādheḥ kṣayaḥ "the waning of a disease". There kṣaya is made with ac (SK 3231) and does not mean a dwelling.
Rule. jaya, when it means an instrument (karaṇa), has its first vowel udātta.
Example.
ji + gha → gha in the sense of the instrument, "one conquers by it" (SK 3296): jaya
→ this rule: the first vowel is udātta
→ jáyo 'śvaḥ "the horse, [an instrument of] victory", udātta on ja.
Applies when / blocked by. The Subodhinī gives the counter-example jayo brāhmaṇānām "the victory of the brāhmaṇas". There jaya is the action itself, made with ac (SK 3231).
Rule. vṛṣa and the words of its list have their first vowel udātta. The list is open (ākṛtigaṇa): words of the same kind belong to it even if unlisted.
Example.
vaj + ghañ → vāja; SK 3680 would make it antodātta, since the ghañ-final word vāja contains ā
→ it belongs to the open vṛṣa list: this rule
→ vā́jebhir vājínīvatī, udātta on vā (and on ji).
So also índraṃ vā́ṇīḥ, udātta on vā (and on in).
Applies when / blocked by. The Subodhinī explains vāja as above. It takes vṛṣa itself from vṛṣ "to sprinkle" with the affix ka.
Rule. A word for the standard of comparison (upamāna), used as a name, has its first vowel udātta.
Example.
cañcā iva "like a cañcā [a straw figure]" → kan in the sense "like" (SK 2051)
→ kan is elided by lup (SK 2053): cañcā, now the name of the person compared
→ this rule: the first vowel is udātta
→ cáñcā, udātta on cañ.
Applies when / blocked by. Why "as a name"? agnir māṇavakaḥ "the boy is fire". Why "standard of comparison"? jaitraḥ.
Watch out. kan is n-marked. If the elided kan still counted (pratyayalakṣaṇa), SK 3686 would already give this accent. So this rule is a hint (jñāpaka): in accent rules an elided affix sometimes does not count.
dā + kta → (dat, SK 3077) dáttaḥ (a name: udātta on the first syllable da), against kṛtám, hṛtám (affix accent: udātta on the final)A word in kta normally takes the udātta of its affix on its last syllable: kṛtám "done". Six rules make certain kta words ādyudātta instead. One (SK 3693) covers two-vowel names; the other five concern single words. The running example is the name dáttaḥ, from dā with kta, with its udātta on da. Against it stand kṛtám and hṛtám, which are not names and keep the udātta on the last syllable.
Rule. A name of two vowels ending in a niṣṭhā affix has its first vowel udātta, provided that vowel is not ā.
Example.
dā + kta → dat replaces dā (SK 3077): datta, a name of two vowels; the first is a, not ā
→ this rule: the first vowel is udātta
→ dáttaḥ, udātta on da.
Applies when / blocked by.
cintitaḥ, which has three.trātaḥ.kṛtám and hṛtám, which are not names. They keep the affix accent, with the udātta on the last syllable.Rule. śuṣka "dry" and dhṛṣṭa "bold" have their first vowel udātta. The rule is for their use other than as names. As names, SK 3693 would already give this, says the Subodhinī.
Example.
śuṣ + kta → k replaces t (SK 3030): śuṣka
→ this rule: the first vowel is udātta
→ atasáṃ ná śúṣkam "like dry brushwood", udātta on śu (and on saṃ, na).
Rule. āśita, when it means the agent, has its first vowel udātta.
Example.
āśita denoting the agent ("one who has eaten" / "fed")
→ this rule: the first vowel is udātta
→ kṛṣánn ít phā́la ā́śitam "the ploughshare, just by ploughing, [makes one] fed", udātta on ā of āśitam (and on ṣan, it, phā).
Applies when / blocked by. The Subodhinī gives two ways to form the word. Either kta comes after aś with āṅ in the agent sense, as a fixed form. Or āśita is the object of the causative "to feed", the one who had been the eater. Scholars discuss this.
Rule. rikta "empty" optionally has its first vowel udātta.
Example.
ric + kta → rikta
→ this rule (one option): the first vowel is udātta
→ ríktaḥ, udātta on ri.
In the other option it keeps the affix accent, riktáḥ. This form is not stated by Bhaṭṭoji or his commentators; it is inferred, as with kṛtám.
Applies when / blocked by. As a name, rikta is always ādyudātta by SK 3693. Bhaṭṭoji explains this by pūrvavipratiṣedha: in this conflict the earlier rule prevails. The Subodhinī gives the scopes. SK 3693 alone covers a name like guptaḥ. This rule alone covers a non-name like rikto ghaṭaḥ "an empty jar". Both meet in a name like Rikta, and the earlier wins.
Rule. In the Veda, juṣṭa "welcome" and arpita "set, fixed" optionally have their first vowel udātta.
Example. Bhaṭṭoji gives none here. The forms are those of SK 3698: júṣṭaḥ, árpitam.
Applies when / blocked by. "Optionally" is Bhaṭṭoji's vā, carried from SK 3696. That source is not stated by Bhaṭṭoji or his commentators; it is inferred. The Subodhinī forms juṣṭa from juṣ with kta, and arpita from ṛ with ṇic, the augment p, and kta.
Rule. In a mantra, juṣṭa and arpita always have their first vowel udātta.
Example. júṣṭo dámūnāḥ "the welcome house-friend", udātta on ju (and on da). ṣáḷara āhur árpitam "they call [him] set in the six-spoked [wheel]", udātta on ar (and on ṣa).
Applies when / blocked by. Bhaṭṭoji says the sūtra could be left out (śakyam akartum). He gives three reasons.
júṣṭaḥ and árpitam already follow from SK 3697.arpitā́ḥ ṣaṣṭír ná calācalā́saḥ an antodātta arpita is seen in a mantra, with the udātta on tā. This rule would wrongly exclude it.Watch out. The Subodhinī draws the moral: the rule is not merely unneeded but would be faulty if made.
asmad + ṅas → (aś; mama; pararūpa) máma (udātta on the first ma, not the ending SK 3658 would give); kam + ṇiṅ + yat (SK 2842) → kā́mya (udātta kā)The last three rules of this stretch also give a first-syllable udātta. Two concern the pronouns yuṣmad "you" and asmad "I" before the genitive and dative singular. The third concerns two-vowel words in yat. The running examples are máma "of me", udātta on its first ma, and kā́myā "desirable", udātta on kā.
Rule. Before the ending ṅas (genitive singular), yuṣmad and asmad have their first syllable udātta. Both end in the affix madik (uṇādi 136) and would otherwise be antodātta, as the Subodhinī says.
Example.
asmad + ṅas → aś replaces ṅas (SK 399); mama replaces the part up to m (SK 398); the rest is lost (SK 385)
→ a + a merge by pararūpa (SK 191); SK 3658 would make that merged vowel, the ending, udātta
→ this rule instead: the first syllable is udātta
→ nahi ṣas táva no máma "for neither of you nor of me …", udātta on ma of mama and ta of tava.
Rule. Before ṅe (dative singular) too, yuṣmad and asmad have their first syllable udātta.
Example.
yuṣmad + ṅe → tubhyam by SK 382 and SK 394
→ this rule: the first syllable is udātta
→ túbhyaṃ hinvānáḥ "being sped for you", udātta on tu (and on naḥ).
So also máhyaṃ vā́taḥ pavatām "let the wind blow for me", udātta on ma (and on vā).
Watch out. The Subodhinī says the rule is made separate from SK 3699 for a reason. As one rule, yuṣmad and asmad might be paired one-to-one with ṅas and ṅe (yathāsaṃkhya). Split, each pronoun takes the udātta before each ending.
Rule. A word of two vowels ending in the affix yat has its first syllable udātta. A word formed from nau "boat" is excepted. This is an exception to SK 3729, which makes yat, a t-marked affix, svarita.
Example.
kam + ṇiṅ + yat → yat after a vowel-final root (SK 2842): kāmya, two vowels
→ this rule: the first syllable is udātta
→ yuñjánty asya kā́myā "they yoke his two desirable [bays]", udātta on kā (and on ñjan).
Applies when / blocked by. "Two vowels" (dvyac) is carried from SK 3693, as the Subodhinī says. Why "not from nau"? navatíṃ nāvyā́nām "ninety of the navigable [rivers]". nāvya "to be crossed by boat" takes yat by SK 1643, but it is not ādyudātta. By the rule it keeps the svarita of SK 3729 on vyā: nāvyā̀nām. The 1915 print marks it nāvyā́nām, with the udātta on vyā. Either way the first syllable nā is not udātta, and that is the point.
Watch out. The Subodhinī reads the exception as a hint (jñāpaka): in accent rules a consonant counts as absent. Otherwise the exception would be idle. The n of nāvya could not carry an accent, and ā would not count as the first vowel.
īḍ + ṇyat (ṛhalor ṇyat, SK 2872) → īḍya → ī́ḍyaḥ (udātta ī́)Part I ended with yato 'nāvaḥ (SK 3701), which makes a two-vowel word in yat ādyudātta, that is, udātta on its first syllable. These three sūtras continue the word accents of the sixth book. Each again puts the udātta on the first syllable: of certain ṇyat words, optionally of veṇu and indhāna, and optionally of six nouns such as tyāga. As everywhere in this chapter, the udātta is marked with an acute (á), an independent svarita with a grave (à), and an anudātta is left unmarked. The running example is ī́ḍyaḥ "to be praised", from īḍ "praise" with ṇyat; its first syllable ī́ is udātta.
Rule. A word made with the affix ṇyat from īḍ "praise", vand "salute", vṛ "choose", śaṃs "praise" or duh has its first syllable udātta. Bhaṭṭoji's gloss is ādir udāttaḥ, "the first is udātta". The Subodhinī identifies vṛ as vṛṅ "choose".
Example.
īḍ + ṇyat (ṛhalor ṇyat, SK 2872) → īḍya
→ the first syllable is udātta (this rule) → ī́ḍyaḥ, udātta ī́.
Bhaṭṭoji's four Vedic examples:
ī́ḍyo nū́tanair utá "to be praised also by the recent seers": udātta ī́, nū́ and tá.
ājúhvāna ī́ḍyo vándyaś ca "invoked, to be praised and to be saluted": udātta jú, ī́ and ván.
śréṣṭhaṃ no dhehi vā́ryam "bestow on us the best treasure, worth choosing": udātta śré and vā́.
ukthám índrāya śáṃsyam "a hymn to be recited for Indra": udātta kthá, í and śáṃ.
In each, the ṇyat word has the udātta on its first syllable: ī́ḍya, vándya, vā́rya, śáṃsya.
Applies when / blocked by. Why is this rule needed after SK 3701? The Subodhinī answers. SK 3701 names yat. The affix ṇyat carries two marker letters, ṇ and t, so the mention of yat does not include it.
Watch out. Should vā́ryam not take kyap by etistuśāsvṛdṛjuṣaḥ kyap (SK 2857)? The Subodhinī says no. That rule takes only vṛñ, not this vṛṅ "choose". So vṛṅ takes ṇyat by SK 2872, and this rule gives its accent.
Rule. In veṇu "bamboo" and indhāna "kindling", the first syllable is optionally udātta. Bhaṭṭoji glosses: ādir udātto vā, "the first is optionally udātta".
Example.
índhāno agním "kindling the fire": udātta í and gní. Here índhāna takes the first-syllable option.
Applies when / blocked by. The Subodhinī explains what the option does for each word.
veṇu is made with an uṇādi affix marked with n (a nit affix). So it would be ādyudātta in every case by ñnity ādir nityam (SK 3686). The Subodhinī's citation of that uṇādi rule is garbled in the 1915 print. That it means uṇādi 318, ajivṛrībhyo nic ca, is not stated by Bhaṭṭoji or his commentators; it is inferred.
indhāna has two possible affixes. With cānaś, which is c-marked (cit), it is antodātta, with the udātta on its last syllable. With śānac it is madhyodātta, with the udātta in the middle. There the participle affix is anudātta (SK 3730), and the udātta a of the infix śnam is lost (SK 2469). So indhāna could not be ādyudātta in any other way; this rule gives it as an option.
Rule. In tyāga, rāga, hāsa, kuha, śvaṭha and kratha, the first syllable is optionally udātta. Bhaṭṭoji glosses ādir udātto vā. He adds that the first three end in ghañ and the other three in the ac of the paca-group (see SK 2896). His "optionally" is the vibhāṣā of SK 3703 carried forward (anuvṛtti). That source is not stated by Bhaṭṭoji or his commentators; it is inferred. The Subodhinī's "in the other option" (pakṣe) confirms that the rule is optional.
Example. Bhaṭṭoji gives no examples, so the six words of the sūtra serve. The Subodhinī supplies the derivations.
tyaj + ghañ → g for j (cajoḥ ku ghiṇṇyatoḥ, SK 2863) and vṛddhi → tyāga
→ this rule → tyā́gaḥ (udātta tyā́); in the other option tyāgáḥ (udātta gá).
rañj + ghañ → the n is lost (SK 3187) → g for j (SK 2863) and vṛddhi, as for tyāga → rāga → rā́gaḥ or rāgáḥ.
has + ghañ → hāsa → hā́saḥ or hāsáḥ.
kuha + ac → kúhaḥ or kuháḥ; likewise śváṭhaḥ or śvaṭháḥ, kráthaḥ or kratháḥ.
Applies when / blocked by. What is the other option? The Subodhinī says the three ghañ words are then antodātta by karṣātvato ghaño 'nta udāttaḥ (SK 3680). The three ac words are then antodātta by the accent of the affix. The Subodhinī glosses kuha "astonish", śvaṭha "speak badly" and kratha "injure".
śara + matup (a river name; m → v; ā by śarādīnāṃ ca) → śarāvatī → śarā́vatī (udātta rā́)Three sūtras fix the accent of feminine names made with the affix matup, whose m becomes v (-vat, feminine -vatī). If the name ends in -āvatī, the ā is udātta. If it ends in -avatī or -īvatī, the final ī is udātta. The running example is the river name śarā́vatī, with the udātta on rā́.
Rule. In a name (saṃjñā) of something feminine, the ā just before the affix matup is udātta. Bhaṭṭoji calls it a "feminine name" (strīnāman). This means the name of a feminine thing, such as a river, not a woman's name.
Example.
śara + matup + ṅīp (a river name; the Subodhinī: the m becomes v by mādupadhāyāḥ, and a is lengthened by śarādīnāṃ ca)
→ śarāvatī
→ the ā before vatī is udātta (this rule) → śarā́vatī, udātta rā́.
Likewise udumbarā́vatī, udātta rā́.
Bhaṭṭoji marks only the ā; the other syllables are not given.
Watch out. The sūtra says pūrvam, in the neuter, though it qualifies a vowel. The Subodhinī explains that the neuter refers to the sound-form itself. Some read pūrvaḥ, in the masculine.
Rule. The final vowel of the word avatī is udātta.
Example.
vetra + matup + ṅīp → v for m → vetravatī
→ the final ī is udātta (this rule) → vetravatī́, udātta tī́.
Applies when / blocked by. Bhaṭṭoji says why the rule is needed. The feminine affix ṅīp is marked with p (pit), so it would be anudātta. He does not name the rule; it is anudāttau suppitau, SK 3709, taught later in this chapter.
Watch out. The Subodhinī meets two objections. First, the v that replaces m is asiddha, "not yet effective", so is this really the word avatī? Yes: the v counts because this rule depends on it. Second, the Subodhinī adds that in accent rules a consonant counts as absent. So the word here counts as avatī.
Rule. The final vowel of a word ending in īvatī is also udātta. Bhaṭṭoji says "as before" (prāgvat), carrying antaḥ … udāttaḥ from SK 3706.
Example.
ahi + matup + ṅīp → ī for i, v for m → ahīvatī
→ the final ī is udātta → ahīvatī́, udātta tī́.
Likewise munīvatī́, udātta tī́.
Applies when / blocked by. The Subodhinī says the long ī comes from śarādīnāṃ ca, as at SK 3705. The v comes from the rule for names (saṃjñāyām).
Watch out. In the 1915 print the phiṭ-sūtras are printed right after this sūtra, before SK 3708. This chapter reads them here.
phiṭ uccais → uccaíḥ (udātta caí); pāṭalā → pāṭalā́, not pā́ṭalā by laghāvante (phiṭ 42)The phiṭ-sūtras are not Pāṇini's. They are a separate set of rules on the accent of the prātipadika (phiṭ), the nominal stem. The 1915 print places them between SK 3707 and SK 3708. They are numbered here by the 1915 print's running number, "phiṭ 1" to "phiṭ 87". Their first pāda (phiṭ 1–23) puts the udātta on the stem's last vowel. Their second pāda (phiṭ 24–49) puts it, by default, on the first syllable. Many phiṭ rules exist only to keep a word out of a later phiṭ rule. So a rule often names its rival by number. The running examples are uccaíḥ "high", udātta on the last syllable caí, and pāṭalā́, a plant name. Without phiṭ 2, pāṭalā would take the udātta on its heavy first syllable by phiṭ 42.
Rule. The final vowel of a phiṭ is udātta. Bhaṭṭoji: "phiṭ is the prātipadika", the nominal stem. The Subodhinī explains that phiṭ is the earlier teachers' word for the prātipadika.
Example.
uccais → the final vowel is udātta (this rule) → uccaíḥ, udātta caí.
Applies when / blocked by. This is the general rule of the first pāda. Its words antaḥ "the final" and udāttaḥ "is udātta" run on through phiṭ 23. That they come from phiṭ 1 is not stated by Bhaṭṭoji or his commentators; it is inferred. The Subodhinī confirms at phiṭ 23 that "udātta" is carried forward.
Watch out. May a grammarian quote rules that are not Pāṇini's? The Subodhinī raises the objection and answers it. Pāṇini's followers do rely on the phiṭ-sūtras. The Bhāṣya on ādyudāttaś ca (SK 3708) presupposes stems that are antodātta and stems that are wholly anudātta. Such stems are not possible without rules like this one. The 1915 print heads the section "Now the phiṭ-sūtras; first pāda".
Rule. Words meaning pāṭalā (a plant), apālaṅka (a tree), ambā "mother" or sāgara "ocean" have the udātta on the final vowel.
Example. Bhaṭṭoji lists each meaning with its synonyms.
For pāṭalā: pāṭalā́, phaleruhā́, surūpā́, pākalā́.
For apālaṅka: apālaṅká, vyādhighātá, ārevatá, āragvadhá.
For ambā: mātā́.
For sāgara: sāgaráḥ, samudráḥ.
In every word the last syllable is udātta.
Applies when / blocked by. For the pāṭalā words Bhaṭṭoji names the rival: laghāv ante (phiṭ 42), which would put the udātta on a heavy syllable. The Subodhinī applies this remark to the apālaṅka words as well. For mātṛ the rival is unarvannantānām (phiṭ 32), which gives an ṛ-final word the udātta on its first syllable. The Subodhinī says sāgara and samudra also escape phiṭ 42.
Watch out. In the feminines, the stem-final udātta a and the anudātta ā of ṭāp merge into one vowel. That vowel stays udātta (SK 3658). This step is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. Words meaning "house" have the udātta on the final vowel, when they are not feminine.
Example.
geha (neuter) → the final vowel is udātta (this rule) → gehám, udātta hám.
Counter-example (feminine): śā́lā, udātta śā́.
Applies when / blocked by. Without this rule, nabviṣayasya (phiṭ 26) would apply, because geha is neuter. It would make geha ādyudātta.
Watch out. How do we know that śālā is ādyudātta? Bhaṭṭoji says the exclusion in this very rule makes it known (a jñāpaka, an indication). Had feminine house-words been antodātta anyway, excluding them would be pointless.
Rule. guda "gut" also has the udātta on its final vowel, but not in the feminine. "Not in the feminine" is carried from phiṭ 3. The Subodhinī notes that guda does not mean "house", so phiṭ 3 alone would not cover it.
Example.
guda (not feminine) → this rule → gudám, udātta dám.
Counter-example (feminine): āntrébhyas te gúdābhyaḥ "from your entrails, from your guts". Here gúdābhyaḥ has the udātta on its first syllable gú; āntrébhyas has it on ntré.
Applies when / blocked by. Why is the feminine ādyudātta? The 1915 print adds a sentence. svāṅgaśiṭām adantānām (phiṭ 29) gives the first-syllable udātta to the a-final guda first. That is the inner operation (antaraṅga). Only then does the feminine ṭāp come.
chāyā → chāyā́ (udātta yā́), not chā́yā by yāntasyāntyāt pūrvam (phiṭ 62)These three rules give the final udātta to words by the sounds they end in. A feminine word in -dhā or -yā, a word in -kha, and words in -hiṣṭha, -vatsara, -ti, -śat and -tha all take it. Each rule is set against a later phiṭ rule that would otherwise apply. The running example is chāyā́ "shade", with the udātta on its last syllable yā́. Without phiṭ 5, yāntasyāntyāt pūrvam (phiṭ 62) would make it chā́yā.
Rule. In a word restricted to the feminine, a final vowel preceded by dh or y is udātta. The Subodhinī supplies "of a word that is always feminine" (nityastrīliṅga).
Example.
antardhā → the final ā is udātta (this rule) → antardhā́, udātta dhā́.
chāyā → this rule → chāyā́, udātta yā́; likewise māyā́, jāyā́.
Applies when / blocked by. Bhaṭṭoji names the rule each example escapes. For antardhā it is strīviṣayavarṇa- (phiṭ 43), which would make it ādyudātta. For chāyā, māyā and jāyā it is yāntasyāntyāt pūrvam (phiṭ 62).
Watch out. Bhaṭṭoji gives two sets of counter-examples.
Why "feminine"? bā́hyam "outer" is ādyudātta, because its affix yañ is marked with ñ (ñit), and ñnity ādir nityam (SK 3686) makes such a word ādyudātta.
Why "restricted"? íbhyā and kṣatríyā. These words also occur in the masculine, so this rule does not reach them. ibhya is ādyudātta by yato 'nāvaḥ (SK 3701). kṣatriya is madhyodātta by phiṭ 62, with the udātta on trí. The Subodhinī adds that kṣatriya may also be madhyodātta by the accent of its affix.
Rule. A word ending in kha has the udātta on its final vowel, when it does not begin with ś or m. The Subodhinī divides aśmādeḥ as a-śmādeḥ, "not beginning with ś or m".
Example.
nakha → this rule → nakhám "nail", udātta khám.
ukhā́ "a kind of pot", sukhám "happiness", duḥkhám "sorrow", each with the udātta on its last syllable.
Counter-examples: śíkhā "crest" and múkham "face", with the udātta on the first syllable.
Applies when / blocked by. Each example escapes a different rule.
nakha: svāṅgaśiṭām (phiṭ 29), since a nail is a part of the body.
ukhā: khayy uvarṇaṃ kṛtrimākhyā cet (phiṭ 31), which would make its u udātta, since a pot is a made thing.
sukha and duḥkha: nabviṣayasya (phiṭ 26), since they are neuter.
For the counter-examples, mukha is ādyudātta by phiṭ 29. The 1915 print adds "or by nabviṣayasya" (phiṭ 26). śikhā is ādyudātta by one of two routes. The uṇādi rule that forms it calls its affix nit, so SK 3686 applies. Or phiṭ 29 applies before ṭāp, as the inner operation.
Watch out. Bhaṭṭoji quotes the uṇādi rule as śīṅaḥ kho nid dhrasvaś ca. In the uṇādi section (uṇādi 702) the rule reads only śīṅo hrasvaś ca. The Subodhinī says most uṇādi books read it so, as Bhaṭṭoji himself does there. That is why he adds the second route through phiṭ 29.
Rule. Words ending in hiṣṭha, vatsara, ti, śat or tha have the udātta on the final vowel. The heading gives the sūtra as baṃhiṣṭhavatsartiśatthānām. The 1915 print reads hiṣṭhavatsaratiśatthāntānām, "of words ending in …", as the Subodhinī glosses it; that reading is followed here.
Example.
bahula + iṣṭhan "most abundant" → baṃhi replaces bahula (the Subodhinī) → baṃhiṣṭha
→ the final vowel is udātta (this rule) → baṃhiṣṭhá, udātta ṣṭhá.
saṃvatsaráḥ "year", udātta rá.
saptatíḥ "seventy", aśītíḥ "eighty", udātta tí.
catvāriṃśát "forty", udātta śát.
abhyū̀rṇvānā́ prabhṛthásyāyóḥ: the thá of prabhṛthá is udātta. The grave on ū̀ marks the independent svarita that arises from abhi + ū. The nā́ is udātta.
Applies when / blocked by. Bhaṭṭoji names the rule each word escapes.
baṃhiṣṭha would be ādyudātta, because iṣṭhan is nit (SK 3686).
saptati, aśīti and catvāriṃśat would take the heavy-syllable udātta of laghāv ante (phiṭ 42).
For saṃvatsara and prabhṛtha, "they say" that the accent an indeclinable first member keeps is set aside here.
Watch out. In two mantras baṃhiṣṭha is ādyudātta after all: báṃhiṣṭhair áśvaiḥ suvṛ́tā ráthena "with most bountiful horses, with a well-rolling chariot", and yád báṃhiṣṭhaṃ nā́tivídhe. Here the udātta is on báṃ. Bhaṭṭoji puts this down to vyatyaya, the free interchange allowed in the Veda. The 1915 print reads parivatsaraḥ for saṃvatsaraḥ; the Subodhinī knows both readings. Bhaṭṭoji also notes that the tha of prabhṛtha is already covered by the thāthādi sūtra (SK 3878).
dakṣiṇa "skilful" → dakṣiṇá (udātta ṇá); in other senses dákṣiṇa (udātta dák)Six rules deal with four adjectives whose accent shifts with their sense. dakṣiṇa is antodātta when it means "skilful". It is ādyudātta in other senses, and in two cases either syllable may carry the udātta. kṛṣṇa "black" is antodātta in the Veda unless it names an animal, and optionally so as a proper name. śukla "white" and gaura "pale" are ādyudātta, optionally in Bhaṭṭoji's view. The running example is dakṣiṇá "skilful", udātta on the last syllable ṇá, against dákṣiṇa, udātta on dák.
Rule. In the sense "good at, skilful" (sādhu), dakṣiṇa has the udātta on its final vowel. The Subodhinī's example of this sense is vīṇāyāṃ dakṣiṇaḥ, "skilled on the vīṇā".
Example.
dakṣiṇa "skilful" → this rule → dakṣiṇá, udātta ṇá.
In other senses → dákṣiṇa, udātta dák.
Bhaṭṭoji gives no example sentence.
Applies when / blocked by. Bhaṭṭoji says what happens in the other senses. When dakṣiṇa expresses relative position ("southern, right"), it is a sarvanāman, a pronominal word. It is then ādyudātta by svāṅgaśiṭām (phiṭ 29). In any other sense, laghāv ante (phiṭ 42) makes its heavy syllable dak udātta, since its last two syllables are light.
Watch out. The 1915 print adds a line from the lexicon (kośa). It says dakṣiṇa is used of the straightforward, the generous, and one who follows another's wish.
Rule. When dakṣiṇa names a part of the body, its first vowel is optionally udātta. Otherwise its final vowel is. Bhaṭṭoji: the first and the last are udātta by turns (paryāyeṇa). dakṣiṇa is carried from phiṭ 8.
Example.
dakṣiṇo bāhuḥ "the right arm"
→ this rule → dákṣiṇaḥ (udātta dák) or dakṣiṇáḥ (udātta ṇá).
Applies when / blocked by. The alternative final udātta is the general one of phiṭ 1, still running. It is not phiṭ 8's, which is limited to the sense "skilful". This is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. Why does the sūtra say ākhyā, "the name"? Bhaṭṭoji: for a person seated facing west, the left hand is the "southern" (dakṣiṇa) one. There dakṣiṇa states the hand's position; it is not the name of the limb. So the option does not apply. This reading of his remark is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. In the Veda the same alternation holds even when dakṣiṇa does not name a body part. Bhaṭṭoji: "this rule is for a sense other than a body part" (asvāṅga). The Subodhinī: in the Veda the first and the last vowel of dakṣiṇa are udātta by turns.
Example.
dakṣiṇaḥ (Veda, any sense) → this rule → dákṣiṇaḥ (udātta dák) or dakṣiṇáḥ (udātta ṇá).
Rule. In the Veda, kṛṣṇa "black" has the udātta on its final vowel, unless it is the name of an animal. "In the Veda" is carried from phiṭ 10, as the Subodhinī says. Bhaṭṭoji: this antodātta is prescribed where varṇānāṃ taṇa- (phiṭ 33) would give ādyudātta.
Example.
kṛṣṇa → this rule → kṛṣṇá, udātta ṣṇá.
kṛṣṇó nonāva vṛṣabháḥ "the black bull has bellowed": udātta ṣṇó and bhá.
kṛṣṇānāṃ vrīhīṇām "of the black rice-grains".
As the name of an animal: kṛṣṇo rātryai "a black antelope for Night". Here the word keeps the first-syllable udātta of phiṭ 33, kṛ́ṣṇaḥ, udātta kṛ́. This form is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. In kṛṣṇó nonāva vṛṣabháḥ, the printed svarita signs are irregular. The accent given here is the one Bhaṭṭoji's argument needs: kṛṣṇa antodātta.
Rule. As a proper name, kṛṣṇa takes the final udātta only optionally. Bhaṭṭoji: "'of kṛṣṇa' is still in force" (kṛṣṇasyety eva).
Example.
ayáṃ vā kṛṣṇó aśvinā "this Kṛṣṇa, O Aśvins": udātta yáṃ and ṣṇó.
kṛṣṇarṣiḥ "the seer Kṛṣṇa": here kṛṣṇa is the seer's name.
Applies when / blocked by. Bhaṭṭoji shows only the antodātta form. The other form would be the first-syllable udātta of phiṭ 33, kṛ́ṣṇa. This is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. The reading vā in the mantra is doubtful; the sense, "you two, O Aśvins", would suit vāṃ. It is kept as vā here.
Rule. The first syllable of śukla "white" and gaura "pale" is udātta. Some say this is invariable. Bhaṭṭoji holds that "optionally" (vā, from phiṭ 12) is still in force.
Example.
gaura → this rule → gaúraḥ (udātta gaú); in the other option gauráḥ (udātta rá).
Bhaṭṭoji's evidence: sáro gauró yáthā piba "drink, as a gaura bull drinks at a lake". Here gauró has the udātta on its last syllable ró. So the rule cannot be invariable.
Other udāttas: sá and yá.
Applies when / blocked by. The Subodhinī explains how "some" meet this evidence. They carry nāmadheyasya, "as a name", from phiṭ 12 into this rule. In the mantra gaura is not a name, so it is antodātta by phiṭ 1. Elsewhere, at SK 3932, the Subodhinī itself applies the rule optionally: "the first and the last of gaura are udātta".
jyeṣṭha "eldest" → jyeṣṭhá (udātta ṣṭhá), against jyéṣṭhaḥ "best" (nit iṣṭhan, SK 3686; udātta jyé)The last ten rules of the first pāda give the final udātta to single words or small sets. Some apply only in the Veda, some only in one sense. Each again sets aside a rule that would put the udātta elsewhere. The pāda ends with an optional svarita. The running example is jyeṣṭhá "eldest", udātta on ṣṭhá, against jyéṣṭhaḥ "best", udātta on jyé.
Rule. In the Veda, aṅguṣṭha "thumb", udaka "water", baka "heron" and vaśā "cow" have the udātta on the final vowel.
Example.
aṅguṣṭha (Veda) → this rule → aṅguṣṭhá, udātta ṣṭhá.
udaka (Veda) → udakám, udātta kám.
Applies when / blocked by. Bhaṭṭoji: for aṅguṣṭha, svāṅgānām akurvādīnām (phiṭ 52) would make the second syllable udātta. The rule is undertaken to make it antodātta. The Subodhinī adds that baka is likewise taken out of prāṇināṃ kupūrvam (phiṭ 30). It also says that in ordinary language udaka belongs to the kardama-list (phiṭ 59). There its first or second syllable is udātta.
Watch out. Why is vaśā named? Bhaṭṭoji: as a restriction (niyama), "only in the Veda". So in ordinary usage vaśā is ādyudātta, "they say". Why a restriction? vaśā would be antodātta anyway by phiṭ 1, so naming it only confines that accent to the Veda. This is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. In the Veda, pṛṣṭha "back" also has the udātta on its final vowel. Bhaṭṭoji adds: optionally in ordinary language (bhāṣā). "In the Veda" and "the final" are carried from phiṭ 14.
Example.
pṛṣṭha (Veda) → this rule → pṛṣṭhám, udātta ṣṭhám.
pṛṣṭha (ordinary language) → pṛṣṭhám or pṛ́ṣṭham, udātta pṛ́.
Applies when / blocked by. The Subodhinī: in the other alternative pṛṣṭha is ādyudātta by svāṅgaśiṭām (phiṭ 29), as a part of the body.
Rule. arjuna has the udātta on its final vowel when it is the name of a grass.
Example.
arjuna (a grass) → this rule → arjuná, udātta ná.
Counter-example (the Subodhinī): arjuno vṛkṣaḥ "the arjuna tree" → árjunaḥ, udātta ár.
Applies when / blocked by. Bhaṭṭoji: this is an exception (apavāda) to the ādyudātta of unarvannantānām (phiṭ 32), which covers words ending in una. The tree keeps that ādyudātta.
Rule. arya has the udātta on its final vowel when it means "master" (svāmin). The heading gives the sūtra as āryasya. The 1915 print reads aryasya, as the Subodhinī does, and that reading is followed here.
Example.
arya "master" → this rule → aryá, udātta yá.
Applies when / blocked by. Bhaṭṭoji: the rule is stated because ādyudātta would otherwise come, either by yāntasyāntyāt pūrvam (phiṭ 62) or by yato 'nāvaḥ (SK 3701). The Subodhinī says this arya is the fixed form (nipātana) of aryaḥ svāmivaiśyayoḥ (SK 2851).
Watch out. SK 3701 concerns a two-vowel word in yat. That fits arya (ṛ + yat), not ārya. This supports the print's reading. The point is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. āśā "hope" has the udātta on its final vowel, provided it does not mean "direction". Bhaṭṭoji: the rule is there to exclude āśā as the name of a direction.
Example.
āśā "hope" → this rule → āśā́, udātta śā́.
In the sense "direction": índra ā́śābhyas pári "Indra … from the quarters". Here ā́śābhyas has the udātta on its first syllable ā́. The other udāttas are í and pá.
Applies when / blocked by. Bhaṭṭoji draws a jñāpaka from the exclusion. Because the rule excludes the sense "direction", a word meaning "direction" is ādyudātta. The mantra shows that ādyudātta, not this rule.
Watch out. The Subodhinī's gloss here says āśā is "ādyudātta" when it does not mean direction. That is probably a slip for "antodātta". The whole first pāda teaches antodātta, and the Subodhinī itself treats this āśā as antodātta at SK 3656. Scholars discuss this.
Rule. Names of lunar mansions (nakṣatra) that take the feminine affix āp have the udātta on the final vowel. The 1915 print reads the sūtra nakṣatrāṇām.
Example.
āśleṣā, anurādhā → this rule → āśleṣā́, anurādhā́, udātta ṣā́, dhā́.
jyeṣṭhā, śraviṣṭhā, dhaniṣṭhā → jyeṣṭhā́, śraviṣṭhā́, dhaniṣṭhā́, udātta ṣṭhā́.
Applies when / blocked by. Bhaṭṭoji names the rivals. For āśleṣā and anurādhā, laghāv ante (phiṭ 42) would make a heavy syllable udātta. jyeṣṭhā, śraviṣṭhā and dhaniṣṭhā end in the nit affix iṣṭhan. So they would be ādyudātta by ñnity ādir nityam (SK 3686), as the Subodhinī names it.
Watch out. The Subodhinī derives the iṣṭhan words. śraviṣṭhā is "the most famed", from śravavatī "she who has fame". dhaniṣṭhā is from dhanavatī. Before iṣṭhan the matup is elided (vinmator luk).
Rule. The final vowel is not udātta for the name Kṛttikā, where a k-class consonant (ku) stands before the final. Bhaṭṭoji: "the final is not udātta; Kṛttikā is the lunar mansion". So this rule takes back phiṭ 19's final udātta for this one name.
Example.
kṛttikā → phiṭ 19 blocked (this rule) → laghāv ante (phiṭ 42) → kṛ́ttikā, udātta kṛ́.
The Subodhinī supplies the result by phiṭ 42; Bhaṭṭoji gives no accented form.
Watch out. Bhaṭṭoji reports another view. Some explain the rule as "of words restricted to an āp that has a k-class consonant before it". On that view āryikā and bahulikā also lose the final udātta. Bhaṭṭoji reports this view without adopting it.
Rule. ghṛta "ghee" and the words of its list have the udātta on the final vowel. Bhaṭṭoji: the list is an open class (ākṛtigaṇa), known from usage rather than fixed.
Example.
ghṛta → this rule → ghṛtám mimikṣe, udātta tám.
Applies when / blocked by. The Subodhinī: the rule applies where nabviṣayasya (phiṭ 26) would give the first-syllable udātta. That presupposes that ghṛta is restricted to the neuter. That is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. In the sense of age, jyeṣṭha "eldest" and kaniṣṭha "youngest" have the udātta on the final vowel.
Example.
vṛddha + iṣṭhan → jya replaces vṛddha (SK 2013) → jyeṣṭha "eldest"
→ this rule → jyeṣṭhá, udātta ṣṭhá.
yuvan + iṣṭhan → kan replaces yuvan (SK 2019) → kaniṣṭhá "youngest", udātta ṣṭhá.
jyeṣṭhá āha camasā́ "the eldest said: 'the cups …'": udātta ṣṭhá and sā́.
kaniṣṭhá āha catúraḥ "the youngest said: 'four …'": udātta ṣṭhá and tú.
Counter-examples (not age): jyéṣṭhaḥ "best" and kániṣṭhaḥ "least", udātta jyé and ká.
Applies when / blocked by. Bhaṭṭoji: in the counter-examples the word is only ādyudātta, because iṣṭhan is nit (SK 3686). The Subodhinī calls this rule an exception (apavāda) to that nit accent. It derives the counter-examples too. jyeṣṭha "best" has jya for praśasya (SK 2009, SK 2011). kaniṣṭha "least" has kan for alpa.
Watch out. The same two words thus carry their udātta at opposite ends. The sense decides: age gives the last syllable, rank gives the first.
Rule. The final vowel of bilva and tiṣya is optionally svarita. Bhaṭṭoji: in the other alternative it is udātta.
Example.
bilva → this rule → bilvà (independent svarita on và) or bilvá (udātta vá).
tiṣya → tiṣyà or tiṣyá.
Bhaṭṭoji gives no examples; these are the sūtra's own words.
Applies when / blocked by. The Subodhinī explains the other alternative: "udātta, because 'udātta' is carried forward" (udātta ity anuvṛtteḥ). This is the evidence that "udātta" runs through the whole first pāda.
Watch out. The 1915 print closes the first pāda here: "thus ends the first pāda of the phiṭ-sūtras".
vána (neuter-only, not is-final) → váne (udātta vá, phiṭ 26); páñca (udātta páñ, phiṭ 28)The second pāda opens with a heading: from here the first syllable is udātta. The rules under it name the kinds of word that take this accent. These are feminine-only words in a short vowel and neuter-only words. Then come two-vowel grass and grain words and numerals in n or r. Then come a-final body parts and pronouns, animal words with a k-class consonant, -una, -ṛ and -van words, and colour words. One rule, phiṭ 31, names a non-initial vowel instead. The running examples are váne "in the tree", from vána, and páñca "five"; in each the first syllable is udātta.
Rule. "Now the first, before śakaṭi." This is a heading (adhikāra). In the rules before śakaṭiśakaṭyoḥ (phiṭ 69), the first syllable is udātta. Bhaṭṭoji: "a heading, as far as śakaṭiśakaṭyoḥ". The 1915 print heads this sūtra "second pāda".
Example. A heading has no example of its own. Under it, phiṭ 25 gives báliḥ, udātta on the first syllable bá.
Applies when / blocked by. The heading supplies "the first" (ādiḥ). "Is udātta" still comes from phiṭ 1. Bhaṭṭoji's glosses on the next rules ("the first is udātta") and the Subodhinī at phiṭ 23 show this. Neither says so at this sūtra.
Watch out. The heading does not reach every rule in its range. The Subodhinī says ādiḥ is not construed at phiṭ 42 or phiṭ 44. Phiṭ 31 names its own vowel, so ādiḥ plays no part there either.
Rule. A word that ends in a short vowel and is restricted to the feminine has its first syllable udātta.
Example.
bali → the first syllable is udātta (this rule) → báliḥ, udātta bá.
tanu "body" → tánuḥ, udātta tá.
Watch out. The reading baliḥ is doubtful. vali "fold" is feminine, while bali "offering" is masculine, so váliḥ may be meant. The accent would be the same.
Rule. A word restricted to the neuter has its first syllable udātta, unless it ends in is. Bhaṭṭoji: nap means "neuter" (napuṃsaka). The Subodhinī: "of a word that is always neuter".
Example.
vana "tree, wood" (neuter only) → this rule → vána
→ váne ná vāyáḥ "like a young bird in the tree": váne has the udātta on its first syllable vá.
Counter-example (ending in is): sarpíḥ "ghee", with the udātta on pí.
Applies when / blocked by. The Subodhinī says sarpis is made with the uṇādi affix isi (uṇādi 265). That its udātta falls on pí is not stated by Bhaṭṭoji or his commentators; it is inferred. Bhaṭṭoji's own sarpír at phiṭ 30 supports it. Three rules of the first pāda set this rule aside: phiṭ 3 (geha), phiṭ 6 (sukha, duḥkha) and phiṭ 21 (ghṛta).
Watch out. The 1915 print accents the mantra váne ná vāyáḥ, with váne ādyudātta as the rule requires; that reading is followed here. The Ṛgveda word-by-word text (padapāṭha) divides vā yaḥ. The accent of these syllables is the same either way.
Rule. Words for grasses and grains that have two vowels have the first syllable udātta. Bhaṭṭoji: aṣ means ac, a vowel. The Subodhinī: aṣ is the eastern grammarians' name for ac.
Example.
kuśa (a grass) → this rule → kúśāḥ, udātta kú.
kā́śāḥ (a grass), mā́ṣāḥ "beans", tílāḥ "sesame", udātta on the first syllable.
Counter-example (more than two vowels): godhū́māḥ "wheat", udātta dhū́.
Applies when / blocked by. The Subodhinī: godhūma is madhyodātta by laghāv ante (phiṭ 42).
Rule. A numeral ending in n or r has the first syllable udātta.
Example.
pañcan "five" → this rule → páñca, udātta páñ.
Bhaṭṭoji's second example is catvāraḥ "four". Its accent is left unmarked here; see below.
Watch out. The Subodhinī argues that catvāraḥ does not show this rule. It takes the udātta of the augment ām (SK 331), on vā. So the Subodhinī comments on catuṣkapāla as the example instead. That compound is ādyudātta by the accent of its prior member (SK 3763). The 1915 print, too, reads catvāraḥ. Scholars discuss this.
Rule. a-final words for parts of one's own body (svāṅga), and a-final śiṭ words, have the first syllable udātta. Bhaṭṭoji: śiṭ means a sarvanāman, a pronominal word.
Example.
karṇa "ear" → this rule → kárṇābhyāṃ chúbukād ádhi "from the two ears, from the chin": udātta kár, chú and á.
óṣṭhāv iva mádhu "like two lips … honey": udātta ó and má.
viśva "all" (a pronoun) → víśvo víhāyāḥ: udātta ví in each word.
Applies when / blocked by. Earlier rules cite this one as the rule they set aside or fall back on: phiṭ 4 (guda), phiṭ 6 (nakha, mukha, śikhā), phiṭ 8 (dakṣiṇa as a pronoun) and phiṭ 15 (pṛṣṭha).
Rule. In words for living beings, the syllable before a k-class consonant (ku) is udātta. Bhaṭṭoji: "the first syllable, before a ku, is udātta". So the syllable must be both the first and the one before the ku.
Example.
kāka "crow" → this rule → kā́kaḥ, udātta kā́.
vṛ́kaḥ "wolf", udātta vṛ́; śúkeṣu me "on the parrots … my", udātta śú.
Counter-example (not living beings): kṣīrám sarpír mádhūdakám "milk, ghee, honey, water". The udāttas are rám, pír, má and kám. The da before the k of udaka is not udātta, because water is not a living being.
Applies when / blocked by. The Subodhinī: udakám has the final udātta by phiṭ 14.
Watch out. Some texts add "Why 'before a ku'? dhanuḥ." The 1915 print lacks this sentence. dhanuḥ "bow" is not a living being, so it cannot be the counter-example the question needs. It may be a slip for dhenuḥ "cow". This is a conjecture, not stated by Bhaṭṭoji or his commentators.
Rule. In the name of something made (kṛtrima), a u-vowel followed by a khay consonant is udātta. khay is the group of voiceless stops (k, kh, c, ch, ṭ, ṭh, t, th, p, ph).
Example.
kanduka "ball" → the u before k is udātta (this rule) → kandúkaḥ, udātta dú.
Applies when / blocked by. Bhaṭṭoji does not gloss "name of something made" here. At phiṭ 6 he applies this rule to ukhā, "since it is a made thing", which supports the reading. This rule names the vowel that is udātta, so the heading's "first syllable" plays no part.
Rule. Words ending in una, ṛ or van have the first syllable udātta. The Subodhinī divides unarvan into the three endings.
Example.
Ending in una: váruṇaṃ vo riśā́dasam "Varuṇa, destroyer of foes, for you": udātta vá (and śā́).
Ending in ṛ: svásāraṃ tvā kṛṇavai "let me make you my sister": udātta svá.
Ending in van: pī́vānaṃ meṣám "the fat ram": udātta pī́ (and ṣám).
Applies when / blocked by. Two earlier rules set this one aside: phiṭ 2 for mātṛ and phiṭ 16 for arjuna as a grass.
Rule. Colour words ending in ta, ṇa, ti, ni or t have the first syllable udātta.
Example. Bhaṭṭoji gives one word for each ending, in order:
eta (ta) → étaḥ "variegated", udātta é.
hariṇa (ṇa) → háriṇaḥ "yellowish", udātta há.
śiti (ti) → śítiḥ "white", udātta śí.
pṛśni (ni) → pṛ́śniḥ "dappled", udātta pṛ́.
harit (t) → hárit "green", udātta há.
Applies when / blocked by. Phiṭ 11 overrides this rule for kṛṣṇa in the Veda, unless it names an animal.
cañcā iva → (kan, SK 2051, elided by lup, SK 2053) cáñcā (udātta cáñ: a man like a straw figure)The heading of phiṭ 24 continues. Five rules name single words or meaning-sets with the first syllable udātta: muni-type words, akṣa, ardha, pītadru and grāma. Three rules then deal with names given by comparison, where the affix has been elided by lup. Such a name is ādyudātta, except for some animal, tree and mountain names, and a king's name only under a condition. The running example is cáñcā, "a man like a cañcā", a straw figure. Its first syllable cáñ is udātta.
Rule. In a word ending in a short vowel, the short vowel that stands first is udātta, if it is not ṛ. The rule holds "in the sense of habitual disposition" (tācchīlye). Bhaṭṭoji: "excluding ṛ, the short vowel that is first in a word ending in a short vowel is udātta." The 1915 print reads anṛttācchīlye, that is anṛt + tācchīlye. That reading is followed here.
Example.
muni "sage" → the first short vowel is udātta (this rule) → múniḥ, udātta mú.
Applies when / blocked by. Why "not ṛ"? The Subodhinī: to exclude nṛśaṃsaḥ "cruel", whose first vowel is a short ṛ.
Watch out. Neither Bhaṭṭoji nor the Subodhinī explains tācchīlye. Reading it as "habitual disposition" is not stated by Bhaṭṭoji or his commentators; it is inferred. Both muni and nṛśaṃsa can be taken as words of disposition.
Rule. akṣa has the first syllable udātta when it does not mean "gaming" (devana), that is, a die.
Example.
akṣa "axle" → this rule → ákṣaḥ
→ na + ákṣaḥ → one udātta vowel for both (SK 3658) → tásya nā́kṣaḥ "its axle … not". The udātta of á survives in the merged nā́; tá is udātta too.
In the sense of gaming: akṣaír mā́ dīvyaḥ "do not play with dice". Here the first syllable of akṣa is not udātta; kṣaí is.
Applies when / blocked by. The quotation tásya nā́kṣaḥ breaks off before its verb. It still shows the point: the merged vowel keeps the udātta of akṣa's first syllable.
Rule. ardha has the first syllable udātta when it does not express an equal share, that is, when it means "a part".
Example.
ardha "a part" → this rule → árdho grāmasya "a part of the village", udātta ár.
Counter-example (an equal share): ardhaṃ pippalyāḥ "half of the long pepper".
Watch out. Bhaṭṭoji does not say where the udātta of ardham "an exact half" falls. He says only that this rule does not apply. The pair contrasts the masculine ardhaḥ "a part" with the neuter ardham "a half".
Rule. Words meaning pītadru, a kind of pine, have the first syllable udātta. "Meaning" (-artha) covers pītadru and its synonyms.
Example.
pītadru → this rule → pī́tadruḥ, udātta pī́.
sarala → sáralaḥ, udātta sá.
Watch out. Bhaṭṭoji writes pītadruḥ saralaḥ with no stop between. This may be two examples, or one: "pītadru, that is, sarala". The accents are the same on either reading.
Rule. grāma and the words of its list also have the first syllable udātta. The Subodhinī supplies "the first is udātta", which Bhaṭṭoji leaves unstated.
Example.
grāma "village" → this rule → grā́maḥ, udātta grā́.
sómaḥ, udātta só; yā́maḥ "watch", udātta yā́.
Rule. A name for the thing compared (upameya) whose affix has been elided by lup has the first syllable udātta. lup is one of the named kinds of elision of an affix.
Example.
cañcā iva "like a cañcā" → kan (ive pratikṛtau, SK 2051) → kan elided by lup (lum manuṣye, SK 2053), since a man is meant → cañcā
→ this rule → cáñcā, udātta cáñ.
Bhaṭṭoji: cañcā iva cañcā, "one like a cañcā is called cañcā". Only the name, the second cañcā, takes the accent.
Counter-example (the Subodhinī): agnir māṇavakaḥ "the boy called Agni". Here there is no lup.
Applies when / blocked by. The Subodhinī names SK 2051 for the kan. At SK 2053 Bhaṭṭoji glosses cañcā as "a man made of grass" and ties the kan to the one taught "in a name". Scholars discuss this.
Watch out. A variant reading of the sūtra is sphigantasya. Bhaṭṭoji: sphik is the earlier grammarians' name for lup.
Rule. Phiṭ 39 does not apply to names of particular trees and mountains, nor to vyāghra, siṃha and mahiṣa. Bhaṭṭoji: "when these are names of a thing compared, their first syllable is not udātta."
Example.
tāla iva tālaḥ "one like a palmyra palm is called tāla".
merur iva meruḥ "one like Meru is called Meru".
vyāghraḥ "tiger", siṃhaḥ "lion", mahiṣaḥ "buffalo", as names of a person compared.
In none of these is the first syllable udātta. Bhaṭṭoji says only where the udātta is not, so the forms are left unmarked.
Applies when / blocked by. The Subodhinī: the sūtra says "particular" (viśeṣa) trees and mountains. So the generic words still take phiṭ 39: in vṛkṣa ivāyaṃ vṛkṣaḥ "this one, like a tree, is called vṛkṣa" and in parvataḥ, the first syllable is udātta.
Rule. The name of a particular king, given by comparison with lup, takes phiṭ 39 only if it is yamanvā. Bhaṭṭoji: yamanvā means vṛddha. vṛddha is Pāṇini's name for a word whose first vowel is a vṛddhi vowel (SK 1335).
Example.
āṅga iva āṅgaḥ "this one, like the Āṅga, is Āṅga" (the Subodhinī)
→ first vowel ā, so vṛddha → phiṭ 39 applies → ā́ṅgaḥ, udātta ā́.
Counter-example: aṅgāḥ. Its first vowel a is not a vṛddhi vowel, so the rule does not apply.
Some texts read the counter-example as āṅgāḥ, which would itself be vṛddha; the 1915 print reads aṅgāḥ, followed here.
Applies when / blocked by. The Subodhinī calls this rule a restriction (niyama) on phiṭ 39, which would already cover a king's name. It reports two forms of the restriction without choosing. Some "say" it means: "a particular king's name, only if vṛddha". The Phiṭ-vṛtti, a commentary on the phiṭ-sūtras, shows it as: "a vṛddha name, only if a particular king's".
kalyāṇa (final ṇa light; heavy kal and yā) → a heavy syllable is udātta; kukkuṭa → kukkúṭaḥ (light penultimate kú)These five rules close Part II. laghāv ante (phiṭ 42) is the rule that so many earlier rules set aside. In a long word with light syllables at its end, it makes a heavy syllable udātta. A syllable is heavy (guru) if its vowel is long or a consonant cluster follows it; otherwise it is light (laghu). Around it stand rules for feminine and colour words, for birds, for names of seasons and living beings, and for grain words. The running examples are kalyāṇa "auspicious", whose heavy syllables kal and yā stand before a light final, and kukkúṭaḥ "cock", whose light syllable before the final, kú, is udātta.
Rule. In a word with many vowels (bahvaṣ = bahvac), when the final syllable is light, or the last two are, a heavy syllable is udātta.
Example.
kalyāṇa: final ṇa light; heavy kal and yā → a heavy syllable is udātta → kalyāṇaḥ.
kolāhala: last two syllables ha, la light; heavy ko and lā → kolāhalaḥ.
Which heavy syllable takes the udātta is not stated by Bhaṭṭoji, so the two words are left unmarked.
Applies when / blocked by. The heading's "first syllable" is not construed here, the Subodhinī says. So a heavy syllable that is not initial can also take the udātta. That is why other works say the heavy syllable of vṛṣākapi (SK 494) is udātta. Earlier rules set this rule aside at phiṭ 2, phiṭ 7 and phiṭ 19. It applies at phiṭ 8 (dákṣiṇa), phiṭ 20 (kṛ́ttikā) and phiṭ 27 (godhū́māḥ).
Watch out. The Subodhinī says the two examples are "by turns" (paryāyeṇa) ādyudātta and madhyodātta. This may mean one each: kályāṇaḥ and kolā́halaḥ. Or it may mean that either heavy syllable may be udātta in each word. Scholars discuss this. The Subodhinī also reports, as others' view, that "the first of the heavy syllables" must not be read into the rule. The commentator Haradatta, under SK 497, calls sāraṅga and kalmāṣa madhyodātta by this rule.
Rule. The first syllable is udātta in three kinds of word. These are words restricted to the feminine, colour words, and words in which something stands before akṣu.
Example.
Feminine only: mallikā (jasmine) → mállikā, udātta má.
Colour words: śyénī, udātta śyé; háriṇī, udātta há.
Something before akṣu: tara + akṣu → tárakṣuḥ "hyena", udātta tá.
Applies when / blocked by. Phiṭ 5 sets this rule aside for antardhā, which becomes antardhā́.
Watch out. Bhaṭṭoji explains akṣupūrva: "those which have something before the word akṣu".
Rule. In words for birds (śakuni), a light syllable before the final is udātta. Bhaṭṭoji: "a light syllable before [the final] is udātta."
Example.
kukkuṭa "cock": the syllable before the final is the light ku → this rule → kukkúṭaḥ, udātta kú.
tittiri "partridge" → tittíriḥ, udātta tí.
Applies when / blocked by. The Subodhinī: "before" means "before the final". The heading's "first syllable" is not construed here, because the sūtra says pūrvam. Both examples are madhyodātta.
Watch out. The 1915 print adds a third example, khañjarīṭaḥ. Its syllable before the final, rī, is heavy, so it does not fit the rule. The print's own Subodhinī says "these two", which fits the two examples given here.
Rule. In the name of a season (ṛtu) or of a living being (prāṇin), the udātta that a rule would otherwise give does not come. Bhaṭṭoji: "the udātta obtained according to the rule (yathālakṣaṇam) does not take place." The 1915 print reads -prāṇyākhyāyām, as the Subodhinī does; that reading is followed here.
Example.
vasantaḥ "spring" and kṛkalāsaḥ "lizard". The barred udātta does not come.
Applies when / blocked by. Neither Bhaṭṭoji nor the Subodhinī names the barred rule, or the accent that results. In both words laghāv ante (phiṭ 42) would make a heavy syllable udātta: san in vasanta, lā in kṛkalāsa. That this is the barred rule is not stated by Bhaṭṭoji or his commentators; it is inferred. The words are therefore left unmarked.
Rule. Grain words that are vṛddha and end in k, and grain words that end in ṣ, have the first syllable udātta. The 1915 print reads dhānyānāṃ, "of grain words", as the Subodhinī does; that reading is followed here.
Example.
Ending in k: śyāmāka (a millet; first vowel ā, so vṛddha) → śyā́mākāḥ, udātta śyā́.
Ending in ṣ: rājamāṣa (a bean) → rā́jamāṣāḥ, udātta rā́. The 1915 print reads māṣāḥ here.
Applies when / blocked by. The Subodhinī objects to the examples. śyāmāka already has this accent from the uśīra-list (phiṭ 67), and māṣa from phiṭ 27. So naiṣādakāḥ and rājamāṣāḥ should have been cited. The rājamāṣāḥ given here is in fact the example it asks for. The Subodhinī prints the list's name as uśīnara-; that it means uśīra- (phiṭ 67) is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. At phiṭ 67 the Subodhinī turns the point round. There it calls the mention of śyāmāka in phiṭ 67 doubtful, because this rule already covers it. Scholars discuss this. The Subodhinī also reports the Phiṭ-vṛtti, which reads the rule without ca and makes "the heavy syllable" udātta, citing naiṣādakāḥ and kālākṣāḥ.
kekaya (janapada, vowel-final) → (ādiḥ udātta) kékayaḥ; kṛṣi (two vowels, ik-final) → kṛ́ṣiḥThe phiṭ-sūtras of Śāntanava give the accent of stems that no affix explains. Their second pāda (section) is governed by the heading ādiḥ, "the first syllable [is udātta]" (phiṭ 24). That heading is still in force here. Three more word-classes take the udātta on their first syllable, and the second pāda closes.
The running examples are two words. kekaya is the name of a country and ends in a vowel, so its first syllable is udātta: kékayaḥ, with the udātta on ké. kṛṣi "ploughing" has two vowels and ends in i, so it too is udātta on the first syllable: kṛ́ṣiḥ, with the udātta on kṛ́.
Rule. In a word for a country (janapada) that ends in a vowel, the first syllable is udātta. The phiṭ-sūtras call a vowel aṣ; it is Pāṇini's ac. The Subodhinī glosses aṣāntānām as ajantānām, "ending in a vowel".
Example.
kekaya (a country name, ending in a)
→ the heading ādiḥ (phiṭ 24): the first syllable is udātta
→ kékayaḥ: the udātta is ké; ka and yaḥ are anudātta.
Applies when / blocked by. Bhaṭṭoji adds that the rule is stated because yāntasyāntyāt pūrvam (phiṭ 62) would otherwise apply. That rule would put the udātta on the syllable before the final ya, that is, the second syllable, ka. The 1915 print lacks this sentence.
Watch out. Scholars discuss this example. The Subodhinī objects that phiṭ 62, being the later rule, ought to win (paratvāt) in kekaya. So, it reports, aṅgāḥ and vaṅgāḥ should be the examples.
Rule. Take a word that begins with a consonant and ends in the syllable la, where the l does not follow a conjunct consonant. Its first syllable is udātta, or optionally the syllable before the final. Bhaṭṭoji explains that hay is the phiṭ name for Pāṇini's hal, a consonant.
Example.
palala → pálalam (udātta on the first syllable pá) or palálam (udātta on lá, the syllable before the final).
śalala → śálalam (udātta śá) or śalálam (udātta lá).
Applies when / blocked by. Bhaṭṭoji gives two counter-examples. ekalaḥ begins with a vowel, so the rule does not reach it. In mallaḥ the final la follows the conjunct ll, so again the rule does not apply. The Subodhinī reports that some carry janapadānām, "of country names", forward from phiṭ 47. They cite pañcālāḥ and kośalāḥ, with palālam as the counter-example.
Watch out. The words antaḥ pūrvaṃ vā could also be read "the final, or the one before it". Bhaṭṭoji gives no gloss. The reading taught here is the Subodhinī's: the first syllable, or optionally the one before the final.
Rule. In a word of two vowels that ends in an ik vowel (i, u, ṛ, ḷ), the first syllable is udātta. dvyaṣ is the phiṭ form of dvyac, "having two vowels".
Example.
kṛṣi (two vowels, ṛ and i; ends in i)
→ the heading ādiḥ: the first syllable is udātta
→ kṛ́ṣiḥ: the udātta is kṛ́.
Applies when / blocked by. This sūtra ends the second pāda of the phiṭ-sūtras. The 1915 print closes it here with a colophon.
Watch out. Scholars discuss this example. The Subodhinī reports that some carry janapadānām forward here too. They cite the country names kuravaḥ and cedayaḥ. For them kṛṣi is a counter-example, because in akṣair mā dīvyaḥ kṛṣim it kṛṣasva the word kṛṣi is antodātta (udātta on its final).
lalāṭa (svāṅga, three vowels, l-initial) → (dvitīyam udātta) lalā́ṭam; pādapa (heavy-initial, pa-final) → pādápaḥThe third pāda of the phiṭ-sūtras opens with two headings. dvitīyam, "the second syllable [is udātta]", holds until phiṭ 66. tryacām, "of words with three vowels", holds until phiṭ 57. Under both, five word-classes take the udātta on their second syllable.
The running examples are lalāṭa "forehead" and pādapa "tree". lalāṭa names a body part, has three vowels and begins with l. Its second syllable is udātta: lalā́ṭam, with the udātta on lā́. pādapa has three vowels, ends in pa and begins with a heavy syllable. Its second syllable is udātta: pādápaḥ, with the udātta on dá.
Rule. This is a heading (adhikāra). From here on, "the second [syllable is udātta]". It holds up to, but not including, īṣāntasya hayādeḥ (phiṭ 66). The third pāda of the phiṭ-sūtras begins here.
Example. The heading supplies the udātta lā́ of lalā́ṭam (phiṭ 52), and every other second-syllable udātta from phiṭ 52 to phiṭ 65.
Applies when / blocked by. The word "udātta" is not in the sūtra. Bhaṭṭoji supplies it at phiṭ 52: dvitīyam udāttam.
Rule. This is a second heading: "of words with three vowels (tryac)". It holds up to, but not including, the sūtra that begins with makara (phiṭ 57).
Example. la-lā-ṭa has three vowels, so it falls under this heading. The rules of phiṭ 52–56 apply only to such words.
Watch out. The two headings end at different points. tryacām stops before phiṭ 57; dvitīyam runs on to phiṭ 66. So in phiṭ 57–65 the second syllable can be named for words of any length.
Rule. In a three-vowel word for a body part (svāṅga), the second syllable is udātta. The words must not begin with a k-class consonant (ku), with r or with v.
Example.
lalāṭa (body part; three vowels; begins with l)
→ dvitīyam (phiṭ 50): the second syllable is udātta
→ lalā́ṭam: the udātta is lā́.
Applies when / blocked by. Bhaṭṭoji gives three counter-examples: kapola "cheek" begins with k, rasanā "tongue" with r, vadana "mouth" with v. The Subodhinī says each then follows a rule of its own. kapólaḥ has its udātta on pó by laghāv ante (phiṭ 42). rásanā and vádanam have it on the first syllable, rá and vá, by svāṅgaśiṭām (phiṭ 29).
Watch out. The counter-examples are not unaccented words. This rule simply does not decide them; another rule does.
Rule. In a three-vowel word beginning with m, the second syllable is udātta.
Example.
malaya → maláyaḥ: the udātta is lá, the second syllable.
makara → makáraḥ: the udātta is ká.
Watch out. The Subodhinī calls these poor examples, because a rule of its own covers each word. makara is named in phiṭ 57, and malaya falls under phiṭ 62. For malaya, phiṭ 62 picks the same syllable, the one before the final ya. The Subodhinī would cite mahendra instead.
Rule. In a word for a pot-herb (śāka) that begins with ś, under the three-vowel heading, the second syllable is udātta.
Example.
śītanyā → śītányā: the udātta is tá.
śatapuṣpā → śatápuṣpā: the udātta is again tá.
Applies when / blocked by. The Subodhinī reports that some read sādīnām, "beginning with s", and cite sarṣapā.
Watch out. śa-ta-puṣ-pā has four syllables, although the rule stands under the three-vowel heading. Neither Bhaṭṭoji nor the Subodhinī remarks on this.
Rule. In a three-vowel word that ends in pa and begins with a heavy (guru) syllable, the second syllable is udātta. A heavy syllable has a long vowel, or a short vowel before a conjunct consonant.
Example.
pādapa (first syllable pā, heavy) → pādápaḥ: the udātta is dá, the second syllable.
ātapa (first syllable ā, heavy) → ātápaḥ: the udātta is tá, the second syllable.
Applies when / blocked by. Bhaṭṭoji gives two words the rule does not reach. anūpa begins with a light syllable, a. nīpa has only two syllables. The Subodhinī says, "so they say", that both take a final udātta from compound rules: anūpám by anor apradhānakanīyasī (SK 3923) and nīpám by ner anidhāne (SK 3926). The udātta is on pám in each.
Rule. In a three-vowel word ending in yuta, ani or aṇi, the second syllable is udātta.
Example.
ending in yuta: ayuta → ayútam: the udātta is yú.
ending in ani: dhamani → dhamániḥ: the udātta is má.
ending in aṇi: vipaṇi → vipáṇiḥ: the udātta is pá.
Applies when / blocked by. The 1915 print reads yute in the sūtra; the Subodhinī reads yuta, as taught here.
makara → (ādir vā; else dvitīyam) mákaraḥ / makáraḥ; sugandhitejana → súgandhitejanāḥ / sugándhitejanāḥ / sugandhitéjanāḥThe three-vowel heading has ended, but "the second syllable" is still in force. Five sūtras now give a choice. The udātta falls on the first syllable or on the second, and in one word on a third place as well.
The running examples are makara "sea-monster" and sugandhitejana. makara may be mákaraḥ, udātta on má, or makáraḥ, udātta on ká. sugandhitejana has three possible places: sú, gán or té.
Rule. Twelve words take the udātta on the first syllable, or else on the second. They are makara, varūḍha, pārevata, vitastā, ikṣu, ārji, drākṣā, kalā, umā, kāṣṭhā, peṣṭhā and kāśī. Bhaṭṭoji: "of these the first or the second is udātta".
Example.
makara → mákaraḥ (udātta má) or makáraḥ (udātta ká).
varūḍha → várūḍhaḥ (udātta vá) or varū́ḍhaḥ (udātta rū́).
Applies when / blocked by. The three-vowel heading stops just before this sūtra (phiṭ 51). That is why two-vowel words, ikṣu and umā, can be named here. The heading "the second" (phiṭ 50) is still in force, which supplies the alternative.
Watch out. The Subodhinī reads pārāvata for pārevata and peṣṭā for peṣṭhā. It also says that kāśīnānām would be the expected form; the missing n is a sage's licence (ārṣa).
Rule. In the Veda, words other than makara and the rest also take the udātta on the first or the second syllable. Bhaṭṭoji: "it is begun for words other than makara etc.; according to usage (lakṣyānusārāt) the first and the second are to be understood as udātta". Which of the two a given word takes is learned from the attested forms.
Example. Bhaṭṭoji gives none. The Subodhinī names kāśyapaḥ, kardamaḥ, kulaṭā, udakam and gāndhāriḥ, without accent marks.
Applies when / blocked by. "The first … or the second" is carried forward from phiṭ 57, as the Subodhinī's gloss shows.
Watch out. The last four of the Subodhinī's words may belong to the next rule, the kardamādi list. The Subodhinī does not settle this.
Rule. For kardama "mud" and the words listed with it (kardamādi), the first or the second syllable is udātta. Bhaṭṭoji: "the first or the second is udātta".
Example. Bhaṭṭoji gives none. The Subodhinī elsewhere calls udaka "water" one of this list.
Applies when / blocked by. The rule is not limited to the Veda. The Subodhinī says of udaka that this holds "in the spoken language" (bhāṣāyām). That "in the Veda" stops at phiṭ 58 is not stated by Bhaṭṭoji or his commentators; it is inferred from that remark. That "the first or the second" comes from phiṭ 57 is likewise inferred.
Rule. In sugandhitejana, the syllable te is udātta, or else the first or the second syllable. Bhaṭṭoji: "the first, the second and the syllable te: these three are udātta by turns (paryāyeṇa)". The word has one udātta, in any of the three places.
Example.
sugandhitejana → súgandhitejanāḥ (udātta sú, the first syllable)
or sugándhitejanāḥ (udātta gán, the second)
or sugandhitéjanāḥ (udātta té).
Applies when / blocked by. The Subodhinī says that "the first" means the first syllable of the whole word, su.
Rule. In a neuter word ending in phala "fruit", the first or the second syllable is udātta. nap is the phiṭ name for the neuter (napuṃsaka), as the Subodhinī explains.
Example.
rājādanaphala (neuter) → rā́jādanaphalam (udātta rā́) or rājā́danaphalam (udātta jā́).
Applies when / blocked by. The Subodhinī's counter-example is dāsīphalo vṛkṣaḥ. Here the word is a bahuvrīhi, a compound that describes something outside itself. It qualifies vṛkṣa "tree", so it is masculine, and the rule does not apply. That the alternatives come from phiṭ 57 is not stated by Bhaṭṭoji or his commentators; it is inferred.
kulāya (ya-final) → (antyāt pūrvam) kulā́yaḥ; halīṣā (hay-initial, īṣā-final) → (ādir vā) hálīṣā; mahiṣī (wife) → máhiṣīThese seven sūtras fix the udātta by the shape of the word or by naming the words. One rule takes the syllable before the last. One takes nā and a light syllable. Two name lists that take the penultimate or the second syllable. Phiṭ 66 then ends the heading "the second", and the last two rules return to the first syllable.
The running examples are three. kulāya "nest" ends in ya, so the syllable before the last is udātta: kulā́yaḥ, udātta lā́. halīṣā "plough-pole" begins with a consonant and ends in īṣā, so its first syllable may be udātta: hálīṣā, udātta há. mahiṣī meaning "wife" is udātta on the first syllable: máhiṣī, udātta má.
Rule. In a word ending in the syllable ya, the syllable before the last is udātta. The Subodhinī: "of a word ending in the syllable ya, the one before the last is udātta".
Example.
kulāya → kulā́yaḥ: the udātta is lā́.
talaya → taláyaḥ: the udātta is lá.
valaya → valáyaḥ: the udātta is lá.
Applies when / blocked by. The 1915 print gives only kulāyaḥ. This is the rule that phiṭ 47 sets aside for the country name kekaya, by Bhaṭṭoji's account.
Rule. In a word ending in tha, the syllable nā and a light (laghu) syllable are udātta. A light syllable has a short vowel not followed by a conjunct consonant.
Example.
sanāthā sabhā "an assembly that has a protector"
→ sanā́thā: the udātta is nā́, the syllable before the last.
Applies when / blocked by. The Subodhinī reports another reading of the sūtra, āntasya ca nālaghunī. It explains: "of a word ending in ā, and of the word nā, the light syllables before the last are udātta". Its examples are nā́nā, dívā and múdhā. Each has one udātta, on nā́, dí and mú.
Watch out. The sūtra does not say where the light syllable stands. Nor does it say whether the light sa of sanāthā is udātta as well. Only nā is marked here. The single udātta of the Subodhinī's examples supports this.
Rule. In śiśumāra "porpoise", udumbara "fig tree", balīvarda "bull", uṣṭrāra and purūravas, the syllable before the last is udātta, or else the second. Bhaṭṭoji: "the one before the last is udātta, or the second".
Example. Bhaṭṭoji gives no example. Applied to śi-śu-mā-ra, the rule names mā (before the last) or śu (the second): śiśumā́ra or śiśúmāra. These forms are not stated by Bhaṭṭoji or his commentators; they are inferred.
Applies when / blocked by. The ca "and" brings in "the one before the last" of phiṭ 62, as Bhaṭṭoji's gloss shows. The option "or the second" is Bhaṭṭoji's gloss; the sūtra does not say it. The Subodhinī gives śiṃśumāra as a variant reading.
Rule. In the four names sāṅkāśya, kāmpilya, nāsikya and dārvāghāṭa, the second syllable is optionally udātta. Bhaṭṭoji: "the second is udātta, optionally".
Example. Bhaṭṭoji gives none.
Applies when / blocked by. The Subodhinī notes how the words are formed. sāṅkāśya and kāmpilya end in the affix ṇya, by vuñchaṇ- (SK 1292). dārvāghāṭa ends in the affix aṇ, by a vārttika under SK 2966. It does not say what the accent is in the other case, and neither does Bhaṭṭoji.
Rule. In a word that begins with a consonant and ends in īṣā, the first syllable is optionally udātta. This sūtra is the limit of the heading "the second" (phiṭ 50), so that heading no longer applies here.
Example.
halīṣā → hálīṣā: the udātta is há, the first syllable.
lāṅgalīṣā → lā́ṅgalīṣā: the udātta is lā́.
Applies when / blocked by. The Subodhinī: "of a word ending in īṣā and beginning with a hal, the first is optionally udātta". Neither Bhaṭṭoji nor the Subodhinī states the accent in the other case.
Rule. In thirteen words the first syllable is udātta. They are uśīra, dāśera, kapāla, palāla, śaivāla, śyāmāka, śārīra, śarāva, hṛdaya, hiraṇya, araṇya, apatya and devara. Bhaṭṭoji: "of these the first is udātta". The heading ādiḥ (phiṭ 24) is still in force.
Example. Bhaṭṭoji gives none. By the rule, hṛdaya "heart" has its udātta on the first syllable hṛ́: hṛ́dayam. This form is not stated by Bhaṭṭoji or his commentators; it is inferred.
Applies when / blocked by. The Subodhinī explains why the words are named. hṛdaya is named to stop the second-syllable udātta of phiṭ 52, since it is a three-vowel body-part word. Most of the others are named to stop laghāv ante (phiṭ 42). The division into thirteen words is not stated by Bhaṭṭoji or his commentators; it is inferred (dāśera + kapāla, not dāśeraka + pāla).
Watch out. The Subodhinī calls two names questionable (cintya). śyāmāka already has this accent by dhānyānāṃ ca vṛddhakṣāntānām (phiṭ 46). devara already has it by phiṭ 42.
Rule. mahiṣī and aṣāḍhā take the udātta on the first syllable, if they name a wife and a brick respectively. The Subodhinī supplies "respectively" (yathāsaṃkhyam): mahiṣī goes with "wife", aṣāḍhā with "brick".
Example.
mahiṣī (the chief queen, a wife) → máhiṣī jāyā: the udātta is má.
aṣāḍhā (a Vedic brick) → áṣāḍhā upadadhāti "he lays down the aṣāḍhā bricks": the udātta is á.
Applies when / blocked by. The Devanāgarī of the heading reads āṣāḍhā, with a long first vowel. The 1915 print reads aṣāḍhā, the name of the brick; the transliteration and this section follow the print. This sūtra ends the third pāda of the phiṭ-sūtras.
śakaṭi → (akṣaram akṣaraṃ paryāyeṇa) śákaṭiḥ / śakáṭiḥ / śakaṭíḥThe fourth pāda of the phiṭ-sūtras opens with a new kind of rule. In a few named words, each syllable in turn (akṣaram akṣaraṃ paryāyeṇa) may be the udātta. The word still has only one udātta, but any of the named syllables may carry it. Phiṭ 69 is also the limit of the heading ādiḥ, "the first syllable", which began at phiṭ 24.
The running example is śakaṭi. It has three syllables, and each may be the udātta: śákaṭiḥ (udātta śá), śakáṭiḥ (udātta ká) or śakaṭíḥ (udātta ṭí).
Rule. In śakaṭi and śakaṭī, each syllable by turns is udātta. Bhaṭṭoji adds only the word "udātta". The fourth pāda of the phiṭ-sūtras begins here.
Example.
śakaṭi → śákaṭiḥ (udātta śá) / śakáṭiḥ (udātta ká) / śakaṭíḥ (udātta ṭí).
śakaṭī → śákaṭī (udātta śá) / śakáṭī (udātta ká) / śakaṭī́ (udātta ṭī́).
Bhaṭṭoji gives the two words without marks and does not spell out the three forms.
Applies when / blocked by. The Devanāgarī of the heading reads akṣarakṣaraṃ. The transliteration gives akṣaram akṣaraṃ "each syllable", the wording Bhaṭṭoji repeats at phiṭ 70.
The heading athādiḥ prāk śakaṭeḥ (phiṭ 24) ran "up to śakaṭi". So from this sūtra on, "the first syllable" is no longer understood.
Watch out. "In turn" does not give a word three udāttas. Each form has one udātta; the three forms are alternatives.
Rule. When goṣṭhaja is the name of a brāhmaṇa, each syllable by turns is udātta. Bhaṭṭoji repeats the wording of phiṭ 69: akṣaram akṣaraṃ paryāyeṇodāttam.
Example.
goṣṭhajo brāhmaṇaḥ "the brāhmaṇa named Goṣṭhaja"
→ góṣṭhajaḥ (udātta gó) / goṣṭhájaḥ (udātta ṣṭhá) / goṣṭhajáḥ (udātta já).
Applies when / blocked by. Elsewhere the word means "born in a cow-pen": goṣṭhajáḥ paśuḥ "a beast born in a cow-pen". Then it has only the final udātta, já. Bhaṭṭoji says this comes from the accent by which a final member ending in a kṛt affix keeps its own accent (kṛduttarapadaprakṛtisvara). This is presumably gatikārakopapadāt kṛt (SK 3873); that is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. In pārāvata "dove", every syllable except the penultimate (upottama) is udātta in turn. Bhaṭṭoji: "the rest are udātta in turn (krameṇa)". The penultimate of pā-rā-va-ta is va.
Example.
pārāvata → pā́rāvataḥ (udātta pā́) / pārā́vataḥ (udātta rā́) / pārāvatáḥ (udātta tá).
va is never the udātta.
Applies when / blocked by. "In turn" presupposes the rule of phiṭ 69. Bhaṭṭoji's śeṣam "the rest" shows this.
Rule. In dhūmrajānu, muñjakeśa, kālavāla and sthālīpāka, the syllables other than dhū, ja, la and sthā are udātta in turn. Bhaṭṭoji: "of these four, leaving out the four beginning with dhū, the remaining ones are udātta in turn".
Example. Bhaṭṭoji gives the four words unmarked. One excluded syllable belongs to each word: dhū to dhūmrajānu, ja to muñjakeśa, la to kālavāla, sthā to sthālīpāka. Applying the rule gives:
dhūmrajānu → dhūmrájānuḥ (udātta mrá) / dhūmrajā́nuḥ (jā́) / dhūmrajānúḥ (nú).
muñjakeśa → múñjakeśaḥ (udātta múñ) / muñjakéśaḥ (ké) / muñjakeśáḥ (śá).
sthālīpāka → sthālī́pākaḥ (udātta lī́) / sthālīpā́kaḥ (pā́) / sthālīpākáḥ (ká).
The pairing of word and excluded syllable, and these forms, are not stated by Bhaṭṭoji or his commentators; they are inferred.
Watch out. kālavāla has two syllables la. The text does not say which one is left out, so its forms are not given.
Rule. The sūtra names kapikeśa and harikeśa, "in the Veda". Bhaṭṭoji gives only the two words: kapikeśaḥ, harikeśaḥ.
Example. No accented form is given here.
Watch out. Neither Bhaṭṭoji nor his commentators says what this rule does. The natural reading is that "each syllable in turn" (phiṭ 69) still holds, as in phiṭ 70–72. That is not stated by Bhaṭṭoji or his commentators; it is inferred.
nyaṅ → (svarita) nyàṅ; tila + yat → tilya → (antaḥ svaritaḥ, not the ādyudātta of SK 3701) tilyàmThe phiṭ-sūtras now turn from udātta to svarita. A svarita that a rule gives a syllable on its own is an independent svarita; it is marked here with a grave accent (à). A svarita that only follows an udātta (SK 3660) is a dependent one and stays unmarked. Four sūtras name words with an independent svarita, on the first or on the final syllable.
The running examples are nyaṅ, which is svarita as a whole: nyàṅ. And tilya, made from tila "sesame" with the affix yat. yato 'nāvaḥ (SK 3701) would make it udātta on the first syllable. Phiṭ 76 instead gives its final syllable the svarita: tilyàm, svarita on yà.
Rule. The words nyaṅ and svar are svarita. Bhaṭṭoji: "this is clear" (spaṣṭam).
Example.
nyàṅṅ uttānáḥ: nyàṅ is svarita; the udātta of uttānáḥ is ná.
vyàcakṣayat svàḥ: svàḥ is svarita; vyà also carries a svarita sign.
The marks other than nyàṅ and svàḥ follow the signs of the 1915 print.
Applies when / blocked by. In nyàṅṅ uttānáḥ the ṅ of nyaṅ is doubled before a vowel.
Rule. In nyarbuda and vyalkaśa the first syllable is svarita. Bhaṭṭoji: "is svarita".
Example.
nyàrbudāya svā́hā "svāhā to the nyarbuda, a high numeral": the svarita is nyà.
svā́hā is udātta on svā́, as a particle (phiṭ 80).
pākadūrvā vyàlkaśā: the svarita is vyà.
Applies when / blocked by. "svarita" is carried from phiṭ 74; that is not stated by Bhaṭṭoji or his commentators; it is inferred from Bhaṭṭoji's gloss. The 1915 print gives the gloss but not the examples.
Rule. In tilya, śikya, matya, kārṣmarya, dhānya, kanyā, rājanya and manuṣya, the final syllable is svarita. Bhaṭṭoji: "is svarita".
Example.
tila + yat → tilya: "a field where sesame grows" → tilyàm: the svarita is yà.
vaiśvānaryā śikyàm ādatte "with the Vaiśvānarī verse he takes up the sling": svarita kyà.
prabhinnāya matyàm anvyāsyati: svarita tyà.
kārṣmaryàm upayāti: svarita ryà.
dhānyàm asi dhinuhi "you are grain; gladden …": svarita nyà.
rājanyò vajrasya rūpam "the rājanya is the form of the thunderbolt": svarita nyò.
pratīcīṃ manuṣyā̀ḥ: svarita ṣyā̀ḥ.
Two more examples, vajraḥ kārṣmaryor vajreṇa and māvīravīkanyā, are left unmarked, because their word division is uncertain.
Applies when / blocked by. Bhaṭṭoji says the rule applies where yato 'nāvaḥ (SK 3701) would give the first syllable the udātta, as in tilya (tila + yat). This rule sets that aside.
Watch out. The 1915 print reads the sūtra differently, and its only example is tilyam. The sūtra in the heading spells karṣmarya; the examples spell kārṣmarya.
Rule. In the Veda, bilva, bhakṣya and vīrya have svarita on the final syllable. Bhaṭṭoji: "they have svarita on the final" (antasvaritāni).
Example.
tato bilvà ud atiṣṭhat "from that the bilva tree arose": the svarita is và.
There is no example for bhakṣya. A second example, balaṃ virujavīryam, is left unmarked, because its word division is uncertain.
Applies when / blocked by. "Final" comes from phiṭ 76 and "svarita" from phiṭ 74; that is not stated by Bhaṭṭoji or his commentators; it is inferred from his gloss.
svāhā (nipāta) → (ādyudātta) svā́hā; eva → (antaḥ) evá; yathā at the end of a pāda → anudātta: táṃ nemím ṛbhávo yathāEight sūtras settle the accent of small words. Four small words, tvat, tva, sama and sima, are wholly anudātta. Particles (nipāta) and the verbal prefixes (upasarga) take the udātta on the first syllable. A list headed by eva takes it on the final, and a list headed by vācā on both syllables. The ca-group is anudātta, and so is yathā at the end of a verse-quarter.
The running examples are three. The particle svāhā is udātta on its first syllable: svā́hā, udātta svā́. eva is udātta on its final: evá, udātta vá. And in táṃ nemím ṛbhávo yathā, yathā closes the verse-quarter and has no udātta.
Rule. tvat, tva, sama and sima are "not high" (anuccāni), that is, anudātta. The Subodhinī, in the 1915 print: "all these are anudātta".
Example.
starī́r u tvat: tvat is anudātta. The udātta is rī́; ru carries only the dependent svarita that follows it.
utá tvaḥ páśyan "and one, seeing …": tvaḥ is anudātta, between the udāttas tá and pá.
nábhantām anyaké same "let all the others burst asunder": same is anudātta. Its sa carries only the dependent svarita after the udātta ké.
símasmai: the texts print an udātta on sí.
Applies when / blocked by. tvaḥ is a form of tva. The reading tvaṃ, found in some copies, would be the pronoun "you" instead.
Watch out. The printed símasmai goes against this rule, which makes sima anudātta. The sign is recorded as printed, not corrected. Compare phiṭ 79.
Rule. In the Atharvaveda, sima has its final syllable udātta. The Subodhinī, in the 1915 print: "the word sima is udātta in the Atharvaveda".
Example.
vāsas tanute simásmai "she spreads her garment for all": the udātta is má, the final of sima.
This line is from the Ṛgveda.
Applies when / blocked by. Bhaṭṭoji explains why a Ṛgveda line is cited. "In the Atharvaveda" states what usually holds (prāyikam). Or it means "of sima as seen there" (tatra dṛṣṭasya). Either way the final udātta goes with the word, and so it occurs in the Ṛgveda too.
Rule. Particles (nipāta) are ādyudātta: the udātta falls on the first syllable.
Example.
svāhā (a particle) → svā́hā: the udātta is svā́.
Applies when / blocked by. The same rule gives yáthā its udātta on yá when it does not end a verse-quarter (phiṭ 85).
Rule. The upasargas, except abhi, are also ādyudātta. Bhaṭṭoji gives no comment. The Subodhinī: "leaving out the word abhi, the upasargas are ādyudātta".
Example. No example stands here. The rule can be seen in úpa tvāgne divé-dive (SK 3717), udātta ú, and in the first prá of práprāyám (phiṭ 87), udātta prá.
Applies when / blocked by. The ca "and" brings in "ādyudātta" from phiṭ 80. That is not stated by Bhaṭṭoji or his commentators; it is inferred from the ca.
Rule. In eva and the words listed with it, the final syllable is udātta. Bhaṭṭoji notes a variant reading, evamādīnām.
Example.
evá (udātta vá), evám (udātta vám), nūnám (udātta nám), sahá (udātta há).
sahá te putra sūribhiḥ "together with your patrons, O son".
Applies when / blocked by. The sūtra says only "the final". That the final is udātta is not stated by Bhaṭṭoji or his commentators; it is inferred from the sense and from Bhaṭṭoji's remark on saha.
Watch out. Bhaṭṭoji reports that the earlier authorities (prāñcaḥ) call saha ādyudātta. They say so under sahasya saḥ (SK 1009). He calls that view open to question (cintyam). It conflicts with this rule, which makes sahá final-udātta.
Rule. In vācā and the words listed with it, both syllables are udātta.
Example. Bhaṭṭoji gives none.
Applies when / blocked by. Bhaṭṭoji explains the word "both". It sets aside anudāttaṃ padam ekavarjam (SK 3650), by which a word is anudātta except for one syllable. Without "both", only one udātta could stand.
Rule. ca and the words listed with it (cādayaḥ) are anudātta. Bhaṭṭoji: "this is clear" (spaṣṭam).
Example. Bhaṭṭoji gives none.
Watch out. cādayaḥ is presumably the ca-group of particles (A 1.4.57). If so, this rule sets aside phiṭ 80 for them, and they have no udātta. Both points are not stated by Bhaṭṭoji or his commentators; they are inferred.
Rule. The word yathā at the end of a metrical quarter (pāda) is anudātta. The Subodhinī: "the word yathā is anudātta at the end of a pāda".
Example.
táṃ nemím ṛbhávo yathā "as the Ṛbhus bend the felly": yathā ends the quarter and is wholly anudātta. The udāttas are táṃ, mím and bhá.
Applies when / blocked by. Bhaṭṭoji asks why "at the end of a pāda" is said, and answers with yáthā no áditiḥ kárat "so that Aditi may make it for us". Here yathā does not end the quarter. It is ādyudātta as a particle (phiṭ 80): udātta yá. That "anudātta" is carried from phiṭ 84 is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. The accents of the first line follow the 1915 print, which leaves yathā wholly unmarked. Note also the two senses of pāda: here a quarter of a verse, elsewhere a section of the phiṭ-sūtras.
paṭu + paṭu (prakāra) → (final of the latter member udātta) paṭupaṭúḥ; dive + dive (nitya) → (latter member wholly anudātta) divé-diveThe last two phiṭ-sūtras give the accent of doubled words. When a word is doubled to express likeness, the final of the second member is udātta. In other doublings, the second member is wholly anudātta. With this the phiṭ-sūtras end.
The running examples are paṭupaṭuḥ "somewhat clever", udātta on ṭú, the final of the second paṭu. And divé-dive "day by day": the first dive keeps its udātta vé, and the second dive is wholly anudātta.
Rule. In a doubling for likeness (prakāra) and the like, the final vowel of the latter member is udātta. Bhaṭṭoji: "the final of the latter is udātta" (parasyānta udāttaḥ). The Subodhinī names the doubling: that of prakāre guṇavacanasya (SK 2147) and its kind.
Example.
paṭu + paṭu (doubled for likeness: "somewhat clever")
→ the final of the latter member is udātta
→ paṭupaṭúḥ: the udātta is ṭú; the rest is anudātta.
Applies when / blocked by. The 1915 print sets parasyānta udāttaḥ inside the sūtra; it is taught here as Bhaṭṭoji's gloss. Either way those words supply the operation.
Watch out. The Subodhinī raises an objection. By karmadhārayavad uttareṣu (SK 2146) such a doubling is treated as a karmadhāraya, a compound whose members refer to the same thing, one describing the other. So samāsasya (SK 3734) would already make it antodātta, and this rule seems pointless. It replies that the phiṭ-sūtras were current before Pāṇini's grammar.
Rule. In the other doublings, the latter member is wholly anudātta. Bhaṭṭoji: "'the rest' means the latter member of a word doubled for nitya and the like". nitya is constant repetition (nityavīpsayoḥ, SK 2140). The Subodhinī: in a doubling other than the likeness doubling, the latter member is wholly anudātta.
Example.
pra + pra + ayam
→ the first prá is udātta, as an upasarga (phiṭ 81)
→ the second pra is wholly anudātta; its a merges with the a of ayam
→ práprāyám: the udāttas are prá and yám; the second pra is low.
dive + dive "day by day" → divé-dive: the udātta is vé; the second dive is wholly anudātta. Its di carries only the dependent svarita after an udātta (SK 3660).
Applies when / blocked by. The 1915 print closes with a colophon: here ends the fourth pāda of the phiṭ-sūtras composed by Śāntanavācārya.
agni (affix ni, uṇādi 490) → (pratyaya ādyudātta) agníḥ; kṛ + tavya → kartávyam; yajña + sya (sup) → (anudātta) yajñásyaWith the phiṭ-sūtras done, the 1915 print opens a new heading: "now the remaining accent rules" (atha svaraprakaraṇaśeṣāḥ). The SK returns to the Aṣṭādhyāyī. Two sūtras of the third book give the basic accent of an affix (pratyaya). An affix is udātta on its first vowel. But case endings (sup) and affixes marked with p (pit) are anudātta.
The running examples show both rules. In agni the affix is ni, so its first vowel is udātta: agníḥ, udātta ní. In kartavya the affix is tavya: kartávyam, udātta tá. In yajñasya the case ending stays anudātta: yajñásya, udātta jñá, not the ending.
Rule. An affix (pratyaya) is ādyudātta: its first vowel is udātta. Bhaṭṭoji: "an affix is to be ādyudātta, and only that (eva)". The heading pratyayaḥ comes from SK 180 (A 3.1.1).
Example.
aṅg + ni (the affix ni, per the Subodhinī)
→ the n of the root is elided (aṅger nalopaś ca, uṇādi 490) → agni
→ the affix's first vowel is udātta
→ agníḥ: the udātta is ní.
kṛ + tavya → kartavya
→ the first vowel of tavya is udātta
→ kartávyam: the udātta is tá.
Applies when / blocked by. The 1915 print sets the heading atha svaraprakaraṇaśeṣāḥ before this sūtra. The Subodhinī says the affix in kartavyam must be tavya, not tavyat. tavyat is marked with t (tit), and a tit affix will be taught as svarita (tit svaritam, SK 3729).
Watch out. Bhaṭṭoji's eva may be no more than emphatic. The rule itself is general; the next sūtra gives an exception to it.
Rule. The case endings (sup) and the affixes marked with p (pit) are anudātta. Bhaṭṭoji: "this is an exception (apavāda) to the preceding rule" (SK 3708).
Example.
yajña + ṅas (genitive singular, a sup ending, which becomes sya)
→ the ending is anudātta
→ yajñásya: the udātta is jñá; sya is low.
yucch + śap + tip → ná yó yúcchati "who does not falter":
→ the root is udātta by dhātoḥ (SK 3671): yú
→ śap (a) is pit, so anudātta; after the udātta it becomes a dependent svarita (SK 3660)
→ tip (ti) is pit, so anudātta; after the svarita it becomes pracaya (SK 3668).
Bhaṭṭoji: "since śap and tip are anudātta, [they become] svarita and pracaya". The Subodhinī gives the steps and the rule numbers.
Watch out. pracaya is not a fourth accent. It is the level pitch that anudātta syllables take after a svarita in continuous recitation (SK 3668).
anya + akac (cit) → (final of the whole udātta) anyaké; agni + ḍhak (kit) → āgneyáḥThree sūtras of the sixth book move the udātta to the final. The words "the final is udātta" are carried on from karṣātvato ghaño 'nta udāttaḥ (SK 3680). This is not stated by Bhaṭṭoji or his commentators for these sūtras; it is inferred. A word formed with an affix marked with c (cit) is udātta on its final. So is a word formed with a taddhita marked with k (kit).
The running examples are two. anya "other" takes the affix akac (cit), and the whole word becomes final-udātta: anyaké, udātta ké. agni takes the taddhita ḍhak (kit): āgneyáḥ, udātta yá.
Rule. The final vowel of what has a cit affix is udātta. Bhaṭṭoji: "the final is to be udātta". A vārttika extends this: "together with the base (saprakṛteḥ), for the sake of bahuc and akac". Bhaṭṭoji: "when a cit affix is present, the final of the whole of base and affix is to be udātta".
Example.
anya + akac (inserted before the last vowel by avyayasarvanāmnām akac prāk ṭeḥ, SK 2026, per the Subodhinī) → anyaka
→ the final of the whole word is udātta
→ nábhantām anyaké same "let all those others burst": the udātta of anyaké is ké.
yaké sárasvatīm ánu: yaka from yad with akac; the udātta is ké.
takát súte: takat from tad with akac; the udātta is kát. The signs make su udātta as well.
Applies when / blocked by. The Subodhinī explains the vārttika. akac is itself made cit. That would be pointless for a one-vowel affix, unless the whole word were meant. Its example for bahuc is bahupaṭavaḥ.
Watch out. The ke of anyaké is the final of the whole word, not of the affix alone. The word division of takát súte is not settled by the texts.
Rule. The final vowel of a cit taddhita is udātta. Bhaṭṭoji: "although this is already established by the preceding rule (SK 3710), it is taught to block the accent of a ñit affix (ñitsvara)".
Example.
kuñja + cphañ (gotre kuñjādibhyaś cphañ, SK 1099) → kauñjāyana
→ then ñya (SK 1100), which gets the name tadrāja (SK 2072), and its luk, loss without trace, in the plural (SK 1193), per the Subodhinī
→ the final of the cit taddhita is udātta
→ kauñjāyanā́ḥ "the many descendants of Kuñja": the udātta is nā́.
Applies when / blocked by. The affix cphañ is marked with ñ as well as c. The Subodhinī explains that without this rule the ñit accent would win, being the later rule (paratvāt). The ñit accent is presumably ñnity ādir nityam (SK 3686), which puts the udātta on the first syllable. That identification is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. The final vowel of a kit taddhita is udātta. Bhaṭṭoji: "the final of a kit taddhita is udātta". "taddhita" is carried on from SK 3711.
Example.
agni + ḍhak (agner ḍhak, SK 1236, per the Subodhinī) → āgneya
→ the final of the kit taddhita is udātta
→ yad āgneyáḥ: the udātta is yá.
vāc + ṭā → (monosyllable before su; third case on) vācā́; aṣṭā + bhis → aṣṭābhírThe last six sūtras of this part make a case ending (vibhakti) udātta after certain stems. Normally a case ending is anudātta (SK 3709). Here it carries the udātta itself: after tisṛ, after a stem of one syllable, optionally after a one-syllable final member, after añc in the Veda, after a named list, and after aṣṭā.
The running examples are two. vāc "speech" has one syllable, so the third-case ending is udātta: vācā́, udātta cā́. aṣṭā "eight", in its long-ā form, makes the ending of bhis udātta: aṣṭābhír, udātta bhír.
Rule. After tisṛ, the final vowel of the ending jas (nominative plural) is udātta. Bhaṭṭoji: "the final is udātta".
Example.
tisṛ + jas (as) → ṛ becomes r (aci ra ṛtaḥ, SK 299) → tisraḥ
→ the final of jas is udātta
→ tisró dyā́vaḥ savitúḥ "three are the heavens of Savitṛ": the udātta of tisró is ró.
Applies when / blocked by. The Subodhinī gives the reason. tisṛ replaces the antodātta tri (SK 298), so it is antodātta too. jas is anudātta by SK 3709. Once r replaces ṛ, jas would become svarita by udāttasvaritayor yaṇaḥ svarito 'nudāttasya (SK 3657). This rule blocks that and makes it udātta.
Rule. sau means "before su", the seventh-case plural ending. After a stem that is monosyllabic (ekāc) before that su, a vibhakti from the third case onward is udātta.
Example.
vāc + ṭā (third case) → vāc has one syllable before su (vākṣu)
→ the ending is udātta
→ vācā́ virūpaḥ: the udātta is cā́, the ending.
Applies when / blocked by. Bhaṭṭoji gives three counter-examples.
Why "before su"? In rājñā the stem rājñ- is monosyllabic only because its a is lost. Before su it has two syllables. So the ending is not udātta: rā́jño nú te, udātta rā́. The Subodhinī adds that rājan is ādyudātta, being formed with kanin (uṇādi 154).
Why "monosyllabic"? vidhaté rā́jani tvé: rājan has two syllables, so ni stays low. That rājani is the word meant is not stated by Bhaṭṭoji; it is inferred. The udātta té of vidhaté comes from another rule (SK 3719).
Why "from the third onward"? ná dadarśa vā́cam: the second-case ending am is not udātta; the udātta is vā́.
Rule. After a final member (uttarapada) that is antodātta and monosyllabic, a vibhakti from the third case onward is optionally udātta. This holds only in a compound not prescribed under the nitya heading. ekācaḥ and tṛtīyādir vibhaktiḥ run on from SK 3714 (Subodhinī).
Example.
parama + vāc → the compound is antodātta by samāsasya (SK 3734), so vāc is an antodātta one-syllable final member
→ paramavāc + ṭā
→ paramavācā́: the udātta is optionally cā́, the ending.
Otherwise the form is paramavā́cā, udātta vā́. This second form is not stated by Bhaṭṭoji or his commentators; it is inferred from the word "optionally".
Applies when / blocked by. The Subodhinī gives counter-examples. agnicit is an upapada compound: its prior word agni is the word in whose presence the affix of cit is taught (SK 3001). It is prescribed under the nitya heading, so the option does not apply. The heading is meant, not mere "nitya": a compound that is nitya only because it has no dissolving phrase of its own words still takes the option, as in avācā brāhmaṇena. In rājadṛṣadā the final member is not monosyllabic. It also says uttarapada is named so that "monosyllabic" describes the final member, not the whole compound.
Rule. In the Veda, a vibhakti after añc is udātta, if it is not a sarvanāmasthāna. The sarvanāmasthāna endings are the strong endings: su, au, jas, am, auṭ, and śi in the neuter.
Example.
dadhi + añc (kvin, SK 373) → n lost (SK 415), a lost (SK 416), the vowel lengthened (SK 417) → dadhīc, per the Subodhinī
→ dadhīc + ṅas → the ending is udātta
→ índro dadhīcáḥ: the udātta of dadhīcáḥ is cáḥ.
pratyañc + śas → pratīcó bāhū́n: the udātta of pratīcó is có.
Applies when / blocked by. Bhaṭṭoji names the rule this overrides. By cau (SK 3652) the final of the prior member would have been udātta. And "from the third case onward" is still carried forward from SK 3714. So "not a sarvanāmasthāna" is said to bring in śas, the second-case plural, as in pratīcó.
Watch out. The printed form is unclear; the form meant is the second-case plural bāhū́n, udātta hū́n.
Rule. After ūṭh, idam, the six words beginning with pad, ap, pum, rai and div, a vibhakti that is not a sarvanāmasthāna is udātta. The Subodhinī: ekācaḥ "monosyllabic" and antodāttāt "antodātta" run on here. Bhaṭṭoji names antodāttāt as carried forward.
Example.
ūṭh: praṣṭhauháḥ, praṣṭhauhā́: the udātta is on the ending, háḥ and hā́.
idam: ebhír nṛ́bhir nṛ́tamaḥ "most manly with these men": the udātta is bhír.
the pad-words: padbhyā́ṃ bhū́miḥ "the earth from his feet": udātta bhyā́.
dadbhír ná jihvā́ "like the tongue with the teeth": udātta bhír.
áharahar jāyate māsí-māsi "is born day after day, month after month": udātta sí.
mánaś cin me hṛdá ā́ "the mind, in my heart": udātta dá.
ap: apā́ṃ phénena "with the foam of the waters": udātta pā́.
pum: abhrātéva puṃsáḥ "like a brotherless woman, toward men": udātta sáḥ.
rai: rāyā́ vayám "we, with wealth": udātta yā́. rāyó dhartā́ vivásvataḥ: udātta yó.
div: úpa tvāgne divé-dive "to you, Agni, day after day": udātta vé.
Applies when / blocked by. A vārttika adds that ūṭh counts only as the penultimate sound (upadhā). So the rule does not apply in akṣadyúvā, akṣadyúve. The Subodhinī explains praṣṭhauha. ṇvi comes by SK 3409–3410 and ūṭh by vāha ūṭh (SK 329). The compound is antodātta by SK 3873, and ūṭh itself is monosyllabic.
idam does not take this accent in re-mention (anvādeśa), because "antodātta" is carried forward. There idam is replaced by an anudātta aś (SK 350): prá te babhrū́ … mā́bhyāṃ gā́ ánu. The udātta mā́ comes from merging mā with ābhyām, not from this rule.
The six pad-words are those of paddannomāshṛnniś- (SK 228): pad, dat, nas, mās, hṛd, niś. The Subodhinī: only these six, not the later words of that list, because "monosyllabic" runs on.
Watch out. The printed accent of apā́ṃ is unclear; it is given here in the form the rule requires, udātta on the ending.
Rule. After aṣṭan "eight" in its long-ā form, a vibhakti from śas onward is udātta. Bhaṭṭoji: "a vibhakti from śas onward is udātta". These are the endings that are not sarvanāmasthāna.
Example.
aṣṭan + bhis → long ā (aṣṭana ā vibhaktau, SK 371) → aṣṭābhis
→ the ending is udātta
→ aṣṭābhír daśabhiḥ "with eight, with ten": the udātta of aṣṭābhír is bhír.
Applies when / blocked by. The Subodhinī calls this an exception to jhaly upottamam (SK 3683). That rule still governs the short form aṣṭabhiḥ, with the udātta on the penultimate ṭá: aṣṭábhiḥ. aṣṭan is antodātta as one of the ghṛtādi (phiṭ 21). The Subodhinī seems to say that the carried-on "antodātta" is therefore met here too, though its printed wording is uncertain.
Watch out. Why say "long"? The Subodhinī answers that the word is a jñāpaka, a wording that shows something not stated outright. It shows that the ā of SK 371 is optional. If ā were obligatory, "long" would exclude nothing.
jñā + śnā + śatṛ → jānat + ṅīp → jānatī́; netṛ + ṅīp → netrī́The last rules of Part III made a case ending (vibhakti) udātta after certain stems, for example after a one-syllable stem (SK 3714). This group adds two more kinds of stem. The first is a present participle in śatṛ without an inserted n. The second is a stem whose udātta vowel has turned into a yaṇ (y, v, r, l). After them, the following element takes the udātta. That element is either the feminine ī (a nadī) or a case ending from śas onward. The third rule denies this after the yaṇ of ūṅ and of a root. There the ending becomes an independent svarita instead. The running examples are jānatī́ "knowing (feminine)" and netrī́ "she who leads". In both, the udātta stands on the last syllable, tī and trī.
The case endings are counted in a fixed order: su, au, jas, am, auṭ, śas, ṭā …. "From śas onward" means the accusative plural śas and every ending after it. So it leaves out the first five endings.
Rule. Take a stem made with the affix śatṛ (the present participle in -at) that has no inserted n (num). The stem must already be antodātta, with its last syllable udātta. After it, a nadī is udātta. So is a vowel-initial case ending from śas onward. "antodātta" is carried forward from SK 3715, and "from śas onward" from SK 3716. Here the nadī is the feminine affix ṅīp, heard as ī.
Example.
jñā + śnā + śatṛ
→ jā replaces jñā (SK 2511), and the ā of nā is lost (SK 2483) → jānat
→ the śatṛ stem is antodātta by the accent of the affix → jānat, udātta on nat
→ jānat + ṅīp → the ī is udātta by this rule → jānatī́, udātta on tī.
Bhaṭṭoji quotes it in a Vedic line: ácchā rávaṃ prathamā́ jānatī́ "knowing, foremost, toward the call".
With a case ending: kṛṇvaté. Here the dative ending -e carries the udātta, té. The Subodhinī forms kṛṇvat by SK 2332.
Applies when / blocked by.
dádhatī it is not: SK 3673 makes the first syllable of a reduplicated stem udātta. So the udātta stays on da, and the ī is not udātta.tudántī the root tud takes the class sign śa, and num comes by SK 445. The śatṛ follows the a of śa, so SK 3730 makes it anudātta. The a of śa and the a of at merge, and the merged vowel is udātta by SK 3658. So the udātta stands on dán, and the ī stays anudātta.Watch out. The Subodhinī draws a lesson from the word "without num" (anumaḥ). A śatṛ stem with num could be antodātta only through the merged vowel of SK 3658. So the word shows that this merger counts for the present rule. A wording that only makes sense if some principle holds is called a jñāpaka, an "indicator" of that principle.
Rule. Suppose an udātta vowel has been replaced by a yaṇ, and a consonant (hal) stands before that yaṇ. After it, a nadī and a case ending from śas onward are udātta. Without this rule SK 3657 would make them svarita, as a vowel after the yaṇ of an udātta.
Example.
netṛ is made with tṛc, a c-marked affix, so it is antodātta by SK 3710: udātta on tṛ
→ netṛ + ṅīp → ṛ becomes r before the vowel (SK 47) → netr + ī
→ the r stands for an udātta ṛ and follows the consonant t → the ī is udātta → netrī́.
Bhaṭṭoji gives eṣā́ netrī́ "she, the leader" and codayitrī́ sūnṛ́tānām "she who stirs up glad songs". In both, the udātta is on trī.
With a case ending: savitṛ + ṅe → savitré, udātta on tré. The line is ṛtáṃ devā́ya kṛṇvaté savitré "… the truth, for the god Savitṛ who brings it about".
Watch out. The Subodhinī would add n to this rule as well. Then the feminine ī after an n that stands for an udātta would be udātta too, as in vākpatnī. Bhaṭṭoji does not say this.
Rule. The previous rule does not apply after the yaṇ of the feminine affix ūṅ, or after the yaṇ of a root (dhātu). Only the case endings from śas onward are meant here. The Subodhinī explains why: no nadī can follow these two.
Example.
brahmabandhū + ṭā (instrumental singular)
→ the ū of ūṅ is udātta by the affix accent (Subodhinī) → brahmabandhū́ + ā
→ ū becomes v → brahmabandhvā
→ the ending is denied the udātta by this rule → SK 3657 makes it svarita → brahmabandhvā̀, an independent svarita on vā.
A root example: subhū is a kvip word, antodātta by SK 3873. subhū + ṅe → the ū becomes v (SK 281) → subhvè, an independent svarita on vè. The line is sét pṛ́śniḥ subhvè.
Watch out. An independent svarita is a svarita made by a rule in its own right, here by SK 3657. It is marked with a grave (vā̀). A svarita that merely follows an udātta is a dependent one and is left unmarked.
abdi + matup → abdimā́n; ṣaṣ + bhis → ṣaḍbhírThis group gives the udātta to two more things that follow a stem. One is the affix matup (-mat or -vat, "having"). The other is the genitive plural nām. Then SK 3725 makes a consonant-initial ending udātta after certain numerals. The running examples are abdimā́n "having waters", with the udātta on mā, and ṣaḍbhír "by six", with the udātta on the ending bhír.
Rule. After an antodātta stem that ends in a short vowel, the affix matup is udātta. So is matup after the augment nuṭ (an added n). "antodātta" is still carried forward from SK 3715.
Example.
abdi "that in which the waters are placed" takes ki (SK 3271) and is antodātta (Subodhinī)
→ abdi + matup → matup is udātta → abdimā́n, udātta on mā.
udani is antodātta as one of the ghṛtādi words (phiṭ 21), so its matup is udātta too: udanimā́m̐.
Bhaṭṭoji's line: yó abdimā́m̐ udanimā́m̐ íyarti.
With nuṭ: akṣi + matup → akṣan (SK 322) → the n is lost → matup takes nuṭ → m becomes v (SK 1897) → akṣaṇvántaḥ, udātta on ván.
The line is akṣaṇvántaḥ kárṇavantaḥ sákhāyaḥ "friends having eyes, having ears".
Applies when / blocked by. The stem must be antodātta. In mā́ tvā vidad íṣumān "let no archer find you", iṣu is ādyudātta, so matup stays anudātta. The Subodhinī explains that iṣu's affix is n-marked.
Watch out. There is a maxim: "in an accent rule a consonant counts as non-existent" (svaravidhau vyañjanam avidyamānavat). Bhaṭṭoji says it does not hold here. So marut, which ends in t, is not treated as ending in a short u. Its matup stays anudātta: marútvān indra "Indra, accompanied by the Maruts". Likewise niyútvān vāyav ā́ gahi "Vāyu, with your teams, come here".
A vārttika (a supplementary rule) adds re: "and after re". rayi becomes re before matup, so no short vowel is left. Yet matup is still udātta: revā́m̐ íd revátaḥ "the wealthy one, of the wealthy", udātta on vā́m̐ and vá.
Rule. After such an antodātta stem, the ending nām is optionally udātta. "Short" and "matup" are carried forward from SK 3722, but matup now means "before matup". So the stem must end in a vowel that is short before matup.
Example.
sumati is a bahuvrīhi compound, antodātta by SK 3906 (Subodhinī)
→ sumati + nām → the i is lengthened → sumatīnām
→ nām is optionally udātta → sumatīnā́m, udātta on nā́m.
Bhaṭṭoji's line: cétantī sumatīnā́m "heeding good thoughts".
Watch out. The Subodhinī explains "short before matup". In sumatīnām and agnīnām the vowel is long, because it is lengthened before nām. What counts is that it is short before matup. The wording also keeps the rule away from tisṛṇām. There the ending is udātta without option, by SK 3725.
Rule. In the Veda, after the feminine ī (ṅī), nām is udātta variously (bahulam). Bhaṭṭoji glosses bahulam here as "optionally".
Example.
devasenā́nām abhibhañjatīnā́m "of the divine armies, as they break [the foe]". In abhibhañjatīnā́m the ending nā́m is udātta.
Because the rule is optional, it does not apply in jáyantīnāṃ marúto yantu "let the Maruts go [at the head] of the conquering [armies]". There nām is anudātta, and the udātta is on já.
Rule. After a numeral of the class called ṣaṭ, and after tri "three" and catur "four", a case ending that begins with a consonant is udātta. "Case ending" is carried forward from SK 3714. The Subodhinī's examples show that the class ṣaṭ covers pañcan, saptan, aṣṭan and navan as well as ṣaṣ.
Example.
ṣaṣ + bhis → ṣaḍbhiḥ → the consonant-initial bhis is udātta → ṣaḍbhíḥ.
Bhaṭṭoji's line: ā́ ṣaḍbhír hūyámānaḥ "being invoked by six", udātta on bhír.
tri + bhis → tribhíḥ: tribhíṣ ṭváṃ deva "you, O god, by three", udātta on bhíṣ.
Applies when / blocked by.
catasraḥ paśya (Subodhinī; compare SK 3682).go + ām → gávām; div + bhis → dyúbhiḥ, against divéThe previous rules, from SK 3714 on, gave the udātta to a case ending after many stems. These three rules take it back after certain stems. Then the udātta stays on the stem. The running examples are gávām "of cows", udātta on gá, where SK 3714 would make the ending udātta. And dyúbhiḥ "by days", udātta on dyú, against divé "on the day", where the vowel-initial ending keeps its udātta.
Rule. "What was stated before" does not apply after seven kinds of stem. They are go "cow", śvan "dog", a word ending in a or ā in the nominative singular (sau), rāṭ, aṅ, kruṅ and kṛt. The Subodhinī takes "what was stated before" as the ṣāṣṭhika accent, the accent "taught in the sixth book". This is read here as the udātta rules for the ending, for nām and for matup, from SK 3714 on. So this rule blocks even SK 3722: in vīravattamam the matup after antodātta vīra stays anudātta.
Example.
go + ām → gávām, udātta on gá: gávāṃ śatā́ "hundreds of cows".
go + bhyas → góbhyaḥ: góbhyo gātúm "a way for the cows". SK 3714 would otherwise make the ending udātta.
After śvan: śúnaś cic chépam "Śunaḥśepa". The word cit is inserted inside the name in the recited text. As a compound of the vanaspatyādi list, both members keep their own accent: śú and ché.
After a nominative in a: tad has sa and kim has ka in the nominative singular. So both count. tébhyo dyumnám "splendour for them". téṣāṃ pāhi śrudhī́ hávam "drink of these, hear the call". In both, the udātta stays on te.
Applies when / blocked by. rāṭ, aṅ, kruṅ and kṛt have no example in the kaumudī. The Subodhinī supplies them:
rājā.prāñcā. In the sense of motion the n is lost, and then the ending is udātta: prācā, prāce.kruñcā.kṛtā, kṛte.Bhaṭṭoji does not mark these forms.
Watch out. sau here is the nominative singular, not the locative plural su of SK 3714. The Subodhinī shows why. Before the locative plural, tad and kim end in e (teṣu, keṣu), not in a. Then the rule could not reach tebhyaḥ at all.
Rule. After div "sky, day", a case ending that begins with a jhal is not udātta. A jhal is any consonant other than a nasal or a semivowel. "not" is carried forward from SK 3726, and "begins with" (ādiḥ) from SK 3725 (Subodhinī).
Example.
div + bhis → dyubhiḥ → the ending begins with bh, a jhal, so it is not udātta → dyúbhiḥ, udātta on dyú.
Bhaṭṭoji's line: dyúbhir aktúbhiḥ "by days and by nights".
The ending would otherwise be udātta by SK 3714 or SK 3717 (Subodhinī).
Applies when / blocked by. A vowel-initial ending is not covered. In úpa tvāgne divé-dive "to you, Agni, day after day", the ending -e is udātta by SK 3717: divé.
Rule. After nṛ "man", the same prohibition holds optionally. Bhaṭṭoji states the effect: a jhal-initial ending after nṛ is optionally udātta.
Example.
nṛ + bhis → nṛ́bhiḥ: the prohibition applies, and the udātta stays on nṛ́.
Bhaṭṭoji's line: nṛ́bhir yemānáḥ "being guided by men". The ending bhi is only a dependent svarita, and the udātta of yemānáḥ is on náḥ.
In the other option the ending is udātta. Bhaṭṭoji gives no example of it.
kim + at → kvà; vṛdh + śap + śānac → várdhamānamThese five rules close the accents of affixes (pratyayasvara). An affix with t as its marker is svarita. A verb ending that replaces a la is anudātta after certain stems. Then come three rules for the aorist in sic, the perfect ending thal, and r-marked affixes. The running examples are kvà "where?", an independent svarita on its only syllable. And várdhamānam "growing", where the ending stays anudātta and the udātta is on vár.
Rule. An affix that has t as its marker (it) is svarita. Bhaṭṭoji says the rule is "explained by its wording alone".
Example.
kim + at (SK 1959) → kva replaces kim (SK 1960) → kva + a
→ the at is svarita by this rule → kvà.
Bhaṭṭoji's line: kvà nūnám "where now?". The svarita on kvà stands at the head of the phrase, so it is an independent svarita.
That the svarita of at survives on the single a left by the merger is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. A lasārvadhātuka is anudātta after four things. A lasārvadhātuka is a sārvadhātuka ending that replaces a la (a tense or mood sign). The four are:
It does not apply after the roots hnu and iṅ. Bhaṭṭoji calls the rule an exception (apavāda) to the affix accent. The 1915 print spells the last word anhviṅoḥ; the example is hnute.
Example.
kartā, kartārau, kartāraḥ "he, they two, they will do". Bhaṭṭoji marks no accents here.ya āste "who sits".abhícaṣṭe ánṛtebhiḥ. The ending te stays anudātta. The udāttas are on bhí and on á.púrubhujā canasyátam. The class sign śap counts as an a in upadeśa (Subodhinī). So the ending tam is anudātta. The udāttas are on pú and syá.vṛdh + śap + śānac (with the augment m, SK 3101) → vardhamāna
→ śānac is c-marked, but this rule makes it anudātta → várdhamānam, udātta on vár.
The line is várdhamānaṃ své dáme. The Subodhinī adds that the augment m does not stand in the way.
Applies when / blocked by.
abhí vṛdhé gṛṇītáḥ. Here the ending tas keeps its udātta, táḥ.ható vṛtrāṇy āryā. After han loses its nasal, ha ends in a. But han is not taught with a final a. So tas keeps its udātta: ható, udātta on tó.katīha pacamānāḥ "how many are cooking here?" The affix is cānaś, not a la (Subodhinī).śiśye. The perfect ending is ārdhadhātuka by SK 2172.hnute; yad adhīte. The Subodhinī says yat is added to stop the verb's own loss of accent (SK 3937). Then the ending keeps its udātta.indhé rā́jā the ending te keeps the affix accent, ndhé. Bhaṭṭoji notes that this vārttika is in the Bhāṣya on SK 3651.Watch out. The root must be ṅ-marked in its upadeśa. In śṛṇute jaritur havam the ending is ṅ-marked only by extension (atideśa). So this rule does not apply there (Subodhinī).
Rule. In a form ending in sic, the aorist sign s, the first syllable is optionally udātta.
Example.
yā + luṅ, with tam for thas → sic for the aorist sign, with iṭ and sak (SK 2377), no augment a (SK 3546) → yāsiṣṭam.
Bhaṭṭoji's line: yāsiṣṭáṃ vartír aśvinā "you two have come to the dwelling, O Aśvins". The line marks the udātta on ṣṭáṃ. This is the other option, where the rule has not applied.
By the rule itself the udātta would stand on the first syllable, yā. The Subodhinī says an example of that option must be sought elsewhere.
Rule. Take a word ending in the perfect ending thal (-tha) with the augment iṭ (-itha). In it, optionally, the iṭ is udātta, or the final, or the first syllable. "Optionally" is carried forward from SK 3677. The word vā brings in "the first", which the sūtra does not name (Subodhinī). When none of the three is udātta, SK 3676 (liti) makes the vowel before the affix udātta.
Example.
lū + thal with iṭ → luluvitha. Bhaṭṭoji says that all four syllables are udātta in turn:
→ the first: lúluvitha
→ the vowel before the affix, by SK 3676: lulúvitha
→ the iṭ: luluvítha
→ the final: luluvithá.
The four forms are not spelled out by Bhaṭṭoji or his commentators; they are inferred from his statement.
Watch out. The word seṭi "with iṭ" limits the rule to thal that has iṭ (Subodhinī). Where thal has no iṭ, this rule does not apply.
Rule. In a word ending in an affix that has r as its marker, the penultimate syllable is udātta. riti is a locative word, and in accent rules a locative names "a word ending in" the item. The Subodhinī draws this from the word pūrvam in SK 3675. With this rule the accents of affixes end (iti pratyayasvarāḥ, "here end the accents of affixes").
Example.
ā + hu + anīyar → āhavanīya
→ the syllables are ā-ha-va-nī-ya; the penultimate is nī → āhavanī́ya.
Bhaṭṭoji's example: yad āhavanī́ye, udātta on nī.
The Subodhinī takes anīyar here in the sense of place. In the compound with ā, the kṛt-final member keeps its own accent.
yajña + śrī → yajñaśríyam; citrá + śravas → citráśravas-Here the accents of compounds (samāsa) begin, under the heading samāsasvarāḥ, "the accents of compounds". A compound has a prior member (pūrvapada) and a final member (uttarapada). SK 3734 gives the default: the last syllable of the compound is udātta. SK 3735 opens a long heading. Under it the prior member keeps its own accent (prakṛtyā, "in its original form"). Then the prior member's udātta survives, and the rest of the compound is anudātta. The running examples are yajñaśríyam "glory of the sacrifice", with the compound's udātta on śri. And citráśravas- "whose fame is bright", where citra keeps its udātta on trá.
Rule. Of a compound, the final syllable is udātta. The sūtra says only "of a compound"; "the final is udātta" is Bhaṭṭoji's gloss.
Example.
yajñasya śrīḥ → yajña + śrī (a tatpuruṣa with a sixth-case prior member) → yajñaśrī
→ the final ī is udātta by this rule → yajñaśrī́
→ + am → iy replaces ī before the vowel → the i of iy takes the udātta, by nearest likeness (Subodhinī) → yajñaśríyam, udātta on śri.
Applies when / blocked by. This is the default for every compound. Every rule from SK 3735 on sets it aside in its own field.
Rule. In a bahuvrīhi, a prior member that contains an udātta or a svarita keeps its original form. In the accent section, "original form" means its own accent (Subodhinī). A bahuvrīhi is a compound that describes something outside itself: "whose fame is bright". The words prakṛtyā pūrvapadam are a heading (adhikāra). They run through every rule of this part and on into Part V.
Example.
citra "bright" is antodātta by phiṭ 1: citrá
→ citra + śravas "fame" → bahuvrīhi "whose fame is bright" → citraśravas-
→ citra keeps its accent → citráśravas-, udātta on trá.
Bhaṭṭoji's line: satyáś citráśravastamaḥ "true, most widely famed". The Subodhinī adds the effect on what follows. The syllable after trá is svarita by SK 3660. The syllables after that are pracaya, unmarked monotone, by SK 3668.
Applies when / blocked by. Why "containing an udātta or a svarita"? A wholly anudātta prior member leaves the compound with only the accent of SK 3734. sama is wholly anudātta by phiṭ 78. So sama + pāda → samapādáḥ, udātta on the final dáḥ.
kiri + kāṇa → kiríkāṇaḥ; gamana + cira → gámanaciramA tatpuruṣa is a compound whose final member is the main word, and the prior member qualifies it, often in a case relation. SK 3736 lists seven kinds of prior member that keep their accent in a tatpuruṣa. SK 3737–3745 then name particular final members, each with a condition of sense. In every form below only the prior member's udātta is marked; the rest of the compound is anudātta. A counter-example falls back on SK 3734, and it is left unmarked unless the text marks it. The running examples are kiríkāṇaḥ "one-eyed through a kiri", with the udātta of kiri on rí. And gámanaciram "a going long delayed", with the udātta of gamana on gá.
Rule. In a tatpuruṣa, seven kinds of prior member keep their own accent:
"In a tatpuruṣa" (tatpuruṣe) runs on through SK 3757.
Example.
túlyaśvetaḥ. tulya takes yat and is ādyudātta by SK 3701, so the udātta is on tú. The compound comes by SK 749.kiriṇā kāṇaḥ → kiríkāṇaḥ "one-eyed through a kiri". The Subodhinī: kiri is antodātta by its k-marked uṇādi affix, so the udātta is on rí.patayán mandayátsakham. Bhaṭṭoji explains mandayati as "in the exhilarating one", that is, in Indra. The Subodhinī makes mandayat antodātta, so the udātta is on yát.śastrī́śyāmā "dark as a knife". śastrī ends in the ṅīṣ of the gaurādi list and is antodātta (Subodhinī), so the udātta is on strī́.muhūrtásukham "happiness lasting a moment". muhūrta is antodātta (Subodhinī), so the udātta is on tá.bhojyoṣṇam. bhojya ends in ṇyat and is svarita-final by SK 3729. The accent after its merger with uṣṇa is not stated, so the form is left unmarked.Applies when / blocked by. A vārttika limits the indeclinables: "among indeclinables, only nañ, ku and the nipātas". nañ is the negative a-/an-, and a nipāta is a particle. Example: ayajño vā eṣaḥ "this is indeed no sacrifice". Bhaṭṭoji marks no accent here.
Why the limit? snātvākālakaḥ has the indeclinable snātvā as prior member, but it is not in the list. So the compound takes the final udātta of SK 3734 (Subodhinī).
Watch out. The Subodhinī explains why nañ is named, though particles are covered anyway. Its accent must win even over a later rule, such as the kṛt accent in akaraṇiḥ. ku is named because it is not in the list of particles.
Rule. In a tatpuruṣa, a prior member that names a colour keeps its accent before a final member that names a colour. The final member must not be eta "variegated".
Example.
lohita + kalmāṣa → lóhitakalmāṣaḥ "red-speckled".
lohita ends in the affix itan. The Subodhinī: itan is n-marked, so lohita is ādyudātta. The udātta is on ló.
kṛṣṇasāraṅgaḥ: kṛṣṇa ends in the affix nak. The Subodhinī gives kṛṣṇa two accents. It is antodātta in the Veda (phiṭ 11) and ādyudātta in ordinary speech (phiṭ 33). So this form is left unmarked.
Applies when / blocked by.
paramakṛṣṇaḥ.kṛṣṇatilāḥ "black sesame".kṛṣṇaitaḥ.Rule. Before gādha "firm bottom, depth" or lavaṇa "salt", in a tatpuruṣa that denotes a measure, the prior member keeps its accent. "Measure" (pramāṇa) means any fixing of extent, not only length.
Example.
aritra + gādha → arítragādham udakam "water as deep as an oar reaches". aritra ends in itra and is madhyodātta, udātta in the middle, on rí. Bhaṭṭoji states this.
go + lavaṇa → gólavaṇam "as much salt as is given to a cow". go ends in the uṇādi affix ḍo and is antodātta (Subodhinī). Since go has one syllable, the udātta is on gó.
Applies when / blocked by. Without the sense of measure the rule does not apply: paramagādham.
Rule. Before dāyāda "heir", a prior member that names what is inherited keeps its accent, in a tatpuruṣa.
Example.
dhana + dāyāda → dhánadāyādaḥ "heir to wealth".
dhana ends in the uṇādi affix kyu and is ādyudātta by the affix accent. So the udātta is on dhá.
Applies when / blocked by. A prior member that does not name the inheritance does not keep its accent. paramadāyādáḥ "the chief heir" takes only the compound's final udātta, on dáḥ (Subodhinī).
Rule. Before cira "long-lasting" or kṛcchra "difficult", a prior member that names an obstruction keeps its accent, in a tatpuruṣa.
Example.
gamana + cira → gámanaciram "a going long delayed".
vyāharaṇa + kṛcchra → vyāháraṇakṛcchram "a speaking that is difficult".
Both prior members end in lyuṭ. By SK 3676 (liti), the syllable before an l-marked affix is udātta. So the udātta is on gá and on há.
Bhaṭṭoji explains the sense. A going that comes late for want of means, and is attended with hardship, becomes an obstruction.
Applies when / blocked by. Without the sense of obstruction: mūtrakṛcchram "strangury".
Rule. Before pada in the sense "pretext", the prior member keeps its accent, in a tatpuruṣa.
Example.
mūtra + pada → mū́trapadena prasthitaḥ "he set off on the pretext of urinating".
mūtra ends in ghañ. A ghañ-final word is ādyudātta by SK 3686, so the udātta is on mū́. That last step is not stated by Bhaṭṭoji or his commentators; it is inferred.
uccāra + pada → uccārápadena "on the pretext of evacuating". uccāra also ends in ghañ. Bhaṭṭoji says it is antodātta by SK 3878. So the udātta is on rá.
Applies when / blocked by. Without the sense "pretext": viṣṇupadam "Viṣṇu's step".
Rule. Before nivāta in the sense "shelter from the wind", the prior member keeps its accent, in a tatpuruṣa.
Example.
kuṭī + nivāta → kuṭī́nivātam "a hut that shelters from the wind". kuṭī ends in the ṅīṣ of the gaurādi list. So the udātta is on ṭī́.
kuḍya + nivāta → kuḍyánivātam "a wall that shelters from the wind". kuḍya ends in ḍyak; others say yak. So the udātta is on dyá.
Bhaṭṭoji names only the affixes. The udātta syllables they give are not stated by Bhaṭṭoji or his commentators; they are inferred from the affix accent.
Applies when / blocked by. In rājanivāte vasati "he lives at the king's side", nivāta means "side". Bhaṭṭoji says the word is fixed by usage in that sense. So the rule does not apply.
Rule. Before śārada in a sense other than "seasonal", the prior member keeps its accent, in a tatpuruṣa. ārtava means "what is in a season". Here śārada means "new".
Example.
rajju + śārada → rájjuśāradam udakam "water fresh from the rope".
Bhaṭṭoji analyses it with other words: rajjoḥ sadya uddhṛtam "just drawn up by the rope".
rajju is formed by uṇādi 15 as ādyudātta. So the udātta is on rá.
Applies when / blocked by. In the seasonal sense: uttamaśāradam "the finest autumnal [thing]".
Rule. Before adhvaryu "officiating priest" or kaṣāya, in a tatpuruṣa that names a class (jāti), the prior member keeps its accent.
Example.
kaṭha + adhvaryu → kaṭhā́dhvaryuḥ "an adhvaryu of the Kaṭha school".
kaṭha ends in the ac of the pacādi list, so it is antodātta: kaṭhá. Its ṇini (SK 1484) is lost by SK 1487.
Then the udātta a and the anudātta a of adhvaryu merge, and the merged ā is udātta by SK 3658: ṭhā́.
These two accent steps are not stated by Bhaṭṭoji or his commentators; they are inferred.
dauvārika + kaṣāya → dauvārikákaṣāyam "the kaṣāya of door-keepers". dauvārika "appointed at the door" ends in ṭhak and is antodātta, Bhaṭṭoji says. So the udātta is on ká.
Applies when / blocked by. Without the sense of a class: paramādhvaryuḥ "the chief adhvaryu".
Rule. Before sadṛśa or pratirūpa "like", in a tatpuruṣa of likeness, the prior member keeps its accent.
Example.
pitṛ + sadṛśa → pitṛ́sadṛśaḥ "like the father".
The Subodhinī: pitṛ ends in tṛc, which is c-marked, so it is antodātta by SK 3710. The udātta is on tṛ́.
Bhaṭṭoji gives no example with pratirūpa.
Applies when / blocked by. In paramasadṛśaḥ the sense is honouring, not likeness. So the rule does not apply.
gṛha + pati → gṛhápatiḥ; bhuvana + pati → bhúvanapatiḥThe heading "in a tatpuruṣa" is still in force. Each rule names more final members, again with a condition of sense. In the middle stand three rules on pati "lord". SK 3752 keeps the prior member's accent before pati. SK 3753 denies it for four words, and SK 3754 makes it optional for bhuvana. The running examples are gṛhápatiḥ "lord of the house", with the udātta of gṛha on há. And bhúvanapatiḥ "lord of the world", with the udātta of bhuvana on bhú, in the option where it is kept.
Rule. When the final member is a dvigu that denotes a measure, the prior member keeps its accent, in a tatpuruṣa. A dvigu is a compound with a numeral as its prior member.
Example.
saptasama "whose measure is seven years" is a dvigu: sapta samāḥ pramāṇam asya.
A vārttika, pramāṇe lo dvigor nityam, makes its measure affix mātrac vanish (luk).
prācyaś cāsau saptasamaś ca → prācya + saptasama → prā́cyasaptasamaḥ "an Easterner seven years old".
prācya is ādyudātta, Bhaṭṭoji says. The Subodhinī gives the reason: it ends in yat (SK 3701). So the udātta is on prā́.
Applies when / blocked by.
vrīhiprasthaḥ "a prastha of rice".paramasaptasamam. There saptasama is a group of seven years, a samāhāra-dvigu, not a measure (Subodhinī).Rule. Before vāṇija "trader", a prior member naming the place to be gone to, or the merchandise, keeps its accent, in a tatpuruṣa.
Example.
madra + vāṇija → madrávāṇijaḥ "a trader to the Madras". madra ends in the affix rak. The Subodhinī: it is antodātta by the affix accent. So the udātta is on drá.
go + vāṇija → góvāṇijaḥ "a cattle-trader". go is antodātta (Subodhinī), so the udātta is on gó.
Bhaṭṭoji calls the compound a saptamī-samāsa, a compound with a seventh-case prior member.
Applies when / blocked by. Without these senses: paramavāṇijaḥ "the chief trader".
Rule. Before mātra, upajña, upakrama or chāyā, in a neuter tatpuruṣa, the prior member keeps its accent.
Example.
bhikṣāyās tulyapramāṇam "of a measure equal to alms" → bhikṣā́mātram "as much as alms". bhikṣā ends in the affix a of SK 3280, so its udātta is on kṣā́.pā́ṇinyupajñam "first devised by Pāṇini". pāṇini is ādyudātta, so the udātta is on pā́.nandópakramam "first undertaken by Nanda". nanda ends in the ac of the pacādi list, so it is antodātta, nandá. Its udātta a and the anudātta u of upakrama merge into an udātta o by SK 3658: dó.íṣucchāyam "the shade of [many] arrows". iṣu is ādyudātta because its affix is n-marked. The udātta is on í.The Subodhinī makes bhikṣā antodātta. That its udātta survives on kṣā́ after the feminine ā, and the whole accent step for nanda, are not stated by Bhaṭṭoji or his commentators; they are inferred.
Applies when / blocked by. The compound must be neuter: not kuḍyacchāyā "the shade of a wall", which is feminine.
Rule. Before sukha "pleasant" or priya "agreeable", in a tatpuruṣa meaning "good for", the prior member keeps its accent.
Example.
gamana + priya → gámanapriyam "going that is agreeable".
gamana + sukha → gámanasukham "going that is pleasant".
gamana ends in lyuṭ. By SK 3676 the syllable before the affix is udātta, so the udātta is on gá.
Applies when / blocked by. Without the sense "good for": paramasukham "the highest happiness".
Rule. The same holds when affection (prīti) is meant. Before sukha or priya, the prior member keeps its accent, in a tatpuruṣa.
Example.
brāhmaṇásukhaṃ pāyasam "milk-rice, a delight to Brāhmaṇas".
chātrápriyo 'nadhyāyaḥ "a day without recitation, dear to students".
Bhaṭṭoji says brāhmaṇa and chātra are antodātta by the affix accent. So the udātta is on ṇá and on trá.
Applies when / blocked by. Without affection: rājasukham "the king's ease".
Watch out. Why a separate rule, when SK 3749 has the same final members? The Subodhinī: the split stops a one-to-one pairing (yathāsaṃkhya). Otherwise "good for" might go only with sukha, and "affection" only with priya.
Rule. Before svāmin "owner", a prior member naming the property keeps its accent, in a tatpuruṣa.
Example.
go + svāmin → gósvāmī "owner of cattle". go is antodātta, so its one syllable gó is udātta.
Applies when / blocked by. Without the sense of property: paramasvāmī "supreme master".
Rule. Before pati in the sense of lordship, the prior member keeps its accent, in a tatpuruṣa.
Example.
gṛha ends in ka (SK 2906) and is antodātta by the affix accent (Subodhinī): gṛhá
→ gṛha + pati → gṛha keeps its accent → gṛhápatiḥ, udātta on há.
Bhaṭṭoji's line: dámūnā gṛhápatir dáme "the household one, lord of the house, in the house".
Rule. Before pati meaning lordship, four prior members do not keep their accent. They are bhū "earth", vāc "speech", cit "thought" and didhiṣū.
Example.
bhuvaḥ patiḥ "lord of the earth" → bhūpatiḥ. Likewise vākpatiḥ, citpatiḥ, didhiṣūpatiḥ.
The Subodhinī says all four prior members are antodātta. That is the accent this rule withholds.
Bhaṭṭoji does not state what accent the compounds take instead. So the forms are left unmarked.
Rule. In the same setting, before pati meaning lordship, bhuvana "world" keeps its accent only optionally.
Example.
bhuvana ends in the uṇādi affix kyun (uṇādi 238). The Subodhinī: being n-marked, it is ādyudātta by SK 3686: bhúvana
→ bhuvana + pati → where the rule applies → bhúvanapatiḥ "lord of the world", udātta on bhú.
The accent of the other option is not stated.
Rule. Before āśaṅka "suspected", ābādha "obstructed" or nedīyas "nearer", when supposition is meant, the prior member keeps its accent, in a tatpuruṣa. Bhaṭṭoji defines supposition (saṃbhāvana) as judging that something exists.
Example.
gámanāśaṅkam asti "there is a suspected going".
gámanābādham "a going that is obstructed".
gámananedīyaḥ "a going that is nearer".
That is, one supposes that the going is suspected, obstructed or near. gamana ends in lyuṭ and takes the udātta before the affix, on gá (Subodhinī, SK 3676).
Applies when / blocked by. Without supposition: paramanedīyaḥ "the very nearest".
Rule. Before pūrva in the sense "formerly", the prior member keeps its accent, in a tatpuruṣa. Inside the compound, pūrva stands for "having been before".
Example.
āḍhyo bhūtapūrvaḥ "one who was formerly rich" → āḍhyápūrvaḥ.
The Subodhinī: āḍhya is formed with ka and is antodātta by SK 3878. So the udātta is on ḍhyá.
Applies when / blocked by. Without the sense "formerly": paramapūrvaḥ.
Rule. Before savidha, sanīḍa, samaryāda, saveśa and sadeśa in the sense of nearness, the prior member keeps its accent, in a tatpuruṣa. All five mean "the vicinity of" (Subodhinī).
Example.
madrásavidham, madrásaveśam, madrásadeśam "the vicinity of the Madras". madra is antodātta, so the udātta is on drá.kāśmī́rasamaryādam "the vicinity of Kāśmīra". The Subodhinī: kāśmīra is madhyodātta by phiṭ 42. The udātta is on mī́.gāndhārasanīḍam is left unmarked. The Subodhinī read gāndhāri here, not gāndhāra.Applies when / blocked by. Without the sense of nearness:
saha maryādayā "together with a boundary" → samaryādaṃ kṣetram "a field with its boundary".caitrasamaryādam "Caitra's bounded field".gamana + śreṣṭha → gámanaśreṣṭham; kumāra + śramaṇā → kumāráśramaṇā, then kumāra + pratyenas → kúmārapratyenāḥSK 3758 keeps the accent of vispaṣṭa and similar words before a word for a quality. The next four rules are for the karmadhāraya, a tatpuruṣa whose two members refer to the same thing. SK 3759 keeps the accent of an action noun before "best", "worse" and the like. SK 3760–3762 deal with kumāra "young, maiden". Normally kumāra keeps its final udātta. Before pratyenas, and optionally before a guild-name, the udātta moves to its first syllable. The running examples are gámanaśreṣṭham "going, the best", udātta on gá. Then kumāráśramaṇā "a maiden ascetic", udātta on rá, against kúmārapratyenāḥ, udātta on kú.
Rule. vispaṣṭa "distinct" and the words of its list keep their accent as prior members before words that denote a quality.
Example.
vispaṣṭa + kaṭuka → víspaṣṭakaṭukam "distinctly pungent".
Bhaṭṭoji: vispaṣṭa is ādyudātta by SK 3783, the rule for a gati (a prefix such as vi) before a participle. So the udātta is on ví.
The list is: vispaṣṭa, vicitra, vyakta, saṃpanna, paṇḍita, kuśala, capala, nipuṇa.
Applies when / blocked by.
paramalavaṇam.vispaṣṭabrāhmaṇaḥ.Watch out. This compound is not a karmadhāraya, says the Subodhinī. vispaṣṭa qualifies the quality itself, while kaṭuka names a thing that has it. So the two do not refer to the same thing, and the compound is made by the general rule for joining two case-forms. The Subodhinī's list also has paṭu; the kaumudī's list does not.
Rule. In a karmadhāraya, a prior member that names the action itself (bhāva) keeps its accent before certain final members. They are a word containing the substitutes śra, jya or kan, the word avama "lowest", and a word containing pāpa "bad". The three substitutes come before the comparative and superlative affixes (SK 2009, 2011, 2019).
Example.
gamana "going" ends in lyuṭ in the action sense. By SK 3676 the udātta is on gá (Subodhinī).
→ gámanaśreṣṭham "going, the best"
→ gámanajyāyaḥ "going, the better"
→ gámanāvamam "going, the lowest"
→ gámanakaniṣṭham "going, the least"
→ gámanapāpiṣṭham "going, the worst".
Applies when / blocked by.
gamanaśobhanam.gamanaśreyaḥ, gamana means "that by which one goes", a means of going (SK 3293). So the rule does not apply.Rule. kumāra also keeps its own accent as prior member in a karmadhāraya. "In a karmadhāraya" is carried forward from SK 3759.
Example.
kumāra is antodātta, Bhaṭṭoji says. The Subodhinī gives the reason: it ends in the c-marked ac of the pacādi list. So its last syllable is udātta: kumārá
→ kumārī śramaṇā "a maiden who is an ascetic" → in the compound kumārī takes its masculine form kumāra (puṃvadbhāva, Subodhinī) → kumāra + śramaṇā (SK 752) → kumāráśramaṇā, udātta on rá.
Watch out. The Subodhinī reports two views. In its own view only the compound taught by SK 752 (kumāraḥ śramaṇādibhiḥ) is meant. Others take in every karmadhāraya with kumāra. Scholars discuss this; Bhaṭṭoji says only "in a karmadhāraya".
Rule. Before pratyenas, in a karmadhāraya, the first syllable of kumāra is udātta. kumāra is carried forward from SK 3760. The Subodhinī: the accent that SK 3760 would keep on kumāra falls on its first syllable instead.
Example.
pratigatam eno 'sya "he to whom sin has returned" → pratyenas
→ kumāra + pratyenas → the first syllable of kumāra is udātta → kúmārapratyenāḥ, udātta on kú; the rest is anudātta.
Rule. Before a word for a guild (pūga), the previous rule holds optionally. Then the first syllable of kumāra is udātta in a karmadhāraya. Bhaṭṭoji glosses pūga as gaṇa, a group.
Example.
With the rule: kúmāracātakāḥ "young Cātakas", kúmārajīmūtāḥ "young Jīmūtas", udātta on kú.
Without it, Bhaṭṭoji says, SK 3760 alone applies. Then kumāra keeps its final udātta: kumārácātakāḥ, kumārájīmūtāḥ, udātta on rá.
The Subodhinī: cātaka and the rest are names of guilds, whose affix ñya (SK 2066) is lost in the plural (SK 1193).
Watch out. On the view that SK 3760 covers only the compound of SK 752, the Subodhinī notes a different result. In the other option the compound is then antodātta. Scholars discuss this; the forms above follow Bhaṭṭoji's sentence.
pañcan + kapāla → páñcakapālaḥ; bahu + aratni → bahvàratniḥThree rules keep the prior member's accent in a dvigu, a compound with a numeral as its prior member. SK 3763 names the final members: an ik-final word, a word for time, and three vessel-words. SK 3764 makes it optional for bahu "many". SK 3765 adds two measures, also optionally. The running examples are páñcakapālaḥ "prepared in five dishes", where pañcan keeps its udātta on páñ. And bahvàratniḥ "measuring many cubits", where the u of bahú becomes v and the following a becomes an independent svarita.
Rule. In a dvigu, the prior member keeps its accent before certain final members:
Bhaṭṭoji: pañcan is ādyudātta by phiṭ 28. So the udātta is on páñ, and in dáśamāsyaḥ on dá.
Example.
pañca aratnayaḥ pramāṇam asya "whose measure is five cubits" → páñcāratniḥ. The measure affix is lost by the vārttika quoted at SK 3746 (Subodhinī).daśa māsān bhūtaḥ "who has lived ten months" → dáśamāsyaḥ; pañca māsān bhūtaḥ → páñcamāsyaḥ. The affix yap comes by SK 1746, dvigor yap, inside the heading of SK 1744. For daśan the initial udātta, dá, is not stated by Bhaṭṭoji or his commentators; it is inferred from phiṭ 28.páñcakapālaḥ "prepared in five dishes", páñcabhagālaḥ, páñcaśarāvaḥ.Applies when / blocked by.
pañcāśvaḥ "bought for five horses".paramāratniḥ.Rule. bahu "many" keeps its accent in such a dvigu only optionally. bahu is antodātta: bahú, udātta on hú.
Example.
bahu + aratni → u becomes v before the vowel → bahvaratni
→ the udātta u has become a yaṇ, so the following anudātta a becomes svarita by SK 3657, as Bhaṭṭoji says → bahvàratniḥ, an independent svarita on và.
bahúmāsyaḥ, bahúkapālaḥ: the udātta stays on hú.
In the other option the compound has its final udātta (Subodhinī). Bhaṭṭoji does not give those forms.
Rule. Before diṣṭi or vitasti "span", two measures, the prior member of a dvigu optionally keeps its accent.
Example.
páñcadiṣṭiḥ "measuring five diṣṭis", páñcavitastiḥ "measuring five spans". The udātta is on páñ.
In the other option the compound is antodātta (Subodhinī).
bhrāṣṭra + pakva → bhrā́ṣṭrapakvaḥ; śini + vāsudeva → śínivāsudevāḥThe last rules of this part cover three kinds of compound. SK 3766 restores the accent of a seventh-case prior member before three participles in kta and the ghañ-noun bandha. SK 3767 keeps the accent of four upasargas in the avyayībhāva, an indeclinable compound. SK 3768–3770 are for the dvandva, a compound of "X and Y". The running examples are bhrā́ṣṭrapakvaḥ "cooked in a frying pan", udātta on bhrā́. And śínivāsudevāḥ "the Śinis and the Vāsudevas", udātta on śí.
Rule. A seventh-case prior member that does not denote time keeps its accent before siddha "accomplished", śuṣka "dried", pakva "cooked" or bandha "bound".
Example.
sāṅkāśyásiddhaḥ "accomplished in Sāṅkāśya". Bhaṭṭoji names the affix ṇya. That this puts the udātta on śyá is not stated by Bhaṭṭoji or his commentators; it is inferred from the affix accent.ātapaśuṣkaḥ "dried in the sun". Its accent is not stated, so it is left unmarked.bhrā́ṣṭrapakvaḥ "cooked in a frying pan". bhrāṣṭra ends in ṣṭran, which is n-marked, so it is ādyudātta (Subodhinī). The udātta is on bhrā́.cakrábandhaḥ "bound on a wheel". cakra is antodātta, Bhaṭṭoji says. The udātta is on krá.Applies when / blocked by. The prior member must not denote time: not pūrvāhṇasiddhaḥ "accomplished in the forenoon".
Watch out. Why is this rule needed, when SK 3736 already keeps a seventh-case prior member's accent? Bhaṭṭoji: the kṛt accent had blocked that accent, and this rule restores it. The Subodhinī names the blocking rule: SK 3878, which makes a final member in kta or ghañ antodātta.
Rule. pari, prati, upa and apa keep their accent as prior members before a final member that names something excluded, or a part of the day or the night. Bhaṭṭoji: the upasargas are ādyudātta.
Example.
páritrigartaṃ vṛṣṭo devaḥ "the god has rained except over Trigarta". Here pari means "except". The udātta is on pá.ápatrigartam "Trigarta excepted": udātta on á.prátipūrvāhṇam "towards the forenoon": udātta on prá.úpapūrvāhṇam "near the forenoon", úpapūrvarātram "near the first part of the night": udātta on ú.Applies when / blocked by.
agniṃ prati "towards the fire" → pratyagni.Rule. Take a dvandva of plural words for rājanyas (kṣatriyas of anointed lineage) that refers to the Andhakas and the Vṛṣṇis. In it the prior member keeps its accent.
Example.
śvāphalkácaitrakāḥ "the descendants of Śvaphalka and of Citraka". The Subodhinī forms śvāphalka with aṇ in the sense of a descendant. That this puts the udātta on lká is not stated by Bhaṭṭoji or his commentators; it is inferred.śínivāsudevāḥ "the Śinis and the Vāsudevas". śini is ādyudātta, Bhaṭṭoji says, and here it stands for Śini's descendants. The udātta is on śí.Applies when / blocked by.
dvaipyabhaimāyanāḥ. dvaipya means "those who live on an island", and bhaimāyana "a young descendant of Bhaimi". These are Andhakas and Vṛṣṇis, but not of anointed lineage.saṃkarṣaṇavāsudevau, a dual.vṛṣṇīnāṃ kumārāḥ → vṛṣṇikumārāḥ "the princes of the Vṛṣṇis".kurupañcālāḥ.Rule. In a dvandva, a prior member that is a numeral keeps its accent. "In a dvandva" is carried forward from SK 3768.
Example.
tráyodaśa "thirteen". trayas replaces tri by SK 809, and Bhaṭṭoji says it is stated as ādyudātta. The udātta is on trá.dvā́daśa "twelve". dvi has one vowel, which is udātta by phiṭ 1. The ā that replaces its i by SK 808 keeps that udātta: dvā́. This step is not stated by Bhaṭṭoji or his commentators; it is inferred.Watch out. The Subodhinī read antodāttaḥ "antodātta" in place of ādyudāttaḥ here. It calls that reading the Kāśikā's view and itself concludes that trayas is ādyudātta, from the Vedic usage tráyodaśa ca me.
Rule. In a dvandva of words for pupils (antevāsin) named after a teacher, the prior member keeps its accent. In such a word the teacher's name is the upasarjana, the subordinate part.
Example.
pāṇinīya + rauḍhīya → pāṇinī́yarauḍhīyāḥ "the pupils of Pāṇini and those of Rauḍhi".
Bhaṭṭoji: both are madhyodātta by the accent of the affix cha (-īya). So pāṇinīya has its udātta on nī́, and that udātta is kept.
Applies when / blocked by.
pāṇinīyadevadattau "a Pāṇinīya and Devadatta".chāndasavaiyākaraṇāḥ "students of the Veda and grammarians".āpiśalapāṇinīye śāstre "the treatises of Āpiśali and of Pāṇini".mahat (antodātta) + vrīhi → mahā́vrīhiḥ (udātta on hā, the rest anudātta)This part goes on with the compound accents of the sixth book. By default a compound has its udātta on its last syllable (samāsasya, SK 3734). The heading of SK 3735, prakṛtyā pūrvapadam, overrides that default: the prior member (pūrvapada) keeps its own accent (prakṛtyā, "by its original form"). Then the prior member keeps its own udātta, and every other syllable of the compound is anudātta. The heading runs through SK 3797.
The six sūtras of this group name particular words. They are the dvandvas of a list, mahat "great", kṣullaka, uṣṭra "camel", go "cow", and seven fixed compounds with an open list.
The running example is mahat before vrīhi "rice". mahat is antodātta: its udātta is on its last syllable.
mahat + vrīhi → mahā for mahat (SK 807) → mahāvrīhi
→ mahat keeps its own accent, on its last syllable → mahā́vrīhiḥ.
The udātta is on hā. The marks follow this chapter's convention: acute for the udātta, grave for an independent svarita, no mark for an anudātta.
Rule. In a dvandva of the kārtakaujapa list, the prior member keeps its own accent. A dvandva is a compound of co-ordinate members, "A and B". The word dvandve "in a dvandva" is carried forward from SK 3768.
Example.
kārta + kaujapa (dvandva) → kārta keeps its own accent → kārtákaujapau (udātta on tá).
sāvarṇi + māṇḍūkeya (dvandva) → sāvarṇi keeps its own accent → sā́varṇimāṇḍūkeyau (udātta on sā).
The rest of each compound is anudātta.
Bhaṭṭoji explains the first pair. kārta is "this belongs to Kṛta", and kaujapa "this belongs to Kujapa". Both end in the affix aṇ. An affix carries the udātta on its first vowel (SK 3708), so kārta has it on ta. The Subodhinī says sāvarṇi ends in iñ. A word that ends in a ñ-marked affix is ādyudātta (SK 3686), so sāvarṇi has it on sā. The exact syllables are not stated by Bhaṭṭoji or his commentators; they are inferred from these grounds.
Applies when / blocked by. The rule holds only for the dvandvas of its list. Bhaṭṭoji does not reproduce the list.
Watch out. The Subodhinī derives kārta and kaujapa in the sense of a descendant, not "this belongs to". Scholars discuss this. The Subodhinī derives māṇḍūkeya with ḍhak (SK 1122).
Rule. mahat, as prior member, keeps its own accent before ten words. They are vrīhi, aparāhṇa, gṛṣṭi, iṣvāsa, jābāla, bhāra, bhārata, hailihila, raurava and pravṛddha. Bhaṭṭoji states that mahat is antodātta. The Subodhinī says it is given so by nipātana in uṇādi 241. A nipātana is a finished form stated in a sūtra.
Example.
mahat + vrīhi → mahā for mahat (SK 807, as the Subodhinī notes) → mahāvrīhi
→ mahat keeps its final udātta → mahā́vrīhiḥ (udātta on hā).
In the same way: mahā́parāhṇaḥ, mahā́gṛṣṭiḥ, mahā́hailihilaḥ.
mahā + iṣvāsa → ā and i merge into e → maheṣvāsa
→ the single substitute of an udātta and another vowel is udātta (SK 3658) → mahéṣvāsaḥ (udātta on he).
The Subodhinī allows a svarita here instead, by SK 3659: mahèṣvāsaḥ (independent svarita on he).
Applies when / blocked by. This accent holds only in the compound taught by name (pratipadokta) in sanmahat… (SK 740). A pratipadokta compound is one that a sūtra teaches specifically, not by a general rule. So the rule does not reach a sixth-case compound. mahato vrīhiḥ "the rice of a great one" → mahadvrīhiḥ. Its accent is not stated.
Watch out. Why name pravṛddha, if SK 3780 might cover a kta-final word anyway? The Subodhinī answers: SK 3780 too takes only the compound taught by name, the śreṇyādi type (SK 738). So pravṛddha needs its place here.
Rule. kṣullaka keeps its own accent before vaiśvadeva. By the word ca "also", mahat does the same.
Example.
kṣullaka + vaiśvadeva → kṣullaka keeps its final udātta → kṣullakávaiśvadevam (udātta on ká).
mahat + vaiśvadeva → mahā → mahā́vaiśvadevam (udātta on hā).
Bhaṭṭoji derives kṣullaka. kṣulla is "that which takes (lāti) hunger (kṣudh)". The Subodhinī adds the affix ka (SK 2915), and d becomes l (SK 117). The taddhita ka in the sense "unknown" and the like (SK 2028) then gives kṣullaka. Bhaṭṭoji says that word is antodātta.
Rule. uṣṭra "camel" keeps its own accent before sādin and vāmin.
Example.
uṣṭra + sādin → uṣṭra keeps its first-syllable udātta → úṣṭrasādī "camel-rider" (udātta on u).
uṣṭra + vāmin → úṣṭravāmī (udātta on u).
Bhaṭṭoji says uṣṭra is uṣ with ṣṭran, and is ādyudātta. The Subodhinī names the root uṣ "to burn" and the uṇādi rule (uṇādi 601). That the first-syllable udātta comes from the n-marker of ṣṭran (SK 3686) is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. go "cow" keeps its own accent before sāda, sādi and sārathi.
Example.
go + sāda → go keeps its udātta → gósādaḥ (udātta on go).
go + sādi → gósādiḥ; go + sārathi → gósārathiḥ (udātta on go).
go has one syllable, so its udātta can stand only there.
Applies when / blocked by. The Subodhinī says what this rule overrides. For sāda and sādin it overrides the kṛt accent, the accent a kṛt-final member would have. For sārathi it overrides the compound accent (SK 3734).
Watch out. The Subodhinī reads the second member as sādin, with ṇini: gosādī. Bhaṭṭoji gives gosādiḥ.
Rule. In seven named compounds, and in dāsībhāra and the rest, the prior member keeps its own accent. The seven are kurugārhapata, riktaguru, asūtajaratī, aślīladṛḍharūpā, pārevaḍavā, taitilakadrū and paṇyakambala.
Example.
kurūṇāṃ gārhapatam "the householdership of the Kurus" → kurúgārhapatam. kuru ends in the affix u; the Subodhinī says the affix accent makes it antodātta (udātta on rú).
A vārttika (a supplementary statement) adds vṛji: vṛ́jigārhapatam. Bhaṭṭoji says vṛji is ādyudātta (udātta on vṛ).
rikto guruḥ "an empty teacher" → ríktaguruḥ. rikta is ādyudātta by rikte vibhāṣā (SK 3696) (udātta on ri).
asūtā jaratī "an old woman who has not given birth" → ásūtajaratī (udātta on a).
aślīlā dṛḍharūpā "unlovely, firm of form" → áślīladṛḍharūpā (udātta on a). aślīla is a compound with the negative a-, so it is ādyudātta. ślīla is "having loveliness (śrī)": the affix lac, and l in place of r.
pāre vaḍavā iva "like a mare on the far bank" → pārévaḍavā. By nipātana this is a compound in the sense of "like", and the case-ending is kept. pāra is antodātta, being in the ghṛtādi list (phiṭ 21). That the merged e of pāre is then udātta by SK 3658 is not stated by Bhaṭṭoji or his commentators; it is inferred (udātta on re).
taitilānāṃ kadrūḥ "the Kadrū of the Taitilas" → taitilákadrūḥ. taitila ends in aṇ. The udātta on lá, the affix vowel (SK 3708), is not stated by Bhaṭṭoji or his commentators; it is inferred.
dāsyā bhāraḥ "a slave-woman's load" → dāsī́bhāraḥ (udātta on sī). Also deváhūtiḥ (udātta on va). The Subodhinī makes dāsī and deva antodātta.
Applies when / blocked by. A second vārttika limits paṇyakambala to a name: páṇyakambalaḥ (udātta on pa). paṇya ends in yat and is ādyudātta (SK 3701, as the Subodhinī notes). For a blanket merely to be sold, only the compound accent holds: paṇyakambaláḥ (udātta on the last syllable). Bhaṭṭoji's reason: the kṛtya accent is only for the compound taught by name. A kṛtya is an affix such as yat, giving the sense "to be done"; paṇya is "to be sold". The Subodhinī identifies the accent and the compound as SK 3736 and SK 749.
Watch out. dāsībhārādi is an open list (ākṛtigaṇa). Bhaṭṭoji says any tatpuruṣa that should keep the prior member's accent, with no rule of its own, belongs to it. A Vedic example: sá rāyé sá púraṃdhyām. puraṃdhi is "that in which the body (pura) is placed": the affix ki (SK 3271). The case-ending of puram is kept, a Vedic feature. pura is ādyudātta by phiṭ 26. So the udātta of púraṃdhyām is on pu.
yūpāya dāru → yū́padāru (yūpa nit: udātta yū); kaṣṭaṃ śritaḥ → kaṣṭáśritaḥ (udātta tá, against thāthādi's final udātta)These six sūtras still keep the prior member's own accent. Now the prior member is chosen by its case. A word in the fourth case (dative) keeps its accent before three kinds of final member. A word in the second case (accusative) or the third case (instrumental) keeps it before a kta-final word. kta is the past-participle affix ta, as in kṛtam "done". One rule takes a karmadhāraya instead.
Later in this chapter, SK 3878 (thātha…) makes a final member in kta and some other affixes antodātta, after a gati, kāraka or upapada. Several of these rules are exceptions (apavāda) to it.
The first running example is yūpāya dāru "wood for a sacrificial post".
yūpāya dāru → compound → yūpadāru
→ yūpa keeps its own accent → yū́padāru (udātta on yū; dāru anudātta).
The Subodhinī says yūpa has the n-marked affix pa (uṇādi 307), so it is ādyudātta.
The second is kaṣṭaṃ śritaḥ "resorted to hardship".
kaṣṭam + śrita → compound → kaṣṭaśrita
→ kaṣṭa keeps its final udātta → kaṣṭáśritaḥ (udātta on ṭa).
SK 3878 would instead put the udātta on the last syllable.
Rule. A fourth-case prior member keeps its own accent when the final member names what serves its purpose (tadartha). Bhaṭṭoji glosses: the final member is that which is for the referent of the fourth-case word.
Example.
yūpāya dāru "wood for a sacrificial post" → yūpadāru
→ yūpa keeps its first-syllable udātta → yū́padāru (udātta on yū).
The Subodhinī gives the ground: yūpa has the n-marked uṇādi affix pa, so it is ādyudātta.
Rule. A fourth-case prior member keeps its own accent before artha. "Fourth case" is carried forward from SK 3777.
Example.
deva + artha "for the gods" → a and a merge → devārtha
→ deva keeps its final udātta → devā́rtham (udātta on dā).
The Subodhinī says deva has the affix ac of the paca list, so its udātta is on its last syllable. The merged ā is then udātta by SK 3658. That last step is not stated by Bhaṭṭoji or his commentators; it is inferred.
Rule. A fourth-case prior member keeps its own accent before a kta-final word too.
Example.
gobhyo hitam "good for cows" → gohitam
→ go keeps its udātta → góhitam (udātta on go).
The Subodhinī says the fourth case comes by SK 631.
Rule. In a karmadhāraya before a kta-final word, a prior member that does not end in a niṣṭhā keeps its own accent. A karmadhāraya is a tatpuruṣa whose two members refer to the same thing. A niṣṭhā is a word ending in kta or ktavatu.
Example.
śreṇi + kṛta "made into a guild" → śreṇi keeps its first-syllable udātta → śréṇikṛtāḥ (udātta on śre).
pūga + kṛta "made into a band" → pūga keeps its final udātta → pūgákṛtāḥ (udātta on ga).
Bhaṭṭoji says śreṇi is ādyudātta and pūga antodātta. The Subodhinī gives the ground for śreṇi: its affix ni is n-marked (uṇādi 491).
Applies when / blocked by. Why "in a karmadhāraya"? śreṇyā kṛtam "done by a guild" → śreṇikṛtam, a third-case compound. Why "not a niṣṭhā"? kṛtākṛtam "done and undone", whose prior member kṛta is a niṣṭhā (SK 739). The text gives no accent for these two.
Watch out. The Subodhinī says only the śreṇyādi compound (SK 738) is meant. It is the compound taught by name with a kta-final word.
Rule. Before a kta-final word, a second-case prior member keeps its own accent. This holds when the compound denotes what is not abandoned (ahīna). The rule is an exception to thātha… (SK 3878).
Example.
kaṣṭaṃ śritaḥ "resorted to hardship" → kaṣṭa keeps its final udātta → kaṣṭáśritaḥ (udātta on ṭa).
grāma + gata "gone to the village" → grāma keeps its first-syllable udātta → grā́magataḥ (udātta on grā).
Bhaṭṭoji says kaṣṭa is antodātta, and grāma is ādyudātta by its n-marked affix. The Subodhinī names that affix: man (uṇādi 140).
Applies when / blocked by. Why "not abandoned"? kāntārātītáḥ "who has passed beyond the wilderness". Here SK 3878 applies, with the udātta on the last syllable. A vārttika adds "when there is no upasarga": sukhaprāptáḥ "who has attained happiness", with pra. Here too the last syllable is udātta. The Subodhinī states both of these final accents.
Watch out. The Subodhinī says anupasarga should replace ahīna in the sūtra, as the wider condition.
Rule. A third-case prior member keeps its own accent before a kta-final word that denotes the object (karman). So the kta must be passive.
Example.
tvayā ūtāḥ "protected by you" → tva + ūta → a and ū merge into o → tvotāsaḥ
→ the merged vowel is udātta (SK 3658) → tvótāsaḥ (udātta on tvo).
The Subodhinī explains these steps. It adds that tā is then a dependent svarita, the syllable just after an udātta. The syllables after it are pracaya, an unaccented level run. Neither is marked.
rudra + hata "struck by Rudra" → rudráhataḥ (udātta on dra). Bhaṭṭoji says rudra ends in the affix rak. The udātta on the affix vowel (SK 3708) is not stated by Bhaṭṭoji or his commentators; it is inferred.
mahārāja + hata "struck by the great king" → mahārājáhataḥ (udātta on ja). The Subodhinī says mahārāja ends in ṭac (SK 788) and is antodātta.
Applies when / blocked by. Why "denoting the object"? rathena yātaḥ "gone by chariot" → rathayātaḥ. The Subodhinī explains: yā is a root of motion, so its kta names the agent (SK 3086). The text gives no accent for it.
puras (antodātta) + hita → puróhitam (udātta ro); parā + añc (kvin) → párāñcaḥ (udātta pa)A gati is a word such as an upasarga (pra, sam, ni) or puras, joined to a root. These five sūtras let a gati prior member keep its own accent. They override two rules taught later in this chapter. One is thātha… (SK 3878), the final udātta of a kta-final member. The other is the kṛt accent (SK 3873), by which a kṛt-final member keeps its own accent after a gati.
The first running example is puras "in front" with hita "placed", the kta form of dhā.
puras + hita → compound → purohita
→ puras keeps its final udātta → puróhitam (udātta on ro).
The Subodhinī says puras has the affix asi (SK 1975) and is antodātta by the affix accent.
The second is parā with añc "to bend, to go" and the affix v (kvin, SK 373).
parā + añc + kvin → parāñc
→ parā keeps its first-syllable udātta → párāñcaḥ (udātta on pa).
Rule. Before a kta-final word in the object sense, a gati that stands immediately before the root keeps its own accent. This is an exception to thātha… (SK 3878).
Example.
puras + hita → puras keeps its final udātta → puróhitam (udātta on ro).
The Subodhinī notes that four accents compete here. They are the compound's final udātta, the accent of an indeclinable prior member, the kṛt accent and SK 3878. The prior member's own accent prevails.
Applies when / blocked by. Why "immediately before" (anantara)? abhyúddhṛtaḥ. Here abhi does not stand next to the root, so it does not keep its accent. The Subodhinī explains: ud first joins hṛta and keeps its own udātta. Then abhi joins uddhṛta. So the word keeps the udātta of ud (udātta on u).
Watch out. After a kāraka prior member, only SK 3878 applies: dūrādāgatáḥ "come from afar" (udātta on the last syllable). A kāraka is a participant in the action, such as the source "from afar". Bhaṭṭoji's reason is that SK 3878 is satiśiṣṭa. It is taught for the outer compound, dūrāt + āgata, which already exists. So it wins over the accent settled inside āgata.
Rule. Before a kṛt affix that begins with t, is n-marked and is not tu, the adjacent gati keeps its own accent. This is an exception to the kṛt accent (SK 3873).
Example.
pra + bhūti (ktin: begins with t, n-marked) → pra keeps its udātta → prábhūtau (udātta on pra).
Bhaṭṭoji's verse: ágne rāyó nṛ́tamasya prábhūtau. Its udāttas are on a, yo, nṛ and pra.
sam + gati → sam keeps its udātta: sáṅgatiṃ góḥ. Its udāttas are on sa and go.
The Subodhinī names the affix of prabhūti as ktin.
Applies when / blocked by. Why "beginning with t"? prajalpākaḥ, with ṣākan (SK 3135). Why "n-marked"? prakartā, with tṛc. Why "not tu"? āgantuḥ, with tun. The text gives no accent for these three.
Watch out. The Subodhinī says the initial t is judged on the affix as first taught (upadeśa). So a form with an inserted i, such as pralapitā, still counts.
Rule. Before a tavai-final word, two things happen together (yugapat). The adjacent gati keeps its own accent. And the last vowel of the tavai-word is udātta. This is an exception to the kṛt accent. tavai is a kṛt affix taught in the sense of a kṛtya (SK 3441).
Example.
anu + etavai → anvetavai; vai becomes vā before u
→ anu keeps its first-syllable udātta, and the end of tavai is udātta → ánvetavā́ u "to follow after".
The two udāttas are on a of anu and on vā. The Subodhinī says anu is ādyudātta by phiṭ 81.
Watch out. The sūtra does not say "udātta". The Subodhinī draws it from prakṛtyā, which is still in force. It says yugapat forbids taking the two accents in turn. One word here carries two udāttas.
Rule. A gati that does not end in ik (i, u, ṛ, ḷ) keeps its own accent before añc with the affix v. The Subodhinī says the affix is kvin (SK 373), of which only v is not a marker.
Example.
parā + añc + kvin → parāñc → parā keeps its first-syllable udātta → yé párāñcas tān "those who are turned away, them". The udāttas are on ye and pa.
Applies when / blocked by. Why "not ending in ik"? pratyáñco yantu "let them go back". prati ends in i, so the kṛt accent applies (udātta on tyañ).
Why "with the affix v"? udañcanam. The text gives no accent for it.
Watch out. This rule also beats the cu accent (cau, SK 3652), which would put the udātta before an añc that has lost its a. Bhaṭṭoji says this rule alone applies, "being later": jahí vṛ́ṣṇyāni kṛṇuhī́ párācaḥ. The udāttas are on hi, vṛ, hī and pa. In párācaḥ the udātta is on pa, not on rā. The Subodhinī, after Haradatta, reads "being later" as "because it is wanted". Scholars discuss this.
Rule. ni and adhi also keep their own accent before añc with v, although they end in ik.
Example.
ni + añc + kvin → the udātta i of ni becomes y → nyañc
→ the following a of añc becomes svarita (SK 3657) → nyàṅṅ uttānáḥ "turned downward, stretched out".
nyàṅ has no udātta of its own: it is an independent svarita. In uttānáḥ the udātta is on nā.
Also adhyaṅ. The text gives no accent for it.
Watch out. The Subodhinī adds that here too this rule alone applies, over the cu accent.
īṣat (antodātta) + kaḍāra → īṣátkaḍāraḥ (udātta ṣat) or, by the compound accent, īṣatkaḍāráḥ; rājan (ādyudātta) + brāhmaṇa → rā́jabrāhmaṇaḥ (udātta rā)SK 3788 adds the word anyatarasyām "optionally". From here through SK 3797 the prior member keeps its accent only optionally. In the other option another accent applies, often the compound's final udātta (SK 3734). Bhaṭṭoji's glosses say vā "optionally". That this vā is the anyatarasyām of SK 3788 carried forward is not stated by Bhaṭṭoji or his commentators; it is inferred.
The first running example is īṣat "slightly" with kaḍāra "tawny".
īṣat + kaḍāra → īṣat keeps its final udātta → īṣátkaḍāraḥ (udātta on ṣat)
or → the compound accent → īṣatkaḍāráḥ (udātta on the last syllable).
The second is rājan "king" with brāhmaṇa.
rājan + brāhmaṇa → rāja → rājan keeps its first-syllable udātta → rā́jabrāhmaṇaḥ (udātta on rā).
Rule. īṣat, as prior member, optionally keeps its own accent. Bhaṭṭoji says īṣat is antodātta. The Subodhinī says this is by the phiṭ accent.
Example.
īṣat + kaḍāra "slightly tawny" → īṣátkaḍāraḥ (udātta on ṣat).
Other option → īṣatkaḍāráḥ (udātta on the last syllable). The Subodhinī gives this second option.
Applies when / blocked by. With a kṛt final member, only the kṛt accent applies: īṣadbhedaḥ, with khal. The Subodhinī gives two reasons. The kṛt accent is later; or only the compound taught by name, īṣad guṇavacanena, is meant. Bhaṭṭoji gives neither.
Rule. A prior member that names a quantity of gold optionally keeps its own accent before dhana "wealth".
Example.
"two *suvarṇa*s are its quantity" → dvisuvarṇa
dvisuvarṇa + dhana "wealth of two *suvarṇa*s" → dvisuvarṇa optionally keeps its own accent → dvisuvarṇadhanam.
The text does not say which syllable of dvisuvarṇa is udātta. The Subodhinī gives the other option, the compound accent. It also gives the scale: five *kṛṣṇala*s make a māṣa, sixteen *māṣa*s a suvarṇa, four *suvarṇa*s a pala.
Applies when / blocked by. Why "gold"? prasthadhanam, a measure of grain. Why "quantity"? kāñcanadhanam, gold but no quantity. Why "before dhana"? niṣkamālā.
Watch out. Even in a bahuvrīhi there is only the option. A bahuvrīhi is a compound that describes something outside itself. The text gives no bahuvrīhi example here. In a bahuvrīhi, SK 3735 would make the prior member's accent compulsory. This rule is later, so it keeps the option. The Subodhinī says the rule reaches a bahuvrīhi because there is no "tatpuruṣa" heading here.
Rule. prathama optionally keeps its own accent when it means "new", "just begun" (aciropasampatti). Bhaṭṭoji glosses this as newness.
Example.
prathama + vaiyākaraṇa "one who has just begun to study grammar" → prathamávaiyākaraṇaḥ (udātta on ma).
Bhaṭṭoji says prathama is prath with the affix amac. amac is c-marked, so citaḥ (SK 3710) would make the word antodātta. That the udātta is on ma is not stated by Bhaṭṭoji or his commentators; it is inferred.
Applies when / blocked by. Why "new"? prathamo vaiyākaraṇaḥ "the first grammarian". The Subodhinī says the compound in that sense is always antodātta.
Rule. katara and katama "which (of two, of many)" optionally keep their own accent in a karmadhāraya.
Example.
katara + kaṭha "which of the two is a Kaṭha?" → katara keeps its final udātta → katarákaṭhaḥ (udātta on ra).
The Subodhinī says katara and katama are antodātta, because their affixes ḍatarac and ḍatamac are c-marked.
Watch out. The word "karmadhāraya" is there for the next rule. Bhaṭṭoji explains: here the limit already follows, because only the compound taught by name is meant. The Subodhinī names it: katarakatamau jātiparipraśne (SK 742), which is a karmadhāraya anyway.
Rule. ārya optionally keeps its own accent before brāhmaṇa or kumāra, in a karmadhāraya.
Example.
ārya + kumāra → ārya keeps its own accent → āryàkumāraḥ.
ārya + brāhmaṇa → āryàbrāhmaṇaḥ.
Bhaṭṭoji says ārya ends in ṇyat and so is antasvarita: its last vowel is svarita. The Subodhinī gives the rule: ṇyat is t-marked, and a t-marked affix is svarita (SK 3729). So these compounds have no udātta at all. yà is an independent svarita, and every other syllable is anudātta.
Applies when / blocked by. Why "ārya"? paramabrāhmaṇaḥ. Why "brāhmaṇa and kumāra"? āryakṣatriyaḥ. "In a karmadhāraya" still holds here. The text gives no accent for these.
Rule. rājan also optionally keeps its own accent before brāhmaṇa or kumāra, in a karmadhāraya.
Example.
rājan + brāhmaṇa → rājabrāhmaṇa → rājan keeps its first-syllable udātta → rā́jabrāhmaṇaḥ (udātta on rā).
rājan + kumāra → rā́jakumāraḥ (udātta on rā).
The Subodhinī says rājan has the n-marked affix kanin, so it is ādyudātta.
Watch out. Why a separate sūtra, and not ārya and rājan together? Bhaṭṭoji says the split (yogavibhāga) is for the next rule. Then SK 3794 takes rājan alone. The Subodhinī adds a second purpose: it stops a one-to-one pairing of ārya, rājan with brāhmaṇa, kumāra.
Rule. rājan as a sixth-case word optionally keeps its own accent before pratyenas.
Example.
rājñaḥ pratyenāḥ "the king's pratyenas" → rājapratyenas
→ rājan keeps its first-syllable udātta → rā́japratyenāḥ (udātta on rā).
The Subodhinī glosses pratyenas as "one to whom sin has gone back": one answerable to the king for an offence.
Applies when / blocked by. Why "sixth case"? "Elsewhere it does not apply." The Subodhinī names the case: the karmadhāraya rājā cāsau pratyenāś ca, "a king who is also a pratyenas".
Rule. Before a kta-final word, in a compound that means "constantly", the prior member optionally keeps its own accent.
Example.
nitya + prahasita "constantly laughing" → nitya keeps its first-syllable udātta → nítyaprahasitaḥ (udātta on ni).
Bhaṭṭoji says this is a second-case compound by kālāḥ (SK 690). nitya ends in tyap and is ādyudātta. hasita is antodātta by thātha… (SK 3878). So in the other option the udātta is on the last syllable: nityaprahasitáḥ.
Applies when / blocked by. Why "constantly"? muhūrtaprahasitaḥ "laughing for a moment". The Subodhinī says it has the final udātta of SK 3878. It lays out the chain. The compound accent gives way to SK 3781. SK 3781 gives way to SK 3878. This rule is an optional exception to SK 3878 in turn.
Rule. The word grāma "village" optionally keeps its own accent before a word for an artisan (śilpin).
Example.
grāma + nāpita "the village barber" → grāma keeps its first-syllable udātta → grā́manāpitaḥ (udātta on grā).
Bhaṭṭoji says grāma is ādyudātta. The Subodhinī calls the compound a sixth-case compound.
Applies when / blocked by. Why "grāma"? paramanāpitaḥ "the excellent barber". Why "an artisan"? grāmarathyā "the village street".
Watch out. The Subodhinī says grāma means the word itself, while śilpin means any word for an artisan. So nāpita "barber" qualifies, though the sūtra does not name it.
Rule. rājan, used to praise, optionally keeps its own accent before a word for an artisan.
Example.
rājan + nāpita "a barber fit for a king" → rā́janāpitaḥ (udātta on rā).
rājan + kulāla "a potter fit for a king" → rā́jakulālaḥ (udātta on rā).
The Subodhinī explains the praise: the barber is skilled enough to serve the king.
Applies when / blocked by. Why "to praise"? rājanāpitaḥ again, now simply "the king's barber". Its accent is not stated. Why "an artisan"? rājahastī "the king's elephant".
Watch out. The example and the counter-example are the same word. Only the meaning decides the accent.
mukuṭe kārṣāpaṇam → múkuṭekārṣāpaṇam (udātta mú); go + ballava → góballavaḥ (udātta gó)SK 3798 opens a new heading, ādir udāttaḥ: "the first vowel of the prior member is udātta". The prior member no longer keeps its own accent. It is made ādyudātta, whatever accent it had before, and the rest of the compound is anudātta. This heading governs every rule left in this part. The option of SK 3788 has stopped, so these rules are compulsory unless they say "optionally".
The five sūtras of this group choose the compound by what it denotes. It may be a customary due, a person engaged in a task, a superintendent, a wretched artisan, or a person held in contempt.
The first running example is mukuṭe kārṣāpaṇam "the kārṣāpaṇa due at the mukuṭa".
mukuṭe + kārṣāpaṇa → seventh-case compound, the case-ending kept → mukuṭekārṣāpaṇa
→ the first vowel of the prior member is udātta → múkuṭekārṣāpaṇam (udātta on mu).
The second is go with ballava "cowherd".
go + ballava → góballavaḥ (udātta on go).
Rule. This is a heading (adhikāra): the first vowel of the prior member is udātta. A heading is a rule whose words are read into the rules that follow it. Bhaṭṭoji says only "this is a heading".
Example. The heading has no example of its own. Each rule after it applies it: múkuṭekārṣāpaṇam (SK 3799), góballavaḥ (SK 3800).
Applies when / blocked by. The Subodhinī explains the grammar. The word pūrvapadam "the prior member" is still in force from SK 3735. Here it is turned by sense into the sixth case: "the first vowel of the prior member". Throughout this section a first-case word for the prior member stands in that sixth-case sense.
Watch out. Bhaṭṭoji does not say how far the heading runs. The rules after it treat the prior member as ādyudātta. Where SK 3817 names another vowel, that vowel takes the place of the first.
Rule. A seventh-case word, or a word for a taker (hārin), as prior member, is ādyudātta before a word for a customary due (dharmya). The final member must not be haraṇa. Bhaṭṭoji defines both words. A hārin is one who receives what is to be given. A dharmya is a due fixed by custom.
Example.
mukuṭe + kārṣāpaṇa "the kārṣāpaṇa due at the mukuṭa" → múkuṭekārṣāpaṇam (udātta on mu).
hale + dvipadikā "the dvipadikā due on the plough" → háledvipadikā (udātta on ha).
These are seventh-case compounds by SK 721, and the case-ending is kept by SK 968.
yājñika + aśva "the horse due to the sacrificial priest" → yā́jñikāśvaḥ (udātta on yā).
vaiyākaraṇa + hastin "the elephant due to the grammarian" → vaíyākaraṇahastī (udātta on vai).
Bhaṭṭoji describes the custom. In some places a kārṣāpaṇa is given at the mukuṭa, but a horse to sacrificial priests.
Applies when / blocked by. Why "a customary due"? stamberamaḥ "elephant" and karmakaravarddhitakaḥ "the labourer's rice-ball". The Subodhinī explains: the rice-ball is given so that the labourer will work, not by custom. Why "not haraṇa"? vāḍavaharaṇam. A vāḍava is the stallion, "the one of the mare". Its haraṇa is the food given to it after mating, to build up its body. The Subodhinī gives this word its final udātta by SK 3884.
Watch out. The words "not haraṇa" teach something more (a jñāpaka, a hint built into the wording). The kṛt accent comes later in the text than this rule. Even so, the hārin accent blocks it. Otherwise no exclusion would be needed. So vā́ḍavahāryam has the hārin accent (udātta on vā).
Rule. In a compound that denotes one engaged (yukta), the prior member is ādyudātta. Bhaṭṭoji says yukta is one intent on what is to be done.
Example.
go + ballava "cowherd" → the first vowel of go is udātta → góballavaḥ (udātta on go).
The Subodhinī says ballava and like words denote keepers of cows and other animals.
Watch out. This rule is compulsory. The "optionally" of SK 3788 stops at SK 3797. The Subodhinī says yukta here comes from yuj in the sense of devotion to a task (yuja samādhau). The niyukta of SK 3809 comes from yuj "to join" (yujir yoge).
Rule. Before adhyakṣa "superintendent", the prior member is optionally ādyudātta.
Example.
go + adhyakṣa "superintendent of cows" → the compound, as given, is gavādhyakṣa, with gava for go
→ optionally the first vowel of the prior member is udātta → gávādhyakṣaḥ (udātta on ga).
Bhaṭṭoji gives only the finished example. He does not show the step from go to gava, and the other option is not stated.
Applies when / blocked by. The Subodhinī explains why the rule exists. An adhyakṣa is also one engaged in a task. So SK 3800 would make the accent compulsory. This rule makes it optional.
Rule. pāpa "wretched", before a word for an artisan, is ādyudātta. The Subodhinī reads "optionally" into this rule, from SK 3801.
Example.
pāpa + nāpita "a wretched barber" → pā́panāpitaḥ (udātta on pā).
Applies when / blocked by. Only the compound taught by name in pāpāṇake kutsitaiḥ (SK 733) is meant. That is a karmadhāraya, where pāpa expresses contempt. So the rule does not apply in a sixth-case compound, such as "the barber of a sinner".
Watch out. The Subodhinī says pāpa means the word itself, but śilpin means any word for an artisan.
Rule. When contempt (kṣepa) is meant, the prior member is ādyudātta before four kinds of final member. They are a word for a descendant (gotra), a word for a pupil (antevāsin), māṇava "boy", and brāhmaṇa.
Example.
Descendant: bhāryā + sauśruta → bhā́ryāsauśrutaḥ "the wife-ruled Sauśruta" (udātta on bhā). The contempt is that Suśruta's descendant is ruled by his wife.
Pupil: kumārī + dākṣa → kúmārīdākṣāḥ "Dākṣa students for the sake of a girl" (udātta on ku).
Pupil: odana + pāṇinīya → ódanapāṇinīyāḥ "Pāṇinīya students for the sake of rice" (udātta on o).
Bhaṭṭoji explains: they study the treatises of Dākṣi and others only to win a girl or a meal.
māṇava: bhikṣā + māṇava → bhíkṣāmāṇavaḥ (udātta on bhi), a boy only in the thought "I shall get alms".
brāhmaṇa: bhaya + brāhmaṇa → bháyabrāhmaṇaḥ (udātta on bha), one who becomes a brāhmaṇa out of fear.
Applies when / blocked by. Why these four? dāsīśrotriyaḥ. Why "contempt"? paramabrāhmaṇaḥ "the excellent brāhmaṇa". The text gives no accent for these two.
Watch out. The Subodhinī names the author of the Dākṣa texts as Dakṣa, not Dākṣi. It glosses bhayabrāhmaṇaḥ as one who is not a brāhmaṇa but acts like one for fear of the king's punishment.
madhu + maireya → mádhumaireyaḥ (udātta má); danta + lekhaka → dántalekhakaḥ (udātta dán)These five sūtras still make the prior member ādyudātta. Now the final member decides. It may be the drink maireya, a vessel meant for some food, one of four words (go, biḍāla, siṃha, saindhava) used as a standard of comparison, or an aka-noun. An aka-noun is a member ending in the affix aka, for example one made with ṇvul.
The first running example is madhu "honey" with maireya, an intoxicating drink.
madhu + maireya → the first vowel of madhu is udātta → mádhumaireyaḥ (udātta on ma).
The second is danta "tooth" with lekhaka "painter".
danta + lekhaka → dántalekhakaḥ "a tooth-painter" (udātta on dan).
Rule. A word for an ingredient (aṅga) of maireya is ādyudātta as prior member before maireya. Bhaṭṭoji says maireya is a particular intoxicating drink.
Example.
madhu + maireya "maireya made of honey" → mádhumaireyaḥ (udātta on ma).
Bhaṭṭoji explains: the drink is a product of honey, so honey is its ingredient.
Applies when / blocked by. Why "ingredient"? paramamaireyaḥ "excellent maireya". Why "maireya"? puṣpāsavaḥ "flower liquor". The text gives no accent for these.
Watch out. The Subodhinī says the plural aṅgāni shows that the meaning is meant, not the word aṅga itself.
Rule. A name of a food (bhakta), as prior member, is ādyudātta before a word for something meant for that food. Bhaṭṭoji says bhakta means food.
Example.
bhikṣā + kaṃsa "a bowl for alms-food" → bhíkṣākaṃsaḥ (udātta on bhi).
bhājī + kaṃsa "a bowl for bhājī" → bhā́jīkaṃsaḥ (udātta on bhā).
Bhaṭṭoji says bhikṣā and the rest are particular foods. The Subodhinī calls these sixth-case compounds.
Applies when / blocked by. Why "a name of a food"? samāśaśālayaḥ "rice for eating". samāśa means the bare act of eating, not a food. Why "meant for it"? bhikṣāpriyaḥ "one to whom alms-food is dear". Bhaṭṭoji says this is a bahuvrīhi, and its prior member is antodātta: bhikṣā́priyaḥ (udātta on kṣā).
Watch out. The 1915 print reads samāśālayaḥ, one syllable short. Bhaṭṭoji's explanation needs samāśa + śālayaḥ, so it is read here as samāśaśālayaḥ. The Subodhinī adds that "name" (ākhyā) takes in words for particular foods, such as bhikṣā, not only words meaning "food".
Rule. Before go "cow", biḍāla "cat", siṃha "lion" or saindhava, used as the standard of comparison (upamāna), the prior member is ādyudātta.
Example.
dhānyaṃ gaur iva "grain like a cow" → dhānya + go → the affix ṭac (SK 729) → dhānyagava
→ the first vowel of dhānya is udātta → dhā́nyagavaḥ (udātta on dhā).
Bhaṭṭoji says the compound is of the vyāghrādi type (SK 735). That is a compound of comparison in which the thing compared stands first: here dhānya "grain". It means grain heaped in the shape of a cow.
Also góbiḍālaḥ, tṛ́ṇasiṃhaḥ, sáktusaindhavaḥ (udātta on go, tṛ, sa).
Applies when / blocked by. Why "as the standard of comparison"? paramasiṃhaḥ "a supreme lion". The text gives no accent for it.
Watch out. ṭac is c-marked, so citaḥ (SK 3710) would put the udātta on the last syllable. The Subodhinī says this rule sets that accent aside.
Rule. Before a final member ending in aka, in a compound that denotes a livelihood (jīvikā), the prior member is ādyudātta.
Example.
danta + lekhaka "a tooth-painter", one who lives by painting teeth → dántalekhakaḥ (udātta on dan).
Bhaṭṭoji says the compound is made by nityaṃ krīḍājīvikayoḥ (SK 711). The Subodhinī says lekhaka has ṇvul (SK 2895).
Applies when / blocked by. Why "ending in aka"? ramaṇīyakartā "a maker of lovely things". Why "a livelihood"? ikṣubhakṣikāṃ me dhārayasi "you owe me a sugar-cane eating". The text gives no accent for these.
Rule. Before an aka-final member, in a compound that names a game of the eastern people, the prior member is ādyudātta.
Example.
uddālaka + puṣpa + bhañjikā → úddālakapuṣpabhañjikā, the game "breaking the uddālaka flowers" (udātta on u).
Bhaṭṭoji says the ṇvul is by saṃjñāyām (SK 3286).
Applies when / blocked by. Why "of the eastern people"? jīvaputrapracāyikā, a game of the northerners. Why "a game"? tava puṣpapracāyikā "your turn at gathering flowers". Bhaṭṭoji calls this ṇvul in the sense of a turn. But SK 3288 prescribes ṇvuc for that sense, so the reading ṇvul here is doubtful; it may be a slip for ṇvuc.
Watch out. Here prācām is a sixth-case word: "of the eastern people". It does not mean "in the view of the eastern teachers", as the word often does. The Subodhinī explains Bhaṭṭoji's gloss in this way.
chatra + dhṛ + aṇ → chátradhāraḥ (udātta chá); tantu + vā + aṇ → tántuvāyaḥ (udātta tán); puṣpa + hārin → púṣpahārī (udātta pú)These seven sūtras make the prior member ādyudātta before agent nouns. Three take a final member in aṇ (SK 2913, SK 2914): chatradhāra, tantuvāya. One takes pāla "keeper". Two take a final member in ṇini, as in puṣpahārin "flower-taker". The last is a list of named compounds.
The running examples:
chatra + dhṛ + aṇ → chatradhāra "parasol-bearer", one appointed → chátradhāraḥ (udātta on cha).
tantu + vā + aṇ → tantuvāya "weaver", an artisan → tántuvāyaḥ (udātta on tan).
puṣpa + hārin → púṣpahārī "a taker of flowers" (udātta on pu).
Rule. Before an aṇ-final member, in a compound that denotes one appointed (niyukta), the prior member is ādyudātta.
Example.
chatra + dhṛ + aṇ (SK 2913) → chatradhāra "the parasol-bearer"
→ the first vowel of chatra is udātta → chátradhāraḥ (udātta on cha).
The Subodhinī names the affix rule, karmaṇy aṇ.
Applies when / blocked by. Why "one appointed"? kāṇḍalāvaḥ "a reaper of stalks". The text gives no accent for it.
Watch out. The Subodhinī separates niyukta from the yukta of SK 3800. niyukta is from yuj "to join" (yujir yoge) and means "put in charge". Such a person need not be devoted to the task.
Rule. Before an aṇ-final member, in a compound that denotes an artisan, the prior member is ādyudātta. But not if the aṇ comes after kṛ "to make".
Example.
tantu + vā + aṇ → y is inserted after the ā → tantuvāya "a weaver"
→ the first vowel of tantu is udātta → tántuvāyaḥ (udātta on tan).
The Subodhinī gives the affix rule (SK 2914) and the inserted y (SK 2761).
Applies when / blocked by. Why "an artisan"? kāṇḍalāvaḥ. Why "not after kṛ"? kumbhakāraḥ "a potter". The text gives no accent for these.
Watch out. The Subodhinī asks why this rule is needed, since SK 3809 might already apply. Its answer: the rule exists for the exclusion of kṛ.
Rule. In a name (saṃjñā) too, before an aṇ-final member, the prior member is ādyudātta. The exclusion of kṛ still holds.
Example.
tantuvāya, the name of a creature → tántuvāyo nāma kṛmiḥ "the creature called tantuvāya" (udātta on tan).
Applies when / blocked by. Bhaṭṭoji says "not after kṛ" still holds: rathakāro nāma brāhmaṇaḥ "the brāhmaṇa named rathakāra". The text gives no accent for it.
Rule. go, tanti and yava, as prior members, are ādyudātta before pāla "keeper".
Example.
go + pāla "cowherd" → gópālaḥ (udātta on go).
tanti + pāla → tántipālaḥ (udātta on tan).
yava + pāla "keeper of the barley" → yávapālaḥ (udātta on ya).
The Subodhinī analyses gopāla as "he protects cows". It derives tanti from tan "to extend" with ktin.
Applies when / blocked by. Bhaṭṭoji says the rule is for compounds that do not mean "appointed". In that sense SK 3809 would already apply; that reason is not stated by Bhaṭṭoji or his commentators; it is inferred. Why "go" and the others? vatsapālaḥ "calf-keeper". Why "pāla"? gorakṣaḥ "cow-protector". The text gives no accent for these.
Rule. Before a ṇini-final member, the prior member is ādyudātta.
Example.
puṣpa + hārin "a taker of flowers" → the first vowel of puṣpa is udātta → púṣpahārī (udātta on pu).
Bhaṭṭoji does not name the rule that gives ṇini here.
Rule. A prior member that is a standard of comparison (upamāna) is ādyudātta before a ṇini-final member. But this holds only when the base of the ṇini is a root meaning sound.
Example.
uṣṭra + krośin "one who cries out like a camel" → úṣṭrakrośī (udātta on u).
dhvāṅkṣa + rāvin "one who caws like a crow" → dhvā́ṅkṣarāvī (udātta on dhvā).
The Subodhinī says the ṇini comes by kartary upamāne (SK 2989).
Applies when / blocked by. Why "a base meaning sound"? vṛkavañcī "one who cheats like a wolf". vañc "to cheat" does not mean sound. The Subodhinī says only the kṛt accent applies there.
Why the word "base" (prakṛti)? Bhaṭṭoji says the root alone must mean sound, without its upasarga. So the rule does not apply in gardabhoccārī "one who brays like an ass". ud + car means "to utter", but car alone does not.
Watch out. Bhaṭṭoji says naming upamāna divides the work of this rule and SK 3813. The details are not stated by Bhaṭṭoji or his commentators; they are inferred. A prior member that is not a comparison follows SK 3813 before any ṇini-final member. A comparison is ādyudātta only when the root means sound.
Rule. yuktārohin and the other compounds of its list are ādyudātta.
Example.
yuktārohin → yúktārohī (udātta on yu).
āgatayodhin → ā́gatayodhī (udātta on ā).
kṣīrahotṛ → kṣī́rahotā (udātta on kṣī).
Applies when / blocked by. The Subodhinī asks why the rule is needed, since SK 3813 would cover a ṇini-final member. It answers that the list fixes both members. The accent holds only with these prior members and these final members.
Watch out. kṣīrahotā is a sixth-case compound (SK 703), says the Subodhinī. Three rules compete there. The kṛt accent (SK 3873) overrides the compound accent. manktin… (SK 3885) overrides the kṛt accent. This rule overrides SK 3885.
kuṭī + ja → kúṭījaḥ (udātta kú); indra + prastha → índraprasthaḥ (udātta í)The last seven sūtras of this part take five final members. They are ja "born", grāma "village", ghoṣa and its list, śālā "hall", and prastha. One rule moves the udātta to the next-to-last vowel of a long prior member.
The running examples:
kuṭī + jan + ḍa → kuṭīja "born in a hut" → the first vowel of kuṭī is udātta → kúṭījaḥ (udātta on ku).
indra + prastha → índraprasthaḥ (udātta on i).
Rule. Before ja, a prior member that ends in a long vowel is ādyudātta. So are kāśa, tuṣa, bhrāṣṭra and vaṭa. The Subodhinī says dīrgha "long" here means a prior member ending in a long vowel.
Example.
kuṭī + ja "born in a hut" → kúṭījaḥ (udātta on ku).
kāśa + ja "born in kāśa grass" → kā́śajaḥ (udātta on kā).
tuṣa + ja "born in chaff" → túṣajaḥ (udātta on tu).
bhrāṣṭra + ja "born in a frying pan" → bhrā́ṣṭrajaḥ (udātta on bhrā).
vaṭa + ja "born in a banyan" → váṭajaḥ (udātta on va).
The Subodhinī says ja is jan with the affix ḍa (SK 3007).
Rule. Before ja, a prior member with many vowels (bahvac) has its next-to-last vowel udātta. That vowel takes the place of the first vowel of SK 3798.
Example.
upasara + ja "born of the first mating" → the next-to-last vowel of upasara is udātta → upasárajaḥ (udātta on sa).
āmalakī + ja "born in the āmalakī tree" → āmalákījaḥ (udātta on la).
The Subodhinī explains upasara: the first approach of a male to a female for breeding, with ap (SK 3248).
Applies when / blocked by. Why "many vowels"? dagdhajāni tṛṇāni "grasses sprung up on burnt ground". dagdha has only two vowels. The text gives no accent for it.
Watch out. āmalakī ends in a long ī, so SK 3816 would make it ādyudātta. The Subodhinī says this rule wins, being later.
Rule. Before grāma, the prior member is ādyudātta. But not when the prior member denotes those who dwell there.
Example.
malla + grāma "the band of the Mallas" → mállagrāmaḥ (udātta on mal). Bhaṭṭoji says grāma here means a collection.
deva + grāma "a village owned by the gods" → dévagrāmaḥ (udātta on de).
Applies when / blocked by. Why "not those who dwell there"? dākṣigrāmaḥ "the dwelling-place of the Dākṣis". The text gives no accent for it.
Watch out. The Subodhinī reads the excluded word as the dwelling-place, not the dwellers. Bhaṭṭoji glosses it as a prior member that denotes the dwellers. Both give the same counter-example.
Rule. Before ghoṣa and the other words of its list, the prior member is ādyudātta.
Example.
dākṣi + ghoṣa → dā́kṣighoṣaḥ (udātta on dā).
dākṣi + kaṭa → dā́kṣikaṭaḥ (udātta on dā).
dākṣi + hrada "the Dākṣis' lake" → dā́kṣihradaḥ (udātta on dā).
hrada is one of the words of the ghoṣādi list.
Applies when / blocked by. The Subodhinī reports two views. Some carry forward "not those who dwell there" from SK 3818; others do not. Where dākṣighoṣa means "the ghoṣa where the Dākṣis dwell", the first view gives no ādyudātta, the second does. Bhaṭṭoji gives only the examples.
Rule. chātri and the other words of its list are ādyudātta as prior member before śālā "hall".
Example.
chātri + śālā → chā́triśālā (udātta on chā).
vyāḍi + śālā → vyā́ḍiśālā (udātta on vyā).
Applies when / blocked by. A tatpuruṣa ending in śālā may be neuter, with śāla. The Subodhinī gives the rule (SK 828). Then tatpuruṣe śālāyāṃ napuṃsake (SK 3857) would make śāla ādyudātta. Bhaṭṭoji says this rule still applies, by the precedence of the earlier rule (pūrvavipratiṣedha): chā́triśālam (udātta on chā).
Watch out. pūrvavipratiṣedha reverses the usual order. Normally the later of two conflicting rules wins. Here the earlier one, SK 3820, wins. The Subodhinī credits this to Vāmana and Haradatta. The 1915 print reads pūrvapratiṣedhena in Bhaṭṭoji's sentence, where the Subodhinī uses the usual pūrvavipratiṣedha; the sense is the same.
Rule. Before prastha, a prior member that is not vṛddha is ādyudātta. karkī and the other words of its list are excepted. A vṛddha word is one whose first vowel is a vṛddhi vowel (ā, ai, au). By SK 1338 (eṅ prācāṃ deśe), a word for an eastern place whose first vowel is e or o counts as vṛddha too (see SK 3822).
Example.
indra + prastha → the first vowel of indra is udātta → índraprasthaḥ (udātta on i).
Applies when / blocked by. Why "not vṛddha"? dākṣiprasthaḥ: dākṣi has ā as its first vowel. Why "except karkī and the rest"? karkīprasthaḥ, makarīprasthaḥ. The text gives no accent for these. That dākṣi is vṛddha because of its ā is not stated by Bhaṭṭoji or his commentators; it is inferred.
Watch out. The Subodhinī says the sūtra word is a-vṛddham "not vṛddha", because the next rule is made for vṛddha words.
Rule. mālā and the other words of its list are ādyudātta before prastha, too. Bhaṭṭoji says this rule is for vṛddha words, which SK 3821 left out.
Example.
mālā + prastha → the first vowel of mālā is udātta → mā́lāprasthaḥ (udātta on mā).
śoṇā + prastha → śóṇāprasthaḥ (udātta on śo). The Subodhinī says śoṇa is vṛddha by SK 1338.
Watch out. This rule ends Part V. Part VI goes on under the same heading, ādir udāttaḥ, with nagara (SK 3823).
brahma + nagaram → bráhmanagaram; gupta + armam → gúptārmamA compound has two members. The first is the pūrvapada, the prior member; the second is the uttarapada, the later member. The heading ādir udāttaḥ (SK 3798) is still running. Under it the first syllable of the pūrvapada becomes udātta; such a word is called ādyudātta. By SK 3650 every other syllable of the compound is then anudātta. As throughout this chapter, the acute marks the udātta, the grave marks an independent svarita, and anudātta is left unmarked.
Three sūtras close the heading. Two of them name the uttarapada: nagara "town" and arma. The third lists words that are excluded before arma. The running examples are bráhmanagaram, udātta on bra, and gúptārmam, udātta on gu.
Rule. Before nagara, a pūrvapada other than mahat and nava is ādyudātta. This does not happen when the town belongs to the northerners.
Example.
brahma + nagaram → the pūrvapada is neither mahat nor nava, and the town is not northern
→ the first syllable of the pūrvapada is udātta (ādir udāttaḥ, SK 3798)
→ bráhmanagaram, udātta on bra; every other syllable anudātta (SK 3650).
Applies when / blocked by. "pūrvapada" is carried from bahuvrīhau prakṛtyā pūrvapadam (SK 3735). "First syllable udātta" is the heading of SK 3798. Bhaṭṭoji explains anudīcām only by "if that is not of the northerners".
Counter-examples, with no accent stated:
mahānagaram.navanagaram.kārtikanagaram.Watch out. The Subodhinī says "other than mahat and nava" applies to each word separately. So the rule fails if the pūrvapada is mahat (mahānagaram) and also if it is nava (navanagaram).
Rule. Before arma, a pūrvapada of two or three vowels that ends in a or ā is ādyudātta. mahat and nava are still excluded, carried from SK 3823.
Example.
gupta + armam → gupta has two vowels and ends in a
→ first syllable of the pūrvapada udātta → gúptārmam, udātta on gu.
kukkuṭa + armam → three vowels, a-final → kúkkuṭārmam, udātta on ku.
Applies when / blocked by. Each condition has its counter-example. No accent is stated for them.
bṛhadarmam.kapiñjalārmam.mahārmam, navārmam.Watch out. Why must mahat be excluded by name, when it does not end in a? The Subodhinī explains. Before a word in apposition, mahat becomes mahā (SK 807). mahā ends in ā, so the rule would reach it. nava "new" is an ordinary a-final word. The Subodhinī gives this remark at SK 3823, but it is read here with the present rule, since only this rule has the a-final condition.
Rule. Before arma, the pūrvapadas bhūta, adhika, saṃjīva, madra, aśman and kajjala are not ādyudātta. arma is carried from SK 3824.
Example. The rule only forbids. Bhaṭṭoji states no accent for its forms:
bhūtārmam, adhikārmam, saṃjīvārmam, kajjalārmam.
For madra and aśman: madrārmam, aśmārmam, madrāśmārmam.
Applies when / blocked by. Bhaṭṭoji says madra and aśman are named "for the combination and for its separated members" (saṃghāta-vigṛhīta). So each word counts alone, and the pair madrāśma counts too.
The Subodhinī narrows this. The prohibition is really for madra alone and for madrāśma. madrāśma ends in a by the compound-final affix of SK 796. aśman alone is not a-final, so SK 3824 never reached it. The example aśmārmam is given only to follow the Vṛtti, a running commentary on the sūtras. Others think the whole remark has no basis in the Bhāṣya; scholars discuss this.
Vārttika. In the Veda, divodāsa and the like are added to the ādyudātta section:
dívodāsāya dāśúṣe "for Divodāsa, the worshipper".
The udātta is on di, the first syllable of divodāsa. The next word has its own udātta, on śu.
The Subodhinī adds that the sixth-case ending of divas is kept (aluk), by a vārttika under SK 979.
Watch out. aśmārmam is not a true example of the rule, on the Subodhinī's view. Learn the rule from madrārmam and madrāśmārmam.
sarva + śvetaḥ → sarváśvetaḥA new heading now moves the udātta to the other end of the pūrvapada. Under antaḥ the final syllable of the pūrvapada becomes udātta; such a word is antodātta. The heading runs up to SK 3845. Seven rules follow in this group. Each names a pūrvapada or an uttarapada and a meaning.
The running example is sarváśvetaḥ "white all over". The udātta is on va, the last syllable of sarva. Everything else is anudātta.
Rule. This is an adhikāra, a heading. From here on, the final syllable of the pūrvapada is udātta. The heading runs up to uttarapadādiḥ (SK 3845).
Applies when / blocked by. antaḥ takes the place of ādiḥ in SK 3798. "pūrvapada" still comes from SK 3735, and "udātta" from SK 3798. Bhaṭṭoji's gloss at SK 3827, "the pūrvapada is antodātta", confirms this.
Watch out. The Subodhinī reads the pūrvapada as standing in the sense of the sixth case. So the rules mean "the final of the pūrvapada". It makes the same remark for the ādyudātta section at SK 3798.
Rule. The pūrvapada sarva "all" is antodātta when it expresses the totality of a quality. The Subodhinī defines this: no other quality is present.
Example.
sarva + śvetaḥ → sarva expresses that the whiteness is total
→ the final of sarva udātta (SK 3826) → sarváśvetaḥ "white all over", udātta on va.
sarvámahān "great in every respect", udātta on va.
The Subodhinī calls sarvaśvetaḥ a karmadhāraya by SK 726, the whiteness pervading all the parts.
Applies when / blocked by. Three counter-examples. No accent is stated for them.
paramaśvetaḥ "supremely white". Here too the whiteness is total, since it pervades its substrate. But the word is not sarva.sarvasauvarṇaḥ "made entirely of gold". The totality is of the product, not a quality.sarvaśvetaḥ = sarveṣāṃ śvetataraḥ "whiter than all".Watch out. The same form sarvaśvetaḥ has two meanings. "White all over" takes the rule. "Whiter than all" does not. The Subodhinī explains the second. It is a sixth-case compound with a word in -tara, the tara being dropped (a vārttika cited under SK 703). There the totality is of the people who have the quality, not of the quality.
Rule. In a name (saṃjñā), before giri "mountain" or nikāya "group", the pūrvapada is antodātta.
Example.
añjanā + giriḥ → a name → final of the pūrvapada udātta → añjanā́giriḥ, udātta on nā.
The long ā of añjanā- comes from SK 1038, as the Subodhinī notes.
mauṇḍi + nikāyaḥ → mauṇḍínikāyaḥ, udātta on ḍi.
Applies when / blocked by. Not names, so no rule: paramagiriḥ "the highest mountain", brāhmaṇanikāyaḥ "a body of brāhmaṇas".
Rule. Before kumārī "maiden", when an age is conveyed, the pūrvapada is antodātta.
Example.
vṛddha + kumārī → an age ("old") is conveyed → final of the pūrvapada udātta
→ vṛddhákumārī "an old maiden", udātta on dha.
Applies when / blocked by. Counter-example: paramakumārī "an excellent maiden". Here no other age is conveyed, only maidenhood.
Watch out. How can a "maiden" be "old"? Bhaṭṭoji answers. Here kumārī means only "not joined to a man". In that sense it can stand in apposition with vṛddha. The age that counts is the one the other word names, not maidenhood itself.
The Subodhinī fills this out. kumārī normally names the first age, and its feminine ī is taught for that (SK 478). But it has a second ground of use, being unmarried. When only that second ground is used, vṛddha can qualify it. The feminine ī does not lapse, because it was added earlier. The Subodhinī calls vṛddhakumārī a compound of qualifier and qualified, with vṛddhā shortened to vṛddha by SK 746.
Rule. Before udaka "water", in a compound meaning a mixture, the pūrvapada is antodātta. Bhaṭṭoji glosses akevala "not pure" as miśra "mixed".
Example.
guḍa + udakam "water mixed with jaggery"
→ the final a of guḍa is udātta (this rule); the u of udaka is anudātta
→ the two vowels merge into o. The accent rule applies first, then the merger.
→ a single substitute for an udātta and an anudātta is udātta (SK 3658) → guḍódakam, udātta on ḍo.
→ or, because the anudātta stood at the head of a pada, the o is svarita (SK 3659) → guḍòdakam, independent svarita on ḍo.
Applies when / blocked by. Not a mixture: śītodakam "cold water". The Subodhinī calls it a karmadhāraya.
Watch out. The Subodhinī fixes the word division as a-kevale. Otherwise Pāṇini would have said kevala udake.
Rule. When a dvigu is the uttarapada, in the name of a sacrifice (kratu), the pūrvapada is antodātta. A dvigu is a compound whose first member is a numeral, such as trirātra "three nights".
Example.
garga + trirātraḥ "the three-night rite of the Gargas"
→ trirātra is a dvigu, and the whole names a rite
→ final of garga udātta → gargátrirātraḥ, udātta on ga (the second syllable).
The Subodhinī calls it a sixth-case compound. It derives trirātra with the compound-final affix of SK 787.
Applies when / blocked by.
atirātraḥ. The Subodhinī calls it a compound with ati, "a rite that has passed beyond the night".bilvasaptarātraḥ, the seven nights of the bilva offering.Rule. Before sabhā "assembly", in a neuter compound, the pūrvapada is antodātta.
Example.
gopāla + sabhā → neuter by SK 827 → gopālasabham → final of gopāla udātta
→ gopālásabham "an assembly of cowherds", udātta on la.
strī + sabhā → strī́sabham "an assembly of women", udātta on strī.
Applies when / blocked by.
brāhmaṇasenam "an army of brāhmaṇas".rājasabhā "a king's assembly hall".Watch out. Only the neuter that is taught for sabhā by name counts (pratipadokta, SK 826–827). So the rule does not reach ramaṇīyasabhaṃ brāhmaṇakulam "a brāhmaṇa family whose hall is pleasant". The Subodhinī calls ramaṇīyasabha a bahuvrīhi. It is neuter only because the noun it qualifies, kula, is neuter.
devadatta + puram → devadattápuramUnder the same heading antaḥ, four sūtras deal with pura "town" and bila "opening". Two give the accent, one widens it, one blocks it. The running example is devadattápuram, udātta on ta, the last syllable of devadatta.
Rule. Before pura, in the eastern country, the pūrvapada is antodātta. The Subodhinī reads prācām as a region, not as the view of eastern grammarians.
Example.
devadatta + puram → a town in the east → final of the pūrvapada udātta
→ devadattápuram, udātta on ta.
nāndī + puram → nāndī́puram, udātta on dī.
Applies when / blocked by. Not eastern: śivapuram. Bhaṭṭoji gives no gloss of his own here.
Rule. Before pura, in a compound that begins with ariṣṭa or gauḍa, the pūrvapada is antodātta. pura is carried from SK 3833.
Example.
ariṣṭa + puram → ariṣṭápuram, udātta on ṣṭa.
gauḍa + puram → gauḍápuram, udātta on ḍa.
Applies when / blocked by. Why does the sūtra say pūrva "first"? So that it also applies here:
ariṣṭāśrita + puram → ariṣṭāśritápuram, udātta on ta.
gauḍabhṛtya + puram → gauḍabhṛtyápuram, udātta on tya.
The Subodhinī explains. With pūrva the condition is a bahuvrīhi, "a compound in which ariṣṭa or gauḍa comes first". So the pūrvapada may be longer, and its final is udātta. Without pūrva, ariṣṭa would be separated from pura by āśrita, and the rule would fail.
Watch out. The udātta falls on the last syllable of the whole pūrvapada, not on ariṣṭa itself.
Rule. Before pura, hāstina, phalaka and mārdeya are not antodātta. This blocks SK 3833.
Example. No accent is stated for these forms:
hāstinapuram, phalakapuram, mārdeyapuram.
Applies when / blocked by. Bhaṭṭoji explains mārdeya. It is "the offspring of Mṛdi", with the affix ḍhak, because the word is in the śubhrādi list (SK 1126). The Subodhinī's derivation instead assumes the stem mṛdu; scholars discuss this.
Watch out. The rule says only what does not happen. Which accent these compounds then take is not stated here.
Rule. Before bila "opening", the pūrvapadas kusūla "granary", kūpa "well", kumbha "jar" and śāla are antodātta.
Example.
kusūla + bilam → kusūlábilam "the opening of a granary", udātta on la.
kūpábilam "the opening of a well", udātta on pa.
kumbhábilam "the opening of a jar", udātta on bha.
śālábilam, udātta on la.
Applies when / blocked by.
sarpabilam "a snake's hole".kusūlasvāmī "the owner of the granary".pūrva + pañcālāḥ → pūrvápañcālāḥThree sūtras make a direction word antodātta, such as pūrva "eastern, earlier" or apara "western, later". Each names the uttarapada it must stand before. The third also brings in sarva. The running example is pūrvápañcālāḥ "the eastern Pañcālas", udātta on va, the last syllable of pūrva.
Rule. A direction word is antodātta before a word for a village, a country or a narrative, or before the word cānarāṭa. For cānarāṭa the form itself is meant. For the others, any word with that meaning counts (Subodhinī).
Example.
Before village names (compounds by SK 727):
pūrva + iṣukāmaśamī → final a of pūrva udātta → a + i merge into e
→ udātta by SK 3658: pūrvéṣukāmaśamī, udātta on ve; or svarita by SK 3659: pūrvèṣukāmaśamī.
aparákṛṣṇamṛttikā, udātta on ra.
Before a country: pūrvápañcālāḥ, udātta on va. The Subodhinī calls it a karmadhāraya by SK 737. pañcāla here names part of the country.
Before a narrative: pūrváyātam, pūrváyāyātam, udātta on va. yāyāta is a work about Yayāti.
Before cānarāṭa: pūrvácānarāṭam, udātta on va.
Applies when / blocked by. Why does the sūtra say śabda "word"? To take in a direction word used of time. pūrva in pūrvayāyātam means "earlier", not "eastern".
Watch out. In pūrvéṣukāmaśamī the text does not say which of the two forms, udātta or svarita, is meant. Both follow from the rules.
Rule. Direction words are also antodātta before a word for a pupil, in which the teacher is subordinate. ca "also" carries "direction words" from SK 3837.
Example.
pūrva + pāṇinīyāḥ "the earlier Pāṇinīyas"
→ pāṇinīya names a pupil, with the teacher Pāṇini subordinate within it
→ final of pūrva udātta → pūrvápāṇinīyāḥ, udātta on va.
The Subodhinī calls it a karmadhāraya by SK 737.
Applies when / blocked by.
pūrvāntevāsī "an earlier pupil".pūrvapāṇinīyaṃ śāstram "the ancient Pāṇinian treatise". Here pūrva means "ancient".Watch out. Bhaṭṭoji gives no reason why pūrvāntevāsī fails. It is read here as follows: antevāsin names a pupil with no teacher in the word, so no teacher is subordinate.
Rule. Before an uttarapada whose vṛddhi is taught under the heading uttarapadasya (SK 1396), sarva and the direction words are antodātta.
Example.
sarvapañcāla + vuñ → the affix vuñ (SK 1349) → vṛddhi of the later member (SK 1398)
→ sarvapāñcālaka → final of sarva udātta → sarvápāñcālakaḥ, udātta on va.
aparápāñcālakaḥ (vṛddhi by SK 1399), udātta on ra.
Applies when / blocked by. Why name the heading? To keep out sarvabhāsaḥ and sarvakārakaḥ. The ā of kāraka is vṛddhi by SK 254, not by a rule under that heading.
Watch out. The word uttarapada in the sūtra points to the heading uttarapadasya, as the Subodhinī explains. "Any later member with vṛddhi" would wrongly take in sarvakārakaḥ.
viśva + karman → viśvákarmāA bahuvrīhi is a compound that describes something outside itself, such as śuklakarṇa "white-eared" (SK 3846). In a bahuvrīhi the pūrvapada normally keeps its own accent (SK 3735). These five rules are exceptions (apavāda): they make the pūrvapada antodātta instead. SK 3840 also opens a heading bahuvrīhau "in a bahuvrīhi", which runs up to the avyayībhāva rule (SK 3855).
The running example is viśva. Alone it is ādyudātta, víśva, because its affix kvan is marked with n (Subodhinī). In the name viśvákarmā the udātta moves to śva, the last syllable of viśva.
Rule. In a bahuvrīhi that is a name, the pūrvapada viśva is antodātta. The Subodhinī says "name" qualifies the compound, not the word viśva.
Example.
viśva + karman → a bahuvrīhi, and a name
→ by SK 3735 viśva would keep its own udātta on vi
→ this rule overrides that: final of viśva udātta → viśvákarmā, udātta on śva.
viśvádevaḥ, udātta on śva.
In the Veda: ā́ viśvádevaṃ sátpatim. Here ā is udātta, viśvádevam has its udātta on śva, and sátpatim on sa.
Applies when / blocked by.
viśvadevāḥ "they are all, and they are gods". Bhaṭṭoji's analysis makes it a karmadhāraya. No accent is stated.víśvadevaḥ, udātta on vi. The pūrvapada keeps its own accent by SK 3735.Watch out. Bhaṭṭoji adds: "The bahuvrīhi heading runs up to the avyayībhāva." So SK 3841–3854 all concern bahuvrīhis.
Rule. Before udara "belly", aśva "horse" or iṣu "arrow", in a bahuvrīhi that is a name, the pūrvapada is antodātta. Bhaṭṭoji says "in a name" continues from SK 3840.
Example.
vṛka + udaraḥ "whose belly is like a wolf's"
→ final a of vṛka udātta → a + u merge into o, udātta by SK 3658
→ vṛkódaraḥ, udātta on ko.
hari + aśvaḥ → final i of hari udātta
→ i becomes y before the vowel (SK 47)
→ the a after a y that replaces an udātta becomes svarita (SK 3657)
→ haryàśvaḥ, independent svarita on the a after y.
mahéṣuḥ: the merged e is udātta, on he.
Applies when / blocked by. The Subodhinī says this too is an exception to SK 3735. vṛka alone is ādyudātta by a phiṭ-sūtra (phiṭ 30); hari alone is ādyudātta too.
Watch out. The Subodhinī objects to mahéṣuḥ. mahat is antodātta anyway, so SK 3735 would already give this accent. It suggests suvarṇapuṅkheṣuḥ as the true example. A merged vowel at the head of the uttarapada may also be svarita by SK 3659.
Rule. Before udara, aśva or iṣu, in a bahuvrīhi expressing censure (kṣepa), the pūrvapada is antodātta. Here no name is required. Bhaṭṭoji glosses kṣepa as nindā "blame".
Example.
ghaṭa + udaraḥ "pot-bellied" → ghaṭódaraḥ, udātta on the merged ṭo.
kaṭuka + aśvaḥ → kaṭukā́śvaḥ, udātta on the merged kā.
calācala + iṣuḥ "one whose arrows waver" → calācaléṣuḥ, udātta on the merged le.
Applies when / blocked by. In anudaraḥ "bellyless" this rule meets nañsubhyām (SK 3906). That rule makes the uttarapada antodātta after the negative a(n). Each rule has its own field, as the Subodhinī says: this one in ghaṭodaraḥ, SK 3906 in ayaśaḥ. Where both reach the same form there is vipratiṣedha, a conflict, and SK 3906 applies: anudaráḥ, udātta on raḥ.
Watch out. Censure does not let this rule win over SK 3906. Here vipratiṣedha applies, so the later rule, SK 3906, prevails: anudaráḥ.
Rule. Before bandhu "kin", in a bahuvrīhi, a pūrvapada that ends in a nadī vowel is antodātta. nadī is the name of a feminine long ī or ū (SK 266).
Example.
gārgī + bandhuḥ → gārgī ends in feminine ī, a nadī
→ alone gārgī is ādyudātta, gā́rgī, by the accent of its ñ-marked affix (SK 3686)
→ this rule overrides that: final of gārgī udātta → gārgī́bandhuḥ, udātta on gī.
Applies when / blocked by.
bráhmabandhuḥ, udātta on bra. brahman is ādyudātta by its n-marked affix manin (Subodhinī), and keeps that accent.gā́rgīpriyaḥ, udātta on gā. The pūrvapada keeps its own accent (SK 3735).Rule. In a bahuvrīhi, a pūrvapada that ends in a niṣṭhā affix and begins with an upasarga is optionally antodātta. The niṣṭhā affixes are kta and ktavatu. An upasarga is a prefix such as pra.
Example.
pra + dhauta + pādaḥ "one whose feet are washed clean"
→ pradhauta ends in kta and begins with pra
→ in one option, final of pradhauta udātta → pradhautápādaḥ, udātta on ta.
→ in the other option it keeps the accent of the prefix (Subodhinī) → prádhautapādaḥ, udātta on pra.
Applies when / blocked by.
prasevakamukhaḥ. prasevaka ends in ṇvul.śúṣkamukhaḥ "dry-faced", udātta on śu. śuṣka is ādyudātta by SK 3694 and keeps it.Watch out. Bhaṭṭoji's gloss says only "a pūrvapada ending in a niṣṭhā". The upasarga condition is in the sūtra, and his second counter-example tests it.
śukla + karṇaḥ → śuklakárṇaḥA third heading moves the udātta into the later member. Under uttarapadādiḥ the first syllable of the uttarapada becomes udātta. The bahuvrīhi heading of SK 3840 still runs. The words named are body-parts: karṇa "ear", kaṇṭha "throat", pṛṣṭha "back", grīvā "neck", jaṅghā "shank" and śṛṅga "horn".
The running example is śuklakárṇaḥ "white-eared". The udātta is on kar, the first syllable of karṇa.
Rule. This is a heading. From here on, the first syllable of the uttarapada is udātta.
Applies when / blocked by. Bhaṭṭoji sets two limits.
The heading antaḥ of SK 3826 ends here.
Watch out. The Subodhinī takes uttarapada as a separate word with its sixth-case ending dropped: "of the uttarapada, the first". That is why its two parts can have different limits. "udātta" is supplied from Bhaṭṭoji's glosses on the next rules ("karṇa is ādyudātta").
Rule. In a bahuvrīhi, karṇa after a word for a colour or a mark is ādyudātta.
Example.
śukla + karṇaḥ "white-eared" → śukla names a colour
→ first syllable of the uttarapada udātta → śuklakárṇaḥ, udātta on kar.
śaṅku + karṇaḥ "with a peg-mark on the ear" → śaṅkukárṇaḥ, udātta on kar.
Applies when / blocked by. The Subodhinī explains the "mark". It is a brand, such as one shaped like a peg, put on cattle's ears to tell them apart. It must mean this, since colour is named separately.
śvetapādaḥ.śobhanakarṇaḥ.In these counter-examples the pūrvapada keeps its own accent (Subodhinī).
Watch out. By SK 1036 one would expect śaṅkūkarṇaḥ, with long ū, as the Subodhinī's remark implies. The short u is kept as transmitted.
Rule. karṇa is also ādyudātta in a bahuvrīhi that is a name or expresses a comparison. karṇa is carried from SK 3846.
Example.
Name: maṇi + karṇaḥ → maṇikárṇaḥ, udātta on kar.
Comparison: go + karṇaḥ "cow-eared", one whose ears are like a cow's → gokárṇaḥ, udātta on kar.
Applies when / blocked by. ca "also" adds these two conditions to the colour-or-mark condition of SK 3846.
Rule. kaṇṭha, pṛṣṭha, grīvā and jaṅghā as uttarapada are ādyudātta in a bahuvrīhi that is a name or a comparison. "Name or comparison" is carried from SK 3847.
Example.
Names:
śitikáṇṭhaḥ, udātta on kaṇ.
kāṇḍapṛ́ṣṭhaḥ, udātta on pṛ.
sugrī́vaḥ, udātta on grī.
nāḍījáṅghaḥ, udātta on jaṅ.
Comparisons:
kharakáṇṭhaḥ "donkey-throated", udātta on kaṇ.
gopṛ́ṣṭhaḥ "cow-backed", udātta on pṛ.
aśvagrī́vaḥ "horse-necked", udātta on grī.
gojáṅghaḥ "cow-shanked", udātta on jaṅ.
Applies when / blocked by. The Subodhinī notes the starting accents. kaṇṭha and pṛṣṭha are ādyudātta alone (phiṭ 29). grīvā and jaṅghā do not end in a, so alone they are antodātta. The rule gives all four an udātta on the first syllable.
Watch out. The Subodhinī doubts sugrīvaḥ. After su, the later rule SK 3906 ought to make the compound antodātta. Then this rule would have its field in words like daśagrīva. The example is kept as Bhaṭṭoji gives it.
Rule. In a bahuvrīhi, śṛṅga "horn" is ādyudātta when a stage of life is meant. It is also ādyudātta in a name or a comparison, by ca.
Example.
Stage of life:
udgataśṛ́ṅgaḥ "one whose horns have come up", udātta on śṛṅ.
dvyaṅgulaśṛ́ṅgaḥ "one whose horns are two fingers long", udātta on śṛṅ.
Name: ṛṣyaśṛ́ṅgaḥ, udātta on śṛṅ.
Comparison: meṣaśṛ́ṅgaḥ "with horns like a ram's", udātta on śṛṅ.
Applies when / blocked by. Bhaṭṭoji defines the stage of life. It is the age of a cow or like animal, marked by the sprouting of its horns. Not a stage of life: sthūlaśṛṅgaḥ "thick-horned".
na me jarā → a + jara → ajáram; su + karman → sukármāṇaḥnañ is the negative particle a(n) "not". su means "good, well". By nañsubhyām (SK 3906), a bahuvrīhi after nañ or su has its uttarapada antodātta. The five rules here are exceptions to that. They make the uttarapada ādyudātta instead.
The running examples are ajáram "ageless", udātta on ja, and sukármāṇaḥ "skilled in work", udātta on kar.
Rule. After nañ, in a bahuvrīhi, the uttarapadas jara, mara, mitra and mṛta are ādyudātta.
Example.
na me jarā "I have no old age" → a + jara → first syllable of the uttarapada udātta → ajáram, udātta on ja.
amáram, udātta on ma.
amítram ardaya "harass the enemy": udātta on mi.
śrávo devéṣv amṛtam "immortal fame among the gods". By the rule, amṛ́tam, udātta on mṛ.
Applies when / blocked by. The Subodhinī calls this an exception to SK 3906.
brāhmaṇamitraḥ.aśatrúḥ. By SK 3906 the uttarapada is antodātta, udātta on truḥ (Subodhinī).Watch out. In the Vedic quotation the accent marks put no udātta on mṛ, though the rule requires it. The quotation is given unmarked on amṛtam, and the rule's form beside it. This is not resolved.
Rule. After su, in a bahuvrīhi, an uttarapada ending in man or as is ādyudātta. loman "hair" and uṣas "dawn" are excluded. This is an exception to SK 3906.
Example.
su + karman → karman ends in man → first syllable of the uttarapada udātta
→ sukármāṇaḥ suyújaḥ "skilled in work, well yoked": udātta on kar.
sá no vakṣad animānáḥ suváhmā: in suvahmā the udātta is on vah.
śivā́ paśúbhyaḥ sumánāḥ suvárcāḥ "kindly to the cattle, well-minded, splendid": udātta on man and on var.
supéśasas karati "will make well adorned": udātta on pe.
Applies when / blocked by.
kṛtakarmā.surā́jā, udātta on rā by the accent marks. rājan is formed with kanin, not man (Subodhinī). Bhaṭṭoji does not say how this counter-example is accented.sulomā, sūṣāḥ.Watch out. With the compound-final affix kap, kapi pūrvam (SK 3907) applies instead, "because it is later" (paratvāt). That rule makes the syllable just before kap udātta, not su:
sukarma + ka → sukarmákaḥ, udātta on ma.
susrotas + ka → susrotáskaḥ, udātta on tas.
The Subodhinī adds that sukarmāṇaḥ would already be ādyudātta by SK 3853. This rule is still needed for longer words such as suprathimā, and outside the Veda.
Rule. kratu and the rest are also ādyudātta after su, in a bahuvrīhi. su is carried from SK 3851, as Bhaṭṭoji's gloss shows.
Example.
sā́mrājyāya sukrátuḥ "of good resolve, for sovereignty": in sukratuḥ, udātta on kra. In sāmrājyāya the accent marks put the udātta on sā.
suprátīkaḥ "of fair appearance", udātta on pra.
suhávyaḥ "well invoked", udātta on ha.
suprátūrtim anehásam "very swift, unassailable": in supratūrtim, udātta on pra.
Applies when / blocked by. The list beginning with kratu is not given in the text.
Rule. In the Veda, after su, in a bahuvrīhi, an uttarapada that is already ādyudātta and has two vowels stays ādyudātta.
Example.
ádhā sváśvāḥ "then, having good horses": aśva is ādyudātta, so in svaśvāḥ the udātta is on the a after sv.
suráthā̃ ātithigvé "with good chariots, for Atithigva": in surathā̃, udātta on ra.
Bhaṭṭoji explains: aśva and ratha are ādyudātta by the accent of an n-marked affix. The Subodhinī names the affixes, kvan and kthan.
Applies when / blocked by.
yā́ subāhúḥ "she who has fair arms". bāhu is antodātta (Subodhinī). The accent marks put the udātta on the final húḥ, as SK 3906 gives.sugúr asat suhiraṇyáḥ "he shall be rich in cows, rich in gold". Bhaṭṭoji notes that hiraṇya has three vowels. The accent marks put the udātta on the final ṇyaḥ.Watch out. "After su" and "bahuvrīhi" are not in the sūtra. They come from SK 3851 and SK 3840, as Bhaṭṭoji's gloss shows. "In the Veda" is in the sūtra itself.
Rule. In the Veda, after su, in a bahuvrīhi, vīra "hero" and vīrya "heroism" are also ādyudātta.
Example.
suvī́reṇa rayíṇā "with wealth in good heroes": udātta on vī.
suvī́ryasya gómataḥ "of wealth of good heroism, rich in cattle": udātta on vī.
Applies when / blocked by. Bhaṭṭoji explains why vīrya is named. vīrya ends in the affix yat. Naming it is a jñāpaka, an indication drawn from the wording. It shows that yato 'nāvaḥ (SK 3701) does not make vīrya ādyudātta. If it did, "the preceding rule" would already give suvī́ryasya.
Watch out. "The preceding rule" is read here as SK 3853. vīrya has two vowels, so SK 3853 would cover it if it were ādyudātta.
upa + kūlam → upakū́lamThe bahuvrīhi heading has ended. uttarapadādiḥ (SK 3845) still runs: the first syllable of the uttarapada is udātta. Two rules name other compound types. An avyayībhāva is an indeclinable compound, usually with an indeclinable first, such as upakūlam "near the bank". A dvigu has a numeral first, such as dvikaṃsa "bought for two kaṃsas".
The running example is upakū́lam, udātta on kū, the first syllable of kūla.
Rule. In an avyayībhāva, the uttarapadas kūla, tīra, tūla, mūla, śālā, akṣa and sama are ādyudātta. Bhaṭṭoji gives only examples; the Subodhinī states the rule.
Example.
upa + kūlam → an avyayībhāva → first syllable of the uttarapada udātta → upakū́lam, udātta on kū.
upatī́ram, udātta on tī.
upatū́lam, udātta on tū.
upamū́lam, udātta on mū.
upaśā́lam, udātta on śā.
upa + akṣam → the a of akṣa is udātta → a + a merge into ā → upā́kṣam, udātta on pā.
suṣámam, niḥṣámam, udātta on ṣa.
Applies when / blocked by.
upakumbham.paramakūlam.Watch out. In suṣamam the s of sama has become ṣ. The Subodhinī says that change is asiddha, treated as not yet done. So the word still counts as sama. Bhaṭṭoji adds that these forms are in the list beginning with tiṣṭhadgu. The Subodhinī says that is how they get the name avyayībhāva.
Rule. In a dvigu, the uttarapadas kaṃsa, mantha, śūrpa, pāyya and kāṇḍa are ādyudātta.
Example.
dvi + kaṃsaḥ "bought for two kaṃsas" → a dvigu → first syllable of the uttarapada udātta
→ dvikáṃsaḥ, udātta on kaṃ.
dvimánthaḥ, udātta on man.
dviśū́rpaḥ, udātta on śūr.
dvipā́yyam, udātta on pāy.
dvikā́ṇḍam "measuring two kāṇḍas", udātta on kāṇ.
Applies when / blocked by. Not a dvigu: paramakaṃsaḥ.
Watch out. These dvigus are made in a taddhita sense, such as "bought for". The Subodhinī says the taddhita affix is then dropped by luk in each. So the uttarapada looks bare, though an affix stood there.
brāhmaṇa + śālā → brāhmaṇaśā́lamA tatpuruṣa is a compound whose later member is the main one, such as brāhmaṇaśālā "a hall of brāhmaṇas". From here to SK 3873, except SK 3872, the rules concern tatpuruṣas. The Subodhinī says tatpuruṣe of SK 3857 is stated "for what follows". It continues even across rules that do not repeat it.
Three rules deal with neuter tatpuruṣas. Two make śālā "hall" and kanthā ādyudātta. The third moves the udātta back to the pūrvapada. The running example is brāhmaṇaśā́lam, udātta on śā.
Rule. In a neuter tatpuruṣa ending in śālā, the uttarapada is ādyudātta.
Example.
brāhmaṇa + śālā → a tatpuruṣa, neuter by SK 828 (Subodhinī) → brāhmaṇaśālam
→ first syllable of the uttarapada udātta → brāhmaṇaśā́lam "a hall of brāhmaṇas", udātta on śā.
Applies when / blocked by.
dṛḍhaśālaṃ brāhmaṇakulam "a brāhmaṇa household whose hall is firm". This is a bahuvrīhi. The pūrvapada keeps its own accent, and dṛḍha, a niṣṭhā form, is antodātta (Subodhinī).brāhmaṇasenam.brāhmaṇaśālā.Watch out. The Subodhinī raises a doubt. The bahuvrīhi would fail anyway, since only a specifically taught neuter counts (compare SK 3832). It answers: the word tatpuruṣa is needed for the rules that follow. Being stated here, it makes that other argument unnecessary.
Rule. kanthā as uttarapada is also ādyudātta in a neuter tatpuruṣa. "Neuter tatpuruṣa" is carried from SK 3857, as Bhaṭṭoji's gloss shows.
Example.
sauśami + kanthā "the kanthā of Sauśami" → a neuter tatpuruṣa → sauśamikántham, udātta on kan.
āhvarakántham, udātta on kan.
The Subodhinī makes them neuter by saṃjñāyāṃ kanthośīnareṣu, a rule for names, so they are place names.
Applies when / blocked by. Not neuter: dākṣikanthā.
Rule. In a neuter tatpuruṣa ending in kanthā, the first syllable of the pūrvapada cihaṇa and the rest is udātta.
Example.
cihaṇa + kanthā → a neuter tatpuruṣa → first syllable of the pūrvapada udātta
→ cíhaṇakantham, udātta on ci.
mándurakantham, udātta on man.
Applies when / blocked by. Bhaṭṭoji explains the restated ādi. "First" is already current from SK 3845. It is said again so that the pūrvapada becomes ādyudātta.
Watch out. Up to SK 3858 the "first syllable" belonged to the uttarapada. Here, and only here, it belongs to the pūrvapada (Subodhinī).
putra + celam → putracélamThis group has eleven rules. Ten of them make a tatpuruṣa uttarapada ādyudātta; the eleventh (SK 3867) blocks SK 3866. Each sets a condition: contempt, comparison, mixture, a name, or a certain kind of pūrvapada. The group closes the ādi heading at SK 3870.
The running example is putracélam "a son no better than a rag". The udātta is on ce, the first syllable of cela.
Rule. When contempt is expressed, the uttarapadas cela, kheṭa, kaṭuka and kāṇḍa are ādyudātta.
Example.
putra + celam "a son (no better than) a rag" → contempt → first syllable of the uttarapada udātta
→ putracélam, udātta on ce.
nāgarakhéṭam "a town (as feeble as) grass", udātta on khe.
dadhikáṭukam "curds (as tasteless as) something gone off", udātta on ka.
prajākā́ṇḍam "offspring (as painful as) an arrow", udātta on kā.
Applies when / blocked by. Bhaṭṭoji explains the contempt. It lies in the son's likeness to a rag, and so on. The compound is formed because these words are in the vyāghrādi list (SK 735). No contempt: paramacelam "the best garment".
Watch out. The Subodhinī's glosses give the images: kheṭa is a name for grass, and kāṇḍa for an arrow.
Rule. cīra "rag" is ādyudātta when it is the standard of comparison (upamāna).
Example.
vastraṃ cīram iva "a garment like a rag" → vastracī́ram, udātta on cī.
kambalacī́ram "a blanket like a rag", udātta on cī.
The Subodhinī says the compound is by the vyāghrādi rule, as before (SK 735).
Applies when / blocked by. Not a standard of comparison: paramacīram "the best rag".
Rule. When a mixture is expressed, the uttarapadas palala, sūpa "broth" and śāka "vegetables" are ādyudātta. The Subodhinī glosses palala as meat.
Example.
ghṛtena miśraṃ palalam "palala mixed with ghee" → ghṛtapálalam, udātta on pa.
ghṛtasū́paḥ "broth mixed with ghee", udātta on sū.
ghṛtaśā́kam "vegetables mixed with ghee", udātta on śā.
Bhaṭṭoji says the compound is by bhakṣyeṇa miśrīkaraṇam (SK 697).
Applies when / blocked by. Not a mixture: paramapalalam "the best palala".
Rule. In a tatpuruṣa that is a name, the uttarapadas kūla, sūda, sthala and karṣa are ādyudātta.
Example. Bhaṭṭoji says these are names of villages.
dākṣikū́lam, udātta on kū.
śāṇḍisū́dam, udātta on sū.
dāṇḍāyanasthálam, udātta on stha.
dākṣikárṣaḥ, udātta on ka.
Applies when / blocked by. Not a name: paramakūlam "the best bank".
Watch out. The Subodhinī reads dāṇḍāyanasthalī, with feminine sthalī. It says the feminine counts as sthala by a paribhāṣā, a rule about how other rules are read. The form sthalam is kept here.
Rule. In a tatpuruṣa that is not a karmadhāraya, the uttarapada rājya "kingdom" is ādyudātta. A karmadhāraya is a tatpuruṣa whose two members refer to the same thing, such as paramarājya "supreme kingdom".
Example.
brāhmaṇa + rājyam "the kingdom of the brāhmaṇas" → not a karmadhāraya
→ brāhmaṇarā́jyam, udātta on rā.
Applies when / blocked by. A karmadhāraya: paramarājyam "the supreme kingdom".
Vārttika. The accent of an indeclinable prevails over the accent of cela, rājya and the rest. This is by pūrvavipratiṣedha, where the earlier rule wins a conflict.
kucelam "a wretched garment", kurājyam "a wretched kingdom".
The Subodhinī explains. The indeclinable accent is given by SK 3736. Each rule has its own field: SK 3736 in niṣkauśāmbiḥ, the cela/rājya rules in bhāryācelam, brāhmaṇarājyam. In kucela and kurājya both apply, and the earlier wins. So ku keeps its own accent, and the uttarapada is not made ādyudātta.
Watch out. "And the rest" in the vārttika also takes in vargya and the like of SK 3865 (Subodhinī). Bhaṭṭoji does not mark the accent of kucelam, kurājyam.
Rule. vargya and the rest are also ādyudātta as uttarapada in a tatpuruṣa that is not a karmadhāraya. "Not a karmadhāraya" holds here too, as Bhaṭṭoji says.
Example.
arjuna + vargyaḥ "one of Arjuna's party" → arjunavárgyaḥ, udātta on var.
vāsudevapákṣyaḥ "one of Vāsudeva's side", udātta on pak.
Applies when / blocked by. A karmadhāraya: paramavargyaḥ.
Watch out. The Subodhinī notes that there is no separate list "vargya and the rest". That is why Bhaṭṭoji adds: vargādi is a sub-list inside the digādi list, whose words take yat (SK 1429). So vargya, pakṣya are varga, pakṣa with yat.
Rule. In a tatpuruṣa, putra "son" after a word for a male is ādyudātta.
Example.
dāśaki + putraḥ "the son of Dāśaki" → dāśaki names a male
→ dāśakipútraḥ, udātta on pu.
māhiṣapútraḥ "the son of Māhiṣa", udātta on pu.
Applies when / blocked by.
kaunaṭimātulaḥ "the maternal uncle of Kaunaṭi".dākṣīputraḥ "the son of Dākṣī". dākṣī is feminine.Rule. putra is not ādyudātta after words naming a teacher, a king, an officiating priest, a relation by marriage, or a kinsman. putra is carried from SK 3866.
Example. Bhaṭṭoji gives no accent for these:
Teacher: ācāryaputraḥ, upādhyāyaputraḥ, śākaṭāyanaputraḥ.
King: rājaputraḥ, īśvaraputraḥ, nandaputraḥ.
Priest: ṛtvikputraḥ, yājakaputraḥ, hotuḥputraḥ.
Relation by marriage: śyālaputraḥ "the son of the wife's brother".
Kinsman: jñātiputraḥ, bhrātuṣputraḥ "the brother's son".
Applies when / blocked by. Because the sūtra says "names of" (ākhyā), synonyms and particular kinds are included. The Subodhinī joins ākhyā to each of the five words. Bhaṭṭoji defines two terms. saṃyukta are relations by marriage. jñāti are kinsmen through mother and father.
Watch out. The rule only cancels SK 3866. Which accent the compound then takes is not stated. In hotuḥputraḥ the sixth-case ending is kept by SK 981. In bhrātuṣputraḥ the ṣ comes from SK 144 (Subodhinī).
Rule. In a tatpuruṣa, cūrṇa "flour" and the rest are ādyudātta after a sixth-case word naming something not living.
Example.
mudgānāṃ cūrṇam → mudga "a bean" is not a living being
→ mudgacū́rṇam "mudga-bean flour", udātta on cū.
Applies when / blocked by. A living being: matsyacūrṇam "fish meal".
Rule. The six words kāṇḍa, cīra, palala, sūpa, śāka and kūla are also ādyudātta after a sixth-case word naming something not living. That condition is carried from SK 3868.
Example.
darbhakā́ṇḍam "a stalk of darbha grass", udātta on kā.
darbhacī́ram "a strip of darbha", udātta on cī.
tilapálalam "sesame palala", udātta on pa.
mudgasū́paḥ "mudga-bean broth", udātta on sū.
mūlakaśā́kam "radish greens", udātta on śā.
nadīkū́lam "the bank of a river", udātta on kū.
Applies when / blocked by.
rājasūdaḥ "the king's cook".dattakāṇḍam.Watch out. The Subodhinī counts the six from kāṇḍa (SK 3860) to kūla (SK 3863). Here they are ādyudātta without their earlier conditions. kāṇḍa needs no contempt, cīra no comparison, the three foods no mixture, and kūla no name.
Rule. In a tatpuruṣa denoting a forest, kuṇḍa is ādyudātta.
Example.
darbha + kuṇḍam → used of a forest → darbhakúṇḍam "a darbha thicket", udātta on ku.
Bhaṭṭoji says kuṇḍa here is used by resemblance.
Applies when / blocked by. The literal sense: mṛtkuṇḍam "a clay pot".
Watch out. This is the last rule of the ādi heading of SK 3845. From SK 3871 the uttarapada keeps its own accent instead.
kumbhī + bhagālam → kumbhībhagā́lamA last heading, prakṛtyā "in its original form", runs from SK 3871 into Part VII. Here no new udātta is given. A member keeps the accent it had before it entered the compound, and the other member loses its accent. In SK 3874 both members keep their accents.
The running example is kumbhībhagā́lam. bhagāla keeps its own udātta on the middle syllable gā. kumbhī loses its accent.
Rule. In a tatpuruṣa, an uttarapada that means bhagāla keeps its own accent.
Example.
kumbhī + bhagālam → bhagāla keeps its own accent, udātta on the middle syllable
→ kumbhībhagā́lam, udātta on gā.
kumbhīnadā́lam, udātta on dā.
Bhaṭṭoji says: "These are madhyodātta", with the udātta on the middle syllable. The Subodhinī derives that accent from phiṭ 42.
Applies when / blocked by. The Subodhinī takes bhagāla by its meaning, not its form. So any word with that meaning counts. For this it cites the commentarial explanation alone. Bhaṭṭoji adds that prakṛtyā is a heading up to antaḥ (SK 3877).
Watch out. Bhaṭṭoji's third example, kumbhīpālam, cannot be madhyodātta, since pāla has two syllables. Scholars discuss this example; the form is left unmarked here.
Rule. In a bahuvrīhi after śiti, a colour word, an uttarapada that never has many vowels keeps its own accent. bhasat is excluded.
Example.
śiti + pādaḥ → pāda never has more than two vowels
→ pāda keeps its own ādyudātta, from the vṛṣādi list (SK 3691)
→ śitipā́daḥ, udātta on pā.
śiti + aṃsaḥ → aṃsa is ādyudātta by its n-marked affix → śityáṃsaḥ, udātta on tyaṃ, the a of aṃsa.
Applies when / blocked by. Bhaṭṭoji calls this an exception to SK 3735, where the pūrvapada keeps its accent.
darśanī́yapādaḥ. SK 3735 applies, and darśanīya keeps its udātta on nī (SK 3733, Subodhinī).śítibhasat. śiti keeps its own ādyudātta, on śi (phiṭ 33, Subodhinī).Watch out. This one rule is a bahuvrīhi in the middle of tatpuruṣa rules. Its "bahuvrīhi" does not carry on to SK 3873 (Subodhinī).
Rule. In a tatpuruṣa, an uttarapada ending in a kṛt affix keeps its own accent after a gati, a kāraka or an upapada. A gati is a prefix such as pra joined to a verb. A kāraka is a participant in the action, such as its object. An upapada is a word a kṛt rule names beside the root.
Example.
After a gati:
pra + kārakaḥ → kāraka keeps the accent of its l-marked affix ṇvul: the syllable before the affix → prakā́rakaḥ, udātta on kā.
praháraṇam, udātta on ha.
After a kāraka:
śóṇā dhṛṣṇū́ nṛvā́hasā "the two bays, bold, carrying men". nṛ is the object in "carrying men"; vāhas keeps its udātta on vā.
idhmapravráścanaḥ, udātta on vra.
After an upapada:
uccaiḥkā́ram (SK 3381), udātta on kā.
īṣatkáraḥ, udātta on ka.
In each the pūrvapada loses its accent.
Applies when / blocked by. Not after these: devasya kārakaḥ. Bhaṭṭoji says its sixth case marks a residual relation (śeṣa). The Subodhinī adds that it is not the sixth case of the object. If it were, deva would be a kāraka, and SK 709 would forbid the compound too.
Bhaṭṭoji's gloss says "in a tatpuruṣa". The Subodhinī explains that the word is carried from SK 3857 across the rules in between. It also says the "bahuvrīhi" of SK 3872 does not continue, because it is not marked svarita, the sign of a heading.
Watch out. Is the word kṛt needed? Bhaṭṭoji reports two views.
prapacatitarām the compound is made with the -tara form, and ām is added afterwards. The accent of ām prevails, being taught when the compound accent already exists (satiśiṣṭa).prapacatideśya and the like. These then take the accent of the indeclinable pūrvapada (SK 3736), as the Subodhinī explains.The Subodhinī reports further discussion; scholars discuss this.
Rule. In vanaspati and the rest, both members at once keep their own accent. So each compound has two udāttas.
Example.
vánaspátiṃ vána ā́ "the lord of the wood, in the wood": in vanaspatim, udātta on va and on spa.
bṛ́haspátiṃ yáḥ: udātta on bṛ and on pa. bṛhat is made ādyudātta here by nipātana, a form fixed by the list.
harṣayā śácīpátim "gladden the lord of Śacī": udātta on śa and on pa. śacī is ādyudātta as a śārṅgaravādi word (SK 527), as śácībhir naḥ shows.
tánūnápād ucyate "Tanūnapāt is invoked": udātta on ta and on na.
nárāśáṃsaṃ vājínam "Narāśaṃsa, the strong": udātta on na and on śaṃ. The long ā is by nipātana.
śúnaḥśépam: udātta on śu and on śe.
Applies when / blocked by. The Subodhinī gives the sources of each accent. vana is ādyudātta by phiṭ 26, and pati by its affix. tanū and napāt are made ādyudātta by nipātana. nara is nṝ with ap (SK 3232), and śaṃsa is śaṃs with ghañ; both are ādyudātta. In śunaḥśepa both members are ādyudātta, śepa by phiṭ 29.
Watch out. SK 3650 normally leaves only one udātta in a pada. Here the rule itself prescribes two, by the word yugapat "at once".
índra + váruṇa → índrāváruṇau (two udāttas, í and vá); against indra + agní → indrāgníbhyām (one udātta, on gní)By SK 3734 a compound normally has one udātta, on its final syllable. The heading prakṛtyā (SK 3871) lets a member keep its own original accent instead. Its last two rules are about the dvandva, the compound of co-ordinated members, "X and Y". When the members are names of deities, both keep their own udātta at the same time. That stops when the second member begins with an anudātta syllable, with four listed exceptions. The running example is índrāváruṇau "Indra and Varuṇa", with two udātta syllables, í and vá. Against it stands indrāgníbhyām, whose only udātta is gní.
Rule. In a dvandva of deity names, both members keep their original accent at once. "Both at once" (ubhe yugapat) is carried forward from SK 3874. "Keep their original accent" (prakṛtyā) is the heading of SK 3871.
Example.
indra (affix ran, uṇādi 186: udātta on the first syllable, í) + varuṇa (affix unan, uṇādi 333: udātta on the first syllable, va)
→ the first member takes ānaṅ (SK 922) → indrā-varuṇa
→ both members keep their udātta → ā́ yá índrāváruṇau.
In the compound the udātta syllables are í and vá. The rest is unaccented: ru is the dependent svarita after vá (SK 3660), and the others are anudātta.
A second mantra: índrābṛ́haspátī vayám. bṛhaspati belongs to the vanaspati group of SK 3874, so it already has two udāttas of its own. The compound therefore has three udātta syllables: í, bṛ́ and pá (the syllable spa). The Subodhinī counts the three.
Applies when / blocked by. "Deity": plakṣanyagrodhau "the plakṣa and the nyagrodha" names two trees, not deities, so the rule does not apply. "Dvandva": agniṣṭomaḥ is not a dvandva, so again it does not apply. Bhaṭṭoji does not state the accent of these two.
Watch out. The svarita on ru, after the udātta vá, is only a dependent svarita (SK 3660). It is not a third accent, and the two udātta syllables stay í and vá.
Rule. The double accent of SK 3875 does not hold when the uttarapada begins with an anudātta syllable. The uttarapada is the final member of a compound. The exceptions are pṛthivī, rudra, pūṣan and manthin: after these the double accent stays.
Example.
indra + agni (affix ni, uṇādi 490: udātta on the last syllable, so the first syllable a is anudātta)
→ the uttarapada begins with an anudātta → no double accent
→ indrāgníbhyāṃ káṃ vṛ́ṣaṇaḥ. The compound has one udātta, gní.
The excepted words, which keep both accents:
dyā́vāpṛthivī́ janáyan "producing heaven and earth". dyāvā is made ādyudātta by nipātana, so dyā́ is udātta. pṛthivī is antodātta, so vī́ is udātta.sómārudráu. rudra ends in the affix rak (uṇādi 179) and is antodātta. The udātta syllables are só and drá.índrāpūṣáṇau. pūṣan is antodātta by nipātana (uṇādi 157). The udātta syllables are í and ṣá.śukrāmanthínau. manthin ends in in and is antodātta, so thí is udātta. Bhaṭṭoji does not state the accent of śukra.Applies when / blocked by. All four excepted words are antodātta, so each begins with an anudātta. Without the exception, this rule would remove their double accent.
Watch out. Why does the sūtra say "uttarapada"? In this section the dvandva itself is named in the seventh case. Without the word, "beginning with an anudātta" would describe the whole dvandva. The prohibition would then wrongly reach candrāsūryau "moon and sun", whose first member candra is antodātta (Subodhinī). The phrase "beginning with an anudātta" also divides the ground: SK 3875 takes the rest, this rule takes these cases. The Subodhinī adds that the exclusion of pṛthivī and the rest shows a maxim: in accent rules a consonant counts as absent.
pra + bhid + ghañ → prabhedáḥ (udātta on the final dá, where the kṛt accent of SK 3873 would have kept ñit bhéda ādyudātta)Now a new heading begins: antaḥ, "the final syllable is udātta". Every rule from here to the end of this stretch makes the final syllable of the uttarapada udātta. The first rules work in a tatpuruṣa, a determinative compound. They apply after a gati (a prefix construed with a verb), a kāraka (a word for a role in the action) or an upapada (a word beside the root). Normally the kṛt member there keeps its own accent by SK 3873. The running example is prabhedáḥ "splitting". bheda has the ñ-marked affix ghañ, so by SK 3686 it would be udātta on its first syllable. This rule moves the udātta to the last syllable, dá.
Rule. This is an adhikāra, a heading. It supplies "the final syllable [of the uttarapada] is udātta" to the rules that follow. Bhaṭṭoji says only "this is an adhikāra".
Example. A heading has no example of its own. Its first application is SK 3878: pra + bhid + ghañ → prabhedáḥ, udātta on the final syllable dá.
Applies when / blocked by. At SK 3871 Bhaṭṭoji said that the heading prakṛtyā runs "up to antaḥ". So this rule closes the prakṛtyā section. The words "of the uttarapada" come from Bhaṭṭoji's own gloss at SK 3881, "the uttarapada is antodātta". Antodātta means "having the udātta on the final syllable".
Watch out. Do not confuse this antaḥ with two later sūtras. SK 3913 is also antaḥ, but there it is the word antar "within". SK 3914 antaś ca is the word anta "end".
Rule. After a gati, a kāraka or an upapada, an uttarapada ending in tha, atha, ghañ, kta, ac, ap, itra or ka is antodātta. "After a gati, a kāraka or an upapada" carries on from SK 3873.
Example.
pra + bhid + ghañ → pra-bheda → the final syllable of the uttarapada udātta → prabhedáḥ (udātta dá).
One example for each ending, with the udātta on the final syllable of the uttarapada:
prabhṛthásyāyóḥ. In prabhṛtha the udātta is thá. bhṛtha has the affix kthan (uṇādi 159–160). Being n-marked, it would otherwise be ādyudātta (Subodhinī).āvasatháḥ (thá); the affix atha is uṇādi 396.prabhedáḥ (dá).dhartā́ vajrī́ puruṣṭutáḥ "the upholder, the vajra-bearer, praised in many places". In puruṣṭuta the udātta is tá.prakṣayáḥ (yá), with ac by SK 3231.pralaváḥ (vá), with ap by SK 3232.pralavitrám (trá).govṛṣáḥ (ṣá). ka comes because vṛṣa is in the mūlavibhujādi list.Applies when / blocked by. Bhaṭṭoji repeats "only after a gati, a kāraka or an upapada". In sustutaṃ bhavatā "well praised by you", su is a karmapravacanīya, not a gati (SK 555). So this rule does not apply. The Subodhinī says only the accent of the indeclinable first member applies there.
Watch out. Two examples are chosen with care, the Subodhinī explains. puruṣṭuta is analysed as puruṣu stutaḥ "praised among many", with a seventh case. A third-case analysis would bring in tṛtīyā karmaṇi (SK 3782). Then the first member would keep its accent. And kṣaya "dwelling" is ādyudātta by SK 3689, so prakṣayáḥ really needs this rule.
Rule. After su "well", and after an upamāna (a standard of comparison), an uttarapada ending in kta is antodātta.
Example.
su + kṛta → the final of the kta-member udātta → sukṛtá:
ṛtásya yónau sukṛtásya "in the seat of ṛta, of the well-made". In this line the udātta syllables are tá, yó and tá; in sukṛtá the udātta is the final tá.
śaśa iva plutam "leapt like a hare" → śaśaplutáḥ (udātta tá).
Applies when / blocked by. The Subodhinī names the rules set aside. In sukṛta, gatir anantaraḥ (SK 3783) would give the first member's accent. In śaśapluta, tṛtīyā karmaṇi (SK 3782) would do the same.
Rule. In a name (saṃjñā), an uttarapada ending in kta after a gati, a kāraka or an upapada is antodātta. ācita and the rest are excepted.
Example.
upa + hūta (hve + kta) → in a name → upahūtáḥ śākalyaḥ "Śākalya called Upahūta" (udātta tá).
pari + jagdha (jagdh replacing ad) → parijagdháḥ kauṇḍinyaḥ "Kauṇḍinya called Parijagdha" (udātta dhá).
Applies when / blocked by. Here too the Subodhinī says gatir anantaraḥ (SK 3783) would otherwise apply. The excepted words are ācitam and āsthāpitam. Bhaṭṭoji does not state their accent.
Rule. In pravṛddha and the rest, the uttarapada ending in kta is antodātta.
Example.
pra + vṛddha → pravṛddháḥ (udātta ddhá).
pra + yuta → prayutáḥ (udātta tá).
Applies when / blocked by. Bhaṭṭoji says this rule is undertaken for words that are not names. In a name, SK 3880 already gives the result (Subodhinī). The list pravṛddhādi is an ākṛtigaṇa, a list defined by its pattern and left open.
deva + datta (blessing name) → devadattáḥ (udātta ttá); māṣa + ūna → māṣonám (udātta nám)These rules still give the final udātta to the uttarapada. Now the condition is narrower: the first member is a kāraka, or a word in a particular case. Several rules are exceptions to SK 3782 (tṛtīyā karmaṇi) or SK 3736, which would let the first member keep its accent. The running examples are devadattáḥ "god-given", a name given in blessing, with the udātta on ttá, and māṣonám "short by a māṣa", with the udātta on nám.
Rule. In a tatpuruṣa that is a name given in blessing, after a kāraka only datta and śruta are antodātta. This is a niyama, a restricting rule. It limits what SK 3880 would give. "In a name" carries on from SK 3880. "In a tatpuruṣa" and "ending in kta" are supplied by the Subodhinī.
Example.
deva + datta, a child prayed for with "may the gods give him" → devadattáḥ (udātta ttá).
viṣṇu + śruta "heard by Viṣṇu" → viṣṇuśrutáḥ (udātta tá).
Applies when / blocked by. Each word of the sūtra has a purpose.
saṃbhūtó rāmāyaṇaḥ. sam is a gati, not a kāraka, so the restriction does not touch it. It then stays antodātta, presumably by SK 3880. The Subodhinī says that without "kāraka" the restriction would reach gatis too, since the heading of SK 3873 names all three.devapālitaḥ. By this restriction SK 3880 does not apply. tṛtīyā karmaṇi (SK 3782) applies instead.saṃśrutáḥ (udātta tá).devaiḥ khātā → devakhātā "dug by the gods" takes SK 3782.Watch out. The restriction means "only in a blessing". So in anāhato nadati devadattaḥ "Devadatta sounds without being struck", the final is not udātta. Here devadatta is the name of a conch, not a blessing. The first member keeps its own accent by SK 3782.
Rule. In a compound meaning "done by one in such a state", the final member ending in kta is antodātta. It is an apavāda, an exception, to tṛtīyā karmaṇi (SK 3782).
Example.
supta + pralapita, "the babbling of one asleep" → suptapralapitám (udātta tám).
pramatta + gīta, "the singing of one intoxicated" → pramattagītám (udātta tám).
Applies when / blocked by. The Subodhinī: an itthambhūta is one who has come into such a condition. The sense is "done by him".
Watch out. Is babbling "done"? Bhaṭṭoji answers that kṛ in kṛtam means action in general. It does not mean only bringing into being what did not exist. So babbling and singing also count as "done".
Rule. After a kāraka, a final member ending in the affix ana (lyuṭ) is antodātta when it denotes the action itself (bhāva) or the object (karman). "After a kāraka" carries on from SK 3882. The Subodhinī calls the rule an exception to SK 3873.
Example.
payas + pāna (lyuṭ for the action; an upapada compound by SK 3291) → payaḥpānáṃ sukham "drinking milk is pleasant" (udātta nám).
rājan + bhojana (lyuṭ for the object) → rājabhojanā́ḥ śālayaḥ "rice to be eaten by kings" (udātta nā́).
Applies when / blocked by.
hastādāyaḥ.dantadhāvanam "tooth-cleaner". Here lyuṭ denotes the instrument.nidarśanam.Bhaṭṭoji does not state the accent of these three.
Rule. After a kāraka, in a tatpuruṣa, these final members are antodātta: words in man and in ktin, and vyākhyāna, śayana, āsana, sthāna, yājaka and the rest, and krīta. Bhaṭṭoji calls it an exception to the kṛt accent (SK 3873). For krīta, the Subodhinī says it overrides SK 3782 instead.
Example. Each compound has its udātta on the final syllable:
rathavartmá "chariot-track" (vartman with manin).pāṇinikṛtíḥ "Pāṇini's work".chandovyākhyānám "explanation of the metres".rājaśayanám "the king's bed"; rājāsanám "the king's seat"; aśvasthānám "horse-stand".brāhmaṇayājakáḥ "sacrificer for brāhmaṇas". yājakādi means the list of SK 703.gokrītáḥ "bought with a cow".Applies when / blocked by. Not after a kāraka: prabhūtau saṅgatim. There the accent is by tādau ca niti kṛty atau (SK 3784), and the first member keeps its accent.
Rule. puṇya after a seventh-case word is antodātta, in a tatpuruṣa.
Example.
adhyayane puṇyam "meritorious in study" → compound by splitting off saptamī from SK 717 (Subodhinī) → adhyayanapuṇyám (udātta yám).
Applies when / blocked by. Without this rule, tatpuruṣe tulyārtha- (SK 3736) would apply. Not after a seventh case: vedena puṇyam → vedapuṇyam "meritorious through the Veda". Its accent is not stated.
Rule. After a third-case word, final members meaning "deficient" (ūna), and kalaha "quarrel", are antodātta. It is an exception to the rule that a third-case first member keeps its own accent. The compounds themselves are made by SK 693 (Subodhinī).
Example.
māṣa + ūna → māṣonám "short by a māṣa" (udātta nám).
māṣavikalám "wanting by a māṣa" (udātta lám).
vākkalaháḥ "a quarrel with words" (udātta há).
Applies when / blocked by. Some (kecit) read artha as the word artha itself: dhānyārtháḥ (udātta thá). On their view the word ūna stands for its meaning, so all words meaning "deficient" are taken.
Watch out. Bhaṭṭoji says the word "third case" in the sūtra is only for clarity. These compounds are specially taught with a third-case first member, so that much is already known.
Rule. After a third-case word, miśra "mixed" without an upasarga is antodātta. This does not hold where sandhi, an alliance, is meant. Bhaṭṭoji: sandhi is becoming one in purpose through a pact. "After a third-case word" carries on from SK 3887.
Example.
tila + miśra → tilamiśrā́ḥ "mixed with sesame" (udātta śrā́).
sarpirmiśrā́ḥ "mixed with ghee" (udātta śrā́).
Applies when / blocked by.
guḍadhānāḥ "grains with jaggery".tilasaṃmiśrāḥ.brāhmaṇamiśro rājā "a king allied with brāhmaṇas", one who has joined them by a pact.Their accent is not stated.
Watch out. The exclusion "without an upasarga" is itself a jñāpaka, a hint. A jñāpaka is a detail of wording from which an unstated principle is inferred. Here it shows that where SK 693 mentions miśra, miśra with an upasarga is also included. The Subodhinī reports a second view: sandhi is the perception that the joined things stay distinct.
na kārṇaveṣṭakikam → akārṇaveṣṭakikám (udātta kám); na pacaḥ (unable) → apacáḥ (udātta cáḥ)nañ is the negative particle, which appears as a- or an- at the head of a compound. In a tatpuruṣa with nañ, the particle normally keeps its own accent, the udātta á-. The seven rules of this group move the udātta to the final syllable of the uttarapada instead. Each names a kind of uttarapada, or a sense such as inability, reviling or a name used in abuse. The running examples are akārṇaveṣṭakikám "a face not suited to earrings", with the udātta on the final kám, and apacáḥ "one unable to cook", with the udātta on the final cáḥ.
Rule. After nañ that negates a quality, a final member ending in a taddhita is antodātta. A taddhita is an affix added to a noun stem. Its sense must be "suited to" (sampādin), "deserving" (arha), "good for" (hita) or "able for" (alam).
Example.
karṇaveṣṭakābhyāṃ sampādi "suited to earrings" → ṭhañ → kārṇaveṣṭakikam
→ na kārṇaveṣṭakikam → akārṇaveṣṭakikám (udātta kám): "a face not suited to earrings".
chedam arhati "deserves cutting" → ṭhak → chaidikaḥ → acchaidikáḥ (udātta káḥ).
na vatsebhyo hitaḥ "not good for calves" → cha → avatsīyáḥ (udātta yáḥ).
na santāpāya prabhavati "not able for affliction" → ṭhañ → asāntāpikáḥ (udātta káḥ).
The Subodhinī names the affix in each case.
Applies when / blocked by.
gārdabharathikaḥ "deserving a donkey-cart"; dvigārdabharathikaḥ. The Subodhinī reads the second as vigārdabharathikaḥ, whose indeclinable first member keeps its accent.agārdabharathikaḥ "one other than a gārdabharathika". Here a- means "other than".apāṇinīyaḥ "not a student of Pāṇini's work".avoḍhā "not fit to carry", whose affix tṛc is a kṛt.Bhaṭṭoji does not state their accent.
Watch out. What is the "quality" here? Bhaṭṭoji says it is the ground of application (pravṛttinimitta) of the taddhita's meaning: being suited, being deserving, and so on. The rule applies where that quality is denied. In akārṇaveṣṭakikám the face is denied "suitedness to earrings". In agārdabharathikaḥ nothing is denied; another person is meant.
Rule. After such a nañ, a final member ending in the taddhita affix ya or yat is antodātta. This holds when the affix does not mean "for that" (tadartha). "After nañ negating a quality" carries on from SK 3889.
Example.
pāśānāṃ samūhaḥ "a collection of nooses" → ya → pāśyā → na pāśyā → apāśyā́ (udātta yā́).
danta + yat "belonging to the teeth" → dantya → adantyám (udātta yám).
Applies when / blocked by.
apādyam. Its yat is in the sense "for that" (Subodhinī).adeyam. Here yat is the kṛt affix after dā.dantyād anyad adantyam "other than what belongs to the teeth".Their accent is not stated.
Watch out. avāmadevyam does not take this rule. Its affix is ḍya, which carries the marker ḍ. Bhaṭṭoji cites a maxim: when an affix with certain markers is named, one with other markers is not included. So ḍya does not count as ya.
Rule. After nañ, a final member ending in ac or ka is antodātta when inability is meant. Bhaṭṭoji adds that "in the negation of a quality" still holds.
Example.
na + paca (ac, the pacādi affix) "not able to cook" → apacáḥ (udātta cáḥ).
na + vilikha (ka) → avilikháḥ (udātta kháḥ).
Applies when / blocked by.
apaco dīkṣitaḥ "the consecrated man does not cook". He is able, but the śāstra forbids it (Subodhinī).anyo 'yaṃ pacād apacaḥ "this one, other than a cook, is an apaca".Their accent is not stated.
Rule. The same final members, in ac or ka after nañ, are also antodātta in reviling (ākrośa).
Example.
apacáḥ + jālmaḥ → apacó jālmaḥ "the wretch is a non-cook" (udātta có, which is cáḥ changed in sandhi).
avikṣipáḥ (udātta páḥ).
Watch out. This is not about real inability. Bhaṭṭoji: he is reviled thus, "he cannot cook". The Subodhinī adds that he can in fact cook. The insult is the condition.
Rule. In a name, when abuse is meant, the uttarapada after nañ is antodātta. "After nañ" carries on from SK 3889; "in reviling" carries on from SK 3892.
Example.
na + devadatta → adevadattáḥ (udātta ttá): "a non-Devadatta", said in reproach.
Applies when / blocked by. The Subodhinī: the man is Devadatta, but he does not do the deed expected of him. So he is reproached in this way.
Rule. After nañ, these final members are antodātta: words ending in a kṛtya affix, in ukañ or in iṣṇuc, and the words of the cārvādi list. A kṛtya is an affix of the gerundive, "to be done".
Example.
akartavyáḥ "not to be done" (udātta yáḥ).anāgāmukáḥ (udātta káḥ); ukañ by SK 3134.analaṅkariṣṇúḥ (udātta ṣṇúḥ); iṣṇuc by SK 3116.acārúḥ, asādhúḥ (udātta rúḥ, dhúḥ).Applies when / blocked by. khiṣṇuc is also included under iṣṇuc. It has two markers, kh and c. But it is taught as beginning with i, and that has a purpose only if it counts here. Example: anāḍhyambhaviṣṇúḥ (udātta ṣṇúḥ), with khiṣṇuc by SK 2974.
The cārvādi list contains a gaṇasūtra, a rule placed inside a list: rājāhnoś chandasi, "rājan and ahan, in the Veda". So in the Veda: arājā́ (udātta jā́) and anāháḥ (udātta háḥ). arājā takes no compound-final affix, by SK 956 (Subodhinī).
Watch out. In ordinary speech (bhāṣā), Bhaṭṭoji says, these two have only the accent of nañ. The udātta then presumably stays on a-, as in árājā.
Rule. After nañ, in a tatpuruṣa, these final members are optionally antodātta: words ending in tṛn, and anna, tīkṣṇa and śuci.
Example. In the option this rule gives, the final syllable is udātta:
akartā́ (udātta tā́), anannám (udātta nnám), atīkṣṇám (udātta kṣṇám), aśucí (udātta cí).
Applies when / blocked by. In the other option, Bhaṭṭoji says, the accent of the indeclinable comes. So nañ keeps its own udātta á-.
idaṃ prathamaṃ yasya → idaṃprathamáḥ (udātta má), and with kap idaṃprathamákaḥ (still má)A bahuvrīhi is a compound that describes something outside itself: "he whose first [act] is this". SK 3896 sets up a new heading, bahuvrīhau "in a bahuvrīhi". It runs to just before SK 3912. The final-syllable heading antaḥ still holds. So in each rule the uttarapada of a bahuvrīhi becomes antodātta. The running example is idaṃprathamáḥ "he for whom this act is the first", with the udātta on má, the final of prathama. With the compound-final affix kap (-ka) it is idaṃprathamákaḥ, and the udātta stays on má.
Rule. In a bahuvrīhi, after idam, etad or tad, the words prathama "first" and an ordinal (pūraṇa) are antodātta. This holds when acts are being counted.
Example.
idaṃ prathamaṃ yasya "he for whom this [act] is the first" → idaṃprathamáḥ (udātta má).
etaddvitīyáḥ (udātta yáḥ); tatpañcamáḥ (udātta máḥ).
The Subodhinī: the first act is, for instance, a going or an eating.
With kap:
idaṃprathama + kap → the uttarapada is the operand, so its final keeps the udātta → idaṃprathamákaḥ (udātta má, before ka).
Applies when / blocked by.
anena prathamaḥ "first by this [act]" → idaṃprathamaḥ, a tatpuruṣa made by splitting SK 692.yatprathamaḥ.tāni bahūny asya → tadbahuḥ "he who has many of those".ayaṃ prathamaḥ pradhānaṃ yeṣām → idaṃprathamāḥ "those among whom this one is the first, the leading one".idaṃprathamāḥ meaning idaṃpradhānāḥ "those whose chief is this one". Here prathama means "chief", not the number one.Their accent is not stated.
Watch out. With kap, the udātta does not move to ka. Bhaṭṭoji's reason: the uttarapada undergoes the operation (it is the kāryin). The Subodhinī adds: kap ends the compound, not the uttarapada. So the final of prathama, before kap, is udātta.
Rule. In a bahuvrīhi, stana "breast, teat" after a numeral is antodātta.
Example.
dvi + stana → final a of stana udātta → feminine ṅīṣ (SK 510, Subodhinī) → dvistanī́ (udātta nī́).
The udātta a is lost before ī, and its udātta presumably passes to the ī (SK 3651).
catur + stana → catuḥstanā́ (udātta nā́).
Applies when / blocked by.
darśanīyastanā. The Subodhinī: the first member keeps its own accent; darśanīya has the affix anīyar.dviśirāḥ "two-headed".Watch out. On catuḥstanā the Subodhinī disagrees with Bhaṭṭoji. It says catur keeps its own accent here, giving cátuḥstanā (udātta cá). On that view the word is not antodātta, so it takes no ṅīṣ. Bhaṭṭoji gives the form as an example of this rule.
Rule. In the Veda (chandas), the rule of SK 3897 is optional (vibhāṣā).
Example.
dvistanāṃ karoti "he makes [it] two-teated".
Bhaṭṭoji does not say which option this shows, so no accent is marked here.
Applies when / blocked by. In the Veda, stana after a numeral may be antodātta, or may not. The Subodhinī: what the previous sūtra stated is optional in the Veda.
Rule. In a bahuvrīhi that is a name, mitra and ajina as uttarapada are antodātta.
Example.
deva + mitra (a name) → devamitráḥ (udātta tráḥ).
kṛṣṇa + ajina (a name) → kṛṣṇājinám (udātta nám).
Applies when / blocked by. Not a name: priyamitraḥ. Its accent is not stated.
A vārttika, a supplementary rule, adds: ṛṣipratiṣedho mitre. With mitra, the accent is barred when a seer is named. So viśvāmitra ṛṣiḥ "the seer Viśvāmitra" does not take it. The Subodhinī explains. This rule, coming later, would set aside bahuvrīhau viśvaṃ saṃjñāyām (SK 3840). The vārttika stops that, so SK 3840 applies: viśvā́mitraḥ (udātta vā́). The long ā comes from SK 1049.
Rule. In a bahuvrīhi, antara after a word for what intervenes (a screen) is antodātta.
Example.
vastram antaraṃ vyavadhāyakaṃ yasya "he for whom a cloth is the screen" → vastrāntaráḥ (udātta ráḥ).
Applies when / blocked by. Not a screen: ātmāntaraḥ "of another nature". Here antara means "other" (Subodhinī). Its accent is not stated.
gaura + mukha → gauramukháḥ (udātta kháḥ), against go + mukha → gómukhaḥ (first member's own udātta gó, mukha unaccented)Still in the bahuvrīhi, five rules deal with two kinds of uttarapada. The first is mukha "face" when it is a body part. The second is a word in kta, the past participle, after certain first members. Where the rule does not apply, the first member keeps its own accent by bahuvrīhau prakṛtyā pūrvapadam (SK 3735). The running example is gauramukháḥ "fair-faced", with the udātta on the final kháḥ. Against it stands gómukhaḥ "cow-faced". There go keeps its own udātta, gó, and mukha has none.
Rule. In a bahuvrīhi, mukha is antodātta when it denotes a body part (svāṅga).
Example.
gaura + mukha → mukha a body part → gauramukháḥ (udātta kháḥ).
Applies when / blocked by. Not a body part: dīrghamukhā śālā "a hall with a long front". Here mukha is the door-area of the hall (Subodhinī). Its accent is not stated.
Watch out. svāṅga is taken in its technical definition, which begins "not liquid …". It does not simply mean "one's own limb" (svam aṅgam). Otherwise, the Subodhinī says, the rule would reach the hall as well.
Rule. SK 3901 does not apply after an avyaya (an indeclinable), after a word for a direction, or after go, mahat, sthūla, muṣṭi, pṛthu and vatsa. Then the first member keeps its own accent (pūrvapadaprakṛtisvara). The Subodhinī says every first member here is antodātta, by SK 3735.
Example. mukha is unaccented in each; the udātta is the final syllable of the first member:
uccaírmukhaḥ (udātta caí); uccaiḥ is from uṇādi 690.prā́ṅmukhaḥ (udātta prā́ṅ). pra is udātta by SK 3786, and the merged prā keeps it.gómukhaḥ (udātta gó).mahā́mukhaḥ (udātta hā́).sthūlámukhaḥ (udātta lá).muṣṭímukhaḥ (udātta ṭí).pṛthúmukhaḥ (udātta thú).vatsámukhaḥ (udātta sá).Applies when / blocked by. When go, muṣṭi or vatsa is a standard of comparison, SK 3903 would also offer an option. Bhaṭṭoji says this rule sets that option aside. The Subodhinī explains: both rules apply there, and the earlier one wins by pūrvavipratiṣedha. That is the principle that, in some conflicts, the earlier rule prevails.
Rule. In a bahuvrīhi, mukha as a body part is optionally antodātta after a word in a niṣṭhā affix, or after a word for a standard of comparison.
Example.
prakṣālita + mukha "one whose face is washed" → prakṣālitamukháḥ (udātta kháḥ).
siṃha + mukha "lion-faced" → siṃhamukháḥ (udātta kháḥ).
Applies when / blocked by. Bhaṭṭoji names the other forms of prakṣālitamukhaḥ. By SK 3844 the first member may be antodātta: the udātta falls on tá of prakṣālita. And through the first member's own accent, the gati accent on pra may also come (probably by SK 3735). For siṃhamukhaḥ, the other option keeps the first member's accent (Subodhinī).
Rule. In a bahuvrīhi, a final member in kta is antodātta after a word for a genus (jāti), a time (kāla), or sukha and the rest. It is not so after a word for a covering. And kṛta, mita and pratipanna are excepted.
Example.
sāraṅgo jagdho yena "he by whom a sāraṅga bird has been eaten" → sāraṅgajagdháḥ (udātta dháḥ).
māsajātáḥ, sukhajātáḥ, duḥkhajātáḥ (udātta táḥ in each).
Applies when / blocked by.
putrajātaḥ.vastracchannaḥ.kuṇḍakṛtaḥ, kuṇḍamitaḥ, kuṇḍapratipannaḥ.In all of these the first member keeps its own accent (Subodhinī). putra has the udātta on its last syllable. vastra and kuṇḍa have it on the first; kuṇḍa is so by phiṭ 26.
Watch out. Normally a niṣṭhā word stands first in a bahuvrīhi (SK 899). This sūtra speaks of the kta word as following. Bhaṭṭoji takes that as a jñāpaka: in such compounds the niṣṭhā word is placed second.
Rule. jāta after a word for a genus, a time, or sukha and the rest is optionally antodātta.
Example.
danta + jāta → dantajātáḥ (udātta táḥ).
māsa + jāta → māsajātáḥ (udātta táḥ).
Applies when / blocked by. In the other option the first member keeps its accent (Subodhinī). danta is ādyudātta by phiṭ 29: dántajātaḥ (udātta dán). māsa is ādyudātta by phiṭ 38: mā́sajātaḥ (udātta mā́).
Watch out. māsajātaḥ was also an example of SK 3904. This rule makes the accent of jāta optional, so both forms are correct.
na vrīhiḥ yasya → avrīhíḥ (udātta hí); with kap avrī́hikaḥ (udātta vrī́, the syllable before the last of vrīhi)In a bahuvrīhi after nañ or su, the uttarapada is antodātta. When the compound-final affix kap (-ka) is added, the udātta moves. After a long vowel it falls just before kap. After a short vowel it falls on the syllable before the last of the uttarapada. bahu "many" then borrows the treatment of nañ, except guṇa and the rest when they denote parts. The running example is avrīhíḥ "one who has no rice", udātta on the final hí. With kap it is avrī́hikaḥ, udātta on vrī́.
Rule. In a bahuvrīhi, after nañ and su, the uttarapada is antodātta.
Example.
na vrīhiḥ yasya "one who has no rice" → avrīhíḥ (udātta hí).
su + māṣa → sumāṣáḥ (udātta ṣáḥ).
Watch out. Does the rule apply to the uttarapada or to the whole compound? Scholars discuss this. The Subodhinī sets out both views. If compound-final affixes belong to the uttarapada, the added a of anṛcaḥ (SK 940) also becomes udātta. If they belong to the compound, the whole compound is the operand.
Rule. In such a bahuvrīhi, when kap follows, the syllable before kap is udātta. Bhaṭṭoji glosses: of the compound ending in the uttarapada that follows nañ or su.
Example.
a + brahmabandhū + kap (SK 833) → the vowel before kap udātta → abrahmabandhū́kaḥ (udātta dhū́).
su + kumārī + kap → sukumārī́kaḥ (udātta rī́).
Applies when / blocked by. Both examples end in long vowels. The Subodhinī: short vowels are taken by the next rule. Without this rule, SK 3906 would make kap itself udātta.
Rule. When the uttarapada, or the compound, ends in a short vowel, the syllable before the last is udātta. This is when kap follows, after nañ or su, in a bahuvrīhi. The short vowel is reckoned before kap is added (Subodhinī).
Example.
a + vrīhi + kap → vrīhi ends in short i → the syllable before the last, vrī, udātta → avrī́hikaḥ (udātta vrī́).
su + māṣa + kap → sumā́ṣakaḥ (udātta mā́).
Applies when / blocked by. "Before" (pūrvam) already runs on from SK 3907. Bhaṭṭoji says the new mention is a niyama. For a short-final member, only the syllable before the last is udātta. SK 3907 does not also apply.
Watch out. Take ajñakaḥ, "one who has no knowledge" (nāsti jñā yasya). Bhaṭṭoji: because of the niyama, SK 3907 does not apply here. So SK 3906 simply makes the compound with kap antodātta: ajñakáḥ (udātta káḥ).
Rule. In a bahuvrīhi, after bahu that expresses the plurality of what the uttarapada denotes, the following word takes the accent it would have after nañ.
Example.
bahu + vrīhi + kap → as after nañ, SK 3908 → bahuvrī́hikaḥ (udātta vrī́).
bahu + mitra + kap → bahumítrakaḥ (udātta mí).
Applies when / blocked by. Not the plurality of the uttarapada's meaning: bahuṣu māno 'sya "he who has honour among many" → bahumānaḥ. Here bahu keeps its own accent, the Subodhinī says. bahu is antodātta (uṇādi 29), so the udātta is on hu.
Watch out. "Plurality of the uttarapada" means plurality of its meaning. A word as such cannot be plural (Subodhinī).
Rule. After bahu, in a bahuvrīhi, guṇa and the rest are not antodātta when they denote parts.
Example.
bahuguṇā rajjuḥ "a rope of many strands".
bahvakṣaraṃ padam "a word of many syllables".
bahvadhyāyaḥ "[a work] of many chapters".
Bhaṭṭoji states only the negation, so no accent is marked.
Applies when / blocked by. Not parts: bahuguṇo dvijaḥ "a twice-born of many qualities". The qualities are study, learning, good conduct and the like. The list guṇādi is an ākṛtigaṇa.
pra + pṛṣṭha → prapṛṣṭháḥ (udātta ṣṭháḥ); pari + anta → paryantáḥ (udātta táḥ), against ni + anta → nyàntaḥ (independent svarita yà)From here the first member is an upasarga, a prefix such as pra, pari or ni. SK 3911 is the last rule of the bahuvrīhi heading. With SK 3912 the word samāse "in a compound" ends that heading, so the rules now hold in any compound. Every upasarga except abhi is udātta on its first syllable (phiṭ 81). Where these rules do not apply, the upasarga keeps that accent. The running examples are prapṛṣṭháḥ, with the udātta on the final ṣṭháḥ, and paryantáḥ, with the udātta on táḥ. Against them stands nyàntaḥ. There ni keeps its udātta, and the a after the y becomes an independent svarita, marked grave: yà.
Rule. In a bahuvrīhi, after an upasarga, a final member denoting a permanent (dhruva) body part is antodātta. parśu "rib" is excepted. Bhaṭṭoji: dhruva means "of one constant form".
Example.
pra + pṛṣṭha → prapṛṣṭháḥ (udātta ṣṭháḥ).
pra + lalāṭa → pralalāṭáḥ (udātta ṭáḥ).
Applies when / blocked by.
darśanīyapṛṣṭhaḥ.praśākho vṛkṣaḥ "a tree with spreading branches".udbāhuḥ "[one who cries out] with arms raised". The arms are raised only at the moment of crying out (Subodhinī).viparśuḥ.In these the first member keeps its own accent (Subodhinī). Their accent is not marked.
Watch out. A body part has no link with an action. So "upasarga" here just points to pra and the rest, the Subodhinī says.
Rule. In any compound, vana "forest" after an upasarga is antodātta. "After an upasarga" carries on from SK 3911.
Example.
pra + vana → n becomes ṇ by SK 1050 → pravaṇa
→ final syllable udātta → tásyédime pravaṇé.
In pravaṇé the udātta is the final syllable ṇé.
Applies when / blocked by. The Subodhinī: whether pravaṇa is a bahuvrīhi or a tatpuruṣa, the first member would otherwise keep its own accent.
Watch out. The word samāse "in a compound" is there to end the heading "in a bahuvrīhi" (Subodhinī). Bhaṭṭoji glosses it "in any compound whatever". At SK 3896 he had said that bahuvrīhau runs up to this rule.
Rule. vana after antar "within" is antodātta. "vana" and "in a compound" carry on from SK 3912.
Example.
antar + vana → ṇ by SK 1050 → antarvaṇáḥ deśaḥ "a region within the forest" (udātta ṇáḥ).
Applies when / blocked by. Bhaṭṭoji: the rule is for antar when it is not an upasarga. When antar is an upasarga, SK 3912 already gives the result (Subodhinī).
Watch out. The sūtra's antaḥ is the word antar, whose fifth-case ending has been dropped as a Vedic licence (Subodhinī). It is not the heading antaḥ of SK 3877.
Rule. In any compound, after an upasarga, the word anta "end" is antodātta.
Example.
pari + anta → i becomes y → paryantáḥ (udātta táḥ).
sam + anta → samantáḥ (udātta táḥ).
Applies when / blocked by. upasargāt "after an upasarga" is Bhaṭṭoji's gloss. samāse "in a compound" of SK 3912, which the Subodhinī says ends the bahuvrīhi heading, still holds here. The Subodhinī on SK 3915 agrees, since it allows both a bahuvrīhi and a tatpuruṣa with anta.
Watch out. This antaḥ is the word anta in the first case. It is not the antar of SK 3913.
Rule. After ni and vi, SK 3914 does not apply: anta is not antodātta.
Example.
ni + anta → ni keeps its own udātta (phiṭ 81); anta is anudātta
→ i becomes y (yaṇ), so the udātta i is gone
→ the anudātta a after it becomes svarita by SK 3657 → nyàntaḥ.
vi + anta → vyàntaḥ in the same way.
In each, the syllable yà carries an independent svarita, marked grave.
Applies when / blocked by. The first member keeps its own accent. The Subodhinī: in a bahuvrīhi this is by SK 3735, in a tatpuruṣa by SK 3736.
Watch out. The svarita here is independent. It is not the dependent svarita that follows an udātta. SK 3657 makes an anudātta vowel svarita when it follows a y or v that replaced an udātta vowel.
Rule. After pari, a final member for what exists on both sides, and maṇḍala "circle", are antodātta. Bhaṭṭoji: what exists on both sides is kūla "bank" and the like.
Example.
pari + kūla → parikūlám (udātta lám).
pari + maṇḍala → parimaṇḍalám (udātta lám).
Applies when / blocked by. The Subodhinī: parikūla may be a bahuvrīhi, a prādi compound, or an avyayībhāva "excluding the bank". A prādi compound has pra and the like as first member. An avyayībhāva is an indeclinable compound. In the first two the first member would keep its accent. In the avyayībhāva, SK 3767 and SK 3855 would apply. This rule sets all of these aside.
Rule. After pra, in a name, a final member that is not a body part is antodātta.
Example.
pra + gṛha → pragṛhám (udātta hám).
Applies when / blocked by. A body part: prapadam "the forepart of the foot". The Subodhinī: the first member keeps its own accent, so it is prápadam, udātta on prá (phiṭ 81).
Rule. nirudaka and the rest are antodātta.
Example.
nir + udaka "waterless" → nirudakám (udātta kám).
nir + upala "stoneless" → nirupalám (udātta lám).
Applies when / blocked by. The Subodhinī: nirudaka is a bahuvrīhi, "from which the water has gone", or a prādi compound. As an avyayībhāva, "absence of water", samāsasya (SK 3734) would already give the final udātta.
apa + mukha → apamukhám (udātta khám); anu + jyeṣṭha (prior member principal) → anujyeṣṭháḥ (udātta ṣṭháḥ)The last eight rules of this stretch each name one upasarga or particle and what may follow it. Each makes that final member antodātta, under a condition of sense. Several of them fill gaps that SK 3911 leaves: compounds that are not bahuvrīhis, body parts that are not permanent, and words that are not body parts. The running examples are apamukhám, with the udātta on the final khám, and anujyeṣṭháḥ "one who follows the eldest", with the udātta on ṣṭháḥ.
Rule. After abhi, mukha is antodātta.
Example.
abhi + mukha → abhimukhám (udātta khám).
abhimukhā́ śālā "a hall facing [something]" (udātta khā́).
Applies when / blocked by. SK 3911 would already give abhimukhám. Bhaṭṭoji says this rule is for compounds that are not bahuvrīhis, for mukha when not permanent, and for mukha when not a body part. The hall shows the last: there mukha means the front.
Watch out. Why does Bhaṭṭoji name SK 3911 and not SK 3901? The Subodhinī: SK 3901 cannot apply after an indeclinable, because SK 3902 forbids it.
Rule. After apa, mukha is antodātta, in a compound.
Example.
apa + mukha → apamukhám (udātta khám).
Applies when / blocked by. The Subodhinī: as a prādi compound or a bahuvrīhi, the first member would keep its accent. As an avyayībhāva, "excluding the face", SK 3767 would apply. This rule sets them aside.
Watch out. Why not one sūtra, "after abhi and apa"? Bhaṭṭoji: the split (yogavibhāga) is for the next rule. Only apa, not abhi, carries on into SK 3921.
Rule. After apa, these final members are antodātta: sphiga, pūta, vīṇā, añjas, adhvan, kukṣi, words for the plough, and nāman.
Example. Each compound has the udātta on its final syllable:
apasphigám (gám), apapūtám (tám), apavīṇám (ṇám), apāñjáḥ (ñjáḥ), apādhvā́ (dhvā́), apakukṣíḥ (kṣíḥ).
Words for the plough: apasīrám (rám), apahalám (lám).
apanāmá (má).
Applies when / blocked by. adhvan is named for the case where upasargād adhvanaḥ (SK 953) does not apply. When that compound-final ac is added, the final udātta already follows from citaḥ (SK 3710) (Subodhinī). So apādhvā is the form without ac. sphiga, pūta and kukṣi are named for non-bahuvrīhi compounds, impermanent uses, and non-body-part uses. Otherwise SK 3911 would already cover them.
Watch out. Bhaṭṭoji draws a jñāpaka from adhvan. If compound-final affixes were obligatory, naming adhvan would be pointless. So such affixes are anitya, not always applied.
Rule. After adhi, a final member for what stands above is antodātta.
Example.
adhyārūḍho dantaḥ "a tooth grown on top of a tooth" → adhidantáḥ (udātta táḥ).
Applies when / blocked by. Not standing above: adhikaraṇam "that in which something is placed". The Subodhinī: it denotes what is below. The a of kar is udātta by the accent of a kṛt final member: adhikáraṇam (udātta ká).
Rule. After anu, a final member that is not the principal sense, and kanīyas "younger", are antodātta.
Example.
anugato jyeṣṭham "one who follows the eldest" → anujyeṣṭháḥ (udātta ṣṭháḥ). It is a prādi compound in which the prior member's sense is principal.
anugataḥ kanīyān "the younger one, who follows" → anukanīyā́n (udātta yā́n). Here the final member's sense is principal.
Applies when / blocked by. kanīyas is named for the case where it is principal. When it is not principal, "non-principal" already covers it (Subodhinī). Counter-example: anugato jyeṣṭhaḥ "the eldest, who follows" → anujyeṣṭhaḥ. Here jyeṣṭha is principal, so the rule does not apply. Its accent is not stated.
Watch out. The same word anujyeṣṭha has two senses. Only the sense in which jyeṣṭha is subordinate takes the final udātta.
Rule. After anu, puruṣa is antodātta when it means "mentioned afterwards" (anvādiṣṭa). "After anu" carries on from SK 3923.
Example.
anvādiṣṭaḥ puruṣaḥ "the man mentioned afterwards" → anupuruṣáḥ (udātta ṣáḥ).
Applies when / blocked by. Not "mentioned afterwards": anugataḥ puruṣaḥ "the man who follows" → anupuruṣaḥ. Its accent is not stated. The Subodhinī: when one thing has been mentioned, another mentioned after it is anvādiṣṭa.
Rule. After ati, a final member not ending in a kṛt affix, and the word pada, are antodātta.
Example.
aṅkuśam atikrāntaḥ "gone beyond the goad" → atyaṅkuśáḥ nāgaḥ "an elephant beyond the goad" (udātta śáḥ).
atipadā́ gāyatrī "a gāyatrī that goes beyond the pada" (udātta dā́).
Applies when / blocked by. A kṛt member other than pada: atikāraḥ "an excellent maker". Its accent is not stated.
A vārttika corrects the condition: ater dhātulope iti vācyam. Instead of "not a kṛt", read "when a verbal root is elided". The Subodhinī explains: in atyaṅkuśa, ati stands for atikrānta, and the root kram is simply not said.
atigārgyaḥ "an excellent Gārgya". No root is elided, though gārgya is not a kṛt word.atikārukáḥ "one who has surpassed the artisan" (udātta káḥ). A root is elided, though kāru is a kṛt word.So the vārttika cures both over-coverage and under-coverage (Subodhinī).
Rule. After ni, the uttarapada is antodātta when the sense is not nidhāna. Bhaṭṭoji: nidhāna is concealment, non-manifestness. anidhāna is what is other than that, manifestation.
Example.
ni + mūla → nimūlám (udātta lám).
ni + akṣa → i becomes y → nyakṣám (udātta kṣám).
Applies when / blocked by. Concealment: nihito daṇḍaḥ "a stick put away" → nidaṇḍaḥ. Here ni expresses nidhāna. Its accent is not stated. The Subodhinī: inside a compound, pra and the like express the participant together with its action. That is how ni alone can mean "put away".
Watch out. The heading antaḥ does not end here. It runs on into the next rules, beginning with SK 3927 after prati.
prati + aṃśu (pratigato 'ṃśuḥ) → pratyaṃśúḥ (udātta on the final śú)These four sūtras are the last tatpuruṣa rules under the heading antaḥ (SK 3877). Under that heading the named uttarapada, the later member of the compound, has its final syllable udātta. tatpuruṣe, "in a tatpuruṣa", is named in SK 3927 and carried into SK 3928–3930. The running example is pratyaṃśúḥ, "one whose share has gone away".
prati + aṃśu → compound in the sense pratigato 'ṃśuḥ
→ ordinary sandhi: the i of prati becomes y before the vowel (SK 47) → pratyaṃśu
→ final syllable of aṃśu udātta (SK 3927, under antaḥ) → pratyaṃśú
→ with the ending su → pratyaṃśúḥ: the udātta is on śú, the last syllable.
Rule. In a tatpuruṣa, after prati, the word aṃśu and the others of its list (aṃśvādi) have the final syllable udātta. Such a word is called antodātta, "udātta at the end".
Example.
pratyaṃśúḥ (pratigato 'ṃśuḥ): the udātta is on śú.
pratijanáḥ: the udātta is on ná.
pratirājā́: the udātta is on jā́, the final syllable of the stem pratirājan, which appears as jā in the nominative.
Applies when / blocked by. Bhaṭṭoji adds: there is no ṭac in pratirājā́, because samāsāntas are non-obligatory. A samāsānta is an affix added at the end of a compound. The Subodhinī explains why the list proves this. ṭac by SK 788 is cit (marked with c), and a cit affix makes the compound antodātta by itself (SK 3710). If ṭac were always added, listing rājan here would be pointless. So the listing is a jñāpaka: a wording that proves an unstated point.
Rule. In a tatpuruṣa, after upa, an uttarapada of two vowels has its final syllable udātta. ajina "skin" does so too. The words of the list gaura and the rest (gaurādi) are excepted.
Example.
upa + deva → upadeváḥ: the udātta is on vá.
upa + indra → (a + i merge to e) → upendráḥ: the udātta is on drá.
upa + ajina → upājinám: the udātta is on nám. ajina is named separately in the sūtra.
Applies when / blocked by. Bhaṭṭoji's counter-examples. Why "except gaura and the rest"? upagauraḥ, upataiṣaḥ: these are in the list, so the rule does not apply. Why "in a tatpuruṣa"? upasomaḥ, "one whom soma has reached" (upagataḥ somo 'sya). That is a bahuvrīhi, a compound that names something outside its own members.
Rule. In a tatpuruṣa, after su, the uttarapada has its final syllable udātta when censure (avakṣepaṇa) is meant.
Example. su + pratyavasita → supratyavasitáḥ: the udātta is on tá, the last syllable before the ending.
Bhaṭṭoji explains the puzzle. su here means only praise. But the sentence as a whole means censure, because it is said out of spite. The Subodhinī adds that su is a karmapravacanīya by SK 555: a particle that marks a relation and is not a prefix of the verb. It also sets the scene: the words are said to someone who sits at ease when a misfortune has come.
Applies when / blocked by. Why "after su"? kubrāhmaṇaḥ "a bad brāhmaṇa": ku is not su. Why "when censure is meant"? suvṛṣaṇam: plain praise, so the rule does not apply.
Watch out. The Subodhinī says the thāthādi rule (SK 3878), which also makes an uttarapada antodātta, is not in question here. It gives no reason. Presumably su, as a karmapravacanīya, is not a gati, and that rule needs one. A gati is a word such as pra, joined to a verb and named "gati" by its own rules.
Rule. In the tatpuruṣa utpuccha, the member puccha after ud optionally has its final syllable udātta. vibhāṣā means "optionally".
Example. ud + puccha (utkrāntaḥ pucchāt, "gone out from the tail") → utpuccháḥ: the udātta is on chá. In the other option the final udātta does not come.
Applies when / blocked by. Bhaṭṭoji shows that the option works against two different starting points.
So "this is an option in both cases".
Watch out. Why "in a tatpuruṣa"? The bahuvrīhi utpucchaḥ, "one by whom the tail is raised" (udastaṃ pucchaṃ yena saḥ), does not take this rule.
dvi + pād (pāda, a lost, SK 879) → dvipā́t (udātta on the final pā́): dvipā́c cátuṣpāc caráthāyaThree sūtras close the compound accents. The first two still work under antaḥ, now in the bahuvrīhi. Both are options: vibhāṣā is carried from SK 3930 into SK 3931–3933, as the Subodhinī says at SK 3933. The last sūtra moves the udātta, in the Veda, to the first syllable of the later member. The running example is dvipā́t, "two-footed".
dvi + pāda → bahuvrīhi "that which has two feet"
→ the a of pāda is lost (SK 879) → dvipād
→ final syllable udātta, optionally (SK 3931) → dvipā́d
→ nominative singular → dvipā́t: the udātta is on pā́.
In the mantra dvipā́c cátuṣpāc caráthāya the udātta syllables are pā́, the first cá and rá.
Rule. In a bahuvrīhi, after dvi "two" and tri "three", when pād "foot", dat "tooth" or mūrdhan "head" follows, the compound optionally has its final syllable udātta.
Example.
dvipā́c cátuṣpāc caráthāya "the two-footed, the four-footed, for moving": the udātta of dvipā́t is on pā́.
tripā́d ūrdhváḥ "three-footed, upward": the udātta of tripā́t is on pā́.
dvidán: the udātta is on dán. The Subodhinī: dat replaces danta by SK 880.
trimūrdhā́naṃ saptáraśmim "the three-headed, seven-rayed": the udātta of trimūrdhā́nam is on dhā́. In saptáraśmim the udātta is on ptá; the next syllable, ra, is a dependent svarita, left unmarked.
Applies when / blocked by. Bhaṭṭoji draws two points from the word mūrdhan.
dvi + mūrdhan + ṣa → dvimūrdha → final udātta (option) → dvimūrdháḥ: the udātta is on dháḥ. Likewise trimūrdháḥ.
Watch out. Why "after dvi and tri"? kalyāṇamūrdhā: the rule does not apply. The Subodhinī says kalyāṇa keeps its own accent there (phiṭ 42). Why "in a bahuvrīhi"? dvimūrdhā, "the head of two" (dvayor mūrdhā), is a tatpuruṣa.
Rule. saktha "thigh" optionally has its final syllable udātta, provided it does not follow a word ending in kra. Here saktha is the stem with the samāsānta ṣac (SK 852), as the Subodhinī says.
Example.
gaurasaktháḥ "white-thighed": in this option the udātta is on tháḥ.
ślakṣṇasaktháḥ "smooth-thighed": likewise on tháḥ.
In the other option, says the Subodhinī, gaura keeps its own accent. gaura has both its first and last syllables udātta (phiṭ 13). So the compound then has three accents.
Applies when / blocked by. Why "not after a word ending in kra"? cakrasaktháḥ. Here the final is udātta obligatorily, because ṣac is cit (SK 3710). So this rule is a prāptavibhāṣā: an option on an accent that would come anyway.
Watch out. The printed text of Bhaṭṭoji's counter-question has ākrāntāt kim?. The sūtra is ca + akrāntāt, so the grammar requires akrāntāt: that is the word asked about.
Rule. In the Veda, the first syllable of the following member, saktha, is variously (bahulam) udātta. The Subodhinī explains why the sūtra says "following" at all. The bahuvrīhi is also in context, and one might think the bahuvrīhi's first syllable is meant.
Example. ajisáktham ālabheta: the udātta is on sák, the first syllable of saktha.
Applies when / blocked by. A vārttika in verse explains bahulam. A vārttika is a supplementary statement to a sūtra. It says: the first of the later member (parādi), the last of the later member (parānta), and the last of the earlier member (pūrvānta) are seen as udātta. So is the first of the earlier member (pūrvādi). Hence there is interchange (vyatyaya), and the sūtra says bahulam. Bhaṭṭoji's examples:
tuvijātā́ urukṣáyā "of mighty birth, of wide dwelling". The udātta syllables are tā́ and kṣá; kṣá is the first syllable of kṣaya.ní yéna muṣṭihatyáyā "by which, with fist-blows". The udātta syllables are ní, yé and tyá; tyá is the final of the stem muṣṭihatyā.yás tricakráḥ "who, three-wheeled". The udātta syllables are yás and kráḥ, the final of cakra.viśvā́yur dhehi "grant full life". The udātta is on śvā́: the final of viśva, merged with the ā of āyus.Watch out. The Subodhinī questions the first example. It argues that urukṣaya is really a case of bāhulaka (variable application) of other rules. It gives codayitrī sūnṛtānām as the true example of parādi.
With this sūtra Bhaṭṭoji closes the accents of compounds (samāsasvarāḥ).
agní + am → agním (udātta gní) + īḷe → agním īḷe (īḷe after a non-tiṅ pada: wholly anudātta)The 1915 print opens a new section here, "the accents of tiṅ-final words" (tiṅantasvarāḥ). A tiṅanta is a finite verb form, a word ending in a verb-ending (tiṅ). Two headings taught with the pronoun substitutes govern the whole section. padāt (SK 402) means "after a pada", a full word. anudāttaṃ sarvam apādādau (SK 403) means "wholly anudātta, when not at the start of a verse-quarter (pāda)". Being wholly anudātta is called nighāta. These three sūtras set up the basic rule (SK 3935) and its first exception (SK 3936).
The running example is the first words of the Ṛgveda.
agni + am → am is anudātta, as a case-ending (sup)
→ i + a become one vowel by SK 194 → agnim
→ the single substitute is udātta by SK 3658 → agním: the udātta is on gní
agním + īḷe → īḷe follows a pada that is not a verb → nighāta (SK 3935) → agním īḷe: īḷe is wholly anudātta.
In recitation its ī becomes a dependent svarita after the udātta (SK 3660). A dependent svarita is a falling pitch that only follows an udātta; it is left unmarked.
Rule. After a pada ending in a verb-ending, gotra "family" and the other words of its list (gotrādi) are wholly anudātta. This holds in two senses: contempt (kutsana) and repetition (ābhīkṣṇya).
Example.
pacati gotram "he torments his family" (contempt). The Subodhinī glosses: he torments his own family.
pacatipacati gotram (repetition). The verb is doubled by SK 2140. The Subodhinī: at a wedding, "he makes the family happy again and again".
In both, gotram is wholly anudātta. Likewise pravacana, prahasana, prakathana, pratyāyana and the rest.
Applies when / blocked by. Why "gotra and the rest"? pacati pāpam "he cooks badly": pāpa is not in the list. Why "in contempt and repetition"? khanati gotraṃ sametya kūpam "the family, having come together, digs a well": neither sense is present.
Watch out. Bhaṭṭoji: the two senses qualify the list itself (pāṭhaviśeṣaṇa). So wherever the gotrādi list is named elsewhere, it counts only in these senses. SK 3964 is such a place. The Subodhinī gets this by splitting the rule in two (yogavibhāga). "tiṅo gotrādīni" gives the anudātta. "kutsanābhīkṣṇyayoḥ" then acts as a paribhāṣā, a rule about how other rules are read.
Rule. A tiṅanta that follows a pada not ending in tiṅ is wholly anudātta. Bhaṭṭoji says it "is struck down" (nihanyate). The Subodhinī: nighāta is the grammarians' usual name for being wholly anudātta.
Example. agním īḷe "I praise Agni". The udātta is on gní. īḷe follows agnim, which is not a verb, so īḷe is nighāta. Its ī is only a dependent svarita.
Applies when / blocked by. This is the rule that SK 3936–3973 hold back, case by case. The Subodhinī adds that the Bhāṣya (the great commentary on Pāṇini's sūtras) rejects the word atiṅ "not ending in tiṅ" as unnecessary.
Watch out. "Wholly anudātta" holds only away from the start of a verse-quarter (SK 403). A verb that opens a pāda keeps its accent.
Rule. A tiṅanta in luṭ, the periphrastic future, is not made wholly anudātta. The Subodhinī: "the prohibitions begin here." From here to SK 3973 the rules mostly name cases where SK 3935 does not apply; SK 3944–3945 work the other way.
Example.
śvaḥ kartā́ "he will do it tomorrow": the udātta is on tā́.
śvaḥ kartā́rau "the two will do it tomorrow": the udātta is again on tā́.
Applies when / blocked by. The Subodhinī explains each form.
kartā́rau.Watch out. śvaḥ is left unmarked because its accent is not discussed. Unmarked does not mean anudātta there.
yád agne syā́m aháṃ tvám (syā́m keeps its udātta syā́ after the particle yad; agne, an āmantrita after a pada, anudātta)From here to SK 3973 the pattern is the same. A particle (nipāta), or a condition, keeps a tiṅanta from nighāta. The verb then keeps its own accent. Bhaṭṭoji's gloss "a tiṅanta … is not anudātta" supplies the subject and the "not", presumably from SK 3935 (tiṅ) and SK 3936 (na). yuktam, "connected with", runs on from SK 3937.
Often the text does not say which syllable a protected verb keeps. It is then marked here from the general rules: a root such as pac or gam has the udātta on its root syllable (SK 3671), and śap and tip are anudātta (SK 3709). The Subodhinī itself states this accent of pacati at SK 3957.
Two sūtras of this group work the other way (SK 3944–3945). In praise they cancel the protection, and the verb is anudātta after all. The running example is a Vedic line: "if, Agni, I were you".
yád (a particle) + agne + syā́m
→ agne is an āmantrita, a vocative, after a pada → wholly anudātta (SK 3654)
→ syām is connected with yad → not nighāta (SK 3937) → keeps its udātta on syā́
→ yád agne syā́m aháṃ tvám: the udātta syllables are yá, syā́, há and tvá.
Rule. A tiṅanta connected with the particles yad, yadi, hanta, kuvit, net, cet, caṇ, kaccit or yatra is not made anudātta.
Example.
yád agne syā́m aháṃ tvám "if, Agni, I were you": syā́m keeps its udātta on syā́.
yúvā yádī kṛtháḥ: kṛtháḥ keeps its udātta on thá.
kuvíd aṅgá ā́san: ā́san keeps its udātta on ā́.
ácittibhiś cakṛmā́ kác cit "whatever we have done through thoughtlessness": cakṛmā́ keeps its udātta on mā́. The Subodhinī: cakṛma is antodātta by the affix accent, and kac cit is two particles.
putrā́so yátra pitáro bhávanti "where the sons become fathers": bhávanti keeps its udātta on bhá.
Applies when / blocked by. The Subodhinī: yukta means direct connection. In yatra kva ca te mano dakṣaṃ dadhata uttaram, yatra goes directly with uttaram, so the rule does not apply to the verb.
Watch out. The Subodhinī asks why the sūtra says "particles" (nipātaiḥ). In yat kūjati śakaṭam "the cart, going, creaks", yat is the participle of i "to go", not the particle. The Subodhinī says the word also proves something larger. In this section neither the "specifically named" form nor association with neighbouring words decides which word is meant. That is why the derived yāvat (with the affix vatup) is also taken at SK 3943.
Rule. A tiṅanta connected with naha is not anudātta when there is a pratyārambha. Bhaṭṭoji: a pratyārambha is a beginning joined with a prohibition.
Example. naha bhokṣyáse "you shall not eat at all!". The udātta is on kṣyá, the future suffix sya. The Subodhinī: the ending se is anudātta after a suffix taught with final a (SK 3730). It also gives the scene. Someone was told "eat!", then forbidden again and again in anger or mockery. When he asks "will I get to eat?", this is the answer.
Applies when / blocked by. Why "in a pratyārambha"? naha vai tásmiṃl loké dákṣiṇam icchanti. Here icchanti is nighāta by SK 3935. The Subodhinī gives the accents of the other words: tásmin first-syllable udātta, loké final udātta, dákṣiṇam first-syllable udātta (phiṭ 29).
Rule. A tiṅanta connected with satya "truly" is not anudātta in a question.
Example. satyaṃ bhokṣyáse? "will you really eat?". The Subodhinī: the verb is udātta in the middle, on kṣyá.
Applies when / blocked by. Why "in a question"? satyám íd vā́ u táṃ vayám índraṃ stavāma "truly, indeed, we shall praise that Indra". The udātta syllables are tyá, íd, vā́, táṃ, yá and ín. The verb stavāma is wholly anudātta by SK 3935, even though satya precedes.
Rule. A tiṅanta connected with aṅga is not anudātta when there is no contrariness (aprātilomya). The Subodhinī: contrariness is acting against someone's wish, so the condition amounts to "in compliance".
Example. aṅga kurú "go on, do it!": aṅga expresses permission. The udātta is on rú. The Subodhinī: kuru is antodātta by the accent of the vikaraṇa u; the ending hi is dropped.
Applies when / blocked by. Why "no contrariness"? aṅga kūjasi vṛṣala, idānīṃ jñāsyasi jālma "so you make a noise, vṛṣala! Now you will find out, wretch!". This is said in indignation. Bhaṭṭoji: one who does what is not wanted is contrary. So kūjasi is nighāta by SK 3935.
Rule. A tiṅanta connected with hi is also not anudātta. The Subodhinī: "no contrariness" (SK 3940) applies here too. So in hi kūja vṛṣala, said in indignation, the rule does not apply.
Example.
ā́ hí ṣmā yā́ti: the verb yā́ti keeps its udātta on yā́. The other udātta syllables are ā́ and hí.
ā́ hí ruhátam: the verb ruhátam keeps its udātta on há.
The Subodhinī: sma is anudātta though a particle, since it belongs to the cādi list; its long ā is Vedic.
Watch out. The printed text has ruhaṃtam, with a nasal sign. As a verb, second person dual, the grammar requires ruhatam.
Rule. In the Veda, with hi, even more than one tiṅanta is kept from anudātta, when they are in mutual expectancy (sākāṅkṣa). Mutual expectancy means each clause needs the other to complete the sense.
Example. anṛtaṃ hi matto vádati, pāpmā cainaṃ yunā́ti "since the intoxicated man speaks untruth, evil attaches itself to him". The Subodhinī: hi means "since". Both verbs keep their accent.
Both accents are worked out from the rules named.
Applies when / blocked by. The Subodhinī: by api "even", a single tiṅanta is also covered. In the Veda the arrangement follows usage.
Rule. A tiṅanta connected with yāvat or yathā is not anudātta.
Example. yáthā cit káṇvam ā́vatam: the verb ā́vatam keeps its udātta on ā́. The other udātta syllables are yá and ká. The Subodhinī: āvatam is the loṭ, second person dual, of av.
Applies when / blocked by. The Subodhinī: the ending of yāvadyathābhyām is the third case, "with", not the fifth case, "after". So the rule also applies when the particle follows the verb: devadattaḥ pacati yāvat. yāvat is taken both as the particle and as the derived word in vatup, as SK 3937 showed.
Rule. In praise (pūjā), a tiṅanta that stands immediately after yāvat or yathā is anudātta. The na of this sūtra cancels the protection of SK 3943. The Subodhinī: "the meaning is that it is anudātta."
Example.
yāvat pacati śobhanam "how well he cooks!": pacati is wholly anudātta.
yathā pacati śobhanam: likewise.
Applies when / blocked by. Why "in praise"? yāvad bhuṅkte: no praise, so SK 3943 keeps the verb's accent. Why "immediately after"? yāvad devadattaḥ pacati śobhanam: devadattaḥ stands between. Here SK 3943 again prohibits the nighāta.
Watch out. The Subodhinī adds that "immediately after" could have been left out. SK 3945 allows only an upasarga in between, so nothing would over-apply.
Rule. The same holds when an upasarga stands between yāvat or yathā and the verb. In praise, the verb is then anudātta too. An upasarga is a prefix such as pra, construed with the root.
Example. yāvat prápacati śobhanam. The Subodhinī: pra has its first syllable udātta, prá, and pacati is anudātta.
Applies when / blocked by. Bhaṭṭoji: "immediately after" still holds. The verb with its upasarga must follow at once. In yāvad devadattaḥ prapacati śobhanam, devadattaḥ stands between. So SK 3943 keeps the verb from becoming anudātta.
aho devadattaḥ pácati śobhanam (with aho, in praise: pacati keeps its root udātta pá)Five sūtras keep the verb accented with particles of praise, haste and permission. With aho outside praise, and with purā, the protection is only optional. The running example is "ah, how well Devadatta cooks!".
aho devadattaḥ pacati śobhanam (praise)
→ pacati follows a non-verb → nighāta would come (SK 3935)
→ aho in praise prohibits it (SK 3947)
→ pacati keeps the root accent → pácati: the udātta is on pá; śap and tip are anudātta as pit affixes (SK 3709).
The Subodhinī states this accent of pacati at SK 3957.
Rule. In praise, a tiṅanta connected with tu, paśya "see!", paśyata "see!" (plural) or aha is not made anudātta.
Example. ā́d áha svadhā́m ánu púnar garbhatvám eriré "then indeed, after their own nature, they took on again the state of an embryo". The particle is áha. The verb eriré keeps its final udātta, ré. The other udātta syllables are ā́, á of áha, dhā́, á of ánu, pú and tvá.
The Subodhinī adds simpler examples: paśya māṇavako bhuṅkte, māṇavakas tu bhuṅkte. Here tu expresses wonder, so praise of the eating is understood.
Applies when / blocked by. Why "in praise"? paśya mṛgo dhāvati "look, a deer is running": a mere statement of fact.
Watch out. Why is "praise" stated again, when SK 3944 already said it? The Subodhinī: the praise of SK 3944 carried a cancelling force. Carried on, it would cancel the prohibitions of SK 3936 and the following rules. Stated afresh, it is a plain prohibition of nighāta. It is also not limited to "immediately after".
Rule. In praise, a tiṅanta connected with aho is also not anudātta.
Example. aho devadattaḥ pácati śobhanam "ah, how well Devadatta cooks!": pácati keeps the udātta on its root syllable pá.
Watch out. Why a separate sūtra, when aha was already in SK 3946? The Subodhinī: so that the option of the next rule holds for aho alone, not for tu and the rest.
Rule. In the remaining case, a tiṅanta connected with aho is optionally anudātta. The Subodhinī glosses "the remaining case" as "other than praise".
Example. aho kaṭaṃ kariṣyati "oh, he will make a mat!". Two options:
kariṣyáti: the udātta is on ṣyá, the suffix sya.The Subodhinī: something not wonderful is presented out of envy as if wonderful. So this is not praise.
Watch out. Bhaṭṭoji's gloss, as transmitted, says "in praise" (pūjāyām). The Subodhinī's account of "other than praise" fits a reading apūjāyām; scholars discuss this. Either way the example, said out of envy, gets the option.
Rule. A tiṅanta connected with purā is optionally anudātta when haste (parīpsā, glossed tvarā) is meant.
Example. adhīṣva māṇavaka purā vidyotate vidyut "study, boy: the lightning will soon flash". purā here means "soon". The present ending (laṭ) has a future sense by SK 2783. The Subodhinī: study during lightning is forbidden, so "study quickly" is conveyed. Two options:
vídyotate: dyotate is wholly anudātta, and vi keeps its own udātta, ví;vidyótate: the root syllable dyó is udātta, and vi becomes anudātta by SK 3978.The Subodhinī states only the accent of dyotate; the marks on vi presumably come from phiṭ 81 and SK 3978.
Applies when / blocked by. Why "when haste is meant"? na tena sma purādhīyate "formerly it was not studied by him". Here purā means "long ago", and the present ending has a past sense by SK 2778.
Rule. A tiṅanta connected with nanu is not anudātta when permission is requested (anujñaiṣaṇā).
Example. nanu gácchāmi bhoḥ "well, I am going, sir!". Bhaṭṭoji: the meaning is "permit me, as I am going". gácchāmi keeps the udātta on its root syllable gá. The Subodhinī: the present ending is used for what is close to the present (SK 2789).
Applies when / blocked by. Why "when permission is requested"? akārṣīḥ kaṭaṃ tvam? nanu karomi "did you make the mat?" "of course I made it". This is a reply to a question. The present ending karomi is past in sense by SK 2780.
Watch out. The printed text has ākarṣīḥ in the question. That is not a form of kṛ, and the reply needs one; the grammar requires akārṣīḥ "did you make". The Subodhinī: the iti in the sūtra shows that nanu is a single particle, not na + nu.
kiṃ dvijaḥ pácaty ā́hosvid gacchati (the first verb keeps its root udātta pá; ā́hosvit ādyudātta)Seven sūtras protect the verb in questions and with indefinite words. Two of them are options (SK 3952, SK 3957), and one is a niyama, a rule that restricts what another rule already gives (SK 3953). The running example is "is the brāhmaṇa cooking, or is he going?".
kiṃ dvijaḥ pacati → kim asks about the action → pacati is not nighāta (SK 3951)
→ pácati: the udātta is on the root syllable pá
āhosvit → a single particle, udātta on its first syllable (the Subodhinī) → ā́hosvit
→ kiṃ dvijaḥ pácaty ā́hosvid gacchati.
The accent of gacchati is disputed; see SK 3951.
Rule. A tiṅanta connected with kim, when kim asks about the action, is not anudātta. The verb must have no upasarga and must not be negated.
Example. kiṃ dvijaḥ pácaty ā́hosvid gacchati "is the brāhmaṇa cooking, or is he going?". pácati keeps the udātta on pá.
Applies when / blocked by. Bhaṭṭoji's counter-examples:
kiṃ bhaktaṃ pacaty apūpān vā "is it rice he cooks, or cakes?": the question is about a means of the action, not the action.kiṃ paṭhati "what, he reads!": this is censure, not a question.kiṃ prapacati uta prakaroti: the verb has an upasarga.kiṃ dvijo na pacati: the verb is negated.Watch out. The second verb, gacchati, is disputed. The Subodhinī reports two views. Haradatta (a commentator on the grammar) holds that kim goes only with the first verb, so gacchati is nighāta. Others hold that both verbs are in doubt, so both are protected. So gacchati is left unmarked here.
Rule. When kim is not used, the whole of SK 3951 holds optionally. The Subodhinī: no rule ever elides kim, so lopa "elision" here means mere non-use. The sense is understood from context.
Example. devadattaḥ pacaty āhosvit paṭhati "is Devadatta cooking, or reading?". Two options:
pácati, with the root syllable pá udātta.Rule. In jest (prahāsa), a tiṅanta in lṛṭ, the simple future, connected with ehi manye, is not anudātta.
Example. ehi mánye bhaktaṃ bhokṣyáse, nahi bhokṣyase, bhuktaṃ tat tv atithibhiḥ "come, you think: 'I shall eat the rice!' You will not eat it: the guests have eaten it". bhokṣyáse keeps its udātta on kṣyá, the suffix sya. The Subodhinī: mánye has its udātta on the first syllable, má. The persons are exchanged in jest by SK 2163: "I think you will eat" stands for "you think I shall eat".
Applies when / blocked by. Bhaṭṭoji: ehi is the loṭ of a verb of motion, so SK 3958 would already give this result. So this sūtra is a niyama: with ehi manye, the protection holds only in jest. Without jest, ehi manyase odanaṃ bhokṣye iti, suṣṭhu manyase "come, you think 'I shall eat the rice': you think rightly". Here bhokṣye is presumably nighāta by SK 3935.
Watch out. The Subodhinī notes that the Bhāṣya gives a different counter-example: ehi manye rathena yāsyasi "come, I think you will go by chariot".
Rule. A tiṅanta connected with jātu "ever", when jātu has nothing before it, is not anudātta.
Example. jā́tu bhokṣyáse "at some time you will eat". The Subodhinī: jātu, a particle, has the udātta on its first syllable, jā́. bhokṣyáse has it on kṣyá.
Applies when / blocked by. Why "with nothing before it"? kaṭáṃ jā́tu kariṣyasi. Here kaṭa precedes jātu. kaṭáṃ has its udātta on the final ṭáṃ (phiṭ accent, the Subodhinī says). jātu keeps its own udātta on jā́. kariṣyasi is presumably wholly anudātta.
Watch out. "With nothing before it" (apūrvam) describes jātu, not the verb. The Subodhinī proves this from the word anantaram in SK 3956. If apūrvam described the verb, adjacency would already follow there, and anantaram would be pointless.
Rule. A tiṅanta connected with a form of kim that is followed by cit, and has nothing before it, is not anudātta. "Nothing before it" is carried from SK 3954. Bhaṭṭoji: a form of kim (kiṃvṛtta) is kim with a case-ending, or with the affixes ḍatara or ḍatama.
Example.
kaścid bhuṅkte "someone is eating".
kataraś cit [bhuṅkte], katamaś cit [bhuṅkte].
bhuṅkte keeps its own accent. Bhaṭṭoji and his commentators do not say which syllable that is, so it is left unmarked.
Applies when / blocked by. Why "followed by cit"? ko bhuṅkte "who is eating?". "With nothing before it" holds too: rāmaḥ kiṃcit paṭhati "Rāma reads something". In both, the verb is nighāta by SK 3935.
Watch out. The Subodhinī: the limit to case-forms and ḍatara/ḍatama keeps out forms such as kiṃtarām.
Rule. A tiṅanta connected with āho or utāho is not anudātta. The verb must follow immediately (anantaram), and the particle must have nothing before it (from SK 3954).
Example. āho bhuṅkte, utāho bhuṅkte "is he really eating?". bhuṅkte keeps its own accent; it is left unmarked, as in SK 3955.
Applies when / blocked by. Bhaṭṭoji: "immediately" holds, and the next sūtra gives an option for the remaining case. Why "with nothing before it"? deva āho bhuṅkte: deva stands before āho, so bhuṅkte is presumably nighāta by SK 3935.
Rule. In the remaining case, a tiṅanta connected with āho or utāho is optionally anudātta. Bhaṭṭoji: the remaining case is where the verb is separated (vyavahita) from the particle.
Example. āho devaḥ pacati "is Deva really cooking?". devaḥ stands between. Two options:
pácati: the Subodhinī says it then has its root accent, with the udātta on pá.āgaccha deva grāmaṃ drakṣyásy enam (lṛṭ after the motion loṭ āgaccha: drakṣyasi keeps its suffix udātta ṣyá)loṭ is the imperative. Three sūtras concern a verb connected with the imperative of a verb of motion, such as āgaccha "come!" (SK 3958–3960). SK 3958 covers a future (lṛṭ) verb, SK 3959 an imperative one, and SK 3960 makes the protection optional when that imperative has an upasarga. SK 3961 extends the option of SK 3960 to a loṭ with hanta. The running example is "come to the village, Deva: you will see him".
āgaccha deva grāmam → āgaccha is the loṭ of gam, a verb of motion
drakṣyasi → lṛṭ of dṛś, connected with āgaccha; its agent is the same Deva
→ not nighāta (SK 3958)
→ drakṣyási: the Subodhinī says the suffix accent puts the udātta on ṣyá
→ āgaccha deva grāmaṃ drakṣyásy enam.
Rule. A tiṅanta in lṛṭ connected with the loṭ of a root meaning motion is not anudātta. The condition: the kāraka must not be wholly different. A kāraka is a role in the action, such as agent or object. Bhaṭṭoji: the lṛṭ must be used of the very kāraka of which the loṭ is used.
Example.
āgaccha deva grāmaṃ drakṣyásy enam "come to the village, Deva: you will see him". drakṣyási has the udātta on ṣyá.
uhyantāṃ devadattena śālayo rāmeṇa bhokṣyante "let the rice be carried by Devadatta; it will be eaten by Rāma". The Subodhinī: vah "carry" implies motion. The agents differ, but the object, the rice, is the same.
Applies when / blocked by. Bhaṭṭoji's counter-examples:
paca deva odanaṃ bhokṣyase 'nnam: paca "cook!" is not a verb of motion.āgaccher deva grāmaṃ drakṣyasy enam: āgaccheḥ is liṅ (the optative), not loṭ.āgaccha deva grāmaṃ paśyasy enam: paśyasi is not lṛṭ.āgaccha deva grāmaṃ, pitā te odanaṃ bhokṣyate: the agent is a different person. The Subodhinī: kāraka means the individual, so two different agents count as different.Watch out. Why "wholly" (sarva)? So that the rule still applies in āgaccha deva grāmaṃ tvaṃ cāhaṃ ca drakṣyāva enam "come to the village, Deva: you and I shall see him". The lṛṭ verb expresses the loṭ's agent and another one besides. That is not wholly different.
Rule. A loṭ verb, too, connected with the loṭ of a verb of motion, is not anudātta. All of SK 3958 except "lṛṭ" carries over, including the kāraka condition.
Example. āgaccha deva grāmaṃ páśya "come, Deva; see the village". páśya keeps the udātta on its root syllable pá (SK 3671), with śap and hi anudātta.
Applies when / blocked by. Bhaṭṭoji's counter-examples:
paca devaudanaṃ bhuṅkṣvainam "cook the rice, Deva; eat it": paca is not a verb of motion.āgaccha deva grāmaṃ paśyasi: paśyasi is not loṭ.āgaccha deva grāmaṃ paśyatv evaṃ rāmaḥ "come, Deva, to the village; let Rāma see it thus": the agent is different, so paśyatu is nighāta.Because of "wholly" it does apply in āgaccha deva grāmaṃ tvaṃ cāhaṃ ca paśyāva "let you and me see it". The printed text has paśyāvaḥ, a present form; the grammar requires the loṭ paśyāva.
Watch out. Why a separate sūtra, and not "lṛṭ and loṭ" in one? Bhaṭṭoji: the split (yogavibhāga) is for the next rule. The Subodhinī: so that the option of SK 3960 applies to loṭ only, not to lṛṭ.
Rule. Such a loṭ verb, when it has an upasarga and is not in the first person, is protected only optionally. The Subodhinī calls it a prāptavibhāṣā: an option on what SK 3959 would give.
Example. āgaccha deva grāmaṃ praviśa "come, Deva; enter the village". Two options:
praviśá: the udātta is on śá, by the suffix accent. The Subodhinī: the gati pra then becomes anudātta by SK 3978.práviśa: viśa is anudātta, and pra keeps its own udātta, prá (presumably phiṭ 81).Applies when / blocked by. Why "with an upasarga"? āgaccha deva grāmaṃ paśya. Why "not in the first person"? āgacchāni deva grāmaṃ praviśāni "let me come, Deva; let me enter the village". In both, SK 3959 applies without option.
Watch out. The Subodhinī: "with an upasarga" means only that an upasarga is present. The upasarga itself does not share in the operation.
Rule. A loṭ verb connected with hanta, with an upasarga and not in the first person, is also protected only optionally.
Example. hanta praviśa "come now, enter!". Two options, presumably marked as in SK 3960: praviśá (udātta on śá, pra anudātta) or práviśa (udātta on prá).
Applies when / blocked by. The Subodhinī: this is a prāptavibhāṣā against SK 3937, whose list includes hanta. Bhaṭṭoji: "with an upasarga" still holds. In hanta kuru "come now, do it!", the verb has no upasarga, so SK 3937 prohibits nighāta outright. Why "not in the first person"? hanta prabhuñjāvahai "come now, let us two eat": again SK 3937 applies outright.
devaḥ pácati cana (cana follows; pacati, not after a gati, keeps its root udātta pá)Up to now the protecting word mostly stood before the verb. In this group it mostly stands after it. SK 3962 is the odd one out: it protects a vocative, not a verb. The running example is devaḥ pácati cana.
devaḥ pacati → pacati follows a non-verb → nighāta would come (SK 3935)
→ cana follows the verb, and no gati precedes it → nighāta is prohibited (SK 3964)
→ devaḥ pácati cana: the udātta is on the root syllable pá.
Rule. An āmantrita (a word of address) after the particle ām, with one pada in between, is not anudātta when the addressee is not near (anantika).
Example. ām pacasi dévadatta3 "yes, you are cooking, Devadatta!". The Subodhinī: since nighāta is prohibited, devadatta has the vocative's first-syllable udātta (SK 3653). The udātta is on dé. The 3 marks a pluta, a vowel drawn out to three morae.
Otherwise devadatta would presumably be nighāta by SK 3654; the pluta is presumably that of calling from afar (SK 95).
Applies when / blocked by. Bhaṭṭoji's counter-examples:
ām prapacasi devadatta3: two padas, pra and pacasi, stand between.ām pacati devadattaḥ: devadattaḥ is a nominative.ām pacasi devadatta: the addressee is near.Watch out. The Subodhinī reports two views of anantika. Some take it as "far". Others take it as "neither far nor near", since otherwise the sūtra would just say "far". The two views differ on whether the pluta udātta should be used.
Rule. In the Veda, a tiṅanta followed by yad, hi or tu is not anudātta.
Example.
udásṛjo yád aṅgiraḥ "when you let them out, O Aṅgiras". The udātta of udásṛjaḥ is on dá, the augment a. yád is udātta. The first syllable of aṅgiraḥ is a dependent svarita.
uśánti hí "for they desire": the udātta syllables are śá and hí.
ākhyāsyā́mi tu te "but I shall tell you": the verb presumably keeps its accent on syā́, the suffix sya, with ā anudātta by SK 3978.
Applies when / blocked by. Bhaṭṭoji: SK 3937, SK 3941 and SK 3946 would already give this. So the sūtra is a niyama. With these particles standing after the verb, only yad, hi and tu count. Others, such as yadi or hanta, do not (the Subodhinī).
Watch out. So in jāye svā rohāvaihi "wife, come, let us two climb to heaven", rohāva is nighāta. ehi, the loṭ of a verb of motion, follows it. SK 3959 would protect rohāva, but this niyama withdraws that protection. The Subodhinī: the nighāta rests on the preceding word svā, by SK 3935.
Rule. When cana, cit, iva, a word of the gotrādi list, a taddhita or an āmreḍita follows, a tiṅanta not preceded by a gati is not anudātta. A taddhita is an affix added to a word; an āmreḍita is the second member of a doubled word.
Example. In each, pácati keeps the udātta on its root syllable pá:
devaḥ pácati cana,
devaḥ pácati cit,
devaḥ pácatīva,
devaḥ pácati gotram,
devaḥ pácatikalpam (the taddhita kalpap),
devaḥ pácatipacati (the second pacati is the āmreḍita).
Applies when / blocked by. Why "not preceded by a gati"? devaḥ prapacati cana: pra precedes, so the rule does not apply.
Watch out. The Subodhinī: only an anudātta taddhita is meant, such as kalpap, which is pit. Elsewhere the taddhita's own accent prevails, as in ayaṃ pacatideśyaḥ. And by SK 3934, a gotrādi word counts only in contempt or repetition.
Rule. When ca, vā, ha, aha or eva follows, a tiṅanta not preceded by a gati is not anudātta. "Not preceded by a gati" is drawn in by ca "and".
Example. devaḥ pácati ca khā́dati ca "Deva both cooks and eats". pácati has the udātta on pá, khā́dati on khā́.
Applies when / blocked by. "Not preceded by a gati" holds: devaḥ prapacati ca prakhādati ca. Bhaṭṭoji: the first verb is still saved, by SK 3966; the second is nighāta.
Watch out. The Subodhinī: the list meant is that of SK 408 (ca, vā, ha, aha, eva), which is near. It is not the cādi list of SK 20, which is remote.
itó vā sātím ī́mahe (with vā: the first tiṅanta īmahe keeps its udātta ī́)Seven of these sūtras protect only the first verb of a series; SK 3973 then protects any verb after a word containing yad. prathamā "the first" is named in SK 3966 and runs into SK 3967–3970 and SK 3972. Bhaṭṭoji reads it as "the first tiṅ-ending". agateḥ does not continue here, the Subodhinī says. With SK 3973, which holds always (nityam), the protections close. The running example is a Vedic line, "from here or … we pray for gain".
itó vā sātím īmahe → vā connects this clause with the next two
→ īmahe is the first verb → not nighāta (SK 3966)
→ itó vā sātím ī́mahe: the udātta syllables are tó, tí and ī́; īmahe keeps ī́.
Rule. In connection with ca or vā, the first tiṅ-ending is not anudātta.
Example.
gāś ca cārayati vīṇāṃ vā vādayati "he grazes the cows and …, he plays the vīṇā or …".
itó vā sātím ī́mahe: ī́mahe keeps its udātta on ī́.
Bhaṭṭoji: īmahe counts as "first" with respect to the verb understood in the next two clauses. The Subodhinī names them: divo vā pārthivād adhi and indraṃ maho vā rajasaḥ.
Applies when / blocked by. Why "in connection" (yoge)? So that the verb is also saved when ca or vā stands before it, as vā does in the mantra. Why "first"? So that the second and later verbs are not covered.
Rule. The first tiṅ-ending connected with ha is not anudātta in kṣiyā. Bhaṭṭoji: kṣiyā is a breach of proper conduct (dharmavyatikrama).
Example. svayaṃ ha rathena yāti3 upādhyāyaṃ padātiṃ gamayati3 "he himself goes by chariot, and makes his teacher go on foot!". The first verb, yāti, is protected; gamayati, not being first, is not. The Subodhinī: riding while the teacher walks breaks proper conduct. The pluta is by SK 3623.
Rule. The first tiṅ-ending connected with aha is not anudātta in viniyoga, and also in kṣiyā. Bhaṭṭoji: viniyoga is commissioning someone for several purposes. The Subodhinī: it is sending someone on more than one errand.
Example.
Commission: tvam aha grāmaṃ gaccha "you, go to the village".
Breach of conduct: svayam aha rathena yāti3 upādhyāyaṃ padātiṃ nayati "he himself goes by chariot, and leads his teacher along on foot!".
Rule. When ca or aha is left out, and eva is used as a restriction (avadhāraṇa), the first tiṅ-ending is not anudātta. The Subodhinī: "left out" means not used; no rule elides them.
Example.
deva eva grāmaṃ gacchatu, deva evāraṇyaṃ gacchatu "let Deva go just to the village; let Deva go just to the forest". Bhaṭṭoji: the meaning is "to the village and to the forest", so ca is left out. The agent is the same.
deva eva grāmaṃ gacchatu, rāma evāraṇyaṃ gacchatu "let Deva go only to the village, Rāma only to the forest". Here aha "only" is left out. The agents differ.
In both, the first gacchatu keeps its accent.
Applies when / blocked by. Why "as a restriction"? deva kveva bhokṣyase "Deva, where will you eat?", meaning "nowhere". This eva expresses impossibility, not restriction.
Rule. When ca, vā, ha, aha or eva is left out, the first tiṅ-ending is optionally not anudātta.
Example.
ca left out: índra vā́jeṣu no 'va "O Indra, help us in the contests". The udātta syllables are ín and vā́. The verb ava is anudātta here. The Subodhinī: this shows only the nighāta option; an example of the other option "is to be sought".
ca left out: śuklā vrīhayo bhavanti, śvetā gā ājyāya duhanti "the rice-grains are white; they milk white cows for the clarified butter".
vā left out: vrīhibhir yajeta, yavair yajeta "he should sacrifice with rice-grains, or with barley". The Subodhinī: in one option the first verb is yájeta, with the udātta on yá; the second is nighāta.
Watch out. In the second example the agents differ, yet ca is left out. The Subodhinī: the condition "same agent" for ca in SK 3969 holds only where eva restricts.
Rule. In the Veda, with vai or vāva, the first tiṅanta is optionally not anudātta. Bhaṭṭoji gives no gloss. The Subodhinī supplies "first" and "optionally".
Example. ahar vai devānām āsīt "the day, indeed, was the gods'". ayaṃ vāva hasta āsīt "this, surely, was the hand". The verb is āsīt in both. No accented copy is given, so nothing is marked. The Subodhinī: vai means evidentness or forbearance; vāva means the well-known or evidentness.
Rule. In the Veda, with eka and anya used in the same sense, "the one … the other", the first tiṅ-ending is not anudātta.
Example.
ajām ekāṃ jinvati, ajām ekāṃ rakṣati "one she-goat he gladdens, the other he guards".
táyor anyáḥ píppalaṃ svādv átti "of those two, one eats the sweet berry". átti keeps its udātta on its first syllable a, merged into dva of svādv átti. The other udātta syllables are tá, nyáḥ and pí.
Applies when / blocked by. Why "used in the same sense"? eko devān upātiṣṭhat "a single one attended on the gods": here eka denotes number. The Subodhinī: association with anya does not settle the sense in this section (as at SK 3937), so the sūtra has to say samarthābhyām.
Rule. A tiṅanta after a pada in which the word yad occurs is always (nityam) not anudātta.
Example.
yo bhuṅkte "he who eats".
yadadryaṅ vāyur vāti "towards which the wind blows". The Subodhinī: yadadryaṅ is yad + añc with kvin, and adri replaces the ṭi of yad (SK 418).
No accented copy is given, so nothing is marked.
Applies when / blocked by. Bhaṭṭoji: the operation is wanted even with a word in between, as vāyuḥ is here. The sūtra is stated with a fifth-case word, "after", which usually means "immediately after". The jñāpaka that allows the gap comes in SK 3974.
Watch out. nityam "always" presumably sets aside the option of SK 3970–3971.
yat prapácati (the verb keeps pá after yat, so the gati pra goes anudātta) against yat kāṣṭhāṃ prapacati (in praise: pra and pacati both anudātta)The last five sūtras turn around. They no longer protect: they impose anudātta. SK 3974–3976 make the praised or censured word anudātta. SK 3977–3978 make a gati anudātta. Then Bhaṭṭoji works through the first verse of the Ṛgveda, accent by accent, and closes the chapter. The running example contrasts two phrases.
yat prapacati → pacati is protected after yat (SK 3937) → keeps the udātta on pá
→ the gati pra stands before a verb that has an udātta → pra anudātta (SK 3978)
→ yat prapácati.
yat kāṣṭhāṃ prapacati (in praise) → the praised verb with its gati is anudātta (SK 3975)
→ pra and pacati are both anudātta, though yat would have protected the verb.
Rule. After kāṣṭha and the rest, words of praise, the word for what is praised is anudātta. Bhaṭṭoji reads the singular pūjanāt as a plural, to match kāṣṭhādibhyaḥ. The Subodhinī: kāṣṭha and the rest are synonyms of "wonderful".
Example. kāṣṭhādhyāpakaḥ "a wonderful teacher": adhyāpaka is anudātta.
A vārttika adds: the m is also elided. dāruṇam + adhyāpakaḥ → dāruṇādhyāpakaḥ; likewise ajñātādhyāpakaḥ.
Applies when / blocked by. Bhaṭṭoji: this is an exception (apavāda) to the final udātta of a compound. It is wanted in a compound, not in dāruṇam adhyāpakaḥ: that is the view of the Vṛtti (a running commentary on the sūtras). The Subodhinī reports that the Bhāṣya, the Vārttika, Kaiyaṭa (a commentator on the Bhāṣya) and Haradatta want it only outside compounds, as in dāruṇam adhyāpakaḥ; scholars discuss this.
Watch out. Why say "what is praised" at all, when praise implies it? Bhaṭṭoji: to secure only the word immediately after. This is the jñāpaka that elsewhere in this section "immediately" is not relied on, even with a fifth case. That jñāpaka is used at SK 3973.
Rule. After kāṣṭha and the rest, a tiṅanta that is what is praised is anudātta. With a gati, the gati is anudātta too; without one, the verb alone.
Example. yat kāṣṭhāṃ prapacati "that he cooks wonderfully": pra and pacati are both anudātta. Bhaṭṭoji: this rule applies where SK 3937 would have blocked the nighāta of SK 3935.
Applies when / blocked by. A vārttika: where "gati" is said here, only an upasarga is meant. So in yat kāṣṭhāṃ śuklīkaroti "that he whitens wonderfully", śuklī is not made anudātta: it is presumably a gati (SK 762) but not an upasarga.
Watch out. The Subodhinī: the word tiṅ shows that SK 3974 did not concern verbs.
Rule. In censure, when a subanta (a word with a case-ending) other than a gotrādi word follows, a tiṅanta, with or without a gati, is anudātta. The Subodhinī: padāt (SK 402) has ceased here.
Example.
pacati pūti "he cooks rottenly".
prapacati pūti: pra and pacati both anudātta.
pacati mithyā "he cooks wrongly".
Applies when / blocked by. Bhaṭṭoji's counter-examples:
prapacati śobhanam "he cooks finely": no censure.pacati kliśnāti "he cooks, he toils": no subanta follows.pacati gotram: gotra is a gotrādi word.Three vārttikas follow.
pacati; pūtir devadattaḥ "he cooks — Devadatta is rotten". The agent is censured, not the cooking.pūtí, with the udātta on tí.pacanti pūti "they cook rottenly".Rule. A gati followed by another gati is anudātta.
Example. abhi + ud + harati → abhyúddharati "he lifts up". abhi is anudātta. ud keeps its udātta, ú, and harati is nighāta. Bhaṭṭoji states only that abhi is anudātta; the rest presumably comes from phiṭ 81 for ud and SK 3935 for harati.
Applies when / blocked by. Why "a gati"? Bhaṭṭoji gives dattaḥ pacati. Neither word is a gati. The point is presumably that a non-gati is not made anudātta.
Why "followed by a gati"? ā́ mandraír indra háribhir yāhí mayū́raromabhiḥ "come, Indra, with your delightful bay horses with peacock hair". The Subodhinī: ā is the gati of yāhi. It is followed by mandraiḥ, which is not a gati, so ā keeps ā́. The udātta syllables are ā́, draí, há, hí and yū́.
Rule. A gati before a tiṅanta that has an udātta is anudātta. The Subodhinī: this is an exception to phiṭ 80, by which particles have the first syllable udātta.
Example. yat prapácati "that he cooks". pacati keeps its udātta, pá, because SK 3937 prohibits its nighāta after yat. So the gati pra is anudātta.
Applies when / blocked by. Why "that has an udātta"? prápacati. Here pacati is nighāta by SK 3935, so pra keeps its udātta, prá. The text does not mark these two syllables; the marks are presumed.
Watch out. Why does the sūtra say tiṅ, the whole verb, and not just the root? Bhaṭṭoji: pra and the rest are gatis only with respect to the action, the root. Without tiṅ, the rule would apply only when the root itself has the udātta. With tiṅ, it also applies when the udātta is on the affix. The Subodhinī's example: yat prakaróti, with the udātta on the affix u, ró; pra is still anudātta.
With this sūtra Bhaṭṭoji closes the accents of tiṅantas (tiṅantasvarāḥ).
The svara-saṃcāra. The 1915 print gives the heading "the manner of the movement of accent" (svarasaṃcāraprakāraḥ). Bhaṭṭoji now shows how the accents move in a Vedic sentence. He takes the first ṛc of the Ṛgveda:
agním īḷe puróhitaṃ yajñásya devám ṛtvíjam | hótāraṃ ratnadhā́tamam.
agni. Mādhava (a commentator on the Veda) says that, if the word is underived, it is antodātta by phiṭ 1 (phiṣaḥ). Bhaṭṭoji: in fact it is antodātta because it is in the ghṛtādi list. If it is derived, it is antodātta by the accent of the affix ni.
agním.
agni + am → am is anudātta as a sup
→ the single substitute by SK 194 → udātta by SK 3658
→ agním: the udātta is on gní.
īḷe. It is nighāta by SK 3935. In recitation (saṃhitā) its ī becomes svarita after the udātta (SK 3660). The next syllable, ḷe, becomes ekaśruti by SK 3668. Ekaśruti, "one tone", is also called pracaya: the unaccented syllables after a svarita run on at a single pitch.
puróhitam.
puras → antodātta by the accent of the affix asi (SK 1975)
hita → dhā + the niṣṭhā ta; hi replaces dhā (SK 3076) → antodātta by the affix accent
→ puras is named gati (SK 768) and compounded (SK 761)
→ four accents become due in turn, each overriding the one before:
the compound's final udātta; the accent of an indeclinable prior member (SK 3736); the accent of a kṛt-final later member (SK 3873); and the thāthādi accent
→ finally gatir anantaraḥ (SK 3783) gives the prior member its own accent
→ puróhitam: the udātta is on ró.
In saṃhitā the u of puras would be pracaya. But it stands before an udātta, so it becomes anudāttatara, "lower than anudātta", by SK 3669. Bhaṭṭoji does not give the forms under each of the four overridden accents.
The other words:
yajñásya: yajña has the accent of the affix naṅ; the udātta is on ñá. The case-ending, a sup, is anudātta and then becomes svarita.devám: deva has the affix ac of the pacādi list. It is antodātta by the phiṭ accent, by the affix accent, or by the accent of a cit. The udātta is on vá.ṛtvíjam: ṛtvij is antodātta by the accent of a kṛt-final later member; the udātta is on tví.hótāram: hotṛ ends in the affix tṛn, which is nit (marked with n), so it is ādyudātta; the udātta is on hó.ratnadhā́tamam: ratna has its first syllable udātta (phiṭ 26). ratnadhāḥ is "he who bestows treasures". It is antodātta by the compound accent, or by the accent of a kṛt-final later member. tamap is pit, so anudātta; its syllables then become svarita and pracaya. The udātta is on dhā́.Bhaṭṭoji ends: the rest is to be worked out according to the śāstra. "Thus the section on accent." A closing verse follows. Only a pointer to Vedic words has been shown here. May it please Bhavānī and Viśvanātha. The 1915 print adds the colophon: thus the Vaidika-svara-prakriyā in the Siddhānta-kaumudī of Bhaṭṭoji Dīkṣita.
á.kvà. A dependent svarita, made by SK 3660 after an udātta, is left unmarked.dādhīcáḥ (SK 3652).subrahmaṇyó3m.lū́yate kedāraḥ svayam eva (SK 3678).lólūyaṃlolūyam.cikī́rṣakaḥ.pākáḥ.ī́ḍyaḥ, páñca.uccaíḥ, gehám.kṣatríyā, kukkúṭaḥ.īḍya.tyāga from tyaj.kuha, kratha.śarāvatī.vetravatī), ṭāp leaves ā (pāṭalā). āp is the name covering the ā-affixes.báṃhiṣṭhair (phiṭ 7).jyéṣṭhaḥ (phiṭ 19, phiṭ 22).kauñjāyana (SK 3711).brahmabandhvā̀, bahvàratniḥ.jānatī́ (SK 3719).jānat.tudantī.brahmabandhū.akṣaṇvántaḥ.kartā.yāsiṣṭam.subhū, rāṭ.bhojya.gámana.kuṭī, śastrī.kirikāṇaḥ.kumāraśramaṇā.pañcakapālaḥ.saptasamam "a group of seven years".upapūrvāhṇam.śinivāsudevāḥ.ayajñaḥ.vispaṣṭa.pāṇinīya.śúnaś cic chépam.kaṭha, nanda, kumāra.hita, kṛta.kumbhakāraḥ, puṣpahārī (SK 3809–3814).dhānyagavaḥ (SK 3806).aśmārmam (SK 3825).sukármāṇaḥ.sukarmákaḥ.prakṣayáḥ, pralaváḥ.payaḥpānám / rājabhojanā́ḥ.vartman and kṛti.āgāmuka, alaṅkariṣṇu, āḍhyambhaviṣṇu.gómukhaḥ.anujyeṣṭháḥ.cakrasaktháḥ, pūtí.hótāram.agním īḷe at the end of SK 3978.| SK / phiṭ | what it does | example |
|---|---|---|
| 3650 | in an accent rule, all vowels of a pada but the one prescribed are anudātta | gopāyátaṃ naḥ (udātta ya) |
| 3651 | when an udātta is elided, the following anudātta becomes udātta | devī́ṃ vā́cam (udātta vī) |
| 3652 | before añc with its a elided, the prior member's last vowel is udātta | devadrī́cīm (udātta drī) |
| 3653 | the first vowel of a vocative is udātta | ágna índra váruṇa |
| 3654 | a vocative after a pada, not at the head of a pāda, is wholly anudātta | imáṃ me gaṅge yamune sarasvati |
| 3655 | a plural vocative before its qualifier optionally counts as absent | dévīḥ ṣaḷ urvīḥ |
| 3656 | a case-form before a vocative joins it for accent | drávatpāṇī śúbhaspatī (udātta śu) |
| 3657 | an anudātta after the yaṇ of an udātta or svarita becomes svarita | abhy abhi hi, khalapvyāśā |
| 3658 | a single substitute for an udātta and another vowel is udātta | vó 'śvāḥ, kvā́varam |
| 3659 | before a pada-initial anudātta that substitute is optionally svarita | vī̀daṃ, divī́yate, tè 'vadan |
| 3660 | in saṃhitā an anudātta after an udātta becomes svarita | agním īḷe (ī svarita) |
| 3661 | not when an udātta or svarita follows | prá yá ārúḥ |
| 3662 | a call from afar is ekaśruti | āgaccha bho māṇavaka |
| 3663 | a mantra in the rite is ekaśruti, except japa, nyūṅkha, sāman | agnír mūrdhā́ … monotone in the rite |
| 3664 | in the rite vauṣaṭ is optionally "higher" | — |
| 3665 | in the Veda ekaśruti is optional, by fixed distribution | saṃhitā: three accents; brāhmaṇa: monotone |
| 3666 | in the Subrahmaṇyā no ekaśruti; a svarita becomes udātta | subrahmaṇyó3m, gārgyó yajate |
| 3667 | there the svarita of deva and brahman becomes anudātta | dévā bráhmāṇa āgacchata |
| 3668 | in saṃhitā anudāttas after a svarita are ekaśruti | imám me gaṅge yamune sarasvati |
| 3669 | an anudātta before an udātta or svarita is sannatara | sarasvati śútudri |
| 3670 | the second copy of a doubled word is anudātta | divé-dive |
| 3671 | the last vowel of a root is udātta | gopāyátam, ási |
| 3672 | svap etc. and hiṃs: first vowel optionally udātta before a vowel-initial ending without iṭ | svápanti, híṃsanti |
| 3673 | the first vowel of a reduplicated stem is udātta there | dádati, dádhānā |
| 3674 | also before an ending with no udātta | dádhāsi |
| 3675 | nine reduplicated roots: the vowel before a pit ending is udātta | juhóti, jāgárṣi |
| 3676 | before an l-marked suffix, the vowel before it is udātta | cikī́rṣakaḥ |
| 3677 | before ṇamul, the first vowel is optionally udātta | lólūyaṃlolūyam |
| 3678 | before the kartṛyak, a vowel-final root's first vowel is optionally udātta | lū́yate / lūyáte |
| 3679 | in a caṅ-final root the penultimate is optionally udātta | cīkáratām / cīkarátām |
| 3680 | a ghañ-final word from karṣ, or one containing ā, is antodātta | pākáḥ, karṣáḥ |
| 3681 | the uñcha words are antodātta | yugé-yuge, bhakṣáḥ |
| 3682 | before śas the last vowel of catur is udātta | catúraḥ |
| 3683 | numerals before a jhal-initial ending: the penultimate is udātta | pañcábhiḥ |
| 3684 | in ordinary speech this is optional | pañcábhiḥ / pañcabhíḥ |
| 3685 | sarva before a case-ending is ādyudātta | sárve |
| 3686 | a word in a ñ- or n-marked affix is ādyudātta | paúṃsyā, cánaḥ |
| 3687 | pathin, mathin before a strong ending are ādyudātta | pánthāḥ |
| 3688 | a tavai form has first and last vowels udātta at once | dā́tavā́ u |
| 3689 | kṣaya "dwelling" is ādyudātta | své kṣáye |
| 3690 | jaya "instrument" is ādyudātta | jáyo 'śvaḥ |
| 3691 | vṛṣa and its open list are ādyudātta | vā́jebhiḥ, vā́ṇīḥ |
| 3692 | an upamāna used as a name is ādyudātta | cáñcā |
| 3693 | a two-vowel niṣṭhā name, first vowel not ā, is ādyudātta | dáttaḥ; not kṛtám |
| 3694 | śuṣka, dhṛṣṭa are ādyudātta | śúṣkam |
| 3695 | āśita meaning the agent is ādyudātta | ā́śitam |
| 3696 | rikta is optionally ādyudātta | ríktaḥ |
| 3697 | in the Veda juṣṭa, arpita are optionally ādyudātta | — |
| 3698 | in a mantra always (a dispensable rule) | júṣṭo dámūnāḥ, árpitam |
| 3699 | yuṣmad, asmad before ṅas: first syllable udātta | táva, máma |
| 3700 | also before ṅe | túbhyam, máhyam |
| 3701 | a two-vowel yat word is ādyudātta, except from nau | kā́myā; not nāvyā̀nām |
| 3702 | ṇyat words from īḍ, vand, vṛ, śaṃs, duh are ādyudātta | ī́ḍyaḥ, vā́ryam |
| 3703 | veṇu and indhāna are optionally ādyudātta | índhāno agním |
| 3704 | tyāga, rāga, hāsa, kuha, śvaṭha, kratha are optionally ādyudātta | tyā́gaḥ / tyāgáḥ |
| 3705 | in a feminine name, the ā before matup is udātta | śarā́vatī |
| 3706 | the final of avatī is udātta | vetravatī́ |
| 3707 | the final of īvatī is udātta too | ahīvatī́ |
| phiṭ 1 | the final vowel of a stem is udātta | uccaíḥ |
| phiṭ 2 | words meaning pāṭalā, apālaṅka, ambā, sāgara are antodātta | pāṭalā́, mātā́ |
| phiṭ 3 | house-words, not feminine, are antodātta | gehám; but śā́lā |
| phiṭ 4 | guda, not feminine, is antodātta | gudám; but gúdābhyaḥ |
| phiṭ 5 | in feminine-only words, a final after dh or y is udātta | chāyā́, antardhā́ |
| phiṭ 6 | kha-final words not beginning with ś or m are antodātta | sukhám; but múkham |
| phiṭ 7 | words in hiṣṭha, vatsara, ti, śat, tha are antodātta | baṃhiṣṭhá, saptatíḥ |
| phiṭ 8 | dakṣiṇa "skilful" is antodātta | dakṣiṇá |
| phiṭ 9 | dakṣiṇa naming a body part: first or last udātta | dákṣiṇaḥ / dakṣiṇáḥ |
| phiṭ 10 | in the Veda the same, in any sense | dákṣiṇaḥ / dakṣiṇáḥ |
| phiṭ 11 | in the Veda kṛṣṇa is antodātta unless it names an animal | kṛṣṇó nonāva vṛṣabháḥ |
| phiṭ 12 | kṛṣṇa as a proper name: antodātta optionally | ayáṃ vā kṛṣṇó aśvinā |
| phiṭ 13 | śukla, gaura are ādyudātta (optionally, Bhaṭṭoji) | sáro gauró yáthā piba |
| phiṭ 14 | in the Veda aṅguṣṭha, udaka, baka, vaśā are antodātta | aṅguṣṭhá, udakám |
| phiṭ 15 | pṛṣṭha is antodātta in the Veda, optionally in ordinary language | pṛṣṭhám / pṛ́ṣṭham |
| phiṭ 16 | arjuna as a grass is antodātta | arjuná; tree árjunaḥ |
| phiṭ 17 | arya "master" is antodātta | aryá |
| phiṭ 18 | āśā not meaning "direction" is antodātta | āśā́; ā́śābhyas (direction) |
| phiṭ 19 | āp-final nakṣatra names are antodātta | jyeṣṭhā́, anurādhā́ |
| phiṭ 20 | but not Kṛttikā | kṛ́ttikā |
| phiṭ 21 | ghṛta and its open list are antodātta | ghṛtám |
| phiṭ 22 | jyeṣṭha, kaniṣṭha of age are antodātta | jyeṣṭhá; but jyéṣṭhaḥ "best" |
| phiṭ 23 | bilva, tiṣya: final optionally svarita, else udātta | bilvà / bilvá |
| phiṭ 24 | heading: before phiṭ 69 the first syllable is udātta | — |
| phiṭ 25 | feminine-only words in a short vowel are ādyudātta | báliḥ, tánuḥ |
| phiṭ 26 | neuter-only words not in is are ādyudātta | váne; but sarpíḥ |
| phiṭ 27 | two-vowel grass and grain words are ādyudātta | kúśāḥ, tílāḥ |
| phiṭ 28 | numerals in n or r are ādyudātta | páñca |
| phiṭ 29 | a-final body parts and pronouns are ādyudātta | kárṇābhyām, víśvaḥ |
| phiṭ 30 | living beings: the first syllable before a ku is udātta | kā́kaḥ, vṛ́kaḥ |
| phiṭ 31 | a made thing: u before a khay is udātta | kandúkaḥ |
| phiṭ 32 | words in una, ṛ, van are ādyudātta | váruṇam, svásāram, pī́vānam |
| phiṭ 33 | colour words in ta, ṇa, ti, ni, t are ādyudātta | étaḥ, hárit |
| phiṭ 34 | short-vowel-final words: the first short vowel, not ṛ, is udātta | múniḥ |
| phiṭ 35 | akṣa not meaning dice is ādyudātta | tásya nā́kṣaḥ; but akṣaír |
| phiṭ 36 | ardha "a part" is ādyudātta | árdho grāmasya |
| phiṭ 37 | words meaning pītadru are ādyudātta | pī́tadruḥ, sáralaḥ |
| phiṭ 38 | grāma and its list are ādyudātta | grā́maḥ, sómaḥ |
| phiṭ 39 | a lup-name of the thing compared is ādyudātta | cáñcā |
| phiṭ 40 | not for particular trees and mountains, vyāghra, siṃha, mahiṣa | tālaḥ, meruḥ |
| phiṭ 41 | a king's name only if vṛddha | ā́ṅgaḥ; not aṅgāḥ |
| phiṭ 42 | many-vowel words with light final(s): a heavy syllable is udātta | kalyāṇaḥ, kolāhalaḥ |
| phiṭ 43 | feminine-only, colour and akṣu words are ādyudātta | mállikā, śyénī, tárakṣuḥ |
| phiṭ 44 | bird words: the light syllable before the final is udātta | kukkúṭaḥ, tittíriḥ |
| phiṭ 45 | season and living-being names: the expected udātta does not come | vasantaḥ, kṛkalāsaḥ |
| phiṭ 46 | grain words that are vṛddha and end in k, or end in ṣ, are ādyudātta | śyā́mākāḥ, rā́jamāṣāḥ |
| phiṭ 47 | country names ending in a vowel: first syllable udātta | kékayaḥ |
| phiṭ 48 | consonant-initial words in unconjoined la: first syllable udātta, or the one before the final | pálalam / palálam |
| phiṭ 49 | two-vowel words ending in ik: first syllable udātta; the second pāda ends | kṛ́ṣiḥ |
| phiṭ 50 | heading: "the second syllable is udātta", up to phiṭ 66 | lalā́ṭam |
| phiṭ 51 | heading: "of three-vowel words", up to phiṭ 57 | la-lā-ṭa |
| phiṭ 52 | three-vowel body-part words not beginning with ku, r, v: second syllable udātta | lalā́ṭam; not kapólaḥ |
| phiṭ 53 | three-vowel words beginning with m: second syllable udātta | maláyaḥ, makáraḥ |
| phiṭ 54 | pot-herb names beginning with ś: second syllable udātta | śītányā, śatápuṣpā |
| phiṭ 55 | three-vowel words in pa with a heavy first syllable: second syllable udātta | pādápaḥ, ātápaḥ |
| phiṭ 56 | three-vowel words in yuta, ani, aṇi: second syllable udātta | ayútam, dhamániḥ |
| phiṭ 57 | makara and eleven more: first syllable udātta, or the second | mákaraḥ / makáraḥ |
| phiṭ 58 | in the Veda, other words too: first or second, by usage | — |
| phiṭ 59 | kardama and the rest: first or second syllable udātta | — |
| phiṭ 60 | sugandhitejana: first, second or te udātta, by turns | súgandhitejanāḥ / sugándhitejanāḥ / sugandhitéjanāḥ |
| phiṭ 61 | neuters in phala: first or second syllable udātta | rā́jādanaphalam / rājā́danaphalam |
| phiṭ 62 | words ending in ya: the syllable before the last udātta | kulā́yaḥ, valáyaḥ |
| phiṭ 63 | words ending in tha: nā and a light syllable udātta | sanā́thā |
| phiṭ 64 | śiśumāra and four more: the syllable before the last, or the second, udātta | illustrative only: śiśumā́ra / śiśúmāra |
| phiṭ 65 | sāṅkāśya and three more: second syllable optionally udātta | — |
| phiṭ 66 | consonant-initial words in īṣā: first syllable optionally udātta; "the second" ends | hálīṣā, lā́ṅgalīṣā |
| phiṭ 67 | uśīra and twelve more: first syllable udātta | illustrative only: hṛ́dayam |
| phiṭ 68 | mahiṣī "wife", aṣāḍhā "brick": first syllable udātta; the third pāda ends | máhiṣī, áṣāḍhā |
| phiṭ 69 | śakaṭi, śakaṭī: each syllable udātta by turns; ādiḥ ends; the fourth pāda begins | śákaṭiḥ / śakáṭiḥ / śakaṭíḥ |
| phiṭ 70 | goṣṭhaja as a brāhmaṇa's name: each syllable by turns | góṣṭhajaḥ / goṣṭhájaḥ / goṣṭhajáḥ |
| phiṭ 71 | pārāvata: each syllable by turns except the penultimate | pā́rāvataḥ / pārā́vataḥ / pārāvatáḥ |
| phiṭ 72 | four words: each syllable by turns except dhū, ja, la, sthā | sthālī́pākaḥ / sthālīpā́kaḥ / sthālīpākáḥ |
| phiṭ 73 | kapikeśa, harikeśa in the Veda (operation not stated) | — |
| phiṭ 74 | nyaṅ and svar are svarita | nyàṅṅ uttānáḥ, svàḥ |
| phiṭ 75 | nyarbuda, vyalkaśa: first syllable svarita | nyàrbudāya, vyàlkaśā |
| phiṭ 76 | tilya and seven more: final syllable svarita, against SK 3701 | tilyàm, rājanyò |
| phiṭ 77 | in the Veda, bilva, bhakṣya, vīrya: final syllable svarita | bilvà |
| phiṭ 78 | tvat, tva, sama, sima are anudātta | utá tvaḥ páśyan |
| phiṭ 79 | sima in the Atharvaveda (and as found there): final udātta | simásmai |
| phiṭ 80 | particles are ādyudātta | svā́hā |
| phiṭ 81 | upasargas except abhi are ādyudātta | úpa, prá |
| phiṭ 82 | eva and the rest: final syllable udātta | evá, sahá |
| phiṭ 83 | vācā and the rest: both syllables udātta, against SK 3650 | — |
| phiṭ 84 | ca and the rest are anudātta | — |
| phiṭ 85 | yathā at the end of a verse-quarter is anudātta | táṃ nemím ṛbhávo yathā; else yáthā |
| phiṭ 86 | doubling for likeness: final of the latter member udātta | paṭupaṭúḥ |
| phiṭ 87 | other doublings: the latter member wholly anudātta; the phiṭ-sūtras end | divé-dive, práprāyám |
| 3708 | an affix is ādyudātta | agníḥ, kartávyam |
| 3709 | sup and pit affixes are anudātta | yajñásya, ná yó yúcchati |
| 3710 | after a cit affix the final of base + affix is udātta | anyaké, yaké |
| 3711 | the final of a cit taddhita is udātta, blocking the ñit accent | kauñjāyanā́ḥ |
| 3712 | the final of a kit taddhita is udātta | āgneyáḥ |
| 3713 | after tisṛ the final of jas is udātta | tisró dyā́vaḥ |
| 3714 | after a stem monosyllabic before su, endings from the third case are udātta | vācā́; not vā́cam |
| 3715 | the same, optionally, after an antodātta monosyllabic final member outside a nitya compound | paramavācā́ |
| 3716 | in the Veda, a non-sarvanāmasthāna ending after añc is udātta | dadhīcáḥ, pratīcó |
| 3717 | after ūṭh, idam, the pad-words, ap, pum, rai, div: non-sarvanāmasthāna endings udātta | padbhyā́m, divé-dive |
| 3718 | after long-ā aṣṭā, endings from śas on are udātta | aṣṭābhír |
| 3719 | after an antodātta śatṛ stem without num, the nadī and a vowel-initial ending from śas on are udātta | jānatī́, kṛṇvaté |
| 3720 | after a yaṇ standing for an udātta vowel, with a consonant before it, the nadī and the ending are udātta | netrī́, savitré |
| 3721 | not after the yaṇ of ūṅ or of a root; the ending becomes an independent svarita | brahmabandhvā̀, subhvè |
| 3722 | matup is udātta after a short-final antodātta stem and after nuṭ | abdimā́n, akṣaṇvántaḥ; not marútvān |
| 3723 | nām is optionally udātta after such a stem | sumatīnā́m |
| 3724 | in the Veda, nām after ṅī is udātta variously | abhibhañjatīnā́m; jáyantīnām |
| 3725 | a consonant-initial ending after ṣaṭ, tri, catur is udātta | ṣaḍbhír, tribhíṣ |
| 3726 | no udātta ending after go, śvan, an a-final nominative, rāṭ, aṅ, kruṅ, kṛt | gávām, tébhyaḥ |
| 3727 | no udātta jhal-initial ending after div | dyúbhiḥ; against divé |
| 3728 | after nṛ the same, optionally | nṛ́bhiḥ |
| 3729 | a t-marked affix is svarita | kvà |
| 3730 | a lasārvadhātuka is anudātta after tās, an anudāttet, a ṅit or an a-final in upadeśa; not after hnu, iṅ (nor vid, indh, khid) | várdhamānam; indhé |
| 3731 | in a sic form the first syllable is optionally udātta | yāsiṣṭáṃ (other option) |
| 3732 | in a word in thal with iṭ, the iṭ, the final or the first is optionally udātta | lúluvitha … luluvithá |
| 3733 | in a word ending in an r-marked affix, the penultimate is udātta | āhavanī́ye |
| 3734 | the final of a compound is udātta | yajñaśríyam |
| 3735 | in a bahuvrīhi the prior member keeps its accent (heading of 6.2) | citráśravastamaḥ; samapādáḥ |
| 3736 | in a tatpuruṣa seven kinds of prior member keep their accent | kiríkāṇaḥ, śastrī́śyāmā |
| 3737 | a colour word before a colour word other than eta | lóhitakalmāṣaḥ |
| 3738 | before gādha, lavaṇa in a measure | arítragādham, gólavaṇam |
| 3739 | before dāyāda, what is inherited | dhánadāyādaḥ |
| 3740 | before cira, kṛcchra, an obstruction | gámanaciram |
| 3741 | before pada "pretext" | uccārápadena |
| 3742 | before nivāta "shelter from the wind" | kuṭī́nivātam |
| 3743 | before śārada "new" | rájjuśāradam |
| 3744 | before adhvaryu, kaṣāya naming a class | kaṭhā́dhvaryuḥ, dauvārikákaṣāyam |
| 3745 | before sadṛśa, pratirūpa in likeness | pitṛ́sadṛśaḥ |
| 3746 | before a dvigu denoting a measure | prā́cyasaptasamaḥ |
| 3747 | before vāṇija, the destination or the goods | madrávāṇijaḥ, góvāṇijaḥ |
| 3748 | before mātra, upajña, upakrama, chāyā in a neuter compound | pā́ṇinyupajñam, íṣucchāyam |
| 3749 | before sukha, priya meaning "good for" | gámanasukham |
| 3750 | the same when affection is meant | brāhmaṇásukham |
| 3751 | before svāmin, the property | gósvāmī |
| 3752 | before pati meaning lordship | gṛhápatiḥ |
| 3753 | not for bhū, vāc, cit, didhiṣū before pati | bhūpatiḥ |
| 3754 | bhuvana before pati, optionally | bhúvanapatiḥ |
| 3755 | before āśaṅka, ābādha, nedīyas in supposition | gámanāśaṅkam |
| 3756 | before pūrva "formerly" | āḍhyápūrvaḥ |
| 3757 | before savidha and four others meaning "near" | madrásavidham |
| 3758 | vispaṣṭa and its list before a quality word | víspaṣṭakaṭukam |
| 3759 | in a karmadhāraya, an action noun before śra, jya, kan, avama, pāpa | gámanaśreṣṭham |
| 3760 | kumāra in a karmadhāraya | kumāráśramaṇā |
| 3761 | before pratyenas the first syllable of kumāra is udātta | kúmārapratyenāḥ |
| 3762 | the same before a guild-name, optionally | kúmāracātakāḥ / kumārácātakāḥ |
| 3763 | in a dvigu before an ik-final, a time-word, kapāla, bhagāla, śarāva | páñcakapālaḥ, páñcāratniḥ |
| 3764 | bahu in such a dvigu, optionally | bahvàratniḥ, bahúmāsyaḥ |
| 3765 | before diṣṭi, vitasti in a dvigu, optionally | páñcavitastiḥ |
| 3766 | a non-time seventh-case word before siddha, śuṣka, pakva, bandha | bhrā́ṣṭrapakvaḥ, cakrábandhaḥ |
| 3767 | pari, prati, upa, apa before what is excluded or a part of day or night | páritrigartam, úpapūrvāhṇam |
| 3768 | a plural dvandva of Andhaka and Vṛṣṇi kṣatriyas | śínivāsudevāḥ |
| 3769 | a numeral in a dvandva | tráyodaśa, dvā́daśa |
| 3770 | a dvandva of pupils named after teachers | pāṇinī́yarauḍhīyāḥ |
| 3771 | in the dvandvas of the kārtakaujapa list the prior member keeps its accent | kārtákaujapau, sā́varṇimāṇḍūkeyau |
| 3772 | mahat (antodātta) keeps its accent before vrīhi and nine others | mahā́vrīhiḥ, mahéṣvāsaḥ |
| 3773 | kṣullaka, and mahat, keep their accent before vaiśvadeva | kṣullakávaiśvadevam, mahā́vaiśvadevam |
| 3774 | uṣṭra keeps its accent before sādin, vāmin | úṣṭrasādī |
| 3775 | go keeps its accent before sāda, sādi, sārathi | gósādaḥ, gósārathiḥ |
| 3776 | seven named compounds and the open list dāsībhārādi keep the prior member's accent | kurúgārhapatam, dāsī́bhāraḥ, páṇyakambalaḥ (as a name) |
| 3777 | a fourth-case prior member keeps its accent before what serves its purpose | yū́padāru |
| 3778 | the same before artha | devā́rtham |
| 3779 | the same before a kta-final word | góhitam |
| 3780 | in a karmadhāraya before a kta-final word, a non-niṣṭhā prior member keeps its accent | śréṇikṛtāḥ, pūgákṛtāḥ |
| 3781 | a second-case prior member keeps its accent before a kta-final word (not abandoned) | kaṣṭáśritaḥ, grā́magataḥ |
| 3782 | a third-case prior member keeps its accent before a passive kta-final word | tvótāsaḥ, rudráhataḥ |
| 3783 | a gati next to the root keeps its accent before a passive kta | puróhitam |
| 3784 | the gati keeps its accent before a t-initial n-marked kṛt other than tu | prábhūtau, sáṅgatim |
| 3785 | before tavai the gati keeps its accent and the end of tavai is udātta, together | ánvetavā́ u |
| 3786 | a gati not ending in ik keeps its accent before añc + v | párāñcaḥ, párācaḥ |
| 3787 | ni and adhi do the same | nyàṅ, adhyaṅ |
| 3788 | īṣat optionally keeps its accent | īṣátkaḍāraḥ / īṣatkaḍāráḥ |
| 3789 | a quantity of gold optionally keeps its accent before dhana | dvisuvarṇadhanam |
| 3790 | prathama "new" optionally keeps its accent | prathamávaiyākaraṇaḥ |
| 3791 | katara, katama in a karmadhāraya optionally keep their accent | katarákaṭhaḥ |
| 3792 | ārya before brāhmaṇa, kumāra optionally keeps its accent (antasvarita) | āryàkumāraḥ |
| 3793 | rājan likewise | rā́jabrāhmaṇaḥ |
| 3794 | sixth-case rājan before pratyenas optionally keeps its accent | rā́japratyenāḥ |
| 3795 | before a kta-final word meaning "constantly", the prior member optionally keeps its accent | nítyaprahasitaḥ |
| 3796 | grāma before an artisan-word optionally keeps its accent | grā́manāpitaḥ |
| 3797 | rājan in praise before an artisan-word optionally keeps its accent | rā́janāpitaḥ |
| 3798 | heading: the first vowel of the prior member is udātta | — |
| 3799 | a seventh-case or hārin prior member is ādyudātta before a customary due (not haraṇa) | múkuṭekārṣāpaṇam, yā́jñikāśvaḥ |
| 3800 | ādyudātta in a compound denoting one engaged | góballavaḥ |
| 3801 | optionally ādyudātta before adhyakṣa | gávādhyakṣaḥ |
| 3802 | pāpa before an artisan-word is ādyudātta (optionally, per the Subodhinī) | pā́panāpitaḥ |
| 3803 | ādyudātta before a descendant, a pupil, māṇava, brāhmaṇa, in contempt | bhíkṣāmāṇavaḥ, bháyabrāhmaṇaḥ |
| 3804 | an ingredient is ādyudātta before maireya | mádhumaireyaḥ |
| 3805 | a name of a food is ādyudātta before what is meant for it | bhíkṣākaṃsaḥ |
| 3806 | ādyudātta before go, biḍāla, siṃha, saindhava as standard of comparison | dhā́nyagavaḥ, tṛ́ṇasiṃhaḥ |
| 3807 | ādyudātta before an aka-final member, for a livelihood | dántalekhakaḥ |
| 3808 | ādyudātta before an aka-final member, for an eastern game | úddālakapuṣpabhañjikā |
| 3809 | ādyudātta before an aṇ-final member, for one appointed | chátradhāraḥ |
| 3810 | the same for an artisan, unless the aṇ is after kṛ | tántuvāyaḥ |
| 3811 | the same in a name, unless the aṇ is after kṛ | tántuvāyo nāma kṛmiḥ |
| 3812 | go, tanti, yava are ādyudātta before pāla | gópālaḥ, yávapālaḥ |
| 3813 | ādyudātta before a ṇini-final member | púṣpahārī |
| 3814 | a comparison is ādyudātta before ṇini only if the root means sound | úṣṭrakrośī, dhvā́ṅkṣarāvī |
| 3815 | yuktārohin and the rest are ādyudātta | yúktārohī, kṣī́rahotā |
| 3816 | a long-vowel-final prior member, and kāśa etc., are ādyudātta before ja | kúṭījaḥ, váṭajaḥ |
| 3817 | before ja, a many-vowelled prior member has its next-to-last vowel udātta | upasárajaḥ, āmalákījaḥ |
| 3818 | ādyudātta before grāma, unless the prior member names the dwellers | mállagrāmaḥ, dévagrāmaḥ |
| 3819 | ādyudātta before ghoṣa and the rest | dā́kṣighoṣaḥ |
| 3820 | chātri and the rest are ādyudātta before śālā, even in the neuter | chā́triśālā, chā́triśālam |
| 3821 | a non-vṛddha prior member is ādyudātta before prastha (not karkī etc.) | índraprasthaḥ |
| 3822 | mālā and the rest (vṛddha) are ādyudātta before prastha | mā́lāprasthaḥ, śóṇāprasthaḥ |
| 3823 | before nagara, a pūrvapada other than mahat, nava is ādyudātta, unless the town is northern | bráhmanagaram |
| 3824 | before arma, a 2- or 3-vowel a-final pūrvapada is ādyudātta | gúptārmam, kúkkuṭārmam |
| 3825 | bhūta, adhika, saṃjīva, madra, aśman, kajjala are not ādyudātta before arma; vārttika: divodāsa etc. in the Veda | bhūtārmam; dívodāsāya |
| 3826 | heading: the pūrvapada is antodātta, up to SK 3845 | — |
| 3827 | sarva is antodātta when it means the totality of a quality | sarváśvetaḥ |
| 3828 | in a name, before giri or nikāya | añjanā́giriḥ, mauṇḍínikāyaḥ |
| 3829 | before kumārī, when an age is conveyed | vṛddhákumārī |
| 3830 | before udaka, when a mixture is meant | guḍódakam / guḍòdakam |
| 3831 | before a dvigu, in the name of a sacrifice | gargátrirātraḥ |
| 3832 | before sabhā, in a compound neuter by name | gopālásabham, strī́sabham |
| 3833 | before pura, in the eastern country | devadattápuram |
| 3834 | before pura, in a compound beginning with ariṣṭa or gauḍa | ariṣṭápuram, ariṣṭāśritápuram |
| 3835 | hāstina, phalaka, mārdeya are not antodātta before pura | hāstinapuram |
| 3836 | kusūla, kūpa, kumbha, śāla are antodātta before bila | kusūlábilam |
| 3837 | a direction word before a village, country, narrative, or cānarāṭa | pūrvápañcālāḥ |
| 3838 | a direction word before a word for a pupil with the teacher subordinate | pūrvápāṇinīyāḥ |
| 3839 | sarva and direction words before an uttarapada with vṛddhi of the uttarapadasya heading | sarvápāñcālakaḥ |
| 3840 | viśva is antodātta in a bahuvrīhi that is a name; opens the bahuvrīhi heading | viśvákarmā |
| 3841 | before udara, aśva, iṣu, in a bahuvrīhi that is a name | vṛkódaraḥ, haryàśvaḥ |
| 3842 | the same in censure; anudaraḥ goes to SK 3906 | ghaṭódaraḥ; anudaráḥ |
| 3843 | a nadī-final pūrvapada before bandhu | gārgī́bandhuḥ |
| 3844 | a niṣṭhā-final pūrvapada with an upasarga, optionally | pradhautápādaḥ / prádhautapādaḥ |
| 3845 | heading: the uttarapada is ādyudātta, ādi up to SK 3871 | — |
| 3846 | karṇa after a colour or mark word, in a bahuvrīhi | śuklakárṇaḥ |
| 3847 | karṇa in a name or a comparison | maṇikárṇaḥ, gokárṇaḥ |
| 3848 | kaṇṭha, pṛṣṭha, grīvā, jaṅghā in a name or a comparison | śitikáṇṭhaḥ, aśvagrī́vaḥ |
| 3849 | śṛṅga for a stage of life, a name or a comparison | udgataśṛ́ṅgaḥ |
| 3850 | after nañ, jara, mara, mitra, mṛta are ādyudātta | ajáram, amítram |
| 3851 | after su, a man- or as-final uttarapada, not loman, uṣas; with kap, SK 3907 | sukármāṇaḥ; sukarmákaḥ |
| 3852 | after su, kratu and the rest | sukrátuḥ |
| 3853 | in the Veda, after su, a two-vowel ādyudātta uttarapada stays so | sváśvāḥ, suráthā̃ |
| 3854 | in the Veda, after su, vīra and vīrya; a jñāpaka on SK 3701 | suvī́ryasya |
| 3855 | in an avyayībhāva, kūla, tīra, tūla, mūla, śālā, akṣa, sama | upakū́lam, upā́kṣam |
| 3856 | in a dvigu, kaṃsa, mantha, śūrpa, pāyya, kāṇḍa | dvikáṃsaḥ |
| 3857 | śālā in a neuter tatpuruṣa; tatpuruṣe runs on | brāhmaṇaśā́lam |
| 3858 | kanthā in a neuter tatpuruṣa | sauśamikántham |
| 3859 | before kanthā, the pūrvapada cihaṇa etc. is ādyudātta | cíhaṇakantham |
| 3860 | cela, kheṭa, kaṭuka, kāṇḍa when contempt is expressed | putracélam |
| 3861 | cīra as standard of comparison | vastracī́ram |
| 3862 | palala, sūpa, śāka when a mixture is expressed | ghṛtapálalam |
| 3863 | kūla, sūda, sthala, karṣa in a name | dākṣikū́lam |
| 3864 | rājya in a non-karmadhāraya tatpuruṣa; vārttika: the avyaya accent prevails | brāhmaṇarā́jyam; kurājyam |
| 3865 | vargya and the rest in a non-karmadhāraya tatpuruṣa | arjunavárgyaḥ |
| 3866 | putra after a word for a male | dāśakipútraḥ |
| 3867 | not after names of a teacher, king, priest, in-law or kinsman | ācāryaputraḥ |
| 3868 | cūrṇa and the rest after a sixth-case word for a non-living thing | mudgacū́rṇam |
| 3869 | the six words kāṇḍa … kūla under the same condition, without their own | darbhakā́ṇḍam, nadīkū́lam |
| 3870 | kuṇḍa meaning a forest; the ādi heading ends | darbhakúṇḍam |
| 3871 | heading prakṛtyā: an uttarapada meaning bhagāla keeps its accent | kumbhībhagā́lam |
| 3872 | after śiti in a bahuvrīhi, a never-many-vowelled uttarapada keeps its accent, not bhasat | śitipā́daḥ; śítibhasat |
| 3873 | a kṛt-final uttarapada after a gati, kāraka or upapada keeps its accent | prakā́rakaḥ, uccaiḥkā́ram |
| 3874 | in vanaspati and the rest, both members keep their accents | vánaspátim, bṛ́haspátim |
| 3875 | in a dvandva of deity names both members keep their accent | índrāváruṇau (í, vá) |
| 3876 | not if the uttarapada begins with an anudātta, except pṛthivī, rudra, pūṣan, manthin | indrāgníbhyām (gní); dyā́vāpṛthivī́ |
| 3877 | heading antaḥ: the uttarapada's final syllable is udātta | — |
| 3878 | after a gati, kāraka or upapada: tha, atha, ghañ, kta, ac, ap, itra, ka finals are antodātta | prabhedáḥ, puruṣṭutáḥ |
| 3879 | a kta member after su or an upamāna is antodātta | sukṛtá, śaśaplutáḥ |
| 3880 | in a name, a kta member after a gati, kāraka or upapada is antodātta, except ācitādi | upahūtáḥ |
| 3881 | pravṛddha and the rest (not names) are antodātta | pravṛddháḥ, prayutáḥ |
| 3882 | niyama: after a kāraka, only datta and śruta, in a blessing | devadattáḥ, viṣṇuśrutáḥ |
| 3883 | "done by one in such a state": the kta member is antodātta | suptapralapitám |
| 3884 | after a kāraka, an ana member of action or object is antodātta | payaḥpānám, rājabhojanā́ḥ |
| 3885 | after a kāraka: man, ktin, vyākhyāna, śayana, āsana, sthāna, yājakādi, krīta | pāṇinikṛtíḥ, gokrītáḥ |
| 3886 | puṇya after a seventh-case word is antodātta | adhyayanapuṇyám |
| 3887 | after a third-case word, "deficient" and kalaha are antodātta | māṣonám, vākkalaháḥ |
| 3888 | after a third-case word, miśra without an upasarga, except alliance | tilamiśrā́ḥ |
| 3889 | after nañ negating a quality: taddhitas of sampādin, arha, hita, alam | akārṇaveṣṭakikám |
| 3890 | the same for taddhita ya and yat not meaning "for that" | apāśyā́, adantyám |
| 3891 | after nañ, ac and ka members meaning inability | apacáḥ |
| 3892 | the same in reviling | apacó jālmaḥ |
| 3893 | after nañ, a name used in abuse | adevadattáḥ |
| 3894 | after nañ: kṛtya, ukañ, iṣṇuc members and cārvādi | akartavyáḥ, acārúḥ, arājā́ (Veda) |
| 3895 | after nañ, optionally: tṛn, anna, tīkṣṇa, śuci | akartā́, aśucí |
| 3896 | heading bahuvrīhau; prathama or an ordinal after idam, etad, tad, counting acts | idaṃprathamáḥ, idaṃprathamákaḥ |
| 3897 | stana after a numeral | dvistanī́, catuḥstanā́ |
| 3898 | optional in the Veda | dvistanāṃ karoti |
| 3899 | mitra and ajina in a name; vārttika: not for a seer | devamitráḥ, kṛṣṇājinám; viśvā́mitraḥ |
| 3900 | antara "screen" after a word for what intervenes | vastrāntaráḥ |
| 3901 | mukha as a body part | gauramukháḥ |
| 3902 | not after an avyaya, a direction word, go, mahat, sthūla, muṣṭi, pṛthu, vatsa | gómukhaḥ, prā́ṅmukhaḥ |
| 3903 | optionally after a niṣṭhā word or an upamāna | prakṣālitamukháḥ, siṃhamukháḥ |
| 3904 | a kta member after a genus, time or sukhādi, not a covering, not kṛta, mita, pratipanna | sāraṅgajagdháḥ, māsajātáḥ |
| 3905 | jāta there is optionally antodātta | dantajātáḥ / dántajātaḥ |
| 3906 | after nañ and su in a bahuvrīhi | avrīhíḥ, sumāṣáḥ |
| 3907 | with kap, the syllable before kap is udātta | abrahmabandhū́kaḥ |
| 3908 | with a short-final member, the syllable before the last (niyama) | avrī́hikaḥ, sumā́ṣakaḥ; ajñakáḥ |
| 3909 | after bahu of the uttarapada's plurality, as after nañ | bahuvrī́hikaḥ, bahumítrakaḥ |
| 3910 | not guṇa and the rest denoting parts | bahuguṇā rajjuḥ |
| 3911 | after an upasarga, a permanent body part other than parśu | prapṛṣṭháḥ, pralalāṭáḥ |
| 3912 | in any compound, vana after an upasarga; ends bahuvrīhau | pravaṇé |
| 3913 | vana after antar | antarvaṇáḥ |
| 3914 | anta after an upasarga | paryantáḥ, samantáḥ |
| 3915 | not after ni and vi | nyàntaḥ, vyàntaḥ |
| 3916 | after pari, what is on both sides, and maṇḍala | parikūlám, parimaṇḍalám |
| 3917 | after pra, a non-body-part member in a name | pragṛhám; prápadam |
| 3918 | nirudaka and the rest | nirudakám, nirupalám |
| 3919 | mukha after abhi | abhimukhám, abhimukhā́ śālā |
| 3920 | mukha after apa | apamukhám |
| 3921 | after apa: sphiga, pūta, vīṇā, añjas, adhvan, kukṣi, plough-words, nāman | apādhvā́, apanāmá |
| 3922 | after adhi, what stands above | adhidantáḥ |
| 3923 | after anu, a non-principal member, and kanīyas | anujyeṣṭháḥ, anukanīyā́n |
| 3924 | after anu, puruṣa "mentioned afterwards" | anupuruṣáḥ |
| 3925 | after ati, a non-kṛt member and pada; vārttika: when a root is elided | atyaṅkuśáḥ, atipadā́, atikārukáḥ |
| 3926 | after ni, when concealment is not meant | nimūlám, nyakṣám |
| 3927 | after prati, in a tatpuruṣa, aṃśu and the rest are antodātta | pratyaṃśúḥ, pratirājā́ |
| 3928 | after upa, in a tatpuruṣa, a two-vowel uttarapada and ajina are antodātta, except gaurādi | upadeváḥ, upājinám |
| 3929 | after su, in censure, the uttarapada is antodātta | supratyavasitáḥ |
| 3930 | in the tatpuruṣa utpuccha, optionally antodātta | utpuccháḥ |
| 3931 | bahuvrīhi of dvi/tri + pād, dat, mūrdhan: optionally antodātta | dvipā́t, dvimūrdháḥ |
| 3932 | saktha, not after a kra-final word, optionally antodātta | gaurasaktháḥ |
| 3933 | in the Veda, the first syllable of saktha variously udātta; the vārttika on interchange | ajisáktham, viśvā́yuḥ |
| 3934 | after a verb, gotrādi words are wholly anudātta in contempt and repetition | pacati gotram |
| 3935 | a verb after a non-verb is wholly anudātta (nighāta) | agním īḷe |
| 3936 | a luṭ verb is not nighāta | śvaḥ kartā́ |
| 3937 | with yad, yadi, hanta, kuvit, net, cet, caṇ, kaccit, yatra: not nighāta | yád agne syā́m |
| 3938 | with naha, in a pratyārambha: not nighāta | naha bhokṣyáse |
| 3939 | with satya, in a question: not nighāta | satyaṃ bhokṣyáse? |
| 3940 | with aṅga, without contrariness: not nighāta | aṅga kurú |
| 3941 | with hi: not nighāta | ā́ hí ṣmā yā́ti |
| 3942 | in the Veda, with hi, even several verbs in mutual expectancy | vádati … yunā́ti |
| 3943 | with yāvat or yathā: not nighāta | yáthā cit káṇvam ā́vatam |
| 3944 | in praise, a verb right after yāvat/yathā is anudātta after all | yāvat pacati śobhanam |
| 3945 | the same with an upasarga in between | yāvat prápacati śobhanam |
| 3946 | with tu, paśya, paśyata, aha, in praise: not nighāta | … eriré |
| 3947 | with aho, in praise: not nighāta | aho devadattaḥ pácati śobhanam |
| 3948 | with aho, in the remaining case: optionally anudātta | kariṣyati / kariṣyáti |
| 3949 | with purā, in haste: optionally anudātta | vídyotate / vidyótate |
| 3950 | with nanu, asking permission: not nighāta | nanu gácchāmi bhoḥ |
| 3951 | with kim asking about the action, no upasarga, not negated: not nighāta | kiṃ dvijaḥ pácati |
| 3952 | the same when kim is not used: optionally | devadattaḥ pacati / pácati |
| 3953 | lṛṭ with ehi manye, in jest only (niyama) | ehi mánye … bhokṣyáse |
| 3954 | with jātu that has nothing before it | jā́tu bhokṣyáse |
| 3955 | with a form of kim + cit, nothing before it | kaścid bhuṅkte |
| 3956 | right after āho/utāho, nothing before them | āho bhuṅkte |
| 3957 | separated from āho/utāho: optionally anudātta | āho devaḥ pacati / pácati |
| 3958 | lṛṭ with the loṭ of a verb of motion, kāraka not wholly different | āgaccha deva grāmaṃ drakṣyási |
| 3959 | loṭ with the loṭ of a verb of motion | āgaccha deva grāmaṃ páśya |
| 3960 | the same loṭ with an upasarga, not first person: optionally | praviśá / práviśa |
| 3961 | loṭ with hanta, with an upasarga, not first person: optionally | hanta praviśá / práviśa |
| 3962 | a vocative after ām, one pada between, addressee not near: not anudātta | ām pacasi dévadatta3 |
| 3963 | in the Veda, a verb followed by yad, hi, tu (niyama) | uśánti hí |
| 3964 | a verb followed by cana, cit, iva, gotrādi, a taddhita, an āmreḍita, no gati before | devaḥ pácati cana |
| 3965 | a verb followed by ca, vā, ha, aha, eva, no gati before | pácati ca khā́dati ca |
| 3966 | with ca/vā: the first verb is not anudātta | itó vā sātím ī́mahe |
| 3967 | with ha, in a breach of conduct: the first verb | svayaṃ ha rathena yāti3 |
| 3968 | with aha, in a commission or a breach of conduct: the first verb | tvam aha grāmaṃ gaccha |
| 3969 | ca/aha left out, eva restrictive: the first verb | deva eva grāmaṃ gacchatu … |
| 3970 | ca and the rest left out: the first verb optionally | vrīhibhir yájeta … |
| 3971 | in the Veda, with vai/vāva: the first verb optionally | ahar vai devānām āsīt |
| 3972 | in the Veda, with eka/anya in one sense: the first verb | táyor anyáḥ … átti |
| 3973 | after a pada containing yad: always not anudātta | yo bhuṅkte |
| 3974 | after kāṣṭha and the rest, the praised word is anudātta | kāṣṭhādhyāpakaḥ |
| 3975 | the praised verb, with its upasarga, is anudātta | yat kāṣṭhāṃ prapacati |
| 3976 | in censure of the action, before a non-gotrādi subanta, the verb is anudātta | pacati pūti |
| 3977 | a gati before a gati is anudātta | abhyúddharati |
| 3978 | a gati before a verb with an udātta is anudātta; then the svara-saṃcāra | yat prapácati; agním īḷe puróhitam |