After this chapter you will be able to read a rule of the Aṣṭādhyāyī the way Pāṇini meant it to be read. None of the thirteen rules below changes a sound. Each one fixes how some other rule is to be understood: which sound a rule that merely says "guṇa" or "short" works on; where an added sound (an augment) is put; what a rule means when it says "of ik", "when a vowel follows" or "after ud"; which of several possible substitutes is chosen, and how much of the old sound it displaces; and, when two rules both want to act on the same word, which one goes first. A rule about how other rules are read is a paribhāṣā, and this is the chapter of paribhāṣās. Every rule is shown at work on a concrete word.
mid + guṇa → med-, śrīpā → śrīpamMany rules of the grammar say only "guṇa of mid" or "short for a neuter stem", without saying which sound of the word is meant. The two rules of this group supply the missing word. Wherever a rule speaks of guṇa or vṛddhi, the change falls on an ik vowel; wherever a rule speaks of short, long or protracted, the change falls on a vowel. So the i of mid becomes e, and the final ā of śrīpā, used as a neuter, becomes a.
Rule. Wherever a rule prescribes guṇa or vṛddhi by using those very words, the word ikaḥ, "of ik", presents itself in that rule. The guṇa or vṛddhi therefore comes in place of an ik vowel (i, u, ṛ, ḷ) of the item concerned. Bhaṭṭoji's word for "presents itself" is upatiṣṭhate: the genitive ikaḥ becomes part of the other rule's sentence, and each rule construes it in its own way. The words guṇa and vṛddhi are carried into this sūtra by anuvṛtti from SK 16 and SK 17.
Example.
mid (a root) → the rule mider guṇaḥ (A 7.3.82) says "guṇa of mid", naming no sound → ikaḥ presents itself: the guṇa falls on the ik of mid, its i → the i is replaced by its guṇa, e (why e and not a is settled at SK 39) → med-.
Likewise the rule sārvadhātukārdhadhātukayoḥ (A 7.3.84), which prescribes guṇa for a root before certain affixes: its working on the i of ci and the u of stu is shown at SK 39 (cetā, stotā).
Applies when / blocked by. Only where the words guṇa or vṛddhi actually prescribe something. Not at a mere restatement: vṛddhir yasyācām ādis tad vṛddham (A 1.1.73) uses the word vṛddhi to describe a word, not to prescribe a change, and a paribhāṣā does not present itself at a restatement. Not where the words are absent: tyadādīnām aḥ (A 7.2.102) prescribes an a without calling it guṇa, so ikaḥ does not come; which sound that a replaces is settled by SK 42 instead.
Watch out. The sūtra's ikaḥ stands for its own form. What is inserted into the other rules is that genitive word itself, so read "guṇa of mid" as "guṇa of the ik of mid".
Rule. And wherever a rule prescribes an operation by using the words hrasva, dīrgha or pluta, the word acaḥ, "of a vowel", presents itself in that rule. The shortening, lengthening or protraction therefore falls on a vowel (an ac) of the item concerned. The words ac and hrasva-dīrgha-pluta come by anuvṛtti from SK 4; acaḥ, like ikaḥ in SK 34, stands for its own form.
Example.
śrīpā used as a neuter stem → the rule hrasvo napuṃsake prātipadikasya (A 1.2.47) says "short for a neuter stem", naming no sound → acaḥ presents itself: the shortening falls on a vowel, here the final ā → ā is replaced by short a → śrīpa-, seen in the neuter word śrīpam.
Applies when / blocked by. Only where the word hrasva (or dīrgha, pluta) is actually heard. In diva ut (A 6.1.131), which prescribes the short u by naming the sound, the paribhāṣā does not act. And the operation needs a vowel to fall on: in supāt brāhmaṇakulam the neuter stem supād does not end in a vowel, so no shortening happens.
bhav·i·tā, vār·ī·ṇiAn augment is a sound added to an item; nothing is replaced. The grammar states each augment with a marker: iṬ has the marker ṭ, kuK the marker k, nUM the marker m. The two rules of this group read those markers as instructions about position. A ṭ-marked augment goes at the front of its item, a k-marked one at the end, and an m-marked one after the item's last vowel. So iṬ gives bhav·i·tā, and nUM placed after the i of vāri gives, in the end, vārīṇi.
Rule. An element marked with ṭ (a ṭit) and an element marked with k (a kit), prescribed for some item, are respectively the initial part and the final part of that item. "Respectively" means first with first, second with second; this pairing is called yathāsaṃkhya and is supplied by the rule yathāsaṃkhyam anudeśaḥ samānām (A 1.3.10). In the sūtra's ṭakitau the a after ṭ is only for pronunciation; the markers are ṭ and k.
Example.
bhavitā → the rule ārdhadhātukasyeḍ valādeḥ (A 7.2.35) prescribes the augment iṬ for the affix tṛC (seen here as -tā) → the ṭ of iṬ is a marker, so iṬ is ṭit → the i becomes the initial part of that affix → bhav·i·tā.
prāṅ + ṣaṣṭha → the rule ṅṇoḥ kuk ṭuk śari (A 8.3.28) prescribes the augments kuK and ṭuK for a word ending in ṅ or ṇ → the k of kuK is a marker, so kuK is kit → the k becomes the final part of prāṅ → prāṅ·k·ṣaṣṭha.
Applies when / blocked by. Bhaṭṭoji's commentators make two points. First, this rule is an exception — an apavāda, a special rule that sets aside a general rule within the special rule's own field — only to the rule that immediately follows it in the Aṣṭādhyāyī, ṣaṣṭhī sthāneyogā (SK 38). The maxim behind this, purastād apavādā anantarān vidhīn bādhante nottarān: an exception stated earlier blocks only the rules that immediately follow it, not later ones. Second, this rule does not move affixes. A ṭit affix such as Ṭa (careṣ ṭaḥ, A 3.2.16) or Ṭak (gāpoṣ ṭak, A 3.2.8) still comes after its base, because the later rule paraś ca (A 3.1.2), read with pratyayaḥ (A 3.1.1), says that an affix follows; being later, that rule prevails.
Watch out. The sūtra pairs two things with two things, and Bhaṭṭoji's commentators note that such a pairing is not how a compound works in ordinary speech; the grammar allows it here, as in eco 'yavāyāvaḥ (A 6.1.78), by first taking the two pairs as wholes and then, by the yathāsaṃkhya rule, matching member to member.
Rule. An augment marked with m (a mit) is placed after the last among the vowels of the item it is prescribed for, and it is the final part of that item. Bhaṭṭoji stresses that the sūtra's acaḥ is a genitive, "among the vowels": look at all the vowels of the item, pick the last, and put the augment after it. The word anta, "final part", is carried forward from SK 36 — only that half of ādyantau, because the svarita rests on that half alone (see SK 46).
Example.
vāri + the neuter plural ending → the augment nUM is prescribed for the stem (aṅga, the base to which the ending is being added) → nUM is mit → its n goes after the last vowel of vāri, the i, as the stem's final part → vāri·n- → the stem now ends in n, so sarvanāmasthāne cāsambuddhau (A 6.4.8), which lengthens the vowel of an n-final stem before such an ending, can act → vārīn- → the finished word is vārīṇi (the change of n to ṇ is taught elsewhere).
vaha + liha- (formed with the affix KHaŚ by vahābhre lihaḥ, A 3.2.32) → the augment mUM is prescribed by arur dviṣad- (A 6.3.67) → mUM is mit → its m goes after the last vowel of vaha, as its final part → vaham·liha- → the pada vaham now ends in m, so mo 'nusvāraḥ (A 8.3.23), which turns a pada-final m into anusvāra, can act → vahaṃliha- (nominative vahaṃlihaḥ).
Applies when / blocked by. The genitive reading of acaḥ matters. If acaḥ were read as an ablative, "after the last vowel", the sūtra could not work in rules such as śe mucādīnām (A 7.1.59), where no last vowel is available for it. Read as a genitive of specification, "among the vowels", it can; scholars discuss the details.
Watch out. The mit belongs to the whole item — not to what follows it, and not to nothing. If it were the initial part of the next element, vāri would not end in n, and there would be no lengthening in vārīṇi. If it belonged to nothing, vaham would not be a pada ending in m, and there would be no anusvāra in vahaṃlihaḥ. The Mahābhāṣya puts it in two words: samudāyabhakto mit, "the mit belongs to the aggregate". Bhaṭṭoji will say the same of nUM later: prescribed for the stem, it is a part of the stem.
iko yaṇ aci → dadhy atraThe rule iko yaṇ aci (A 6.1.77) has three words: ikaḥ, a genitive, "of ik"; yaṇ, the thing prescribed; aci, a locative, "when a vowel". The three rules of this group say what such endings mean in a sūtra. A genitive means "in place of". A locative points at the element immediately before the thing named. An ablative points at the element immediately after it. Read with them, iko yaṇ aci says: "in place of an ik standing immediately before a vowel, a yaṇ." That is why dadhi + atra gives dadhy atra: the i goes, and y stands where it was.
Rule. A genitive in a sūtra, when its relation has not been otherwise determined, is connected with sthāna: "of X" means "in the place of X". And sthāna here means prasaṅga: the occasion on which X was due to occur. Two names go with this. What was due to occur and is turned back is the sthānin; what turns it back is the ādeśa, the substitute. So in iko yaṇ aci the genitive ikaḥ means: where an ik was due, a yaṇ occurs instead.
Example.
dadhi + atra → the i is an ik, and it is due to stand before the vowel a → ikaḥ says "in place of the ik": the i is the sthānin → a yaṇ comes as its ādeśa (which of the four, SK 39) → the i is removed and y stands in its place → dadhy atra.
Applies when / blocked by. Only for a genitive whose relation is not already fixed. In ūdupadhāyā gohaḥ (A 6.4.89) the word upadhā, "penult", fixes the genitive gohaḥ as a partitive genitive: "ū for the penult of goh". There is no sthāna-sense there; if there were, ū would come in place of the final sound of goh (SK 42), and in place of any penult whatever. A paribhāṣā restricts only where nothing has yet been decided.
Watch out. In ordinary speech sthāna means a place one occupies, as in "Kailāsa is the sthāna of Śiva". If the sūtra meant that, yaṇ would come "in the location of the ik", and the ik would stay put with y beside it. That is why Bhaṭṭoji glosses sthāna as prasaṅga. His commentators illustrate the sense from a ritual injunction — "where darbha grass would be in place, strew śara reeds": the darbha is displaced by the reeds, and just so the i by the y. They also cite the Mahābhāṣya: between one word and another there are only three relations, immediate sequence, proximity and prasaṅga, and this sūtra chooses the third. Scholars discuss how exactly the word sthāne is built into the sūtra.
Rule. When a rule's condition is stated in the locative — "when X" — the operation applies to the element immediately before X, with no other sound between. The sūtra's tasmin does not mean itself; it imitates the locative words of other sūtras, aci, hali and the like. Nirdiṣṭa means "uttered without a break" (nir, without a gap; diś, to utter): the X must be uttered, in actual speech, right after the element that changes.
Example.
dadhi + atra → the condition of iko yaṇ aci is aci, "when a vowel" → the a of atra is uttered right after the i → the operation falls on the element immediately before that a, the i → dadhy atra.
samidham → the a of the ending stands after the i of -idh-, but the dh lies between → that i is not immediately before the a → no yaṇ for that i.
dadhi udakam → the u of udakam stands next to the vowel i, but after it → a locative condition looks backward only → the u is not replaced on account of the i.
Applies when / blocked by. Without this rule, aci could only be read as "when a vowel is present" (a locative absolute, sati-saptamī), because one vowel cannot literally be the place of another sound; and then yaṇ would apply across a gap and forwards, giving the two wrong results above. Only separation by another sound is barred, not separation by time. A pause is excluded separately, by the saṃhitā heading (SK 28); rules outside that heading, such as ānaṅ ṛto dvandve (A 6.3.25) and devatādvandve ca (A 6.3.26), act even across a pause, as when agnāviṣṇū is recited with a break (avagraha) between its members. The rule does not operate where the locative names a meaning rather than a sound, as in atiśāyane tamap (A 5.3.55); nor in certain rules of syntax such as kartṛkarmaṇoḥ kṛti (A 2.3.65), in deference to actual usage. Scholars discuss these cases, and whether the word nirdiṣṭa is needed at all.
Rule. When a rule's condition is stated in the ablative — "after X" — the operation applies to the element immediately after X, with no other sound between. Tasmāt imitates the ablative words of sūtras such as dvyantarupasargebhyo 'pa īt (A 6.3.97) and udaḥ sthāstambhoḥ pūrvasya (A 8.4.61); nirdiṣṭa is carried forward from SK 40. The rule is a restriction (niyama): only what follows, and only when nothing intervenes.
Example.
agnim īḻe → the rule tiṅ ṅatiṅaḥ (A 8.1.28) takes away the accent of a verb (a tiṅ-ending word) after a word that is not a verb → the verb īḻe follows agnim → it loses its accent (this loss is called nighāta).
īḻe agnim → the verb comes first → nothing follows the non-verb → the rule does not touch it.
ut·pra·sthānam → udaḥ sthāstambhoḥ pūrvasya (A 8.4.61) prescribes, for sthā and stambh after ud, a substitute savarṇa with the preceding sound → here pra stands between ud and sthā → something intervenes → no change.
dvi + ap → dvyantarupasargebhyo 'pa īt (A 6.3.97) gives ī to ap after dvi, antar or an upasarga → ap follows dvi directly → the rule applies to ap; which sound of ap the ī replaces is settled by SK 44.
vāg ghariḥ, cetāOnce the sthānin is known, there are often several candidates for the substitute. "A yaṇ" is any of y, v, r, l; "guṇa" is any of a, e, o; "the sound savarṇa with g" is any of the five sounds of the k-class. This group has one rule, which picks the candidate most like the sthānin. For the h of hariḥ after vāg the choice is gh, which shares every effort with h; for the guṇa of i it is e, which shares its place.
Rule. When something is due to be replaced (sthāne, again in the sense of prasaṅga), and more than one substitute is in prospect for the one sthānin, the one most like the sthānin (antaratama, "most similar") becomes the substitute. Likeness is measured in four ways: sthāna, place of articulation; artha, meaning; guṇa, effort of articulation; pramāṇa, measure, one mātrā or two. Bhaṭṭoji adds a paribhāṣā: where likeness of more than one kind is possible, likeness by sthāna is the stronger.
Example. One case of each kind.
Place: dadhi + atra → the sthānin i is palatal → of the yaṇ sounds y, v, r, l, the palatal one is y → dadhy atra.
Meaning: kroṣṭu → tṛjvat kroṣṭuḥ (A 7.1.95) gives the u-final word kroṣṭu a substitute ending in the affix tṛC → among such words the one with the same meaning, kroṣṭṛ, is chosen → kroṣṭṛ-.
Effort: vāg + hariḥ → the sthānin h has the external efforts ghoṣa, nāda, saṃvāra and mahāprāṇa (chapter 1, SK 12) → the substitute must be savarṇa with the preceding g, so it is one of k, kh, g, gh, ṅ → k has śvāsa, aghoṣa, vivāra and alpaprāṇa, sharing nothing; kh shares only mahāprāṇa; g and ṅ share three efforts but lack mahāprāṇa; gh shares all four → vāg ghariḥ.
Measure: adas → adaso 'ser dād u do maḥ (A 8.2.80) replaces a vowel of adas → for a short vowel the substitute is u, for a long one ū.
Place beats measure: ci + tā, stu + tā → sārvadhātukārdhadhātukayoḥ (A 7.3.84) prescribes guṇa for the i and for the u → guṇa is a, e or o (SK 17) → by measure a is nearest, one mātrā like i and u; by place e shares the palate with i and o shares the lips with u → place is the stronger → cetā, stotā.
Applies when / blocked by. The superlative ending -tama in antaratama is what excludes the near-misses. Bhaṭṭoji's commentators raise the objection themselves: why not g, or ṅ, which agree with h in three efforts? Because only gh agrees in all four, so only gh is similar in the highest degree. And for this nearness, sameness of place in some respect is enough: e counts as sharing place with i although it also has the throat. Total sameness of place is required only for the name savarṇa (SK 10), not here.
Watch out. The word sthāna is used in two senses in this chapter. In this sūtra as written it means prasaṅga, as in SK 38. But the word sthāne also comes down from SK 38 by anuvṛtti (its svarita marks it for that, SK 46), and read a second time inside this sūtra, with its ending changed to the instrumental, it means place of articulation: "most similar by place". That second sentence is the paribhāṣā "likeness by sthāna is the stronger".
tyadādīnām aḥ → yaḥ, saḥ; aster bhūḥA rule says "of yad, a" or "of as, bhū". Does the substitute replace one sound of the sthānin, or all of it? Four rules answer. By default the substitute replaces only the last sound (SK 42), so yad becomes ya-. If the operation belongs to a following element, it replaces that element's first sound (SK 44). A substitute of more than one sound, or one marked with ś, replaces the whole (SK 45), so as becomes bhū. But a substitute marked with ṅ, though it has more than one sound, again replaces only the last sound (SK 43). SK 43 is placed here beside SK 45, the rule it excepts, out of its Aṣṭādhyāyī order.
Rule. A substitute prescribed for an item designated by a genitive comes in place of the final al of that item — its last sound only. The words ṣaṣṭhī and sthāne come from SK 38 by anuvṛtti, ṣaṣṭhī now read as "by a genitive", and "designated" and "being prescribed" are understood.
Example.
yad → tyadādīnām aḥ (A 7.2.102), "of tyad and the rest, a" → the genitive designates the whole word yad, but the substitute takes only its last sound, d → ya- → with its ending, yaḥ.
tad → likewise ta-, seen in saḥ (the t becomes s by a later rule).
Applies when / blocked by. Two words of the sūtra are guards. "Substitute": an augment is not a substitute, so the iṬ of A 7.2.35 (SK 36) does not come in place of the final ṛ of the affix tṛC; it is added in front of it. "Of an al": under the heading padasya, "of a pada" (A 8.1.16), the rule vasusraṃsudhvaṃsvanaḍuhāṃ daḥ (A 8.2.72) prescribes d; in paramānaḍudbhyām that d replaces only the last sound of the pada paramānaḍuh, not the entire pada. The exceptions to this rule are SK 44 and SK 45; SK 43 then restores it for one class.
Rule. When an operation belongs to a following element — fixed as such by an ablative condition, SK 41 — it takes effect on the initial al of that element, not on its final. Bhaṭṭoji calls this an apavāda of SK 42. The word alaḥ is carried forward too, so the operation still hits one sound, not the whole element.
Example.
dvi + ap → dvyantarupasargebhyo 'pa īt (A 6.3.97): after dvi, antar or an upasarga, ī for ap → ap is the following element (SK 41) → SK 42 would have ī replace the final p; this rule sends it to the initial a → dvi + īp- → the i and the ī, being savarṇa (SK 10), become one long ī by a later sandhi rule, akaḥ savarṇe dīrghaḥ (A 6.1.101) → dvīpam.
Without alaḥ, the ī would replace the whole of ap.
Rule. A substitute that is anekāl — consisting of more than one al — or śit, marked with ś, comes in place of the whole sthānin. Bhaṭṭoji calls the sūtra clear, since no further word needs to be carried into it, and an apavāda of SK 42.
Example.
as (the root) → aster bhūḥ (A 2.4.52), "of as, bhū" → bhū has more than one sound, so it is anekāl → SK 42 would let it replace only the s; by this rule it replaces the whole root → bhū.
Other cases named by Bhaṭṭoji's commentators: the endings taught by tas-thas-tha-mipāṃ tāṃ-taṃ-ta-āmaḥ (A 3.4.101), and idama iś (A 5.3.3).
Applies when / blocked by. One might object that "for as comes bhū" would obviously mean the whole root; but SK 42 would cut the substitute down to the final sound, so this exception is needed. This rule also beats SK 44 when both apply. At the stage aṣṭā + as — the word aṣṭan has taken ā by aṣṭana ā vibhaktau (A 7.2.84), and the ending jas or śas follows — the rule aṣṭābhya auś (A 7.1.21) prescribes auś in place of that ending. auś is prescribed for the following element, so SK 44 offers it to the initial a of as; but auś, which Bhaṭṭoji's commentators count as anekāl, is also offered to the whole ending. Which rule yields? Each has its own field — SK 44 in dvīpam, SK 45 in aster bhūḥ and idama iś — so they are of equal strength, and between rules of equal strength the later one prevails: vipratiṣedhe paraṃ kāryam (A 1.4.2), the para principle of SK 46. A 1.1.55 is later than A 1.1.54, so the whole ending as is replaced. The same holds in ato bhisa ais (A 7.1.9).
Watch out. Bhaṭṭoji's alo 'ntya- is a quotation of part of the sūtra alo 'ntyasya, not a word of ordinary Sanskrit; his commentators defend the shortening as a quotation, on the model of prāg rīśvarān nipātāḥ (SK 19), which quotes part of adhir īśvare (A 1.4.97).
Rule. A substitute marked with ṅ (a ṅit) also comes in place of the final al only — even though it consists of more than one sound. It is an apavāda of SK 45. The word antyasya is carried forward from SK 42.
Example.
dadhi before the instrumental singular ending → asthidadhisakthyakṣṇām anaṅ udāttaḥ (A 7.1.75) prescribes the substitute anaṄ for dadhi → anaṄ has more than one sound (a, n; the ṅ is a marker), so SK 45 would have it replace the whole of dadhi → but it is ṅit, so by this rule it replaces only the final i → dadh·an- → the finished instrumental is dadhnā (the a of an is dropped by a later rule).
Other ṅit substitutes: avaṄ and the like; every one of them is anekāl.
Applies when / blocked by. Why this rule at all, when SK 42 already gives final replacement? Because SK 45 would take it away: every ṅit substitute is anekāl. Here Bhaṭṭoji's commentators define the apavāda. A rule undertaken where another rule would necessarily have applied is that other rule's apavāda (yena nāprāpte yo vidhir ārabhyate, "the rule that is begun where that one has not failed to apply"). SK 45 would necessarily apply to anaṄ; SK 43 is undertaken there all the same; so SK 43 is the apavāda and blocks SK 45 among the ṅit substitutes. Why the special rule is the stronger: a rule about particulars seizes them at once, by its own express word, whereas the general rule reaches them only by way of the general, and so is slow; besides, SK 43 would have nothing left to do if SK 45 won, while SK 45 keeps its field in the non-ṅit substitutes of A 3.4.101 and aster bhūḥ. Hence the maxim that the rule of the particular is the stronger.
tudati, dadhnā, ubhaye devamanuṣyāḥTwo things remain. How does a reader know that a word of one sūtra continues into the sūtras after it? And when two rules both want to act on the same stage of a word, which goes first? SK 46 answers the first question: the word carries the svarita accent. Bhaṭṭoji then appends the ranking that answers the second: a later rule beats an earlier one, a nitya rule beats a later one, an antaraṅga rule beats those, and an apavāda beats them all. In tudati the affix śa goes in before guṇa can act; in dadhnā the final-only replacement beats the whole replacement; in ubhaye devamanuṣyāḥ the name sarvanāma holds although a later rule would make it optional.
Rule. A word that Pāṇini pronounced with the svarita accent (SK 7) is adhikṛta, "put in charge": it is an adhikāra, and it recurs by anuvṛtti in the sūtras that follow. Adhikāra means employment — as a man "appointed over the village" is employed there, so a word's employment is its continuing into later rules. The svarita is not marked in the written text; like the nasality of marker vowels (SK 3), it is known from the author's declaration (pratijñā). How far the anuvṛtti runs is known only from the received interpretation (vyākhyāna).
Example.
sthāne in SK 38 → pronounced with the svarita → it recurs in SK 39, where it yields the second sentence "most similar by place" → cetā, with e and not a.
anta inside ādyantau (SK 36) → the svarita rests on that half alone → only anta recurs in SK 37: the mit is a final part → vārīṇi.
pūrvatrāsiddham (SK 12) → an adhikāra met in chapter 1 → read into every rule of the Tripādī (the last three quarters of the Aṣṭādhyāyī, A 8.2–8.4).
Applies when / blocked by. One might say that interpretation alone would tell us where a word recurs, just as it tells us where it stops; Bhaṭṭoji's commentators answer that the Mahābhāṣya shows the sūtra's uses at many points.
Watch out — the ranking of rules. Bhaṭṭoji appends three paribhāṣās to this sūtra.
The first: "Among para, nitya, antaraṅga and apavāda, each later one is stronger than the earlier." The four names are these.
Nitya over para:
tud + ti → two rules present themselves: the guṇa of a light penult (the u of tud is a light vowel in the penult, SK 31), which is the later rule, and the affix śa (tudādibhyaḥ śaḥ, A 3.1.77) → śa applies whether or not the guṇa has been done, while the guṇa, once śa is in, no longer applies → śa is nitya, so śa goes in first → tud·a·ti → śa is treated as ṅit (sārvadhātukam apit, A 1.2.4), and before a ṅit affix guṇa is forbidden (kṅiti ca, A 1.1.5) → tudati, with no guṇa.
The reason a nitya rule is strong: the nitya rule never fails to apply, so if one of the two must be imagined not to act, it should be the rule that fails anyway.
Antaraṅga over para:
ubhaya + the plural ending → sarvādīni sarvanāmāni (A 1.1.27) gives ubhaya the name sarvanāma → the later rule prathamacaramatayālpārdhakatipayanemāś ca (A 1.1.33) would make that name optional → but the name by A 1.1.27 depends on no case ending, so it is antaraṅga → it holds invariably, setting the later rule aside → ubhaye devamanuṣyāḥ.
The reason an antaraṅga rule is strong is a further paribhāṣā, from the Mahābhāṣya on vāha ūṭh (A 6.4.132): asiddhaṃ bahiraṅgam antaraṅge — when an antaraṅga rule has presented itself, the bahiraṅga rule is treated as not existing.
Apavāda over para:
dadhi + the instrumental ending → anaṄ (A 7.1.75) → SK 45, the later rule, would replace the whole of dadhi; SK 43 replaces the final i only → SK 43 is the apavāda, by the maxim yena nāprāpte → dadhnā.
The reason an apavāda is strong was given at SK 43.
Antaraṅga over nitya:
grāmaṇī used as a neuter, before the locative singular ending → the augment nUM by iko 'ci vibhaktau (A 7.1.73) is nitya; the shortening by hrasvo napuṃsake (A 1.2.47, SK 35) is antaraṅga → the shortening goes first → grāmaṇi → then nUM → grāmaṇini kule. Had nUM gone first, the stem would have ended in n, not in a vowel, and no shortening could have followed.
Apavāda over antaraṅga:
daitya + ariḥ → a before a vowel: ād guṇaḥ (A 6.1.87), being antaraṅga, presents itself even against the later akaḥ savarṇe dīrghaḥ (A 6.1.101) → but the lengthening of savarṇa vowels is its apavāda → daityāriḥ.
śrī + īśaḥ → iko yaṇ aci (A 6.1.77), being antaraṅga, presents itself likewise → the savarṇa lengthening is its apavāda → śrīśaḥ.
The second appended paribhāṣā is the one just quoted, asiddhaṃ bahiraṅgam antaraṅge. The third is akṛtavyūhāḥ pāṇinīyāḥ, "Pāṇini's followers do not set out the arrangement in advance". Taken literally — vyūha being the separating of base and affix — it would make the whole grammar pointless, so Bhaṭṭoji supplies a qualification: seeing that a condition (nimitta) of an operation is about to perish, they do not carry out the operation that condition would occasion. The illustration is the form seduṣaḥ, to be explained when the word sedivas is examined later. Scholars discuss whether this paribhāṣā has any basis.
This ends the paribhāṣā-prakaraṇa.
| SK | Sūtra | What it does | Example |
|---|---|---|---|
| 34 | iko guṇavṛddhī | Guṇa or vṛddhi prescribed by name falls on an ik. | mid → med- |
| 35 | acaś ca | Hrasva, dīrgha or pluta prescribed by name falls on a vowel. | śrīpā → śrīpam |
| 36 | ādyantau ṭakitau | A ṭit augment is the initial part, a kit augment the final part. | bhav·i·tā; prāṅ·k·ṣaṣṭha |
| 37 | mid aco 'ntyāt paraḥ | A mit augment goes after the last vowel, as a final part. | vāri·n- → vārīṇi; vahaṃlihaḥ |
| 38 | ṣaṣṭhī sthāneyogā | An undetermined genitive means "in place of"; sthāna = prasaṅga. | iko yaṇ aci: dadhy atra |
| 40 | tasminn iti nirdiṣṭe pūrvasya | A locative condition targets the element immediately before. | dadhy atra; not samidham |
| 41 | tasmād ity uttarasya | An ablative condition targets the element immediately after. | agnim īḻe loses accent; not ut·pra·sthānam |
| 39 | sthāne 'ntaratamaḥ | The most similar candidate is the substitute; place outranks the rest. | vāg ghariḥ; cetā |
| 42 | alo 'ntyasya | A substitute replaces only the last sound. | yad → yaḥ |
| 44 | ādeḥ parasya | For a following element, the first sound instead. | dvi + ap → dvīpam |
| 45 | anekālśit sarvasya | A multi-sound or ś-marked substitute replaces the whole. | as → bhū |
| 43 | ṅic ca | A ṅ-marked substitute replaces only the last sound. | dadhi + anaṄ → dadhnā |
| 46 | svaritenādhikāraḥ | A svarita-marked word is a heading that recurs; para < nitya < antaraṅga < apavāda. | sthāne into SK 39; tudati, ubhaye, dadhnā |