अच्सन्धिप्रकरणम्

iko yaṇaci — iko yaṇaci. "ikaḥ" is a genitive (ṣaṣṭhī). Hence, by the paribhāṣā "ṣaṣṭhī sthāneyogā" (A 1.1.49), the sense "in place of" (sthāne) is obtained. And "sthāna" has been said [earlier] to mean prasaṅga — the occasion [for a substitute], i.e., where the original would otherwise stand. Since it is impossible for one varṇa to be the locus (adhikaraṇa) of another varṇa [in a literal locative sense], "aci" is a sati-saptamī (a conditional locative, "when an ac is present"). By the paribhāṣā "tasminn iti nirdiṣṭe pūrvasya" (A 1.1.66), the sense "of what precedes, when an ac has been uttered with no other varṇa intervening" is obtained. And thus "when an ac follows (aci parataḥ)" results as the sense. "saṃhitāyām" is in force as a governing heading (adhikāra); and from that is obtained the absence of any temporal interval exceeding half a mātrā. This resultant meaning [Bhaṭṭoji] states with "ikaḥ sthāne" and so on. This [whole explanation] follows the older view that adheres to the literal wording of the sūtra.

In the Bhāṣya on the sūtra "saṃhitāyām," however, taking "aci" as a locative in the contactual sense of the locus (aupaśleṣika adhikaraṇa) — which is a form of proximity (sāmīpika) — and relying on [the paribhāṣā] "tasmin," the saṃhitā-adhikāra is rejected (pratyākhyāta) [as unnecessary]. For the Bhāṣya there runs thus: "This rule [saṃhitāyām] can be left unstated. How? The locus (adhikaraṇa) is of three kinds: pervasive (vyāpaka), contactual (aupaśleṣika), and domain-based (vaiṣayika). But what other connection can a sound have with a sound than contact (upaśleṣa)? [So] 'iko yaṇaci' means 'of the ik that is in contact with an ac.' There, even without mention of saṃhitā, [the operation] will occur precisely in saṃhitā." — upa means "in proximity," śleṣa means "connection," [so] upaśleṣa [is proximity-connection]; the locus constituted by that is aupaśleṣika, that is, proximity-based (sāmīpika). And thus it results as "of the ik standing in proximity to an ac." And proximity means the absence of any interval, both temporally and in respect of varṇa. And so, since in the absence of saṃhitā the said proximity is itself absent, the non-occurrence of yaṇ is established on that very ground; hence the saṃhitā-adhikāra need not be prescribed. Such is the purport of the Bhāṣya.

And thus, by parity of reasoning, in "tasminn iti nirdiṣṭe pūrvasya" (A 1.1.66) too the locative (saptamī) is in the contactual-locus (aupaśleṣika) sense. From that, the sense results: when the referent of a locative-ending word such as "aci" — something beginning with the vowel a — is uttered, the operation applies to the preceding [item] that is present [to it] by the relation of proximity. And the qualification "unintervened" (avyavahita) need not be supplied to "uttered," because that [immediacy] is already obtained precisely from the saptamī taken in the proximity-locus sense; for proximity there consists in the absence of interval both temporally and in respect of varṇa. And thus, there [in A 1.1.66], the syllable "nir" [in nirdiṣṭa] need not be taken separately for the sake of the sense of immediacy (nairantarya). This is made clear by Kaiyaṭa on the sūtra "snukramoḥ" and on the sūtra "tatra ca dīyate." The threefold nature of the locus will be clarified [later] in the kāraka section, at "ādhāro 'dhikaraṇam" (A 1.4.45).

sudhī — From the root "dhyai," in the sense of thinking (i.e., dhyāyati): with the kvip suffix, the y undergoes samprasāraṇa [becoming i]; then by pūrvarūpa [the single substitute in the form of the preceding vowel, for i + ā]; and by "halaḥ" (A 6.4.2) the vowel is lengthened — [yielding] the word dhī. Those who think well (suṣṭhu dhyāyanti) are sudhiyaḥ; or, those whose intellect (dhī) is excellent (su = śobhanā) are sudhiyaḥ. The analytical paraphrase (vigraha) is "sudhībhir upāsyaḥ" (to be worshipped by the wise). The compound is by "kartṛkaraṇe kṛtā bahulam" (A 2.1.32); and by "supo dhātuprātipadikayoḥ" (A 2.4.71) there is luk of the bhis-ending. When "sudhī upāsya" stands, the construal [with the sūtra] is that y is the substitute for ī. The secondary signification (lakṣaṇā) by which the pratyāhāras denote the individual sounds subsumed under what they denote is elaborated at the sūtra "nājjhalau" (A 1.1.10); to bring that out, ī has been given as the example [of an ik member].

Objection: Why should not v, r, l be [the substitutes] for ī, since they are all equally yaṇ, yaṇ-hood being non-differentiating? To this [Bhaṭṭoji] replies with "sthānata āntaryāt" (by nearness of place). Because of the shared palatal (tālu) place of articulation, in place of ī, by "sthāne 'ntaratamaḥ" (A 1.1.50), it is precisely y that occurs; not v, r, l, because they have different places of articulation. For this very reason, the Bhāṣya on the sūtra "yathāsaṃkhyam anudeśaḥ samānām" (A 1.3.10) [says]: "What is the example here [of yathāsaṃkhya]? iko yaṇaci — dadhy atra, madhv atra. This is not [a valid example]: it is established even by 'sthāne 'ntaratama' (A 1.1.50)." And the Bhāṣya on the sūtra "sthāne 'ntaratamaḥ" (A 1.1.50), conversely, says: "What is the example here? iko yaṇaci — dadhy atra. This is not [a valid example]: it is established even by saṃkhyātānudeśa" — meaning, by the yathāsaṃkhya-sūtra. [Each Bhāṣya passage thus refers its own would-be example to the other rule.]

Objection: Here, by the word "yaṇ," four sounds are taken — the non-nasal y, v, l, and r — because, the yaṇ being the item to be brought about (bhāvyamāna, i.e., the substitute), it does not take its homogeneous sounds (savarṇa); [whereas] even granting that the [phonetic] qualities (guṇa) are non-distinguishing [so that savarṇas would be included], y, v, l amount to six [each in nasal and non-nasal form] and r makes seven. But by the word "ik" sixty-six are taken. Being thus of unequal number, how can the yathāsaṃkhya-sūtra operate here? If [this is asked] — No: because, once unified under [the shared properties of] ik-hood, yaṇ-hood, and the like, they are equal [in the number of classes].

Objection: Since by the ṛ-varṇa and the ḷ-varṇa thirty [homogeneous sounds] each present themselves, there arises the liability (prasaṅga) of r as the substitute for the ḷ-varṇas and l as the substitute for the ṛ-varṇas; so the matter is not resolved by the yathāsaṃkhya-sūtra. If [this is objected] — listen: from the yathāsaṃkhya-sūtra it is obtained that "for what is delimited by ṛ-hood (ṛtva), r occurs; for what is delimited by ḷ-hood (ḷtva), l occurs." And the generic property ṛtva is not present in the ḷ-varṇas, nor is the generic property ḷtva present in the ṛ-varṇas. It is only by force of the rule of homogeneity (sāvarṇya) between the ṛ- and ḷ-varṇas that ṛtva is [figuratively] imposed upon the ḷ, and ḷtva upon the ṛ, for the sake of the operation. And thus, by taking the real (vāstava) ṛtva and ḷtva [rather than the imposed], the operation of yathāsaṃkhya here is unobstructed. But let this rest for now.

khari ca — khari ca. To [the question] "by what [rule]?" he says "khari ca." From *jhalāṃ jaś jhaśi* [A 8.4.53] "jhalām" is carried forward (anuvṛtti), and from *abhyāse carca* [A 8.4.54] "car" is carried forward. He states this with "when a khar follows" (khari pare) etc. "Thus, of the j": because of proximity (āntarya) by place of articulation — this is the idea. It should be understood that chatva (ś → ch) [comes] thereafter, not before, because chatva is asiddha with respect to cutva (ścutva) and cartva.

But in the sūtra-order approved by the Bhāṣya on the sūtra *halo yamāṃ yami lopaḥ* [A 8.4.64]: once j is produced by cutva (ścutva), then chatva of ś (ś → ch), then cartva of j (j → c) — not cartva before chatva, because chatva is asiddha with respect to cartva.

"tac-chivaḥ": the analysis is "he, that Śiva" (sa ca asau śivaḥ), or "the Śiva of him" (tasya śivaḥ). But kutva by *coḥ kuḥ* [A 8.2.30] does not [apply], because ścutva is asiddha.

"chatvam amī": in the sūtra *śaś cho 'ṭi* [A 8.4.63], instead of "aṭi" one should read "amī" — this is the meaning; that is, the sūtra should be read as "śaś cho 'mī." "tac ślokena": the analysis is "that śloka," or "the śloka of it." Since l lies outside the aṬ [pratyāhāra], chatva of the ś that follows it is not obtained by the [received] reading of the sūtra; hence this vārttika [prescribing the reading amī]. "vāk-ścyotati": here kutva does apply, because the c, being natural (svābhāvika) and not produced by ścutva, is not asiddha. Here, since the c lies outside the aM [pratyāhāra], chatva of the ś that follows it does not apply.

sthānivadādeśo 'nalvidhau — Having placed in mind the doubt "In sudh-y, since the y that replaces ī is, by sthānivadbhāva, a vowel (ac) and is followed by a vowel (ac-paraka), how can the dh be doubled?", he states the sūtra that establishes sthānivadbhāva: sthānivadādeśaḥ.

That one who takes the place of a teacher — a teacher's son and the like — acquires the teacher's properties by taking his place is established from ordinary usage (loka); and in the case of śara-grass and the like taking the place of kuśa-grass [in ritual substitution] it is established by Vedic principle. But here in the śāstra, because of the statement svaṃ rūpaṃ śabdasya [aśabda-saṃjñā] (A 1.1.68) [that each word denotes its own form], the properties of the sthānin would not accrue to substitutes; so, to secure them, sthānivadādeśaḥ is taught.

"sthāna means occasion (prasaṅga)," as has been said. That in whose place another is enjoined is the sthānin; that which, being enjoined, displaces what would otherwise obtain is the ādeśa (substitute). sthānivat = "like the sthānin"; the suffix vati [is added] by tena tulyaṃ… (A 5.1.115). The substitute becomes equal to the sthānin — i.e. it comes to possess the sthānin's properties. "al" is a synonym for varṇa (sound). vidhi = that which is enjoined = an operation (kārya). An operation grounded in a sound = al-vidhi; [its negation] = an-al-vidhi. The sense conveyed [by "analvidhau"] is: "when an operation other than a sound-grounded one is to be done." The upshot: when a sound-grounded operation is to be done, [the substitute] is NOT like [the sthānin].

Since "alāśraya" is a general expression, it comprehends all [four types]: an operation [conditioned] by a sound (alā), an operation on what follows a sound (alaḥ parasya), an operation of a sound (alaḥ), and an operation on a sound (ali).

(1) An operation "by a sound": as in vyūḍhoraskena. Here the s, which replaces the visarga by visarjanīyasya saḥ (A 8.3.34) — taken to be [still] a visarga [by sthānivadbhāva] — would get ṇatva by [the rule permitting] the intervention of aṭ/aḍ (aḍvyavāya); but it does not [since ṇatva is here an al-vidhi].

(2) An operation "on what follows a sound": as in dyauḥ. By diva aut (A 7.1.84) the au that replaces the v, being a consonant (hal) by sthānivadbhāva, would make the following su subject to halṅyādi-lopa (A 6.1.68, elision of su after a hal); but it does not.

(3) An operation "of a sound" [i.e. a substitution replacing a sound]: as in dyukāmaḥ. The u that replaces the v [of div → dyu], being a v by sthānivadbhāva, would undergo elision by lopo vyor vali (A 6.1.66); but it does not.

(4) An operation "on a sound" [locative]: as in ka iṣṭaḥ. [iṣṭa is] from √yaj with kta. Here the i (the saṃprasāraṇa replacing the y of yaj), being a haś by sthānivadbhāva, would trigger the u-substitution (utva) by haśi ca (A 6.1.114); but it does not.

And "al" here [in the rule] is taken to be both the al that is the sthānin AND the al that is a component of the sthānin (sthāny-avayava). Hence the upshot: in an operation grounded in the al that is the substitute's sthānin, or in the al that is a component of the sthānin, [the substitute] is not sthānivat. Of these, for an operation on the sthānin-al, vyūḍhoraskena has already been given as an example. Or [take another]: from the root dhivi "to please" (dhivi prīṇane), in laṭ, third person, when jhi [receives its anta-substitute], and when the vikaraṇa's u [prescribed by dhinvikṛṇvyor a ca, A 3.1.80] becomes v by yaṇ — the augment iṭ, [which would apply because] that v is ārdhadhātuka by sthānivadbhāva and itself begins with a val, would come but does not; for the iṭ, proceeding by treating the substitute v as ārdhadhātuka, depends on the al that is the sthānin of the v (namely the u).

For an operation grounded in an al that is a COMPONENT of the sthānin: as in pratidīvya. Here the "ya" that substitutes for ktvā [i.e. lyap], being ārdhadhātuka and beginning with a val by sthānivadbhāva, would take the iṭ augment, but does not; for [the concern about] iṭ in the domain of "beginning with val" is grounded in an al that is a component of the sthānin.

All this he states with "A substitute is like its sthānin, but not in an operation grounded in a sound-form of the sthānin (sthāny-al-āśraya)." Why "of the sthānin" (sthāni-) in "sthāny-al-āśraya"? [Consider] rāmāya. Here the lengthening (dīrgha) by supi ca (A 7.3.102) depends on an al (the y-form) belonging to the SUBSTITUTE [the case-ending ya]; yet, when that lengthening is to be done, the substitute [ya] is still treated as a sup by sthānivadbhāva — because, although the dīrgha depends on the y-form al of the substitute, it does NOT depend on an al of the sthānin. Nor should one object: "From √nī with ṇvul, with the 'aka' substitute and vṛddhi, giving nai-aka, the ai — being of the property of ī by sthānivadbhāva — would block the āy-substitution (āyādeśa), since ī has no āy-substitution." For here only a śāstra-enjoined operation prompted by the sthānin is extended (atideśa); and an "āy-form" of ī is not enjoined by the śāstra, so there is no extension of it to the ai that replaces ī. Let that be.

On "by this" (anena): by the sthānivat-sūtra just cited, one need not fear — in sudh-y here — the prohibition of doubling the dh by anaci [ca], a rule meaning "no doubling before a vowel," [applied] by taking the y (replacing ī) to function, via sthānivadbhāva, as a vowel (ac). Why [need one not fear it]? He says: "because it is denied in an al-vidhi" (analvidhau tanniṣedhāt) — i.e. because sthānivattva is denied [there]. For the prohibition of doubling the y proceeds by relying — through the y — on the vowel-hood (ac-tva) belonging to the al (ī) that is the sthānin of the y-substitute, and is therefore grounded in the al of the sthānin.

acaḥ parasmin pūrvavidhau — Objection (nanu): In sudh-y [where the ī of sudhī has become y before a vowel], granted that by the [following] sūtra 'let it not be sthānivat' [na padānta… (A 1.1.58), which withholds sthānivadbhāva in doubling and the rest] there is no sthānivadbhāva; yet by this following sūtra sthānivattva would indeed apply. — Intending to raise this doubt, [Bhaṭṭoji] cites the following sūtra of that kind: acaḥ parasmin.

Since [the sthānivadbhāva] is already established by the sthānivat-sūtra [sthānivad ādeśo 'nalvidhau, A 1.1.56], why is this sūtra [needed]? To this he says 'for the sake of an al-operation' (alvidhyartham): the meaning is, for the sake of sthānivadbhāva even in an operation that takes an al (the actual sound of the sthānin) as its basis (alāśraya-vidhi) — since A 1.1.56 grants sthānivat everywhere except in an alvidhi. Thereby vavraśca is established.

[Derivation of vavraśca:] The root is o-vraścū 'to cut' (chedana). In the liṭ (perfect), with tiP replaced by ṇaL, doubling (dvirvacana) applies. By liṭy abhyāsasya… (A 6.1.17), saṃprasāraṇa of the r (repha) in the abhyāsa (reduplicative syllable) → ṛ; then pūrvarūpa (by saṃprasāraṇāc ca, A 6.1.108, the ac following the saṃprasāraṇa vowel is absorbed into it); then, by uraṇ raparaḥ (A 1.1.51), the aṇ-substitute of the ṛ is followed by r (rapara); then halādiḥ śeṣaḥ (A 7.4.60), only the initial consonant of the reduplicate remaining. There, in the abhyāsa, a further saṃprasāraṇa of the v does NOT apply; for the 'a' that stands in place of the ṛ (the ṛ being itself a saṃprasāraṇa of r), being by sthānivadbhāva [treated as] a saṃprasāraṇa, is prohibited by na saṃprasāraṇe saṃprasāraṇam (there is no saṃprasāraṇa where a saṃprasāraṇa already stands). By the previous sūtra [A 1.1.56], however, sthānivadbhāva is not possible here, because this prohibition of saṃprasāraṇa is an operation resting on the al (the actual sound of the sthānin) [and A 1.1.56 excludes an alvidhi]. This is the point.

From the previous sūtra, 'sthānivad ādeśaḥ' is carried forward (anuvṛtti) here. The word 'acaḥ' construes with 'ādeśaḥ': 'a substitute of an ac' (aca ādeśaḥ). 'parasmin' is a sati-saptamī (locative absolute); hence 'conditioned by a following element' (para-nimittaka) is obtained, and that qualifies the substitute. So he says: 'a substitute of an ac conditioned by a following element' (para-nimitto 'jādeśaḥ). 'vidhi' — what is prescribed — means an operation (kārya); pūrva-vidhi is an operation upon what precedes.

And although 'being prior' (pūrvatva) is relative to some limit (sāvadhika), and three things are present here — the sthānin, the substitute (ādeśa), and the conditioning follower (para-nimitta) — of these, the sthānin cannot be the limit, since it has been removed by the substitute; nor can the substitute be the limit, nor the conditioning follower. For [if either were the limit,] in vaiyākaraṇa an unwanted result would follow: when the āyaṬ substitute of the ai — which precedes the yaṇ-substitute of the i (ikāra-sthānika) and precedes the ā that conditions that yaṇ — is to be made [by na yvābhyām…], the yaṇ-substitute would, by sthānivadbhāva, [wrongly] become an ac. Nevertheless, priority here is intended only relative to the sthānin; for even though the sthānin has been removed by the substitute, priority relative to it is possible by recourse to 'what was formerly there' (bhūtapūrva-gati). This he states with 'preceding the ac that was the sthānin' (sthānibhūtād acaḥ pūrvatvena), etc.

Here the atideśa is such that what happens when the sthānin is present happens also when the substitute is present, and what does not happen [with the sthānin] does not happen with the substitute either — so that even a non-śāstric (aśāstrīya) absence of an operation is transferred (atideśa). Of these two, the first [the transfer of a positively stated operation] has already been exemplified by vavraśca, since there the prohibitory operation na saṃprasāraṇe saṃprasāraṇam is śāstric. For the second [the transfer of a non-śāstric absence of an operation], the example is gaṇayati.

[Derivation of gaṇayati:] gaṇa 'to count' (saṃkhyāna) is an a-ending root of the cur-ādi class. From it comes ṇic; then, by ato lopaḥ (A 6.4.48), elision of the [final] a; then tiP, śaP, guṇa of the ṇi (ṇer guṇaḥ), and the ay-substitute — giving the form gaṇayati. Here, before ṇic, the penultimate a following the g (the upadhā) does NOT undergo at upadhāyāḥ (A 7.2.116, vṛddhi of the penultimate a before a ṇit affix), because by the present sūtra the elision of the [final] a (al-lopa) is sthānivat: when the [elided] a is present [as sthānin], the a following the g loses its penultimate (upadhā) status, so upadhā-vṛddhi cannot operate. And although this absence of vṛddhi is non-śāstric, it is [nonetheless] transferred (atideśa) in the case of the al-lopa [by sthānivadbhāva]. Nor may it be objected that here the a following the g is not prior to the sthānin a, because it is separated [from it] by ṇ (vyavadhāna); for priority here is common to both the separated and the unseparated (vyavahita and avyavahita), as is indicated (liṅga) by the prohibition regarding a vowel (svara) in the following sūtra — and that will become clear there itself; enough of this.

'iti yaṇaḥ' — the meaning is: by this sūtra, in sudh-y, when the doubling (dvitva) of the dh is to be prohibited, the sthānivadbhāva of the y that stands in place of ī (īkāra-sthānika) would [wrongly] be obtained; and so the sūtra that prohibits it [the following one] is begun.

na padāntadvirvacanavareyalopasvarasavarṇānusvāradīrgha-jaścarvidhiṣu — na padāntadvirvacana…. [The words] "[a substitute is] sthānivat" (sthānivad ādeśaḥ, from 1.1.56) and "of an ac, when [a rule concerning] what follows [applies]" (acaḥ parasmin, from 1.1.57) are both carried forward (anuvṛtti). The construction is: an ac-substitute occasioned by a following element (para-nimitta) is not sthānivat. padānta, dvirvacana, vare, yalopa, svara, savarṇa, anusvāra, dīrgha, jaś, and car — [these form] a dvandva; their vidhis = their prescribed operations (vidhāna); the compound [with vidhi] is a genitive in the objective sense (karmaṇi ṣaṣṭhī). Thence one obtains "when padānta and the rest are to be effected (vidheya)." By "vare" is meant the ac-substitute [occurring] before [the affix] vara; the dvandva is irregular/archaic (ārṣa), and there is non-elision (aluk) of the locative case-ending [in vare]. The word vidhi connects with each member individually: "in the padānta operation," "in the dvirvacana operation," and so forth. The end of a pada = its final member (carama-avayava); "in the operation of padānta" = in an operation that has padānta as its object (padānta-karmaka), that is, when an operation is [to be done] on the final member of a pada, and when an operation such as dvirvacana [is to be done]. That he [Bhaṭṭoji] states with padasya carama-, and so on.

But it is not glossed as "in an operation in place of a padānta (padāntasya sthāne vidhau)"; for on that [reading], [the form] eṣo yan hasati would not be established (asiddhi). To wit: the word-division is eṣaḥ yan. From the root iṇ (√i "to go"), [with] the śatṛ [affix] of laṭ and luk of śap, the i becomes yaṇ (y) by iṇo yaṇ (6.4.81) [yielding yan]. Here the elision of the su [nominative ending, in eṣaḥ] by etattadoḥ [su-lopo... hali] (6.1.132) does not take place, because that elision is prescribed [only] when a consonant (hal) follows; and here, when that [elision] is to be done, the ya-kāra — which stands in place of the i — is [treated as] an ac by sthānivadbhāva. Nor should one apprehend the prohibition "na padānta[...]" [as applying here]; for that prohibition [operates] only where what is being effected (vidhīyamāna) becomes the final member of a pada, whereas here the su-lopa to be effected is not a member of a pada (pada-anavayava). On the interpretation "in an operation in place of a padānta," however, the sthānivadbhāva of the ya-kāra would not be established here, since the [su-]elision is prescribed in place of the pada-final sa-kāra [of su]. Here, though, when the u-substitution (utva) by haśi ca (6.1.114) is to be done, the ya-kāra is not sthānivat, because the u being effected [itself] becomes the final member of the pada [eṣaḥ → eṣo]. And thus, for "in the padānta operation," the example is this [case of] utva in eṣo yan; whereas for the case of su-lopa it is a counter-example (pratyudāharaṇa) — this is clear in the Bhāṣya, and clearer still in the [Bhāṣya-Pradīpa-]Uddyota. The [further] example for "in the padānta operation," [namely] kāni santi and the like, will become clear just below (anupadam) in the base text (mūla) itself.

In [the case of] dvirvacana (doubling), the example is sudh-y [= sudhy]. Nor should it be objected that here, when doubling is to be done, sthānivadbhāva would after all obtain in the prohibition of it, even in the absence of the ya-kāra's sthānivadbhāva; for, since the doubling by anaci ca (8.4.47) is non-conditioned (anaimittika), the sthānivadbhāva of the ya-kāra is not required (anapekṣita) in its domain, and hence the prohibition there would be purposeless (vaiyarthya); therefore, by the word dvirvacana here, it is precisely the prohibition of doubling [expressed] by "aci na" ("no doubling when an ac follows") that is intended — this is the idea.

For vare [before the affix vara], for example: yāyāvaraḥ. By yaś ca yaṅaḥ, the affix varac [is added] after the root yā ending in yaṅ (the intensive); doubling (dvitva) by sanyaṅoḥ (6.1.9). When yāyāya-vara stands, by ato lopaḥ [iṭi ca] (6.4.48) the a of yaṅ is elided; then the ya-kāra is elided by lopo vyoḥ [vali] (6.1.66). Here, when the elision of the ā is to be done by ato lopa iṭi ca (6.4.48) on the strength of an ac-initial ārdhadhātuka (ajādi-ārdhadhātuka), the earlier al-elision is not sthānivat.

For yalopa (elision of y), for example: yātiḥ. From the root yā, in the intensive (yaṅ), [there is] doubling; [then the affix] ktic. When yāyāya-ti stands, [there is] a-elision (ato lopaḥ); [then] y-elision by lopo vyoḥ [vali]. Because of the sthānivadbhāva of the al-elision [here], [there follows] ā-elision (āto lopaḥ); [then again] y-elision by lopo vyoḥ. The form is yātiḥ. Here, the al-elision is not sthānivat when the y-elision is to be done. Nor should it be objected that in vāyvoḥ too, when the y-elision by lopo vyoḥ is to be done, there would be a prohibition of the sthānivadbhāva of the va-kāra that is the substitute for u; for by the vārttika "in [the environments of] svara, dīrgha, and yalopa, [only] an ac-substitute that is [itself] an elision (lopa) is not sthānivat," [the prohibition] is restricted [to lopa-substitutes], and here the va-kāra is not of the nature of an elision.

For the svara operation, for example: cikīrṣakaḥ. From cikīrṣa (the desiderative, sannanta [of √kṛ]), [the affix] ṇvul; the aka-substitution [for ṇvul]; the a of san [i.e. of cikīrṣa] is elided by ato lopaḥ. Here, when the udātta accent of the ī is to be effected by [a rule] ending in "vit," the al-elision is not sthānivat. Although the ī is not immediately (a-vyavahita) prior to the a that was the sthānin of the al-elision, so that no occasion for sthānivadbhāva arises here [in any case], nevertheless, from this very indication (jñāpaka), [it is learned] that in the previous sūtra [1.1.57] the "priority" (pūrvatva) is common to both the mediate and the immediate (vyavahita-avyavahita-sādhāraṇa); its purpose has been stated in the previous sūtra itself.

For the savarṇa operation, for example: śiṇḍḍhi. From the root śiṣ, of the rudhādi class, the loṭ [imperative] second-person singular; sip [tiṅ-ending], [and] śnam [the vikaraṇa]; [thus] śinaṣ-si; hi-substitution [of sip]; dhi-substitution [of hi]; ṣṭutva [retroflexion, dh → ḍh, by ṣṭunā ṣṭuḥ, 8.4.41]; jaś-substitution (jaśtva) of the ṣ [→ ḍa-kāra]; [so] śinaḍḍhi; al-elision of śna[m] [by śnasor allopaḥ, 6.4.111]; anusvāra [of the n] by naś cāpadāntasya [jhali] (8.3.24); by anusvārasya yayi [parasavarṇaḥ] (8.4.58), the para-savarṇa of that [anusvāra] is ṇa-kāra [homorganic with the following ḍ]. The form is śiṇḍḍhi. Here, when the para-savarṇa is to be done, the al-elision is not sthānivat.

But in truth (vastutas tu) this is not an example for the savarṇa operation. The elision of the a of śnam here is an ac-substitute; its sthānin is precisely the a of śnam; and when that [a] is present [by sthānivadbhāva], there is no occasion at all for the anusvāra of the na-kāra [since n would then be followed by the vowel a, not by jhal]. And thus, since the anusvāra is never seen to be preceded (pūrvatva) by an ac that is its sthānin, when its para-savarṇa is to be done, the sthānivadbhāva of the al-elision by acaḥ parasmin (1.1.57) does not present itself — so what [use] is there in prohibiting it? As for what is said in the Tattvabodhinī — "the na-kāra that is the sthānin of the anusvāra is seen to be prior to the a of śnam; hence the anusvāra standing in its place is also seen to be prior to that [a], since priority is admitted even through the sthānin (sthāni-dvārā)" — if that were so, in titāpayati the augment puk (pugāgama) would not occur. [Derivation:] by [the vārttika] tatkaroti tadācaṣṭe, ṇic [is added to titau]; by iṣṭhavadbhāva, the u [the ṭi] is elided by ṭeḥ (6.4.155); by aco ñṇiti (7.2.115), the vṛddhi of the a after the ta-kāra is ā; [then] the augment puk; from titāpi, [comes] laṭ, tip, śap; guṇa; ay-substitution [e → ay by eco'yavāyāvaḥ, 6.1.78]; the form is titāpayati. Here, when the augment puk is to be added to the ā obtained by the vṛddhi [taught] in aco ñṇiti, if the sthānivadbhāva of the u-elision were to hold, then, owing to the interposition by the u, the [ā] would not be immediately followed by ṇi (ṇi-parakatva being absent), and so the augment puk would not occur — because, although the ā's priority to the u that is the sthānin of the elision is naturally absent, it would [on that view] be present through the sthānin (sthāni-dvārā): thus [the Tattvabodhinī's position] is faulted in the Siddhāntaratnākara. In the Śabdaratna, the commentary on the Prauḍhamanoramā, however, it is elaborated that the mention of savarṇa in this sūtra serves to prohibit the sthānivadbhāva that would [otherwise] arise through the ablative compound "[an operation] concerning what follows from the preceding (pūrvasmāt parasya vidhiḥ)" — [as in the derivation of] pādayati (pādam ācaṣṭe "relates the [word] pāda"), then kvip [yielding] pāt, [and then] in pāt hasati and the like, when the pūrva-savarṇa by jhayo haḥ [8.4.62] is to be done.

For the anusvāra operation, for example: śiṃṣanti. From the root śiṣ, in laṭ, [the ending] jhi; jh → ant by jho 'ntaḥ [7.1.3]; al-elision of śna[m] by śnasor allopaḥ; anusvāra by naś cāpadāntasya [jhali]; [so] śiṃṣanti. But here there is no para-savarṇa, because the ṣa-kāra is not a yay [i.e. not among the sounds before which anusvāra → para-savarṇa]. Here, when the anusvāra is to be done, the al-elision is not sthānivat.

For the dīrgha operation, for example: pratidīvnā. When the lengthening (dīrgha) of the i after the da-kāra is to be done by hali ca (8.2.77), the al-elision is not sthānivat.

For the jaś operation, for example: sagdhiś ca me. From ad ("to eat," bhakṣaṇa), [the affix] ktin; by bahulaṃ chandasi, the ghasl̥ substitute [for ad]; upadhā-elision by ghasibhasor hali ca (6.4.100); s-elision by jhalo jhali (8.2.26); dh-substitution of the ta-kāra by jhaṣas tathoḥ [dho 'dhaḥ] (8.2.40); by jhalāṃ jaś jhaśi (8.4.53), through jaś-substitution, the gha-kāra becomes ga-kāra. samānā gdhiḥ = eating [in common], sa-gdhiḥ; the sa-substitution [for samāna] by samānasya chandasi…. Here, when the jaś-substitution is to be done, the upadhā-elision is not sthānivat.

For the car operation, for example: jakṣatuḥ. From ghas, in liṭ [perfect], [the ending] atus; doubling; by halādiḥ śeṣaḥ (7.4.60); jaś-substitution [of the abhyāsa] by abhyāse car [ca] (8.4.54); the ja-kāra [i.e. the abhyāsa consonant → cu] by kuhoś cuḥ (7.4.62); upadhā-elision by gamahana… (6.4.98); by khari ca (8.4.55), car-substitution, [the gh becomes] ka-kāra; ṣa [for the s] by śāsivasi… [ghasīnāṃ ca] (8.3.60). Here, when the car-substitution is to be done, the upadhā-elision is not sthānivat.

In the Bhāṣya, however, relying on [the maxim] "[a substitute is] not sthānivat [only] in the domain of pūrvatrāsiddham (8.2.1)," the [environments] dvirvacana, savarṇa, anusvāra, dīrgha, jaś, and car are dispensed with (pratyākhyāta). Moreover, [the Bhāṣya gives] the answer "dīrgha [belongs] to the [other] ācāryas." The order of the five sūtras (pañcasūtrī) — anusvārasya yayi parasavarṇaḥ (8.4.58), vā padāntasya (8.4.59), khari ca (8.4.55), vāvasāne (8.4.56), aṇo 'pragṛhyasyānunāsikaḥ (8.4.57) — is the sūtra-order approved by the Bhāṣya on the sūtra halo yamāṃ yami (8.4.64). And thus, in śiṇḍḍhi there is no occasion for para-savarṇa: when the para-savarṇa is to be done, the jaś-substitute standing in place of the ṣa-kāra is asiddha, so [the anusvāra] is not followed by a yay. Hence in every way śiṇḍḍhi is not an example for the savarṇa operation — but let this rest (āstāṃ tāvat).

iti sthāni[vat] — [commenting on Bhaṭṭoji's closing "iti sthānivadbhāvaniṣedhaḥ":] the meaning is that, by this sūtra, in sudh-y, when the prohibition of doubling is to be effected, [there is] a prohibition of the sthānivadbhāva of the ya-kāra. And thus, since [the following sound] is not an ac, the prohibition by anaci ca (8.4.47) is absent, and so the doubling of the dha-kāra is unobstructed — this is the idea; and so it stands as sudh-dh-y [i.e. with the doubled dh].

jhalāṃ jaś jhaśi — On [the Kaumudī's word] "self-evident": [the sūtra] is self-explanatory in the sense "in place of a jhal, jaś occurs when a jhaś follows." There is nothing whatever to be explained here, since there is no anuvṛtti of any further word [that would need interpretation] — this is the meaning. On [the phrase] "the dh": that is, the first dh — this is the meaning. On "becomes d": the idea is that [d is chosen] on account of nearness of place [of articulation] (sthānata āntaryāt).

adarśanaṃ lopaḥ — In order to raise the doubt that the y in sud-dh-y might undergo saṃyogānta-lopa (the deletion of the final [consonant] of a conjunct), [the author] states the lopa-designation sūtra: adarśanaṃ lopaḥ. Since the context is the teaching of words (śabdānuśāsana), by "appearance" (darśana) here is intended [audible] hearing (śravaṇa) that has a word (sound) for its object. Non-appearance (adarśana) is the absence of [such] appearance; the compound is [formed] in the sense of absence [as an avyayībhāva]. From [the sūtra] sthāne 'ntaratamaḥ (A 1.1.50) the word sthāne is carried forward (anuvṛtti); and "sthāna" has [already] been said to mean "occasion of applicability" (prasaṅga). The meaning is: when, according to the śāstra, there is occasion for the hearing of some word (sound), the [resulting] non-hearing is termed lopa.

saṃyogāntasya lopaḥ — Accordingly [Bhaṭṭoji] says "prasaktasya" (of what is [otherwise] applicable). There, since the lopa is of the nature of non-audition (śravaṇa-abhāva), it follows by implication that there is not even an utterance for the sake of being heard. Why "of what is [otherwise] applicable"? [So that,] in cases like "dadhi," relying on the pratyaya-lakṣaṇa of the non-audible lopa of the kvip-suffix, the tuk-augment [that would come] by "hrasvasya piti kṛti tuk" (A 6.1.71) should not [wrongly] occur.

saṃyogāntasya lopaḥ. "padasya" (of a pada) is supplied as a governing heading (adhikāra). The analysis (vigraha) is "saṃyogo 'nto yasya" (that which has a saṃyoga as its end) [a bahuvrīhi]. The construal is: there is lopa of the pada ending in a saṃyoga. And it is not the whole pada that undergoes lopa, but rather, by the paribhāṣā "alo 'ntyasya" (A 1.1.52), only its final sound. This [Bhaṭṭoji] states with "saṃyogāntam" etc. Here the mention of "anta" in the sūtra is merely for the sake of clarity; because, since the saṃyoga is a qualifier of the pada, the sense "ending in that" is already obtained by [the paribhāṣā] "yena vidhis tadantasya" (A 1.1.72).

As for the [alternative] view that the mention of "anta" is for obtaining the analysis "saṃyogāv antau yasya" (that which has two saṃyogas as its ends) — otherwise, in "sudṛṣatprāsāda," the t before p would undergo lopa — that is not so, because the term saṃyoga is collectively-applying (vyāsajya-vṛtti) [holding of the consonant-group together, not of each member]; and to state a purpose for the "anta"-mention, presupposing that it applies to each individually (pratyeka-vṛtti), is pointless. This is set out at length in the Śabdaratna.

"iti yalope" [Bhaṭṭoji's phrase]: when, in "sudh y," the lopa of the y is [otherwise] applicable by this sūtra, the prohibition of it is now undertaken. yaṇaḥ pratiṣedho vācyaḥ — meaning: the prohibition of the saṃyoga-final-lopa for a yaṇ must be stated. [The words] "the saṃyoga-final-lopa of the y does not occur" are to be supplied [as the effect of] this vārttika. This vārttika is rejected (pratyākhyāta) in the source-work (ākara, the Mahābhāṣya).

Now here, since the y neither follows an ac nor is not-followed-by-an-ac (being in fact followed by an ac), the doubling is unavailable [both by "aco rahābhyāṃ dve" (A 8.4.46), which needs a preceding ac, and by "anaci ca" (A 8.4.47), which needs no following ac]; so [Bhaṭṭoji] states the doubling-rule [by vārttika]: yaṇo mayo dve vācye. The construal is: by this vārttika there is doubling of the y. Objection: If "yaṇaḥ" is a fifth-case (ablative) and "mayaḥ" a sixth-case (genitive), then what is obtained is "doubling of the may that follows a yaṇ"; but in the case at hand the y is neither what-follows-a-yaṇ nor a may. So how, by this vārttika, is there doubling of it? To this he says "maya iti" [etc.]. By [the word] "pakṣe" (on one alternative) it is indicated that the interpretation is intended in both ways (ubhayathā); the idea being that there is no ground for deciding between them (vinigamanā-viraha). In this vārttika too the word "vā" (optionally) is carried forward (anuvṛtti) from "yaro 'nunāsike 'nunāsiko vā" (A 8.4.45). And so [Bhaṭṭoji] states the result with "tad iha" [etc.]: "tat" is here an indeclinable [= therefore]. By what has been elaborated in this much of the passage, here — that is, in "sudh y" — owing to the optional doubling of both y and dh, four forms come about.

"ekadham ekayam": that which has one dh is ekadha; likewise ekaya. When neither dh nor y is doubled, the first form has one dh and one y. "dvidhaṃ dviyam": that which has two dh's is dvidha; likewise dviya. When both dh and y are doubled, the second form has two dh's and two y's. "dvidham ekayam": when the dh is doubled but not the y, [that is] the third form. "ekadhaṃ dviyam": when the dh is not doubled but the y is, [that is] the fourth form.

"suddhyupāsya": here the prohibition "na bhūsudhiyoḥ" (A 6.4.85) does not apply, because it is prescribed [only] before a vowel-initial sup-ending [whereas upāsya is a separate word]. Nor does "iko 'savarṇe" (A 6.1.127) [apply], because it is debarred by "na samāse" within a compound.

"maddhvari": Madhu is the name of a particular asura; his ari (enemy) is maddhvari — that is, Hari [Viṣṇu]. [madhu + ari, with yaṇ and doubling.] Here, for the u after the dh, the substitute v comes either by nearness of place (āntarya) or by the yathāsaṃkhya-paribhāṣā. Nor should it be objected that here there is no equality of place, since v has an additional dental place (u being labial, v dento-labial): for although complete (yāvat) equality of place is what conditions homogeneity (sāvarṇya), for [determining] the nearest (āntaratamya) even some (yatkiñcit) equality of place is a conditioning factor. Otherwise, in "cetā," "stotā," and the like, the e, o [as guṇa-substitutes] for i, u would not come about.

"dhātraṃśa": here it is only the t that is doubled, not the r — as will be stated just below at "aco rahābhyām" (A 8.4.46). "lākṛti": the analysis is "that whose form (ākṛti) is like the form of the ḷ-varṇa." Here, before the ā, the substitute for the ḷ-varṇa is l, by equality of the dental place. Objection: Since [v too shares] the dental place, and since there is no cause for overstepping the first [candidate], let its [the ḷ's] substitute be simply v — for it was said at "maddhvari" that even some equality of place conditions the [determination of the] nearest. [Reply:] This should not be said. For where all four yaṇs — y, v, r, l — are candidates to be prescribed: [only in the case of u] is the prescription of v, on the ground of labial-place equality, unrivalled (nissapatna) and possessed of scope (sāvakāśa), l there not obtaining because of the difference of its dental form of place. But in "lākṛti," for the ḷ-varṇa, both v and l are applicable. Here, relying on śabda-para-vipratiṣedha (conflict resolved in favour of the rule stated later in the wording), the l-prescription — being later (paratva) and being an exception (apavāda) — overrides the v-prescription. For if, on the ground that there is no cause for overstepping the first candidate, v were the substitute here too, then the l-prescription would have no scope at all (niravakāśa). Therefore here the substitute for the ḷ-varṇa is precisely l. But let this rest for now.

nādinyākrośe putrasya — [From above,] "dve" (doubling) is carried forward (anuvṛtti), and also "yar." ākrośa means reproach, censure (nindā). ādinī [in the sūtra] ends in the feminine ṅī (ṅīṣ), with an elided locative (lupta-saptamīka) [i.e., understood as "when ādinī follows"]. The meaning: when the word ādinī follows, the yar — namely the t (ta-kāra) — which is a constituent (avayava) of the word putra, does not undergo doubling (dvitva), when reproach (ākrośa) is to be understood. He states this with "putra-śabdasya…," which means "of the constituent of the word putra."

On putrādinī tvam asi pāpe: "putrādinī" [means] she whose habit (śīla) is to devour (√ad, attum) sons (putrān). The suffix ṇini [is added] by supy ajātau (3.2.78, "supy ajātau ṇinis tācchīlye"), and the feminine ṅīṣ by ṛnnebhyaḥ (4.1.5). [Construe:] "O wicked one (he pāpe)! you (tvam) are a putrādinī" — meaning "a slayer of [your] son(s)" (putra-ghātinī). Since such reproach is generally possible only with respect to women, in the sūtra "ādinī" is intended specifically as the ṅī-ending (feminine) form — so says Haradatta; Mādhava too holds thus. Here, the doubling of the t — the t following the u [of putra], which would be obtained by anaci ca (8.4.47) — is prohibited. But there is nowhere any occasion (prasakti) for the doubling of the r (repha); this will be stated just below.

On "ākrośe kim": i.e., what is the purpose of [specifying] ākrośe? Amara, in the section on indeclinables, [notes]: "kiṃ [is used] in questioning and in contempt (jugupsana)" — [glossing the interrogative use of "kim"]; and likewise it is to be understood in later cases of this type. On "tattva-kathane": when the word "putrādinī" is applied, with respect to a woman whose sons die of themselves, in stating an actual state of affairs (vastu-sthiti), there is no prohibition of the doubling, because in that case no censure (nindā) is conveyed.

tatpare ca — this is a vārttika. "tatpara" [means] that after which the word ādinī (tat = ādinī-śabda) follows (para); [thus, by bahuvrīhi]. The meaning: when a word putra that is itself followed by the word ādinī (ādinī-śabda-paraka) also follows [i.e., a second putra], the yar of [the first] word putra's constituent is not doubled. On "putra-putrādinī tvam…": the analysis (vigraha) is "she who devours the sons of a son" (putrasya putrān atti). Here, the word ādinī does not [directly] follow the first word putra, because of the intervening second word putra (vyavadhāna). Therefore, since the prohibition of doubling was not obtained by the previous sūtra, this [vārttika] is undertaken.

vā hatajagdhayoḥ. The meaning: when the word hata or the word jagdha follows, the yar of the constituent of the word putra optionally (vā) undergoes doubling. On "puttrahatī" [the source reads "putrahatī"]: this is the form when the t is doubled (ta-kāra-dvitva); the analysis is "she by whom a son has been slain" (putro hato yayā). Some [hold that] the feminine ṅīṣ [is added] by asvāṅga-pūrvapadāt. But in truth (vastutas tu), since "jāti-pūrvāt" is carried forward (anuvṛtti) there, it is proper [to derive] the ṅīṣ from [hatī's] belonging to the gaurādi class (gaurādi-tva). Since the audacity of son-slaying is possible only among women, only the feminine form is exemplified. On "putrahatī": this is the form when there is no doubling. Similarly putrajagdhī and puttrajagdhī. Since the optional doubling is already established generally by anaci ca (8.4.47), they say (āhuḥ) that this vārttika serves the purpose of a restriction (niyama): namely, that of the word putra [standing before] a kta-ending word, [the optional doubling holds] only when [the kta-word] is hata or jagdha, and only in the sense of reproach (ākrośa).

triprabhṛtiṣu śākaṭāyanasya — When three, four, or more consonants are conjoined, the doubling of the yar that follows the first vowel [applies] in the view of Śākaṭāyana; but in another view it does not — this is the meaning. In indra there is the optional doubling of the n; in rāṣṭra there is the optional doubling of the ṣ.

aco rahābhyāṃ dve — From [the sūtra] yaro 'nunāsike, "yar" in the genitive and "vā" ("optionally") are carried forward (anuvṛtti). "acaḥ" is a fifth case [ablative] used in the sense of direction (dig-yoga, i.e., "after a vowel"); "parābhyām" is to be supplied (śeṣa). "rahābhyām" too is a fifth case [ablative]; "parasya" is to be supplied. He states this with "acaḥ parābhyām..." and so on.

haryyanubhavaḥ: the analysis (vigraha) is hareḥ anubhavaḥ ("the experience of Hari"). When hari-anubhava stands, the i-kāra that follows the repha becomes yaṇ (y, by iko yaṇ aci); that [y] is doubled.

Then he gives an example of [a yar] following the ha-kāra: nahyyasti. When nahi-asti stands, the i-kāra that follows the ha-kāra becomes yaṇ (y); that is doubled.

In both cases here the ya-kāra does not follow a vowel (ac) and is itself followed by a vowel (ac); hence the doubling, which would not otherwise obtain [by anaci ca, which requires a consonant preceded by a vowel and followed by a non-vowel], is [specifically] prescribed. Here the doubling of the repha and the ha-kāra by anaci ca does not take place: in the doubling section, since "rahābhyām" names [the r and the h] directly (sākṣāt-śruta) as the conditioning ground (nimitta) in their capacity as repha and ha-kāra, their being subject to the operation signified by the word "yar" is blocked — by the maxim "between what is directly stated (śruta) and what is [merely] inferred (anumita), the directly stated is stronger." [The forms therefore stand as:] hary-y-anubhavaḥ, nah-y-y-asti.

halo yamāṃ yami lopaḥ — He states the rule of elision of the first ya-kāra in both [examples] with "halo yamāṃ yami lopaḥ." From "jhayo ho 'nyatarasyām" [A 8.4.62], "anyatarasyām" (optionally) is carried forward (anuvṛtti); and that [anyatarasyām], being an indeclinable in the guise of a case-form (vibhakti-pratirūpaka avyaya), functions in the sense of "or" (vā). The plural in "yamām" (of the yamas) is with reference to the plurality of usages, since in any single usage a plurality of yamas is impossible. "halaḥ" is a fifth-case (ablative) by association with a directional sense (digyoga); "parasya" ("of the following") is to be supplied. He states this with "halaḥ parasya yamaḥ…" etc. By this sūtra, in both examples, when the first ya-kāra is elided the single-ya-kāra form results; but on the alternative without elision the double-ya-kāra form results. Objection: the sūtra "halo yamām" need not be undertaken for this purpose, since, the doubling of the ya-kāra by "aco rahābhyām" [A 8.4.46] being optional, the double-ya form is established when doubling occurs and the single-ya form when it does not. Anticipating [the view] that this sūtra has no purpose here, he says "iti lopa…" Then what is its fruit? To this he says "lopārambha…": by "diti-aditi-āditya-patyuttarapadāṇ ṇyaḥ" [A 4.1.85], from the word "āditya" — which [already] ends in the ṆyA affix in the residual sense (śeṣārtha) — there is [again] the ṆyA in the sense of a deity (devatārtha). When "āditya + ya" stands, by "yasyeti ca" [A 6.4.148] the a [of ādity-a] before the ya-kāra is elided, and the ya-kāra [standing] after the ta-kāra is elided by this sūtra. Only when that [elision] occurs does the single-ya-kāra form "ādityam" arise, not otherwise; hence the undertaking of the sūtra is not fruitless. This is the point. Here the elision by "āpatyasya ca taddhite 'nāti" [A 6.4.151] does not [apply], because [this ṆyA] has the sense of "born of" etc. (jātādyartha), so it is not a patronymic (āpatya); here, rather, this [present] sūtra — [and] not that [āpatya rule] — has been adduced as being the proper (nyāyya) one. Objection: even so, in "māhātmyam" — in the sense "the state of a great soul (mahātman)," where, by "guṇavacana-brāhmaṇādibhyaḥ" [A 5.1.124] the affix ṣyaÑ [is added], with Ṭi-elision and ādivṛddhi — the ma-kāra after the ta-kāra ought also to be elided, since it is a yama and is followed by a yama in the form of a ya-kāra. To this he says "yamām…": from the yathāsaṅkhya-vijñāna — vijñāna being "that by which [something] is known," i.e. a sūtra (per Amara, "vijñānaṃ śilpa-śāstrayoḥ"), hence "from the yathāsaṅkhya sūtra" ["yathāsaṅkhyam anudeśaḥ samānām," A 1.3.10] — and thus, since the elision of a ma-kāra is obtained only when a ma-kāra follows, there is no elision when a ya-kāra follows. This is the meaning.

eco'yavāyāvaḥ — eco'yavāyāvaḥ. The analysis (vigraha) is: ay, av, āy, and āv (a dvandva). The word aci ("when an ac follows") is carried forward (anuvṛtti) into this rule from iko yaṇ aci (6.1.77). By the yathāsaṃkhya paribhāṣā (the principle of serial, term-for-term correspondence), one obtains: for e — ay; for o — av; for ai — āy; for au — āv. This he states with ecaḥ kramāt ("in place of ec, in order"). But according to the manner of the Bhāṣya on the yathāsaṃkhya sūtra [yathāsaṃkhyam anudeśaḥ samānām, 1.3.10], the assignment here is rather to be understood through the antaratama paribhāṣā (the principle of the nearest/most-similar substitute, sthāne'ntaratamaḥ, 1.1.50) itself. Here it would have been [more] appropriate to frame the sūtra as ico'ci yaṇayavāyāvaḥ [i.e., "in place of ic, when an ac follows, yaṇ, ay, av, āy, āv occur"].

tasya lopaḥ — In these substitutes [ay, av, āy, āv], in order to raise a doubt about the lopa [that would follow] once the ya-kāra and va-kāra received the it-designation by hal antyam [1.3.3, "a final consonant is it"], he states: tasya lopaḥ. This sūtra stands at the end of the it-designation section; there the word "tat" ("that") refers to the it mentioned nearby. He states this with "tasya itaḥ" ["of that it"]. The meaning is: by this sūtra, the lopa of the ya and va need not be feared.

To [the question] why, he says "uccāraṇa..." — if here there were lopa of the ya and va, then they simply would not have been uttered in the sūtra [that prescribes these substitutes]; for by the maxim "it is better not to touch mud from a distance than to wash it off [after being soiled]," this is the idea.

Then, [to the question] what use the it-designation is for these two, he says "evaṃ ca": since in the stated manner there is no lopa, the it-designation too does not apply here to the ya-kāra and va-kāra, because it would be fruitless — this is the meaning.

He illustrates in order with "haraye" and so on. When hare-e, viṣṇo-e, nai-akaḥ, pau-aka stand, the e-kāra and the rest [become] ay, av, āy, āv respectively [→ haraye, viṣṇave, nāyakaḥ, pāvakaḥ].

vānto yi pratyaye — vānto yi pratyaye. yi is a locative (saptamī). Thereby one obtains "beginning with y (yakārādi)," on the strength of the vārttika yasmin vidhis tadādāv algrahaṇe. There, yasmin is meant as a word in the locative; al is a synonym for varṇa (sound); the sense is: when an operation is [taught] with a sound denoted by a locative-ending word, that operation is to be understood as [applying] to what begins with that sound. This is an apavāda (exception) to [the paribhāṣā] yena vidhis tadantasya (1.1.72). vāntaḥ = that which has a va-kāra at its end. The substitutes av and āv taken up in the previous sūtra are what is intended. And when the question arises, "for which two [vowels] do these two [substitutes] occur?", "of o and au" is obtained by implication (arthāt), since in the previous sūtra it was established that these two [av, āv] are the substitutes precisely for those two [o, au]. This he states with yakārādau, etc.

gavyam — when "go-ya" stands, since the o is not followed by an ac (ac-parakatva being absent), eco'yavāyāvaḥ (6.1.78) does not apply, and so the substitute av is here prescribed [by the present rule]. And here "ya" does have affix-status; this he indicates with gopayasor yat — [supply:] "by this sūtra, after the word go, in the sense of modification (vikāra), the affix yat [is added]." nāvyam — when "nau-ya" stands, since the au is not followed by an ac, eco'yavāyāvaḥ (6.1.78) does not apply, and this [rule] is the [governing] statement; and here "ya" has affix-status, which he indicates with nauvaya[dharma]…. In gavyam and nāvyam, the va-kāra is not elided by lopaḥ śākalyasya (8.3.19) or by [vā] hali sarveṣām (8.3.22), because those two rules have the end of a pada (padānta) as their domain, whereas here pada-hood is precluded (bādha) by bha-designation.

gor yūtau chandasy upasaṃkhyānam — In chandas, i.e. the Veda, when the word yūti follows, in place of the o of the component go, av comes as the vānta substitute: such an upasaṃkhyāna — an additional statement (adhikavacana) — is to be made; thus the author of the vārttika [Kātyāyana] instructs [the sūtrakāra]. The example is ā no mitrāvaruṇā ghṛtair gavyūtim ukṣatam. gavyūti = pasture, the ground over which cattle range (gopracārabhūmi). In the sense of locus (adhikaraṇa) — "in which cows are yoked/mingled (yūyante miśryante)" — the affix ktin [is added] to the root yu: so Bhaṭṭabhāskara in his Veda-bhāṣya. Here, since the word yūti does not have affix-status, when it follows, vānto yi pratyaye (6.1.79) does not apply; hence this [supplementary] statement.

adhvaparimāṇe ca — the meaning is: even when a specific measure of road/distance (mārgaparimāṇaviśeṣa) is to be conveyed, when the word yūti follows, the upasaṃkhyāna of the vānta substitute av for the o of the component go is to be made. This is [needed] so that it apply also in ordinary (laukika) usage. Although the previous vārttika [gor yūtau chandasi] ought to have been adduced only in the Vedic-derivation section (vaidikaprakriyā), it is nonetheless adduced here for ease of grasping the anuvṛtti [that carries it into this second vārttika]. gavyūtiḥ — the extent of road over which the word go is yoked, mingled, [and] heard, that much road is a gavyūti, [namely] a pair of krośas (krośayuga): "gavyūti (feminine) is a pair of krośas," says the Amarakośa. The affix ktin [is added] to the root yu in the locus sense. How is there a long vowel (dīrgha) in the root yu here? To this he says ūtiyūti [SK 3274]: the idea is that the long vowel comes precisely from the [irregular] formation (nipātana). nipātana is the [direct] citation of a form whose derivation is [taken as already] accomplished (siddhaprakriya).

Objection: In gavyam and nāvyam the elision of v by lopaḥ śākalyasya and hali sarveṣām is absent because pada-hood is precluded by bha-hood; yet in gavyūtiḥ the elision [of v] might occur, because there go has pada-hood on the strength of an inner case ending (antarvartinī vibhakti), and because — since the word yūti is not a y-initial (yajādi) sup-affix (svādi) — bha-hood does not obtain when it follows. To this he replies with vānta ity atra, etc.: In vānto yi pratyaye, before the va-kāra — or in gor yūtau [chandasi], before the cha-kāra — a va-kāra is fused in (praśliṣyate) in the preceding portion; this is the construction. Objection: Then in v-vāntaḥ and v-chandasi, why is the [fused-in] va-kāra not heard? He answers: by its elision through lopo vyor vali — [supply:] "[so it becomes] concealed (antarhita)." Objection: Once the praśleṣa is done, what is gained? He answers with tena ("thereby"): In the sūtra vānto yi, the [va-kāra] being fused in before the va-kāra is a qualifier (viśeṣaṇa) of vānta; and because it is a qualifier, by the tadantavidhi (1.1.72) one obtains "ending in va-kāra." There, by the force of stating vānta-hood over again of what is [already] vānta, one obtains that the vānta substitute is one possessing an audible (śrūyamāṇa) va-kāra. And in gor yūtau this very mention of vānta recurs (anuvṛtti), so there too one obtains that the vānta substitute possesses an audible va-kāra: even the [va-kāra] fused in before the cha-kāra in gor yūtau is a qualifier of the vānta carried forward from the previous sūtra; so, vānta-hood being obtained by the tadantavidhi, by the force of restating vānta-hood of the vānta [substitute], one obtains "possessing an audible va-kāra" — this is the sense. Objection: Even granting this much — that the va-kāra of the av substitute in gavyūti is audible — the apprehension of [its] elision is not [thereby] removed. He answers: "that is to say, the va-kāra is not elided" — the statement that the va-kāra is audible amounts to the sense "the va-kāra is not elided," since non-audition (aśravaṇa) is precisely the meaning of the word lopa (elision).

Although in the sūtra vānto yi the praśleṣa of the va-kāra has no purpose in its examples gavyam and nāvyam — for there the said elision does not apply, pada-hood being precluded by bha-hood; and in [forms] like lavyam the elision does not apply because there is no pada-hood in the first place (svata eva); and in gavyati ("desires a cow," gām icchati), though it ends in the affix kyac, there is no occasion for elision of the va-kāra, since pada-hood is absent there by the restriction na kye — nevertheless in the vārttika gor yūtau, when the [praśleṣa] is carried into it (anuvṛtti), it does have a purpose, [namely] in gavyūtiḥ. And thus the praśleṣa is properly [placed] in the vārttika itself; it is precisely on account of this awkwardness (asvarasa) that [Bhaṭṭoji] said "or from the cha-kāra (chakārād vā)."

But in truth (vastutas tu) the praśleṣa of a va-kāra is untenable, since on that supposition the fused-in [va-kāra] in the sūtra and in the vārttika would [wrongly] be heard/pronounced. Nor can it be said that its non-audition is due to elision by lopo vyor [vali]; for that would conflict with the Bhāṣya on the sūtra lopo vyor vali, which [states] "because there is no example of the elision of a va-kāra, the mention of va-kāra [in that sūtra] need not be made." Therefore the elision of the fused-in va-kāra here is to be described as [purely] Vedic (chāndasa). In the Prakriyāprakāśa, however, it is said, on the maxim "an operation dependent on a [technical] designation (saṃjñā) is non-obligatory (anitya)," that there is no va-kāra elision in gavyūtiḥ. Others, again (anye tu), say that on the authority of the Bhāṣya — [which states] that in the sūtra iko guṇavṛddhī (1.1.3), in ato lāntasya, the va-kāra is indicated by its elision (luptanirdiṣṭa) — there is, in some places (kvacit), an elision of a va-kāra.

dhātos tan-nimittasyaiva — Objection: In oyate and auyata too, the vān-substitute [prescribed by vānto yi, A 6.1.79] would apply to the o and the au. Anticipating this, [Bhaṭṭoji] restricts that sūtra vānto yi by [the present rule] dhātos tan-nimittasyaiva. The expressions "ecaḥ" [of an ec] and "vānto yi pratyaye" [the vān-substitute, before a y-initial suffix] are carried forward (anuvṛtti). "tan-nimitta" means "that whose occasioning ground (nimitta) is that y-initial suffix." The sense is: when a y-initial suffix follows, the vān-substitute occurring for an ec of the dhātu applies only to that ec which is occasioned by the y-initial suffix, not to any other. He states this with "yādau pratyaye", etc.

lavyam — from lūñ "to cut"; the suffix yat by aco yat (A 3.1.97); guṇa of the ū [of lū], giving o, by sārvadhātukārdhadhātukayoḥ (A 7.3.84). Since that o is a component of the dhātu and is occasioned by the y-initial suffix, the vān-substitute [applies, giving lav-yam]. avaśyalāvyam — the suffix ṇyat after lūñ by or āvaśyake (A 3.1.125); vṛddhi of the ū, giving au, by aco ñṇiti (A 7.2.115); avaśyam is an indeclinable; the compound is formed by inclusion in the mayūravyaṃsakādi class; the m [of avaśyam] is elided by the [rule] that the m of avaśyam is elided before a kṛtya suffix. Here the au, being a component of the dhātu and occasioned by the y-initial suffix, [undergoes] the vān-substitute [giving āv, avaśyalāvyam].

Objection: In lavyam and avaśyalāvyam the vān-substitute is already established by vānto yi itself; hence this sūtra serves only as a restriction (niyama), "only of that occasioned by it" — such is the settled position. He asks the purpose of that restrictive rule with "tan-nimittasyaiva iti kim" — i.e., what is the purpose of the restriction?

oyate — from veñ "to weave" (tantu-santāna); laṭ in the passive (karman); ātmanepada by bhāvakarmaṇoḥ (A 1.3.13); the suffix yak [sārvadhātuke yak, A 3.1.67]; by vacisvapiyajādīnām (A 6.1.15), saṃprasāraṇa of the va [giving u], and pūrvarūpa [the single substitute being the preceding u]; lengthening of the u to ū by akṛtsārvadhātukayoḥ (A 7.4.25); then the ū combined with the preceding preposition ā yields, by ād guṇaḥ (A 6.1.87), the guṇa o [ā + ū → o, oyate]. That o, though by parādivadbhāva [treatment as the initial of the following] a component of the dhātu, is not occasioned by the y-initial suffix; hence no vān-substitute.

Objection: But here, in accordance with the Bhāṣya on the dhātu-elision sūtra, the o produced by ād guṇa, depending on two padas [the preposition and the root], is bahiraṅga, and hence asiddha (regarded as not having taken effect) when the vān-substitute — an antaraṅga operation — is to be performed; so, there being no o [at that stage], there is no occasion for the vān-substitute at all [and oyate is not a clean illustration]. Out of this uneasiness he states [a better example], auyata.

auyata — from the bare veñ, in the passive, with the ending taṅ; the ātmanepada, etc., as before; the augment āṭ by āḍ ajādīnām (A 6.4.72); vṛddhi [au] by āṭaś ca (A 6.1.90). That au too, though by parādivadbhāva a component of the dhātu, is not occasioned by the y-initial suffix; hence no vān-substitute. Here the vṛddhi is not bahiraṅga, since it does not depend on two padas [the āṭ augment belongs to the same pada].

By the maxim siddhe saty ārambho niyamārthaḥ [when [something] is already established, a fresh undertaking serves a restrictive purpose], the restrictive-rule status [of this sūtra] is established; but the word eva ("only") in tan-nimittasyaiva serves to exclude the opposite restriction. For in the absence of eva, [the sense could be:] "if the vān-substitute [applies] to an ec occasioned by a y-initial suffix, then [it applies] to an ec of the dhātu only" — i.e., the opposite restriction ("only of the dhātu") might be understood. And in that case, in bābhravya the vān-substitute would not apply: bābhravya = descendant of Babhru; the suffix yañ by madhubabhrvor brāhmaṇakauśikayoḥ (A 4.1.106); guṇa of the u [of babhru], giving o, by or guṇaḥ (A 6.4.146); since that o, though occasioned by the y-initial suffix, is not a component of a dhātu [babhru being a nominal stem], the vān-substitute would [wrongly] fail. Therefore, to secure the restriction to the intended occasioning ground, [there stands] the word eva in tan-nimittasyaiva.

kṣayyajayyau śakyārthe — kṣayyajayyau śakyārthe. The meaning is: in the sense of śakya, the words kṣayya and jayya function as denoters.

Objection: no operation to be prescribed is seen here [the sūtra merely gives two words]. In response he says "yānta[-ādeśa-nipātana]." A nipātana is the direct mention of a word-form whose derivation is [taken as already] accomplished, without a specific individual rule (prātisvika-vidhi). And so the following prescription results here: when a suffix beginning with y, in the śakya sense, follows the root kṣi ("kṣaye," to destroy) or ji ("jaye," to conquer), the yānta-ādeśa [the ay-substitute, by *eco 'yavāyāvaḥ*, A 6.1.78] applies. Framing an individual rule "in the śakya sense kṣi and ji take the yānta [substitute]" would be cumbersome (gaurava); hence the nipātana [is adopted] instead — this is the idea.

"kṣayya": under the adhikāra "kṛtyāḥ," the suffix yat is added to the root kṣi by *aco yat* (A 3.1.97). And that [yat] conveys the śakya sense, per *śaki liṅ ca* [the kṛtya-suffixes convey the sense of śakya and of the liṅ (potential)]; its meaning is "in the śakya sense the liṅ occurs, and by *ca* the kṛtya-[suffixes as well]." By *sārvadhātukārdhadhātukayoḥ* (A 7.3.84), guṇa replaces the i [of kṣi], giving e. Since that [e] is not followed by a vowel (ac) [but by y], the ay-substitute [which *eco 'yavāyāvaḥ* prescribes only before a following vowel] is not [otherwise] obtained; here it is fixed by nipātana. "jayya" too is as before.

"Why 'in the śakya sense'?": the question is, what is the purpose of "śakyārthe"? "kṣetum" etc.: kṣeyaṃ pāpam = "sin fit to be destroyed"; jeyaṃ manaḥ = "a mind fit to be conquered" — such is the construal. The yat [here] comes by *arhe kṛtyatṛcaś ca* [in the sense of "worthy/deserving," the kṛtya-suffixes and tṛc occur]; and that [yat] is not in the śakya sense, so there is no yānta-ādeśa here — this is the meaning [hence kṣeya, jeya, with guṇa e retained].

krayyas tadarthe — krayyas tadarthe. This too serves to fix the yānta-ādeśa (the ay-substitute) by nipātana [like the previous kṣayya/jayya sūtra]. He explains the word "tadartha" with "tasmai" etc. In the word "krayya," the root krīñ which is the base of the yat-suffix — its expressible meaning, namely "kraya" (buying) in the form of exchange of goods (dravya-vinimaya) — that is the prakṛti-artha (meaning of the base), intended by the word "tat." "This is for that" = tadartha = an object meant for buying (kraya-artha). And so it results: "when an object meant for buying is to be understood, and a suffix beginning with y follows the root krīñ [the ay-substitute applies]." Framing a rule "the yānta [substitute] of krīñ in the tadartha sense" would be cumbersome (gaurava); hence nipātana is resorted to. And the "being meant for buying" is here intended as productive of the [distinguishing] result (phalopadhāyaka); he says this from "kretāraḥ" up to "krayyam." If mere fitness (yogyatā) were taken, "tadarthe" would fail to distinguish [would be non-exclusive] — this is the idea.

To krīñ, in the object sense (karmaṇi), the suffix yat by *aco yat* [A 3.1.97]; by *sārvadhātuka[ārdhadhātukayoḥ]* [A 7.3.84] guṇa gives e (krī → kre). Here, since the e is not followed by a vowel (ac), the ay-substitution is [prescribed] by this [nipātana]. "kreyam anyat": it means grain and the like, stored in a house etc. for the purpose of eating and so on; the yat [there comes] by *arhe kṛtyatṛcaś ca* [in the sense of "worthy," the kṛtya-suffixes and tṛc]. Here the ay-substitution does not apply, because the result-producing "being meant for buying" is absent there — this is the meaning [hence kreya, with guṇa e retained and no ay].

lopaḥ śākalyasya — In hare ehi, viṣṇo iha, śriyai udyataḥ, gurau utka, once the ay, av, āy, av substitutes for the ec-vowels (e, o, ai, au) respectively have been made, [this rule] prescribes the optional (vaikalpika) elision of the resulting y and v — that is, lopaḥ śākalyasya. From bho bhago agho apūrvasya yo 'śi (A 8.3.17), 'apūrvasya' and 'aśi' are carried forward (anuvṛtti). From vyor laghuprayatnataraḥ… (A 8.3.18), 'vyoḥ' is carried forward, meaning 'of the v and y.' And 'apūrvasya' is a qualifier of 'vyoḥ': 'apūrva' means that which has an a-varṇa as its predecessor (a-varṇaḥ pūrvo yasmāt). 'Of the a-preceded v and y' — the singular [in apūrvasya] is meant to construe distributively with each (pratyekam). The adhikāra 'padasya' [A 8.1.16] is qualified by the a-preceded y and v; and because it is so qualified, tadanta-ness is obtained (by the tadantavidhi): 'of a pada ending in [such a] y or v.' Hence the meaning: a y or v preceded by an a-varṇa — the pada ending in that undergoes elision, when a sonant (aśi) follows. By the alo 'ntyasya paribhāṣā (A 1.1.52), 'of the final [sound] of such a pada' is obtained. From the mention of Śākalya it is understood that [this elision] does not occur in another view (matāntara) [i.e. it is optional]. So [Bhaṭṭoji] states the resultant [rule] with 'of the y and v preceded by an a-varṇa' etc. Although 'vyoḥ' is carried forward [in the order v-then-y], because in the alphabet (varṇa-samāmnāya) the y comes first, [Bhaṭṭoji] said 'ya-va' [in that order].

Objection: Why should the mention of Śākalya not be for the sake of a person-based restriction (vyavasthā) — namely that only persons of Śākalya's lineage follow this elision, and not others? — Response: This cannot be maintained; for it is the settled doctrine — in the Bhāṣya on the sūtra na vā… ('vibhāṣā'), and where iko hrasvo 'ṅyo gālavasya, prācām avṛddhāt, and the like are taken up — that the mention of teachers such as Gālava, and the mention of particular regions such as 'of the easterners' (prācām) and 'of the northerners' (udīcām), is made simply to signify optionality (vikalpa), without distinction, and not to regulate usage according to the particular lineage or the particular region.

Objection (nanu): In hare ehi, once, after the ay-substitute, the y is elided, then by om āṅoś ca (A 6.1.95) pararūpa (single-substitution as the following vowel) would result; and in viṣṇo iha, śriyai udyataḥ, gurau utka, once, after the av, āy, av substitutes, the v/y is elided, guṇa would result by ād guṇaḥ (A 6.1.87). — To this [Bhaṭṭoji] says 'pūrvatra' [pūrvatrāsiddham]: because the elision of the v and y is asiddha, [the preceding vowel] remains separated by them (vyavahita) and so is not immediately followed by a vowel (ac-parakatva-abhāva); hence the apprehended vowel-sandhi (ac-sandhi) does not occur. Thus the ik-sandhi and the ec-sandhi have both been expounded.

Apprehending an overextension (atiprasaṅga) in both, [Bhaṭṭoji] resolves it with 'kāni' etc. '[Although]… nevertheless…' is to be supplied: the construction is that in kāni santi and kau staḥ, although the yaṇ and āv substitutes would be obtained, nevertheless they do not occur. Objection: Here the i and au are followed by s (sakāra-paraka), not by a vowel; so how could the yaṇ and āv be obtained? — To this he says: 'by the sthānivadbhāva of the al-lopa of as.' From the root as (of the ad-ādi class), in the laṭ, third person plural, the form is santi; in the third person dual, the form is staḥ. In both, by śnasor allopaḥ (A 6.4.111), there is elision of the initial a of the root — this is the situation. There, since the al-lopa is sthānivat and hence counts as an ac, the i and au become followed by a vowel, so the yaṇ and āv substitutes would be obtained.

Objection: sthānivad ādeśo 'nalvidhau (A 1.1.56) [makes the substitute sthānivat except in an alvidhi]; so sthānivadbhāva is not possible here — because the yaṇ and āv operations, which proceed by taking as their conditioning follower (para-nimitta) the a that was the sthānin of the al-lopa, rest on the al of the sthānin (sthāny-al-āśraya) [and are thus an alvidhi, excluded by A 1.1.56]. — Response: because sthānivadbhāva is [nonetheless] established by acaḥ parasmin pūrvavidhau (A 1.1.57). The point: the al-lopa, being prescribed when a kṅit [affix] follows, is a substitution for an ac conditioned by a following element (para-nimittaka ac-ādeśa); and so, for the yaṇ and āv operations upon the i and au — which are seen as preceding the a that was its sthānin — acaḥ parasmin operates [restoring sthānivadbhāva even in this alvidhi].

ekaḥ pūrvaparayoḥ — [First an objection carried over from the preceding discussion:] Then why do the yaṇ and āv substitutes (yaṇāv al-ādeśau) for al not take place here? To this he replies: because of the prohibition of that [sthānivadbhāva] in operations upon a pada-final, by the sūtra na padāntadvirvacana… (A 1.1.58). The meaning: the yaṇ and āv substitutes that come in place of the pada-final i and au, being effected as the last constituent of a pada, [would each require treating the substituted vowel as sthānivat]; but the sthānivadbhāva of a para-conditioned vowel-substitute (paranimittaka aj-ādeśa) and of the [attendant] elision (lopa) is denied [in pada-final operations by A 1.1.58] — hence [the yaṇ-substitution does not apply here]. Now, in order to set out the ekādeśa-sandhi, he says: "having established ekaḥ pūrvaparayoḥ as an adhikāra." The meaning: this sūtra, consisting of two words, having been recited for the sake of its anuvṛtti into the following [rules], several sandhis will now be prescribed.

ādguṇaḥ — "āt" is a fifth case [ablative, "after an a-varṇa"], not a ta-para specification [i.e., the final t is the ablative ending, not a marker restricting the vowel's length]. "ekaḥ pūrvaparayoḥ" ["a single [substitute] in place of the preceding and the following"] is in force as an adhikāra. He states this with "avarṇād aci..." and so on.

upendra: when upa-indra stands, of the a-kāra following the pa-kāra and of the i-kāra following that — the preceding and following [vowels], whose places of articulation are throat (kaṇṭha) and palate (tālu) — the single guṇa of that same nature, viz. the e-kāra, [is the substitute], by closest homogeneity (āntaratamya).

Then, to recall the implied denotation (lakṣaṇā) stated in the context of the pratyāhāras — namely that [a pratyāhāra also] denotes what is denotable by its own [members] — he gives an example involving a long [vowel]: rameśa. When ramā-īśa stands, in place of the ā-kāra and the ī-kāra, as before, the single guṇa e-kāra [is substituted].

gaṅgodakam: when gaṅgā-udaka stands, in place of the ā-kāra and the u-kāra, whose places are throat and lips (kaṇṭha-oṣṭha), the single guṇa of that same nature, viz. the o-kāra [is substituted].

In kṛṣṇarddhi, when kṛṣṇa-ṛddhi stands, ādguṇa presents itself [as applicable]. Here there are two sthānins: the a-kāra and the ṛ-kāra. For these two, all three guṇas — the a-kāra, the e-kāra, and the o-kāra — present themselves [as candidates], since [each] shares throat-place homogeneity with the a-kāra without distinction.

uraṇ raparaḥ — But with ḷ as the sthānin (that which is to be replaced), no [substitute] has homogeneity of place [with ḷ]; hence, when the expectation arises "which guṇa takes place there?", this [rule] is begun. As for [Bhaṭṭoji's] "it has been stated (ityuktam)": supply "in the aṇudit sūtra" [aṇuditsavarṇasya cāpratyayaḥ, A 1.1.69, whereby ṛ denotes the thirty homogeneous sounds]. The word "uḥ" is the genitive singular of ṛ. By the paribhāṣā ṣaṣṭhī sthāneyogā [A 1.1.49], "in place of" (sthāne) is obtained. Since this is a mere restatement (anuvāda), even though that paribhāṣā does not [directly] apply, the word sthāne recurs (anuvṛtti) from there. Hence he says: "the aṇ [that comes] in place of that." The word sthāne recurs also from sthāne 'ntaratamaḥ (A 1.1.50). It has been said that sthāna means "occasion" (prasaṅga); the intended sense is "in the condition of being an occasion [for substitution]." Hence he says: "it operates only as raparaḥ."

Here "r" is intended as a pratyāhāra. And so it is obtained that [the substitute] operates as one followed by repha (r) or followed by lakāra (l). He shows the distribution between the two by "there (tatra)": the construction is — between the repha-followed and la-followed [substitutes], in kṛṣṇarddhiḥ [it is] ar, in tavalkāra [it is] al. To [the question] "whence this distribution?" he says: "from maximal homogeneity (āntaratamya)." When the three guṇas [a, e, o] present themselves [as candidate substitutes for ṛ], the aṇ, namely a — since it is repha-crested or la-crested, i.e. of the form ar or al — [is fixed thus]: by homogeneity of place, in its a-portion with the sthānin a, in its repha-portion with ṛ, and in its la-portion with ḷ. So, because of the place-homogeneity of a and ṛ, in place of ṛ ar alone occurs; and in place of ḷ al alone occurs. But e and o, [though] guṇas, simply do not occur, because they lack place-homogeneity with ṛ and ḷ. Nor should it be objected: "how is it that e and o would not [themselves] be rapara?" — for here the aṇ is grasped by the earlier ṇ [i.e. the short pratyāhāra aṇ = a, i, u, which excludes e, o], as indicated by formations such as praśāstṝṇām; enough on this.

As for "in the alternative there is doubling": from the root ṛdh, with the affix ktin, the t [of ti] becoming dh by jhaṣas tathor dho 'dhaḥ [A 8.2.40], the form ṛddhiḥ — [with a doubled dh] — is the natural form. There, upon the ar substitution [in kṛṣṇa + ṛddhi → kṛṣṇarddhi], the dh following the repha optionally (kadācit) undergoes doubling by aco rahābhyāṃ [dve] — this is the meaning.

jharo jhari savarṇe — From jhayo ho 'nyatarasyām [A 8.4.62], anyatarasyām (optionally) is carried forward (anuvṛtti). From halo yamāṃ yami lopaḥ [A 8.4.64], "hal" and "lopa" (elision) are carried forward. Hence he says: "a jhar following a hal," etc. He states that here three forms result, by "when there is no doubling," etc.: when the dh — namely the root-final dh of ṛdh following the repha — does not undergo doubling by aco rahābhyāṃ, and does undergo elision by jharo jhari, the form has a single dh (ekadhakāra) — this is the meaning. As for "when [elision] does not take place (asati)": "when there is no doubling" is drawn along [from before]; when that same dh of ṛdh has no doubling and there is no elision, the form has two dh (dvidhakāra); likewise, when that same dh of ṛdh has doubling and there is elision, the form again has [just] two dh — this is the meaning. As for "when [elision] does take place (sati)": when that same dh of ṛdh has doubling and no elision, the form has three dh (tridhakāra) — this is the meaning. Although, in the dvidha alternative the first dh, and in the tridha alternative the middle dh, become d by jaśtva [via jhalāṃ jaśo 'nte, A 8.2.39], nevertheless — with a view to the [stage of] derivation, or by the principle of what was formerly [present] (bhūtapūrvagatyā) — the designations "twofold" and "threefold" are used, and there is no fault.

Now, regarding tavalkāra, he says "yaṇ": on the interpretation "a may [consonant] following a yaṇ is doubled," the k is doubled [by that rule], while the l is doubled by anaci ca — this is the meaning. As for "thereby (tena)": [it means] "by the optionality of doubling of both the l and the k." As for "doubling of the l alone": when the l is doubled and the k is not, the first form has two l and one k — this is the meaning. As for "of neither (na ubhayoḥ)": when neither the l nor the k is doubled, the third form has one l and one k — this is the meaning. As for "of both (ubhayor api)": when both the l and the k are doubled, the fourth form has two l and two k — this is the meaning. As for "in tavalkāra and the like": by "and the like (ādi)," forms such as mamalkāra are included.

vṛddhir eci — vṛddhir eci. From ādguṇaḥ (A 6.1.87) the word āt, ending in the fifth (ablative), is carried forward (anuvṛtti). [The words] ekaḥ… pūrvaparayoḥ (A 6.1.84) are the governing adhikāra. He states this with "ādeci…" and the rest. "guṇāpavādaḥ" (an exception to guṇa) — the intention being that this [rule] is undertaken precisely where ādguṇaḥ would [otherwise] have applied. "kṛṣṇaikatvam" — [the compound] kṛṣṇasya ekatvam is a genitive-tatpuruṣa (ṣaṣṭhī-samāsa). The prohibition [of such a compound] by pūraṇaguṇa… (A 2.2.11) is, however, anitya (non-invariable), as will be stated in the mūla at that very place. Others [hold that] "kṛṣṇa", ending in the vocative (sambuddhi), is a separate word. Here, in place of the a and the e — both of which have kaṇṭha-tālu (guttural-palatal) as their place of articulation — the single substitute is an ai of that same [guttural-palatal] nature. When "gaṅgā-ogha" stands, in place of the ā and the o — both having kaṇṭha-oṣṭha (guttural-labial) as their place — [the single substitute] is an au of that same nature. So too in "devaiśvaryam" and "kṛṣṇautkaṇṭhyam". But in truth (vastutas tu), [the point] that words such as saṅkhyā (number-words) are not guṇa-denoting (guṇavacana) we shall state at that very place.

etyedhatyūṭhsu — etyedhatyūṭhsu. The analysis (vigraha) is: eti, and edhati, and ūṭh. The forms eti and edhati are specifications of roots by means of śtip (the tip used in citing a root), per "[īk]-śtipau dhātunirdeśe"; the roots intended are iṇ "in the sense of going" (iṇ gatau) and edh "in the sense of thriving" (edha vṛddhau). "[vṛddhir] eci" is carried forward by anuvṛtti [from A 6.1.88]. By the paribhāṣā "yasmin vidhis tadādāv algrahaṇe" there is grasping-as-that-beginning (tadādigrahaṇa), so that "beginning with ec (ejādau)" is obtained; and this [qualification] qualifies eti and edhati only, not ūṭh — since ūṭh's beginning with ec is impossible. "ekaḥ pūrvaparayoḥ" (A 6.1.84) is in force as an adhikāra; and from "ād guṇaḥ" (A 6.1.87) the word "āt", ending in the ablative (fifth) case, is carried forward. This he states with "avarṇāt" etc.

upaiti — the root iṇ, laṭ; tip; luk of śap [iṇ being adādi]. By "sārvadhātuka[ārdhadhātukayoḥ]" the ikāra takes guṇa, [becoming] the ekāra: eti. When "upa-eti" stands, by this [rule] vṛddhi gives upaiti.

upaidhate — the root edh, laṭ; ātmanepada, the ta-substitute; śap; by "ṭita ātmanepadānāṃ [ṭer e]" the etva [gives edhate]. When "upa-edhate" stands, by this [rule] vṛddhi gives upaidhate.

praṣṭhauh[aḥ] — "one who draws at the front" (praṣṭhaṃ vahati) is praṣṭhavāṭ: by "vahaś ca" the affix ṇvi; by "ver apṛktasya" elision of the v [of ṇvi]; by "ata upadhāyāḥ" vṛddhi of the penult [vah → vāh]. Then, saṃprasāraṇa continuing by anuvṛtti from "śasivasoḥ saṃprasāraṇam", by "vāha ūṭh" (A 6.4.132) [the semivowel] becomes ūṭh, which bears the saṃjñā saṃprasāraṇa; then by "saṃprasāraṇāc ca" (A 6.1.108) there is pūrvarūpa. When "praṣṭha-ūh" stands, by this [rule] there is vṛddhi, [namely] the aukāra: praṣṭha + ūh → praṣṭhauh.

Objection: Here the mention of eti and edhati is pointless (vyartha), since in upaiti and upaidhate vṛddhi can be had by "vṛddhir eci" (A 6.1.88) alone [they begin with ec]. The mention of ūṭh too is pointless, [obtainable] by "vṛddhir eci" alone; [for] the ūṭh-mention in "vāha ūṭh" is itself purposeless, since [the result] is accomplished by saṃprasāraṇa [alone], while the guṇa follows by pratyayalakṣaṇa — just as the vārttika-author has stated, "vṛddhir ejgrahaṇāt". So the ūṭh-mention [in this sūtra] should not be made; let the sūtra [6.4.132] read simply "vāha", meaning "let there be saṃprasāraṇa of the vāh that is a bha-stem", so that only saṃprasāraṇa is enjoined. Then, the v becoming u by saṃprasāraṇa and pūrvarūpa having occurred, when "praṣṭha-uh" stands, relying on pratyayalakṣaṇa with the ṇvi affix [treated as] following, the u undergoing laghūpadhaguṇa to become o, when "praṣṭha-oh" stands, vṛddhi follows by "vṛddhir eci" [→ praṣṭhauh]. Such is the vārttika's meaning; therefore the sūtra "etyedhatyūṭhsu" is [held to be] pointless.

Response: To this he [Bhaṭṭoji] says "pararūpaguṇāpavādaḥ" — the meaning being that for eti and edhati this rule is an exception to "eṅi pararūpam", and before ūṭh an exception to "ād guṇaḥ". [And the pūrvapakṣa fails:] for if only saṃprasāraṇa were enjoined for vāh and not ūṭh, then, the v having become u by saṃprasāraṇa and pūrvarūpa having taken place, the laghūpadhaguṇa of that u would not occur — because, with respect to that guṇa's being performed, the pūrvarūpa, which depends on the saṃprasāraṇa, which depends on the bha-saṃjñā, which depends on the external (bahirbhūta) śas affix, is bahiraṅga and therefore asiddha here [relative to the antaraṅga guṇa]. And so, guṇa being absent: when "praṣṭha-u-āh" stands, and after pūrvarūpa "praṣṭha-uh" stands, by "ād guṇaḥ" guṇa gives o, and "praṣṭhoha" would [wrongly] result. But when the ūṭh that bears the saṃjñā saṃprasāraṇa is enjoined [by "vāha ūṭh"]: "praṣṭha-ū-āh" standing, and after pūrvarūpa "praṣṭha-ūh", since laghūpadhaguṇa is inapplicable [ūṭh being long ū, not a light penult] so that there is no o but only ū present, and since "vṛddhir eci" is thereby inapplicable [there being no ec] while "ād guṇaḥ" would obtain — to block that, by the vṛddhi of "etyedhatyūṭhsu", "praṣṭhauh" is [correctly] established. Hence the ūṭh-mention is necessary. So the vārttika cited under the "vāha ūṭh" sūtra is refuted (dūṣita) in the Bhāṣya. Let this rest for now (āstāṃ tāvat).

upetaḥ — the root iṇ, laṭ; tas; luk of śap. By "sārvadhātukam apit [being treated as ṅit]" the ending tas is ṅit, so when it follows the ikāra does not take guṇa; and here, the iṇ-root lacking ejāditva [it is short i, not e], there is no vṛddhi before it [→ upa-itaḥ → upetaḥ by ād guṇaḥ].

mā bhavān predidhat ("may your honour not thrive") — the root edh, ṇic; luṅ; tip. By "nityaṃ ṅitaḥ" elision of the ikāra [of ṇi]; by "ṇiśri-dru-sru-bhyaḥ [kartari caṅ]" the caṅ; by "ṇau caṅy upadhāyā hrasvaḥ" the penult is shortened [e → i]; by "caṅi" the doubling (dvitva) of the second, ac-initial "dhi" portion; by "abhyāse carca[ḥ]" jaśtva in the reduplicative syllable (abhyāsa) [→ didh]; by "na māṅyoge" the absence of the āṭ augment [mā (māṅ) being in construction]; and, because of construction with the word bhavat, the third person (prathama-puruṣa). (This will become clear in the tiṅanta section.) When "pra-ididhat" stands, "etyedhati" gives no vṛddhi, the edh-root lacking ejāditva [it now begins with i]; whereas in the absence of the qualification ejāditva vṛddhi would wrongly apply here too. Nor should it be objected that, the ekāra having become the ikāra by "ṇau caṅi", this is no longer the edh-root — because a thing altered in one part is not [thereby] a different thing (ekadeśa-vikṛtasya ananyatvāt). Here [Bhaṭṭoji] did not adduce "predidhat" without the particle mā (māṅ) as the counter-example: for then, the augment āṭ [applying] by "āḍ ajādīnām" and vṛddhi by "āṭaś ca", "aididhat" would result — and there, because of edhati's ejāditva (ejāditvāt), vṛddhi is exactly what is desired. But the word bhavat [in "mā bhavān predidhat"] has a purpose that is only to-be-considered (cintya-prayojana).

Objection: In ā + ihi, by "ād guṇaḥ" [we get] ehi; then "ava-ehi" stands. Now the single-substitute guṇa — the ekāra [of ehi] — by "antādivac ca" (A 6.1.85) behaving as the beginning of what follows (parādivadbhāva), and by "ādyantavad ekasmin" (A 1.1.21) behaving as a substitute standing for the whole (vyapadeśivadbhāva), [and the iṇ-root having ejāditva,] when that follows, [the rule] — blocking the pararūpa of "eṅi pararūpam" — would apply the vṛddhi of "etyedhatyūṭhsu", so that the desired "avehi" would not result [we should get avaihi]. Nor may it be supposed that [avehi] is secured by the pararūpa of "omāṅoś ca"; for the vṛddhi of "etyedhati", just as it is an exception to "eṅi pararūpam", would likewise be an exception to "omāṅ[oś ca]", by parity with the "yena nāprāpti" maxim.

Response: To this he [Bhaṭṭoji] says "purastāt" — the maxim being "earlier-stated exceptions block the immediately-following rules, not the [later] subsequent ones" (purastād apavādā anantarān vidhīn bādhante nottarān); i.e., exceptions read earlier block only the contiguous subsequent rules, not the ones separated [from them by intervening rules]. And in the present case, after "etyedhatyūṭhsu" some sūtras are read, then "eṅi pararūpam", then again several sūtras, and then "omāṅoś ca". So, by the stated maxim, the vṛddhi of "etyedhati" blocks only "eṅi pararūpam", not also the pararūpa of "omāṅoś ca". This is the purport.

In truth, however (vastutas tu): the vṛddhi of "etyedhati" is not an exception to "omāṅoś ca" at all — because in cases like upaiti the vṛddhi of "etyedhati" is undertaken (ārambha) even where that [omāṅoś-ca pararūpa] does not obtain. Hence this is no domain for the "purastād-apavāda" maxim. And so, in "ava-ehi", it is proper to say rather: blocking "etyedhati", the pararūpa of "omāṅoś ca" [applies] because it is the later rule (paratvāt). This too is clearly stated in the Bhāṣya: "Since 'eṅi pararūpam' does not otherwise obtain (nāprāpta), the vṛddhi of 'etyedhati' is undertaken [there] as its exception; but as for 'āṅi pararūpam' [omāṅoś ca], the vṛddhi is undertaken both where [that pararūpa] obtains and where it does not, and so is not its exception." As for the alternative view (pakṣāntara) stated in the Bhāṣya — "Or else, [by] 'earlier-stated exceptions block the immediately-following rules', the vṛddhi of etyedhati blocks only 'eṅi pararūpam', not also the pararūpa of 'omāṅoś ca'" — that rests merely on a concessive assumption (abhyupagamavāda), namely that the vṛddhi of etyedhati is an exception to both sūtras, "eṅi pararūpam" and "omāṅoś ca"; or else on disregarding the generality of what-is-to-be-blocked (bādhya-sāmānya-acintā). Enough of this at length (alaṃ bahunā).

akṣād ūhinyām upasaṅkhyānam — "āt", "aci", and "vṛddhiḥ" are carried forward by anuvṛtti, and "ekaḥ pūrvaparayoḥ" as well. [The root] ūha is "in the sense of conjecture" (ūha vitarke); the reckoning (ūhana) is ūha, and she who has it (sā asyā asti) is ūhinī [by the matvarthīya ini]. The meaning: after the word akṣa, when the word ūhinī follows, vṛddhi is the single substitute of the two preceding-and-following vowels. In akṣauhiṇī [the further affixes are] the matvarthīya ini, and — the stem ending in n (nānta) — ṅīp. The analysis is "the ūhinī of the akṣas" (akṣāṇām ūhinī); an army distinguished by a particular measure is an akṣauhiṇī. By "pūrvapadāt saṃjñāyām [aṇaḥ]" ṇatva [n → ṇ]. When "akṣa-ūhinī" stands, guṇa being [otherwise] due, by this vārttika there is vṛddhi.

svād īrer iṇoḥ — the meaning: after the word sva, when the words īra and īrin follow, vṛddhi is the single substitute of the two vowels. svaira: from īra "to move" (īra gatau), ghañ in the action-sense (bhāve), analysis "movement by one's own will" (svena chandena īraḥ); the compound by "kartṛkaraṇe kṛtā bahulam". When "sva-īra" stands, guṇa being due, by this vārttika vṛddhi. svena eritum — in the sense "whose nature is to move about (īritum, saṃcaritum) at his own will", the affix ṇini by "supy ajātau [ṇini]" [→ svairin], meaning "moving at his own will" (svacchandacārin), an upapada compound. When "sva-īrin" stands, guṇa being due, by this [vārttika] vṛddhi.

prād ūhoḍhoḍhy-eṣaiṣyeṣu — the meaning: after the upasarga pra, when these — ūha, ūḍha, [ūḍhi,] eṣa, eṣya — follow, vṛddhi is the ekādeśa of the two vowels. prauha — ūha "in the sense of conjecture", ghañ in the bhāva-sense, a gati-compound; when "pra-ūha" stands, by this vārttika vṛddhi. prauḍha — vah "to convey" (vaha prāpaṇe), kta; by "vaci svapi[yajādīnām]" saṃprasāraṇa; pūrvarūpa; by "ho ḍhaḥ" ḍhatva; by "jhaṣas tathor dho'dhaḥ" dhatva; ṣṭutva; by "ḍho ḍhe lopaḥ" [elision of the first ḍh] and by "ḍhralope [pūrvasya dīrgho'ṇaḥ]" lengthening [→ ūḍha]. When "pra-ūḍha" stands, guṇa being due, by this [vārttika] vṛddhi.

Objection: In proḍhavān too — from vah with the affix ktavatu, after ḍhatva, dhatva, ṣṭutva, ḍha-lopa, and lengthening, "pra-ūḍhavat" standing — vṛddhi by "prād ūḍha…" would apply, the word ūḍha being heard there too as following. Response: To this he cites "arthavat [-grahaṇe…]" — the meaning being that when a meaningful word can be grasped, a meaningless word is not grasped. This paribhāṣā is established from the mention of "rūpa" in "svaṃ rūpam [śabdasya…]", as is clear in the Bhāṣya on that very sūtra. "tena" — i.e., by the restriction that only a meaningful item is grasped. "ktāntam eva" — the affix kta is meaningful, being enjoined for bhāva and karman, whereas the affix ktavatu is enjoined for the agent (kartṛ); hence, the kta which is a portion of that [ktavatu] being meaningless, in "prād ūḍha…" the ūḍha is not grasped, so no vṛddhi occurs before it, but only guṇa [→ proḍhavān]. prauḍhi — from vah, ktin; ḍhatva, dhatva, ṣṭutva, ḍha-lopa, lengthening [→ ūḍhi]; when "pra-ūḍhi" stands, blocking guṇa, by this [vārttika] vṛddhi.

iṣa-icchāyām etc. — in all three [root-]classes these roots have a short penult (hrasvopadha): iṣ "in the sense of wishing", a tudādi; iṣ "in the sense of going", a divādi; iṣ "in the sense of repetition", a kryādi. "teṣām" is a genitive of specification (nirdhāraṇa). From any one among them, with ghañ and with ṇyat, laghūpadhaguṇa having been done, the forms eṣa and eṣya are established [iṣ → eṣ]. When "pra-eṣa" and "pra-eṣya" stand, blocking the pararūpa of "eṅi pararūpam", by this [vārttika] vṛddhi [→ praiṣa, praiṣya]. By this, the [alleged] redundancy (gatārthatva) of mentioning eṣa and eṣya — [the claim that they are] already covered by "vṛddhir eci" — is set aside. Objection: If so, how do the usages preṣa and preṣya arise [with guṇa, not vṛddhi]? Response: To this he says "yas tu" — the [roots] īṣ "in the sense of gleaning" (uñcha) and īṣ "in the senses of going, injuring, and seeing", having a long penult (dīrghopadha) so that laghūpadhaguṇa is absent, establish īṣa and īṣya; and since these two are not grasped in this vārttika [which lists only eṣa, eṣya], there being no vṛddhi before them, by "ād guṇaḥ" guṇa applies, and so the two forms preṣa and preṣya also exist. By association (sāhacarya) with the word eṣa, the word eṣya too is grasped only as a kṛt-ending nominal; therefore in a tiṅ-ending [finite verb] and in a lyap-ending [absolutive] there is no vṛddhi — [as in] agnaye preṣya dūtam ("having sent a messenger to Agni"), preṣya gataḥ.

ṛte ca tṛtīyāsamāse — the meaning: in a tṛtīyā-compound, after an a-ending [word], when the word ṛta follows, vṛddhi of the two vowels [occurs]. "sukhena ṛtaḥ" shows the analysis: ṛ "to go" (ṛ gatau); kta in the agent-sense (kartari) by "gatyarthākarmaka[…]"; the instrumental (tṛtīyā) because [sukha] is among "prakṛti" etc.; the compound by "sup[ā] sup[ā]". When "sukha-ṛta" stands, guṇa being due, by this [vārttika] vṛddhi, [namely] the ākāra; with raparatva the form is sukhārta. Since the counter-example for the mention "samāsa" is virtually shown already by the cited analytic phrase, he asks the purpose of the portion "tṛtīyā": "Why tṛtīyā?" paramarta — "that which is supreme and is ṛta", a karmadhāraya; [here] "ād guṇaḥ" [applies → paramarta].

pra-vatsatara-kambala-vasana-[ṛṇa]-daśānām ṛṇe — the genitive is used in the ablative sense (pañcamyarthe ṣaṣṭhī). The meaning: for pra, vatsatara, kambala, vasana, ṛṇa, and daśa, when the word ṛṇa follows, vṛddhi is the ekādeśa of the two vowels. prārṇam — in the analysis "pra-gata ṛṇa" [advanced debt], a gati-compound by "prādayo gatādyarthe prathamayā"; when "pra-ṛṇa" stands, blocking guṇa, by this vārttika vṛddhi, the ākāra; [then] raparatva [→ prārṇa]. vatsatara = a young calf (śiśuvatsa), analysis "the ṛṇa of it" (tasya ṛṇam); the ṛṇa contracted with reference to a vatsatara, or for the purpose of acquiring one, is vatsatarārṇa — and similarly for the following [members]. By the word "ādi" [in the analysis] there are grasped kambalārṇa, vasanārṇa, ṛṇārṇa, and daśārṇa, everywhere a genitive compound (ṣaṣṭhī-samāsa). How is the word ṛṇa construed when [another] word ṛṇa follows? To this he says "ṛṇasya…" [the other ṛṇa contracted to discharge a ṛṇa is ṛṇārṇa]. How does the word daśārṇa come to apply to a particular country and river? To this he says "ṛṇaśabda…" — the word ṛṇa [here] denotes inaccessible ground (durga-bhūmi) and water (jala): so the country in which there are ten ṛṇas [= inaccessible grounds] is the region daśārṇa, and the river in which there are ten-fold ṛṇas [= waters] is daśārṇā.

upasargād ṛti dhātau — upasargād ṛti. From ādguṇaḥ (A 6.1.87) the word āt, ending in the fifth (ablative), is carried forward. And that [āt] is a qualifier of upasarga; hence, by the tadantavidhi, "from [an upasarga] ending in a" (akārāntāt) is obtained. "ṛti" is a qualifier of dhātu; by [the paribhāṣā] "that in which an operation is prescribed [is taken as beginning with it]" (yasmin vidhis tadādividhiḥ), "beginning with ṛ" (ṛkārādau) is obtained. From vṛddhir eci (A 6.1.88) the word vṛddhi too is carried forward. ekaḥ pūrvaparayoḥ (A 6.1.84) is the adhikāra. He states this with "avarṇāntāt…" and the rest. "ekādeśaḥ" — "of the two vowels, preceding and following" (pūrvaparayor acoḥ) is to be supplied; otherwise there would be a substitution for the whole of the upasarga and the dhātu [rather than for the two vowels only]. "upārcchati, prārcchati" — when "upa ṛcchati" [and] "pra ṛcchati" stand, guṇa being [otherwise] applicable, this vṛddhi [prevails instead] on the ground of paratva (being the subsequent rule).

antādivacca — Now, in order to raise [as a prospect] the visarga that is prescribed at the pada-end for that r (repha), and thereby to establish its pada-final status, [Bhaṭṭoji] states antādivacca. The sūtra ekaḥ pūrvaparayoḥ (6.1.84) is carried forward (anuvṛtti). By the paribhāṣā of respective correspondence (yathāsaṃkhya), the expression antādivat is construed in order [anta with the preceding, ādi with the following]. Consequently it is obtained that the single substitute (eka ādeśa) which occurs in place of both the preceding and the following is like the final of the preceding and like the initial of the following.

Objection: Although the single substitute has two sthānins (originals) — the preceding and the following — and since each of those two sounds is single and indivisible (akhaṇḍa), with respect to them the statement that the ekādeśa is "like the final and the initial" is incoherent (asaṅgata) [for a single indivisible sound has no separable final or initial part]. — Response: Nevertheless, the intended sense is this: the single substitute occurring for the preceding and following sounds functions as the final (anta) of the preceding aggregate (samudāya) that is constituted by the first sthānin — the first sthānin becomes the bearer of that [anta's] operation; and [it functions] as the initial (ādi) of the following aggregate constituted by the second sthānin — the second sthānin becomes the bearer of that [ādi's] operation.

There, an example of being like the final of the preceding: kṣīrapeṇa. When kṣīrapa-ina stands, by ād guṇaḥ (6.1.87) the single substitute is e (ekāra). There, because [that e] is like the final of the preceding, the following segment "pe" is monosyllabic (ekāc), and hence by ekāj-uttarapade ṇaḥ (8.4.12) the ṇatva (the substitution ṇ for n) applies. Here, in the case of the ekādeśa, the sthānivat rule (sthānivad-ādeśo'nalvidhau, 1.1.56) does not operate: for the ṇatva, which operates by resorting to the monosyllabic-uttarapada status of "pe" — depending upon the a following the p (which is the sthānin of the substitute) — would have as its basis the single sound (al) of the sthānin [and sthānivadbhāva is barred in an al-vidhi]. Although the ṇatva rule primarily (prādhānyena) resorts to the monosyllabic-uttarapada status, and resorts to the single sound of the sthānin (the a after p) only as the vowel-ness (ac-ness) qualifying it, nonetheless the prohibition "an-alvidhau" does apply here, because what is intended in that prohibition is a resorting of any kind whatever (yathākathañcit) to the single sound of the sthānin. Otherwise — in pratidīvya, where the iṭ augment is to be made by resorting, in virtue of the val [initial] present in the sthānin, to the ktvā-substitute lyap as an ārdhadhātuka — the prohibition "an-alvidhau" would fail to obtain, since there the val is not resorted to as primary. By this [reasoning], the view that "this sūtra [antādivacca] is already accomplished (gatārtha) by the sthānivat rule" is refuted (nirasta).

An example of being like the initial of the following: khaṭvā. Here, from the word khaṭva, ṭāp (feminine ā) is added by ajādyataṣṭāp (4.1.4); the savarṇa-dīrgha (homogeneous long vowel, by akaḥ savarṇe dīrghaḥ 6.1.101) is the long ā. Because that [ā, the single substitute], by being like the initial of the following, [counts as] the ṭāp, the su (nominative singular) that follows it undergoes elision by halṅyāb…lopaḥ (6.1.68). This too cannot be accomplished by sthānivadbhāva, because [the fact] that the halṅyādi-lopa here has as its basis the single sound of the sthānin (sthāny-al-āśraya) will be stated by the author of the Mūla [the Kaumudī] later.

Objection: This being so, it should be said that in ayaje indram — [derived] from √yaj (yaj) in laṅ, first person singular ātmanepada, with the ending iṭ, with the vikaraṇa śap, with ād-guṇa, and with the aṭ (aḍ) augment — since the guṇa [e], being a single substitute, is, by being like the initial of the following, an i-kāra, before that i the savarṇa-dīrgha would occur [wrongly yielding ayajīndram]. — Response: For here [in antādivadbhāva] only the properties of an aggregate of sounds (al-samudāya-dharma) are extended (atidiśyante) — such as being a prātipadika, being a subanta, being a pratyaya, and so on — and not the properties of a mere single sound (varṇa-mātra-dharma), such as being an a, being short (hrasvatva), and the like. And it is said in the Bhāṣya (Mahābhāṣya): "na vā, atādrūpyātideśāt" ("Or rather not, since it is not an extension of [identical phonetic] form"). Because the properties residing in a mere single sound qua final/initial are not extended, in cases of the type "ayaja indram" the savarṇa-dīrgha and the like do not occur. Enough of this over-elaboration.

kharavasānayor visarjanīyaḥ — On [the Kaumudī phrase] "iti rephasya" [carried over from the preceding rule]: in upārcchati, since the r of the single substitute ār is pada-final by being like the final of the preceding [upa] (pūrva-antavattva), and since visarga is thereby obtained, [that phrase] construes with what follows (i.e., the present rule). To the question "how is visarga obtained?" he states kharavasānayoḥ. From ro ri (8.3.14) the genitive "raḥ" (of the r) is carried forward (anuvṛtti); and that is a qualifier of "padasya" (of a pada), which is under adhikāra. By the paribhāṣā yena vidhis tadantasya (1.1.72), it is obtained that [the rule applies] to a pada ending in r (rephānta). The meaning: when a khar or avasāna follows, the visarga is substituted for a pada ending in r; and by alo'ntyasya (1.1.52) it applies to the final sound [of the pada]. The "following-ness" (paratva) of avasāna — which is of the nature of an absence (abhāva-rūpa) — is only mentally constructed (buddhi-kṛta). He states the upshot: "khary avasāne ca." On "padānte": [the qualifier] "vidyamānasya" (of the [r] actually present) is to be supplied. But if the designation avasāna were [taken as] that of the final sound [itself], then the construal of the word avasāna here [would be] as stated under virāmo'vasānam (1.4.110).

On "iti visargaḥ": the meaning is that, since visarga is obtained in cases such as upārcchati by this [rule], the avoidance (parihāra) of that [visarga] is now being stated. On "antavat": the meaning is that the visarga does not occur for the r which has attained pada-final status by antavadbhāva (being-like-the-final). To "why?" he says "ubhayatharkṣu": the idea (bhāva) is that otherwise, there too, a visarga would have been indicated [by the tradition, but it is not].

Objection (nanu): here the word "dhātau" [in upasargād ṛti dhātau] is purposeless (vyartha), since [the sense of] "dhātau" is already established by the mention of "upasarga" alone, because the designation upasarga is enjoined only when there is connection with an action (kriyā-yoga) [and an upasarga thus presupposes a dhātu]. — Nor should it be argued: "In uparkāra [< upagata ṛkāra], since there is connection with an action and hence upasarga-hood, vṛddhi would [wrongly] be obtained, so the mention of the dhātu here serves to prevent that (tan-nivṛtty-artham)" — because, by the paribhāṣā "the prā-di particles are gati/upasarga only in relation to that action with which they are connected," [the upa here] is an upasarga only in relation to the implied action of "coming near" (gamana-kriyā), [not in relation to the ṛ-sound], while in the present rule it is precisely upasarga-hood conditioned by a following ṛ-sound (ṛkārādi-nimittaka) that is intended. Having raised this doubt, he replies with "upasargeṇaiva…"

The construal [of the Kaumudī]: "when the dhātu is [already] implied by the mention of upasarga alone" — this implication arising from presumption (arthāpatti) of the form "otherwise it would be unaccountable" (anyathānupapatti) — then, this being achieved, the [re-]statement of the word "dhātau" serves the purpose of re-enjoining [the vṛddhi] (punar-vidhāna). To "how is the re-enjoining obtained?" he says "yoga-vibhāgena." The word "yoga" is a synonym of "sūtra." The sūtra is split into "upasargād ṛti" and "dhātau." There, in the first [portion] upasargād ṛti, taking "dhātau" as supplied by implication (arthāt), the stated meaning is obtained; and in the second [portion] "dhātau" also, the prior "upasargād ṛti" is carried forward (anuvṛtti). Thus this [second] sūtra comes to have the same meaning as the prior one — hence the re-enjoining is obtained. To "what is the purpose of this re-enjoining?" he says "tena," i.e., "by the re-enjoining" [the optional prakṛti-bhāva of ṛty akaḥ is set aside, the vṛddhi now being obligatory].

vā supy āpiśaleḥ — "na bhavati" ["it does not occur"] — the meaning is that even the later (para) prakṛtibhāva is set aside by the force of the re-enjoining (punarvidhāna). vā supi. The previous sūtra "upasargād ṛti dhātau" [A 6.1.91] is carried forward (anuvṛtti); "āt" is carried forward from "ād guṇaḥ" [A 6.1.87], and "vṛddhiḥ" from "vṛddhir eci" [A 6.1.88]. By "āt" (a-final), "upasargāt" is qualified, hence the tadanta-vidhi [upasarga ending in a]. By the paribhāṣā on the mention of an affix (pratyayagrahaṇa-paribhāṣā), the word "sup" intends a root having a subanta as its base [a denominative]. And "ekaḥ pūrvaparayoḥ" [A 6.1.84] is in force as adhikāra. He states this with "avarṇāntāt…" etc. "sub-dhātau" means "when a root having a subanta base follows," since a root that is [itself] a subanta is impossible. "ekādeśaḥ" — "of the two vowels, prior and posterior (pūrvaparayor acoḥ)" is to be supplied. Although for [roots] such as "ṛccha gatau" both root-hood and subanta-hood are possible in the kvip [affix], nevertheless, because in the Bhāṣya on this sūtra "upārkārīyati" and the like are stated as its examples, the explanation "[a root] having a subanta base" alone is appropriate. Objection: since [optionality] is established just by taking "vā," the mention of Āpiśali is purposeless. To this he says "āpiśali-grahaṇam…": it serves to convey [things] such as that this view is approved even by the teacher Āpiśali. "prārṣabhīyati" — in the sense "he wishes an ṛṣabha (bull) for himself," root-hood [is obtained] by "supa ātmanaḥ kyac" [A 3.1.8] and "sanādyantāḥ [dhātavaḥ]" [A 3.1.32]; luk of the sup by "supo dhātuprātipadikayoḥ" [A 2.4.71]; ī-substitution by "kyaci ca" [A 7.4.33]; laṬ, tiP, śaP; pararūpa. When "pra ṛṣabhīyati" stands, by this [rule] vṛddhi = ā, with raparatva [the substitute for ṛ being followed by r]. "prarṣabhīyati" — on the alternative without vṛddhi, [there is] guṇa by "ād guṇaḥ," with raparatva. "sāvarṇyāt" — because of the homogeneity of the ṛ-varṇa and the ḷ-varṇa, the ḷ [-initial] is also included by "ṛti" [in the governing rule]. "upālkārīyati" — in the sense "he wishes an ḷ-sound for himself," the kyaC etc. as before; the distinction is the laparatva [the substitute being followed by l]. "upalkārīyati" — because of the absence of vṛddhi, here [there is] guṇa, with laparatva. "taparatvāt" — because, "ṛti" being a taparakaraṇa, only [an ṛ] of that same measure (duration) is grasped, this optional vṛddhi does not occur before a long ṝ; rather, guṇa alone. "ṛkārīyati" — in the sense "he wishes an ṛ-sound for himself," the kyaC etc. as before; because of the absence of vṛddhi, here [there is] guṇa only.

eṅi pararūpam — eṅi pararūpam. "upasargāt" and "dhātau" are carried forward (anuvṛtti). "āt" [a-varṇa], being carried forward, is a qualifier of "upasargāt"; hence, by the tadantavidhi, [the rule concerns an upasarga] ending in a-varṇa. By the paribhāṣā yasmin vidhis [tadādāv algrahaṇe] [the operation is enjoined on what begins with the stated sound], there is the tadādigrahaṇa [so "eṅi" means "before a dhātu beginning with eṅ"]. And ekaḥ pūrvaparayoḥ (A 6.1.84) [one single substitute in place of the preceding and following] is in effect (adhikṛta). He states this with "avarṇāntāt" [after (an upasarga) ending in a-varṇa], etc. In [the phrase] "ekādeśaḥ" supply "pūrvaparayoḥ" [in place of the preceding and following].

prejate — from ejṛ "to shine" (dīpti). When "pra-ejate" is formed, overriding the vṛddhi [that would come] by vṛddhir eci (A 6.1.88), by this [rule] the pararūpa e [is the single substitute → prejate]. (The root ejṛ "to tremble" (kampana), on the other hand, is parasmaipada.) upoṣati — from uṣa "to burn" (dāha); laṭ, tip, śap; guṇa of the light penult [uṣ → oṣ]. When "upa-oṣati" is formed, overriding vṛddhi, by this [rule] the o [is the pararūpa single substitute → upoṣati].

Objection: eḍaka is a ram (meṣa); "he wishes it for himself" → eḍakīyati [denominative]. When "upa-eḍakīyati" is formed, by this [rule] pararūpa would apply obligatorily (nitya); yet vṛddhi too is desired — [giving] upaiḍakīyati. On this he says "iha" [here...]. In eṅi pararūpam, "vā supi" ["optionally, before a nominal"] is carried forward; and it is not to be construed as forming a single sentence [with eṅi pararūpam]. For if it were [one sentence], the meaning would only be: "after an upasarga ending in a-varṇa, when a sub-dhātu beginning with eṅ follows, pararūpa is optionally the single substitute"; and in that case, in prejate and upoṣati, pararūpa would not be obligatory. Therefore it must be explained by splitting the sentence (vākyabheda). "eṅi pararūpam" is the first sentence, meaning: after an upasarga ending in a-varṇa, when an eṅ-initial dhātu follows, pararūpa is the single substitute [obligatorily]. "vā supi" is the second sentence, into which "eṅi pararūpam" recurs, as do "dhātau", "upasargāt", etc.; and thence it is obtained that, in the domain of the aforesaid pararūpa, when a sub-dhātu follows, pararūpa optionally applies. He states this with "tena" [thereby]: by resorting to sentence-splitting in the manner stated, pararūpa is optional (pākṣika) before an eṅ-initial sub-dhātu, but obligatory (nitya) before any other dhātu.

eve cāniyoge. He explains the word niyoga: niyoga is avadhāraṇa [restrictive emphasis]. avadhāraṇa is the exclusion of connection with anything else (anyayogavyavaccheda). The meaning is: when the word eva is in a sense other than that [restriction], and follows an a-vowel, pararūpa is the single substitute in place of the preceding and following. kveva bhokṣyase — meaning "we are not certain [about] where [you will eat]." When "kva-eva" is formed, overriding vṛddhi, by this vārttika the pararūpa e [→ kveva]. "anavakḷptau" means "in the sense of non-restriction (anavadhāraṇa)." tavaiva — meaning "of you only, not of another"; here, since the word eva has restrictive sense, pararūpa does not apply before it [→ tavaiva, by vṛddhi].

Now, in order to prescribe pararūpa of the ṭi in certain cases, he states the ṭi designation: aco 'ntyādi ṭi (A 1.1.64). "acaḥ" is a genitive of selection (nirdhāraṇa); the singular is generic (jāti). "antya" is that which is at the end; "antyādi" is that whose beginning is the last one — such is the analysis. He states the upshot: "among the vowels" [the last vowel together with what follows it is called ṭi]. "śakandhvādiṣu" — in the domain of the śakandhu-class, in order to derive those [forms], pararūpa of the preceding and following is to be prescribed. And thence it follows by implication: for the words śaka etc., when an ac follows the ṭi, pararūpa is the single substitute in place of the ṭi and the following ac. For, were "āt" [the a-varṇa restriction] carried forward [into this pararūpa], several forms in the śakandhvādi group such as sīmanta would not be derivable (asiddhi). He says this with "and that, of the ṭi..."

He recites the śakandhvādi group with "śakandhu", etc. Analysis: śaka = certain regions; andhu = a well (kūpa). When "karka-andhu" is formed, pararūpa [→ karkandhu]. kulaṭā — from aṭa "to go" (gati); the suffix ac by the pacādi [class]; ṭāp [feminine ā]; analysis: "the roamer (aṭ) among families (kula)"; she who roams to [others'] houses for alms or for adultery is a kulaṭā. But if analyzed as "she who roams the houses (kula, accusative)" — with aṇ in the sense of object (karman) — then by ṭiḍḍhāṇañ... (A 4.1.15) the feminine would take ṅīp [giving a different form]. When "kula-aṭa" is formed, pararūpa [→ kulaṭā]. "sīmantaḥ keśaveśe" — this is a vārttika included within the group; keśa-veśa = a particular arrangement (parting) of the hair; when that [sense] is intended, pararūpa [applies] in place of the ṭi of the word sīman and the a-vowel in the word anta [→ sīmanta]; but were "āt" carried forward, this would not be derivable. "sīmānta" [with long ā] is different — meaning the region at a village boundary (grāmānta-pradeśa). manīṣā — analysis manas + īṣā; here pararūpa ī in place of the ṭi [of manas] and the ī [of īṣā → manīṣā]; from īṣ "to go" (gati), the suffix a by guroś ca halaḥ (A 3.3.103), and ṭāp; "manīṣā means intellect (buddhi)," says Amara. halīṣā — "īṣā is the pole of a plough," says Amara; analysis: the īṣā of the plough (hala); the [apparent] redundancy is to be resolved as in the case of "karikalabha." Here, when guṇa presents itself, pararūpa [applies → halīṣā]; likewise in lāṅgalīṣā. patañjali — analysis: "he before whom, on account of his being worthy of salutation, the añjali (folded hands) is falling (patat)"; here pararūpa a in place of the ṭi [of patat] and the a [of añjali → patañjali]. But some analyze it as "añjaleḥ patan" [falling from the añjali], on the strength of a tradition (aitihya) that [Patañjali] emerged from the añjali of a certain sage at the time of the sandhyā-worship in the region called Gonarda; the compound is [then] formed by inclusion in the mayūravyaṃsakādi class.

"sāraṅgaḥ paśupakṣiṇoḥ" — this too is a vārttika within the group; but it is not found in the Bhāṣya. The word sāra [means] "the excellent/best part"; Amara: "sāra [is used] for strength, for the firm part, for the proper, [as] neuter for cloth and sky, [and] in three genders." Analysis: "whose limbs (aṅga) are excellent (sāra)." When "sāra-aṅga" is formed, [with] savarṇadīrgha [ā] presenting itself, by this [rule] pararūpa [→ sāraṅga]. Amara: "sāraṅga [means] the cātaka bird and the deer in the masculine, [and] dappled in three genders." "This is an ākṛtigaṇa" — meaning, this group is to be determined by [family] type (ākṛti), i.e., by [words] of this kind; that is, words of this kind are to be included in this group according to usage in the world (loka-prayoga). He states the fruit of that [open-endedness]: mārtaṇḍa. "He whose egg (aṇḍa) was dead" is Mṛtaṇḍa, a certain sage; when "mṛta-aṇḍa" is formed, overriding savarṇadīrgha, by this [rule] pararūpa [→ mṛtaṇḍa]; then, with aṇ in the sense of "descendant," mārtaṇḍa. But the savarṇadīrgha in "parāmārtāṇḍam āsthat" is Vedic (chāndasa). "otvoṣṭhayoḥ" [a group-vārttika]: after an a-varṇa, when the word otu or the word oṣṭha follows, pararūpa is to be stated in place of the two vowels of the preceding and following. sthūlotu — Amara: "otu means cat (biḍāla, mārjāra)"; when "sthūla-otu" is formed, overriding vṛddhi, pararūpa [applies] optionally (pākṣika). bimboṣṭha — analysis: "he whose two lips (oṣṭha) are red like the bimba fruit."

omāṅośca — The analysis (vigraha) is "om ca āṅ ca" ("and om and āṅ"). "āt" [from ādguṇaḥ], "pararūpam" [from eṅi pararūpam], and "ekaḥ pūrvaparayoḥ" are carried forward (anuvṛtti). He states this with "omi..." and so on.

śivāyoṃ namaḥ: here, blocking the vṛddhi [that would come by vṛddhir eci, since a + o would give au], para-rūpa [applies, giving o] → śivāya + om = śivāyoṃ.

śivehi: when ā(ṅ)-ihi stands, by guṇa the form is ehi [āṅ (ā) + ihi → e → ehi]. Then, when śiva-ehi stands, blocking vṛddhi, para-rūpa, viz. the e-kāra [is the single substitute] → śivehi.

Objection: it should not be urged that "when śiva-a(ṅ)-ihi stands, once savarṇa-dīrgha is done [a + ā → ā, giving śivā] and thereafter guṇa [śivā + ihi → śivehi], the form śivehi is established, so the mention of āṅ [in the sūtra] is pointless." [Response:] Because, by the maxim "an operation [involving] a dhātu and an upasarga is antaraṅga," guṇa applies first [between āṅ and ihi], by "prathamād guṇaḥ," yielding, when guṇa is done, śiva-ehi; and then, vṛddhi [a + e → ai] presenting itself, [the mention of āṅ] serves to cancel that [vṛddhi]. But in the Bhāṣya it is settled that the mention of "āt" serves to cancel savarṇa-dīrgha, in [the derivation] ā ṛśyāt, guṇaḥ, aśryāt, adya-aśryāt → adyaśryāt.

avyaktānukaraṇasyāta itau — avyaktānukaraṇasyāta itau. "pararūpa" is carried forward (anuvṛtti) [from *eṅi pararūpam*, A 6.1.94]. "avyakta" = a sound produced by the falling of a tree and the like, in which the division into distinct phonemes is indistinct (asphuṭa-varṇa-vibhāga). Its "anukaraṇa" (imitation) = a word resembling it, which represents it and likewise has an unclear phonemic division. The a-sound that is a member of it — before the word iti, in place of that a-sound and of the i [of iti], pararūpa comes: this is the meaning.

Here it must be explained as "when the first i in the word iti follows"; otherwise the whole word iti would be liable to become the substitute. He states this with "of a sound" (dhvaneḥ) etc. Here *alo 'ntyasya* (A 1.1.52) does not apply, because of the prohibition *nānarthake 'lo 'ntyavidhiḥ* [there is no operation confined to the final sound in the case of a meaningless element — the a here being a meaningless imitative sound].

"paṭat-iti" is the display of the procedural string (the pre-sandhi analytic form); one must not mistake it for the form obtaining where pararūpa fails to apply, since here pararūpa is obligatory (nitya). Or it is a statement in the non-saṃhitā (un-combined) condition. "paṭiti" is the [finished] example; "the tree falls" (vṛkṣaḥ patati) and the like is to be supplied [as the sense].

He captures, as to its purport, the vārttika *anekāj-grahaṇaṃ kartavyam* ["the mention 'having more than one vowel' should be made"] by [the words] "ekāco na." "ekāc" = that which has a single ac (vowel); the aforesaid pararūpa does not apply to an anukaraṇa that is such [monosyllabic] — this is the meaning. "śrad iti": here, because it is monovocalic (ekāc), there is no pararūpa.

Objection: in "paṭatpaṭat-iti," the desired form is paṭatpaṭeti; but there, by pararūpa via *avyaktānukaraṇasya*, only "paṭatpaṭiti" would result.

tasya param āmreḍitam — And the experts say: "Sometimes application, sometimes non-application, sometimes an option, and sometimes something else entirely; where 'bahula' [is enjoined], this is to be ascertained in the world by following the usage of the learned (śiṣṭa)." Now, what is called āmreḍita? To that he says "tasya param āmreḍitam." This sūtra is read immediately after "sarvasya dve" [A 8.1.1]; and so, by "tasya" ("of that"), "of the doubled [form]" is obtained [by anuvṛtti]. In construction with the word "para" denoting a part (avayava), the genitive [in "tasya"] is by an indicator (jñāpaka) from this very [context]. He states this with "dvir-uktasya…" etc. "paṭatpaṭeti" — when "paṭat paṭat iti" stands, the pararūpa [single substitution in the form of the latter], an i-kāra, [replaces] the ta-kāra and the i-kāra; and then [there is] "ād guṇaḥ."

jhalāṃ jaśo'nte — Now, on the alternative in which pararūpa does not apply, in paṭatpaṭat [+ iti] this [sūtra] prescribes d for the t: jhalāṃ jaśo'nte. [The word] padasya is in force by adhikāra; hence he says [in the Kaumudī] "at the end of a pada." On [the example] paṭatpaṭad: the idea is that, on account of sameness of place [of articulation], the jaś-substitute for t is d. It should be borne in mind here that, in the absence of any prohibition of pararūpa for the ac (vowel), under the rule that optionally prescribes pararūpa as the single substitute for the t alone, on the alternative in which that [t] does not take pararūpa, since the ac would then obtain pararūpa by the preceding sūtra, the form paṭatpaṭiti would result.

akaḥ savarṇe dīrghaḥ — akaḥ savarṇe. "akaḥ" is ablative (fifth case). From iko yaṇ aci (A 6.1.77), "aci" recurs (anuvṛtti); ekaḥ pūrvaparayoḥ (A 6.1.84) is in effect (adhikṛta). He states this with "akaḥ savarṇe 'ci", etc. daityāriḥ — when "daitya-ari" is formed, in place of the two a's, one long ā, because of sameness of place [of articulation]. When "śrī-īśa" is formed, in place of the two ī's, one ī. When "viṣṇu-udaya" is formed, [in place] of the two u's, [one] ū.

He asks the purpose of the anuvṛtti of "aci": "aci kim." kumārī śete — were "aci" not carried forward, the ī and the ś, being homogeneous through sameness of the palatal place and the open (vivṛta) effort, would [wrongly] yield a single substitute, the long ī; to exclude that, "aci" must be carried forward.

Objection: even though the ī and the ś have sameness of place and effort, there is no homogeneity, because of the prohibition nājjhalau (A 1.1.10); hence "akaḥ savarṇe" does not apply in "kumārī śete" [anyway], and the anuvṛtti of "aci" to exclude that is pointless. Response (nājjhalau...): the prohibition of homogeneity by nājjhalau concerns only the sounds of the varṇasamāmnāya [the phoneme-inventory], not the long and pluta vowels as well — because from ādir antyena sahetā (A 1.1.71) it is understood that only the varṇasamāmnāya-sounds are denoted by the word "ac." Hence, since the ī and the ś do possess homogeneity [at this level], there would be overapplication in "kumārī śete"; therefore the anuvṛtti of "aci" is necessary.

Objection: even granting that only the varṇasamāmnāya-sounds are denoted by "ac," still the short, long, and pluta vowels alike are grasped by the a etc. presented by "ac," on the strength of the aṇudit sūtra [aṇudit savarṇasya cāpratyayaḥ, A 1.1.69]; hence the ī and the ś do not [after all] incur homogeneity, so the anuvṛtti of "ac" is pointless. Response (grahaṇaka...): the grahaṇaka-sūtra aṇudit savarṇasya... (A 1.1.69), only once it has itself come into existence (labdhātmaka), is fit to operate in [rules] such as asya cvau (A 7.4.32); and at the stage when nājjhalau operates, this grahaṇaka-śāstra has not [yet] come into existence — for, being composed with the word "savarṇa," it comes into existence only after the meaning of the word "savarṇa" is known. And the sūtra tulyāsya[prayatnaṃ savarṇam] (A 1.1.9), which enjoins the savarṇa designation, though conveying its sense generally, attains capacity for its own effect only after excluding the domain of its exception nājjhalau and coming to rest elsewhere. And it has been said: "the utsarga (general rule) settles in after the domain of the apavāda (exception) has been marked off."

And thus, since the aṇudit sūtra has not come into existence prior to the operation of the tulyāsya-sūtra together with the prohibition nājjhalau — [and] therefore in nājjhalau the mention of "ac" does not grasp the long and pluta vowels, [and] therefore there is no prohibition of homogeneity for the ī and the ś, [and] therefore homogeneity obtains — "akaḥ savarṇe" would apply in "kumārī śete"; to exclude that, the anuvṛtti of "aci" must be resorted to. This has [already] been elaborated on the occasion of commenting on the passage beginning "even though the prohibition of homogeneity by nājjhalau..." "ako 'ki dīrghaḥ would be the better formulation" — and thus the word "savarṇa" need not be included, which is a simplification (lāghava); while in dadhyukāra, because of resorting to yathāsaṃkhya (respective correspondence), there is no overapplication. And so [under this formulation] even the anuvṛtti of "aci" need not be resorted to.

The vārttika "ṛti ṛ vā" he recites, construing it with the word "savarṇa" carried over from "akaḥ savarṇe": ṛti savarṇe ṛ vā. "akaḥ" recurs, and "ekaḥ pūrvaparayoḥ." Meaning: in place of an ak [here, ṛ] and a following homogeneous ṛ, the single substitute is optionally "ṛ." hotṛkāra — when "hotṛ-ṛkāra" is formed, by this [vārttika], in place of the two ṛ's there arises a certain sound: an ṛ distinct from [the ordinary] ṛ, of dual inner nature like Nṛsiṃha [the man-lion], possessing two ra-sounds (repha). In the alternative where this [special ṛ] does not [arise], he shows the form: hotṝkāra, [with] the long [ṝ] by akaḥ savarṇe [dīrghaḥ].

ḷti savarṇe ḷ vā — in place of an ak and a following homogeneous ḷ, the single substitute is optionally "ḷ." hotḷkāra — when "hotṛ-ḷkāra" is formed, in place of the ṛ and the ḷ there arises a certain sound: a ḷ of dual inner nature like Nṛsiṃha, possessing two la-sounds. "In the alternative": where the said dual-natured sound does not [arise], the dīrgha by "akaḥ savarṇe" in place of the ṛ and the ḷ becomes just the ṛ-sound [ṝ], because ṛ and ḷ are homogeneous [ṛ-ḷvarṇayoḥ...], and because ḷ has no long form. For this very reason, in "hotṛ-ḷkāra," in the savarṇadīrgha alternative, the Bhāṣya cites just "hotṝkāra."

Now, in "ṛti ṛ vā" and "ḷti ḷ vā," he analyzes the nature of the sound to be enjoined: "in both cases the sound to be enjoined is a two-fold sound of two mātrās." He substantiates this with "in the first, in the middle..." — "one mātrā": since consonants have half a mātrā each, each single repha has half a mātrā [so the two rephas together have one mātrā]. "on both sides of the ac-portion": after "abhitaḥ" supply "the two rephas"; "ajbhakteḥ" [of the ac-portion] uses the singular with generic intent; meaning, the other mātrā belongs to the two short-ṛ portions present before and after (above and below) the two rephas. "of the second": supply "to be enjoined"; "the rest as before" means the two la-sounds have one mātrā, and the two ḷ-portions present on both sides of them have the other mātrā. This is clear in the Bhāṣya and Kaiyaṭa on the sūtra tulyāsya[prayatnaṃ savarṇam] (A 1.1.9). By this, the older text [view] that "when dīrgha presents itself, a short ṛ and ḷ are here [instead] enjoined" is refuted. "pākṣika" means optional (vaikalpika). "prakṛtibhāva" means remaining in one's unmodified form — that is, non-application of sandhi.

eṅaḥ padāntādati — From [the sūtra] ami pūrvaḥ (A 6.1.107) the word pūrva is carried forward (anuvṛtti); and [the heading] ekaḥ pūrvaparayoḥ (A 6.1.84) is in force by adhikāra. Hence he states [the rule in the Kaumudī] beginning with "padāntāt...". In hare-ava, once this is established, [the rule] overrides the ay-substitute (ayādeśa) and the pūrva-rūpa, namely e, [is the single substitute] → hare 'va. In viṣṇo-ava it overrides the av-substitute (avādeśa) and the pūrva-rūpa, namely o [→ viṣṇo 'va].

sarvatra vibhāṣā goḥ — sarvatra vibhāṣā. "padāntāt" is carried forward (anuvṛtti). From *prakṛtyāntaḥ pādam* [A 6.1.115] "prakṛtyā" is carried forward. "prakṛtyā" = it stands in its natural, unmodified form (nirvikāra-svarūpa) — this is the meaning. From *yajuṣy uraḥ* [A 6.1.117] the [word] "yajuṣi" has ceased (nivṛtta); to indicate that, "sarvatra" is employed, and by it "in secular usage and in the Veda [both]" is obtained. He states this with "in the world" (loke) etc.

"prakṛti-bhāva" means standing in [its] natural state; and so neither pūrvarūpa nor the ava-substitution applies. "go-agram" is the form with prakṛti-bhāva; with pūrvarūpa, "go 'gram." Here "eṅ" too is carried forward [from *eṅaḥ padāntād ati*, A 6.1.109]. And so, by the maxim *ekadeśa-vikṛtam ananyavad bhavati* (a thing altered in one part is treated as not other [than itself]), there is no over-application in "citragv-agram." Also in "he citrago 'gram" [O Citrago!] there is no prakṛti-bhāva, because only the eṅ that is explicitly mentioned (pratipadokta) [the o of go proper] is taken, whereas in the present case the o [of the vocative citrago] is rule-derived (lākṣaṇika), by *hrasvasya guṇaḥ* [A 7.3.108].

"goḥ": when *go asi* is present, since the o of "go" is not at pada-end [the s / visarga following within the same pada], there is no prakṛti-bhāva. Nor should it be objected: "then, since it is not pada-final, pūrvarūpa too — by *eṅaḥ padāntād ati* [A 6.1.109] — would be hard to obtain"; for precisely on that account pūrvarūpa is [separately] undertaken there by *ṅasiṅasoś ca* [A 6.1.110].

avaṅ sphoṭāyanasya — On "the [word] ati has ceased": one must supply "from [the sūtra] eṅaḥ padāntāt"; and it ceases on the strength of interpretation (vyākhyāna) — that is the idea. The words padāntāt [from eṅaḥ padāntāt], goḥ, and aci are carried forward (anuvṛtti). In the view of the ṛṣi Sphoṭāyana [there is] avaṅ; but not for another [grammarian]. And thereby the option (vikalpa) is established [since naming a particular ṛṣi's opinion implies optionality]. This he states with "when an ac follows," etc. Because [avaṅ] is marked with ṅ and c, it is a substitute for the final [sound] [by ṅic ca, A 1.1.53 — substitution for the last only].

On gavāgram: when "go-agram" stands, avaṅ [replaces] the o after the g [→ gav-agram]. When "gav-agram" stands, savarṇadīrgha [a + a → ā, giving gavāgram]. Objection: one should not say — avaṅ, having operated by depending on the following a-vowel (ac) in the word agra, how can it bring about the savarṇadīrgha which destroys that very [conditioning a], since this contradicts the sannipāta-paribhāṣā [that an operation conditioned by an element does not then destroy that element]? Response: because the non-universality (anityatva) of the sannipāta-paribhāṣā will be stated later, at rāmāya.

On gavi: when "go-i" stands, since the o is not at the end of a pada, there is no avaṅ; nor [is there] prakṛtibhāva or pararūpa by the two preceding sūtras, but rather the substitute ava [by eco 'yavāyāvaḥ]. But were "ati" carried forward [as a condition, i.e., "before short a"], gaveśa and the like would not be derived.

On vyavasthita [-vibhāṣā]: an option (vibhāṣā) that operates by a regulation according to observed usage (lakṣya) — such that in some cases only the "it occurs" portion operates, in some cases only the "it does not occur" portion, and in some cases both — is a "regulated option" (vyavasthita-vibhāṣā); it is resorted to here in [interpreting] the option regarding go in all cases. And so, in gavākṣa, avaṅ [applies] obligatorily (nitya) — this is the meaning. And this is made plain in the Bhāṣya on the sūtra śāc choḥ [in the verse]: "devatrāto galo grāha ity yoge ca sadvidhiḥ / mithas te na vibhāṣyante 'gavākṣaḥ' saṃśitavrataḥ //" ["... they are not mutually optional — gavākṣaḥ, saṃśitavrataḥ"]. On gavākṣa: the analysis (vigraha) is "like the eye (akṣi) of the rays (gavām)"; by "akṣṇo 'darśanāt" the affix ac [is added]. The masculine gender [is] from usage (loka). "gavākṣa means a window (vātāyana)" — thus the Amara[-kośa].

indre ca [obligatorily (nityam)] — goḥ [for go], avaṅ, and aci are carried forward (anuvṛtti). Hence he says "for go (goḥ)." [This rule serves] to remove the optionality (vikalpanivṛttyartha) [of the preceding avaṅ sphoṭāyanasya, A 6.1.123]. As for gavendraḥ: when "go-indra" stands, avaṅ [is substituted for the final of go, yielding gav]; [then] ād guṇaḥ [a + i → e]. → gavendra. [The source then appends the stray colophon-marker samāptam ("finished") and the header atha adādiprakaraṇam ("Now the adādi section").]