Chapter 44 gave the three general rules for the two sets of endings. SK 2157 gives the ātmanepada to a root marked with an anudātta vowel or with ṅ. SK 2158 gives it to a root marked with a svarita vowel or with ñ, when the fruit of the action goes to the agent. SK 2159 gives the parasmaipada to every other root. In the Aṣṭādhyāyī a long run of special rules stands between SK 2157 and SK 2159 (A 1.3.13–1.3.77). This chapter teaches that run, with eight rules from elsewhere slipped in where they are needed. Most of the rules give the ātmanepada to a root that would otherwise take the parasmaipada. They do so only with a certain upasarga, in a certain sense, or with a certain kind of object. Some serve a root marked with ñ or a svarita vowel: they give it the ātmanepada even when the fruit does not go to the agent. A few are restrictions (niyama) on an ātmanepada already due. After this chapter you will be able to say why one says niviśate "he enters" but praviśati, and santiṣṭhate but tiṣṭhati. You will see why muktāv uttiṣṭhate "he strives for liberation" differs from pīṭhād uttiṣṭhati "he rises from the seat". You will know when saṃgacchate is right and when grāmaṃ saṃgacchati, and why utkurute, śāstre nayate and kramate / krāmati take the endings they do. You will build the aorists (luṅ forms) in which sic is kit: samasthita, āhata, udāyata, samagata, upāyata. Later in the chapter come apajānīte, vadate, avagirate, upayacchate, jijñāsate, prayuṅkte, odanaṃ bhuṅkte (but mahīṃ bhunakti), darśayate bhavaḥ, nyāyam apavadate, and the option svaṃ yajñaṃ yajati / yajate. The chapter has sixty-six sūtras, SK 2679 to SK 2744. The printed Kaumudī opens it by recalling SK 2157, with its examples āste "he sits" and śete "he lies". It closes with the words ity ātmanepadaprakriyā, "here ends the working of the ātmanepada".
dṛś → darś). You know SK 2191 and SK 2249 (loss of s), and SK 2250 (h for s before e). You know SK 2360 (tiṣṭha for sthā) and SK 2322 (kram lengthened before a śit affix in the parasmaipada). You know SK 2382 (aṅ in luṅ after sṛ, śās and ṛ) and SK 2390 (i for the final ṝ of a root, heard as ir: gṝ → gir).gacchati, yacchati, and in this chapter saṃgacchate, āyacchate, upayacchate.bhū → bhāvayati "he impels one who is becoming".vyati + lū + laṭ (agent; exchange of action, 2680) → vyatilunīte; against vyati + gam (a going-root, 2681) → vyatigacchantiThe chapter opens with two rules that do not depend on the root at all. SK 2679 gives the ātmanepada whenever the lakāra denotes the bare action or the object. SK 2680 gives it for the agent when an exchange of action is meant: one person does what befits another. The running example is lū "to reap" with vyati: vyatilunīte "he does the reaping that befits another". SK 2681 and SK 2682 then take some roots and some sentences out of SK 2680: vyatigacchanti keeps the parasmaipada.
Rule. When a lakāra stands for bhāva (the action itself) or for karman (the object), the ātmanepada endings replace it. The word "ātmanepada" is carried forward from SK 2157. One of Bhaṭṭoji's commentators says it runs on from there all the way to SK 2159. So it is understood in every rule of this chapter.
The same commentator adds that bhāva, bhāvanā and kriyā all mean "action". Karman here is the object kāraka. The other commentator's gloss is short: "ātmanepada for the lakāra that is in the sense of bhāva or karman".
Example.
bhū "to be" takes no object; by SK 2152 its lakāra may denote bhāva.
bhū + liṭ (bhāva)
→ this rule: an ātmanepada ending
→ babhūve "there was being".
anu + bhū "to experience" takes an object; its lakāra may denote the object.
anu + bhū + liṭ (karman)
→ this rule: an ātmanepada ending
→ anubabhūve "[bliss] was experienced".
One of Bhaṭṭoji's commentators supplies the object ānandaḥ "bliss".
Watch out. bhū is a parasmaipada root: babhūva "he was" (chapter 44). The ātmanepada of babhūve does not come from the root. It comes because the lakāra denotes the action, not the agent.
Rule. The ātmanepada comes for the agent when karmavyatihāra, exchange of action, is to be signalled. Bhaṭṭoji glosses it as kriyāvinimaya: one person does the action that befits another. One of his commentators explains the compound: karman is "action" and vyatihāra is "exchange". Both commentators add a wider sense: doing something to each other also counts, as in saṃpraharante rājānaḥ (SK 2681). One of them gives this as Kaiyaṭa's view.
Example.
vyati + lū + laṭ, the agent, exchange of action
→ this rule: ātmanepada
→ vyatilunīte.
Bhaṭṭoji: "he does the reaping that befits another". Both commentators give the picture: a brāhmaṇa does the reaping of grain that befits a śūdra.
Bhaṭṭoji then runs as "to be" through this rule:
vyati + as + te → the a of as is deleted (SK 2469) → vyati + s + te → vyatiste. There is no ṣ: SK 2472 needs y or a vowel after the s. Both commentators note this.vyati + s + āte → a vowel follows, so SK 2472 gives ṣ → vyatiṣāte; plural vyatiṣate.vyati + s + se → the s of as is lost before s (SK 2191) → vyatise. Both commentators add that SK 2472 has nothing to work on here: the s of as is gone.vyati + s + dhve → the s is lost before dh (SK 2249) → vyatidhve.vyati + s + e → h replaces the s before e (SK 2250) → vyatihe.vyatyasai, laṅ vyatyāsta, liṅ vyatiṣīta.Bhaṭṭoji lists vyatirāte three times, then vyatibhāte three times, then vyatibabhe. One commentator explains: vyatirāte is the laṭ form for the singular, the dual and the plural alike. vyatibhāte is the form of bhā, and vyatibabhe is its liṭ.
Applies when / blocked by. Why does the sūtra say kartari "for the agent"? One commentator first suggests it stops the word "bhāva or karman" from coming in from SK 2679. But a fresh sūtra already does that. So both commentators say kartari is there for the rules that follow; it runs on into them. One of them says the Bhāṣya makes this clear.
Rule. The ātmanepada of SK 2680 does not come after a root meaning going or injuring. The word artha "meaning" matters, say both commentators. Only roots that mean going or injuring on their own are meant. A root that gets such a meaning only from an upasarga is not.
Example.
vyati + gam, exchange of action
→ gam means going: SK 2680 is blocked
→ parasmaipada → vyatigacchanti.
Likewise han "to injure": vyatighnanti.
Applies when / blocked by. Two vārttikas adjust the rule.
The vārttika pratiṣedhe hasādīnām upasaṃkhyānam adds has "to laugh" and the like to the prohibition. Bhaṭṭoji explains these as verbs of sound of the kind of has: vyatihasanti, vyatijalpanti. One of his commentators notes there is no listed group "has and the rest"; that is why Bhaṭṭoji says "of the kind of has".
The vārttika harater apratiṣedhaḥ frees hṛ from the prohibition: saṃpraharante rājānaḥ "the kings strike at one another". One commentator says hṛ means striking only through an upasarga, so this vārttika already follows from the word artha. The other reports the same as what "they say".
Rule. The ātmanepada of SK 2680 also does not come when the root has itaretara or anyonya, "one another", as an upapada, a word standing with it in the sentence. One of Bhaṭṭoji's commentators supplies "no ātmanepada" to complete the sūtra. Bhaṭṭoji gives no gloss of his own; he adds a vārttika.
Example. The vārttika parasparopapadāc ceti vaktavyam adds paraspara, also "one another".
itaretarasya / anyonyasya / parasparasya vyati + lū
→ the exchange is already in the word "one another": no ātmanepada
→ vyatilunanti "they reap one another's [crop]".
Applies when / blocked by. If "one another" already shows the exchange, why prohibit an ātmanepada that would only show it again? One of Bhaṭṭoji's commentators answers. In ordinary speech people might add it anyway, careless of economy. So the prohibition is stated. For the same reason, in vyatilunīte the upasarga vyati and the ātmanepada both signal the exchange together.
ni + viś + te → niviśate; apa + kṝ → (suṭ, 2688) apaskir-a-te → apaskirate (vṛṣo hṛṣṭaḥ)From here on the rules name a root and the upasarga or sense with which it takes the ātmanepada. viś "to enter" takes it after ni, krīḍ "to play" after anu, sam, pari and āṅ. krī and dā are marked with ñ. They have the ātmanepada already when the fruit is the agent's, and these rules give it even when it is not. A string of vārttikas adds roots the sūtras do not name: gam with āṅ, śikṣ, nāth, hṛ, kṝ, nu, pracch, śap. The running example is niviśate "he enters", and apaskirate vṛṣo hṛṣṭaḥ "the bull, delighted, scratches". In the second, SK 2688 adds an s before the k of kṝ.
Rule. After viś preceded by ni, the ātmanepada comes. One of Bhaṭṭoji's commentators notes that the "not" of SK 2681–2682 has stopped here.
Example.
ni + viś + laṭ
→ this rule: ātmanepada
→ niviśate "he enters".
Without ni: praviśati. One commentator gives this counter-example.
Applies when / blocked by. Only the upasarga ni counts. One commentator gives madhūni viśanti bhramarāḥ. The ni at the end of madhūni is not the upasarga, so no ātmanepada.
In laṅ and luṅ the augment a stands between ni and viś: nyaviśata, nyavikṣata. Both commentators say a vārttika gives the ātmanepada even so: aḍvyavāye upasaṃkhyānam, "add the case where aṭ intervenes". One of them adds that on the Bhāṣya's view no vārttika is needed.
Rule. After krī "to buy" preceded by pari, vi or ava, the ātmanepada comes. Bhaṭṭoji: "this is for akartrabhiprāya", the case where the fruit is not meant for the agent. One of his commentators explains. krī is krīñ, marked with ñ. When the fruit is the agent's, it takes the ātmanepada anyway, by SK 2158.
Example.
pari + krī + laṭ, fruit for another
→ SK 2158 does not apply; this rule: ātmanepada
→ parikrīṇīte.
Likewise vikrīṇīte, avakrīṇīte.
Watch out. This is the first rule that serves a ñ-marked root. It does not give the ātmanepada to a parasmaipada root. It widens the ātmanepada of a ñ-marked root to the case where the fruit goes to someone else.
Rule. After ji "to conquer" preceded by vi or parā, the ātmanepada comes.
Example.
vi + ji + laṭ → this rule → vijayate "he excels".
parā + ji + laṭ → this rule → parājayate "he is defeated".
One of Bhaṭṭoji's commentators glosses the two as "he becomes superior" and "he becomes inferior".
Rule. After dā "to give" preceded by āṅ, the ātmanepada comes. The exception is the sense of opening the mouth (āsyaviharaṇa).
Example.
ā + dā + laṭ, "to take"
→ this rule: ātmanepada
→ vidyām ādatte "he takes up learning".
In opening the mouth: mukhaṃ vyādadāti "he opens his mouth".
Applies when / blocked by. Bhaṭṭoji says the word "mouth" is not meant to restrict. Any opening at all keeps the parasmaipada:
vipādikāṃ vyādadāti "he opens up the foot-sore"; vipādikā is a bursting sore of the foot;nadī kūlaṃ vyādadāti "the river breaks open the bank".A vārttika, parāṅgakarmakān na niṣedhaḥ, lifts the exception when the object is another's limb: vyādadate pipīlikāḥ pataṅgasya mukham "the ants open the bird's mouth". One commentator adds that they open it in order to eat.
Both commentators say the rule is for akartrabhiprāya, since dāñ is marked with ñ. One of them draws the consequence. In vyādatte vihagapatir mukhaṃ svakīyam "the king of birds opens his own mouth", the ātmanepada is correct. The fruit goes to the agent, so SK 2158 gives the ātmanepada. This rule, with its exception, does not concern that case.
Rule. After krīḍ "to play" preceded by anu, sam or pari, the ātmanepada comes. Bhaṭṭoji: by ca "and", also after āṅ.
Example.
sam + krīḍ + laṭ → this rule → saṃkrīḍate "he plays".
Likewise anukrīḍate, parikrīḍate, ākrīḍate.
Applies when / blocked by. Not after anu when it is a karmapravacanīya. Bhaṭṭoji derives this from association: anu is named with sam, an upasarga, so it too must be an upasarga. One of his commentators calls the statement a vārttika that follows from reasoning.
māṇavakam anukrīḍati "he plays with the boy"
→ anu means "with", the sense of the third case: a karmapravacanīya by SK 549
→ not an upasarga: no ātmanepada.
Then Bhaṭṭoji gives six vārttikas:
saṃkrīḍate; but saṃkrīḍati cakram "the wheel creaks".āgamayasva tāvat, which he glosses "do not hurry". One commentator renders it "just bear with it a while".dhanuṣi śikṣate "he wishes to become capable in the knowledge of the bow". Both commentators say this śikṣ is śak "to be able" with san. The other root śikṣ "to learn" already takes the ātmanepada, being marked with an anudātta vowel.sarpiṣo nāthate "he wishes: may I have ghee". In the verse nāthase kimu patiṃ na bhūbhṛtām there is no wishing, so Bhaṭṭoji says to read nādhase, with dh. One of his commentators explains why a restriction is needed. nāth has an anudātta marker only so that the noun nāthana can be formed.paitṛkam aśvā anuharante "horses take after the father's manner"; mātṛkaṃ gāvaḥ "cows, the mother's". But mātur anuharati "he takes after his mother": both commentators say mere likeness is meant here, not habit.Rule. After apa, kṝ takes the augment suṭ before its k. The sense is scratching (ālekhana), done by four-footed animals or birds. This rule comes from the sixth book. It stands here because the kṝ vārttika of SK 2687 needs it. One of Bhaṭṭoji's commentators gives the source of the words. "kṝ" is carried forward from kiratau lavane (SK 2539). "suṭ before the k" comes from the heading SK 2553.
Example. A vārttika limits the suṭ, too, to joy and the rest: suḍ api harṣādiṣv eva vaktavyaḥ.
apa + kṝ, a bull scratching in delight
→ this rule: suṭ before the k → apa + s + kṝ
→ the forms of kṝ (kirati, chapter 49), and the ātmanepada by the vārttika under SK 2687
→ apaskirate vṛṣo hṛṣṭaḥ "the bull, delighted, scratches".
So also kukkuṭo bhakṣārthī "the cock, seeking food" and śvā āśrayārthī "the dog, seeking a place to lie": each apaskirate.
Applies when / blocked by. Without joy and the rest: apakirati kusumam "it scatters flowers". Here there is neither ātmanepada nor suṭ. What if joy is present but there is no scratching? Bhaṭṭoji says the ātmanepada would apply, but "they say" it is not accepted without suṭ. One of his commentators hears doubt in "they say"; scholars discuss this.
gajo 'pakirati "the elephant scatters" also has neither. Bhaṭṭoji's commentators read this example differently. One says it only describes the animal's nature. The other says there is joy but no scratching, so suṭ cannot come.
Two more vārttikas add roots:
ānute, āpṛcchate. One commentator supplies "the jackal" with ānute, following the Bhāṣya.kṛṣṇāya śapate "he swears to Kṛṣṇa". Both commentators say kṛṣṇāya is sampradāna by SK 572.sam + sthā + te → santiṣṭhate; luṅ sam-a-sthā-s-ta → (2389: i for ā, sic kit) samasthita; luṅ of han before the ātmanepada: with āṅ, āhan s ta → (sic kit, n deleted, 2697) āhata; with the substitute vadha, han → (2696) avadhiṣṭaThis group gives the ātmanepada to sthā "to stand" with four upasargas and in several senses. Then a chain of rules begins with akarmakāc ca (SK 2693), "and when it has no object". The word akarmaka runs on into the rules for tap "to heat" and for yam and han with āṅ. A vārttika widens it: an object that is one's own limb counts too. Three rules from outside this run follow (A 2.4.44, 1.2.14, 1.2.15). They concern the luṅ of han and yam before the ātmanepada, where sic is kit or vadha replaces han. The running example is santiṣṭhate "it comes to completion", with its luṅ samasthita. Beside it stands the luṅ of han before the ātmanepada: with āṅ, āhata (SK 2697); with the substitute vadha, avadhiṣṭa (SK 2696).
Rule. After sthā preceded by sam, ava, pra or vi, the ātmanepada comes. sthā is otherwise a parasmaipada root: tiṣṭhati.
Example.
sam + sthā + laṭ
→ this rule: ātmanepada → sam + sthā + te
→ tiṣṭha for sthā before śap (SK 2360) → santiṣṭhate "it comes to completion".
One of Bhaṭṭoji's commentators gives the gloss. Likewise avatiṣṭhate, pratiṣṭhate, vitiṣṭhate.
In luṅ Bhaṭṭoji cites SK 2389:
sam + a + sthā + s + ta
→ SK 2389: i for the ā, and sic is kit → sam + a + sthi + s + ta
→ the s of sic drops after the short vowel, before the jhal t (SK 2369) → samasthita.
In the dual and plural the ending begins with a vowel, so the s stays and becomes ṣ (SK 212): samasthiṣātām, samasthiṣata.
Applies when / blocked by. A vārttika adds āṅ in the sense of affirming: āṅaḥ pratijñāyām upasaṃkhyānam. śabdaṃ nityam ātiṣṭhate "he affirms that word is eternal".
Rule. sthā also takes the ātmanepada when it expresses prakāśana, making one's mind known, or stheya, treating someone as arbiter. "sthā" is carried forward from SK 2689. Both of Bhaṭṭoji's commentators explain stheya as the one who decides a disputed matter.
Example.
gopī kṛṣṇāya tiṣṭhate: Bhaṭṭoji glosses "she makes known her intention [to Kṛṣṇa]". One commentator notes that kṛṣṇāya is sampradāna by SK 572.
saṃśayya karṇādiṣu tiṣṭhate yaḥ "he who, in doubt, takes his stand on Karṇa and the rest". Bhaṭṭoji: he resorts to them as deciders.
Applies when / blocked by. In neither sense: caitre tiṣṭhati, one commentator's counter-example. Bhaṭṭoji's commentators differ on the seventh case in karṇādiṣu. One takes it as the locus, the other as the subject matter.
Rule. sthā preceded by ud takes the ātmanepada when it denotes an action that is not upward. Both of Bhaṭṭoji's commentators define "upward action" as one that brings contact with a higher place.
Example.
muktau + ud + sthā + laṭ, striving, not rising
→ this rule: ātmanepada
→ muktāv uttiṣṭhate "he strives for liberation".
Upward: pīṭhād uttiṣṭhati "he rises from the seat".
Applies when / blocked by. A vārttika narrows the rule: īhāyām eva, "only in purposeful bodily activity". Hence grāmāc chatam uttiṣṭhati "a hundred accrues from the village": no ātmanepada. Both commentators take īhā as bodily movement; one adds that it is preceded by desire. That commentator also says the sūtra has the word karman "action" to bring in movement, and Bhaṭṭoji gives the vārttika with that in view.
Rule. sthā preceded by upa takes the ātmanepada when a mantra is the instrument.
Example.
āgneyyā āgnīdhram upa + sthā, the instrument a verse to Agni
→ this rule: ātmanepada
→ āgneyyāgnīdhram upatiṣṭhate "with the āgneyī verse he attends upon the āgnīdhra".
No mantra: bhartāram upatiṣṭhati yauvanena "she attends upon her husband with her youth".
Bhaṭṭoji's commentators differ on what upa-sthā means here, praise or standing near; scholars discuss this.
Applies when / blocked by. A vārttika adds four senses: upād devapūjāsaṅgatikaraṇamitrakaraṇapathiṣv iti vācyam. One commentator says it is for cases where no mantra is the instrument.
ādityam upatiṣṭhate "he worships the sun";gaṅgā yamunām upatiṣṭhate "the Gaṅgā joins the Yamunā";rathikān upatiṣṭhate "he makes friends of the charioteers";panthāḥ srughnam upatiṣṭhate "the road reaches Srughna".Bhaṭṭoji asks about the verse stutyaṃ stutibhir arthyābhir upatasthe sarasvatī "Sarasvatī attended upon the praiseworthy one with fitting praises". The one praised is a king, not a deity. He answers that divinity is attributed to him, or that a king is a portion of a deity.
A second vārttika, vā lipsāyām iti vaktavyam, makes the ātmanepada optional when the sense is wishing to obtain: bhikṣukaḥ prabhum upatiṣṭhate or upatiṣṭhati "the mendicant waits upon the lord".
Rule. sthā preceded by upa also takes the ātmanepada when it is akarmaka, "without an object". One of Bhaṭṭoji's commentators says upa comes from SK 2692 and sthā from SK 2689.
Example.
bhojanakāle upa + sthā, no object
→ this rule: ātmanepada
→ bhojanakāle upatiṣṭhate "he is present at mealtime".
Watch out. The word akarmakāt does not stop here. It runs on into SK 2694, SK 2695 and SK 2699, and into the vārttika artiśrudṛśibhyaś ca under SK 2701.
Rule. After tap "to heat, shine" preceded by ud or vi, the ātmanepada comes, when the root is akarmaka. Bhaṭṭoji: "akarmakāt holds here too". One of his commentators says this is the Bhāṣya's statement.
Example.
ud + tap + laṭ, no object
→ this rule: ātmanepada
→ uttapate "it shines". Likewise vitapate.
Applies when / blocked by. A vārttika adds an object that is one's own limb: svāṅgakarmakāc ceti vaktavyam. Both commentators say its ca keeps the akarmaka case as well. Bhaṭṭoji explains that svāṅga here is plainly "one's own limb". It is not the technical svāṅga of SK 510.
uttapate pāṇim or vitapate pāṇim "he warms [his own] hand".
Not suvarṇam uttapati "he heats or melts the gold": the gold is no limb.
Not caitro maitrasya pāṇim uttapati "Caitra heats Maitra's hand".
Watch out. The last example shows why the plain sense is needed. A hand is a svāṅga by the technical definition of SK 510. But it is not Caitra's own hand, so there is no ātmanepada. Both commentators make this point.
Rule. After yam "to hold" and han "to strike" preceded by āṅ, the ātmanepada comes. Bhaṭṭoji: the root must still be akarmaka, or have one's own limb as its object.
Example.
ā + yam + laṭ, no object or one's own limb
→ this rule: ātmanepada
→ āyacchate. One of Bhaṭṭoji's commentators glosses: "the rope stretches", or "he stretches out [his own] foot".
ā + han + laṭ → this rule → āhate "he strikes". The same commentator supplies "his own belly".
Another's limb: parasya śira āhanti "he strikes another's head".
Applies when / blocked by. Bhaṭṭoji then tests two verses. Bhāravi has ājaghne viṣamavilocanasya vakṣaḥ "he struck the chest of the odd-eyed one". Bhaṭṭi has āhadhvaṃ mā raghūttamam "do not strike the best of the Raghus". In both, the object is another's. Bhaṭṭoji gives four answers:
On the second answer Bhaṭṭoji adds a remark. The rule lyablope pañcamī puts a fifth case where a lyap-form is left out. It covers cases where the meaning is understood without the lyap-form, he says. One commentator draws the point: here the word itself is supplied, so there is no fifth case.
Rule. In luṅ, before ātmanepada endings, vadha optionally replaces han. Bhaṭṭoji's gloss supplies han, vadha and luṅ. They come from SK 2433 and SK 2434 (chapter 45). There vadha replaced han in luṅ without option. One of his commentators remarks that "taṅi", "before a taṅ ending", would have been simpler wording. Taṅ is the grammar's short name for the ātmanepada endings.
Example.
Bhaṭṭoji's forms, without an upasarga:
han + luṅ, ātmanepada
→ this rule: vadha for han
→ avadhiṣṭa, dual avadhiṣātām.
Without the substitute, the next rule applies.
Watch out. This rule is from the second book of the Aṣṭādhyāyī. It and the next two stand here for the luṅ of han and yam, which take the ātmanepada with āṅ in this group.
Rule. sic after han is kit. Bhaṭṭoji supplies "is kit". One of his commentators says "kit" is carried forward from SK 2242. Chapter 44 already used this sūtra: SK 2389 takes its word sic from here.
Example. One commentator sets up the starting point.
ā + a + han + s + ta (luṅ; ā and the augment a merge into ā: āhan s ta)
→ this rule: sic is kit
→ the n of han is lost before a jhal-initial kit affix (SK 2428) → ā + ha + s + ta
→ the s drops after the short vowel (SK 2369) → āhata.
Dual āhasātām, plural āhasata.
Watch out. Bhaṭṭoji names the loss of the nasal; one commentator identifies the rule as SK 2428. The loss of the s is the familiar SK 2369; the kaumudī does not state it.
Rule. sic after yam is kit when yam means gandhana. Bhaṭṭoji explains gandhana as informing: exposing another's faults. He supplies "sic is kit" to complete the sūtra. One of his commentators confirms the supply.
Example.
ud + ā + a + yam + s + ta (luṅ; ā and the augment a merge into ā), "exposing faults"
→ this rule: sic is kit
→ the m is lost (SK 2428), as one commentator says → ud + ā + ya + s + ta
→ the s drops as in āhata → udāyata "he exposed [another's faults]".
In another sense sic is not kit and the nasal stays: udāyaṃsta pādam "he drew up his foot".
sam + gam + te → saṃgacchate; luṅ samagata (sic kit, 2700) / samagaṃsta; sam + vid + jha → saṃvid-at-e → (ruṭ, 2701) saṃvidrate / saṃvidateThis group gives the ātmanepada to gam "to go" and ṛcch with sam, when they have no object. Vārttikas add more roots with sam: vid "to know", pracch, svṛ, ṛ, śru, dṛś, and, after any upasarga, as "to throw" and ūh. In the middle Bhaṭṭoji quotes a verse giving four grounds on which a root that seems to have an object counts as akarmaka. Three rules from outside this run handle forms of these roots (A 1.2.13, 7.1.7, 7.4.23). They are the optional kit of gam, the augment ruṭ after vid, and the short u of ūh. The running example is saṃgacchate "he meets", luṅ samagata or samagaṃsta; and saṃvidrate or saṃvidate "they know".
Rule. After gam and ṛcch preceded by sam, the ātmanepada comes, when they are akarmaka. Bhaṭṭoji: "akarmaka still holds". One of his commentators says this means the own-limb vārttika of SK 2694 has stopped here.
Example.
sam + gam + laṭ, no object
→ this rule: ātmanepada
→ saṃgacchate "he meets", that is, "he becomes united".
With an object: grāmaṃ saṃgacchati "he goes to the village" (Bhaṭṭoji's counter-example, given under SK 2700).
Applies when / blocked by. One of Bhaṭṭoji's commentators says he follows the Bhāṣya's reading of the sūtra, with two roots. The Vṛtti, an older commentary, reads a longer sūtra: samo gamyṛcchipracchisvaratyartiśruvidibhyaḥ.
Most of the roots that longer reading adds come in here through the vārttikas under SK 2700 and SK 2701.
Rule. After gam, a liṅ affix and a sic that begin with a jhal are optionally kit. Both of Bhaṭṭoji's commentators give the sources. "jhal" comes from SK 2612, "liṅ and sic" from SK 2300, and "kit" from SK 2242.
Example.
sam + a + gam + s + ta (luṅ), sic kit
→ the m is lost (SK 2428) → sam + a + ga + s + ta
→ the s drops (SK 2369) → samagata "he met".
Not kit: samagaṃsta.
One commentator gives both steps; the other gives the loss of m. Bhaṭṭoji gives only the forms.
In the āśīrliṅ, the liṅ of blessing: saṃgasīṣṭa (kit) or saṃgaṃsīṣṭa.
Bhaṭṭoji adds the lṛṭ form samṛcchiṣyate. Both commentators say it shows which ṛcch is meant in SK 2699. It is the root ṛcch of the tud class, not the ṛccha that substitutes for ṛ.
Applies when / blocked by. One commentator adds a usage: tac caikyaṃ samagacchata "and that became one". gam here means "become one", so it is akarmaka and takes the ātmanepada.
A vārttika adds three roots with sam: vidipracchisvaratīnām upasaṃkhyānam, "add vid, pracch and svṛ". One commentator says sam carries into it. Bhaṭṭoji says only vett is meant: vid "to know" of the ad class. saṃvitte, saṃvidāte. The commentators give different grounds for this. One appeals to the tradition of explanation. The other appeals to association: vid stands beside pracch and svṛ, which are parasmaipada roots.
Rule. After vett, the at that replaces jh optionally takes the augment ruṭ. Both of Bhaṭṭoji's commentators give the sources. "ruṭ" comes from SK 2442, as for śī in chapter 45. "at" comes from SK 2479, read in the sixth case. "jh" comes from SK 2169.
Example.
sam + vid + jha (laṭ, third person plural, ātmanepada by the vārttika under SK 2700)
→ at for jh (chapter 44) → saṃvid + ata
→ e for the final a (SK 2233) → saṃvid + ate
→ this rule: ruṭ goes at the front of at (SK 36) → saṃvid + r + ate
→ saṃvidrate "they know".
Without the augment: saṃvidate.
The printed Kaumudī also gives loṭ saṃvidratām / saṃvidatām and laṅ samavidrata / samavidata.
Then Bhaṭṭoji gives the other two roots of the vārttika: saṃpṛcchate, saṃsvarate.
Applies when / blocked by. A vārttika adds three more roots with sam: artiśrudṛśibhyaś ceti vaktavyam, "ṛ, śru and dṛś". Bhaṭṭoji says arti takes in two roots here, ṛ of the bhū class and iyarti (ṛ of the hu class), and gives the forms of each. In the aṅ rule (SK 2382), he notes, only iyarti is taken.
mā samṛta, mā samṛṣātām, mā samṛṣata; and with the augment, samārta, samārṣātām, samārṣata.mā samarata, mā samaretām, mā samaranta; and samārata, samāretām, samāranta.Both commentators explain mā samṛta. sic is kit by SK 2368, so there is no guṇa, and after mā no augment comes. One adds that sic drops by SK 2369. Whether samārta is right is a point scholars discuss.
The other two roots: saṃśṛṇute, saṃpaśyate.
Bhaṭṭoji adds that akarmaka still holds here. That is why, "they say", Murāri's rakṣāṃsīti purāpi saṃśṛṇumahe "even before, we heard [they were] rākṣasas" is a slip. Here śru has an object, so it should not take the ātmanepada. Or else supply iti kathayadbhyaḥ "from those saying so". One commentator explains: then rakṣāṃsi goes with the saying, and śru has no object.
The four grounds of akarmaka. How can han, gam and the rest be akarmaka in this section, when they take objects? Bhaṭṭoji answers with a verse:
dhātor arthāntare vṛtter dhātvarthenopasaṃgrahāt | prasiddher avivakṣātaḥ karmaṇo 'karmikā kriyā ||
An action is without an object for four reasons:
vahati bhāram "he carries a load" has an object, but nadī vahati "the river flows" has none;jīvati "he lives", nṛtyati "he dances";megho varṣati "the cloud rains";hitān na yaḥ saṃśṛṇute sa kiṃprabhuḥ "he who does not listen to a well-wisher, a poor master is he".One commentator adds that this settling of akarmaka-status is used also in SK 2152 and similar rules.
A last vārttika, upasargād asyatyūhyor veti vācyam, gives the ātmanepada optionally to as "to throw" and ūh after an upasarga. Bhaṭṭoji: here akarmaka has stopped. One commentator explains why: these two always take an object with an upasarga. bandhaṃ nirasyati / nirasyate "he throws off the bond"; samūhati / samūhate.
Watch out. SK 2700, this rule and SK 2702 lie outside the run (A 1.2.13, 7.1.7, 7.4.23). They stand here for the forms of sam-gam and of the vārttika roots vid and ūh.
Rule. After an upasarga, the vowel of ūh "to reason" is shortened. This happens before an affix that begins with y and is kit or ṅit. A ṅit affix counts as marked with ṅ; like kit, it blocks guṇa. Bhaṭṭoji supplies that condition; one commentator traces it to SK 2649.
Example.
sam + ūh + yāt (āśīrliṅ)
→ this rule: u for ū → samuhyāt
→ brahma samuhyāt "may he reflect well on brahman".
sam + ūh + ya (lyap, the substitute for ktvā)
→ this rule → agniṃ samuhya "having swept around the fire".
One of Bhaṭṭoji's commentators explains that the yās of the āśīrliṅ is kit, and that lyap is kit. Bhaṭṭoji gives only the examples.
ni + hve + śap + te → nihvayate; ud + kṛ + te → utkurute (in the sense gandhana)Two ñ-marked roots come next. hve "to call" is hveñ; kṛ "to do" is kṛñ. Both already take the ātmanepada by SK 2158 when the fruit goes to the agent. These rules give it with certain upasargas and in certain senses, whoever gets the fruit. hve takes it after ni, sam, upa and vi, and after āṅ in rivalry. kṛ takes it in seven senses, after adhi in forbearing or overpowering, and after vi with a sound as object or with no object. The running examples are nihvayate "he calls" and utkurute "he informs against [someone]".
Rule. After hve preceded by ni, sam, upa or vi, the ātmanepada comes. The sūtra's hvaḥ is the fifth case of hvā, says one of Bhaṭṭoji's commentators. That is hveñ after its e has already become ā.
Example.
ni + hve + laṭ
→ this rule: ātmanepada → ni + hve + śap + te
→ ay for e before the vowel of śap (SK 61) → nihvayate "he calls".
The same commentator adds saṃhvayate, upahvayate, vihvayate.
Applies when / blocked by. That commentator also says the rule is for the case where the fruit does not go to the agent. He gives no reason. The likely one is that hveñ is marked with ñ, so, as with krī (SK 2684), SK 2158 covers the other case.
Rule. After hve preceded by āṅ, the ātmanepada comes in the sense of rivalry (spardhā). "hve" is carried forward from SK 2703. One of Bhaṭṭoji's commentators states it in his gloss: "ātmanepada after hveñ with āṅ, in the sense of rivalry".
Example.
ā + hve + laṭ, rivalry
→ this rule: ātmanepada
→ kṛṣṇaś cāṇūram āhvayate "Kṛṣṇa challenges Cāṇūra".
No rivalry: putram āhvayati "he calls his son".
Rule. After kṛ (kṛñ), the ātmanepada comes in seven senses. One of Bhaṭṭoji's commentators adds: "even when the fruit is not the agent's". The senses and Bhaṭṭoji's examples:
utkurute "he informs against [someone]", that is, he denounces. Bhaṭṭoji explains that informing is injury, since it leads to loss of life.śyeno vartikām udākurute "the hawk menaces the quail".harim upakurute "he serves Hari".paradārān prakurute. Bhaṭṭoji: "he rushes rashly upon them", others' wives.edhodakasyopaskurute "he imparts a quality to the fuel and the water".gāthāḥ prakurute "he narrates the verses".śataṃ prakurute "he lays out a hundred for dharma".Example.
ud + kṛ + laṭ, in the sense of injury
→ this rule: ātmanepada
→ the forms of kṛ (chapter 51) → utkurute.
In none of the senses: kaṭaṃ karoti "he makes a mat".
Applies when / blocked by. Both commentators explain why gandhana means injury. The root gandh is read in the sense of ard, and ard means injury.
One of Bhaṭṭoji's commentators says his gloss of paradārān prakurute states only the resulting sense. The other calls rash conduct a prayogopādhi. It is a condition under which the word is used, not the root's own meaning. Otherwise the object paradārān would not fit the verb.
For edhodaka both commentators give two analyses: a compound of edha "fuel" and udaka "water", or edhas "fuel" with daka, another word for water. The sense is the same. The s of upaskurute is the augment suṭ of chapter 51 (SK 2552), where pratiyatna is one of its senses.
Watch out. This rule does not give the ātmanepada to a parasmaipada root. kṛñ has it already by SK 2158 when the fruit is the agent's: kurute. This rule extends it, in these senses, to the fruit that goes to another.
Rule. After kṛ preceded by adhi, the ātmanepada comes in the sense of prasahana. Bhaṭṭoji explains that prasahana is both forbearance and overpowering. The reason is that the root ṣah is listed with both meanings: "ṣah in enduring and in overpowering".
Example.
adhi + kṛ + laṭ, forbearing or overpowering
→ this rule: ātmanepada
→ śatrum adhikurute "he bears with the enemy", or "he overpowers the enemy".
Rule. After kṛ preceded by vi, the ātmanepada comes when its object is a sound. One of Bhaṭṭoji's commentators reads śabdakarmaṇaḥ as a bahuvrīhi: "that whose object is a sound".
Example.
svarān vi + kṛ + laṭ, the object a sound
→ this rule: ātmanepada
→ svarān vikurute "he utters the notes".
The object not a sound: cittaṃ vikaroti kāmaḥ "desire unsettles the mind".
Rule. After kṛ preceded by vi, the ātmanepada also comes when the root has no object. Bhaṭṭoji: "veḥ kṛñaḥ still holds", that is, "kṛ after vi".
Example.
chātrāḥ vi + kṛ + laṭ, no object
→ this rule: ātmanepada
→ chātrā vikurvate "the students undergo a change".
Bhaṭṭoji glosses: "they acquire a change".
Watch out. "Acquire a change" seems to have an object, the change. One of Bhaṭṭoji's commentators answers. The object is contained in the root's meaning. That is the second of the four grounds in the verse at SK 2701.
nī + śap + te → nayate (śāstre nayate, sammānana); kram + śap + te → kramate (ṛci kramate buddhiḥ, vṛtti)nī "to lead" is ṇīñ, marked with ñ. SK 2709 gives it the ātmanepada in seven senses, whoever gets the fruit. SK 2710 is a restriction on the ātmanepada it already has by SK 2158. kram "to step" is a parasmaipada root: krāmati. SK 2711 gives it the ātmanepada in three senses. The next rules govern it with upasargas: upa and parā only, āṅ for rising, vi for stepping, pra and upa for beginning. Without an upasarga, the ātmanepada is optional. The running examples are śāstre nayate "he conveys [the doctrine] in the śāstra" and ṛci kramate buddhiḥ "his intellect moves freely in the ṛc".
Rule. After nī (ṇīñ), the ātmanepada comes when one of seven senses is understood. One of Bhaṭṭoji's commentators adds: "even when the fruit goes to another". Bhaṭṭoji divides the seven. Four are expressed by nī itself: utsañjana, jñāna, vigaṇana and vyaya. The other three are prayogopādhi, conditions of usage. One of his commentators explains the term. It is a sense not expressed by the root but implied; its presence calls for the word, and the ātmanepada signals it.
Example.
nī + laṭ, sense sammānana
→ this rule: ātmanepada → nī + śap + te
→ guṇa before śap (SK 2168) → ne + a + te
→ ay for e (SK 61) → nayate.
The seven senses with Bhaṭṭoji's examples:
śāstre nayate "he conveys [the doctrine] in the śāstra" to his pupils. The pupils' honour follows from that.daṇḍam unnayate "he raises the staff".māṇavakam upanayate "he initiates the boy", bringing him near according to the rite.tattvaṃ nayate "he ascertains the truth".karmakārān upanayate "he takes on workmen", bringing them near by paying wages.karaṃ vinayate "he pays off the tax", the share due to the king.śataṃ vinayate "he spends a hundred" for dharma.Applies when / blocked by. How does upanayate make one a teacher? Bhaṭṭoji explains. The status of teacher (ācāryatva) comes from teaching that follows the initiation. Both commentators agree, and one calls the link indirect. That commentator adds that the fruit of initiation goes to the boy, so the parasmaipada was due; this rule gives the ātmanepada.
Rule. This is a niyama. Take nī with an object located in the agent. Its ātmanepada then comes only if that object is not a part of the body. Bhaṭṭoji says śarīra "body" in the sūtra stands, by lakṣaṇā (indication), for a part of the body.
Example.
krodhaṃ vi + nī + laṭ "he drives away his anger"
→ the fruit, calm of mind, goes to the agent
→ ṇīñ is ñ-marked: ātmanepada already by SK 2158
→ the object, anger, is in the agent but is not a bodily part: this rule lets it stand
→ krodhaṃ vinayate "he subdues his anger".
A bodily part: gaḍuṃ vinayati "he removes a swelling", parasmaipada.
Applies when / blocked by. Bhaṭṭoji is explicit: since SK 2158 already gives the ātmanepada, this sūtra is for restriction. One of his commentators puts it this way: the ātmanepada is restricted here, not enjoined. He also explains "part of the body". The body is the agent itself, so it cannot be "in" the agent; only its parts can.
What of the verse vigaṇayya nayanti pauruṣam "having weighed it, they bring their manliness to bear"? Manliness is not a bodily part, so the ātmanepada would be expected. Bhaṭṭoji answers that here the fruit's going to the agent is not meant to be expressed.
Bhaṭṭoji's commentators construe the sūtra's karmaṇi differently. One takes it as the object kāraka, the other as the fruit going to oneself; scholars discuss this.
Rule. After kram, the ātmanepada comes in three senses. Vṛtti is unimpeded course, sarga is zeal and tāyana is flourishing. One of Bhaṭṭoji's commentators supplies "ātmanepada" to complete the sūtra.
Example.
kram + laṭ, sense vṛtti
→ this rule: ātmanepada → kram + śap + te
→ ṛci kramate buddhiḥ "his intellect moves freely in the ṛc", that is, it is not checked.
adhyayanāya kramate "he sets about study eagerly".kramante 'smin śāstrāṇi "in him the śāstras flourish", that is, they grow abundant.Both commentators gloss tāyana as growth or abundance.
Watch out. In the ātmanepada the vowel stays short: kramate. The long vowel of krāmati comes only in the parasmaipada (SK 2322).
Rule. In the senses of SK 2711, kram with an upasarga takes the ātmanepada only after upa and parā. Bhaṭṭoji: "only after these two, not after another upasarga". SK 2711 alone would already give the ātmanepada. So this rule is a niyama; both of Bhaṭṭoji's commentators treat it so.
Example.
upa + kram + laṭ, sense vṛtti and the rest
→ SK 2711, and this rule allows upa → upakramate.
Likewise parākramate.
Another upasarga: saṃkrāmati, parasmaipada.
Rule. After kram preceded by āṅ, the ātmanepada comes in the sense of rising. One of Bhaṭṭoji's commentators supplies kram and says the sense of rising comes from the upasarga.
Example.
ā + kram + laṭ, rising
→ this rule: ātmanepada
→ ākramate sūryaḥ "the sun comes up", that is, it rises.
Applies when / blocked by. A vārttika narrows the rule: jyotirudgamane iti vācyam, "only for the rising of a heavenly light". So not ākrāmati dhūmo harmyatalāt "smoke rises up from the roof of the mansion".
Rule. After kram preceded by vi, the ātmanepada comes in the sense of moving the feet. One of Bhaṭṭoji's commentators glosses it as placing the feet.
Example.
vi + kram + laṭ, placing the feet
→ this rule: ātmanepada
→ sādhu vikramate vājī "the horse steps out well".
Another sense: vikrāmati sandhiḥ "the joint splits apart"; Bhaṭṭoji: it becomes twofold, it cracks open.
Watch out. kram already means placing the feet. Why state the sense? The other commentator answers: roots have many meanings, so the condition is needed.
Rule. After kram preceded by pra or upa, the ātmanepada comes when the two upasargas mean the same. Bhaṭṭoji: samartha means "having the same meaning". The two mean the same in the sense of beginning.
Example.
pra + kram + laṭ, "beginning"
→ this rule: ātmanepada
→ prakramate "he begins". Likewise upakramate.
Not the same meaning: prakrāmati "he goes forth", upakrāmati "he comes near".
Applies when / blocked by. Bhaṭṭoji explains the form of the word samartha. It is sama + artha. The two a's do not merge into ā, because the word belongs to the śakandhu group, where the second vowel takes the place of both (pararūpa). One of his commentators remarks that propābhyāṃ prārambhe, "after pra and upa in beginning", would have been simpler.
Why no ātmanepada in upakrāmati, given SK 2712? The same commentator answers: SK 2712 holds only in the senses of SK 2711.
Rule. After kram with no upasarga, the ātmanepada comes optionally. Bhaṭṭoji calls it an aprāptavibhāṣā: an option for something not otherwise obtained. One of his commentators explains: without an upasarga kram never gets the ātmanepada from any other rule.
Example.
kram + laṭ, no upasarga, no special sense
→ this rule: optionally ātmanepada
→ kramate or krāmati "he steps".
Applies when / blocked by. In the senses of SK 2711 the ātmanepada is obligatory, says Bhaṭṭoji. His commentators ground this differently. One appeals to pūrvavipratiṣedha, by which here the earlier rule, SK 2711, wins over this later one. The other appeals to the niyama of SK 2712, which holds in those senses. Bhaṭṭoji himself says only "obligatory".
apa + jñā + te → apajānīte (apahnava); vad + śap + te → vadate (śāstre vadate, bhāsana)Two roots fill this group. Jñā "to know" belongs to the krī class and normally takes the parasmaipada: jānāti "he knows". Three rules give it the ātmanepada in particular senses. With apa, in the sense of denial, it gives apajānīte "he denies". Vad "to speak" is also a parasmaipada root: vadati. Four rules give it the ātmanepada. The first lists six senses: śāstre vadate "he speaks brilliantly on the śāstra". Three more concern speakers of articulate speech: speaking together, speaking like another, and contradicting one another.
Rule. After jñā, in the sense of denial (apahnava), the ātmanepada endings come. "Ātmanepada" runs on from SK 2157, the heading of this whole chapter. One of Bhaṭṭoji's commentators supplies it here.
Example.
śatam apajānīte, "he denies the hundred":
apa + jñā + laṭ, agent, third person singular
→ ātmanepada by this rule: the ending ta
→ the krī-class sign nā (śnā, SK 2554, chapter 52) → apa + jñā + nā + ta
→ jā for jñā before this śit affix (SK 2511, chapter 47) → apa + jā + nā + ta
→ ī for the ā of nā before the ending (SK 2497, chapter 46) → apa + jā + nī + ta
→ te for ta (SK 2233)
→ apajānīte "he denies".
Bhaṭṭoji glosses it: "he disowns it" (apalapati). One of his commentators adds that apahnava is apalāpa, disowning.
Rule. After jñā used intransitively (akarmaka, "without an object"), the ātmanepada comes. "Jñā" is carried forward from SK 2717 and "ātmanepada" from SK 2157. One of Bhaṭṭoji's commentators supplies both.
Example.
sarpiṣo jānīte, "he sets to work with ghee as the means":
jñā + laṭ, used without an object → ātmanepada by this rule → śnā, jā, ī and te as for apajānīte (SK 2717) → jānīte.
Bhaṭṭoji's gloss: "he sets to work by means of ghee". Here jñā means "to engage in action", one of his commentators says. Sarpiṣaḥ "of ghee" is a sixth case standing for the third, "by means of ghee". That is by SK 612, jño 'vidarthasya karaṇe, which you met in chapter 16. It lets the means of jñā take the sixth case, when jñā does not mean "to know".
Applies when / blocked by. Bhaṭṭoji's example has bare jñā, with no upasarga. But bare jñā would get the ātmanepada anyway from a later rule, SK 2743. So where is this rule needed? Both of Bhaṭṭoji's commentators answer: in forms with an upasarga, such as sarpiṣo 'nujānīte.
Rule. After jñā preceded by sam or prati, the ātmanepada comes. The condition: the sense must not be ādhyāna, "longing recollection". One of Bhaṭṭoji's commentators defines ādhyāna: remembering with longing (utkaṇṭhā).
Example.
sam + jñā + laṭ → ātmanepada by this rule → saṃjānīte.
śataṃ saṃjānīte: "he looks after the hundred" (avekṣate).
śataṃ pratijānīte: "he acknowledges the hundred" (aṅgīkaroti).
These verbs have an object, śatam. So SK 2718, which needs an intransitive verb, could not have given the ātmanepada here. One of Bhaṭṭoji's commentators points this out.
Applies when / blocked by. Bhaṭṭoji says the word anādhyāne is split off as a rule of its own. This is a yogavibhāga, "splitting a rule": one sūtra is read as two. The first rule gives the ātmanepada after sam and prati. The second says: only when the sense is not ādhyāna.
Why split it? Bhaṭṭoji: by force of the split, even SK 2718 is kept from applying. Look at the counter-example:
mātaraṃ mātur vā saṃjānāti "he longingly remembers his mother".
The object may stand in the second case (mātaram) or the sixth (mātuḥ). The sixth comes when the object is intended as śeṣa, "a mere relation" (SK 606, chapter 16).
One of Bhaṭṭoji's commentators unpacks the argument. With the sixth case, mātuḥ saṃjānāti counts as intransitive. Then SK 2718 would give the ātmanepada, since SK 2718 has no "not in longing recollection". The split prevents this.
Watch out. A prohibition normally blocks only the rule just before it. That maxim is anantarasya, and you meet it again at SK 2732. Here it would block only the "sam or prati" rule. But then the split would achieve nothing that one sūtra could not. So, one of Bhaṭṭoji's commentators argues, the split must reach back further, to SK 2718.
Rule. After vad "to speak", the ātmanepada comes in six senses:
Bhaṭṭoji adds a distinction. Conciliation and solicitation are expressed by the root itself. The other four are prayogopādhi, "conditions of usage". You met this term at SK 2705 and SK 2709: a sense not expressed by the root but present, whose presence calls for the form. One of his commentators refers back to that explanation here. The ātmanepada makes such a condition evident.
Example.
vad + śap + ta → ātmanepada by this rule → te for ta (SK 2233) → vadate.
One of Bhaṭṭoji's commentators fills in the conditions. Shining forth means bringing ever new arguments against each objection. In the knowledge example, vadate means "he deals with" the śāstra. Such dealing is impossible without knowledge, so knowledge is implied. Likewise, disputing over the field is rooted in disagreement.
Applies when / blocked by. Outside these senses the parasmaipada stays. The other commentator gives yat kiṃcid vadati, "he says just anything".
Watch out. Bhaṭṭoji uses one sentence, śāstre vadate, for two senses: shining forth and knowledge. Do not look for a difference in the form. Both senses are conditions of usage, not something the verb itself expresses.
Rule. After vad, the ātmanepada comes when beings of articulate speech (vyaktavāc) speak together (samuccāraṇa). "Vad" is carried forward from SK 2720. Bhaṭṭoji's gloss: "in the speaking together, in concert, of human beings".
One of Bhaṭṭoji's commentators defines "articulate". The sounds are clearly uttered, through the distinction of vowels and consonants. The sam of saṃpravadante means "together", as one.
Example.
sam + pra + vad + laṭ, third person plural, ātmanepada by this rule → saṃpravadante.
saṃpravadante brāhmaṇāḥ, "the brāhmaṇas speak together".
Applies when / blocked by. Bhaṭṭoji's counter-example: saṃpravadanti khagāḥ, "the birds sing together". Birds do not have articulate speech, so the parasmaipada stays.
Rule. After intransitive vad preceded by anu, the ātmanepada comes. Its domain must be beings of articulate speech. Bhaṭṭoji's gloss: "after intransitive vad preceded by anu, having articulate speakers as its domain".
What is carried forward? "Vad" from SK 2720. One of Bhaṭṭoji's commentators adds: "of articulate speakers" continues, now read as a sixth case of domain. "Speaking together" stops here.
Example.
anu + vad + laṭ, no object → ātmanepada by this rule → anuvadate.
anuvadate kaṭhaḥ kalāpasya, "the Kaṭha student speaks like the Kalāpa student".
Both of Bhaṭṭoji's commentators say anu here means likeness. What case is kalāpasya? Both of Bhaṭṭoji's commentators end by calling it a sixth case of the śeṣa kind (SK 606). One of them first weighs SK 630, the rule for words meaning "equal", and sets it aside.
Applies when / blocked by. Bhaṭṭoji gives two counter-examples.
Rule. When the speaking together of articulate speakers is contradictory talk (vipralāpa), the ātmanepada comes optionally. One of Bhaṭṭoji's commentators says "speaking together of articulate speakers" is carried forward from SK 2721. Vipralāpa is contradictory statement (viruddhokti).
Example.
vi + pra + vad + laṭ, third person plural
→ ātmanepada by this rule → vipravadante
→ or, the option not taken, parasmaipada → vipravadanti.
vipravadante or vipravadanti vaidyāḥ, "the physicians contradict one another".
Applies when / blocked by. What kind of option is this? Bhaṭṭoji does not say. One of his commentators calls it an aprāptavibhāṣā: an option on something not otherwise due.
ava + gṝ + śa + te → avagirate; upa + yam → upayacchate, luṅ upa-a-yam-s-ta → (sic optionally kit, 2730) upāyata / upāyaṃstaFour more roots take the ātmanepada with particular upasargas. Gṝ "to swallow" takes it with ava: avagirate. Car "to move" takes it with ud or sam. Dāṇ "to give" takes it with sam in a special construction. Yam takes it with upa when one takes something as one's own: bhāryām upayacchate "he takes a wife". The group ends with a rule from another part of the Aṣṭādhyāyī. It makes the aorist affix sic of upa-yam optionally kit in the sense of marrying. So "Rāma married Sītā" is upāyata or upāyaṃsta.
Rule. After gṝ preceded by ava, the ātmanepada comes. One of Bhaṭṭoji's commentators supplies "ātmanepada". He identifies the root as gṝ of the tud class (tudādi), "to swallow", whose class-sign is śa.
Example.
ava + gṝ + śa + ta → ātmanepada by this rule
→ ir for the root-final ṝ (SK 2390, chapter 44) → ava + gir + a + ta
→ te for ta (SK 2233) → avagirate "he swallows".
Applies when / blocked by. Bhaṭṭoji adds: "But gṛṇāti preceded by ava is not used at all, says the Bhāṣya." Gṛṇāti is the krī-class root of the same sound. One of his commentators explains the remark. One might expect the ātmanepada in avagṛṇāti too. Bhaṭṭoji rules that out: only the tud-class root is meant.
Watch out. The sūtra's graḥ is the fifth case of the quoted root. A word quoting a root normally behaves like the root itself. That maxim is prakṛtivad anukaraṇam. If the quoted root behaved like the root gṝ, SK 2390 would have turned it into gir. The sūtra would then read giraḥ. One of Bhaṭṭoji's commentators says the maxim is not obligatory. This very graḥ is the proof.
Rule. After gṝ (girati) preceded by sam, the ātmanepada comes in the sense of affirming (pratijñāna). "Gṝ" is carried forward from SK 2724. One of Bhaṭṭoji's commentators glosses pratijñāna as abhyupagama, "acceptance".
Example.
sam + gṝ + śa + ta → ātmanepada by this rule → saṃgirate.
śabdaṃ nityaṃ saṃgirate, "he affirms that word is eternal". Bhaṭṭoji's gloss: "he asserts" (pratijānīte).
Applies when / blocked by. Why "in the sense of affirming"? Bhaṭṭoji: saṃgirati grāsam, "he swallows the mouthful". Here the root keeps its own sense, and the parasmaipada.
Rule. After car preceded by ud, the ātmanepada comes when the verb is transitive (sakarmaka, "with an object").
Example.
ud + car + śap + ta → ātmanepada by this rule → uccarate.
dharmam uccarate, "he transgresses dharma". Bhaṭṭoji's gloss: "he goes on, overstepping it".
Applies when / blocked by. Why "transitive"? Bhaṭṭoji: bāṣpam uccarati, "vapour rises", that is, "goes upward". Bāṣpam is the subject here, not an object. The verb has no object, so the parasmaipada stays.
Rule. After car with sam, the ātmanepada comes when the verb is joined to a word in the third case. "Car" is carried forward from SK 2726. One of Bhaṭṭoji's commentators says "transitive" does not carry forward.
Example.
sam + car + laṭ, with the third-case word rathena → ātmanepada by this rule → saṃcarate.
rathena saṃcarate, "he moves about by chariot".
Applies when / blocked by. The third-case word must actually be uttered. One of Bhaṭṭoji's commentators gives rathikāḥ saṃcaranti, "the charioteers move about". "By chariot" is understood from the sense. But no third-case word is heard, so there is no ātmanepada. That, he says, is why the sūtra says yukta, "joined to". Otherwise it could have said just tṛtīyayā, "with the third case".
He adds that the ātmanepada comes even when car has an object. Unlike SK 2726, this rule sets no condition about objects.
Rule. After dāṇ "to give" preceded by sam, joined to a third-case word, the ātmanepada comes. The condition: that third case stands in the sense of the fourth. One of Bhaṭṭoji's commentators says "with sam, joined to a third-case word" is carried forward from SK 2727. The word sā, "that", refers to the third case.
Example.
sam + dāṇ + laṭ, with dāsyā "to a slave-girl" in the third case
→ ātmanepada by this rule
→ the present stem yaccha for dāṇ before a śit affix (SK 2360, chapter 44) → saṃyacchate.
dāsyā saṃyacchate, "he gives to a slave-girl".
Why the third case for a recipient? One of Bhaṭṭoji's commentators cites a statement. With dāṇ, in improper dealings, the third case is used in the sense of the fourth.
Applies when / blocked by. Bhaṭṭoji adds a point about the previous sūtra. Its samaḥ is a sixth case. So both rules apply even when another upasarga stands between sam and the root:
Watch out. Why does the case matter? One of Bhaṭṭoji's commentators explains. A fifth case in a rule means "immediately after", by SK 41, tasmād ity uttarasya. Then ā or pra in between would block the rule. A sixth case shows only order, before and after, not adjacency.
Rule. After yam preceded by upa, the ātmanepada comes in the sense of svakaraṇa. Bhaṭṭoji glosses it as svīkāra, "taking as one's own".
Example.
upa + yam + śap + ta → ātmanepada by this rule
→ ch for the m of yam before the śit affix śap (SK 2400, chapter 44) → upa + yach + a + ta
→ the augment t before ch, after the short a (SK 146, chapter 5), which ends up as c → upa + yacch + a + ta
→ te for ta (SK 2233) → upayacchate.
bhāryām upayacchate, "he takes a wife".
Applies when / blocked by. One of Bhaṭṭoji's commentators sharpens the sense. Svakaraṇa is taking as one's own what was not one's own. Making one's own garment does not count: "he produces his own garment" stays vastram upayacchati.
Watch out. Why is the word svakaraṇa and not svīkaraṇa, with the affix cvi? Both of Bhaṭṭoji's commentators answer: cvi is optional, so it may be absent. One of them adds that the poet Bhaṭṭi uses the verb for accepting in general: upāyaṃsta mahāstrāṇi, "he accepted the great weapons".
Rule. In the sense of marrying (upayamana), the affix sic after yam is optionally kit. Bhaṭṭoji's gloss: "sic after yam is optionally kit in marriage". Sic is the aorist affix, heard as s. A kit affix is one that counts as marked with k, as at SK 2697 and SK 2698 earlier in this chapter. Here that lets the m of yam drop.
This rule is A 1.2.16, from the rules on kit affixes, not from the ātmanepada run (A 1.3). Bhaṭṭoji places it here because it concerns upa-yam. One of his commentators traces its words. "Yam" is carried forward from SK 2698, yamo gandhane. "Sic" comes from SK 2697, hanaḥ sic. "Kit" comes from SK 2242, asaṃyogāl liṭ kit.
Example.
rāmaḥ sītām upāyata, "Rāma married Sītā":
upa + a + yam + s + ta (luṅ, ātmanepada by SK 2729; the augment a)
→ sic is kit by this rule
→ the m of yam is deleted before the kit affix (SK 2428, chapter 45) → upa + a + ya + s + ta
→ sic is deleted after the short vowel (SK 2369, chapter 44) → upa + a + ya + ta
→ a + a merge → upāyata.
The other option, sic not kit: the m stays → upāyaṃsta.
Bhaṭṭoji's gloss: "he wedded" (udavoḍha). One of his commentators supplies the two deletions, by SK 2428 and SK 2369. Bhaṭṭoji gives the forms bare.
Applies when / blocked by. Bhaṭṭoji adds: "But in an upayama that is part of gandhana, kit is obligatory, by pūrvavipratiṣedha." Gandhana is the sense of SK 2698. One of his commentators says such a marriage is one preceded by violence, the rākṣasa marriage.
Pūrvavipratiṣedha means that in a conflict the earlier rule wins. Normally the later rule wins. Here the earlier SK 2698 makes kit obligatory, and it prevails over this option. Both of Bhaṭṭoji's commentators trace this to the Bhāṣya on SK 24, na veti vibhāṣā. One explains why. The Bhāṣya counts this option as an aprāptavibhāṣā. If it also covered the gandhana marriage, where kit is already due, it would be a prāptavibhāṣā, an option on something already due. So the earlier rule must win there.
Watch out. A kit sic is what deletes the m. So the two forms are not free variants in every sense. In ordinary marriage both are correct; in a forcible one only upāyata is.
jñā + san → jijñāsa + śap + te → dharmaṃ jijñāsate; edh (taṅ) + san → edidhiṣa → edidhiṣate (2734)In this group's title, taṅ is a short name for the ātmanepada endings. A root with the affix san is a desiderative: "he wishes to know", "he wishes to grow". You built these stems in chapter 55. Such a stem ends in san, so it is called sannanta, "san-final". Which set of endings does it take? Four rules answer. SK 2731 gives the ātmanepada to the san-finals of four roots of knowing and perceiving: dharmaṃ jijñāsate "he wishes to know dharma". SK 2732 and SK 2733 take it away again after particular upasargas. SK 2734 states the general principle: a san-final takes the ātmanepada just as its base root does. Edh "to grow" takes the ātmanepada, so its desiderative does too: edidhiṣate.
Rule. After the san-finals of jñā "to know", śru "to hear", smṛ "to remember" and dṛś "to see", the ātmanepada comes. Bhaṭṭoji's gloss: "for these, when they end in san, as before". One of his commentators glosses "as before" (prāgvat): the ātmanepada.
Example.
jñā + san → the desiderative stem, built as in chapter 55 → jijñāsa
→ jijñāsa + śap + ta → ātmanepada by this rule → te for ta (SK 2233) → jijñāsate.
dharmaṃ jijñāsate, "he wishes to know dharma".
Likewise śuśrūṣate "he wishes to hear", susmūrṣate "he wishes to remember", didṛkṣate "he wishes to see".
Applies when / blocked by. Both of Bhaṭṭoji's commentators note that some of these roots already take the ātmanepada in particular senses. Jñā takes it by SK 2717–2719. Śru and dṛś take it by a vārttika under SK 2701, earlier in this chapter. In those senses, they say, their san-finals already take the ātmanepada by SK 2734.
So why name them here? Their answer: the rule gives the ātmanepada to their san-finals in every other sense too. A desiderative of jñā takes it even where there is no denial and no intransitive use.
Rule. The ātmanepada does not come after the san-final of jñā when anu precedes it. "San-final" is carried forward from SK 2731.
Example.
anu + jñā + san → anujijñāsa → no ātmanepada (this rule) → parasmaipada → anujijñāsati.
putram anujijñāsati, "he wishes to give his son leave". One of Bhaṭṭoji's commentators glosses it: "he wishes to permit".
Applies when / blocked by. What exactly does this rule cancel? Bhaṭṭoji: only the preceding sūtra, SK 2731. That is by the maxim anantarasya: "a prescription or a prohibition applies to what immediately precedes it".
So the prohibition does not reach this case:
sarpiṣo 'nujijñāsate "he wishes to set about [it] by means of ghee".
Why the ātmanepada? Bhaṭṭoji: SK 2718 gives the ātmanepada to the root jñā itself, before any san, when it has no object, as in sarpiṣo 'nujānīte. So its san-final gets it by SK 2734, "as its base does". This rule does not touch SK 2734.
Watch out. Both forms have anu + jijñāsa. The difference is which rule gave the ātmanepada. With an object (putram), only SK 2731 could give it, and this rule cancels SK 2731. Without an object, SK 2734 gives it, and nothing cancels that.
Rule. After the san-final of śru "to hear", preceded by prati or āṅ, the ātmanepada does not come. Bhaṭṭoji's gloss: "after these two, from śru ending in san, what was stated does not occur". One of his commentators confirms the "not", glossing it "no ātmanepada". "San-final" comes from SK 2731, and "not" from SK 2732. Both commentators carry the same "not" on past this rule, at SK 2734.
Example.
prati + śru + san → pratiśuśrūṣa → no ātmanepada (this rule) → parasmaipada → pratiśuśrūṣati.
Likewise āśuśrūṣati.
Applies when / blocked by. Bhaṭṭoji adds: "After prati used as a karmapravacanīya, it does occur." A karmapravacanīya is a word like prati used to mark a relation, not as an upasarga of the verb. You met it in chapter 16 (SK 552).
devadattaṃ pratiśuśrūṣate, with prati governing devadattam.
One of Bhaṭṭoji's commentators explains. Here prati is a karmapravacanīya, by SK 552. In this rule, by interpretation, prati and āṅ are taken only as upasargas. So the ātmanepada of SK 2731 stands.
Bhaṭṭoji then names two sūtras as already explained: śadeḥ śitaḥ (SK 2362) and mriyater luṅliṅoś ca (SK 2538). In the Aṣṭādhyāyī they stand next in the run, so he only points to them. You met the first in chapter 44 and the second in chapter 49.
Rule. After a san-final root, the ātmanepada comes just as it comes after the root that stands before san. Bhaṭṭoji's gloss: "as with the root that stands before san, so the ātmanepada comes also after the san-final".
One of his commentators analyses pūrvavat: "like the prior one", where "the prior one" is the base (prakṛti) of san.
Example.
edh + san → edidhiṣa (with iṭ, chapter 55)
→ edh takes the ātmanepada (SK 2157), so the san-final does too (this rule)
→ edidhiṣa + śap + te → edidhiṣate "he wishes to grow".
Bhaṭṭoji's other examples:
śiśayiṣate "he wishes to lie down", from śī;nivivikṣate "he wishes to enter", from ni + viś. Viś takes the ātmanepada after ni by SK 2683, earlier in this chapter. Both of Bhaṭṭoji's commentators say so.Applies when / blocked by. Why "as its base"? Bhaṭṭoji: bubhūṣati "he wishes to be". Bhū takes the parasmaipada, so its desiderative does too.
Then comes a point about SK 2362 and SK 2538. Śad takes the ātmanepada before a śit affix. Mṛ "to die" takes it in luṅ and liṅ. Would their san-finals inherit it by this rule? Bhaṭṭoji says no. Two words are carried into those two sūtras: sanaḥ "after a san-final" from this rule, and na "not" from SK 2732. Each sūtra is then read by vākyabheda, "splitting the statement" into two:
So the desideratives śiśatsati and mumūrṣati take the parasmaipada.
Both of Bhaṭṭoji's commentators name the sources of the two words. One spells out the two statements.
Last, Bhaṭṭoji names āmpratyayavat kṛño 'nuprayogasya (SK 2240), taught in chapter 44. It is the next sūtra in the Aṣṭādhyāyī. His example: edhāṃcakre "he grew". The auxiliary kṛ takes the ātmanepada because edh does.
Watch out. "As its base" means the base's own rule, with its own conditions. nivivikṣate takes the ātmanepada only because ni is there, as SK 2683 requires. Without ni, viś is parasmaipada, and so is its desiderative.
pra + yuj + te → prayuṅkte; bhuj + te → odanaṃ bhuṅkteThree roots take the ātmanepada in this group. Yuj takes it after pra or upa: prayuṅkte. A vārttika widens the list of upasargas. Kṣṇu "to sharpen" takes it after sam. Bhuj of the rudh class takes it in every sense but "to protect": odanaṃ bhuṅkte "he eats rice". But mahīṃ bhunakti "he protects the earth".
Rule. After yuj preceded by pra or upa, the ātmanepada comes. The exception: where sacrificial vessels (yajñapātra) are concerned.
Example.
pra + yuj + laṭ, ātmanepada by this rule, with the rudh-class sign (chapter 50) → prayuṅkte.
Likewise upayuṅkte.
Vārttika. Bhaṭṭoji then cites a vārttika: "It should be stated: after an upasarga that has a vowel at its beginning or its end." Pra and upa both qualify, but so do others:
udyuṅkte: ud begins with a vowel;niyuṅkte: ni ends with a vowel.One of Bhaṭṭoji's commentators analyses svarādyanta as a compound of "beginning" and "end". Bhaṭṭoji's two examples show that one vowel, at either place, is enough.
Applies when / blocked by. Why "except with sacrificial vessels"? Bhaṭṭoji: dvandvaṃ nyañci pātrāṇi prayunakti, "he sets out the vessels in pairs, mouths downward". One of his commentators glosses dvandvam as "two by two" and nyañci as "with their openings downward". The parasmaipada stays.
Rule. After kṣṇu "to sharpen" preceded by sam, the ātmanepada comes.
Example.
sam + kṣṇu + laṭ → ātmanepada by this rule → saṃkṣṇute.
saṃkṣṇute śastram, "he sharpens the weapon". One of Bhaṭṭoji's commentators glosses it "he makes sharp" (tīkṣṇīkaroti).
Applies when / blocked by. Why a separate sūtra? Kṣṇu could have been added to SK 2699, samo gamyṛcchibhyām, earlier in this chapter. Both of Bhaṭṭoji's commentators explain. In SK 2699 "intransitive" is carried forward, so it would have denied the ātmanepada when kṣṇu has an object. As its own sūtra, kṣṇu takes the ātmanepada even with an object. Bhaṭṭoji's example has one, śastram, as a hint.
Rule. After bhuj of the rudh class, the ātmanepada comes in any sense other than protecting (avana). One of Bhaṭṭoji's commentators glosses avana as rakṣaṇa, "protecting"; the other as pālana, the same.
Example.
bhuj + laṭ, ātmanepada by this rule, with the rudh-class sign (chapter 50) → bhuṅkte.
odanaṃ bhuṅkte, "he eats rice". Bhaṭṭoji's gloss: "he consumes" (abhyavaharati).
Bhaṭṭoji gives two more lines:
"Here the meaning of bhuj is experiencing (upabhoga)," he says.
Applies when / blocked by. Why "other than protecting"? Bhaṭṭoji: mahīṃ bhunakti, "he protects the earth". The parasmaipada stays.
Which bhuj is meant? Both of Bhaṭṭoji's commentators say: only the rudh-class root of protecting and eating. The tud-class bhuj "to curl" keeps the parasmaipada. One gives bhujati vāsaḥ "the cloth curls" and traces the point to the Bhāṣya. The other gives pāṇiṃ vibhujati.
Watch out. The sūtra says "not protecting", not "eating". One of Bhaṭṭoji's commentators gives two reasons. A word that names one of a root's senses fixes which root is meant, whether it includes or excludes. And the ātmanepada must also reach senses other than eating, such as experiencing. The other commentator adds that bubhuje pṛthivīm cannot mean "he ate the earth", yet it takes the ātmanepada.
paśyanti bhavaṃ bhaktāḥ → paśyati bhavaḥ → (from that root, ṇic) darśayanti bhavaṃ bhaktāḥ → darśayate bhavaḥThis group is about roots that end in ṇi, the affix of the causative (chapter 54). After such a root, SK 2564 already gives the ātmanepada when the fruit goes to the agent. SK 2738 gives it in a special case, even when the fruit goes to someone else. The case is this: the object of the plain verb has become the agent of the causative, and the action is unchanged. Bhaṭṭoji teaches it through four stages, ending in darśayate bhavaḥ "Bhava comes into view". SK 2739 adds gṛdh and vañc in the sense of deceiving. SK 2740 adds kṛñ with mithyā "wrongly", in the sense of repetition. Bhaṭṭoji then recalls SK 2158, the ātmanepada when the fruit goes to the agent, which leads into the last group.
Rule. After a root ending in ṇi (a ṇyanta), the ātmanepada comes when two conditions hold:
1. the action expressed without ṇi is that very action, and no more, that the ṇi-form expresses;
2. what was the object (karman) without ṇi is the agent (kartṛ) of the ṇi-form.
It does not come in the sense of ādhyāna, "longing recollection".
"Ātmanepada" is the heading, carried forward from SK 2157. Aṇau means "in the non-ṇi state", the plain verb. Ṇau means "in the ṇi state", the causative.
How the sūtra is read. The word karma appears once but does two jobs. Both of Bhaṭṭoji's commentators break the sūtra into clauses:
Bhaṭṭoji's gloss carries both senses: "if the action in the non-ṇi state is that very one expressed by the ṇi-final; if what is the object-kāraka in the non-ṇi state is the agent in the ṇi state".
What of anādhyāne? Both commentators start from four clauses and settle on three. Anādhyāne is a paryudāsa, a negation that picks out "something other than" ādhyāna. It joins the first clause. One of them says this is as the Bhāṣya has it. He adds that Bhaṭṭoji's "but not in ādhyāna" only states the resulting sense.
Why the rule is needed. Bhaṭṭoji: SK 2564, ṇicaś ca, already gives the ātmanepada after a ṇic-final. But it does so only when the fruit goes to the agent (kartrabhiprāya). This rule is for the other case, akartrabhiprāya, "the fruit not intended for the agent".
When the fruit does go to the agent, two later rules also reach the form. SK 2744 makes the ātmanepada optional. SK 2754, aṇāv akarmakāt…, taught in the next chapter, gives the parasmaipada. Each is later than this rule, so each would normally win. Bhaṭṭoji says this rule is what is wanted, by pūrvavipratiṣedha: the earlier rule wins, as at SK 2730. Whether that claim holds against both later rules, scholars discuss this.
Which roots. Bhaṭṭoji: "The examples are roots whose bhāva resides in the agent and roots whose kriyā resides in the agent." Here bhāva and kriyā are a narrower pair than the bhāva of SK 2679, the bare action, and the "action" of condition 1. One of his commentators defines the two. A bhāva is a root-sense achieved without bodily movement, such as seeing or hearing. A kriyā is achieved with movement, such as cooking. Bhaṭṭoji's roots may be of either kind, provided the root-sense resides in the agent (kartṛstha). His two examples are seeing (paśyati) and mounting (ārohati).
Example: the four stages of darśayate. Bhaṭṭoji builds the form through four sentences. He calls each a stage (avasthā, kakṣā) and refers to them by number. One of his commentators compares them to the successive courts of a palace.
Stage 1: paśyanti bhavaṃ bhaktāḥ "the devotees see Bhava".
Bhaṭṭoji's gloss: "they make him an object of visual cognition". So "seeing" is taken to contain an impelling part (preraṇāṃśa): the devotees make Bhava an object.
Stage 2: the impelling part is given up → paśyati bhavaḥ.
Bhaṭṭoji's gloss: "he becomes an object [of sight]". Bhava is now the agent of the plain verb.
Stage 3: from that root comes the hetumat ṇic (SK 2576) → darśayanti bhavaṃ bhaktāḥ.
Bhaṭṭoji's gloss: "they see him".
Stage 4: the sense of ṇi is again not intended → darśayate bhavaḥ "Bhava comes into view".
Read Stage 3 carefully. Bhaṭṭoji's word is tataḥ. It does not mean "then". It means "from that": from the root as it stands in Stage 2, whose impelling part has been given up. Both of Bhaṭṭoji's commentators read it so. The root now means "become an object of sight". The ṇic adds an impeller back: "the devotees make Bhava become an object of sight". That is why Stage 3 means the same as Stage 1.
So the four stages pair up. Stages 1 and 3 mean the same. Stages 2 and 4 mean the same. Now test each ṇi-form against the two conditions.
Stage 3, darśayanti: the action is the same as in Stage 1. But Bhava, the object of Stage 1, is still the object, not the agent. The agent is the devotees. Condition 2 fails → no ātmanepada.
Stage 4, darśayate: compare it with Stage 2, the plain verb. The action is the same, "Bhava becomes an object of sight". And Bhava, the object of Stage 1, is the agent here. Both conditions hold → ātmanepada by this rule.
The form itself:
dṛś + ṇic → guṇa of the light penultimate (SK 2189) → darśi
→ darśi + śap + ta, ātmanepada by this rule
→ guṇa of the i before śap (SK 2168) → darśe + a + ta
→ ay for e (SK 61) → darśay + a + ta
→ te for ta (SK 2233) → darśayate.
The same with ārohayate. Bhaṭṭoji: "In the same way, ārohayate hastī is an example."
ārohanti hastinaṃ hastipakāḥ "the elephant-keepers mount the elephant". Gloss: "they make it go low".ārohati hastī. Gloss: "it goes low".ārohayanti hastinaṃ hastipakāḥ. Gloss: "they mount it".ārohayate hastī. Gloss: "it goes low".One of Bhaṭṭoji's commentators explains the picture. To mount, the keepers make the tall elephant go low, by goad-strokes and the like. The printed text gives only the gloss at Stage 3. The form ārohayanti is restored from that commentator's quotation of it, and from another edition of the text. He says hastinaṃ hastipakāḥ is to be understood with it.
A second view. Bhaṭṭoji then offers another account, introduced by yadvā, "or else". One of his commentators calls it the view of "superimposed impelling".
paśyanti bhavaṃ bhaktāḥ, ārohanti hastinaṃ hastipakāḥ.darśayati bhavaḥ, ārohayati hastī. Gloss: "Bhava impels the devotees who are seeing; the elephant impels the keepers who are mounting."darśayate, ārohayate. The meaning is as before.On this view the second stage gets no ātmanepada. Bhaṭṭoji: the sense of ṇic is in excess, so the action is not the same. Both commentators note what follows. Here "that very action" rules out only more, not less. One gives the reason: otherwise the third stage, which has lost an impelling, would fail too.
Which of the two views is the better, scholars discuss this. One of Bhaṭṭoji's commentators favours the second, on the Bhāṣya's example darśayate bhṛtyān rājā.
Applies when / blocked by. Why "not in longing recollection"? Bhaṭṭoji:
smarati vanagulmaṃ kokilaḥ "the cuckoo remembers the forest thicket"
→ smarayati vanagulmaḥ, with the parasmaipada.
His gloss: "it becomes an object in remembrance preceded by longing". The object of the plain verb has become the agent, but the sense is ādhyāna. So this rule does not apply.
Both commentators place this form in the stages: the fourth on the first view, the third on the second. One adds why it has a short a. Smṛ in the sense of ādhyāna is a mit root, so its vowel is shortened before ṇi (SK 2568, chapter 53).
Last, Bhaṭṭoji names bhīsmyor hetubhaye (SK 2594) as already explained. You met it in chapter 54.
Watch out. Two traps.
Rule. After gṛdh and vañc ending in ṇi, the ātmanepada comes in the sense of deceiving (pralambhana). "Ending in ṇi" is carried forward from SK 2738.
Bhaṭṭoji's gloss: "in the sense of cheating (pratāraṇa), after these two when they end in ṇi, as before". One of his commentators says "as before" means the ātmanepada. He adds that roots have many meanings, so these two can mean "cheat".
Example.
gṛdh + ṇic → guṇa (SK 2189) → gardhi → gardhi + śap + ta, ātmanepada by this rule → gardhayate (as for darśayate).
māṇavakaṃ gardhayate or vañcayate, "he deceives the boy".
Applies when / blocked by. Why "in the sense of deceiving"? Bhaṭṭoji gives two counter-examples with the parasmaipada:
One of Bhaṭṭoji's commentators says the rule is for the case where the fruit does not go to the agent.
Bhaṭṭoji then names liyaḥ saṃmānanaśālīnīkaraṇayoś ca (SK 2592) as already explained, in chapter 54.
Rule. After kṛñ ending in ṇi, with the word mithyā "wrongly" as its upapada, the ātmanepada comes in the sense of repetition (abhyāsa). An upapada here is a word standing beside the verb in the sentence. Bhaṭṭoji says "only after one ending in ṇi", so that word still carries forward from SK 2738. One of his commentators says it stops after this rule.
Example.
kṛ + ṇic → vṛddhi before the ṇit affix (SK 254) → kāri → kāri + śap + ta, ātmanepada by this rule → kārayate (as for darśayate).
padaṃ mithyā kārayate. Bhaṭṭoji's gloss: "he pronounces the word, faulty in accent and so on, again and again".
One of Bhaṭṭoji's commentators explains the sense. Here kṛ means "to come out", as a sound comes out. That verb has no object. Its ṇi-form, "to make come out", has one: padam. So kārayati means "pronounces".
Applies when / blocked by. Bhaṭṭoji gives two counter-examples:
Why is the verb not doubled, as repeated actions often are? The same commentator answers. The ātmanepada itself marks the repetition. So there is no doubling by nityavīpsayoḥ (SK 2140, chapter 43).
SK 2158 recalled. Bhaṭṭoji then quotes svaritañitaḥ kartrabhiprāye kriyāphale (SK 2158), from chapter 44. A root with a svarita vowel or the marker ñ takes the ātmanepada when the fruit of the action goes to the agent. His examples: yajate "he sacrifices [for himself]", sunute "he presses [the soma for himself]".
Why "when the fruit goes to the agent"? ṛtvijo yajanti, sunvanti, "the priests sacrifice, press [the soma, for the patron]". One of Bhaṭṭoji's commentators explains. The priests' fee is not the fruit of the sacrifice. The fruit is heaven, and it goes to the patron.
Watch out. Abhyāsa here means "repetition" of an action. It is not the abhyāsa of chapter 44, the first of the two copies made by doubling (SK 2178). Same word, different meaning.
apa + vad + te → nyāyam apavadate (kartrabhiprāya) / apavadati; svaṃ yajñaṃ yaj + ti / te → yajati / yajate (2744)The last four rules turn on kartrabhiprāya: the fruit of the action is intended for the agent. "When the fruit goes to the agent" is carried forward from SK 2158, which Bhaṭṭoji has just recalled. Three rules give the ātmanepada in that case. Vad with apa: nyāyam apavadate "he speaks against the rule". Yam with sam, ud or āṅ: bhāram udyacchate "he lifts the load". Bare jñā with an object: gāṃ jānīte "he knows the cow". When the fruit goes to someone else, the parasmaipada stays: apavadati. The last rule, SK 2744, makes the ātmanepada of five such rules optional. That happens when a word in the sentence already shows the fruit is the agent's: svaṃ yajñaṃ yajati or yajate "he performs his own sacrifice". The group's title uses taṅ again, the short name for the ātmanepada endings. Its upapada is a word uttered nearby, such as svam.
Rule. After vad preceded by apa, the ātmanepada comes when the fruit goes to the agent. Bhaṭṭoji: "only when the fruit is intended for the agent" (kartrabhiprāye ity eva). That condition is carried forward from SK 2158.
Example.
apa + vad + śap + ta, fruit for the agent → ātmanepada by this rule → apavadate.
nyāyam apavadate, "he speaks against the rule". One of Bhaṭṭoji's commentators glosses it: "he rejects it by an explicit statement".
Applies when / blocked by. When the fruit is not the agent's, the parasmaipada stays: apavadati.
Bhaṭṭoji then names ṇicaś ca (SK 2564), from chapter 53, with its example kārayate. It is the next sūtra in the Aṣṭādhyāyī.
Rule. After yam preceded by sam, ud or āṅ, the ātmanepada comes. Two conditions: the domain is not a book (grantha), and the fruit goes to the agent. Bhaṭṭoji first fixes the word division: a-granthe, "not a book".
Example.
sam + yam + śap + ta → ātmanepada by this rule
→ ch for the m of yam before śap (SK 2400), with the augment t before ch (SK 146), as at SK 2729 → yacch
→ te for ta (SK 2233) → saṃyacchate.
One of Bhaṭṭoji's commentators glosses them: "gathers", "lifts up", "fastens on the waist and so on".
Applies when / blocked by. Why "not a book"? Bhaṭṭoji: udyacchati vedam, "he makes an effort to master the Veda". The parasmaipada stays.
Bhaṭṭoji adds: "only when the fruit is intended for the agent". He gives no counter-example. One of his commentators supplies one: vrīhīn saṃyacchati, "he gathers the rice for another".
Watch out. Why name āṅ here? SK 2695, āṅo yamahanaḥ, earlier in this chapter, already gives the ātmanepada after āṅ + yam. SK 2695 covers the verb with no object, or with one's own limb as object. Both of Bhaṭṭoji's commentators answer: āṅ is named here for the verb with other objects. Bhaṭṭoji's example has one, vastram, as a hint.
Rule. After jñā with no upasarga, the ātmanepada comes. The fruit must go to the agent.
Bhaṭṭoji does not repeat "only when the fruit is intended for the agent" here. But the condition holds. One of his commentators counts this rule among the five that SK 2744 makes optional. And Bhaṭṭoji's own example there, svāṃ gāṃ jānāti or jānīte, presupposes it.
Example.
jñā + laṭ, no upasarga, fruit for the agent → ātmanepada by this rule → jānīte.
gāṃ jānīte, "he knows the cow".
Applies when / blocked by. Why "with no upasarga"? Bhaṭṭoji: svargaṃ lokaṃ na prajānāti, "he does not know the heavenly world". With pra, the parasmaipada stays.
Bhaṭṭoji raises a difficulty. The poet Bhaṭṭi writes: itthaṃ nṛpaḥ pūrvam avāluloce tato 'nujajñe gamanaṃ sutasya, "thus the king first deliberated; then the son's departure was permitted". How can anujajñe take the ātmanepada? One of Bhaṭṭoji's commentators spells out the difficulty. It has an upasarga, anu, so this rule cannot apply. It is transitive, so SK 2718 cannot apply either.
Bhaṭṭoji answers: the liṭ is passive. The same commentator names the rule: SK 2679, bhāvakarmaṇoḥ, the first rule of this chapter. The passive takes the ātmanepada. And "by the king" (nṛpeṇa) is supplied by vipariṇāma, change of case. The nṛpaḥ of the first half is carried into the second as nṛpeṇa.
Watch out. SK 2718 and this rule divide the work on jñā between them. Both of Bhaṭṭoji's commentators say so. SK 2718 already covers the intransitive use. So this rule does its work on the transitive one, which is why its example has an object, gām. In turn, SK 2718 does its own work on forms with an upasarga.
Rule. The ātmanepada of five rules becomes optional when a word in the sentence makes evident that the fruit goes to the agent. Bhaṭṭoji's gloss: "the ātmanepada enjoined by the group of five sūtras beginning with SK 2158 comes optionally, when the fruit's going to the agent is shown by a word uttered nearby".
One of Bhaṭṭoji's commentators names the five:
Here upapada means just "a word uttered nearby". The same commentator says it is not the technical upapada of SK 781 (chapter 18), which cannot apply here.
Example.
svaṃ yajñaṃ yaj + laṭ: svam "his own" shows the fruit is the agent's
→ the ātmanepada of SK 2158 becomes optional (this rule)
→ svaṃ yajñaṃ yajati or yajate "he performs his own sacrifice".
Bhaṭṭoji's six pairs:
svaṃ yajñaṃ yajati / yajate "he performs his own sacrifice" (SK 2158);svaṃ kaṭaṃ karoti / kurute "he makes his own mat" (kṛñ is ñit, so SK 2158);svaṃ putram apavadati / apavadate "he speaks against his own son" (SK 2741);svaṃ yajñaṃ kārayati / kārayate "he has his own sacrifice performed" (SK 2564);svaṃ vrīhiṃ saṃyacchati / saṃyacchate "he gathers his own rice" (SK 2742);svāṃ gāṃ jānāti / jānīte "he knows his own cow" (SK 2743).One of Bhaṭṭoji's commentators matches four of these pairs to their rules. The other two follow from the roots: kṛñ is a ñit root, and apa + vad is SK 2741's.
Applies when / blocked by. What kind of option is this? One of Bhaṭṭoji's commentators, following the Bhāṣya on SK 24, calls it a prāptavibhāṣā: the ātmanepada was already due, obligatorily, by the five rules. The other calls it an aprāptavibhāṣā. Which is right, scholars discuss this.
That second commentator also notes that sva "own" is only an instance. Svārthaṃ yajñaṃ yajati "he performs the sacrifice for his own sake" works the same way.
Watch out. A ṇi-final such as darśayate is a special case. Bhaṭṭoji said at SK 2738 that SK 2738 prevails over this option, by pūrvavipratiṣedha.
With SK 2744 the ātmanepada section closes. Bhaṭṭoji's closing words are ity ātmanepadaprakriyā, "here ends the working (prakriyā) of the ātmanepada". Look back at what it has taught. SK 2157 and SK 2158 in chapter 44 gave the ātmanepada by a root's markers. This chapter gave it by sense, by upasarga, by the presence or absence of an object, and by who gets the fruit. Whatever is left over takes the parasmaipada, by SK 2159. Chapter 62 turns to the parasmaipada. It opens with anuparābhyāṃ kṛñaḥ (SK 2745).
sarpiṣo nāthate (SK 2687), SK 2710, SK 2712.samasthita, āhata.kṛṣṇāya śapate (SK 2688), gopī kṛṣṇāya tiṣṭhate (SK 2690).nī + śap + te → nayate.darśayate (SK 2738).babhūve (SK 2679).anubabhūve, SK 2679) or the agent (SK 2680).vyatilunīte.māṇavakam anukrīḍati (SK 549, SK 2687).apaskirate (SK 2688), upaskurute (SK 2705).ājaghne against SK 2695.āhadhvam (SK 2695), upāyaṃsta (SK 2729), anujajñe (SK 2743).saṃśṛṇumahe is called a slip under SK 2701.samuhya (SK 2702).avadhiṣṭa.saṃgasīṣṭa, samuhyāt.saṃvidrate (SK 2701).edidhiṣa (chapter 55; SK 2734).| SK | what it does | example |
|---|---|---|
| 2679 | ātmanepada when the lakāra denotes bhāva or karman | babhūve, anubabhūve |
| 2680 | ātmanepada for the agent in exchange of action | vyatilunīte, vyatiste |
| 2681 | not after roots meaning going or injuring; has etc. added, hṛ freed (vārttikas) | vyatigacchanti; saṃpraharante |
| 2682 | not with itaretara, anyonya (and paraspara) as upapada | parasparasya vyatilunanti |
| 2683 | viś after ni | niviśate / praviśati |
| 2684 | krī after pari, vi, ava, fruit not the agent's | vikrīṇīte |
| 2685 | ji after vi, parā | vijayate, parājayate |
| 2686 | dā after āṅ, except opening the mouth | vidyām ādatte / mukhaṃ vyādadāti |
| 2687 | krīḍ after anu, sam, pari, āṅ; vārttikas for āgami, śikṣ, nāth, hṛ, kṝ | saṃkrīḍate; sarpiṣo nāthate |
| 2688 | suṭ for kṝ after apa in scratching; vārttikas for nu, pracch, śap | apaskirate vṛṣo hṛṣṭaḥ; kṛṣṇāya śapate |
| 2689 | sthā after sam, ava, pra, vi | santiṣṭhate; luṅ samasthita |
| 2690 | sthā in making one's mind known or taking as arbiter | gopī kṛṣṇāya tiṣṭhate |
| 2691 | sthā after ud, not upward, in purposeful activity | muktāv uttiṣṭhate / pīṭhād uttiṣṭhati |
| 2692 | sthā after upa, a mantra the instrument; worship, joining, befriending, road | ādityam upatiṣṭhate |
| 2693 | upa-sthā without an object | bhojanakāle upatiṣṭhate |
| 2694 | tap after ud, vi, without an object or with one's own limb | uttapate, uttapate pāṇim |
| 2695 | yam, han after āṅ, without an object or with one's own limb | āyacchate, āhate |
| 2696 | vadha for han in luṅ before the ātmanepada, optionally | avadhiṣṭa |
| 2697 | sic after han is kit | āhata |
| 2698 | sic after yam is kit in the sense of gandhana | udāyata / udāyaṃsta pādam |
| 2699 | gam, ṛcch after sam, without an object | saṃgacchate / grāmaṃ saṃgacchati |
| 2700 | jhal-initial liṅ and sic after gam optionally kit; vārttika vid, pracch, svṛ | samagata / samagaṃsta; saṃvitte |
| 2701 | ruṭ to the at for jh after vett, optionally; vārttikas ṛ, śru, dṛś; as, ūh | saṃvidrate / saṃvidate; nirasyate |
| 2702 | short u for the ū of ūh after an upasarga, before a y-initial kit/ṅit | samuhyāt, samuhya |
| 2703 | hve after ni, sam, upa, vi | nihvayate |
| 2704 | hve after āṅ in rivalry | cāṇūram āhvayate / putram āhvayati |
| 2705 | kṛ in seven senses, even for another's fruit | utkurute, harim upakurute |
| 2706 | kṛ after adhi in forbearing or overpowering | śatrum adhikurute |
| 2707 | kṛ after vi with a sound as object | svarān vikurute |
| 2708 | kṛ after vi without an object | chātrā vikurvate |
| 2709 | nī in seven senses, even for another's fruit | śāstre nayate, karaṃ vinayate |
| 2710 | niyama: nī with an object in the agent only if not a bodily part | krodhaṃ vinayate / gaḍuṃ vinayati |
| 2711 | kram in unimpeded course, zeal, flourishing | ṛci kramate buddhiḥ |
| 2712 | niyama: in those senses, with an upasarga only after upa, parā | parākramate / saṃkrāmati |
| 2713 | kram after āṅ in the rising of a light | ākramate sūryaḥ |
| 2714 | kram after vi in placing the feet | sādhu vikramate vājī |
| 2715 | kram after pra, upa meaning "begin" | prakramate / prakrāmati |
| 2716 | kram without an upasarga, optionally | kramate / krāmati |
| 2717 | Ātmanepada after jñā in the sense of denial | śatam apajānīte |
| 2718 | Ātmanepada after intransitive jñā (work: with an upasarga) | sarpiṣo jānīte, sarpiṣo 'nujānīte |
| 2719 | Ātmanepada after jñā with sam or prati, not in longing recollection | śataṃ saṃjānīte, pratijānīte; but mātaraṃ saṃjānāti |
| 2720 | Ātmanepada after vad in six senses | śāstre vadate, bhṛtyān upavadate, kṣetre vivadante |
| 2721 | Ātmanepada after vad when articulate speakers speak together | saṃpravadante brāhmaṇāḥ; but saṃpravadanti khagāḥ |
| 2722 | Ātmanepada after intransitive anu + vad, of articulate speakers | anuvadate kaṭhaḥ kalāpasya; but uktam anuvadati |
| 2723 | Optional ātmanepada after vad in contradictory talk | vipravadante / vipravadanti vaidyāḥ |
| 2724 | Ātmanepada after tud-class gṝ with ava | avagirate |
| 2725 | Ātmanepada after gṝ with sam, in affirming | śabdaṃ nityaṃ saṃgirate; but saṃgirati grāsam |
| 2726 | Ātmanepada after transitive ud + car | dharmam uccarate; but bāṣpam uccarati |
| 2727 | Ātmanepada after sam + car joined to an uttered third-case word | rathena saṃcarate |
| 2728 | Ātmanepada after sam + dāṇ with a third case meaning the fourth | dāsyā saṃyacchate, dāsyā saṃprayacchate |
| 2729 | Ātmanepada after upa + yam in taking as one's own | bhāryām upayacchate |
| 2730 | Sic after yam optionally kit in marrying (obligatory in a forcible one) | upāyata / upāyaṃsta |
| 2731 | Ātmanepada after the san-finals of jñā, śru, smṛ, dṛś | dharmaṃ jijñāsate, śuśrūṣate, didṛkṣate |
| 2732 | Not after the san-final of jñā with anu (cancels SK 2731 only) | putram anujijñāsati; but sarpiṣo 'nujijñāsate |
| 2733 | Not after the san-final of śru with prati or āṅ (as upasargas) | pratiśuśrūṣati; but devadattaṃ pratiśuśrūṣate |
| 2734 | A san-final takes the ātmanepada as its base does | edidhiṣate, nivivikṣate; but bubhūṣati, śiśatsati |
| 2735 | Ātmanepada after yuj with pra, upa, not of sacrificial vessels; vārttika: any vowel-initial or vowel-final upasarga | prayuṅkte, udyuṅkte; but pātrāṇi prayunakti |
| 2736 | Ātmanepada after sam + kṣṇu, even with an object | saṃkṣṇute śastram |
| 2737 | Ātmanepada after rudh-class bhuj except in protecting | odanaṃ bhuṅkte; but mahīṃ bhunakti |
| 2738 | Ātmanepada after a ṇi-final when the plain verb's object is its agent and the action is the same; not in longing recollection | darśayate bhavaḥ, ārohayate hastī; but darśayanti bhavaṃ bhaktāḥ, smarayati vanagulmaḥ |
| 2739 | Ātmanepada after ṇi-final gṛdh, vañc in deceiving | māṇavakaṃ gardhayate / vañcayate; but śvānaṃ gardhayati |
| 2740 | Ātmanepada after ṇi-final kṛñ with mithyā, in repetition | padaṃ mithyā kārayate; but sakṛt padaṃ mithyā kārayati |
| 2741 | Ātmanepada after apa + vad when the fruit is the agent's | nyāyam apavadate; otherwise apavadati |
| 2742 | Ātmanepada after yam with sam, ud, āṅ, not of a book, fruit the agent's | bhāram udyacchate, vastram āyacchate; but udyacchati vedam |
| 2743 | Ātmanepada after jñā without upasarga (transitive), fruit the agent's | gāṃ jānīte; but svargaṃ lokaṃ na prajānāti |
| 2744 | The ātmanepada of SK 2158, 2741, 2564, 2742, 2743 optional when a nearby word shows the fruit is the agent's | svaṃ yajñaṃ yajati / yajate, svāṃ gāṃ jānāti / jānīte |