Chapter 61 taught the rules that give a root the ātmanepada endings. This chapter teaches eleven rules that do the opposite. Each one gives the parasmaipada endings where some ātmanepada rule would otherwise have applied. After this chapter you will know why it is anukaroti "he imitates", not anukurute, though kṛñ is marked with ñ. You will know why viramati "he stops" takes the parasmaipada, though ram takes the ātmanepada on its own. You will know eight ṇic-final roots that take the parasmaipada against ṇicaś ca (SK 2564): bodhayati padmam. Roots meaning swallowing or moving do too (nigārayati, calayati), and so does an intransitive root with a sentient agent (śāyayati). Last, you will meet the counter-rule that gives nine ṇic-final roots the ātmanepada back: pāyayate, damayate, vāsayate. With it Bhaṭṭoji closes the whole account of the two padas. The chapter has eleven sūtras, SK 2745 to SK 2755.
ādayati annaṃ baṭunā "he has the food eaten by the boy".lohitāyati / lohitāyate.ārohayate hastī "the elephant goes low", which has no object.anu + kṛñ + laṭ → (ñit, SK 2158 would give anukurute) → anukarotiFour roots, each after a stated upasarga, take the parasmaipada. Each root is marked with ñ or with a svarita vowel. So SK 2158 would give it the ātmanepada whenever the fruit of the action goes to the agent. These rules give the parasmaipada even then. The running example is kṛñ "to do" after anu: anukaroti "he imitates".
Rule. After kṛñ preceded by anu or parā, the parasmaipada comes, in the sense of the agent. Bhaṭṭoji gives the purpose: the parasmaipada is secured "even when the fruit goes to the agent, and in the senses gandhana and the rest".
Example.
anu + kṛñ + laṭ, the fruit going to the agent
→ kṛñ is marked with ñ, so SK 2158 would give the ātmanepada → anukurute
→ this rule: the parasmaipada, the ending tip (heard as ti) → anukaroti "he imitates".
Likewise parā + kṛñ → parākaroti "he rejects".
What the section assumes. Bhaṭṭoji opens the section by recalling SK 2159, śeṣāt kartari parasmaipadam, with the example atti "he eats". That is the default. Each rule from SK 2745 to SK 2754 restricts a root to it; SK 2755 then takes some of that back.
None of these sūtras states the word parasmaipada. It is supplied (śeṣa, "what is left unstated"), as one of Bhaṭṭoji's commentators says under rule after rule. He does not name the rule it comes from. The natural source is SK 2159, the rule Bhaṭṭoji recalls at the head of the section.
The word kartari, "in the sense of the agent", is carried forward too. Bhaṭṭoji shows this here with kartari ity eva, below.
Applies when / blocked by. Why is the rule needed at all? One of Bhaṭṭoji's commentators raises the question. Where the fruit does not go to the agent, SK 2158 does not apply. Then SK 2159 already gives the parasmaipada. So the rule does its work in the other case: when the fruit goes to the agent. That is what Bhaṭṭoji's "even" (api) means.
The rule also overrides SK 2705, gandhanāvakṣepaṇa…, of chapter 61. That rule gives kṛñ the ātmanepada in certain senses. The same commentator lists them: gandhana, avakṣepaṇa, sevana, sāhasikya, pratiyatna, prakathana and upayoga. After anu or parā, kṛñ takes the parasmaipada in these senses too.
Only in the sense of the agent. Bhaṭṭoji adds kartari ity eva. The word kartari is carried forward into this rule. So the parasmaipada does not come for an l-affix that expresses the action itself (bhāva) or the object (karman). There the ātmanepada stands: anukriyate "it is imitated".
The object-agent. One case still needs thought. A karmakartṛ, "object-agent", is a word that is really the object of an action but is presented as its agent. One of Bhaṭṭoji's commentators gives the example anukriyate śabdaḥ svayam eva, "the sound gets imitated of itself". It is an agent, so would this rule give the parasmaipada? Bhaṭṭoji answers no. The ātmanepada comes there by SK 2766, karmavat karmaṇā tulyakriyaḥ, taught in chapter 64.
How does SK 2766 win? There are two ways of reading what that rule extends to the object-agent.
For the second view, Bhaṭṭoji reads this rule more narrowly. Two mentions of kartṛ are carried forward into it: one from kartari śap (SK 2167) and one from SK 2159. Together they mean "where the agent is purely an agent (kartā eva yaḥ kartā), and not an object-agent". Then this rule simply does not reach the object-agent. Which earlier rule lends the first mention, scholars discuss this.
Rule. After kṣip "to throw" preceded by abhi, prati or ati, the parasmaipada comes. Bhaṭṭoji cites the root as kṣipa preraṇe, "kṣip in the sense of throwing", and notes that it is svaritet: marked with a svarita vowel.
Example.
abhi + kṣip + laṭ, the fruit going to the agent
→ kṣip is svaritet, so SK 2158 would give the ātmanepada
→ this rule: the parasmaipada → abhikṣipati "he throws at".
One of Bhaṭṭoji's commentators adds pratikṣipati and atikṣipati.
Applies when / blocked by. Why does Bhaṭṭoji mention that kṣip is svaritet? Bhaṭṭoji's commentators agree on the answer. A svaritet root gets the ātmanepada by SK 2158 when the fruit goes to the agent. This rule is begun to give the parasmaipada even there.
Rule. After vah "to carry" preceded by pra, the parasmaipada comes.
Example.
pra + vah + laṭ, the fruit going to the agent
→ vah is svaritet, so SK 2158 would give the ātmanepada
→ this rule: the parasmaipada → pravahati "flows, carries forward".
Applies when / blocked by. Bhaṭṭoji gives only the form. One of Bhaṭṭoji's commentators supplies the reason, the same as at SK 2746: vah is svaritet, and the rule secures the parasmaipada even when the fruit goes to the agent.
Rule. After mṛṣ preceded by pari, the parasmaipada comes.
Example.
pari + mṛṣ (the div class, "to endure") + laṭ
→ this mṛṣ is svaritet, so both padas would be possible (Bhaṭṭoji's commentators)
→ this rule: the parasmaipada only → parimṛṣyati "he endures".
Applies when / blocked by. There is more than one root mṛṣ. Bhaṭṭoji names a second: "as for the bhū-class root, parimarṣati". One of Bhaṭṭoji's commentators explains that this mṛṣ is parasmaipadin anyway: it takes only the parasmaipada. So parimarṣati needs no help from this rule. Whether parimarṣati may also be traced to the cur-class mṛṣ, scholars discuss this.
A split of the rule. Bhaṭṭoji reports a view of "some" (kecit). They split the sūtra in two (yogavibhāga): first pareḥ "after pari", then mṛṣaḥ. Into the first part they carry vahaḥ forward from SK 2747. Then vah after pari takes the parasmaipada too. One of Bhaṭṭoji's commentators gives the form parivahati. He adds that this split is not found in the Bhāṣya, Patañjali's great commentary.
vi + ram + laṭ → (anudāttet, SK 2157 would give viramate) → viramatiThe root ram "to stop" is marked with an anudātta vowel. So SK 2157 makes it an ātmanepadin root, one that takes the ātmanepada. Three rules give it the parasmaipada after certain upasargas. After vi, āṅ and pari the parasmaipada is fixed; after upa it is fixed when ram has an object, and optional when it has none. The running example is viramati "he stops".
Rule. After ram preceded by vi, āṅ or pari, the parasmaipada comes.
Example.
vi + ram + laṭ
→ ram is anudāttet, so SK 2157 would give the ātmanepada → viramate
→ this rule: the parasmaipada → viramati "he stops".
One of Bhaṭṭoji's commentators adds āramati and pariramati.
Applies when / blocked by. The ātmanepada set aside here is the one of SK 2157 (ram is anudāttet), says one of Bhaṭṭoji's commentators.
Rule. After ram preceded by upa as well, the parasmaipada comes. "After ram" (ramaḥ) is carried forward from SK 2749.
Example.
Bhaṭṭoji's example has an object: yajñadattam uparamati "he makes Yajñadatta stop".
→ The meaning, Bhaṭṭoji says, is uparamayati, the causative.
→ The root has the sense of ṇic included within it (antarbhāvitaṇyartha). So it is transitive here, with no ṇic.
→ This rule: the parasmaipada → uparamati.
Applies when / blocked by. Why a transitive example? One of Bhaṭṭoji's commentators explains. Uparamati "he desists" has no object. For that sense the next rule makes the parasmaipada optional. So a sure example of this rule must have an object. How can ram, "to stop", take an object without ṇic? Because roots have many meanings, the same commentator says.
Watch out. Why not one rule, vyāṅparyupebhyaḥ, with upa in the list of SK 2749? Bhaṭṭoji's commentators agree: so that only upa is carried forward into SK 2751. The option of SK 2751 is for upa alone, not for vi, āṅ or pari.
Rule. After ram preceded by upa, when it is intransitive (akarmaka, "having no object"), the parasmaipada comes optionally. Bhaṭṭoji's gloss: "after ram preceded by upa, when intransitive, parasmaipada optionally".
Upa is carried forward from SK 2750 and ram from SK 2749. One of Bhaṭṭoji's commentators says that parasmaipada is carried forward too, and that Bhaṭṭoji's words upād rameḥ state this.
Example.
upa + ram + laṭ, with no object
→ this rule, optionally: the parasmaipada → uparamati
→ in the other option, SK 2157 stands: the ātmanepada → uparamate.
The meaning of both, Bhaṭṭoji says, is nivartate "he desists".
Watch out. Keep SK 2750 and SK 2751 apart. With an object, as in yajñadattam uparamati, the parasmaipada is fixed. Without one, both padas are correct.
budh + (hetumat) ṇic → bodhi- + laṭ → (ṇicaś ca, SK 2564 would give bodhayate) → bodhayati padmamA stem ending in ṇi takes the ātmanepada by ṇicaś ca (SK 2564) when the fruit goes to the agent. Three rules give certain ṇi-final stems the parasmaipada instead. SK 2752 names eight roots. SK 2753 takes roots by their meaning: swallowing or moving. SK 2754 takes roots by their construction: intransitive, with a sentient agent. The word ṇeḥ, "after what ends in ṇi", runs from SK 2752 through the rest of the chapter. The running example is bodhayati padmam "[the sun] makes the lotus open".
Rule. After budh, yudh, naś, jan, iṅ, pru, dru and sru, when they end in ṇi, the parasmaipada comes. Bhaṭṭoji calls the rule an exception (apavāda) to ṇicaś ca (SK 2564). An apavāda sets aside a more general rule in its own domain.
Example.
budh + ṇic (the causative, SK 2576) → guṇa of the light penultimate (SK 2189) → bodhi
→ bodhi + laṭ, the fruit going to the agent
→ ṇicaś ca (SK 2564) would give the ātmanepada → bodhayate
→ this rule: the parasmaipada → bodhayati padmam "[the sun] makes the lotus open".
One of Bhaṭṭoji's commentators supplies "the sun" as the agent. Budh here means opening: the lotus opens, and the sun makes it open.
Bhaṭṭoji's other examples:
yodhayati kāṣṭhāni "he makes the logs fight";nāśayati duḥkham "he destroys sorrow";janayati sukham "he produces happiness";adhyāpayati vedam "he teaches the Veda" (iṅ, with adhi);prāvayati, meaning prāpayati "he causes to reach";drāvayati, meaning vilāpayati "he melts";srāvayati, meaning syandayati "he makes it trickle".Applies when / blocked by. Why these examples? One of Bhaṭṭoji's commentators explains. The lotus, the agent of the plain verb, is not sentient. So SK 2754, which needs a sentient agent, would not give the parasmaipada. This rule is needed. The logs likewise "fight" of themselves. Yodhayati devadattam is not the example, since SK 2754 would cover it.
Why does Bhaṭṭoji gloss pru, dru and sru as he does? Bhaṭṭoji's commentators agree on the reason. In the sense of motion these roots would get the parasmaipada by SK 2753 anyway. They are named here for their other senses: reaching, melting, trickling.
One of Bhaṭṭoji's commentators notes the phrase prītiṃ bhaktajanasya yo janayate, with the ātmanepada janayate. He calls it an error, or else the ṇic-final of a stem already ending in ṇic.
Rule. After roots meaning swallowing (nigaraṇa) or moving (calana), when they end in ṇi, the parasmaipada comes as well. "When they end in ṇi" (ṇeḥ) is carried forward from SK 2752. Bhaṭṭoji's commentators gloss nigaraṇa as eating; one of them glosses calana as trembling (kampana).
Example.
The ṇi-final stem nigāri (with the upasarga ni), from a root meaning swallowing, + laṭ, the fruit going to the agent
→ ṇicaś ca would give the ātmanepada
→ this rule: the root means swallowing, so the parasmaipada → nigārayati "he makes [someone] swallow".
Bhaṭṭoji's other examples: āśayati "makes eat", bhojayati "feeds", calayati "makes move", kampayati "makes tremble".
Applies when / blocked by. A vārttika makes an exception: adeḥ pratiṣedhaḥ, "[this is to be] prohibited for ad". Ad "to eat" means eating, so this rule would reach it. The vārttika stops it, and ṇicaś ca stands:
ad + ṇic → ādi + laṭ, the fruit going to the agent
→ this rule is barred by the vārttika
→ the ātmanepada → ādayate devadattena "he makes Devadatta eat".
Why devadattena, in the third case? By SK 540 the one made to eat would become a karman (object). Bhaṭṭoji says this is prohibited by the vārttika ādikhādyor na, "not for ad and khād", given under SK 540. So Devadatta is not a karman; he stays the agent and takes the third case.
How far does the vārttika reach? Bhaṭṭoji answers: it prohibits only what this sūtra gives. When the fruit is not intended for the agent (akartrabhiprāya), SK 2159 still gives the parasmaipada. So ādayaty annaṃ baṭunā "he makes the boy eat the food".
One of Bhaṭṭoji's commentators names the maxim behind this: anantarasya, "a prohibition touches only the injunction immediately before it". Here that injunction is SK 2753, not SK 2159.
Watch out. Do not read adeḥ pratiṣedhaḥ as "ad never takes the parasmaipada". With the fruit for the agent: ādayate. With the fruit for another: ādayati.
Rule. The parasmaipada comes after a ṇi-final root of a certain kind. When not in ṇi (aṇau), the root must be intransitive (akarmaka), and its agent must be sentient (cittavatkartṛka, "having an agent with a mind"). Ṇeḥ is again carried forward from SK 2752.
Example.
śete kṛṣṇaḥ "Kṛṣṇa sleeps": śī has no object, and its agent Kṛṣṇa is sentient
→ śī + ṇic → vṛddhi (SK 254) → śai-i → āy for ai before a vowel (SK 61) → śāyi
→ śāyi + laṭ, the fruit going to the agent: ṇicaś ca would give śāyayate
→ this rule: the parasmaipada → taṃ gopī śāyayati "the cowherdess makes him sleep".
Applies when / blocked by. Bhaṭṭoji's commentators give two counter-examples.
vrīhayaḥ śuṣyanti "the rice dries". So tān śoṣayate "he dries it", with the ātmanepada.kaṭaṃ karoti "he makes a mat". So kaṭaṃ kārayati "he has a mat made". This rule does not decide its pada; ṇicaś ca does.Which ṇi is meant? One of Bhaṭṭoji's commentators, the Tattvabodhinī, says it is the causative ṇic (hetumat), as in SK 2752. "When not in ṇi" refers to the same causative ṇi. So a stem that already ends in a causative ṇic is not "not in ṇi". The Tattvabodhinī's example: SK 2738 (chapter 61) settled the ṇi-final ārohayati, "[the elephant] goes low", as intransitive. A second ṇic can be added to it: "he urges on [the elephant] that is going low". That form is outside this rule, so the ātmanepada stands: ārohayate. It looks like the chapter-61 ārohayate hastī, but it has two *ṇic*s and a different meaning.
Some took aṇau to exclude a cur-class ṇic-final instead. The Tattvabodhinī rejects this view as against the Bhāṣya.
Watch out. The test is made on the plain verb, before ṇic: śete kṛṣṇaḥ, not on the causative taṃ gopī śāyayati, which has an object.
pā ("drink" = nigaraṇa, so 2753 would give pāyayati) + ṇic → pā-y-i (yuk) → pāyayateThe last sūtra withdraws some of the parasmaipadas of group C. For nine ṇi-final roots, the parasmaipada does not come. So ṇicaś ca gives them the ātmanepada again when the fruit goes to the agent. A vārttika adds a tenth root, dheṭ. Bhaṭṭoji then recalls four rules from earlier chapters that make the pada optional, and closes the account of the padas. The running example is pā "to drink": pāyayate.
Rule. After pā, dam, yam with āṅ, yas with āṅ, muh with pari, ruc, nṛt, vad and vas, when they end in ṇi, the parasmaipada does not come. Ṇeḥ is carried forward from SK 2752.
Why the rule is needed. A prohibition needs something to prohibit. Bhaṭṭoji shows where the parasmaipada would come from:
Example.
pā + ṇic → the augment yuk (SK 2585) → pāyi
→ pāyi + laṭ, the fruit going to the agent
→ pā means swallowing: SK 2753 would give the parasmaipada
→ this rule: not the parasmaipada
→ ṇicaś ca (SK 2564) stands: the ātmanepada → pāyayate.
Bhaṭṭoji's other forms: damayate, āyāmayate, āyāsayate, parimohayate, rocayate, nartayate, vādayate, vāsayate.
Bhaṭṭoji's commentators explain some of these:
pāyayate: yuk comes by śācchāsāhvāvyāvepāṃ yuk (SK 2585); one commentator adds that it replaces puk.damayate: the vowel is shortened because dam is mit. One commentator gives this, citing the list-rule janījṝṣknasurañjo 'mantāś ca.āyāmayate: no shortening, because yamo 'pariveṣaṇe denies yam the mit status.vāsayate: from vasa nivāse, "vas in the sense of dwelling", of the bhū class. The ad-class vasa ācchādane, "vas in the sense of covering", is transitive. It never gets the parasmaipada under SK 2754, so this rule has nothing to deny it.Applies when / blocked by. Dam was called transitive among the roots of the div class. How can SK 2754 reach it? One of Bhaṭṭoji's commentators says this very prohibition shows that dam is intransitive too. The other says dam is intransitive when its object is intended as the agent.
The vārttika on dheṭ. A vārttika says dheṭa upasaṃkhyānam: dheṭ is to be added to this prohibition. Bhaṭṭoji's example: dhāpayete śiśum ekaṃ samīcī "the two, united, suckle one child". One of Bhaṭṭoji's commentators says it is a mantra, a Vedic line. The other notes that samīcī is a first-case dual.
When the fruit is not the agent's. Bhaṭṭoji adds: when the fruit is not intended for the agent, SK 2159 does give the parasmaipada. His examples:
vatsān pāyayati payaḥ "he makes the calves drink milk";damayantī kamanīyatāmadam "[Damayantī,] taming the pride of loveliness";bhikṣāṃ vāsayati "he makes the alms stay".Bhaṭṭoji's commentators give the reason: by the maxim anantarasya, this rule prohibits only what SK 2753 and SK 2754 give. It does not touch SK 2159. One commentator's text reads bhikṣā vāsayati, "alms make [one] stay"; Bhaṭṭoji's own text has bhikṣām.
The four optional rules recalled. Bhaṭṭoji now recalls four rules, already explained, in the order of the sūtras. Each makes the pada optional:
lohitāyati / lohitāyate;adyutat / adyotiṣṭa;vartsyati / vartiṣyate; vivṛtsati / vivartiṣate;kalptā, kalptāsi / kalpitāse, kalptāse; kalpsyati / kalpiṣyate, kalpsyate; cikḷpsati / cikalpiṣate, cikḷpsate.Then Bhaṭṭoji closes: samāptā padavyavasthā, "the arrangement of the padas is complete". The section on the parasmaipada ends here.
Watch out. This rule does not make these roots ātmanepadin. It only removes the parasmaipada of SK 2753 and SK 2754. So pāyayate when the fruit is the agent's, but vatsān pāyayati payaḥ when it is not.
anukaroti "he imitates".| SK | What it does | Example |
|---|---|---|
| 2745 | parasmaipada after kṛñ with anu, parā, even with the fruit for the agent and in the senses of SK 2705; only in the sense of the agent | anukaroti, parākaroti |
| 2746 | parasmaipada after kṣip with abhi, prati, ati, against SK 2158 | abhikṣipati |
| 2747 | parasmaipada after vah with pra, against SK 2158 | pravahati |
| 2748 | parasmaipada after mṛṣ with pari; the split pareḥ of "some" | parimṛṣyati; parivahati |
| 2749 | parasmaipada after ram with vi, āṅ, pari, against SK 2157 | viramati, āramati |
| 2750 | parasmaipada after ram with upa (transitive) | yajñadattam uparamati |
| 2751 | optional parasmaipada after intransitive ram with upa | uparamati / uparamate |
| 2752 | parasmaipada after ṇi-final budh, yudh, naś, jan, iṅ, pru, dru, sru: apavāda of SK 2564 | bodhayati padmam, adhyāpayati vedam |
| 2753 | parasmaipada after ṇi-final roots of swallowing or moving; vārttika: not for ad | nigārayati, calayati; ādayate devadattena, but ādayaty annaṃ baṭunā |
| 2754 | parasmaipada after a ṇi-final root intransitive with a sentient agent before ṇi | taṃ gopī śāyayati; but śoṣayate |
| 2755 | no parasmaipada after ṇi-final pā, dam, āṅ-yam, āṅ-yas, pari-muh, ruc, nṛt, vad, vas; vārttika adds dheṭ; four optional rules recalled | pāyayate, dhāpayete; but vatsān pāyayati payaḥ |