In chapter 63 the lakāra expressed the bare action (bhāva) or the object (karman). The root then took yak, the ātmanepada and, in the aorist, ciṇ. This chapter teaches a third case. Rice is sometimes said to cook "of itself": pacyate odanaḥ, "the rice cooks". The rice was the object of cooking; now it is the agent (kartṛ). Such an agent is called the karmakartṛ, the "object-agent". Bhaṭṭoji calls the section the karmakartṛ-prakriyā, "the working of the object-agent". After this chapter you will be able to form pacyate odanaḥ, bhidyate kāṣṭham, apāci and abhedi. You will be able to test whether a root's action resides in its object. You will know where the object-treatment is barred (anyonyaṃ spṛśataḥ), restored (udumbaraḥ phalaṃ pacyate) or newly given (tapyate tapas tāpasaḥ). You will build the optional aorists akāri / akṛta and adohi / adugdha / adhukṣata, and avāruddha without ciṇ. You will meet the roots that keep the ātmanepada but take neither yak nor ciṇ: gauḥ payo dugdhe, namate daṇḍaḥ, kārayate, alaṃkurute kanyā, cikīrṣate kaṭaḥ. Last, you will see śyan stand in the place of yak: sṛjyate srajaṃ bhaktaḥ, kuṣyati pādaḥ. The chapter has seven sūtras, SK 2766 to SK 2772.
odanaṃ pacati → (odana meant as agent) pac + laṭ (kartari) → (karmavat: yak, taṅ) pacyate odanaḥ; luṅ cli → ciṇ → apāciBhaṭṭoji opens the section with a short paragraph. Sometimes a speaker wants to say that an action was exceptionally easy (saukaryātiśaya). Then the agent's own activity is not meant to be expressed. Other kārakas are then called agents, because each is independent in its own activity. So a sword, logs or a pot can be agents: sādhv asiś chinatti, "the sword cuts well"; kāṣṭhāni pacanti, "the logs cook"; sthālī pacati, "the pot cooks". These are ordinary agents. The lakāra is in the agent sense, and nothing new happens.
The object can also be meant as the agent. Then, Bhaṭṭoji says, roots that were transitive (sakarmaka, "having an object") become mostly (prāyeṇa) intransitive (akarmaka). After them the lakāra may express the bare action: pacyate odanena, "there is cooking on the part of the rice". Or it may express the agent. For that case Pāṇini gives SK 2766. It treats the object-agent like an object, so yak, the ātmanepada and ciṇ return. The running example is pacyate odanaḥ, "the rice cooks [by itself]", with its aorist apāci.
One of Bhaṭṭoji's commentators explains "mostly". Roots with one object, such as bhid and chid, do become intransitive. Roots with two objects (dvikarmaka) stay transitive, through the other object. Group B returns to these.
Rule. An agent whose action is like the action residing in the object is treated like the object (karmavat). Bhaṭṭoji's gloss: "an agent whose action is similar to the action residing in the object (karmastha) is treated like the object".
Two things are supplied. "Agent" (kartari) is carried forward from SK 2167, kartari śap, and recast in the first case. Bhaṭṭoji's commentators agree on this. And only an agent denoted by the lakāra is meant; Bhaṭṭoji explains this below.
Bhaṭṭoji then says: "This is an extension of operations (kāryātideśa)." An atideśa is a rule that says "treat this like that". Here the agent receives the operations that an object would bring. Bhaṭṭoji lists them: yak, the ātmanepada, ciṇ and the ciṇvat iṭ. One of his commentators names the rule for each: yak by SK 2756, the ātmanepada by SK 2679, ciṇ by SK 2758, the ciṇvat iṭ by SK 2757.
Example.
"[Someone] cooks the rice": odanaṃ pacati. The rice is the object.
→ the cook's activity is not meant; the rice is meant as the agent
→ pac + laṭ, the lakāra in the sense of the agent, the rice
→ this rule: that agent is treated like the object
→ so the ātmanepada comes (SK 2679): the ending ta, made te by SK 2233 → pac + te
→ so yak comes before the sārvadhātuka ending (SK 2756) → pac + ya + te
→ pacyate.
Bhaṭṭoji: "because the agent is expressed, it takes the first case". The ending expresses it, so the noun is nominative: pacyate odanaḥ, "the rice cooks [by itself]". Likewise bhidyate kāṣṭham, "the wood splits [by itself]".
One of Bhaṭṭoji's commentators describes the scene. The rice softens without any special effort of a person, such as blowing on the fire. The wood splits in two without a firm blow of the axe.
In the aorist:
pac + luṅ → cli
→ the agent is treated like the object, so ciṇ replaces cli before ta (SK 2758) → pac + ciṇ + ta
→ the penultimate a takes vṛddhi before the ṇ-marked ciṇ (SK 2282) → pāc + i + ta
→ ta is deleted after ciṇ (SK 2329); the augment aṭ (SK 2206) → apāci.
Likewise from bhid, with guṇa of the light penultimate (SK 2189): abhedi.
With the ciṇvat iṭ in the future (lṛṭ): kāriṣyate ghaṭaḥ, "the pot will get made [of itself]". This example is explained below.
Applies when / blocked by. Why "-vat", "like"? Both of Bhaṭṭoji's commentators answer. Without -vat the rule could be read as naming the agent "karman". Then the root would stay transitive. The impersonal pacyate odanena would be impossible, because a transitive root takes no lakāra in the bare-action sense.
Why an extension of operations? Both commentators contrast it with an extension of rules (śāstrātideśa). The two give the same result. But a rule exists for the sake of its operation. So the extension of operations is the primary one, and it is adopted.
Only the agent the lakāra denotes. Bhaṭṭoji raises an objection. When the lakāra expresses the bare action, the agent is not expressed by the verb. Treated like an object, would it not take the second case? He answers no. Only an agent denoted by the lakāra is treated like the object. The letter "l" is carried forward from a sūtra-text with two l's: vyatyayo bahulam (SK 3433) read together with liṅy āśiṣy aṅ (SK 3434). How that second l arises, scholars discuss this. In the bare-action sense, the lakāra does not present the agent at all. So the agent is not treated like an object, and takes no second case.
Bhaṭṭoji draws a consequence. The kṛtya affixes, kta, and the affixes in the sense of khal never occur in the object-agent sense. They occur only in the bare-action sense: bhettavyaṃ kusūlena, "there is to be splitting on the part of the granary". Whether kta belongs on this list, scholars discuss this.
"Similar", though it is the same action? Bhaṭṭoji raises a second objection. For pac the action in the object is softening (viklitti). For bhid it is splitting in two (dvidhābhavana). That very action now resides in the agent. It is the same action, not one similar to it. Bhaṭṭoji answers: "True." But being object and being agent are two different states. That difference conditions a difference in the action that goes with them. The word "similar" rests on that. Both of his commentators call this difference conditioned (aupādhika), not real.
Why "karmaṇā", "to the object"? Bhaṭṭoji: so that the rule does not apply when the action is like that of an instrument or a locus. These are the opening examples, sādhv asiś chinatti and the rest. Both of his commentators explain the danger. The sword's activity in asinā chinatti, "he cuts with a sword", is the very activity that resides in the agent in asiś chinatti. Without "karmaṇā", the sword-agent too would be treated like the object.
Not for roots whose action resides in the agent. Bhaṭṭoji adds, "moreover": the rule does not apply to roots whose action resides in the agent (kartṛstha). So there is no yak in these:
gacchati grāmaḥ, "the village goes", that is, it is reached without effort;ārohati hastī, "the elephant mounts", that is, it lets itself be mounted;adhigacchati śāstrārthaḥ smarati śraddadhāti ca, "the meaning of the śāstra understands, remembers and believes", that is, it becomes understood, remembered and believed.How do we know where an action resides? Bhaṭṭoji gives the test. Where the object shows a difference produced by the action (kriyākṛta viśeṣa), the action resides in the object. Cooked rice differs from uncooked rice; cut wood differs from uncut wood. A village that has been gone to looks no different from one that has not. So gam is an agent-residing root. One of Bhaṭṭoji's commentators adds that the agent operations, śap and the rest, apply in these examples.
Why kāriṣyate ghaṭaḥ works. Bhaṭṭoji: kṛ means producing (utpādana), and coming into being (utpatti) resides in the object. So kāriṣyate ghaṭaḥ, "the pot will get made [of itself]". If kṛ meant effort (yatna), this would fail. Effort, like knowing and desiring, resides in the agent. Bhaṭṭoji says "and so on" after the example. His commentators supply the other forms: kariṣyate without the ciṇvat iṭ, the aorist akāri / akṛta, and kriyate ghaṭaḥ svayam eva with yak.
Watch out. The word svayam often goes with these forms: pacyate odanaḥ svayam eva. Both of Bhaṭṭoji's commentators, following Kaiyaṭa, say what it means. It means "by itself as the means" (ātmanā karaṇena), not "by itself as the agent". If svayam named the agent, the lakāra would be in the object sense. So read it "of itself".
A second trap is the phrase anuvyavasyamāne 'rthe, "when the matter is being ascertained". It looks like an ātmanepada form of a karmakartṛ, but it is not. Ascertaining resides in the agent, so there is no yak. Śyan comes, and the o of the root is deleted (SK 2510). The -āna is cānaś (SK 3109, taught later), an affix for habit and the like. It is not an ātmanepada ending. Both of Bhaṭṭoji's commentators add that with yak the form would be anuvyavasīyamāna.
kālaḥ udumbaraṃ phalaṃ pacati → (udumbara as agent; pac still transitive through phala) → (duhipacyoḥ: yak, taṅ) udumbaraḥ phalaṃ pacyate; tapas tapati → tapyate tapas tāpasaḥSK 2766 assumed that the root becomes intransitive when its object turns agent. Two-object roots do not: they keep their other object. Two vārttikas under SK 2766 deal with them. The first bars the object-treatment (karmavattva) for roots that stay transitive. The second restores it, variously, for duh and pac. Then SK 2771 grants it to tap "to practise [austerity]", where SK 2766 would not reach. The running examples are udumbaraḥ phalaṃ pacyate, "the udumbara tree ripens [its] fruit [of itself]", and tapyate tapas tāpasaḥ, "the ascetic practises austerity".
Rule. "For [roots that remain] transitive, the prohibition [of being treated like the object] must be stated." When an object turns agent but the root still has another object, there is no karmavattva. One of Bhaṭṭoji's commentators explains: two-object roots stay transitive through the other object, so this prohibition is needed for them.
Example. Bhaṭṭoji gives two:
anyonyaṃ spṛśataḥ, "the two touch each other";ajā grāmaṃ nayati, "the goat leads [itself] to the village".In anyonyaṃ spṛśataḥ, one of Bhaṭṭoji's commentators explains, each person is both agent and object of the touching. Without the vārttika the form would be spṛśyete, with yak and the ātmanepada. The vārttika keeps the ordinary agent form, spṛśataḥ.
In ajā grāmaṃ nayati, the other commentator explains, the goat is the primary object of nī, now meant as agent. The village is still an object, so the root stays transitive. Without the vārttika, yak and the ātmanepada would follow.
Applies when / blocked by. Which of the two examples truly shows the vārttika at work, scholars discuss this.
Rule. "It must be stated that for duh and pac, [even when] transitive, [the object-treatment holds] variously (bahulam)." Bahulam means "variously": the rule applies in some uses and not in others. Both of Bhaṭṭoji's commentators explain the point. Duh "to milk" and pac "to cook" take two objects. When one object turns agent, the root stays transitive through the other. So the previous vārttika would bar the object-treatment. This vārttika restores it. One commentator calls it a counter-exception (pratiprasava), a rule that lets back in what an exception had removed.
Example. Bhaṭṭoji gives the pac example under SK 2769: udumbaraḥ phalaṃ pacyate. Both of his commentators unpack it.
"Time ripens fruit on the udumbara": kālaḥ udumbaraṃ phalaṃ pacati, with two objects. The udumbara tree is the secondary object, the fruit the primary one.
→ the udumbara, the secondary object, is meant as agent
→ the fruit is still an object, so the root stays transitive
→ the previous vārttika would bar the object-treatment
→ this vārttika restores it: the ātmanepada (SK 2679) and yak (SK 2756) → pacyate
→ udumbaraḥ phalaṃ pacyate, "the udumbara tree ripens [its] fruit [of itself]".
The duh example, gauḥ payo dugdhe, comes under SK 2767 in group D.
Applies when / blocked by. Both of Bhaṭṭoji's commentators consider a doubt: is pac really a two-object root? They agree that the vārttika works either way. If pac is not a two-object root, the vārttika is a fresh grant (apūrvavidhi) of the object-treatment for pac.
Rule. The agent of tap is treated like the object, but only when tap has tapas, "austerity", as its object. Bhaṭṭoji completes the sūtra: "the agent is treated like the object". One of his commentators explains the words. The first tapaḥ is in the sixth case, "of tap". "Agent" comes from SK 2167, and "like the object" from SK 2766.
Bhaṭṭoji: "This is for an injunction (vidhi)." A vidhi gives something not otherwise obtained. One of his commentators says why it is needed. Here tap means to earn, to accomplish austerities. Accomplishing resides in the agent, the ascetic, not in the object. So SK 2766 would not reach this agent, and SK 2771 must grant the treatment itself.
Bhaṭṭoji reports that the followers of the Vṛtti (vṛttyanusāriṇaḥ) call the word eva, "only", superfluous. Whether the eva is idle, scholars discuss this.
Example.
"The ascetic practises austerity": tapas tapati tāpasaḥ. The ascetic is the main agent; tapas is the object.
→ this rule: since tap has tapas as object, the agent is treated like the object
→ the ātmanepada (SK 2679) and yak (SK 2756) → tap + ya + te
→ tapyate tapas tāpasaḥ. Bhaṭṭoji: "the meaning is: he earns [it]".
In the aorist:
tap + luṅ → cli
→ ciṇ would come, but SK 2760 (tapo 'nutāpe ca) bars it for tap
→ so sic replaces cli → tap + s + ta
→ the s is deleted between two consonants of the jhal class (SK 2281); aṭ → atapta [tapas tāpasaḥ].
Applies when / blocked by. Why "having tapas as its object"? Bhaṭṭoji's counter-example: uttapati suvarṇaṃ suvarṇakāraḥ, "the goldsmith heats the gold". The object is not tapas, so there is no object-treatment and no yak.
Watch out. This agent was never an object. The ascetic is the ordinary main agent, and tapas stays the object. Yet the verb takes yak and the ātmanepada, because this rule says so. Do not look for an object turned agent here.
After atapta, Bhaṭṭoji turns back to SK 2767 and gives its examples for snu and nam. Then he adds two vārttikas to that rule. These belong with SK 2767 and are taught in group D.
kaṭaḥ svayam eva: kṛ + luṅ, ta → cli → ciṇ (optional) → akāri; or sic, deleted → akṛtaBy SK 2766 the object-agent takes ciṇ in the aorist, as the object does (SK 2758). Three sūtras adjust this. SK 2768 makes ciṇ optional after a root ending in a vowel. SK 2769 makes it optional after duh. SK 2770 bars it after rudh. All three hold only in the object-agent sense. The running example is the aorist of kṛ with the mat (kaṭa) as object-agent: akāri, or akṛta, "[the mat] got made [of itself]".
Rule. After a root ending in a vowel (ac), in the object-agent sense, ciṇ optionally replaces cli when the ending ta follows. Bhaṭṭoji's gloss: "after [a root] ending in ac, ciṇ optionally replaces cli in the object-agent sense, when the [ending] ta follows".
Most of the rule is carried forward. Both of Bhaṭṭoji's commentators give the sources. "Ciṇ" and "when ta follows" come from SK 2513, ciṇ te padaḥ. "Optionally" comes from SK 2328. One of them adds two more: "of cli" from SK 2222, and "after a root" from SK 2629. "Root" is qualified by acaḥ, so it means a root ending in a vowel (tadantavidhi).
Example. One of Bhaṭṭoji's commentators supplies the subject: kaṭaḥ svayam eva, "the mat, of itself".
kṛ + luṅ, the ending ta, the mat as object-agent → cli
→ the root ends in the vowel ṛ; this rule: ciṇ optionally → kṛ + ciṇ + ta
→ vṛddhi of the final ṛ before ciṇ (SK 254), which comes out as ār → kār + i + ta
→ ta is deleted after ciṇ (SK 2329); the augment aṭ → akāri.
In the other option:
kṛ + luṅ, ta → cli is replaced by sic → kṛ + s + ta
→ sic is deleted after a short vowel (SK 2369); aṭ → akṛta.
So: akāri or akṛta, "[the mat] got made [of itself]".
Applies when / blocked by. One of Bhaṭṭoji's commentators gives two counter-examples, where ciṇ is obligatory:
abhedi kāṣṭham, "the wood split": bhid ends in a consonant;akāri ghaṭaḥ kulālena, "the pot was made by the potter": the lakāra is in the object sense.Watch out. The option is between ciṇ and the ordinary sic. The ātmanepada stays in both forms: akṛta has the ātmanepada ta.
Rule. After duh "to milk" too, in the object-agent sense, ciṇ optionally replaces cli when ta follows. One of Bhaṭṭoji's commentators says that everything carried into SK 2768 continues here, except "ending in a vowel".
Example. One of Bhaṭṭoji's commentators supplies the sentence: svayam eva gauḥ payaḥ, "the cow [gave] milk of herself".
duh + luṅ, ta, the cow as object-agent → cli
→ this rule: ciṇ → guṇa of the light penultimate u → adohi.
In the other option, Bhaṭṭoji says, ksa replaces cli. One of his commentators gives the rule: SK 2336.
→ duh + ksa + ta
→ by SK 2365, ksa is optionally deleted → adugdha
→ if it is not deleted → adhukṣata.
So three forms: adohi, adugdha, adhukṣata.
Applies when / blocked by. One of Bhaṭṭoji's commentators adds that "in the object-agent sense" continues. Where the cow is a plain object, ciṇ is obligatory: adohi gaur gopena, "the cow was milked by the cowherd".
Bhaṭṭoji puts more under this sūtra than its own forms. The example udumbaraḥ phalaṃ pacyate belongs to the duh–pac vārttika (group B). The vārttikas on sṛj and yuj are taught in group E. The vārttika on adorning, kṝ and the rest, and san is taught in group D.
Rule. After rudh "to confine", in the object-agent sense, ciṇ does not replace cli. Bhaṭṭoji's gloss: "after this [root], ciṇ does not replace cli". Both of his commentators say "in the object-agent sense" comes from SK 2768. One of them adds that "ciṇ" comes from SK 2513.
Example. One of Bhaṭṭoji's commentators supplies svayam eva.
ava + rudh + luṅ, ta, the cow as object-agent → cli
→ ciṇ would come from the object-treatment; this rule bars it
→ sic replaces cli
→ the s of sic is deleted (SK 2281); aṭ joins ava → avāruddha gauḥ, "the cow got confined [of herself]".
Applies when / blocked by. Bhaṭṭoji: "only in the object-agent sense". Where the cow is a plain object, ciṇ comes: avārodhi gaur gopena, "the cow was confined by the herdsman".
Watch out. SK 2768 could not give rudh an option, since rudh ends in a consonant. So without this rule, ciṇ would be obligatory. This rule removes it altogether.
gāṃ payo dogdhi → (gauḥ, the secondary object, as agent) duh + laṭ → (no yak) śap, deleted → gauḥ payo dugdheFor some roots the object-agent keeps only part of the object-treatment. It takes the ātmanepada, but neither yak nor ciṇ. SK 2767 names three roots: duh, snu and nam. Three vārttikas add more: causatives, śri and brū; then ṇi-stems, śranth, granth and certain intransitives; then roots of adorning, kṝ and its group, and desideratives. The running example is gauḥ payo dugdhe, "the cow yields milk [of herself]".
Rule. For duh, snu and nam, in the object-agent sense, yak and ciṇ do not occur. Bhaṭṭoji's gloss: "for these, in the object-agent sense, yak and ciṇ do not occur".
"In the object-agent sense" is not in the sūtra. One of Bhaṭṭoji's commentators gives two ways to get it. It may come from the sūtra's being uttered together with SK 2766 (samabhivyāhāra). Or it may be carried by a "frog's leap" (maṇḍūkapluti) from SK 2768. A frog's leap carries a word forward past the sūtras in between.
For duh, Bhaṭṭoji says, this rule bars only yak. Ciṇ is accepted optionally. One of his commentators says the option comes from SK 2769, and he knows this sentence only as the reading of some copies.
Example. duh:
"[The cowherd] milks the cow [for] milk": gāṃ payo dogdhi. The cow is the secondary object, the milk the primary one.
→ the cow is meant as agent → duh + laṭ, the lakāra in the cow as object-agent
→ duh stays transitive through the milk, so the object-treatment comes by the duhipacyoḥ vārttika (group B)
→ treated like the object, so the ātmanepada (SK 2679) → duh + te
→ this rule bars yak, so śap comes (SK 2167)
→ śap is deleted after duh, a root of the ad class (SK 2423) → dugdhe
→ gauḥ payo dugdhe, "the cow yields milk [of herself]".
Both of Bhaṭṭoji's commentators say that the ātmanepada here comes from the object-treatment, and is obligatory.
Why the cow and not the milk? Both of Bhaṭṭoji's commentators answer from the verse under SK 2765. With duh and similar roots, the lakāra goes to the secondary object. So payo gāṃ dugdhe, with the milk as agent, is not given as an example.
snu, from Bhaṭṭoji's examples under SK 2771:
prasnute [gauḥ], "[the cow] lets her milk flow". Both of his commentators explain. Snu here means eagerness for the milk to flow, and that eagerness resides in the cow. In vatso gāṃ prasnauti, "the calf makes the cow eager to give milk", the root carries a causative sense. With the cow, its object, meant as agent: prasnute, with the ātmanepada and no yak.
In the aorist Bhaṭṭoji gives two forms: prāsnāviṣṭa and prāsnoṣṭa. Ciṇ is barred, but the ciṇvat iṭ still comes, optionally. One of his commentators says why: "ciṇvat" extends the operation of ciṇ, and "no ciṇ" cannot cancel that extension. Without the ciṇvat iṭ, no other iṭ comes (SK 2323): prāsnoṣṭa.
nam, also under SK 2771:
namate daṇḍaḥ, "the stick bends [of itself]"; aorist anaṃsta. Bhaṭṭoji: "here nam has the causative sense incorporated (antarbhāvitaṇyartha)". It means "to make bend", as in "someone bends the stick". With the stick, its object, meant as agent: namate.
Applies when / blocked by. One of Bhaṭṭoji's commentators raises a doubt about namate daṇḍaḥ. A bent stick may look no different from an unbent one, so does the action reside in the object? He answers that the bending is what is aimed at, and the bending is in the stick.
Watch out. Without yak, the root keeps its ordinary class-sign: śap, deleted after duh. The ātmanepada stays. Do not write duhyate for the object-agent.
Rule. Bhaṭṭoji introduces it: "in the prohibition of yak and ciṇ —". The vārttika: "[roots ending in] the causative ṇic (hetumaṇṇic), śri and brū should be added." Upasaṃkhyāna means adding items to a rule's list. So these too, as object-agents, take the ātmanepada without yak or ciṇ.
Example. One of Bhaṭṭoji's commentators explains kārayate.
"Devadatta makes [it]; Yajñadatta prompts him": kārayati devadattaṃ yajñadattaḥ.
→ Devadatta, the object of the causative, is meant as agent, without Yajñadatta's prompting
→ the ātmanepada; no yak → kārayate, "[Devadatta] makes [of himself, unprompted]".
In the aorist, ciṇ is barred, so caṅ comes after the ṇi-stem (SK 2312) → acīkarata.
Likewise śri "to raise":
ucchrayate daṇḍaḥ, "the staff raises [itself]"; aorist with caṅ (SK 2312) → udaśiśriyata.
Bhaṭṭoji: "the ciṇvat iṭ, however, does occur": kāriṣyate, ucchrāyiṣyate. One of his commentators answers an objection. SK 2757 needs a vowel in upadeśa, and a ṇi-stem might seem not to qualify. But the i of ṇi is a vowel in upadeśa, and so is the vowel of śri.
And brū "to speak":
brūte kathā, "the story tells [itself]"; aorist avocata. One of Bhaṭṭoji's commentators gives the steps of avocata. Vac replaces brū (SK 2453), aṅ replaces cli (SK 2438), and vac takes the augment um (SK 2454).
Rule. Bhaṭṭoji introduces it: "the Bhāradvājīyas read". The Bhāradvājīyas are a school whose reading of the vārttika Bhaṭṭoji cites. Their list: "ṇi, śranth, granth, brū, and [roots that are] intransitive [in the rule giving] the ātmanepada should be added." These, as object-agents, take neither yak nor ciṇ.
Both of Bhaṭṭoji's commentators say ṇi here is any ṇi, not only the causative. So stems in ṇiṅ count too.
Example. A ṇiṅ-stem:
"[The cow] raises its tail": puccham udasyati → with ṇiṅ (SK 2676) → utpucchayate gauḥ.
→ with the causative sense incorporated: utpucchayate gām, "[someone] makes the cow raise its tail"
→ the cow, now an object, is again meant as agent: utpucchayate gauḥ
→ aorist udapupucchata.
Bhaṭṭoji: "because yak and ciṇ are prohibited, [there are] śap and caṅ."
Śranth and granth. Both of these roots may optionally go without ṇic; they belong to the group called ādhṛṣīya. Bhaṭṭoji says they are named for that ṇic-less option. The forms:
granthati grantham, "he composes a book" → granthate granthaḥ, "the book gets composed [of itself]"; aorist agranthiṣṭa;śranthati mekhalāṃ devadattaḥ, "Devadatta loosens the girdle" → śranthate; aorist aśranthiṣṭa.The same roots also belong to the krī class. Bhaṭṭoji: "for the krī-class ones, however": śrathnīte, grathnīte svayam eva. One of his commentators explains: with yak barred, their class-sign śnā comes.
An intransitive root of the ātmanepada rule:
vikurvate saindhavāḥ, "the horses (saindhava, from Sindhu) prance". Bhaṭṭoji: "the meaning is: they gallop". The ātmanepada comes by SK 2707 and SK 2708.
→ the root takes a causative sense: "[someone] makes the horses prance"
→ that prompting is dropped again, and the horses are meant as object-agents: vikurvate saindhavāḥ, with no yak.
In the aorist Bhaṭṭoji gives two sets:
vyakāriṣṭa, vyakāriṣātām, vyakāriṣata;vyakṛta, vyakṛṣātām, vyakṛṣata.One of his commentators explains. Ciṇ is barred, and there is no caṅ, since the stem does not end in ṇi. So sic comes. In the first set, sic takes the ciṇvat iṭ with vṛddhi. In the second, without it, vyakṛta loses its sic after the short vowel (SK 2369).
Rule. "And for [roots] whose action is adorning, for kṝ and the rest (kirādi), and for [roots ending in] san, [the operations of the object are prohibited] except for the ātmanepada." Bhaṭṭoji's gloss: for these there is no yak, no ciṇ, and no ciṇvat iṭ. Only the ātmanepada remains. Bhaṭṭoji adds that kṝ and the rest form a subgroup within the tud class.
Example. Bhaṭṭoji's forms:
alaṃkurute kanyā, "the girl adorns [herself]"; aorist alamakṛta;avakirate hastī, "the elephant gets strewn [with flowers]"; aorist avākīrṣṭa;girate, "[it] gets swallowed"; aorist agīrṣṭa;ādriyate, "[he] gets respected"; aorist ādṛta;cikīrṣate kaṭaḥ, "the mat is wanted-to-be-made"; aorist acikīrṣiṣṭa.One of Bhaṭṭoji's commentators glosses the first two. The girl becomes adorned of herself, without another's effort. "Flowers strew the elephant" is recast with the elephant as agent.
Applies when / blocked by. In cikīrṣate kaṭaḥ, desire resides in the agent, the one who wants. So would the rule not fail? Bhaṭṭoji answers: "here [the action] resides in the object with respect to the action of kṛ." The making resides in the mat. One of his commentators states the objection behind this. A mat looks no different whether wanted or not.
Watch out. These roots look like ordinary ātmanepada forms. alaṃkurute has the class-sign of kṛ, not yak. Do not add yak to them.
kuṣṇāti pādaṃ devadattaḥ → (pāda as agent) kuṣ + śyan + parasmaipada → kuṣyati pādaḥ; or yak + taṅ → kuṣyateFor a few roots śyan, the class-sign of the div class, stands where yak would. Two vārttikas under SK 2769 do this for transitive sṛj and yuj: they treat the agent like the object, variously. For sṛj this agent is the main agent, if he has faith. For yuj, one commentator reads it as an object-agent. SK 2772 does it optionally for the object-agent of kuṣ and raj, and adds the parasmaipada endings. The running example is kuṣyati pādaḥ svayam eva, "the foot gets pulled of itself", with its other option kuṣyate.
Rule. "For sṛj and yuj, [the agent is variously treated like the object], but [with] śyan." Bhaṭṭoji glosses: "the agent of these two, when they are transitive, is variously (bahulam) treated like the object, and śyan, in exception to yak, is to be stated." One of his commentators says this gloss is the Bhāṣya's own wording.
One of Bhaṭṭoji's commentators identifies the roots. They are sṛj "to produce" of the tud class and yuj "to join" of the rudh class. For their agent the object-treatment would not otherwise come; this vārttika grants it, variously.
Applies when / blocked by. Why śyan and not yak? Both of Bhaṭṭoji's commentators answer: for the accent. Śyan is marked with n, and so the word takes the high tone (udātta) on its first syllable. With yak, the yak itself would carry it.
Rule. "It must be stated that for sṛj [this holds] only when the agent is endowed with faith (śraddhā)."
Example. sṛj:
sṛjyate srajaṃ bhaktaḥ, "the devotee makes a garland". Bhaṭṭoji: "the meaning is: he produces [it] with faith". One of his commentators adds that the lakāra is in the main agent.
→ the devotee is endowed with faith, so he is treated like the object
→ the ātmanepada; śyan in the place of yak → sṛj + ya + te → sṛjyate
→ aorist, with ciṇ: asarji.
yuj:
yujyate brahmacārī yogam, "the celibate student is joined to yoga". One of Bhaṭṭoji's commentators explains. In the active, yoga is the agent and the student its object: yoga unites the student. With the student meant as agent, śyan replaces yak. He comes to be joined with yoga of himself, without much toil.
Watch out. sṛjyate and yujyate look like yak-forms, but the ya is śyan. The two differ only in accent.
Rule. For kuṣ "to pull" and raj "to dye", in the object-agent sense, there is optionally not yak but śyan, and the parasmaipada endings as well. Bhaṭṭoji's gloss: "for these two, in the object-agent sense, [there is] not yak but śyan, and parasmaipada as well".
Both of Bhaṭṭoji's commentators say that "not" is carried forward from SK 2767. One of them gives the rest. "Yak" also comes from SK 2767. "In the object-agent sense" comes by a frog's leap from SK 2768. Prācām, "according to the Easterners", makes the rule optional. Bhaṭṭoji adds that the parasmaipada is an exception (apavāda) to the ātmanepada.
Example. One of Bhaṭṭoji's commentators gives the base sentence.
"Devadatta pulls the foot": kuṣṇāti pādaṃ devadattaḥ.
→ the foot, the object, is meant as agent, independent of human effort
→ this rule: śyan instead of yak, and the parasmaipada → kuṣ + ya + ti
→ kuṣyati pādaḥ svayam eva, "the foot gets pulled of itself".
In the other option, neither applies: yak and the ātmanepada → kuṣyate pādaḥ.
Likewise rajyati or rajyate vastram, "the cloth gets dyed [of itself]". Bhaṭṭoji's text as we have it reads vasram; the sense "cloth" is that of vastram.
Applies when / blocked by. Bhaṭṭoji: "where yak is not in question, however, this [rule] does not apply". So in the precative (āśīrliṅ), whose ending is ārdhadhātuka and outside yak's domain, there is no śyan and no parasmaipada: koṣiṣīṣṭa, raṃkṣīṣṭa, with the ātmanepada. Both of his commentators give the reason. The parasmaipada is taught together with śyan, so it cannot come where śyan does not.
Both of Bhaṭṭoji's commentators explain his gloss "not yak but śyan": śyan is put in yak's very place. One of them adds why this matters. Taught alone, śyan would come even before other ārdhadhātuka affixes, where yak never stands.
Watch out. The two options differ in more than the endings. Both of Bhaṭṭoji's commentators note that the participle differs too. With śyan, kuṣyantī takes the nasal num obligatorily (SK 446). With yak, it would be optional. The accent differs as well.
With SK 2772, Bhaṭṭoji closes the section: "thus the procedure of the object-agent".
pacyate odanaḥ.pacyate odanena.| SK | What it does | Example |
|---|---|---|
| 2766 | an agent whose action is like the object-residing action, and which the lakāra denotes, is treated like the object: yak, ātmanepada, ciṇ, ciṇvat iṭ | pacyate odanaḥ, bhidyate kāṣṭham, apāci, abhedi, kāriṣyate ghaṭaḥ |
| 2766 vārttika | no object-treatment for roots that stay transitive | anyonyaṃ spṛśataḥ, ajā grāmaṃ nayati |
| 2766 vārttika | for duh and pac the object-treatment is restored, variously | udumbaraḥ phalaṃ pacyate |
| 2771 | the agent of tap with tapas as object is treated like the object; ciṇ barred by SK 2760 | tapyate tapas tāpasaḥ, atapta |
| 2768 | after a vowel-final root, object-agent, ciṇ is optional before ta | akāri / akṛta |
| 2769 | the same after duh | adohi / adugdha / adhukṣata |
| 2770 | after rudh, object-agent, no ciṇ | avāruddha gauḥ |
| 2767 | for duh, snu, nam, object-agent, no yak and no ciṇ (for duh, ciṇ optional) | gauḥ payo dugdhe, prasnute, namate daṇḍaḥ |
| 2771 vārttika (on SK 2767) | the same for causatives, śri, brū; ciṇvat iṭ still comes | kārayate, ucchrayate daṇḍaḥ, brūte kathā |
| 2771 vārttika (on SK 2767; Bhāradvājīyas) | the same for any ṇi-stem, śranth, granth, brū, and intransitives of the ātmanepada rule | utpucchayate gauḥ, granthate granthaḥ, vikurvate saindhavāḥ |
| 2769 vārttika | for roots of adorning, kṝ and the rest, and san-stems: only the ātmanepada | alaṃkurute kanyā, avakirate hastī, cikīrṣate kaṭaḥ |
| 2769 vārttikas | for sṛj (with a faithful agent) and yuj: object-treatment, variously, with śyan for yak | sṛjyate srajaṃ bhaktaḥ, yujyate brahmacārī yogam |
| 2772 | for kuṣ and raj, object-agent, optionally śyan and the parasmaipada instead of yak | kuṣyati / kuṣyate pādaḥ, rajyati / rajyate vastram; but koṣiṣīṣṭa |