Book 3 - Pada-kāṇḍa (7): Sādhana-samuddeśa (On the Means)

3.7.1 unformalized

स्वाश्रये समवेतानां तद्वदेवाश्रयान्तरे ।
क्रियाणामभिनिष्पत्तौ सामर्थ्यं साधनं विदुः ॥ १ ॥

svāśraye samavetānāṃ tadvadevāśrayāntare |
kriyāṇāmabhiniṣpattau sāmarthyaṃ sādhanaṃ viduḥ || 1 ||

The capacity, in the completing of actions inherent in their own substratum — and likewise in another substratum — they know as the means.

The definition of sadhana, and it is functional through and through. Sva-asraye samavetanam — of actions *inherent in their own substratum* (the cooking that is in the cook); tadvad eva asrayantare — *and likewise in another substratum* (the cooking that reaches the rice). Kriyanam abhinispattau samarthyam sadhanam viduh: the *capacity*, in the *completing* of those actions, they know as the *means*. Not: a means is a thing used. A means is a *capacity*, individuated by its contribution to an action's completion — samarthyam, the word this work has used for what settles every disputed case (2.371's samarthya-laksanam deciding distribution; 3.1.79's capacity explaining why universals cannot be substituted). So the karakas — agent, object, instrument, and the rest, the whole architecture of Sanskrit syntax — are from the outset capacities and not entities. What makes something an agent is not what it is but what it does toward a completion, and the samuddesa's 157 pages will not add anything to that definition; they will only show how far it reaches.

3.7.2 unformalized

शक्तिमात्रासमूहस्य विश्वस्यानेधर्म्मणः ।
सर्वदा सर्वथा भावात् क्वचित् किंचिद् विवक्ष्यते ॥ २ ॥

śaktimātrāsamūhasya viśvasyānedharmmaṇaḥ |
sarvadā sarvathā bhāvāt kvacit kiṃcid vivakṣyate || 2 ||

Since the universe — an aggregate of mere capacities, of manifold properties — exists always and in every way, somewhere something is intended.

The metaphysics behind 3.7.1, and it is 2.250's semantics turned outward. Sakti-matra-samuhasya visvasya aneka-dharmanah: the *universe* is an *aggregate of mere capacities*, of *manifold properties* — sakti-matra: capacities *only*, nothing underneath them bearing them (3.1.22-23's dravyadayah sarvah saktayah; 3.6.14's things without place, parts, motion, support). Sarvada sarvatha bhavat: since it exists *always* and in *every way* — everything is everything, potentially, at all times. Kvacit kimcid vivaksyate: *somewhere*, *something* is *intended*. Out of a totality where every capacity is everywhere available, speech selects — and the selecting is vivaksa (1.155, 2.400's directed eye). This is the exact structural twin of the word's sarvarthya (2.250: one word, all meanings, carved by nimittas): the world is omni-capable as the word is omni-significant, and in both cases intention does the carving. The karaka-system is where the two carvings meet.

3.7.3 unformalized

साधनव्यवहारश्च बुद्ध्यवस्थानिबन्धनः ।
सन्नसन् वार्थरूपेषु भेदो बुद्ध्या प्रकल्प्यते ॥ ३ ॥

sādhanavyavahāraśca buddhyavasthānibandhanaḥ |
sannasan vārtharūpeṣu bhedo buddhyā prakalpyate || 3 ||

And the practice of means is grounded in a state of the intellect; a difference among the forms of objects, whether existent or non-existent, is constructed by the intellect.

The samuddesa's thesis, stated in its third verse. Sadhana-vyavaharah ca buddhy-avastha-nibandhanah: the *practice of means* is *grounded in a state of the intellect* — which thing counts as agent, which as instrument, is a fact about how the situation is being held in mind. Nibandhana: the grounding-word (1.13's words as the basis of practice; 2.297's sabdah loka-nibandhanah). And the reach of the claim: san asan va artha-rupesu bhedo buddhya prakalpyate — a *difference* among the *forms of objects*, whether *existent or non-existent*, is *constructed by the intellect*. Not merely the karaka-assignments: the object-differences themselves, and *whether they exist or not* (3.3.42's figurative existence, granted by mention). So the whole furniture on which the karaka-system operates is buddhi-constructed, and the system's assignments are constructions on constructions. This is 2.428's doctrine — the speaker alone intends the sadhya-sadhana structure — given its ground: he can, because there is nothing there to resist him.

3.7.4 unformalized

बुद्धया समाहितैकत्वान् पञ्चालान् कुरुभिर्यदा ।
पुनर्विभजते वक्ता तदाऽपायः प्रतीयते ॥ ४ ॥

buddhayā samāhitaikatvān pañcālān kurubhiryadā |
punarvibhajate vaktā tadā'pāyaḥ pratīyate || 4 ||

When a speaker, having brought the Pancalas to oneness by the intellect along with the Kurus, divides them again — then a separation is understood.

The thesis of 3.7.3 in a worked example. Buddhya samahita-ekatvan pancalan kurubhih yada: when, *by the intellect*, the *Pancalas* have been *brought to oneness* *with the Kurus* — two peoples held in one thought, as one group. Punah vibhajate vakta: *the speaker divides them again* — and says something like 'the Pancalas came away from the Kurus'. Tada apayah pratiyate: *then* a *separation* is understood — and with it the ablative case, apadana, the karaka of departure. Note the sequence: the *unity* was made by buddhi first, and only against that made unity can the division register as a separation. Nothing in Pancalas and Kurus made them one; the speaker did; and the ablative he then uses reports a departure from a unity he manufactured a moment earlier. The karaka is the residue of the speaker's own prior construing — which is what 3.7.3 said, now visible in a single sentence.

3.7.5 unformalized

शब्दोपहितरूपाश्च बुद्धेर्विषयतां गतान् ।
प्रत्यक्षमिव कंसादीन् साधनत्वेन मन्यते ॥ ५ ॥

śabdopahitarūpāśca buddherviṣayatāṃ gatān |
pratyakṣamiva kaṃsādīn sādhanatvena manyate || 5 ||

And things such as Kamsa, whose forms are conditioned by words and which have come to be objects of the intellect, one considers as means — as though they were perceptible.

The thesis extended to what does not exist. Sabda-upahita-rupan buddheh visayatam gatan: things *whose forms are conditioned by words* and which have *come to be objects of the intellect* — beings that have no existence except through being spoken of and thought (3.3.39's aupacariki satta; 3.3.53's cognition born of designation). Kamsadin: *Kamsa* and the like — the tyrant of the Krsna-legend, long dead and perhaps never living, and the grammarians' own standard example (Panini has a rule for kamsa-vadha, the slaying of Kamsa). Pratyaksam iva sadhanatvena manyate: one considers them *as means* — *as though they were perceptible*. One says 'Krsna killed Kamsa', and Kamsa is the karman, the object, doing full grammatical work. So the karaka-system operates identically on the real and the fabulous, which is decisive for 3.7.3: if karakas were features of events, the non-existent could not bear them. They are buddhi-postings, and buddhi posts them on Kamsa as readily as on the pot.

3.7.6 unformalized

बुद्धिप्रवृत्तिरूपं च समारोप्याभिधातृभिः ।
अर्थेषु शक्तिभेदानां क्रियते परिकल्पना ॥ ६ ॥

buddhipravṛttirūpaṃ ca samāropyābhidhātṛbhiḥ |
artheṣu śaktibhedānāṃ kriyate parikalpanā || 6 ||

And by the expressers, having superimposed the form of the intellect's operation upon the objects, a construction of differences of capacity is made.

The mechanism named, and the verb is 3.1.6's. Buddhi-pravrtti-rupam ca samaropya: having *superimposed* the *form of the intellect's operation* — samaropa, the superimposition by which the word's own universal is laid over the object's (3.1.6's adhyaropa-kalpana). Abhidhatrbhih arthesu: by the *expressers*, upon the *objects*. Saktibheda nam kriyate parikalpana: a *construction of differences of capacity* is *made*. So the karakas are the shape of the mind's own operating, projected onto what it operates about: the intellect proceeds from agent through instrument to object, and it lays that procession on the world and calls it the structure of the event. The verse is the Padakanda's most explicit statement of where its categories come from — not from things, and not arbitrarily either, but from the form of thinking itself, externalized (3.6.23's antahkarana-dharmo va bahir evam prakasate: a property of the inner organ shining forth outwardly, there declined as unstatable, here stated of the karakas).

3.7.7 unformalized

व्यक्तौ पदार्थे शब्दादेर्जन्यमानस्य कर्मणः ।
साधनत्वं तथा सिद्धं बुद्धिरूपप्रकल्पितम् ॥ ७ ॥

vyaktau padārthe śabdāderjanyamānasya karmaṇaḥ |
sādhanatvaṃ tathā siddhaṃ buddhirūpaprakalpitam || 7 ||

When the individual is the word-meaning, the means-hood of an object being produced from a word and the like is likewise established, constructed by the form of the intellect.

The doctrine applied to the hardest karaka: the nirvartya-karman, the object *produced by* the action. Vyaktau padarthe: when the *individual* is the word-meaning. Sabdadeh janyamanasya karmanah sadhanatvam: the *means-hood* of an *object being produced* — the mat that the weaving makes, the sound the striking makes: the thing that does not exist until the action completes, and is nonetheless the action's object throughout. This is the case that most embarrasses any realist account: what is the action reaching toward, while it reaches? Tatha siddham buddhi-rupa-prakalpitam: it is *likewise established*, *constructed by the form of the intellect*. The produced object is a buddhi-posting held in place across the production — exactly like Kamsa (3.7.5), and by the same mechanism. So the karaka-system's most paradoxical member is not an anomaly needing special treatment; it is the ordinary case with its constructedness visible, since here the thing manifestly is not there to be reached.

3.7.8 unformalized

स्वतन्त्रपरतन्त्रत्वे क्रमरूपं च दर्शितम् ।
निरीहेष्ववि भावेषु कल्पनोपनिबन्धनम् ॥ ८ ॥

svatantraparatantratve kramarūpaṃ ca darśitam |
nirīheṣvavi bhāveṣu kalpanopanibandhanam || 8 ||

Independence and dependence, and the form of sequence, have been shown; even among things that are without striving, they rest upon construction.

The two poles of the karaka-system placed. Svatantra-paratantratve: *independence* and *dependence* — the agent is defined by Panini as svatantra, independent (1.4.54: svatantrah karta), and every other karaka is dependent on it. Krama-rupam ca darsitam: and the *form of sequence* has been shown (3.6.4's sequence from time; 3.1.35's Being with sequence). Nirihesu api bhavesu kalpana-upanibandhanam: *even among things that are without striving* — nir-iha: without effort, without intention, without any doing of their own — these *rest upon construction*. The axe does not strive to cut, the fire does not strive to burn, and yet we make the axe an instrument and the fire an agent ('the fire cooks'). Independence and dependence are conferred on the striving-less by construction, and the whole hierarchy of the karakas — that celebrated architecture in which everything orbits an independent agent — is laid on a world in which nothing is doing anything.

3.7.9 unformalized

शक्तयः शक्तिंमन्तश्र सर्वे संसर्गवादिनाम् ।
भावास्तेष्वस्वशब्देषु साधनत्वं निरूप्यते ॥ ९ ॥

śaktayaḥ śaktiṃmantaśra sarve saṃsargavādinām |
bhāvāsteṣvasvaśabdeṣu sādhanatvaṃ nirūpyate || 9 ||

For the proponents of commingling, all entities are capacities and possessors of capacities; among those, which are not expressed by their own words, means-hood is determined.

The position placed in its school. Saktayah sakti-mantah ca sarve samsarga-vadinam bhavah: for the *proponents of commingling* — the samsarga-vadins, those for whom things exist only in relation (3.4.1's samsarga-rupat sambhutah: word-meanings arisen from the form of commingling) — *all* entities are *capacities* and *possessors of capacities*, both at once, nothing being purely one or the other. Tesu asva-sabdesu sadhanatvam nirupyate: among *those*, which are *not expressed by their own words*, *means-hood* is *determined*. Asva-sabda: not having its own word — the capacities are not named directly by any word; they are determined as means only through the case-endings and the construction (3.7.13 will say this outright). So the karakas are capacities in a world of capacities, reached by no word of their own, fixed by the grammar's oblique apparatus. The verse is the samuddesa's placement of itself: this is what the karaka-system looks like from inside a relational ontology.

3.7.10 unformalized

घटस्य दृशिकर्मत्वे महत्त्वादीनि साधनम् ।
रूपस्य दृशिकर्मत्वे रूपत्वादीनि साधनम् ॥ १० ॥

ghaṭasya dṛśikarmatve mahattvādīni sādhanam |
rūpasya dṛśikarmatve rūpatvādīni sādhanam || 10 ||

In the pot's being the object of seeing, largeness and the rest are the means; in colour's being the object of seeing, colour-hood and the rest are the means.

Two worked cases, and their pairing is the argument. Ghatasya drsi-karmatve mahattvadini sadhanam: in the *pot's* being the *object of seeing*, *largeness* and the rest are the *means* — what makes the pot visible is not the pot but its magnitude (the Vaisesika's own doctrine: mahattva is a condition of visibility; atoms are invisible for want of it). Rupasya drsi-karmatve rupatvadini sadhanam: in *colour's* being the object of seeing, *colour-hood* and the rest are the means. So in each case, what does the enabling is a property, and the thing we call the object of seeing is not what enables its own being seen. The karman — the object — is thus doubly derivative: it is a buddhi-posting (3.7.3), and even its availability to the action runs through properties that are themselves capacities. Nothing in the chain is the solid thing the grammar's accusative seems to name.

3.7.11 unformalized

स्वैः सामान्यविशेषैश्च शक्तिमन्तो रसादयः ।
नियतग्रहणा लोके शक्तयस्तास्तथाश्रयैः ॥ ११ ॥

svaiḥ sāmānyaviśeṣaiśca śaktimanto rasādayaḥ |
niyatagrahaṇā loke śaktayastāstathāśrayaiḥ || 11 ||

Tastes and the rest are possessors of capacity by their own generals and particulars; in the world, those capacities are of fixed apprehension, likewise by their substrata.

The fixity within the constructedness. Svaih samanya-visesaih ca sakti-mantah rasadayah: *tastes* and the rest are *possessors of capacity* by their *own* generals and particulars — sweetness has its capacity by being sweet, and by the kind of sweet it is. Niyata-grahanah loke: in the *world*, they are of *fixed apprehension* — the tongue takes taste and not colour, invariably; no confusion has ever occurred. Saktayah tah tatha asrayaih: those *capacities* likewise, by their *substrata*. This is 3.3.29's yogyata (the senses' beginningless fitness for their objects) reappearing in the karaka-system: the capacities are constructed (3.7.6) and they are not arbitrary — each is fixed to its apprehension as the eye is fixed to colour. So the samuddesa's constructivism has the same shape as the Dik-samuddesa's: relative to the frame, exact (3.6.22); and the frame is not the thing's.

3.7.12 unformalized

इन्द्रियार्थमनः कर्तृसंबन्धः साधनं क्वचित् ।
यद्यदा यदनुग्राहि तत्तदा तत्र साधनम् ॥ १२ ॥

indriyārthamanaḥ kartṛsaṃbandhaḥ sādhanaṃ kvacit |
yadyadā yadanugrāhi tattadā tatra sādhanam || 12 ||

The connection of sense-faculty, object, mind and agent is somewhere the means; whatever, at whatever time, is what assists — that, at that time, is the means there.

The general rule, and it is as loose as the samuddesa's thesis requires. Indriya-artha-manah-kartr-sambandhah sadhanam kvacit: the *connection* of *sense-faculty*, *object*, *mind* and *agent* is *somewhere* the means — the whole four-term complex of perception can itself be the instrument, on occasion. And then the formula: yad yada yad-anugrahi tat tada tatra sadhanam — *whatever*, *at whatever time*, is *what assists* — *that*, *at that time*, is the *means* *there*. Three indexicals in seven words: whatever, whenever, wherever. Nothing is an instrument; anything is an instrument when it is assisting. Anugrahi — assisting: the upakara by which the unnameable relation is tracked (3.3.5), by which capacities are brought about (3.6.6), and now by which karakas are assigned. The rule is the samuddesa's centre: means-hood is occasion-relative assistance, full stop.

3.7.13 unformalized

स्वशब्दैरभिधाने तु स धर्मो नाभिधीयते ।
विभक्त्यादिभिरेवासावुपकारः प्रतीयते ॥ १३ ॥

svaśabdairabhidhāne tu sa dharmo nābhidhīyate |
vibhaktyādibhirevāsāvupakāraḥ pratīyate || 13 ||

But when expression is by their own words, that property is not expressed; that assistance is understood by the case-endings alone.

How the karakas actually get into a sentence, and it is not by being named. Sva-sabdaih abhidhane tu sa dharmo na abhidhiyate: when expression is by *their own words* — when one says 'axe', 'Devadatta', 'rice' — *that property* (means-hood, agency, objecthood) is *not expressed*: the noun names the thing and says nothing about its role. Vibhakty-adibhih eva asau upakarah pratiyate: that *assistance* is understood by the *case-endings alone*. The role rides on the ending, not the stem: 'devadattena' says instrument, and 'devadatta' names a man. So the karaka-system is carried entirely by the inflectional apparatus, and the words that name the participants are silent about what they are doing. This is 3.7.9's asva-sabdesu made precise — the capacities have no words of their own — and it explains why the samuddesa is possible at all: the roles are a grammatical layer, assigned by endings, and one can therefore ask what the assignment tracks. The answer, since 3.7.3, is buddhi.

3.7.14 unformalized

निमित्तभावो भावानामुपकारार्थमाश्रितः ।
नतिरावर्जनेत्येवं सिद्धः साधनमिष्यते ॥ १४ ॥

nimittabhāvo bhāvānāmupakārārthamāśritaḥ |
natirāvarjanetyevaṃ siddhaḥ sādhanamiṣyate || 14 ||

The state of being an occasioning ground for things, resorted to for the sake of assistance — 'bending', 'turning toward' — thus established, it is held to be the means.

The definition of 3.7.1 restated through etymology. Nimitta-bhavah bhavanam upakara-artham asritah: the state of being an *occasioning ground* for things, resorted to *for the sake of assistance* (3.7.12's anugrahi: whatever assists is the means). And then two glosses of nimitta by its supposed roots: natih, *bending*; avarjana, *turning toward*. The etymologists derive nimitta from a bending-toward, an inclination of one thing toward another's completion — and Bhartrhari, who spent 2.170-175 showing that derivations are landmarks rather than foundations, uses this one for what it is worth: it pictures the means as a leaning, not a thing. Iti evam siddhah sadhanam isyate: *thus established*, it is held to be the means. What a sadhana *is*, on this account, is an inclination toward an action's completion — which is a relational posture, not an object.

3.7.15 unformalized

स तेभ्यो व्यतिरिक्तो वा तेषामात्मैव वा तथा ।
व्यतिरेकमुपाश्रित्य साधनत्वेन कल्प्यते ॥ १५ ॥

sa tebhyo vyatirikto vā teṣāmātmaiva vā tathā |
vyatirekamupāśritya sādhanatvena kalpyate || 15 ||

That is either distinct from them, or is their very self; having resorted to separation, it is constructed as means-hood.

The ontology of the means, and the two options are offered without decision. Sa tebhyo vyatirikto va: *that* (the nimitta-bhava, the being-a-ground) is either *distinct from* them — a further item over and above the things it relates; tesam atma eva va tatha: or is *their very self* — nothing but the things themselves under a description. The dispute is the tradition's standing one about relation (3.3.6's samyoga and samavaya: are they properties of the relata, or items?) and Bhartrhari declines it as he declined it there. But he says what happens in either case: vyatirekam upasritya sadhanatvena kalpyate — *having resorted to separation*, it is *constructed as* means-hood. Whatever the metaphysics, the *construction* requires taking the means as separate — the instrument must be held apart from the action to count as its instrument. So means-hood is a construction on a separation that may or may not answer to anything, and the grammar proceeds identically either way (2.179's licensed double description).

3.7.16-17 unformalized

संदर्शनं प्रार्थनायां व्यवसाये त्वनन्तरा ।
व्यवसायस्तथारंभे साधनत्वाय कल्पते ॥ १६ ॥
पूर्वस्मिन् या क्रिया सैव परस्मिन् साधनं मता ।
सन्दर्शने तु चैतन्यं विशिष्टं साधनं विदुः ॥ १७ ॥

saṃdarśanaṃ prārthanāyāṃ vyavasāye tvanantarā |
vyavasāyastathāraṃbhe sādhanatvāya kalpate || 16 ||
pūrvasmin yā kriyā saiva parasmin sādhanaṃ matā |
sandarśane tu caitanyaṃ viśiṣṭaṃ sādhanaṃ viduḥ || 17 ||

(16) Perceiving, in the case of seeking; and next, in the case of resolving; and resolve, likewise, in undertaking — serve as means. (17) The very action that was in the prior case is held to be the means in the subsequent; but in perceiving, they know a distinguished consciousness to be the means.

The chain of practical reason, analysed as a series of karaka-assignments. Sandarsanam prarthanayam: *perceiving* is the means in *seeking* — one wants what one has seen; vyavasaye tv anantara: and *next* (perceiving-and-seeking) in *resolving*; vyavasayah tatha arambhe: and *resolve* likewise in *undertaking*. Seeing, wanting, deciding, doing: each stage is the instrument of the next. And the principle: purvasmin ya kriya sa eva parasmin sadhanam mata — *the very action* that was in the *prior* case is held to be the *means* in the *subsequent*. So the same item is action here and instrument there, purely by where in the chain one stands (2.429's cooking that is object, instrument and sadhya at once, assigned by the sentence's construction). And the chain's first term: sandarsane tu caitanyam visistam sadhanam viduh — in *perceiving*, they know a *distinguished consciousness* to be the means. The regress stops at consciousness, which is not itself a stage.

3.7.18 unformalized

निष्पत्तिमात्रे कर्तृत्वं सर्वत्रैवास्ति कारके ।
व्यापारभेदापेक्षायां करणादित्वसंभवः ॥ १८ ॥

niṣpattimātre kartṛtvaṃ sarvatraivāsti kārake |
vyāpārabhedāpekṣāyāṃ karaṇāditvasaṃbhavaḥ || 18 ||

In the mere completing, agency exists in every karaka whatever; when a difference of operation is depended upon, the possibility of being instrument and the rest arises.

The samuddesa's most consequential structural doctrine. Nispatti-matre kartrtvam sarvatra eva asti karake: in the *mere completing*, *agency* exists in *every karaka whatever*. Every participant is an agent: the axe cuts, the wood splits, the fire cooks — each does its own thing, and each is doing it. Agency is not the privilege of the grammatical subject; it is what every contributor exercises. And then: vyapara-bheda-apeksayam karanaditva-sambhavah — when a *difference of operation* is *depended upon*, the possibility of being *instrument and the rest* arises. The differentiation into instrument, object, locus and the rest appears only when one attends to *how* each contributes — which operation each performs. So the karaka-system is a hierarchy imposed on a democracy: all are agents, and the assignment of subordinate roles is a matter of which operation one is attending to. The verses following will make this the samuddesa's organizing insight.

3.7.19 unformalized

पुत्रस्य जन्मनि यथा पित्रोः कर्तृत्वमुच्यते ।
अयमस्यामियं त्वस्मादिति भेदो विवक्षया ॥ १९ ॥

putrasya janmani yathā pitroḥ kartṛtvamucyate |
ayamasyāmiyaṃ tvasmāditi bhedo vivakṣayā || 19 ||

As, in the birth of a son, agency is spoken of for both parents — 'he in her', 'she from him' — the difference is by intention.

The doctrine of 3.7.18 in the plainest possible case. Putrasya janmani yatha pitroh kartrtvam ucyate: as, in the *birth of a son*, *agency* is spoken of for *both parents* — both are agents of the begetting; neither is the instrument of the other. And yet the grammar must assign roles: ayam asyam — *he in her* (locative: she the locus, he the agent); iyam tu asmat — *she from him* (ablative: he the source, she the agent). Two sentences, the same event, opposite assignments. Iti bhedo vivaksaya: the *difference is by intention*. Nothing in the begetting makes the mother a locus rather than an agent; the speaker's construing does (2.428: the speaker alone intends the sadhya-sadhana structure). The example is chosen for its symmetry — no one can claim that biology settles which parent is the karta — and it demonstrates 3.7.18 exactly: both are agents in the mere completing, and the role-difference is imposed by attention.

3.7.20-21 unformalized

गुणक्रियाणां कर्तारः कर्त्रा न्यक्कृतशक्तयः ।
न्यक्तायामपि सम्पूर्णैः स्वैर्व्यापारैः समन्विताः ॥ २० ॥
करणत्वादिभिर्ज्ञाताः क्रियाभेदानुपातिभिः ।
स्वातन्त्र्यमुत्तरं लब्ध्वा प्रधाने यान्ति कर्तृताम् ॥ २१ ॥

guṇakriyāṇāṃ kartāraḥ kartrā nyakkṛtaśaktayaḥ |
nyaktāyāmapi sampūrṇaiḥ svairvyāpāraiḥ samanvitāḥ || 20 ||
karaṇatvādibhirjñātāḥ kriyābhedānupātibhiḥ |
svātantryamuttaraṃ labdhvā pradhāne yānti kartṛtām || 21 ||

(20) The agents of qualities and actions, their capacities subordinated by the agent — even when subordinated, they are furnished with their own complete operations. (21) Known by instrument-hood and the rest, which follow upon the differences of action, having obtained a subsequent independence, they attain agency with respect to the principal.

The mechanism of subordination, and it is precisely described. (20): guna-kriyanam kartarah — the *agents* of qualities and actions (every participant, each an agent by 3.7.18) — kartra nyakkrta-saktayah: their *capacities subordinated by the agent* — nyakkrta: pressed down, demoted, made to bow. The principal agent's presence demotes the rest. And the crucial concession: nyaktayam api sampurnaih svaih vyaparaih samanvitah — *even when subordinated*, they are furnished with their *own complete operations*: the axe's cutting is not diminished by the woodsman's being the karta; it is complete, and it is the axe's. Demotion is a matter of grammatical rank, not of diminished doing. (21): karanatvadibhih jnatah kriya-bheda-anupatibhih — *known* by *instrument-hood* and the rest, which *follow upon* the differences of *action*; svatantryam uttaram labdhva pradhane yanti kartrtam: having obtained a *subsequent independence*, they *attain agency* with respect to the *principal*. The subordinate karakas regain agency at the next level up: within its own operation, the axe is svatantra. So the system is a nested hierarchy of agents, each independent in its own sphere and demoted in the sphere above.

3.7.22-23 unformalized

यथा राज्ञा नियुक्तेषु योद्धृत्वं योद्धृषु स्थितम् ।
तेषु वृत्तौ तु लभते राजा जयपराजयौ ॥ २२ ॥
तथा कर्त्रा नियुक्तेषु सर्वेष्वेकार्थकारिषु ।
कर्तृत्वं करणादित्वैरुत्तरं न विरुध्यते ॥ २३ ॥

yathā rājñā niyukteṣu yoddhṛtvaṃ yoddhṛṣu sthitam |
teṣu vṛttau tu labhate rājā jayaparājayau || 22 ||
tathā kartrā niyukteṣu sarveṣvekārthakāriṣu |
kartṛtvaṃ karaṇāditvairuttaraṃ na virudhyate || 23 ||

(22) As, among those appointed by the king, the state of being fighters stands in the fighters; but when they act, it is the king who obtains victory and defeat. (23) So, among all those appointed by the agent, doing one work — agency, subsequent, is not in conflict with instrument-hood and the rest.

The political simile, and it is the samuddesa's finest. (22): rajna niyuktesu — among those *appointed by the king* — yoddhrtvam yoddhrsu sthitam: the *state of being fighters* stands *in the fighters*: the soldiers do the fighting, entirely and without remainder; the king is not fighting. Tesu vrttau tu labhate raja jaya-parajayau: but *when they act*, it is *the king* who obtains *victory and defeat*. The doing is theirs; the outcome is predicated of him. (23) applies it: tatha kartra niyuktesu sarvesu eka-artha-karisu — so among *all those appointed by the agent*, *doing one work* — kartrtvam karanaditvaih uttaram na virudhyate: *agency*, *subsequent*, is *not in conflict* with instrument-hood and the rest. The axe is an agent (it cuts) and an instrument (in the woodsman's sentence), as the soldier is a fighter and the king's instrument, and no contradiction arises because the two predications belong to different levels. The simile explains what 3.7.20-21 asserted, and it explains why we say 'the king conquered' without lying and 'the soldiers fought' without contradicting ourselves.

3.7.24 unformalized

अनाश्रिते तु व्यापारे निमित्तं हेतुरिष्यते ।
आश्रितावधिभावं तु लक्षणे लक्षणं विदुः ॥ २४ ॥

anāśrite tu vyāpāre nimittaṃ heturiṣyate |
āśritāvadhibhāvaṃ tu lakṣaṇe lakṣaṇaṃ viduḥ || 24 ||

But when the operation is not resorted to, the occasioning ground is held to be the hetu; when the state of being a limit is resorted to, they know the mark to be in the case of the mark.

Two karakas distinguished by what they do not do. Anasrite tu vyapare nimittam hetur isyate: when the *operation is not resorted to* — where the ground contributes without any operation being attributed to it — the occasioning ground is held to be *hetu*, the cause (Panini's karana in its causal sense, and the ablative of cause: 'he suffers *from* illness'). Asrita-avadhi-bhavam tu laksane laksanam viduh: when the state of being a *limit* is resorted to, they know the *mark* to be in the case of the *mark*. So hetu and laksana are the two karakas that operate without operating: one contributes causally without a vyapara of its own, the other locates by being a boundary (3.6.2's avadhi: the limit from which direction separates). The verse is fine taxonomy: having established that every karaka is an agent with its own complete operation (3.7.20), it identifies the two that are exceptions — and they are the two that do not act.

3.7.25 unformalized

द्रव्यादिविषयो हेतुः कारकं नियतक्रियम् ।
कर्ता कर्त्रन्तरापेक्षः क्रियायां हेतुरिष्यते ॥ २५ ॥

dravyādiviṣayo hetuḥ kārakaṃ niyatakriyam |
kartā kartrantarāpekṣaḥ kriyāyāṃ heturiṣyate || 25 ||

The hetu has substance and the rest for its domain; the karaka has a fixed action. An agent depending on another agent is held to be the hetu in the action.

The hetu/karaka distinction sharpened. Dravyadi-visayo hetuh: the *hetu* has *substance and the rest* for its *domain* — it can be anything: a substance, a quality, an action, so long as it grounds. Karakam niyata-kriyam: the *karaka* has a *fixed action* — a karaka is defined by its determinate contribution to a determinate act (3.7.1's samarthyam in the completing). And the causative: karta kartr-antara-apeksah kriyayam hetur isyate — an *agent depending on another agent* is held to be the *hetu* in the action. This is the grammar of causation: when Devadatta makes Yajnadatta cook, Yajnadatta is an agent who depends on another agent, and Devadatta is the hetu. So the causative construction is analysed as an agent-over-agent relation — which is exactly 3.7.22's king and soldiers, now given its morphological home in the nic-affix.

3.7.26 unformalized

क्रियायै करणन्तस्य दृष्टः प्रतिनिधिस्तथा ।
हेत्वर्था तु क्रिया तस्मान्न स प्रतिनिधीयते ॥ २६ ॥

kriyāyai karaṇantasya dṛṣṭaḥ pratinidhistathā |
hetvarthā tu kriyā tasmānna sa pratinidhīyate || 26 ||

For an action, a substitute for the instrument is seen; but the action is for the sake of the hetu — therefore that one is not substituted.

A test that discriminates the two. Kriyayai karanantasya drstah pratinidhih tatha: for an *action*, a *substitute for the instrument* is *seen* — the axe failing, use a hatchet; instruments are replaceable (3.1.78's substitution-rule discriminating what may be substituted from what may not). Hetv-artha tu kriya: but the *action* is *for the sake of* the *hetu* — the hetu is not a means the action uses; the action exists for it. Tasmat na sa pratinidhiyate: *therefore* that one is *not substituted*. One does not substitute a different reason for acting and call it the same act. The test is the same one 3.1.78-79 ran on universals and qualities (what stands near the capacity cannot be replaced; what stands at a remove can), and it places hetu and karana precisely: the instrument serves the action and is fungible; the hetu is what the action serves and is not.

3.7.27 unformalized

प्रतिलोम्यानुलोम्याभ्यां हेतुरर्थस्य साधकः ।
तादर्थ्यमानुलोम्येन हेतुत्वानुगतं तु तत् ॥ २७ ॥

pratilomyānulomyābhyāṃ heturarthasya sādhakaḥ |
tādarthyamānulomyena hetutvānugataṃ tu tat || 27 ||

The hetu is the accomplisher of the meaning by way of the contrary and by way of the favourable; and purposiveness, by way of the favourable, is what follows upon hetu-hood.

The hetu's two modes. Pratilomya-anulomyabhyam hetur arthasya sadhakah: the *hetu* is the *accomplisher of the meaning* by way of the *contrary* and by way of the *favourable* — one acts *because of* a thing one wants (favourable) and *because of* a thing one flees (contrary): both are reasons, and the ablative of cause covers both ('from desire', 'from fear'). Tadarthyam anulomyena hetutva-anugatam tu tat: and *purposiveness* — tadarthya, being-for-that, the dative's sense ('for the sake of') — *by way of the favourable*, is what *follows upon hetu-hood*. So the dative of purpose is a special case of the hetu: the favourable reason, taken as goal. The verse quietly unifies two of Panini's cases (the ablative of cause and the dative of purpose) under one karaka-analysis, distinguished only by direction — and the unification follows from 3.7.26's finding that the hetu is what the action is *for*.

3.7.28 unformalized

सर्वत्र सहजा शक्तिर्यावद्द्रव्यमस्थिता ।
क्रियाकाले त्वभिव्यक्तेराश्रयादुपरकारिणी ॥ २८ ॥

sarvatra sahajā śaktiryāvaddravyamasthitā |
kriyākāle tvabhivyakterāśrayāduparakāriṇī || 28 ||

Everywhere the innate capacity stands as long as the substance does; but at the time of the action, from its manifestation, it becomes assisting through its substratum.

The temporality of capacity, and it settles a real puzzle. Sarvatra sahaja saktih yavad-dravyam avasthita: *everywhere*, the *innate* capacity stands *as long as the substance* does — the axe's cutting-power is in it while it is an axe, whether or not anyone is cutting; capacities do not come and go with their exercise. Kriya-kale tu abhivyakteh asrayad upakarini: but *at the time of the action*, from its *manifestation*, it becomes *assisting* through its *substratum*. So the capacity is permanent and its *assisting* is occasional — it is always there and it helps only when manifested (1.44's manifestation-doctrine: the sphota permanent, the sounds occasioning). This resolves how a karaka can be both a standing property of a thing (the axe is an instrument even in the shed) and an occasion-relative posting (3.7.12's whatever-whenever-wherever): the capacity stands, the assisting happens.

3.7.29 unformalized

कुड्यस्यावरणे शक्तिरस्यादीनां विदारणे ।
सर्वदा स तु सन् धर्मः क्रियाकाले निरूप्यते ॥ २९ ॥

kuḍyasyāvaraṇe śaktirasyādīnāṃ vidāraṇe |
sarvadā sa tu san dharmaḥ kriyākāle nirūpyate || 29 ||

The wall's capacity is in obstructing, that of a sword and the like in cutting; that property, being always existent, is determined at the time of the action.

The doctrine of 3.7.28 in two homely cases. Kudyasya avarane saktih: the *wall's* capacity is in *obstructing* — a wall is for stopping, and it is for stopping when nothing is being stopped. Asyadinam vidarane: that of a *sword* and the like in *cutting*. Sarvada sa tu san dharmah: that property, being *always existent* — the wall does not acquire obstruction-power when someone walks into it. Kriya-kale nirupyate: it is *determined* at the *time of the action* — nirupyate: ascertained, fixed, brought into determinacy (3.3.54's sva-rupena anirupitam: the object's own form left undetermined by the word). So the capacity is eternal and its determination is occasional, and the karaka-assignment records the determination. The examples are chosen for their obviousness: nobody thinks the wall becomes obstructive on impact, and nobody can say what the wall's obstruction *is* except by reference to a walking-into that has not yet happened.

3.7.30 unformalized

स्वाङ्गसंयोगिनः पाशा दैत्यानां वारुणा यथा ।
व्यज्यन्ते विजिगीषूणां द्रव्याणां शक्तयस्तथा ॥ ३० ॥

svāṅgasaṃyoginaḥ pāśā daityānāṃ vāruṇā yathā |
vyajyante vijigīṣūṇāṃ dravyāṇāṃ śaktayastathā || 30 ||

As the Varuna-nooses of the demons, conjoined with their own limbs, are made manifest — so are the capacities of substances made manifest for those who seek to conquer.

The samuddesa's most vivid image for manifestation. The varuna-pasa — Varuna's nooses, the god's binding weapon of the Vedic hymns, which the demons carry sva-anga-samyoginah, *conjoined with their own limbs*: the noose is in them, part of them, unseen until it binds. Vyajyante: they are *made manifest* — the noose does not arrive; it appears out of what was already there. Dravyanam saktayah tatha vijigisunam: so the *capacities of substances*, for those who *seek to conquer*. The image completes 3.7.28-29 with a picture: capacities are latent bindings, carried in the thing, showing when the occasion draws them out. And vijigisunam — for those who seek to conquer — quietly returns the samuddesa to its own subject: the capacities manifest *for* someone with a purpose, and a purpose is what makes an event into a structure of agent and instrument and object. The karaka-system is the world's latent nooses, drawn out by an intention (3.7.2's kvacit kimcid vivaksyate: somewhere, something is intended).

3.7.31 unformalized

तैक्ष्ण्यगौरवकाठिन्यसंस्थानैः स्वैरसिर्यदा ।
छेद्यं प्रति व्याप्रियते शक्तिमान् गृह्यते तदा ॥ ३१ ॥

taikṣṇyagauravakāṭhinyasaṃsthānaiḥ svairasiryadā |
chedyaṃ prati vyāpriyate śaktimān gṛhyate tadā || 31 ||

When the sword, by its own sharpness, weight, hardness and shape, is engaged toward what is to be cut — then it is apprehended as capacity-possessing.

The manifestation of 3.7.30 given its physics. Taiksnya-gaurava-kathinya-samsthanaih svaih: by its *own* *sharpness*, *weight*, *hardness* and *shape* — four properties, each permanently the sword's (3.7.28's sahaja sakti yavad-dravyam avasthita: the innate capacity standing as long as the substance). Asir yada chedyam prati vyapriyate: when the *sword* is *engaged toward* what *is to be cut*. Saktiman grhyate tada: *then* it is *apprehended* as *capacity-possessing*. The verse locates the apprehension precisely: not when the sword exists (it always has the four properties), and not after the cutting (that is too late), but when it is *engaged toward* — vyapriyate, in the middle of its being brought to bear. Capacity is apprehended in the engaging. And the four properties are named because they are what the engaging draws on: the karana-hood is not a fifth property alongside them but their being brought to bear on something.

3.7.32 unformalized

प्राङ् निमित्तान्तरोद्भूतं क्रियायाः कैश्चिदिष्यते ।
साधनं सहजं कैश्चित् क्रियान्यैः पूर्वमिष्यते ॥ ३२ ॥

prāṅ nimittāntarodbhūtaṃ kriyāyāḥ kaiścidiṣyate |
sādhanaṃ sahajaṃ kaiścit kriyānyaiḥ pūrvamiṣyate || 32 ||

Some hold the means of the action to be produced beforehand from another occasioning ground; some hold it innate; others hold the action to be prior.

Three positions on the means' origin, catalogued. Prak nimittantara-udbhutam kriyayah kaiscid isyate: some hold the means *produced beforehand from another occasioning ground* — the sword's cutting-power was made by the smith; capacities have causal histories. Sahajam kaiscit: some hold it *innate* — the capacity is the thing's own, unmade (3.7.28's sahaja). Kriya anyaih purvam isyate: *others* hold the *action* to be *prior* — the capacity is constituted by the action rather than preceding it, so that the sword becomes an instrument by cutting. Three genuine options, and Bhartrhari does not choose. He does not need to: 3.7.28-31 have established that whatever the capacity's origin, it is *apprehended in the engaging*, and that is all the karaka-system requires. The origin-dispute is 3.1.107's substrate-dispute again — canvassed, tolerated, and shown not to touch the doctrine that depends on it.

3.7.33 unformalized

प्रवृत्तिरेव प्रथमं क्वचिदप्यनपाश्रिता ।
शक्तिरेकाधिकरणे स्रोतोवदपकर्षति ॥ ३३ ॥

pravṛttireva prathamaṃ kvacidapyanapāśritā |
śaktirekādhikaraṇe srotovadapakarṣati || 33 ||

The operation itself is first, somewhere not resorting to anything at all; the capacity, in a single substratum, draws downward, like a current.

The third option of 3.7.32 developed, with an image. Pravrttir eva prathamam kvacid api anapasrita: the *operation itself* is *first*, somewhere *not resorting to* anything at all — the doing is primary, leaning on no prior capacity: sometimes there simply *is* a going, and the goer is read off it. Saktir eka-adhikarane sroto-vad apakarsati: the *capacity*, in a *single substratum*, *draws downward*, *like a current*. The image is exact: a river's current is not a thing in the water that makes it flow; it is the flowing, and it carries what is in it. So capacity is the operation's own momentum, drawing its bearer along — which inverts the ordinary picture (thing has power, power produces act) into: act runs, and what runs in it is called its power. The verse is the samuddesa's most process-metaphysical moment, and it is offered as one option among three.

3.7.34 unformalized

अपूर्वं कालशक्तिं वा क्रियां वा कालमेव वा ।
तमेवंलक्षणं भावं केचिदाहुः कथंचन ॥ ३४ ॥

apūrvaṃ kālaśaktiṃ vā kriyāṃ vā kālameva vā |
tamevaṃlakṣaṇaṃ bhāvaṃ kecidāhuḥ kathaṃcana || 34 ||

Some, in some fashion, call that entity of such a character the unprecedented, or the capacity of time, or action, or time itself.

The current of 3.7.33 named, and the names do not agree. Apurvam: the *unprecedented* — the Mimamsa's technical term for the unseen ritual potency that bridges rite and remote fruit (2.119's test-word). Kala-saktim va: or the *capacity of time* (1.3's kala-sakti, on which the six modifications rest). Kriyam va: or *action*. Kalam eva va: or *time itself*. Tam evam-laksanam bhavam kecid ahuh kathamcana: some, *in some fashion*, call *that entity of such a character* by these names. Kathamcana — 'in some fashion': the shrug is 2.171's (kaiscit kathamcid akhyatah, of the etymologists' derivations). Four schools, four names, one phenomenon: the drawing-onward that makes a sequence of events into an action. Bhartrhari records the disagreement and adds nothing, because 3.6.16 already ruled that where every candidate for a first term turns out to be another differentiated item, the reasonable course is to stop looking.

3.7.35 unformalized

नित्याः षट् शक्तयोऽन्येषां भेदाभेदसमन्विताः ।
क्रियासंसिद्धयेऽर्थेषु जातिवत् समवस्थिताः ॥ ३५ ॥

nityāḥ ṣaṭ śaktayo'nyeṣāṃ bhedābhedasamanvitāḥ |
kriyāsaṃsiddhaye'rtheṣu jātivat samavasthitāḥ || 35 ||

For others, six capacities are eternal, endowed with difference and non-difference; for the accomplishing of action they stand established among objects, like universals.

The doctrine that will organize the rest of the samuddesa. Nityah sat saktayah anyesam: for *others*, *six* capacities are *eternal* — the six karakas (agent, object, instrument, recipient, source, locus) taken as permanent powers, not as occasional postings. Bheda-abheda-samanvitah: *endowed with difference and non-difference* — they are distinct from each other and not distinct from what has them (1.2's saktis, apthak and yet functioning prthaktvena iva). Kriya-samsiddhaye arthesu jativat samavasthitah: *for the accomplishing of action* they stand established *among objects*, *like universals*. The comparison to universals is the point: as cowhood stands in cows eternally and manifests in each (3.1.25-26's jati as self-manifesting power), so agenthood stands in things and manifests when they act. The six become a fixed ontological furniture — and the samuddesa will spend its remaining verses both using this scheme and undermining it.

3.7.36 unformalized

द्रव्याकारादिभेदेन ताश्चापरिमिता इव ।
दृश्यन्ते तत्त्वसायां तु षट् शक्तीर्नातिवर्तते ॥ ३६ ॥

dravyākārādibhedena tāścāparimitā iva |
dṛśyante tattvasāyāṃ tu ṣaṭ śaktīrnātivartate || 36 ||

And by difference of substance, configuration and the rest, they are seen as though unlimited; but in the settling of the reality, one does not transgress the six capacities.

The six defended against their apparent multiplicity. Dravya-akaradi-bhedena tah ca aparimita iva drsyante: by *difference of substance*, *configuration* and the rest, they are seen *as though unlimited* — every instrument is a different instrument, every agent a different agent; the world's roles look innumerable. Iva: as though. Tattva-sayyam tu sat saktih na ativartate: but in the *settling of the reality*, one does *not transgress* the *six* capacities. The unlimited variety reduces to six modes. This is the ordinary structure of the work's reductions (2.250's one word carved by nimittas into unlimited meanings; 3.1.33's one satta carved by relata into all the universals) — and its application here yields the karaka-system's central claim: however many participants an event has, they occupy six slots, and the slots are exhaustive.

3.7.37 unformalized

निमित्तभेदादेकैव भिन्ना शक्तिः प्रतीयते ।
षोढा कर्तृत्वमेबाहुस्तत्प्रवृत्तेर्निबन्धनम् ॥ ३७ ॥

nimittabhedādekaiva bhinnā śaktiḥ pratīyate |
ṣoḍhā kartṛtvamebāhustatpravṛtternibandhanam || 37 ||

From difference of occasioning ground, one and the same capacity is understood as differentiated; they call agency alone six-fold — it is the ground of that operation.

The reduction pressed one step further, and this is the samuddesa's deepest structural claim. Nimitta-bhedad eka eva bhinna saktih pratiyate: from *difference of occasioning ground*, *one and the same* capacity is understood as *differentiated* — the six are not six powers but one, carved by nimittas (2.250-251's mechanism exactly). And the identification: sodha kartrtvam eva ahuh: they call *agency alone* *six-fold*. Not: there are six capacities of which agency is one. Rather: there is *agency*, and it appears sixfold. Tat-pravrtteh nibandhanam: it is the *ground of that operation*. This completes 3.7.18 (in the mere completing, agency exists in every karaka whatever): if every karaka is an agent, then the six karakas are six modes of *one* thing, agency, differentiated by the occasions on which it is attended to. Instrument-hood is agency-under-a-nimitta. The whole architecture of Sanskrit syntax reduces to one power, sixfold-appearing.

3.7.38 unformalized

तत्त्वे वा व्यतिरेके वा व्यतिरिक्तं तदुच्यते ।
शब्दप्रमाणको लोकः स शास्त्रेणानुगम्यते ॥ ३८ ॥

tattve vā vyatireke vā vyatiriktaṃ taducyate |
śabdapramāṇako lokaḥ sa śāstreṇānugamyate || 38 ||

Whether in reality or in separation, that is spoken of as separate; the world has the word for its authority, and it is followed along by the science.

The methodological placement of 3.7.37's reduction. Tattve va vyatireke va vyatiriktam tad ucyate: *whether* in *reality* or in *separation* — whether the six are really distinct or only construed as distinct (3.7.15's undecided ontology) — that is *spoken of as separate*: language treats them as six regardless. And the ground of the treatment: sabda-pramanako lokah — the *world* has the *word* for its *authority*: people's ontology is their grammar (2.328's principle: what a word means is what is understood, and there is no other criterion). Sa sastrena anugamyate: *it* is *followed along by the science* — and the grammar follows the world, rather than correcting it (2.142's tat prajnah na vikalpayet; 3.6.25's lokah tatra anugamyate). So the six karakas are in the sastra because they are in the loka, and they are in the loka because that is how words work. The reduction to one agency (3.7.37) is true and changes nothing, because the science's business is the world's usage.

3.7.39 unformalized

परमार्थे तु नैकत्वं पृथक्त्वाद्भिन्नलक्षणम् ।
पृथक्त्वैकत्वरूपेण तत्त्वमेव प्रकाशते ॥ ३९ ॥

paramārthe tu naikatvaṃ pṛthaktvādbhinnalakṣaṇam |
pṛthaktvaikatvarūpeṇa tattvameva prakāśate || 39 ||

But in the highest sense there is no oneness whose character is distinct from separateness; reality itself shines forth in the form of separateness and oneness.

The reduction of 3.7.37 given its metaphysical honesty. Paramarthe tu na ekatvam prthaktvad bhinna-laksanam: in the *highest sense* there is *no oneness* whose character is *distinct from separateness* — the 'one agency' of 3.7.37 is not a unity standing over against the six; oneness and separateness are not two conditions (3.6.26's finding: neither exists without the other, and in the highest sense the difference does not exist at all). Prthaktva-ekatva-rupena tattvam eva prakasate: *reality itself* *shines forth* in the *form of separateness and oneness*. So one must not read 3.7.37 as saying that the truth is one agency and the six are illusion. The truth shines as both — six, and one — and neither is the correction of the other. This is 3.2.12-13's paired tetralemmas applied to the karaka-system: everything true of it is true both ways, and the pair is the doctrine.

3.7.40-41 unformalized

यत् पृथक्त्वमसंदिग्धं तदेकत्वान्न भिद्यते ।
यदेकत्वमसंदिग्धं तत् पृथक्त्वान्न भिद्यते ॥ ४० ॥
द्यौः क्षमा वायुरादित्यः सागराः सरितो दिशः ।
अन्तः करणतत्त्वस्य भागा बहिरवस्थिताः ॥ ४१ ॥

yat pṛthaktvamasaṃdigdhaṃ tadekatvānna bhidyate |
yadekatvamasaṃdigdhaṃ tat pṛthaktvānna bhidyate || 40 ||
dyauḥ kṣamā vāyurādityaḥ sāgarāḥ sarito diśaḥ |
antaḥ karaṇatattvasya bhāgā bahiravasthitāḥ || 41 ||

(40) That separateness which is beyond doubt is not different from oneness; that oneness which is beyond doubt is not different from separateness. (41) Sky, earth, wind, sun, oceans, rivers, directions — these are portions of the inner-organ reality, standing outside.

(40) states the mutual non-difference with a symmetry that is itself the argument: yat prthaktvam asandigdham tad ekatvan na bhidyate — that *separateness* which is *beyond doubt* is *not different from* oneness; and the mirror. Whatever is indubitably many is not other than one, and whatever is indubitably one is not other than many. Then (41) delivers, without warning, one of the most remarkable sentences in the work: dyauh ksama vayur adityah sagarah saritah disah — *sky*, *earth*, *wind*, *sun*, *oceans*, *rivers*, *directions* — antah-karana-tattvasya bhagah bahir avasthitah: are *portions of the inner-organ reality*, *standing outside*. The cosmos is the inner organ's parts, externalized. This is 3.7.6's doctrine (the intellect's operation superimposed on objects) at maximum scale, and 3.6.23's declined option (a property of the inner organ shining forth outwardly) here affirmed without the hedge. The list is deliberately elemental — the things anyone would call most obviously *out there* — and they are named as the mind's own portions standing outside.
contested: 41's antah-karana-tattvasya bhagah admits a weaker reading ('portions of the reality that is [known by] the inner organ') and a stronger ('portions of the inner-organ-reality itself'). The stronger is the natural construal and coheres with 3.7.6 and 3.7.3; we adopt it while noting that the verse stands without argument here

3.7.42 unformalized

कालविच्छेदरूपेण तदेवैकमवस्थितम् ।
स ह्यपूर्वापरो भागः पररूपेण लक्ष्यते ॥ ४२ ॥

kālavicchedarūpeṇa tadevaikamavasthitam |
sa hyapūrvāparo bhāgaḥ pararūpeṇa lakṣyate || 42 ||

That very one stands, single, in the form of the division of time; for that portion, being without prior and posterior, is marked by another's form.

The cosmos of 3.7.41 given its temporal aspect. Kala-viccheda-rupena tad eva ekam avasthitam: *that very one* stands, *single*, in the *form of the division of time* — the same inner-organ reality that stands outside as sky and sun stands also as time's divisions. Sa hi apurvaparo bhagah: for that *portion* is *without prior and posterior* — the portion itself has no before and after (3.1.37's sequence being Being's own self, undivided; 3.2.8's no difference of time in the real). Para-rupena laksyate: it is *marked by another's form* — its temporal divisions come from elsewhere, as the mountain's sides come from the sun (3.6.12). So time's articulation is conferred on what has none, exactly as space's was (3.6.13's bhaga-prakalpana-sakti), and both articulations are of one thing which is the inner organ's own reality. The Kala-samuddesa, two samuddesas ahead, will take this up at length.

3.7.43 unformalized

दृष्टो ह्यव्यतिरेकेऽपि व्यतिरेकोऽन्वयेऽसति ।
वृक्षाद्यर्थान्लयस्तस्माद्विभक्त्यर्थोऽन्य इष्यते ॥ ४३ ॥

dṛṣṭo hyavyatireke'pi vyatireko'nvaye'sati |
vṛkṣādyarthānlayastasmādvibhaktyartho'nya iṣyate || 43 ||

For separation is seen even where there is no separation, when there is no syntactic connection; therefore the meaning of the case-ending is held to be other than the meanings of 'tree' and the rest.

The argument that case-meanings are distinct from stem-meanings. Drsto hi avyatireke 'pi vyatireko 'nvaye 'sati: *separation* is seen even where there is *no separation*, when there is *no syntactic connection* — one can distinguish things that are not in fact distinct, wherever the construction does not link them. Vrksadyarthan layas tasmad vibhakty-artho 'nya isyate: therefore the *meaning of the case-ending* is held to be *other than* the meanings of 'tree' and the rest. The reasoning: the stem 'vrksa' gives a tree; the ending gives a role; and since the role can vary while the tree stays fixed ('the tree', 'to the tree', 'from the tree'), the two meanings are separable — which the distributional method (2.12's anvaya-vyatireka) certifies. This is the grammatical basis for the whole samuddesa: karaka-meanings are borne by the endings (3.7.13), and here that is established by method rather than asserted.
contested: the transcription's vrksadyarthan layah is corrupt; we translate the evident sense (the meanings of 'tree' etc. being distinguished from the case-meaning) and flag the fault

3.7.44 unformalized

सामान्यं कारकं तस्य सप्ताद्या भेदयोनयः ।
षट् कर्माख्यादिभेदेन शेषभेदस्तु सप्तमी ॥ ४४ ॥

sāmānyaṃ kārakaṃ tasya saptādyā bhedayonayaḥ |
ṣaṭ karmākhyādibhedena śeṣabhedastu saptamī || 44 ||

The karaka is the general; the seventh and the rest are the wombs of its differences: six, by the difference of what is called karman and so on — but the seventh is the difference of the remainder.

The system's formal architecture, and it is Panini's. Samanyam karakam: the *karaka* is the *general* — karaka is the genus, and the individual roles are its species (Panini 1.4.23: karake, the heading under which all the role-definitions fall). Tasya saptadya bheda-yonayah: the *seventh* and the rest are the *wombs of its differences* — the case-endings are what generate the distinctions. Sat karma-akhyadi-bhedena: *six*, by the difference of what is called *karman* and so on — the six karakas proper. Sesa-bhedas tu saptami: but the *seventh* is the *difference of the remainder*. This last is precise and it is Panini's own architecture: the genitive (sasthi) is not a karaka at all; it is the sesa, the *remainder* case (2.3.50: sasthi sese), used where no karaka-relation holds. So the system has six roles plus a residue-case for everything else — and the residue is where relation as such lives (3.3.3: the word-meaning connection understood through the sixth case; 2.197-198's genitive for relations whose action has withdrawn).

3.7.45 unformalized

निर्वर्त्यं च विकार्यं च प्राप्यं चेति त्रिधा मतम् ।
तत्रेप्सिततमं कर्म चतुर्धान्यत्तु कल्पितम् ॥ ४५ ॥

nirvartyaṃ ca vikāryaṃ ca prāpyaṃ ceti tridhā matam |
tatrepsitatamaṃ karma caturdhānyattu kalpitam || 45 ||

The object is held to be three-fold: what is to be produced, what is to be modified, and what is to be reached. Among these, the karman is what is most desired; but another is constructed four-fold.

The object's classical taxonomy. Nirvartyam ca vikaryam ca prapyam ca iti tridha matam: the *object* is held to be *three-fold* — nirvartya, *what is to be produced* (the mat that the weaving makes: 3.7.7's janyamana karman, the object that does not exist until the act completes); vikarya, *what is to be modified* (the gold that the smith works: it exists and is changed); prapya, *what is to be reached* (the village one goes to: it exists and is unchanged). The three exhaust the ways an action can bear on an object. Tatra ipsitatamam karma: among these, the *karman* is *what is most desired* — Panini's own definition (1.4.49: kartur ipsitatamam karma, the object is what the agent most desires), and note that it is defined by the *agent's desire*, not by any feature of the thing (3.4.3's dravya defined by pronominal referability; 3.7.3's karakas grounded in buddhi). Caturdha anyat tu kalpitam: but *another* is constructed *four-fold* — a rival taxonomy, unnamed. The object's classification is doctrine-relative, as everything in this kanda has been.

3.7.46 unformalized

औदासीन्येन यत् प्राप्यं यच्च कर्तुरनीप्सितम् ।
संज्ञान्तरैरनाख्यातं यद्यच्चाप्यन्यपूर्वकम् ॥ ४६ ॥

audāsīnyena yat prāpyaṃ yacca karturanīpsitam |
saṃjñāntarairanākhyātaṃ yadyaccāpyanyapūrvakam || 46 ||

What is reached by way of indifference; what is not desired by the agent; what is not named by other designations; and whatever has another as its antecedent —

The object's residual categories, and they qualify 3.7.45's definition sharply. Panini defined the karman as ipsitatamam — *what is most desired* by the agent. But four classes of object are manifestly not desired. Audasinyena yat prapyam: what is reached by way of *indifference* — the grass one treads on while walking to the village (Panini's own example at 1.4.51's akathitam ca provisions); yat ca kartuh anipsitam: what is *not desired* by the agent — the poison one swallows unknowingly, the thief one meets on the road. Samjnantaraih anakhyatam: what is *not named by other designations* — the residue that no other karaka-term covers (the akathita, 'unstated', of Panini's rule). Yat ca api anya-purvakam: and whatever has *another as its antecedent*. So the definition by desire fails for most objects, and the grammar has had to build a residue-category to catch them. The verse is the samuddesa's honest report that its central karaka is defined by a criterion its own instances violate — and 3.7.3's answer stands ready: the karakas are buddhi-postings, and the postings need not answer to the definition.

3.7.47 unformalized

सती वाविद्यमाना वा प्रकृतिः परिणामिनी ।
यस्य नाश्रीयते तस्य निर्वर्त्यत्वं प्रचक्षते ॥ ४७ ॥

satī vāvidyamānā vā prakṛtiḥ pariṇāminī |
yasya nāśrīyate tasya nirvartyatvaṃ pracakṣate || 47 ||

That for which a transforming material cause — whether existent or non-existent — is not resorted to: of that they declare the state of being what-is-to-be-produced.

The nirvartya defined by an absence. Sati va avidyamana va prakrtih parinamini: a *transforming material cause* — whether *existent or non-existent* (the clause is careful: it does not matter whether one holds the material really transforms) — yasya na asriyate: *for which* it is *not resorted to*. Tasya nirvartyatvam pracaksate: of *that* they declare the state of being *what-is-to-be-produced*. So the nirvartya-karman is the object in whose production no material is being attended to: 'he makes a mat' does not present the reeds; the mat simply comes to be. And the contrast is with vikarya (3.7.48), where the material *is* attended to. The distinction is thus not between two kinds of event but between two ways of construing one — whether the stuff is in view — which is 3.7.3's thesis (the karaka-practice is grounded in a state of the intellect) applied to the object's own taxonomy.

3.7.48 unformalized

प्रकृतेस्तु विवक्षायां विकार्यम् कैश्चिदन्यथा ।
निर्वर्त्यं च विकार्यं च कर्म शास्त्रे प्रदर्शितम् ॥ ४८ ॥

prakṛtestu vivakṣāyāṃ vikāryam kaiścidanyathā |
nirvartyaṃ ca vikāryaṃ ca karma śāstre pradarśitam || 48 ||

But when the material cause is intended, it is what-is-to-be-modified; by some, otherwise. Both what-is-to-be-produced and what-is-to-be-modified have been exhibited as the object in the science.

The complement of 3.7.47, and the criterion is named. Prakrteh tu vivaksayam vikaryam: when the *material cause* is *intended*, it is *what-is-to-be-modified*. Vivaksa — intention: the samuddesa's constant (3.7.2's kvacit kimcit vivaksyate; 3.7.19's difference by intention). Whether an object is nirvartya or vikarya depends on whether the speaker means to bring its material into view: 'he makes an ornament' (nirvartya, the gold unattended) versus 'he works the gold into an ornament' (vikarya, the gold in view). Same goldsmith, same afternoon. Kaiscid anyatha: *by some*, *otherwise* — the criterion is disputed. Nirvartyam ca vikaryam ca karma sastre pradarsitam: both have been *exhibited as the object* in the *science* — Panini's system recognizes both, and it does not tell you which is which, because the assignment is the speaker's.

3.7.49 unformalized

यदसज्जायते सद्वा जन्मना यत्प्रकाश्यते ।
तन्निर्वर्त्यं विकार्यं च कर्म द्वेधा व्यवस्थितम् ॥ ४९ ॥

yadasajjāyate sadvā janmanā yatprakāśyate |
tannirvartyaṃ vikāryaṃ ca karma dvedhā vyavasthitam || 49 ||

What, being non-existent, comes to be; or what, being existent, is made manifest by the birth — that is the object established two-fold, as what-is-to-be-produced and what-is-to-be-modified.

The two-fold object given its metaphysical alternatives, and they are 3.3.43's dilemma. Yad asat jayate: what, *being non-existent*, *comes to be* — the asatkarya-vada option, the Vaisesika's: the mat was not, and now is. Sad va janmana yat prakasyate: or what, *being existent*, is *made manifest by the birth* — the satkarya-vada option, the Samkhya's: the mat was in the reeds and the weaving revealed it. Tan nirvartyam vikaryam ca karma dvedha vyavasthitam: *that* is the object established *two-fold*. The verse maps the grammar's two objects onto philosophy's two theories of causation — and note what this implies: the choice between satkarya and asatkarya, over which the schools fought for centuries, is here a choice of *construal*, available to any speaker, and Sanskrit has a case-assignment for each. The metaphysical dispute is a difference in vivaksa (3.7.48), and both parties speak grammatically.

3.7.50 unformalized

प्रकृत्युच्छेदसंभूतं किंचित् काष्ठादिभस्मवत् ।
किंचिद् गुणान्तरोत्पत्त्या सुवर्णादिविकारवत् ॥ ५० ॥

prakṛtyucchedasaṃbhūtaṃ kiṃcit kāṣṭhādibhasmavat |
kiṃcid guṇāntarotpattyā suvarṇādivikāravat || 50 ||

Something arises from the destruction of the material — like ash from wood; something by the arising of another quality — like the modification of gold and the rest.

The vikarya's two modes. Prakrty-uccheda-sambhutam kimcit: *something* arises from the *destruction of the material* — kastha-adi-bhasmavat, *like ash from wood*: the wood is gone, and the ash is what became of it; the material is consumed in the modification. Kimcid gunantara-utpattya: *something* by the *arising of another quality* — suvarnadi-vikaravat, *like the modification of gold*: the gold persists entire and takes a new shape (3.2.4's suvarnadi bhinnam svaih akaraih apayibhih: gold differentiated by its own perishing configurations, and the gold never other). So the vikarya splits: destructive modification, where the material does not survive, and qualitative modification, where it does. And the two examples are the tradition's standing pair — ash and gold — precisely because they mark the limit: at one end the material is annihilated, at the other it is untouched, and the grammar's single category vikarya covers both. What unites them is only that the material is in view (3.7.48), which is a fact about the speaker.

3.7.51 unformalized

क्रियागतविशेषणां सिद्धिर्यत्र न गम्यते ।
दर्शनादनुमानाद्वा तत्प्राप्यमिति कथ्यते ॥ ५१ ॥

kriyāgataviśeṣaṇāṃ siddhiryatra na gamyate |
darśanādanumānādvā tatprāpyamiti kathyate || 51 ||

Where an accomplishment of the particulars belonging to the action is not understood, either from seeing or from inference — that is called what-is-to-be-reached.

The third object, defined by a double absence. Kriya-gata-visesanam siddhih yatra na gamyate: where an *accomplishment* of the *particulars belonging to the action* is *not understood* — nothing about the object shows that the action happened to it. Darsanad anumanad va: either from *seeing* or from *inference* — neither observation nor reasoning finds a mark. Tat prapyam iti kathyate: *that* is called *what-is-to-be-reached*. The village one goes to is unchanged by the going; look at it and infer as you like, and no trace of the arrival is in it. So the prapya-karman is the object that the action leaves as it found it — which makes it, of the three, the one whose objecthood is *purely* a construal: the nirvartya at least comes to be, the vikarya at least changes, and the prapya does nothing at all except get reached. Its karman-hood is entirely the speaker's posting (3.7.3).

3.7.52 unformalized

विशेषलाभः सर्वत्र विद्यते दर्शनादिभिः ।
केषांचित्तदभिव्यक्तिसिद्धिर्दृष्टिविषादिषु ॥ ५२ ॥

viśeṣalābhaḥ sarvatra vidyate darśanādibhiḥ |
keṣāṃcittadabhivyaktisiddhirdṛṣṭiviṣādiṣu || 52 ||

Everywhere a gaining of particulars exists, through seeing and the rest; for some, the establishing of that manifestation is in the case of seeing, poison and so on.

An objection to 3.7.51, and it is a good one. Visesa-labhah sarvatra vidyate darsanadibhih: *everywhere* a *gaining of particulars* exists, through *seeing* and the rest — even the prapya *does* acquire something: being-seen, being-reached, being-arrived-at. Look hard enough and every object is marked by the action that took it. Kesamcit tad-abhivyakti-siddhih drsti-visadisu: for *some*, the *establishing of that manifestation* is in the case of *seeing*, *poison* and so on — where the object visibly registers the act (the poisoned man shows the poison; the seen thing is, on some accounts, altered by being seen). So the third category may collapse: if every object gains a particular, none is merely reached. The objection presses 3.7.51's double absence, and the next verse answers by naming what the prapya does gain.

3.7.53 unformalized

आभासोपगमो व्यक्तिः सोढत्वमिति कर्मणः ।
विशेषाः प्राप्यमाणस्य क्रियासिद्धौ व्यवस्थिताः ॥ ५३ ॥

ābhāsopagamo vyaktiḥ soḍhatvamiti karmaṇaḥ |
viśeṣāḥ prāpyamāṇasya kriyāsiddhau vyavasthitāḥ || 53 ||

Coming-to-appearance, manifestation, and being-endured: these particulars of the object that is being reached stand established in the accomplishing of the action.

The answer to 3.7.52: the prapya's gains are named, and they are all relational. Abhasa-upagamah — *coming-to-appearance*: the object comes into view; vyaktih — *manifestation*: it becomes manifest; sodhatvam — *being-endured*, being-put-up-with, being-undergone. Iti karmanah visesah prapyamanasya kriya-siddhau vyavasthitah: these *particulars* of the object *that is being reached* stand established *in the accomplishing of the action*. Every one of the three is a fact about the object's standing-in-relation to the act, not about the object: coming-to-appearance is appearing to someone, manifestation is manifestation to a cognition, being-endured is enduring an action. So 3.7.52 is granted its datum and denied its conclusion: the prapya does gain particulars, and they are relational particulars, conferred in the act and belonging to no thing (3.6.14's things without place, parts, motion, support; 3.7.14's nimitta-bhava as a leaning).

3.7.54 unformalized

निर्वर्त्यादिषु तत्पूर्वमनुभूय स्वतन्त्रताम् ।
कर्त्रन्तराणां व्यापारे कर्म संपद्यते ततः ॥ ५४ ॥

nirvartyādiṣu tatpūrvamanubhūya svatantratām |
kartrantarāṇāṃ vyāpāre karma saṃpadyate tataḥ || 54 ||

Among what-is-to-be-produced and the rest, having first experienced independence, the object comes to be, thereafter, in the operation of other agents.

The object's own agency, and this is 3.7.18's doctrine reaching the karman. Nirvartyadisu tat purvam anubhuya svatantratam: among the nirvartya and the rest, *having first experienced independence* — the object is *svatantra*, independent, before it is an object: the wood is doing its own thing, the village stands on its own. Svatantrya is Panini's own mark of the agent (1.4.54: svatantrah karta), and it is here predicated of the object in its prior condition. Kartrantaranam vyapare karma sampadyate tatah: in the *operation of other agents*, the object *comes to be*, *thereafter*. Objecthood is a demotion suffered when another agent's operation arrives (3.7.20's nyakkrta-saktayah: capacities subordinated by the agent). So the karman is an agent that has been overtaken. Every object in every sentence was independent a moment before the verb reached it, and its objecthood is the grammar's record of a subordination.

3.7.55 unformalized

तद्व्यापारविवेकेऽपि स्वव्यापारे व्यवस्थितम् ।
कर्मापदिष्टान् लभते क्वचिच्छास्त्राश्रयान् विधीन् ॥ ५५ ॥

tadvyāpāraviveke'pi svavyāpāre vyavasthitam |
karmāpadiṣṭān labhate kvacicchāstrāśrayān vidhīn || 55 ||

Even in the separation from that operation, the object, standing in its own operation, obtains in some cases the rules taught in the science that rest upon it.

The object's retained agency, and it has grammatical consequences. Tad-vyapara-viveke 'pi: even in the *separation from that* operation — even when the other agent's action is withdrawn or not in view — sva-vyapare vyavasthitam: the object stands *in its own operation*: the wood goes on being wood, doing what wood does (3.7.20's sampurnaih svaih vyaparaih samanvitah: furnished with their own complete operations even when subordinated). Karma apadistan labhate kvacit sastrasrayan vidhin: the *object* obtains, in some cases, the *rules taught* in the science that *rest upon it*. Because the object retains its own agency, the grammar can address rules to it — and this is exactly what the karma-kartari construction does (the next verses' subject): 'the rice cooks itself', where the object's own operation surfaces and it takes the agent's endings. The doctrine that every karaka is an agent (3.7.18) is not a metaphysical flourish; it is what makes a whole class of Sanskrit constructions statable.

3.7.56 unformalized

निवृत्तप्रेषणं कर्म स्वक्रियावयवे स्थितम् ।
निवर्तमाने कर्मत्वे स्वे कर्तृत्वेऽवतिष्ठते ॥ ५६ ॥

nivṛttapreṣaṇaṃ karma svakriyāvayave sthitam |
nivartamāne karmatve sve kartṛtve'vatiṣṭhate || 56 ||

The object, its being-directed having ceased, stands in a portion of its own action; the objecthood withdrawing, it stands in its own agency.

The karma-kartari construction analysed, and the analysis is exact. Nivrtta-presanam karma: the *object*, its *being-directed having ceased* — presana: the sending, the directing by another agent; when that withdraws — sva-kriya-avayave sthitam: it *stands in a portion of its own action*. Nivartamane karmatve sve kartrtve 'vatisthate: the *objecthood withdrawing*, it stands *in its own agency*. So 'pacyate odanah' — 'the rice cooks' — is not a passive with a suppressed agent; it is the object having reverted to the agency it always had (3.7.54's svatantrata, experienced first) once the cook is out of view. The construction's morphology (the passive endings with an active sense) records exactly this: objecthood withdrawn, agency resumed. And the doctrine that makes it intelligible has been in place since 3.7.18: agency was never the subject's privilege; it was every karaka's, differently attended to.

3.7.57 unformalized

तानि धात्वन्तराण्येव पचिसिद्ध्यतिवद् विदुः ।
भेदेऽपि तुल्यरूपत्वादेकत्वपरिकल्पना ॥ ५७ ॥

tāni dhātvantarāṇyeva pacisiddhyativad viduḥ |
bhede'pi tulyarūpatvādekatvaparikalpanā || 57 ||

Those they know as simply other roots — like 'pac' and 'sidhyati'; though there is difference, from likeness of form there is a construction of oneness.

A deflationary account of the karma-kartari's morphology. Tani dhatv-antarani eva viduh: those they know as *simply other roots* — the 'cooks' of 'the cook cooks' and the 'cooks' of 'the rice cooks' are two verbs, not one verb in two constructions. Paci-sidhyati-vat: *like* 'pac' (to cook, transitive) and 'sidhyati' (to be accomplished, intransitive) — two roots that mean nearly the same and are undeniably two. Bhede 'pi tulya-rupatvat ekatva-parikalpana: though there is *difference*, from *likeness of form* there is a *construction of oneness*. We take them for one verb because they sound alike (2.169's kupa-supa-yupa: phonetic overlap is semantically inert; 2.177's rule: different meaning-behaviour, different base). So the elegant analysis of 3.7.56 is offered a rival: there is no reversion of agency, just two verbs that rhyme. Bhartrhari reports it without deciding — and the report is itself instructive, since the two accounts are 2.57's two darsanas (whole-first or parts-first) meeting in a single conjugation.

3.7.58 unformalized

एकदेशे समूहे च व्यापाराणां पचादयः ।
स्वभावतः प्रवर्तन्ते तुल्यरूपसमन्विताः ॥ ५८ ॥

ekadeśe samūhe ca vyāpārāṇāṃ pacādayaḥ |
svabhāvataḥ pravartante tulyarūpasamanvitāḥ || 58 ||

In a single portion and in an aggregate of operations, 'pac' and the rest proceed by their own nature, endowed with like form.

The rival account of 3.7.57 developed. Eka-dese samuhe ca vyaparanam pacadayah: in a *single portion* and in an *aggregate* of operations — whether the verb names one act or a complex of them — svabhavatah pravartante: they proceed *by their own nature*: the roots simply do what they do, without a scheme of role-assignments organizing them. Tulya-rupa-samanvitah: *endowed with like form*. So on this view the verb-family's unity is entirely superficial, and its members are separate items that happen to share a shape and a rough sense. The account is the atomist's (2.52's nominalism: word and sentence as mere designations upon order), and its virtue is economy: no reverting agency, no subordination-hierarchy, just roots. Its cost is that it explains nothing about why the forms cluster as they do — which is why 3.7.59-60 return to the structural analysis.

3.7.59-60 unformalized

न्यग्भावनान्यग्भवनं रूहौ शुद्धे प्रतीयते ।
न्यग्भावना न्यग्भवनं ण्यन्तेऽपि प्रतिपद्यते ॥ ५९ ॥
अवस्थां पंचमीमाहुर्ण्यन्ते तां कर्मकर्तरि ।
निवृत्तप्रेषणाद्धातोः प्राकृतेऽर्थे णिजुच्यते ॥ ६० ॥

nyagbhāvanānyagbhavanaṃ rūhau śuddhe pratīyate |
nyagbhāvanā nyagbhavanaṃ ṇyante'pi pratipadyate || 59 ||
avasthāṃ paṃcamīmāhurṇyante tāṃ karmakartari |
nivṛttapreṣaṇāddhātoḥ prākṛte'rthe ṇijucyate || 60 ||

(59) Subordinating and being-subordinated are understood in the pure root; subordinating and being-subordinated are found in the causative as well. (60) They call it the fifth state; in the causative, in the object-as-agent — when the directing has ceased, the causative affix is stated of the root in its natural meaning.

The structure of 3.7.20-21 found at every morphological level. (59): nyag-bhavana nyag-bhavanam rudhau suddhe pratiyate — *subordinating* and *being-subordinated* are understood in the *pure root*: even the simple verb contains the hierarchy — something subordinates, something is subordinated (nyak: down, the same root as 3.7.20's nyakkrta). And: nyanta 'pi pratipadyate — they are found in the *causative* as well: the nic-affix adds a level and the same structure repeats. So subordination is not a feature the causative introduces; it is present in the bare root and iterated by the affix (3.7.25's karta kartr-antara-apeksah: an agent depending on another agent). (60): avastham pancamim ahuh nyante tam karma-kartari — they call it the *fifth state*, in the *causative*, in the *object-as-agent*; nivrtta-presanad dhatoh prakrte 'rthe nij ucyate: when the *directing has ceased*, the causative affix is stated of the root *in its natural meaning*. The verse tracks a construction where the causative morphology appears without causative sense, because the directing (3.7.56's presana) has withdrawn — the morphology outliving its semantics, which is 2.229-231's ledger-shuffling in the verb's own paradigm.
contested: 60's avastham pancamim ('the fifth state') is not further identified and the commentators differ over what series it numbers; we translate the phrase and leave the enumeration open

3.7.61 unformalized

ब्रवीति पचतेरर्थं सिद्ध्यतिर्न विना णिचम् ।
स ण्यन्तः पचतेरर्थो प्रकृते व्यवतिष्ठते ॥ ६१ ॥

bravīti pacaterarthaṃ siddhyatirna vinā ṇicam |
sa ṇyantaḥ pacaterartho prakṛte vyavatiṣṭhate || 61 ||

'Sidhyati' states the meaning of 'pacati' only with the causative affix; that causative-ending stands in the natural meaning of 'pacati'.

The two roots of 3.7.57 shown to be related after all, and by a morphological argument. Braviti pacater artham sidhyatir na vina nijam: 'sidhyati' (is accomplished) states the meaning of 'pacati' (cooks) *only with the causative affix* — sadhayati, the causative of sidh, means what pacati means: one has to add nic to get from the intransitive to the transitive sense. Sa nyantah pacater artho prakrte vyavatisthate: *that causative-ending* stands in the *natural* meaning of 'pacati' — what the one root needs a causative to say, the other says in its bare form. So pac contains, in its natural meaning, what sidh requires an affix to reach: transitivity is built into pac and added to sidh. This is 3.7.59's doctrine (subordinating and being-subordinated present in the pure root) given its evidence — the hierarchy is inside pac already, which is why it needs no nic, and its being inside is visible precisely by comparison with a root that lacks it.

3.7.62 unformalized

केषांचिद्देवदत्तादेर्व्यापारो यः सकर्मके ।
स विना देवदत्तादेः कटादिषु विवक्ष्यते ॥ ६२ ॥

keṣāṃciddevadattādervyāpāro yaḥ sakarmake |
sa vinā devadattādeḥ kaṭādiṣu vivakṣyate || 62 ||

For some, that operation of Devadatta and the rest which is in a transitive verb — that same is intended, without Devadatta and the rest, in mats and the like.

The karma-kartari's semantics restated. Kesamcid devadattader vyaparo yah sakarmake: for some, that *operation* of Devadatta and the rest which is in a *transitive* verb — the weaving that Devadatta does. Sa vina devadattadeh katadisu vivaksyate: *that same* is *intended*, *without* Devadatta and the rest, in *mats* and the like. 'The mat weaves itself' intends the very operation that 'Devadatta weaves the mat' intends — the same weaving, with the weaver simply not meant (3.7.56's nivrtta-presana: the directing having ceased). Vivaksyate: *is intended* — the agent's absence is a fact about the speaker's aim, not about the event. So the construction subtracts nothing from the action; it withdraws attention from one of its participants, and the mat, which was an agent all along (3.7.54's svatantrata experienced first), surfaces.

3.7.63 unformalized

निवृत्तप्रेषणं कर्म स्वस्य कर्तुः प्रयोजकम् ।
प्रेषणान्तरसंबन्धे ण्यन्ते लेनाभिधीयते ॥ ६३ ॥

nivṛttapreṣaṇaṃ karma svasya kartuḥ prayojakam |
preṣaṇāntarasaṃbandhe ṇyante lenābhidhīyate || 63 ||

The object, its directing having ceased, is the occasioner of its own agent; when there is connection with another directing, in the causative it is expressed by the 'la'-affix.

The construction's morphology tracked through a further level. Nivrtta-presanam karma svasya kartuh prayojakam: the *object*, its directing having ceased, is the *occasioner of its own agent* — the mat, freed from the weaver, occasions its own agency (3.7.56 verbatim, and prayojaka is 3.1.25's word for the universal deploying its causes). Presanantara-sambandhe nyante lena abhidhiyate: when there is connection with *another* directing — a second agent arriving above the first — in the *causative* it is expressed by the *la*-affix: the personal endings (Panini's lakaras) carry it. So the sequence is: object directed (accusative), directing withdrawn (object becomes agent), new directing added (causative, and the la-affix marks the new arrangement). Each level of subordination is a morphological layer, and the layers stack exactly as 3.7.59's nyag-bhavana predicted. The grammar is doing hierarchy-bookkeeping, and the hierarchy is 3.7.22's king and soldiers.

3.7.64 unformalized

सदृशादिषु यत्कर्म कर्तृत्वं प्रतिपद्यते ।
आपत्त्यापादने तत्र विषयत्वं प्रतिक्रिये ॥ ६४ ॥

sadṛśādiṣu yatkarma kartṛtvaṃ pratipadyate |
āpattyāpādane tatra viṣayatvaṃ pratikriye || 64 ||

Where an object attains agency in 'sadrsa' and the like — there, in the bringing-about of a state, objecthood belongs to the counter-action.

A further case, and it is technical. Sadrsadisu yat karma kartrtvam pratipadyate: where an *object* attains *agency* in 'sadrsa' (resembling) and the like — verbs of resembling, of becoming-like, where what is compared behaves now as object, now as agent. Apatty-apadane tatra visayatvam pratikriye: there, in the *bringing-about of a state* (apatti: coming-to-be-such), *objecthood* belongs to the *counter-action*. Where the construction expresses a becoming-like, the objecthood shifts to the reciprocal act — resembling being a relation each relatum bears to the other (3.7.19's parents, both agents, roles assigned by intention). The verse extends the karma-kartari analysis to a family of verbs where the agent/object assignment is symmetrical in the world and asymmetrical in every sentence — which is the samuddesa's constant finding since 3.7.19.

3.7.65 unformalized

कुतश्चिदाहृत्य पदमेवं च परिकल्पने ।
कर्मस्थभावकत्वं स्पाद्दर्शनाद्यभिधायिनाम् ॥ ६५ ॥

kutaścidāhṛtya padamevaṃ ca parikalpane |
karmasthabhāvakatvaṃ spāddarśanādyabhidhāyinām || 65 ||

And when a word is fetched from somewhere and so construed, there would be object-located-state-hood for the expressers of seeing and the rest.

A construal-dependent classification. Kutascid ahrtya padam evam ca parikalpane: when a *word* is *fetched from somewhere* and *so construed* — supplied from context, imported into the analysis (2.327's unuttered word supplied by the cognition). Karma-stha-bhavakatvam syad darsanady-abhidhayinam: there would be *object-located-state-hood* for the expressers of *seeing* and the rest. Karma-stha-bhavaka is the grammarians' term for verbs whose state resides in the object rather than the agent: 'he sees the pot' — where does the seeing's result lodge? In the pot (which becomes seen) or in the seer? The classification decides how the verb's affixes behave. And the verse's point is that the classification follows the *construal*: fetch a word from context, construe accordingly, and the verb changes class. Nothing in seeing settles where seeing lodges (3.7.3's buddhi-grounding, at the level of verb-taxonomy).

3.7.66 unformalized

विशेषदर्शनं यत्र क्रिया तत्र व्यवस्थिता ।
क्रियाव्यवस्था त्वन्येषां शब्दैरेव प्रकाश्यते ॥ ६६ ॥

viśeṣadarśanaṃ yatra kriyā tatra vyavasthitā |
kriyāvyavasthā tvanyeṣāṃ śabdaireva prakāśyate || 66 ||

Where the seeing of a particular is, there the action stands settled; but for others, the settling of the action is made manifest by words alone.

Two criteria for locating an action. Visesa-darsanam yatra kriya tatra vyavasthita: where the *seeing of a particular* is, *there* the action stands settled — locate the action where its result shows: the cooking is in the rice, because the rice softens; the difference is visible, so the action is there (3.7.51's criterion for the prapya, run in reverse). Kriya-vyavastha tu anyesam sabdaih eva prakasyate: but for *others*, the settling of the action is made manifest *by words alone* — the action is where the grammar says it is, and no observation is relevant. Two schools: one locates actions by their visible effects, the other by the sastra's assignments. And the samuddesa has been arguing throughout that the second is right (3.7.13: the assistance is understood by the case-endings alone) — not because words are authoritative over the world but because the location of an action is not a fact the world contains (3.7.8's nirihesu api bhavesu kalpana-upanibandhanam: even among the striving-less, it rests on construction).

3.7.67 unformalized

कालभावाध्बदेशानामन्तर्भूतक्रियान्तरैः ।
सर्वैरकर्मकैर्योगे कर्मत्वमुपजायते ॥ ६७ ॥

kālabhāvādhbadeśānāmantarbhūtakriyāntaraiḥ |
sarvairakarmakairyoge karmatvamupajāyate || 67 ||

For time, state, road and place, in connection with all intransitive verbs having other actions included within them, objecthood arises.

How intransitive verbs acquire objects. Kala-bhava-adhva-desanam: for *time*, *state*, *road* and *place* — the four that Panini's rule (1.4.51's kala-adhvanor atyanta-samyoge and its kin) makes accusative even with intransitives: 'he sleeps a month', 'he goes a league'. Antarbhuta-kriyantaraih sarvaih akarmakaih yoge: in connection with *all intransitive* verbs having *other actions included within them* — antarbhuta: absorbed, included (3.3.47's antarbhava, the self-bearing of 'asti'). Karmatvam upajayate: *objecthood arises*. So 'he sleeps a month' has an object because the sleeping has an occupying-of-time included in it, and the month is that included action's object. The intransitive verb is not really intransitive: it contains a second action, and the accusative belongs to that one. The analysis is elegant and it is the samuddesa's method throughout — where the grammar assigns a role that seems unearned, find the included operation that earns it.

3.7.68 unformalized

आधारत्वमिव प्राप्तास्ते पुनर्द्रव्यकर्मसु ।
कालादयो भिन्नकक्ष्यं यान्ति कर्मत्वमुत्तरम् ॥ ६८ ॥

ādhāratvamiva prāptāste punardravyakarmasu |
kālādayo bhinnakakṣyaṃ yānti karmatvamuttaram || 68 ||

Having attained, as it were, the state of being a locus, those — time and the rest — again, in the case of substance-objects, attain a subsequent objecthood of a different level.

The levels of 3.7.67 sorted. Adharatvam iva praptah te: having attained, *as it were*, the state of being a *locus* — time and place naturally present as loci ('he sleeps *in* the night', locative), and the iva marks that this is their default construal. Punah dravya-karmasu: *again*, in the case of *substance-objects* — where the verb already has a substance for its object ('he studies the book a month'). Kaladayah bhinna-kaksyam yanti karmatvam uttaram: time and the rest attain a *subsequent objecthood* of a *different level* — bhinna-kaksya: of another compartment, another tier. So a sentence can carry two objects at different levels: the book at one, the month at another, and they do not compete because they are objects of different included actions (3.7.67's antarbhuta-kriya). Kaksya — level, compartment — is the samuddesa's term for the hierarchy 3.7.20-23 built, and here it does taxonomic work: objecthood is stratified, and the strata do not collide.

3.7.69 unformalized

अतस्तैः कर्मभिर्धातुर्युक्तोऽद्रव्यैरकर्मकः ।
लस्य कर्मणि भावे च निमित्तत्वाय कल्पते ॥ ६९ ॥

atastaiḥ karmabhirdhāturyukto'dravyairakarmakaḥ |
lasya karmaṇi bhāve ca nimittatvāya kalpate || 69 ||

Therefore a root joined with those objects, which are not substances, is intransitive; it serves as the occasioning ground for the 'la'-affix in the object and in the state.

The consequence for morphology. Atah taih karmabhih dhatur yuktah adravyaih akarmakah: *therefore* a root joined with *those objects* — which are *not substances* (time, place, road, state: 3.7.67's four) — is *intransitive*. The verb has an accusative and is still akarmaka, because its accusative is not a substance-object. Lasya karmani bhave ca nimittatvaya kalpate: it serves as the *occasioning ground* for the *la*-affix *in the object and in the state* — that is, it takes the passive/impersonal endings, which is what intransitives do. So the grammar's classification tracks *substance*-objecthood, not accusatives: a verb with a temporal accusative counts as intransitive for affixation. The verse is fine technical bookkeeping and its interest is that the category 'transitive' turns out not to mean 'takes an accusative' — which is exactly what one expects if the karakas are postings assigned per operation (3.7.12) rather than features read off surface case.

3.7.70 unformalized

सर्वं वाकथितं कर्म भिन्नकक्ष्यं प्रतीयते ।
धात्वर्थोद्देशभेदेन तन्नेप्सिततमं किल ॥ ७० ॥

sarvaṃ vākathitaṃ karma bhinnakakṣyaṃ pratīyate |
dhātvarthoddeśabhedena tannepsitatamaṃ kila || 70 ||

Or every unstated object is understood as of a different level; by difference in the indication of the root's meaning, that is not what is most desired.

The akathita placed. Sarvam va akathitam karma bhinna-kaksyam pratiyate: *every unstated object* is understood as of a *different level* — akathita: 'not stated', Panini's own term (1.4.51: akathitam ca — 'and what is unstated' is called karman), the residue-category that catches every object no other rule assigns. Bhinna-kaksyam: they belong to another tier (3.7.68's stratification). Dhatv-artha-uddesa-bhedena tan na ipsitatamam kila: by difference in the *indication of the root's meaning*, *that* is *not what is most desired* — kila: 'so they say', the reportive particle carrying a shade of irony. So the akathita objects fail Panini's own definition of karman (3.7.45's ipsitatamam, what is most desired), and the grammar catches them in a residue-rule anyway. The verse's kila registers the awkwardness: the definition is by desire, most objects are not desired, and the system works by a catch-all. 3.7.46 already listed the undesired classes; here their status is named.

3.7.71 unformalized

प्रधानकर्म कथितं यत् क्रियायाः प्रयोजकम् ।
तत्सिद्धये क्रियायुक्तमन्यत्त्वकथितं स्मृतम् ॥ ७१ ॥

pradhānakarma kathitaṃ yat kriyāyāḥ prayojakam |
tatsiddhaye kriyāyuktamanyattvakathitaṃ smṛtam || 71 ||

The principal object is what is stated, which is the occasioner of the action; what is joined with the action for the accomplishing of that is remembered as unstated.

The two objects distinguished by office. Pradhana-karma kathitam: the *principal object* is what is *stated* — yat kriyayah prayojakam: which is the *occasioner of the action*: the object the action is *for*, whose desirability drives the whole undertaking (3.7.45's ipsitatama, now confined to this one). Tat-siddhaye kriya-yuktam anyat tu akathitam smrtam: what is joined with the action *for the accomplishing of that* is remembered as *unstated*. So in 'he milks the cow of milk' (the classic double-accusative), the milk is the principal — it is what he wants — and the cow is akathita: joined to the action in service of getting the milk. The distinction is teleological through and through: which object is principal depends on which is wanted, and the other is machinery. This is 2.113's prayojana-doctrine (sentence-meaning is purpose) reaching down into the accusative case.

3.7.72 unformalized

दुह्यादिवन्नयत्यादौ कर्मत्वमकथाश्रयम् ।
आख्यातानुपयोगे तु नियमाच्छेव इष्यते ॥ ७२ ॥

duhyādivannayatyādau karmatvamakathāśrayam |
ākhyātānupayoge tu niyamāccheva iṣyate || 72 ||

In 'nayati' and the like, as in 'duhi' and the rest, objecthood rests on what is not stated; but when the finite verb is not employed, the remainder is held to obtain, by rule.

The double-object verbs, and their two families. Duhy-adi-vat nayaty-adau karmatvam akatha-srayam: in *'nayati'* (leads) and the like, as in *'duhi'* (milks) and the rest, *objecthood* rests on *what is not stated* — both families take two accusatives, and in both the second is akathita (3.7.70-71). Akhyata-anupayoge tu niyamat sesa isyate: but when the *finite verb is not employed* — in nominal constructions, participles, compounds, where no finite verb governs — *the remainder* is held to obtain, *by rule*: the genitive, Panini's sesa-case (3.7.44: sasthi sese), takes over. So 'he milks the cow of milk' has two accusatives, and 'the milking of the cow' has a genitive: withdraw the finite verb and the karaka-relations lapse into the residue-case. This is 2.197-198's doctrine exactly — the genitive expresses relations whose generating action has withdrawn — now shown operating in the double-object constructions.

3.7.73 unformalized

अन्तर्भूतणिजर्थानां दुह्यादीनां णिजन्तवत् ।
सिद्धं पूर्वेण कर्मत्वं णिजन्तं नियमस्तथा ॥ ७३ ॥

antarbhūtaṇijarthānāṃ duhyādīnāṃ ṇijantavat |
siddhaṃ pūrveṇa karmatvaṃ ṇijantaṃ niyamastathā || 73 ||

For 'duhi' and the rest, whose causative-meaning is included within, objecthood is established by the prior rule as though they ended in the causative; and the restriction is likewise as for causative-ending forms.

The double-object verbs explained by a hidden causative. Antarbhuta-nij-arthanam duhy-adinam: of *'duhi'* and the rest, whose *causative-meaning is included within* — antarbhuta again (3.7.67's included actions): 'milks' contains a 'causes-to-flow'; the causative is in the root's own meaning without any affix. Nijantavat siddham purvena karmatvam: *objecthood* is *established by the prior rule* *as though they ended in the causative* — the grammar treats them like causatives because they are causatives semantically, and causatives take two objects (the one caused, and the thing done). Nijantam niyamas tatha: and the *restriction* is likewise as for causative-ending forms. So the double accusative of 'duhi' is not an idiosyncrasy: it is the ordinary causative pattern, with the causative buried in the root. This completes 3.7.61's finding (pac contains what sidh needs a nic to say) — the roots' meanings carry morphology invisibly, and the case-assignments reveal it.

3.7.74 unformalized

करणस्य स्वकक्ष्यायां न प्रकर्षाश्रयो यथा ।
कर्मणोऽपि स्वकक्ष्यायां न स्यादतिशयस्तथा ॥ ७४ ॥

karaṇasya svakakṣyāyāṃ na prakarṣāśrayo yathā |
karmaṇo'pi svakakṣyāyāṃ na syādatiśayastathā || 74 ||

As for the instrument, in its own compartment, there is no resort to excellence — so for the object too, in its own compartment, there would be no surpassing.

An objection to the ipsitatama-definition, from the grammar's own logic. Karanasya sva-kaksyayam na prakarsa-asrayo yatha: as for the *instrument*, in its *own compartment*, there is no resort to *excellence* — one instrument is not more an instrument than another; the karaka-roles do not come in degrees. Karmano 'pi sva-kaksyayam na syad atisayas tatha: so for the *object* too, in its own compartment, there would be *no surpassing*. But Panini's definition is ipsitatamam — the *most* desired: a superlative, which requires degrees. So the definition demands a gradation the category cannot support: if objecthood does not admit of more and less, 'most desired' picks out nothing. The objection is sharp, and it targets the definition that 3.7.45, 3.7.70 and 3.7.71 have all leaned on. The answer follows.

3.7.75 unformalized

कर्मणस्त्वाप्तुमिष्टत्व आश्रितेऽतिशयो यतः ।
आश्रीयते ततोऽत्यन्तं भेदः पूर्वेण कर्मणा ॥ ७५ ॥

karmaṇastvāptumiṣṭatva āśrite'tiśayo yataḥ |
āśrīyate tato'tyantaṃ bhedaḥ pūrveṇa karmaṇā || 75 ||

But since, for the object, when the state of being wished-to-be-attained is resorted to, a surpassing is present — therefore it is resorted to; hence there is an utter difference from the prior object.

The answer to 3.7.74, and it relocates the gradation. Karmanas tu aptum istatve asrite atisayo yatah: for the *object*, when the state of *being wished-to-be-attained* is *resorted to*, a *surpassing* is *present* — the degrees are not in objecthood but in *desiredness*: one may want the milk more than the cow, and that is a real gradation. Asriyate tatah: *therefore* it is resorted to — the definition invokes desire precisely because desire admits of degrees where objecthood does not. Atyantam bhedah purvena karmana: hence there is an *utter difference* from the *prior* object — the ipsitatama (the milk) differs in kind from the akathita (the cow), because the first is wanted most and the second is machinery (3.7.71). So 3.7.74's objection is granted and disarmed: objecthood has no degrees, desire does, and the superlative in Panini's definition qualifies the desire. The definition survives by being read exactly.

3.7.76 unformalized

णिजन्ते च यथा कर्ता सक्रियः सन् प्रयुज्यते ।
नयत्यादौ तथा कर्ता निष्कियोऽपि प्रयुज्यते ॥ ७६ ॥

ṇijante ca yathā kartā sakriyaḥ san prayujyate |
nayatyādau tathā kartā niṣkiyo'pi prayujyate || 76 ||

And as, in a causative-ending form, an agent possessing action is employed — so in 'nayati' and the like, an agent devoid of action is also employed.

The parallel completed, and its second half is the odd one. Nijante ca yatha karta sakriyah san prayujyate: in a *causative-ending* form, an agent *possessing action* is employed — in 'he makes Devadatta cook', Devadatta is an agent and he is cooking: the caused agent acts. Nayaty-adau tatha karta niskriyo 'pi prayujyate: in *'nayati'* (leads) and the like, an agent *devoid of action* is *also* employed — 'he leads the goat to the village': the goat is grammatically an agent of a sort (it is the one led, and the leading is its going), and it does nothing; it is dragged. Niskriya: without action (3.7.8's niriha, without striving). So the pattern that 3.7.73 established — the double-object verbs behaving as causatives — extends to cases where the 'caused agent' has no agency whatever. The grammar constructs an agent out of the inert, exactly as 3.7.8 said it does: independence and dependence rest upon construction, even among things that are without striving.

3.7.77 unformalized

भेदवाक्यं तु यत् ण्यन्ते नीदुहितप्रकृतौ च यत् ।
शब्दान्तरत्वान्नैवास्ति संस्पर्शस्तस्य धातुना ॥ ७७ ॥

bhedavākyaṃ tu yat ṇyante nīduhitaprakṛtau ca yat |
śabdāntaratvānnaivāsti saṃsparśastasya dhātunā || 77 ||

But the analytic sentence which is used for a causative-ending form, and that which is used for the base of 'ni' and 'duh' — from its being another word, there is no contact of it with the root.

The paraphrase disowned. Bheda-vakyam tu yat nyante: the *analytic sentence* used for a *causative-ending* form — the gloss 'he causes X to cook' by which the grammar explains 'pacayati'; ni-duhita-prakrtau ca yat: and that used for the *base* of 'ni' and 'duh' (the double-object verbs of 3.7.72-73). Sabdantaratvat na eva asti samsparsah tasya dhatuna: *from its being another word*, there is *no contact* of it with the *root*. The paraphrase is a different word entirely and touches the root not at all. This is 2.35cd's doctrine (the vrtti-analysis employing a verb-like element for what stands otherwise, the paraphrase faithful to the meaning and false to the form) and 2.37's (the resemblance constructed for the sake of reaching the meaning). So the causative analyses of 3.7.61-76, however illuminating, are glosses: they explain by substituting, and what they substitute has no contact with what it explains. The samuddesa audits its own method at the point where it has leaned hardest on it.

3.7.78 unformalized

यथैवैकमपादानं शास्त्रे भेदेन दर्शतम् ।
तथैकमेव कर्मापि भेदेन प्रतिपादितम् ॥ ७८ ॥

yathaivaikamapādānaṃ śāstre bhedena darśatam |
tathaikameva karmāpi bhedena pratipāditam || 78 ||

Just as one and the same ablative is exhibited in the science by way of difference — so one and the same object too is expounded by way of difference.

The taxonomies placed. Yatha eva ekam apadanam sastre bhedena darsitam: as *one and the same* ablative is exhibited in the *science* by way of *difference* — Panini divides apadana into kinds (the moving source, the fixed source, the source of fear, the source of learning) and it is one karaka throughout. Tatha ekam eva karma api bhedena pratipaditam: so *one and the same* object too is *expounded* by way of *difference* — the nirvartya, vikarya and prapya of 3.7.45 are one karman, divided for exposition. So the object's three-fold taxonomy is a teaching-device, not a discovery of three objecthoods (2.176's instruction by the general for brevity's sake; 3.7.36's unlimited-seeming capacities reducing to six). The verse tidies the samuddesa's own apparatus: everything it has distinguished about the karman is distinction *in the science*, and the karman is one.

3.7.79 unformalized

निर्वर्त्यो वा विकार्यो वा प्राप्यो वा साधनाश्रयः ।
क्रियाणामेव साध्यत्वात् सिद्धरूपोऽभिधीयते ॥ ७९ ॥

nirvartyo vā vikāryo vā prāpyo vā sādhanāśrayaḥ |
kriyāṇāmeva sādhyatvāt siddharūpo'bhidhīyate || 79 ||

Whether what-is-to-be-produced, or what-is-to-be-modified, or what-is-to-be-reached — the substratum of the means, since actions alone are what is to be accomplished, is expressed as having an accomplished form.

The object's grammatical form explained. Nirvartyo va vikaryo va prapyo va sadhana-asrayah: whatever kind it is, the object is the *substratum of the means*. Kriyanam eva sadhyatvat: since *actions alone* are *what is to be accomplished* — only actions are sadhya; nothing else is ever brought about (3.7.1's definition: the means is capacity in the completing of *actions*). Siddha-rupo 'bhidhiyate: it is expressed as having an *accomplished* form. So the object, however unaccomplished it may be in fact (the mat not yet woven, 3.7.7's janyamana karman), is grammatically *siddha* — presented as a standing thing, a noun, complete. The sadhya/siddha distinction is the grammar's deepest: verbs present the to-be-accomplished, nouns the accomplished, and every participant in an action is nominally presented as though finished. This is why 3.7.7 needed buddhi to hold the unmade mat in place: the grammar has no way to present an object as unfinished.

3.7.80 unformalized

अहितेषु यता लौल्यात् कर्तुरिच्छोपजायते ।
विषादिषु भयादिभ्यस्तथैवासौ प्रवर्तते ॥ ८० ॥

ahiteṣu yatā laulyāt karturicchopajāyate |
viṣādiṣu bhayādibhyastathaivāsau pravartate || 80 ||

As, in the case of things not conducive, a desire of the agent arises from greed — so, in the case of poison and the like, it proceeds from fear and the rest.

The ipsitatama-definition rescued at its hardest case. 3.7.46 listed the undesired objects, and poison is the standing example: 'he eats poison' has an object that nobody most-desires. The answer: ahitesu yatha laulyat kartur iccha upajayate — in the case of things *not conducive*, a *desire* of the agent arises *from greed*: the glutton eats what harms him, and there *is* a desire — a bad one. Visadisu bhayadibhyah tatha eva asau pravartate: in the case of *poison* and the like, *that* (desire) proceeds *from fear* and the rest. The suicide's desire for poison comes from fear; the desire is real, its source is dread. So the undesired objects are desired after all, under desires the agent would disown — and the definition holds, at the cost of making 'most desired' cover what one takes in terror. It is an honest rescue: it saves the rule and shows what the rule then means.

3.7.81-82 unformalized

प्रधानेतकयोर्यत्र द्रव्यस्य क्रिययोः पृथक् ।
शक्तिर्गुणाश्रया तत्र प्रधानमनुरुध्यते ॥ ८१ ॥
प्रधानविषया शक्तिः प्रत्ययेनभिधीयते ।
यदा गुणे तदा तद्वदनुक्तापि प्रकाशते ॥ ८२ ॥

pradhānetakayoryatra dravyasya kriyayoḥ pṛthak |
śaktirguṇāśrayā tatra pradhānamanurudhyate || 81 ||
pradhānaviṣayā śaktiḥ pratyayenabhidhīyate |
yadā guṇe tadā tadvadanuktāpi prakāśate || 82 ||

(81) Where the capacity of a substance, in the principal and the other, is separately in the two actions — there the capacity resting on the subordinate follows the principal. (82) The capacity whose domain is the principal is expressed by the affix; when it is in the subordinate, then, though unstated, it likewise shines forth.

How karaka-roles propagate between clauses. (81): where a substance has a capacity in *two* actions — one principal, one subordinate — guna-asraya saktih pradhanam anurudhyate: the capacity resting on the *subordinate* *follows the principal*: the role assigned in the main clause governs the role in the subordinate one. (82) gives the morphology: pradhana-visaya saktih pratyayena abhidhiyate — the capacity whose domain is the *principal* is *expressed by the affix* (the finite verb's endings carry it); yada gune tada tadvat anukta api prakasate — when it is in the *subordinate*, then, *though unstated*, it *likewise shines forth*: the subordinate clause's role is not expressed and is understood, carried over from the principal. This is why 'having cooked, he eats' needs no agent for the cooking: the eater's agency propagates backward. The mechanism is 2.305's anupatta (what is not taken up by the word is nonetheless understood) operating across clause-boundaries.

3.7.83 unformalized

पचावनुक्तं यत्कर्म क्त्त्वान्ते भावाभिधायिनि ।
भुजौ शक्त्यन्तरेऽप्युक्ते तत्तद्धर्मः प्रकाशते ॥ ८३ ॥

pacāvanuktaṃ yatkarma kttvānte bhāvābhidhāyini |
bhujau śaktyantare'pyukte tattaddharmaḥ prakāśate || 83 ||

The object not stated for 'paci', in a gerund expressing the state — for 'bhuji', though another capacity is stated, that property of it shines forth.

The propagation of 3.7.81-82 in a worked case. Pacau anuktam yat karma: the *object not stated* for 'paci' (cooks) — in 'pakva bhunkte', 'having cooked, he eats', the cooking's object is not said. Ktvante bhava-abhidhayini: in a *gerund* expressing the *state* — the ktva-affix, which forms the gerund and expresses the action without an agent of its own. Bhujau sakty-antare 'pi ukte: for 'bhuji' (eats), *though another capacity is stated* — the eating has its own object, said. Tat tad-dharmah prakasate: *that property* of it *shines forth* — the unstated cooking-object is understood as the same as the stated eating-object. One cooks what one eats, and no word says so. The verse shows the propagation working across a difference of role (the rice is the cooking's nirvartya and the eating's prapya) — and it works because the roles ride on the buddhi's holding of the situation (3.7.3), not on the words.

3.7.84 unformalized

इषेश्च गमिसंस्पर्शाद् ग्रामे यो लो विधीयते ।
तत्रोषिणैव निर्भोगः क्रियते गतिकर्मणः ॥ ८४ ॥

iṣeśca gamisaṃsparśād grāme yo lo vidhīyate |
tatroṣiṇaiva nirbhogaḥ kriyate gatikarmaṇaḥ || 84 ||

And the 'la'-affix which is prescribed for 'isi', from its contact with 'gami', in 'the village' — there, by 'usi' alone, a separation of the going-object is effected.

A worked case of case-assignment under verb-contact. The construction is 'gramam icchati' — 'he wants the village', meaning he wants to *go* to the village: iseh gami-samsparsat, from *'isi'* (wanting) having *contact with* *'gami'* (going) — the wanting has a going absorbed in it (3.7.67's antarbhuta-kriyantara), and the village is the *going's* object, taking the accusative that 'icchati' alone could not license. Tatra usina eva nirbhogah kriyate gati-karmanah: there, by *'usi'* alone, a *separation* of the *going-object* is effected. The technical detail is obscure in transmission; the principle is not. When two actions fuse in one verb, the case-assignments come from the buried one — which is 3.7.73's finding about 'duhi' (the double accusative coming from an included causative) applied to a different fusion.
contested: the verse's technical apparatus (the la-affix, usi, nirbhoga) is compressed to the edge of recoverability and the transcription may be at fault; we translate the evident structure and note that the specific Paninian derivation intended is not confidently reconstructible

3.7.85 unformalized

पक्त्वा भुज्यत इत्यत्र केषांचिन्न व्यपेक्षते ।
ओदनं पचतिः सोऽसावनुमानात् प्रतीयते ॥ ८५ ॥

paktvā bhujyata ityatra keṣāṃcinna vyapekṣate |
odanaṃ pacatiḥ so'sāvanumānāt pratīyate || 85 ||

In 'having cooked, it is eaten' — for some, 'pacati' does not require the rice; that one is understood by inference.

The propagation of 3.7.83 disputed. Paktva bhujyate iti atra: in 'having cooked, it is eaten'. Kesamcid na vyapeksate odanam pacatih: for *some*, 'pacati' does *not require* the rice — the cooking-verb has no unstated object waiting to be supplied; it is simply a cooking. So 'apeksa', the expectancy that 3.7.83 relied on, is denied. Sa asau anumanat pratiyate: *that one* is understood *by inference* — the rice arrives, but by inference from the situation rather than by any grammatical expectancy. The difference matters: on 3.7.83's account the object is *carried over* by the construction; on this account it is *inferred* by the hearer, and the construction is silent. This is 2.149's mechanism (the corrupt form conveying by inference, through a recovery so practiced it vanishes) offered as a rival to grammatical propagation — and the two are hard to distinguish, since both deliver the rice and neither shows its working.

3.7.86 unformalized

तत्राभिनिविशौ कर्म यत्तिङन्तेऽभिधीयते ।
क्त्वान्तेऽधिकरणत्वेऽपि न तत्रेच्छन्ति सप्तमीम् ॥ ८६ ॥

tatrābhiniviśau karma yattiṅante'bhidhīyate |
ktvānte'dhikaraṇatve'pi na tatrecchanti saptamīm || 86 ||

There, for 'abhi-nivis', the object which is expressed in the finite verb — in the gerund, even where there is locus-hood, they do not desire the locative there.

A case where the propagation is grammatically visible. The verb abhi-nivis (to enter into, to settle upon) takes its object in the accusative by a special rule of Panini's (1.4.47: abhinivisas ca — the locus of abhinivis is called karman, not adhikarana). Tatra abhinivisau karma yat tinante 'bhidhiyate: the *object* which is expressed in the *finite verb*. Ktvante adhikaranatve 'pi na tatra icchanti saptamim: in the *gerund*, *even where there is locus-hood*, they do *not desire the locative* there. So the special accusative propagates into the gerund: even though the thing entered is by nature a locus, and gerunds might be expected to revert to the natural case, the rule's assignment carries across. The verse is evidence for 3.7.81-82's propagation-doctrine against 3.7.85's inference-account: here the carried-over role is *marked*, by the refusal of a case the situation would otherwise take.

3.7.87 unformalized

यन्निर्वृत्ताश्रयं कर्म प्राप्तेरप्रचितं पुनः ।
भक्ष्यादिविषयापत्त्या भिद्यमानं तदीप्सितम् ॥ ८७ ॥

yannirvṛttāśrayaṃ karma prāpterapracitaṃ punaḥ |
bhakṣyādiviṣayāpattyā bhidyamānaṃ tadīpsitam || 87 ||

That object whose substratum is accomplished, but which is not yet augmented by the attaining — being differentiated by the coming-about of the state of being edible and the like — that is what is desired.

The ipsitatama located with precision. Yan nirvrtta-asrayam karma: that object whose *substratum is accomplished* — the thing is there; the rice exists. Prapteh apracitam punah: but which is *not yet augmented by the attaining* — it has not yet been got; the eating has not happened. Bhaksyadi-visaya-apattya bhidyamanam: being *differentiated* by the *coming-about* of the state of being *edible* and the like — apatti: coming-to-be-such (3.7.64's term). Tad ipsitam: *that* is what is *desired*. So the desired object is the existing thing not yet attained, taken under the aspect it will have when attained: the rice, not yet eaten, wanted *as edible*. Every clause is doing work: existing (else there is nothing to want), unattained (else wanting is idle: 3.3.63-64's already-pregnant woman), and aspectually differentiated (one wants it *as* something). The definition by desire, which 3.7.74 attacked and 3.7.75 rescued, is here given its exact content.

3.7.88 unformalized

धातोरर्थान्तरे वृत्तेर्धात्वर्थेनोपसंग्रहात् ।
प्रसिद्धेरविवक्षातः कर्मणोऽकर्मिका क्रिया ॥ ८८ ॥

dhātorarthāntare vṛtterdhātvarthenopasaṃgrahāt |
prasiddheravivakṣātaḥ karmaṇo'karmikā kriyā || 88 ||

From the root's operating in another meaning; from the object's being included within by the root's meaning; from established usage; from non-intention — the action becomes objectless.

The four ways a transitive verb goes intransitive, and they are a complete taxonomy. Dhatoh arthantare vrtteh: from the *root's operating in another meaning* — the verb used in a sense that has no object ('he sees' meaning he understands). Dhatv-arthena upasangrahat karmanah: from the *object's being included within* by the root's meaning — the object absorbed into the verb, so that nothing is left to state (3.3.47's antarbhava: the self-bearing of 'asti', whence it takes no accusative). Prasiddheh: from *established usage* — the verb simply is not used with an object, whatever its sense allows (2.253's prasiddhi: currency deciding). Avivaksatah: from *non-intention* — the speaker does not mean to bring an object into view, and there is none (3.7.62's vina devadattadeh: the agent unmeant). Karmano akarmika kriya: the action becomes *objectless*. Four routes — semantic, absorptive, conventional, intentional — and only the first two are about the verb. The other two are about usage and the speaker, which is the samuddesa's constant finding.

3.7.89 unformalized

भेदा य एते चत्वारः सामान्येन प्रदर्शिताः ।
ते निमित्तादिभेदेन भिद्यन्ते बहुधा पुनः ॥ ८९ ॥

bhedā ya ete catvāraḥ sāmānyena pradarśitāḥ |
te nimittādibhedena bhidyante bahudhā punaḥ || 89 ||

These four differences that have been exhibited in general are again differentiated manifoldly, by difference of occasioning ground and the rest.

The taxonomy's status. Bhedah ya ete catvarah samanyena pradarsitah: these *four differences* that have been exhibited *in general* — 3.7.88's four routes to intransitivity. Te nimittadi-bhedena bhidyante bahudha punah: they are *again* *differentiated manifoldly*, by *difference of occasioning ground* and the rest. The four are not a bottom; each subdivides by nimitta, and the subdivision has no evident end (3.7.36's aparimita iva: seen as though unlimited, before the reduction to six). So the samuddesa reports its own taxonomies as provisional — general exhibitions that further analysis will multiply — which is 3.4.1's charter (kecit padarthah vaksyante samksepena yatha-agamam: certain word-meanings, in brief, according to tradition) honoured in the working.

3.7.90 unformalized

क्रियायाः परिनिष्पत्तिर्यद्व्यापारादनन्तरम् ।
विवक्ष्यते यदा तत्र करणत्वं तदा स्मृतम् ॥ ९० ॥

kriyāyāḥ pariniṣpattiryadvyāpārādanantaram |
vivakṣyate yadā tatra karaṇatvaṃ tadā smṛtam || 90 ||

When the complete accomplishment of the action is intended as immediately following upon whose operation — then instrument-hood is remembered there.

The instrument defined, and the definition is temporal. Kriyayah parinispattih yad-vyaparad anantaram: the *complete accomplishment* of the action *immediately following* upon whose *operation* — the karana is the participant whose contribution is the *last* one, after which the act is done: the axe's stroke, and the tree is felled. Vivaksyate yada: when it is *intended* so. Tatra karanatvam tada smrtam: *then* instrument-hood is remembered *there*. This is Panini's own definition (1.4.42: sadhakatamam karanam, the most effective means) recast: what makes something 'most effective' is that the accomplishment follows immediately on it. And vivaksyate is the samuddesa's constant: the immediacy is a matter of what the speaker construes as immediately-prior, not of any causal fact. The verse sets up 3.7.91, which says this outright.

3.7.91 unformalized

वस्तुतस्तदनिर्देश्यं न हि वस्तु व्यवस्थितम् ।
स्थाल्या पच्यत इत्येषा विवक्षा दृश्यते यतः ॥ ९१ ॥

vastutastadanirdeśyaṃ na hi vastu vyavasthitam |
sthālyā pacyata ityeṣā vivakṣā dṛśyate yataḥ || 91 ||

In reality it is not to be specified; for no thing stands settled — since this intention is seen: 'it is cooked with a pot'.

The samuddesa's most direct statement of its thesis, and it is delivered as a shrug. Vastutah tad anirdesyam: *in reality* it is *not to be specified* — there is no fact about which participant is the instrument. Na hi vastu vyavasthitam: for *no thing stands settled* — nothing in the world is an instrument (3.6.14's things without place, parts, motion, support; 3.7.8's nirihesu api bhavesu kalpana-upanibandhanam). And the evidence, which is a single ordinary sentence: sthalya pacyate iti esa vivaksa drsyate yatah — since *this intention* is *seen*: '*it is cooked with a pot*'. The pot. Not the fire, not the heat, not the cook's stirring — the *pot*, which does nothing but sit there, is Sanskrit's perfectly ordinary instrument of cooking. If the pot can be the karana, then the karana is not the most effective cause, and 3.7.90's definition tracks nothing but what the speaker chose to put in the instrumental. Sthalya pacyate is the samuddesa's kupa-supa-yupa: one everyday phrase, and the realist account of the karakas is finished.

3.7.92 unformalized

करणेषु तु संस्कारभारभन्ते पुनः पुनः ।
विनियोगविशेषांश्च प्रधानस्य प्रसिद्धये ॥ ९२ ॥

karaṇeṣu tu saṃskārabhārabhante punaḥ punaḥ |
viniyogaviśeṣāṃśca pradhānasya prasiddhaye || 92 ||

But in the instruments, they undergo refinements again and again, and particular applications, for the accomplishing of the principal.

The instruments' ritual dimension. Karanesu tu samskaran labhante punah punah: in the *instruments*, they undergo *refinements* *again and again* — samskara: the preparation, purification, making-fit that the rite performs on its implements (1.146's therapeutic triad; 3.3.35's samskara of the object by the word). Viniyoga-visesan ca: and *particular applications* — viniyoga, the deployment to a ritual station (2.258's mantras obtaining difference by their application). Pradhanasya prasiddhaye: for the *accomplishing of the principal*. So instruments are worked on, repeatedly, and assigned to stations, all in service of the rite's principal act. The verse returns the abstract karana-analysis to the sacrificial ground it came from: the ladle is consecrated, and the consecrating is what makes it the instrument — which is 3.7.91's thesis in ritual dress. Nothing is an instrument; things are *made* instruments, and in the rite the making is visible and liturgical.

3.7.93 unformalized

स्वकक्ष्यासु प्रकर्षश्च करणानां न विद्यते ।
आश्रितातिशयत्वं तु परतस्तत्र लक्षणम् ॥ ९३ ॥

svakakṣyāsu prakarṣaśca karaṇānāṃ na vidyate |
āśritātiśayatvaṃ tu paratastatra lakṣaṇam || 93 ||

And among the instruments, in their own compartments, no excellence exists; but the state of possessing a surpassing, resorted to from outside, is the mark there.

The objection of 3.7.74 revisited for the instrument, and answered as the object's was. Sva-kaksyasu prakarsah ca karananam na vidyate: among the *instruments*, in their *own compartments*, no *excellence* exists — one instrument is not more an instrument than another; instrument-hood does not grade (3.7.74's premise, granted). Asrita-atisayatvam tu paratah tatra laksanam: but the state of *possessing a surpassing*, resorted to *from outside*, is the *mark* there. Panini's definition is sadhakatamam — the *most* effective — and the superlative must qualify something. It qualifies not instrument-hood but effectiveness, which is conferred paratah, from outside: by the action's needs, by the speaker's construal. So the karana's definition works exactly as the karman's did (3.7.75: objecthood has no degrees, desire does): the superlative attaches to a gradable property that is not the role itself.

3.7.94 unformalized

स्वातन्त्र्येऽपि प्रयोक्तार आरादेवोपकुर्वते ।
करणेन हि सर्वेषां व्यापारो व्यवधीयते ॥ ९४ ॥

svātantrye'pi prayoktāra ārādevopakurvate |
karaṇena hi sarveṣāṃ vyāpāro vyavadhīyate || 94 ||

Even in their independence, the employers assist only from a distance; for the operation of all is separated by the instrument.

The agent's remoteness, and it explains the whole hierarchy. Svatantrye 'pi prayoktarah arad eva upakurvate: even in their *independence*, the *employers* — the agents, svatantra by Panini's definition (1.4.54) — assist *only from a distance*: arat, from afar (2.82's arad visesaka, the qualifier from afar). The agent does not touch the work. Karanena hi sarvesam vyaparo vyavadhiyate: for the *operation of all* is *separated by the instrument* — the instrument stands between every agent and every act. The woodsman does not fell the tree; the axe does, and he swings the axe. So agency is constitutively mediated: to be an agent is to act *through*, and what one acts through is the karana. This is 3.7.22's king and soldiers given its mechanism — the king obtains victory precisely because he does not fight — and it explains why 3.7.90 could define the instrument as what the accomplishment immediately follows: the instrument is where the action touches the world, and the agent is always one remove back.

3.7.95 unformalized

क्रियासिद्धौ प्रकर्षो वा न्यग्भावस्त्वेव कर्तरि ।
सिद्धौ सत्यां हि सामान्यं साधकत्वं प्रकृष्यते ॥ ९५ ॥

kriyāsiddhau prakarṣo vā nyagbhāvastveva kartari |
siddhau satyāṃ hi sāmānyaṃ sādhakatvaṃ prakṛṣyate || 95 ||

In the accomplishing of the action there is excellence; but in the agent there is only subordination. For when the accomplishment exists, the general means-hood is surpassed.

The hierarchy's paradox, stated exactly. Kriya-siddhau prakarso va: in the *accomplishing of the action* there is *excellence* — the instrument, being nearest the accomplishment (3.7.90), has the surpassing. Nyag-bhavas tv eva kartari: but in the *agent* there is *only subordination* — nyag-bhava: the being-pressed-down (3.7.59's nyag-bhavana). The agent is *subordinated*? The independent one, the svatantra? Yes — and the reason: siddhau satyam hi samanyam sadhakatvam prakrsyate — when the *accomplishment* exists, the *general means-hood* is *surpassed*. Measured against the accomplishment, the agent's contribution is the most distant and the least effective (3.7.94's arat), so in the ranking-by-effectiveness he comes last. The grammar's most independent karaka is efficacy's least. This is the samuddesa's finest structural irony, and it is not a paradox: independence and effectiveness are different scales, and the karta tops one by placing last on the other.

3.7.96 unformalized

अस्यादीनां तु कर्तृत्वे तैक्ष्ण्यादि करणं विदुः ।
तैक्ष्ण्यादीनां स्वतन्त्रत्वे द्वैधात्मा व्यवतिष्ठते ॥ ९६ ॥

asyādīnāṃ tu kartṛtve taikṣṇyādi karaṇaṃ viduḥ |
taikṣṇyādīnāṃ svatantratve dvaidhātmā vyavatiṣṭhate || 96 ||

But in the agency of a sword and the like, they know sharpness and the rest to be the instrument; and in the independence of sharpness and the rest, the self stands two-fold.

The regress of instruments, and it does not terminate. Asy-adinam tu kartrtve taiksnyadi karanam viduh: in the *agency* of a *sword* and the like — when we say 'the sword cuts', the sword is agent — *sharpness* and the rest are the *instrument*: the sword cuts *by* its sharpness (3.7.31's four properties). So the sword, promoted to agent, needs an instrument of its own, and its own property serves. Taiksnyadinam svatantratve dvaidha atma vyavatisthate: and in the *independence* of sharpness and the rest — promote the sharpness to agent ('the sharpness cuts') — the *self* stands *two-fold*: it must now be both agent and (through some further property) instrument. The regress is 3.7.20-21's nesting run downward without a floor: every agent needs an instrument, every instrument can be promoted to agent, and each promotion generates a new instrument. Dvaidha atma — the self standing two-fold — names the splitting that each step requires.

3.7.97 unformalized

आत्मभेदेऽपि सत्येवमेकोऽर्थः स तथा स्थितः ।
तदाश्रयत्वाद् भेदेऽपि कर्तृत्वं बाधकं ततः ॥ ९७ ॥

ātmabhede'pi satyevameko'rthaḥ sa tathā sthitaḥ |
tadāśrayatvād bhede'pi kartṛtvaṃ bādhakaṃ tataḥ || 97 ||

Even when there is a difference of self, one and the same object stands thus; from its resting on that, agency, though there is difference, is therefore what annuls.

The regress of 3.7.96 stopped. Atma-bhede 'pi sati evam ekah arthah sa tatha sthitah: even when there is a *difference of self* — even granting the splitting — *one and the same* object stands thus: the sword and its sharpness are not two things; the splitting was analytical (3.7.15's vyatirekam upasritya: the construction resorting to separation, whatever the metaphysics). Tad-asrayatvad bhede 'pi kartrtvam badhakam tatah: from its *resting on that*, *agency*, though there is difference, is *therefore what annuls*. Agency wins: where sharpness and sword are one thing, the agency-assignment overrides the instrument-assignment, and the regress stops because there is nothing further to promote. Badhaka — annulling: the badha-machinery (2.66, 2.73's hierarchy of evidences) adjudicating between two karaka-assignments to one thing. So the regress is real as analysis and idle as ontology, and the grammar stops it by ranking.

3.7.98 unformalized

यथा च सन्निधानेन करणत्वं प्रतीयते ।
तथैवासन्निधानेऽपि क्रियासिद्धेः प्रतीयते ॥ ९८ ॥

yathā ca sannidhānena karaṇatvaṃ pratīyate |
tathaivāsannidhāne'pi kriyāsiddheḥ pratīyate || 98 ||

And as instrument-hood is understood from proximity — just so, even in the absence of proximity, it is understood from the accomplishing of the action.

The instrument's criterion loosened. Yatha ca sannidhanena karanatvam pratiyate: as instrument-hood is understood from *proximity* — the axe in hand, the pot on the fire (3.7.91's sthalya pacyate). Tatha eva asannidhane 'pi kriya-siddheh pratiyate: *just so*, even in the *absence of proximity*, it is understood from the *accomplishing* of the action. Distant things are instruments too: one conquers by an army not present, cooks by a fire one is not touching. So proximity was never the criterion; the accomplishment's dependence was (3.7.90's parinispattih yad-vyaparad anantaram — and 'immediately following' is a matter of construal, not of distance). The verse removes the last physical residue from the karana's definition: instrument-hood is a functional and construal-relative posting, and nothing about nearness survives in it.

3.7.99 unformalized

सतोकस्य वाभिनिर्वृत्तेरनिवृत्तेश्च तस्य वा ।
प्रसिद्धिं करणत्वस्य स्तोकादीना प्रचक्षते ॥ ९९ ॥

satokasya vābhinirvṛtteranivṛtteśca tasya vā |
prasiddhiṃ karaṇatvasya stokādīnā pracakṣate || 99 ||

From the coming-about of what is little, or from its not ceasing — they declare the establishing of instrument-hood for 'little' and the like.

A hard case: adverbial instrumentals. Stokasya va abhinirvrtteh anivrtteh ca tasya va: from the *coming-about* of what is *little*, or from its *not ceasing*. The construction is 'stokena mukta' — 'released by a little', 'he escaped by a hair' — where 'little' takes the instrumental and is not an instrument in any recognizable sense: no thing is being used. Prasiddhim karanatvasya stokadina pracaksate: they declare the *establishing* of instrument-hood for *'little'* and the like. Two accounts are offered, and both are functional: the littleness *came about* (and its coming-about is what the escape followed on: 3.7.90's criterion), or it *did not cease* (and its persisting is what the escape depended on). Either way, the instrumental marks what the accomplishment hung on — which in this case is a quantity, not a thing. The verse shows the karana-slot accepting what is not remotely an instrument, because the slot tracks dependence and not instrumentality.

3.7.100 unformalized

धर्माणां तद्वता भेदादभेदाच्च विशिष्यते ।
क्रियावधेपवच्छेदविशेषाद् भिद्यते यथा ॥ १०० ॥

dharmāṇāṃ tadvatā bhedādabhedācca viśiṣyate |
kriyāvadhepavacchedaviśeṣād bhidyate yathā || 100 ||

It is distinguished by the difference and the non-difference of the properties from what possesses them — as it is differentiated by a difference in the delimitation of the action's limit.

The general principle behind 3.7.99. Dharmanam tadvata bhedad abhedac ca visisyate: it is *distinguished* by the *difference* and the *non-difference* of the *properties* from *what possesses them* — a property may be taken as one with its bearer or apart from it (3.7.96-97's sword and sharpness), and the karaka-assignment follows the taking. Kriya-avaccheda-visesad bhidyate yatha: *as* it is differentiated by a difference in the *delimitation of the action's limit* — where one draws the action's boundary decides what falls inside it as participant and what stands outside as its instrument. So two variables govern every karaka-assignment: how one takes the property's relation to its bearer, and where one cuts the action. Both are construals (3.7.3's buddhy-avastha-nibandhanah), and between them they generate the whole taxonomy the samuddesa has been cataloguing.

3.7.101-102 unformalized

प्रागन्यतः शक्तिलाभान्न्यग्भावापादनादपि ।
तदधीनप्रवृत्तित्वात् प्रवृत्तानां निवर्तनात् ॥ १०१ ॥
अदृष्टत्वात् प्रतिनिधेः प्रविवेके च दर्शनात् ।
आरादप्युपकारित्वे स्वातन्त्र्यं कर्तुरुच्यते ॥ १०२ ॥

prāganyataḥ śaktilābhānnyagbhāvāpādanādapi |
tadadhīnapravṛttitvāt pravṛttānāṃ nivartanāt || 101 ||
adṛṣṭatvāt pratinidheḥ praviveke ca darśanāt |
ārādapyupakāritve svātantryaṃ karturucyate || 102 ||

(101) From the obtaining of capacity previously from another; from the bringing-about of subordination; from their operating in dependence on that; from the turning-back of those who have set out — (102) from the non-observation of a substitute; and from its being seen in separation — even where the assistance is from a distance, independence is said of the agent.

Six marks of agency, gathered into a definition. The agent is identified by: (i) prag anyatah sakti-labhat — the others *obtaining capacity previously from* him (he empowers them); (ii) nyag-bhava-apadanat — his *bringing about their subordination* (3.7.20's nyakkrta); (iii) tad-adhina-pravrttitvat — their *operating in dependence* on him; (iv) pravrttanam nivartanat — his *turning back those who have set out* (he can stop what he started: the mark of control); (v) adrstatvat pratinidheh — the *non-observation of a substitute* for him (agents are not substitutable: 3.1.78's test, which discriminates what stands near the capacity from what stands at a remove); (vi) prativeke ca darsanat — his *being seen in separation* (he is identifiable apart from the act). And the conclusion: arad api upakaritve svatantryam kartur ucyate — *even where the assistance is from a distance* (3.7.94's arat), *independence* is said of the *agent*. Six operational marks, no metaphysics: the karta is whoever exhibits these, and 3.7.103 will say what that means.

3.7.103 unformalized

धर्मैरभ्युदितैः शब्दे नियमो न तु वस्तुनि ।
कर्तुर्धर्मविवक्षायां शब्दात् कर्ता प्रतीयते ॥ १०३ ॥

dharmairabhyuditaiḥ śabde niyamo na tu vastuni |
karturdharmavivakṣāyāṃ śabdāt kartā pratīyate || 103 ||

By the properties that have arisen, the restriction is in the word, not in the thing; when the agent's property is intended, the agent is understood from the word.

The six marks of 3.7.101-102 placed, and the placing is the samuddesa's thesis. Dharmaih abhyuditaih sabde niyamah na tu vastuni: by the *properties* that have *arisen*, the *restriction* is *in the word*, *not in the thing*. The six marks do not identify a thing that is the agent; they govern when the *word* takes agent-morphology. Kartur dharma-vivaksayam sabdat karta pratiyate: when the *agent's property* is *intended*, the *agent* is *understood from the word*. So agency is a fact about what is intended and expressed, and the marks are the conditions under which the expression is licensed. Na tu vastuni — not in the thing: the phrase is 3.7.91's (vastutah tad anirdesyam, na hi vastu vyavasthitam) applied now to the karta as it was there to the karana. Neither end of the karaka-system rests on anything.

3.7.104 unformalized

एकस्य बुद्ध्यवस्थाभिर्भेदे च परिकल्पिते ।
कर्मत्वं करणत्वं च कर्तृत्वं चोपजायते ॥ १०४ ॥

ekasya buddhyavasthābhirbhede ca parikalpite |
karmatvaṃ karaṇatvaṃ ca kartṛtvaṃ copajāyate || 104 ||

And when a difference is constructed, by states of the intellect, for one and the same thing — objecthood, instrument-hood and agency arise.

The samuddesa's culminating statement, and it gathers a hundred verses. Ekasya buddhy-avasthabhih bhede ca parikalpite: when a *difference* is *constructed*, by *states of the intellect*, for *one and the same thing* — one thing, and the buddhi's states carve it. Karmatvam karanatvam ca kartrtvam ca upajayate: *objecthood*, *instrument-hood* and *agency* *arise*. All three, from one thing, by cognitive states. This is 3.7.3 (the practice of means is grounded in a state of the intellect) now demonstrated rather than asserted, and it explains everything the samuddesa has found: why the same cooking is object, instrument and sadhya at once (2.429); why both parents are agents and the roles are assigned by intention (3.7.19); why the pot is the instrument of cooking (3.7.91). Buddhy-avastha — states of the intellect: not arbitrary, not private, but *states*, which come and go and can be occupied. The karaka-system is a map of the positions a mind can take up toward an event.

3.7.105 unformalized

उत्पत्तेः प्रागसद्भावो बुद्धयवस्थानिबन्धनः ।
अविशिष्टः सताऽन्येन कर्ता भवति जन्मनः ॥ १०५ ॥

utpatteḥ prāgasadbhāvo buddhayavasthānibandhanaḥ |
aviśiṣṭaḥ satā'nyena kartā bhavati janmanaḥ || 105 ||

The prior non-existence before the arising is grounded in a state of the intellect; not distinguished from another existent thing, it becomes the agent of the birth.

The doctrine of 3.7.104 applied to its hardest case: origination. Utpatteh prag asad-bhavah buddhy-avastha-nibandhanah: the *prior non-existence* before the *arising* is *grounded in a state of the intellect* — the pot's not-yet-being is a way the mind holds the situation, not a condition the pot is in (3.3.79's finding: 'the prior state of a non-thing' is not a describable condition; neither the priority nor the state exists in a non-thing). Avisistah sata anyena karta bhavati janmanah: *not distinguished* from another *existent* thing, it *becomes the agent of the birth*. So 'the pot comes to be' has the pot as agent — and the pot, at that moment, is a buddhi-posting indistinguishable from any existing thing, because buddhi does not distinguish them (3.3.86's mind that will not halt in non-being). This is 3.3.43-48's dilemma of origination dissolved by the karaka-analysis: nothing is born, and the birth's agent is a state of the intellect, and the grammar conjugates it.

3.7.106 unformalized

कारणं कार्यभावेन यथा वा व्यवतिष्ठते ।
कार्यशब्दं तदा लब्ध्वा कार्यत्वेनाथ जायते ॥ १०६ ॥

kāraṇaṃ kāryabhāvena yathā vā vyavatiṣṭhate |
kāryaśabdaṃ tadā labdhvā kāryatvenātha jāyate || 106 ||

Or, as a cause stands in the form of the effect — then, having obtained the word 'effect', it thereupon arises as the effect.

The alternative account of origination. Karanam karya-bhavena yatha va vyavatisthate: *or*, as a *cause* stands in the *form of the effect* — the clay standing as the pot, the satkarya-vadin's picture (3.7.49's sat va janmana yat prakasyate). Karya-sabdam tada labdhva karyatvena atha jayate: *then*, *having obtained the word 'effect'*, it *thereupon arises as the effect*. The order is the verse's shock: the thing obtains the *word* first, and *then* arises as the effect. Naming is prior to being. This is not a slip — it is 3.1.6's doctrine (every word expresses its own universal first, and superimposes it on the object) applied to origination: what makes the clay into a pot is its receiving the word 'pot', and the arising follows the naming. The Padakanda's constructivism reaches its limit here, and the limit is that things come to be by being called.

3.7.107 unformalized

यथाहेः कुण्डलीभावो व्यग्राणां च समग्रता ।
तथैव जन्मरूपत्वं सतामेके प्रचक्षते ॥ १०७ ॥

yathāheḥ kuṇḍalībhāvo vyagrāṇāṃ ca samagratā |
tathaiva janmarūpatvaṃ satāmeke pracakṣate || 107 ||

As the snake's becoming-coiled, and the wholeness of things dispersed — just so, some declare, is the birth-form of existent things.

Two similes for a birth that is no birth. Yatha aheh kundali-bhavah: as the *snake's becoming-coiled* — the snake does not acquire a new thing by coiling; it arranges itself, and 'the coil' comes to be without anything coming to be. Vyagranam ca samagrata: and the *wholeness* of things *dispersed* — scattered things gathered, and the 'whole' arises without any new entity: the heap is not a further thing beside the grains. Tatha eva janma-rupatvam satam eke pracaksate: *just so*, some declare, is the *birth-form* of *existent* things. So birth is rearrangement and gathering — configuration, not creation. The similes are chosen for cases where everyone already sees that the 'new thing' is a way of taking the old: nobody thinks a coil was manufactured. And they complete 3.3.61's finding (non-being is not born as being; the two are constructions, not other than one self) with a picture of what birth then *is*.

3.7.108 unformalized

विभक्तयोनि यत्कार्यं कारणेभ्यः प्रवर्तते ।
स्वा जातिर्व्यक्तिरूपेण तस्यापि व्यवतिष्ठते ॥ १०८ ॥

vibhaktayoni yatkāryaṃ kāraṇebhyaḥ pravartate |
svā jātirvyaktirūpeṇa tasyāpi vyavatiṣṭhate || 108 ||

The effect which proceeds from its causes, having a divided womb — its own universal also stands established in the form of the individual.

The effect's universal, and the verse closes the movement by returning to the Jati-samuddesa. Vibhakta-yoni yat karyam karanebhyah pravartate: the *effect* which proceeds *from its causes*, having a *divided womb* — its origin is in several causes, each contributing (yoni: 3.1.3's womb-vocabulary, 2.58's samhita as the womb of the words). Sva jatih vyakti-rupena tasya api vyavatisthate: *its own universal* also stands established *in the form of the individual*. So the effect, however divided its origin, has its own universal, and that universal stands *as* the individual — the pot's pot-hood is not a further item in the pot but the pot's own standing (3.1.29's face manifested by water: the universal itself is what is expressed). This connects the two ends of the Padakanda: the universals of samuddesa 1 and the causal analysis of samuddesa 7 meet in the effect, whose being-a-thing-of-a-kind is the same fact as its having been made.

3.7.109 unformalized

भावष्वेव पदन्यासः प्रज्ञाया वाच एव वा ।
नास्तीत्यप्यपदे नास्ति न चासद्भिद्यते ततः ॥ १०९ ॥

bhāvaṣveva padanyāsaḥ prajñāyā vāca eva vā |
nāstītyapyapade nāsti na cāsadbhidyate tataḥ || 109 ||

The setting-down of a foot is only among existent things — whether of cognition or of speech; even 'it does not exist' does not exist in a footless place, nor is the non-existent thereby differentiated.

The footless station (3.3.80's apada-sthana) recalled and applied to the karaka-system. Bhavesu eva pada-nyasah: the *setting-down of a foot* is only among *existent* things — prajnaya vacah eva va: whether of *cognition* or of *speech*: neither thought nor word can get a footing where nothing is (3.3.78's karana that cannot set its foot in a non-place). Na asti iti api apade na asti: even *'it does not exist'* does *not exist* in a *footless place* — the very denial needs a floor, and where none is, the denial cannot be made either (3.3.80: no ground for saying 'it exists' or 'it does not exist'). Na ca asad bhidyate tatah: nor is the *non-existent* thereby *differentiated*: our denials do not carve nothing into kinds. So the whole apparatus — every karaka, every case, every construal — operates only where there is standing-room, and 3.3.80's station remains where speech stops.

3.7.110 unformalized

बुद्धिशब्दौ प्रवर्तेते यथा भूतेषु वस्तुषु ।
तेषामन्येन तत्त्वेन व्यवहारो न विद्यते ॥ ११० ॥

buddhiśabdau pravartete yathā bhūteṣu vastuṣu |
teṣāmanyena tattvena vyavahāro na vidyate || 110 ||

As intellect and word proceed among things that have come to be — of those there is no practice by any other reality.

The consequence of 3.7.109, and it is a boundary-marker. Buddhi-sabdau pravartete yatha bhutesu vastusu: as *intellect* and *word* proceed among things *that have come to be* — the two operate on the arisen, the constructed, the standing (3.3.82's beings raised up by construction). Tesam anyena tattvena vyavaharo na vidyate: of *those* there is *no practice* by any *other reality*. There is no second reality in which thought and speech could transact — no access to things as they are apart from having come to be. This is 2.297's charter (words are grounded in the world) and 3.3.71's cost (when reality and otherness are both abandoned, practice withdraws) stated together: the domain of buddhi and sabda is the arisen, and it has no outside that they could reach.

3.7.111 unformalized

आकाशस्य यथा भेदश्छायायाश्चलनं यथा ।
जन्मनाशावभेदेऽपि तथा कैश्चित्प्रकल्पितौ ॥ १११ ॥

ākāśasya yathā bhedaśchāyāyāścalanaṃ yathā |
janmanāśāvabhede'pi tathā kaiścitprakalpitau || 111 ||

As the division of ether, and as the moving of a shadow — so, though there is no difference, birth and destruction are constructed by some.

Two similes for constructed events. Akasasya yatha bhedah: as the *division of ether* — akasa is partless, and we speak of its regions (3.1.15's places constructed by the properties of what is conjoined). Chayayah calanam yatha: as the *moving of a shadow* — a shadow does not move; the light and the object move, and the shadow's 'motion' is a way of describing their change. Nothing travels. Janma-nasau abhede 'pi tatha kaiscit prakalpitau: so, *though there is no difference*, *birth and destruction* are *constructed* by some. Both similes are cases where a real appearance is generated by nothing doing what it seems to do — and both are cases nobody disputes. The shadow is especially apt: its motion is perfectly determinate, perfectly predictable, and there is no mover. So with birth: the event is exact and nothing is born (3.7.107's coiling snake).

3.7.112 unformalized

यथैवाकाषनास्तित्वमसन्नूर्तिनिरूपितम् ।
तथैव मूर्तिनास्तित्वमसदाकाशनिश्रयम् ॥ ११२ ॥

yathaivākāṣanāstitvamasannūrtinirūpitam |
tathaiva mūrtināstitvamasadākāśaniśrayam || 112 ||

Just as the non-existence of ether is determined by a non-existent corporeality — so the non-existence of corporeality rests upon a non-existent ether.

The mutual constitution of absences, and its symmetry is the argument. Yatha eva akasa-nastitvam asan-murti-nirupitam: the *non-existence of ether* is *determined* by a *non-existent* corporeality — 'there is no space here' means 'this is occupied', and the occupying body is itself constructed. Tatha eva murti-nastitvam asad-akasa-nisrayam: so the *non-existence of corporeality* rests upon a *non-existent ether* — 'there is nothing here' means 'this is empty space', and the space is constructed too. Each absence is defined by the other's presence, and each presence is itself an absence's construction. So the pair spins: space's absence needs body, body's absence needs space, and neither is anywhere independently. This is 3.3.60's mutual dependence (resting on being, absence is followed; resting on absence, being is followed) given its most concrete case — and the concreteness is the point: emptiness and occupancy, the plainest facts about a place, are each other's shadows.

3.7.113 unformalized

यथा तदर्थैर्व्यापारैः क्रियात्मा व्यपदिश्यते ।
अभेदग्रहणादेष कार्यकारणयोः क्रमः ॥ ११३ ॥

yathā tadarthairvyāpāraiḥ kriyātmā vyapadiśyate |
abhedagrahaṇādeṣa kāryakāraṇayoḥ kramaḥ || 113 ||

As the self of the action is designated by the operations that are for its sake — this sequence of cause and effect is from the grasping of non-difference.

The causal sequence explained by a grasp. Yatha tad-arthaih vyaparaih kriya-atma vyapadisyate: as the *self of the action* is *designated* by the *operations that are for its sake* — the cooking is named by the fetching, lighting, stirring that serve it; the whole is designated through its parts (2.448's uncounted sub-actions displayed by a single verb). Abheda-grahanad esa karya-karanayoh kramah: *this sequence* of *cause and effect* is from the *grasping of non-difference*. The causal order exists because we grasp the parts as *not different* from the whole they serve: the stirring is the cooking, and so the stirring is its cause. Take them as different and no sequence appears — just events. So causation is generated by the same abheda-grahana that generates the action's unity, and cause-and-effect is the inside of a whole we have taken as one (3.7.107's samagrata: the wholeness of things dispersed).

3.7.114 unformalized

विकारो जन्मनः कर्ता प्रकृतिर्वेति संशये ।
भिद्यते प्रतिपत्तॄणां दर्शनं लिङ्गदर्शनैः ॥ ११४ ॥

vikāro janmanaḥ kartā prakṛtirveti saṃśaye |
bhidyate pratipattṝṇāṃ darśanaṃ liṅgadarśanaiḥ || 114 ||

When there is doubt whether the modification is the agent of the birth, or the material cause — the view of the comprehenders is differentiated by the seeing of marks.

The open question, and its resolution is left to evidence. Vikaro janmanah karta prakrtir va iti samsaye: when there is *doubt* whether the *modification* is the *agent of the birth* — the pot, coming to be, being the agent of its own arising (3.7.105) — *or* the *material cause*: the clay, being what does the arising. Bhidyate pratipattrnam darsanam linga-darsanaih: the *view* of the comprehenders is *differentiated* by the *seeing of marks* — different hearers construe differently, according to what indications they find (2.73's linga; 2.315's fourteen determinants). So the question has no answer of its own: the karaka-assignment for origination follows the marks the context supplies, and comprehenders differ. This is 2.135's doctrine (launched in one way, divided out among differing comprehenders each according to his own cognition) applied to the most metaphysically loaded construction in the language.

3.7.115 unformalized

क्लृपि सम्पद्यमाने या चतुर्थी सा विकारतः ।
सुवर्णपिण्डे प्रकृतौ वचनं कुण्डलाश्रयम् ॥ ११५ ॥

klṛpi sampadyamāne yā caturthī sā vikārataḥ |
suvarṇapiṇḍe prakṛtau vacanaṃ kuṇḍalāśrayam || 115 ||

In 'klp', when there is a becoming, the fourth case is from the modification; in the lump of gold, the material cause, the statement rests upon the earring.

The grammar of becoming, and it assigns by the outcome. Klpi sampadyamane ya caturthi sa vikaratah: in the root *klp*, when there is a *becoming*, the *fourth case* (dative) is *from the modification* — 'suvarna-pindah kundalaya kalpate', 'the gold-lump serves for an earring': the dative goes to the *product*, not to the material. Suvarna-pinde prakrtau vacanam kundala-asrayam: in the *lump of gold*, the *material cause*, the *statement* rests upon the *earring*. So when Sanskrit says a material 'becomes' something, the sentence is oriented toward the product: the earring gets the dative, and the gold is the subject. The case-assignment records which end of the transformation the speaker is facing — and 3.7.48 already established that whether the material is in view at all is a matter of vivaksa. The dative here is a small piece of evidence for a large thesis: the grammar of becoming is a grammar of attention.

3.7.116 unformalized

वाक्ये सम्पद्यते कर्ता संघच्व्यन्तस्य कथ्यते ।
वृत्तौ संधीभवन्तीति ब्राह्मणानां स्वतन्त्रता ॥ ११६ ॥

vākye sampadyate kartā saṃghacvyantasya kathyate |
vṛttau saṃdhībhavantīti brāhmaṇānāṃ svatantratā || 116 ||

In a sentence, the agent 'becomes'; of a form ending in 'cvi' with 'sangha', it is stated; in the derivative formation, 'they become united' — the independence of the brahmins.

The cvi-construction (2.279's) revisited for its karaka-structure. Vakye sampadyate karta: in a *sentence*, the *agent* 'becomes' — the analytic phrase makes the transforming thing the agent: 'the brahmins become a group'. Sangha-cvy-antasya kathyate: of a form ending in *cvi* with 'sangha' — the synthetic form, sanghi-bhavanti, where the cvi-affix compresses the becoming. Vrttau sanghi-bhavanti iti brahmananam svatantrata: in the *derivative formation*, 'they *become united*' — the *independence of the brahmins*. The point is that the brahmins remain svatantra, agents, through both constructions: the analytic phrase and the compressed cvi-form assign agency identically, though their morphology differs entirely. So the karaka-structure survives the form-change — which is 2.415's doctrine (the sentence and its meaning shining forth undivided from words that possess sequence) at the level of a single derivation.

3.7.117 unformalized

अत्वं सम्पद्यते यस्त्वं न तस्मिन् युष्मदाश्रया ।
प्रकृतिः पुरुषस्यास्ति प्राकृतः स विधीयते ॥ ११७ ॥

atvaṃ sampadyate yastvaṃ na tasmin yuṣmadāśrayā |
prakṛtiḥ puruṣasyāsti prākṛtaḥ sa vidhīyate || 117 ||

You who become 'you' — in him there is no material cause resting on 'yusmad'; the man has a material nature, and he is prescribed as natural.

A puzzle in the cvi-construction's pronominal cases. Atvam sampadyate yah tvam: *you* who *become 'you'* — the construction 'tvam-bhavati' (becomes you) faces a difficulty: na tasmin yusmad-asraya prakrtih asti — in him there is *no material cause* resting on *'yusmad'* (the pronominal stem for 'you'): there is no such stuff as you-ness for someone to be made of. The cvi-construction normally has a material becoming a product (3.7.115's gold and earring), and here the 'product' is a person and there is no material. Prakrtih purusasya asti prakrtah sa vidhiyate: the *man* has a *material nature*, and he is *prescribed as natural* — the material is the person himself, and the grammar treats him as the prakrti. So the construction is saved by locating the material in the person rather than in the pronoun's sense — which is a technical fix with a philosophical shadow: becoming-you has no stuff, and the grammar supplies one by fiat.

3.7.118 unformalized

पूर्वामवस्थामजहत्संस्पृशन् धर्ममुत्तरम् ।
समूर्च्छित इवार्थात्मा जायमानोऽभिधीयते ॥ ११८ ॥

pūrvāmavasthāmajahatsaṃspṛśan dharmamuttaram |
samūrcchita ivārthātmā jāyamāno'bhidhīyate || 118 ||

Not abandoning the prior state, touching the subsequent property — the meaning's self, as though congealed, is spoken of as 'being born'.

The verse that resolves the origination-problem within the karaka-system, and its central word is chosen with care. Purvam avastham ajahat: *not abandoning* the *prior state* — the thing does not leave what it was (3.3.74's the self of being is not abandoned). Samsprsan dharmam uttaram: *touching* the *subsequent property* — it reaches the new condition without having left the old. Samurcchita iva artha-atma: the *meaning's self*, *as though congealed* — samurcchita: coagulated, curdled, thickened into shape (the word is used of milk becoming curds, and of things taking form). Jayamano 'bhidhiyate: is *spoken of as 'being born'*. So birth is a congealing between two states, holding both — which is 3.3.39's interval (fallen from the prior property, not having attained the subsequent, and in that interval birth is spoken of) now given a positive image. The iva marks it: as though congealed. Nothing thickens; the thickening is how the two-stateness is taken.

3.7.119 unformalized

सव्यापारतरः कश्चित् क्लचिद्धर्मः प्रतीयते ।
संसृज्यन्ते च भावानां भेदवत्योऽपि शक्तयः ॥ ११९ ॥

savyāpārataraḥ kaścit klaciddharmaḥ pratīyate |
saṃsṛjyante ca bhāvānāṃ bhedavatyo'pi śaktayaḥ || 119 ||

Somewhere a certain property is understood as more possessed of operation; and the capacities of things, though possessing difference, are commingled.

Why one property among many gets the karaka-assignment. Sa-vyaparataraah kascit kvacid dharmah pratiyate: *somewhere* a certain *property* is understood as *more possessed of operation* — sa-vyaparatara: comparatively more operative. Among the sword's sharpness, weight, hardness and shape (3.7.31), one stands out as doing the work, and it takes the instrumental. Samsrjyante ca bhavanam bheda-vatyo 'pi saktayah: and the *capacities* of things, *though possessing difference*, are *commingled*. The properties are distinct and they act together, and the assignment picks one — the more-operative — from a fused ensemble. So the karana-assignment is a selection from a commingling (3.4.1's samsarga-rupat sambhutah), made by comparative operativeness, and the comparison is the speaker's (3.7.91's sthalya pacyate: the pot, which does nothing, chosen).

3.7.120 unformalized

विपरीतार्थवृत्तित्वं पुरुषस्य विपर्यये ।
गम्येत साधनं ह्यत्र सव्यापारं प्रतीयते ॥ १२० ॥

viparītārthavṛttitvaṃ puruṣasya viparyaye |
gamyeta sādhanaṃ hyatra savyāpāraṃ pratīyate || 120 ||

In the inversion, the man's operating in a contrary meaning would be understood; for here the means is understood as possessed of operation.

The cvi-construction inverted. Viparita-artha-vrttitvam purusasya viparyaye: in the *inversion*, the *man's* *operating in a contrary meaning* would be understood — reverse the construction and the person is construed as acting oppositely. Gamyeta sadhanam hi atra sa-vyaparam pratiyate: for *here* the *means* is understood as *possessed of operation* — the inverted construction makes the means operative where it was not. The technicality is dense and the principle is 3.7.119's: which element is taken as operative decides the assignment, and inverting the construction inverts the taking. The verse is preparing 3.7.121's example, where a king becomes a servant and the grammar must decide who is doing what.

3.7.121 unformalized

त्वमन्यो भवसीत्येषा तत्र स्यात् परिकल्पना ।
राज्ञि भृत्यत्वमापन्ने यथा तद्वद्गतिर्भवेत् ॥ १२१ ॥

tvamanyo bhavasītyeṣā tatra syāt parikalpanā |
rājñi bhṛtyatvamāpanne yathā tadvadgatirbhavet || 121 ||

'You become another' — that would be the construction there; as when a king has come to servanthood, the movement would be like that.

The example that makes 3.7.120 vivid. Tvam anyo bhavasi iti esa tatra syat parikalpana: *'you become another'* — *that* would be the *construction* there. And the illustration: rajni bhrtyatvam apanne yatha tadvad gatir bhavet — as when a *king* has come to *servanthood*, the *movement* would be *like that*. The king who becomes a servant is the same man and is another; the construction 'the king becomes a servant' has him as agent of his own demotion, and what he becomes is what he was not. The example is chosen because the identity is undeniable (he is the same man) and the otherness is real (he is not a king now) — so 'you become another' is exactly right and exactly paradoxical. And it is 3.7.118's congealing: not abandoning the prior state, touching the subsequent, the meaning's self as though thickened between them.

3.7.122-123 unformalized

सम्भावनात् क्रियासिद्धौ कर्तृत्वेन समाश्रितः ।
क्रियायामात्मसाध्यायां साधनानां प्रयोजकः ॥ १२२ ॥
प्रयोगमात्रे न्यग्भावं स्वातन्त्र्यादेव निश्रितः ।
अविशिष्टो भवत्यन्यैः स्वतन्त्रैर्मुक्तसंशयैः ॥ १२३ ॥

sambhāvanāt kriyāsiddhau kartṛtvena samāśritaḥ |
kriyāyāmātmasādhyāyāṃ sādhanānāṃ prayojakaḥ || 122 ||
prayogamātre nyagbhāvaṃ svātantryādeva niśritaḥ |
aviśiṣṭo bhavatyanyaiḥ svatantrairmuktasaṃśayaiḥ || 123 ||

(122) From possibility, in the accomplishing of the action, he is resorted to as agent; in an action to be accomplished by himself, he is the occasioner of the means. (123) In mere employment, he is settled in subordination by his very independence; he becomes undistinguished from other independent ones whose doubt is removed.

The agent's two faces. (122): sambhavanat kriya-siddhau kartrtvena samasritah — from *possibility*, in the accomplishing of the action, he is *resorted to as agent*: agency is conferred on whoever could do it. Kriyayam atma-sadhyayam sadhananam prayojakah: in an action *to be accomplished by himself*, he is the *occasioner of the means* — he sets the instruments going (3.7.101's six marks). (123) gives the paradox: prayoga-matre nyag-bhavam svatantryad eva nisritah — in *mere employment*, he is *settled in subordination* *by his very independence*. His independence is what subordinates him — because to be the free one is to stand back and let the instruments touch the work (3.7.94's arat, 3.7.95's nyag-bhava in the agent). Avisisto bhavaty anyaih svatantraih mukta-samsayaih: he becomes *undistinguished* from *other independent ones* whose *doubt is removed*. All the free ones are alike in being at one remove. The verse is the samuddesa's last word on the agent, and it is that freedom and efficacy are inversely related.

3.7.124 unformalized

निमित्तेभ्यः प्रवर्तन्ते सर्व एव स्वभूतये ।
अभिप्रायानुरोधोऽपि स्वार्थस्यैव प्रसिद्धये ॥ १२४ ॥

nimittebhyaḥ pravartante sarva eva svabhūtaye |
abhiprāyānurodho'pi svārthasyaiva prasiddhaye || 124 ||

All alike proceed from their occasioning grounds for their own becoming; and even the accommodation of another's intention is for the accomplishing of one's own purpose.

The samuddesa's motivational realism, and it is bracing. Nimittebhyah pravartante sarva eva sva-bhutaye: *all alike* proceed from their *occasioning grounds* *for their own becoming* — every participant acts for itself: the axe for its own cutting, the fire for its own burning, the servant for his own ends (3.1.25's universal deploying the causes for the manifestation of its own self). Abhipraya-anurodho 'pi sva-arthasya eva prasiddhaye: and even the *accommodation of another's intention* is for the *accomplishing of one's own purpose*. The soldier fights the king's war for his own reasons; the instrument serves the agent by doing what instruments do. So the hierarchy of 3.7.22 (the soldiers fight, the king conquers) is not a hierarchy of purposes: each has his own, and the subordination is grammatical rather than motivational. Every karaka is an agent (3.7.18) and every agent acts for itself — which is why the roles must be *assigned* rather than found (3.7.104's buddhy-avastha), since the world offers only a crowd of self-interested operations and no organizing centre.

3.7.125 unformalized

प्रेषणाध्येषणे कुर्वस्तत्समर्थानि चाचरन् ।
कर्तैंव विहितां शास्त्रे हेतुसंज्ञां प्रपद्यते ॥ १२५ ॥

preṣaṇādhyeṣaṇe kurvastatsamarthāni cācaran |
kartaiṃva vihitāṃ śāstre hetusaṃjñāṃ prapadyate || 125 ||

One who, doing the directing and the requesting, and performing what is capable of that — being an agent alone, obtains the designation 'hetu' taught in the science.

The hetu located within agency. Presana-adhyesane kurvan: one *doing* the *directing* and the *requesting* — presana is command, adhyesana is request: the two ways one gets another to act. Tat-samarthani ca acaran: and *performing what is capable of that* — doing whatever brings the other's action about. Karta eva vihitam sastre hetu-samjnam prapadyate: *being an agent alone*, he *obtains* the *designation 'hetu'* taught in the *science*. The point is karta eva: the hetu is not a separate karaka standing beside the agent; it is an *agent*, receiving a further designation when his agency consists in moving another agent (3.7.25's karta kartr-antara-apeksah). So the causative's structure is agency-over-agency (3.7.63's stacking), and hetu names the upper storey rather than a new kind of thing.

3.7.126 unformalized

द्रव्यमात्रस्य तु प्रैषे पृच्छयादेर्लोड् विधीयते ।
सक्रियस्य प्रयोगस्तु यदा स विषयो णिचः ॥ १२६ ॥

dravyamātrasya tu praiṣe pṛcchayāderloḍ vidhīyate |
sakriyasya prayogastu yadā sa viṣayo ṇicaḥ || 126 ||

But in the directing of a mere substance, the imperative of 'prcch' and the like is prescribed; the employment of one possessed of action, however — then that is the domain of the causative affix.

Two kinds of directing, morphologically distinguished. Dravya-matrasya tu praise: in the *directing of a mere substance* — commanding something that is not itself an agent — prcchy-ader lod vidhiyate: the *imperative* of 'prcch' (ask) and the like is prescribed: one uses the imperative directly. Sa-kriyasya prayogah tu yada sa visayo nicah: but the *employment* of *one possessed of action* — directing something that *does* act — *then* that is the *domain of the causative affix*. So the nic-affix is reserved for directing agents, and mere imperatives suffice for directing things. The distinction is exact and it depends on 3.7.18's doctrine: the causative presupposes a second agent, and where the directed party has no action of its own, there is no second agency for the affix to mark.

3.7.127 unformalized

गुणक्रियायां स्वातन्त्र्यात् प्रेषणे कर्मतां गतः ।
नियमात् कर्मसंज्ञायाः स्वधर्मेणाभिधीयते ॥ १२७ ॥

guṇakriyāyāṃ svātantryāt preṣaṇe karmatāṃ gataḥ |
niyamāt karmasaṃjñāyāḥ svadharmeṇābhidhīyate || 127 ||

From independence in the action that is a quality, one who has come to objecthood in the directing — from the restriction of the object-designation, is expressed by his own property.

The caused agent's double status. Guna-kriyayam svatantryat: from *independence* in the action that is a *quality* — the caused agent is svatantra in his own act (Devadatta genuinely cooks). Presane karmatam gatah: in the *directing*, he has *come to objecthood* — but relative to the director, he is the object ('he makes Devadatta cook'). Niyamat karma-samjnayah sva-dharmena abhidhiyate: from the *restriction* of the *object-designation*, he is expressed *by his own property*. So he is agent below and object above, and the grammar's rule fixes which designation the morphology records. This is 3.7.20-21 exactly (subordinated capacities furnished with their own complete operations; agency regained at the next level down), and the causative construction is where the two-level structure is morphologically visible.

3.7.128 unformalized

क्रियायाः प्रेरकं कर्म हेतुः कर्तुः प्रयोजकः ।
कर्मार्था च क्रियोत्पत्तिः संस्कारप्रतिपत्तिभिः ॥ १२८ ॥

kriyāyāḥ prerakaṃ karma hetuḥ kartuḥ prayojakaḥ |
karmārthā ca kriyotpattiḥ saṃskārapratipattibhiḥ || 128 ||

The object is the impeller of the action; the hetu is the occasioner of the agent; and the arising of the action is for the sake of the object, through refinements and apprehensions.

The three-term structure of purposive action. Kriyayah prerakam karma: the *object* is the *impeller* of the action — what one wants is what sets the act going (3.7.45's ipsitatama; 3.7.71's object as the action's occasioner). Hetuh kartuh prayojakah: the *hetu* is the *occasioner of the agent* — the reason moves the doer. Karma-artha ca kriya-utpattih samskara-pratipattibhih: and the *arising of the action* is *for the sake of the object*, through *refinements* and *apprehensions*. So the causal order runs: hetu moves agent, agent acts, action is for the object — and the object, being what is wanted, was the impeller all along. The circle is teleological and it is not vicious: the wanted end starts the process that reaches it, which is what purposive action *is*. Samskara-pratipattibhih — through refinements and apprehensions — names the machinery: the instruments are prepared (3.7.92) and the situation is grasped, and between them the action arises.

3.7.129 unformalized

अनिराकरणात् कर्त्तुस्त्यागाङ्गं कर्मणोप्सितम् ।
प्रेरणानुमतिभ्यां वा लभते सम्प्रदानताम् ॥ १२९ ॥

anirākaraṇāt karttustyāgāṅgaṃ karmaṇopsitam |
preraṇānumatibhyāṃ vā labhate sampradānatām || 129 ||

From the agent's not refusing, what is desired by the object is a limb of the giving; or, from impelling and from consenting, it obtains the state of being the recipient.

The sampradana — the recipient, the dative's karaka — defined by three alternatives, and all three are relational. Anirakaranat kartuh: from the *agent's not refusing* — the recipient is defined *negatively*: he is the one who does not decline. Tyaga-angam karmana ipsitam: what is *desired by the object* is a *limb of the giving*. Or: prerana-anumatibhyam va labhate sampradanatam: from *impelling* and from *consenting*, it obtains the state of being the *recipient* — he asked for it, or he agreed to it. So the recipient is constituted by not-refusing, or by asking, or by consenting: three attitudes, no action. The sampradana is the karaka that does nothing at all — it merely fails to object — and Panini's own definition (1.4.32: karmana yam abhipraiti sa sampradanam, whom the agent intends by the object) is equally attitudinal. This is 3.7.104's thesis at its cleanest: a karaka constituted wholly by a stance.

3.7.130 unformalized

हेतुत्वे कर्मसंज्ञायां शेषत्वे वापि कारकम् ।
रुच्यार्थादिषु शास्त्रेण सम्प्रदानाख्यमुच्यते ॥ १३० ॥

hetutve karmasaṃjñāyāṃ śeṣatve vāpi kārakam |
rucyārthādiṣu śāstreṇa sampradānākhyamucyate || 130 ||

In hetu-hood, in the object-designation, or even in remainder-hood, the karaka is called 'sampradana' by the science, in the case of 'ruc' and the rest.

The sampradana's assignment shown to override its competitors. In verbs like ruc (to please), the thing pleased could be construed as hetu (it is why the pleasing happens), as karman (it is what is affected), or as sesa (the residue-case: 3.7.44's sasthi sese). Hetutve karma-samjnayam sesatve va api karakam: in *hetu*-hood, in the *object*-designation, or even in *remainder*-hood — three available construals. Rucy-arthadisu sastrena sampradana-akhyam ucyate: the karaka is called *'sampradana'* *by the science* in the case of 'ruc' and the rest. Panini legislates (1.4.33: rucy-arthanam priyamanah): the pleased one takes the dative, whatever else he might have been. So the sampradana-assignment here is a *stipulation* — the science choosing among available construals and fixing one. This is 3.7.103's finding (the restriction is in the word, not in the thing) exhibited: the grammar decides, and there was nothing to discover.

3.7.131 unformalized

भेदस्य च विवक्षायां पूर्वा पूर्वा क्रियां प्रति ।
परस्याङ्गस्य कर्मत्वान्न क्रियाग्रहणं कृतम् ॥ १३१ ॥

bhedasya ca vivakṣāyāṃ pūrvā pūrvā kriyāṃ prati |
parasyāṅgasya karmatvānna kriyāgrahaṇaṃ kṛtam || 131 ||

And when a difference is intended, each prior one with respect to the action — since the subsequent limb is an object, the mention of 'action' has not been made.

Why Panini's rule omits a word. Bhedasya ca vivaksayam purva purva kriyam prati: when a *difference is intended*, *each prior one* with respect to the action — in a chain of sub-actions (3.7.16-17's seeing, seeking, resolving, undertaking), each earlier is means to the later. Parasya angasya karmatvat na kriya-grahanam krtam: since the *subsequent limb* is an *object*, the mention of 'action' has *not been made*. The rule does not need to say 'action' because the later stage is already an object, and the object-rule covers it. This is grammatical economy (2.343's laghava) explained by semantic structure: the sutra is short because the chain of 3.7.16-17 makes each stage an object of the last, and the karman-provisions already reach them. The verse shows the samuddesa's doctrine paying for itself in Panini's brevity.

3.7.132 unformalized

क्रियाणां समुदाये तु यदैकत्वं विवक्षितम् ।
तदा कर्म क्रियायोगात् स्वाख्ययैवोपचर्यते ॥ १३२ ॥

kriyāṇāṃ samudāye tu yadaikatvaṃ vivakṣitam |
tadā karma kriyāyogāt svākhyayaivopacaryate || 132 ||

But when oneness is intended in an aggregate of actions — then the object, from its connection with the action, is figuratively spoken of by its own designation.

The complement of 3.7.131. Kriyanam samudaye tu yada ekatvam vivaksitam: when *oneness* is intended in an *aggregate of actions* — the chain taken as one act rather than as stages (2.448's uncounted sub-actions displayed by a single verb). Tada karma kriya-yogat sva-akhyaya eva upacaryate: *then* the object, from its *connection with the action*, is *figuratively* spoken of by its *own designation*. Upacaryate — figuratively (3.3.39's aupacariki satta; 2.253's upacara): where the chain is taken as one, the intermediate objects lose their objecthood and the designation is transferred. So whether the stages have objects depends on whether the chain is taken as many acts or one — which is vivaksa (3.7.2's kvacit kimcit vivaksyate), and the pair 3.7.131-132 shows both takings licensed and both grammatical.

3.7.133 unformalized

भेदाभेदविवक्षा च स्वभावेन व्यवस्थिता ।
तस्माद्गत्यर्थकर्मत्वे व्यभिचारो न दृश्यते ॥ १३३ ॥

bhedābhedavivakṣā ca svabhāvena vyavasthitā |
tasmādgatyarthakarmatve vyabhicāro na dṛśyate || 133 ||

And the intending of difference and non-difference stands settled by its own nature; therefore, in the objecthood of verbs of motion, no deviation is seen.

The stability underneath the relativity. Bheda-abheda-vivaksa ca svabhavena vyavasthita: the *intending of difference and non-difference* stands settled *by its own nature* — the two takings of 3.7.131-132 are not arbitrary; each has its own settled character, and speakers do not confuse them. Tasmat gaty-artha-karmatve vyabhicaro na drsyate: *therefore*, in the *objecthood* of verbs of *motion*, no *deviation* is seen. The gati-verbs ('goes to the village') take their objects invariably, without wobble — because though the assignment is a taking, the taking is stable. This is the samuddesa's version of 3.6.22 (relative to the limit, no inversion) and 3.7.11 (the capacities are of fixed apprehension): constructivism without arbitrariness. Vyabhicara — deviation — is the logician's word for a mark that strays, and its absence here is what makes the grammar statable at all.

3.7.134 unformalized

विकल्पेनैव सर्वत्र संज्ञे स्यातामुभे यदि ।
आरंभेण न योगस्य प्रत्याख्यानं समं भवते ॥ १३४ ॥

vikalpenaiva sarvatra saṃjñe syātāmubhe yadi |
āraṃbheṇa na yogasya pratyākhyānaṃ samaṃ bhavate || 134 ||

If both designations were to obtain everywhere by option — then, by the undertaking, the rejection of the connection would not be equal.

An argument against free option in karaka-assignment. Vikalpena eva sarvatra samjne syatam ubhe yadi: if *both* designations were to obtain *everywhere* by *option* — if the grammar simply let one choose between karman and sampradana, or between hetu and karana, wherever both fit. Arambhena na yogasya pratyakhyanam samam bhavet: then, by the *undertaking*, the *rejection of the connection* would not be *equal*. The reasoning is compressed: if option ran everywhere, the rules that *reject* certain connections (that bar a construal) would be pointless, since option would restore what they barred. Panini's system contains such rejections; therefore option does not run everywhere. This is the argument from the science's economy that 2.443 ran (Panini does not legislate vacuously) — and it protects 3.7.133's stability: the assignments are takings, and the takings are constrained, or the sastra would not need its prohibitions.

3.7.135 unformalized

त्यागरूपं प्रहातव्ये प्राप्ये संसर्गदर्शनम् ।
आस्थितं कर्म यत् तत्र द्वैरूप्यं भजते क्रिया ॥ १३५ ॥

tyāgarūpaṃ prahātavye prāpye saṃsargadarśanam |
āsthitaṃ karma yat tatra dvairūpyaṃ bhajate kriyā || 135 ||

In what is to be abandoned there is the form of relinquishing; in what is to be attained, the seeing of conjunction; where the object is settled there, the action takes on a two-fold form.

The two directions an action can face. Tyaga-rupam prahatavye: in *what is to be abandoned* there is the *form of relinquishing* — the action faces away, and its object is a source. Prapye samsarga-darsanam: in *what is to be attained*, the *seeing of conjunction* — the action faces toward, and its object is a goal. Asthitam karma yat tatra dvairupyam bhajate kriya: where the *object* is *settled* there, the *action* takes on a *two-fold form*. One action, two faces: every movement is a leaving and an arriving, and which the sentence records decides whether the noun takes the ablative or the accusative. The verse introduces the apadana — the ablative's karaka — by locating it in the same action that yields the accusative, differently faced. Bhajate: takes on, wears (3.2.8's tattvam eva avikalpitam vikalpa-rupam bhajate).

3.7.136 unformalized

निर्दिष्टविषयं किंचिदुपात्तविषयं तथा ।
अपेक्षितक्रियं चेति त्रिधापादानमुच्यते ॥ १३६ ॥

nirdiṣṭaviṣayaṃ kiṃcidupāttaviṣayaṃ tathā |
apekṣitakriyaṃ ceti tridhāpādānamucyate || 136 ||

Something whose domain is specified; likewise something whose domain is taken up; and something whose action is expected — thus the ablative is said to be three-fold.

The apadana's taxonomy, and it is Panini's. Nirdista-visayam kimcit: something whose *domain is specified* — the source explicitly named ('he comes from the village'). Upatta-visayam tatha: likewise something whose *domain is taken up* — the source supplied by the construction ('he fears the thief': the thief is the source of the fear, unnamed as a source). Apeksita-kriyam ca iti: and something whose *action is expected* — the source of an action not stated but required ('he abstains from wine': the drinking is expected and absent). Tridha apadanam ucyate: thus the *ablative* is said to be *three-fold*. The three cover the ablative's whole range — physical source, emotional source, abstained-from action — and they are ordered by how much the construction supplies: named, taken up, expected. This is 3.7.78's method (one karaka expounded by way of difference) applied to the case that most obviously needs it.

3.7.137 unformalized

संयोगिभेदादेभिन्नात्मा गमिरेव भ्रमिर्यथा ।
ध्रुवावधिरपायोऽपि समवेतस्तथाध्रुवे ॥ १३७ ॥

saṃyogibhedādebhinnātmā gamireva bhramiryathā |
dhruvāvadhirapāyo'pi samavetastathādhruve || 137 ||

The self of 'gami' is differentiated by difference of what is conjoined — just as 'bhrami' is; the separation too, having a fixed limit, is inherent likewise in the unfixed.

The ablative's limit-condition examined. Samyogi-bhedad bhinnatma gamir eva bhramir yatha: the self of *'gami'* (going) is *differentiated* by difference of *what is conjoined* — as *'bhrami'* (whirling) is: the going is one action, and it differs according to what it goes from and to, as whirling differs by what whirls. Dhruva-avadhih apayo 'pi: the *separation* too has a *fixed limit* — Panini's rule (1.4.24: dhruvam apaye 'padanam: the fixed point in a separation is the apadana). Samavetah tatha adhruve: it is *inherent* likewise in the *unfixed*. So the ablative's 'fixed point' is found even where nothing is fixed — the departure-point of two moving things, the source that is itself in motion. The rule says dhruva, fixed, and the case applies to the unfixed: which means dhruva names a role in the construal (what is taken as the reference-point) and not a property of anything (3.6.2's avadhi; 3.7.103's restriction in the word, not in the thing).

3.7.138 unformalized

द्रव्यत्वभावो न ध्नौव्यमिति सूत्रे प्रतीयते ।
अपायविषयं घ्रौव्यं यत्तु तावद्विवक्षितम् ॥ १३८ ॥

dravyatvabhāvo na dhnauvyamiti sūtre pratīyate |
apāyaviṣayaṃ ghrauvyaṃ yattu tāvadvivakṣitam || 138 ||

'The state of being a substance is not fixity' — this is understood in the sutra; but the fixity whose domain is the separation is what is intended.

The rule's key word disambiguated. Dravyatva-bhavo na dhrauvyam iti sutre pratiyate: '*the state of being a substance is not fixity*' — this is understood in the *sutra*: dhruva does not mean 'is a substance' or 'is immobile'. Apaya-visayam dhrauvyam yat tu tavad vivaksitam: but the *fixity* whose *domain is the separation* is what is *intended* — dhruva means: fixed *for the purposes of this separation*, taken as the point from which. So the ablative's 'fixed point' is fixed relative to the departure being described, and nothing more. The verse is precise grammatical exegesis and it delivers the samuddesa's thesis exactly: what looked like a metaphysical requirement in Panini's rule (the source must be *fixed*) turns out to be a construal-relative role. 3.7.139 gives the examples.

3.7.139 unformalized

सरणे देवदत्तस्य ध्रौव्यं पाते तु वाजिनः ।
आविष्टं यदपायेन तस्याध्रौव्यं प्रचक्षते ॥ १३९ ॥

saraṇe devadattasya dhrauvyaṃ pāte tu vājinaḥ |
āviṣṭaṃ yadapāyena tasyādhrauvyaṃ pracakṣate || 139 ||

In Devadatta's running there is fixity; but in the horse's falling, of the horse. That which is pervaded by the separation — of that they declare the non-fixity.

The relativity of 3.7.138 exhibited. Sarane devadattasya dhrauvyam: in *Devadatta's running* there is *fixity* — 'the horse falls from running Devadatta': Devadatta is running, and he is the dhruva, the fixed point of the falling. Pate tu vajinah: but in the *horse's falling*, of the *horse* — from the horse's side, the horse is what departs. Avistam yad apayena tasya adhrauvyam pracaksate: *that which is pervaded by the separation* — of *that* they declare the *non-fixity*. So the runner is 'fixed' while running and the faller is 'unfixed' while stationary-relative-to-him: fixity is entirely a matter of which party the separation is predicated of. The example is chosen for its absurdity-as-physics and its exactness-as-grammar: a running man is the fixed point, and Sanskrit says so, and the rule is dhruvam apaye 'padanam. Nothing could show more plainly that the karakas are construals (3.7.91's sthalya pacyate, in the ablative).

3.7.140 unformalized

उभावप्यध्रूवौ मेषौ यद्यप्युभयकर्मजे ।
विभागे प्रविभक्ते तु क्रिये तत्र विवक्षिते ॥ १४० ॥

ubhāvapyadhrūvau meṣau yadyapyubhayakarmaje |
vibhāge pravibhakte tu kriye tatra vivakṣite || 140 ||

Both rams are unfixed, even though the separation is produced by the action of both; but when the division is divided out, the two actions are intended there.

The limit-case: two things parting from each other. Ubhau api adhruvau mesau: *both rams* are *unfixed* — two rams butting and springing apart: neither is stationary, and each departs from the other. Yady api ubhaya-karma-je vibhage: even though the *separation* is *produced by the action of both* — the parting is jointly caused; there is no asymmetry in the event whatever. Pravibhakte tu kriye tatra vivaksite: but when the *division is divided out*, the *two actions* are *intended there*. The speaker splits the one parting into two departures, and each ram becomes the other's dhruva: 'this one springs from that', 'that one springs from this'. Two sentences, two ablatives, one symmetrical event. This is 3.7.19's parents (both agents, roles by intention) at the ablative, and it closes the apadana's analysis where the samuddesa closes everything: the world offers a symmetry, the grammar requires an asymmetry, and vivaksa supplies it.

3.7.141-142 unformalized

मेषान्तरक्रियापेक्षमवधित्वं पृथक् पृथक् ।
मेषयोः स्वक्रियापेक्षं कर्तृत्वं च पृथक् पृथक् ॥ १४१ ॥
अभेदेन क्रियैका तु द्विसाध्या चेद्विवक्षिता ।
मेषावपाये कर्तारौ यद्यन्यो विद्यतेऽवधिः ॥ १४२ ॥

meṣāntarakriyāpekṣamavadhitvaṃ pṛthak pṛthak |
meṣayoḥ svakriyāpekṣaṃ kartṛtvaṃ ca pṛthak pṛthak || 141 ||
abhedena kriyaikā tu dvisādhyā cedvivakṣitā |
meṣāvapāye kartārau yadyanyo vidyate'vadhiḥ || 142 ||

(141) Limit-hood, depending on the action of the other ram, belongs to the two rams severally; and agency, depending on their own action, belongs to them severally. (142) But if one action, to be accomplished by two, is intended by way of non-difference — the two rams are agents in the separation, if some other limit exists.

The rams of 3.7.140 analysed at both takings. (141) gives the divided taking: mesa-antara-kriya-apeksam avadhitvam prthak prthak — *limit-hood*, depending on the *other* ram's action, belongs to them *severally*: each is the other's dhruva, in his own sentence. Sva-kriya-apeksam kartrtvam ca prthak prthak: and *agency*, depending on their *own* action, *severally*. Two sentences, each ram agent in one and limit in the other. (142) gives the undivided taking: abhedena kriya eka tu dvi-sadhya ced vivaksita — if *one* action, *to be accomplished by two*, is intended *by way of non-difference*: one parting, jointly done. Then: mesau apaye kartarau yady anyo vidyate avadhih — the two rams are *agents* in the separation, *if some other limit exists*. Both agents, and the ablative must go elsewhere — to a third thing, or nowhere. So the same event yields two grammars: divided, each is the other's source; undivided, both are agents and the source-slot is vacant. Vivaksa decides, and the case-frame follows.

3.7.143 unformalized

गतिर्विना त्ववधिना नापाय इति गम्यते ।
वृक्षस्य पर्णं पततीत्येवं भाष्ये निदर्शितम् ॥ १४३ ॥

gatirvinā tvavadhinā nāpāya iti gamyate |
vṛkṣasya parṇaṃ patatītyevaṃ bhāṣye nidarśitam || 143 ||

But without a limit, motion is not understood as a separation; 'the leaf falls from the tree' — thus it is exhibited in the Bhasya.

The apadana's necessary condition, cited from Patanjali. Gatir vina tu avadhina na apayah iti gamyate: *without a limit*, *motion* is not understood as a *separation* — a falling is just a falling until something is fallen *from*; apaya requires an avadhi, or it is not apaya. And the citation: vrksasya parnam patati iti evam bhasye nidarsitam — '*the leaf falls from the tree*' — thus it is *exhibited in the Bhasya*. Patanjali's own example, and the genitive in it (vrksasya) rather than the expected ablative is the puzzle the Mahabhasya discusses. Bhartrhari cites rather than argues: the point that separation needs a limit is the tradition's, and the example is the tradition's, and 3.7.140-142's analysis of the rams was an extension of it. The samuddesa's method throughout — 2.475's bhasye eva upavarnitah, 'described in the Bhasya itself' — is to find its doctrines already in the gurus.

3.7.144 unformalized

भेदाभेदौ पृथग्भावः स्थितिश्चेति विरोधिनः ।
युगपन्न विवक्ष्वन्ते सर्वे धर्मा वलाहके ॥ १४४ ॥

bhedābhedau pṛthagbhāvaḥ sthitiśceti virodhinaḥ |
yugapanna vivakṣvante sarve dharmā valāhake || 144 ||

Difference and non-difference, separateness and standing — these being contradictory, all the properties are not intended simultaneously in a cloud.

Why one situation cannot bear all its construals at once. Bheda-abhedau prthag-bhavah sthitih ca iti virodhinah: *difference* and *non-difference*, *separateness* and *standing* — these are *contradictory*. Yugapan na vivaksyante sarve dharmah valahake: *all* the properties are *not intended simultaneously* in a *cloud*. The cloud is the example: it is one and many, gathering and dispersing, moving and standing — every contradictory pair applies to it, and no single utterance can intend them all. So vivaksa selects, and must: the properties are all there (3.7.2's sarvada sarvatha bhavat: existing always and in every way), and speech takes one set at a time. This is 2.251's mechanism (transcending simultaneity, the word's plenitude stands settled in succession) applied to the world rather than the word — and it explains why every karaka-assignment excludes the others it could have made.

3.7.145 unformalized

धनुषा विध्यतीत्यत्र विनापायविवक्षया ।
करणत्वं यतो नास्ति तस्मात्तदुभयं सह ॥ १४५ ॥

dhanuṣā vidhyatītyatra vināpāyavivakṣayā |
karaṇatvaṃ yato nāsti tasmāttadubhayaṃ saha || 145 ||

In 'he pierces with a bow', without the intention of separation, since instrument-hood does not exist — therefore those two are together.

A case where two karaka-construals must coexist. Dhanusa vidhyati iti atra: in '*he pierces with a bow*' — the bow is instrumental. Vina apaya-vivaksaya karanatvam yatah na asti: *without* the *intention of separation*, *instrument-hood* does *not exist* — the bow is an instrument only because the arrow *leaves* it: take away the departure and the bow does nothing. So the instrumental presupposes an ablative sense that the case does not express. Tasmat tad ubhayam saha: *therefore* those *two are together* — instrument-hood and separation-hood coexist in the one bow, in the one case-ending. This is a genuine exception to 3.7.144's rule that contradictory construals cannot be simultaneous: here two construals are *both* required, because one is the condition of the other. The verse is careful evidence that the karaka-system's exclusivity has limits.

3.7.146 unformalized

एकैव सा सती शक्तिर्द्विरूपा व्यवतिष्ठते ।
निमित्तं संज्ञयोस्तत्र परया बाध्यतेऽपरा ॥ १४६ ॥

ekaiva sā satī śaktirdvirūpā vyavatiṣṭhate |
nimittaṃ saṃjñayostatra parayā bādhyate'parā || 146 ||

That one and the same existing capacity stands two-formed; being the occasioning ground for the two designations there, the later annuls the earlier.

The resolution of 3.7.145. Eka eva sa sati saktih dvi-rupa vyavatisthate: that *one and the same* existing capacity stands *two-formed* — the bow's single capacity wears both faces (instrument and source), and there are not two powers in it. Nimittam samjnayoh tatra: it is the *occasioning ground* for the *two designations* *there* — one capacity, two available names. Paraya badhyate 'para: the *later* *annuls* the *earlier*. The badha-machinery adjudicates (2.66's annulment; 2.73's hierarchy; 3.7.97's kartrtvam badhakam): where two designations compete for one capacity, the rule ordering decides, and the instrumental wins in 'dhanusa vidhyati'. So the coexistence of 3.7.145 is real at the level of capacity and resolved at the level of designation: the bow is both, and the grammar says one. This is the samuddesa's settled shape — plenitude below, selection above, and the selection made by rule.

3.7.147 unformalized

निर्धारणे विभक्ते यो भीत्रादीना च यो विधिः ।
उपात्तपेक्षितापायः सोऽबुधप्रतिपत्तये ॥ १४७ ॥

nirdhāraṇe vibhakte yo bhītrādīnā ca yo vidhiḥ |
upāttapekṣitāpāyaḥ so'budhapratipattaye || 147 ||

The rule for the ablative in determination, and that for 'bhi', 'tra' and the rest, has its separation taken up or expected — that is for the comprehension of the unlearned.

The apadana's extended cases placed as pedagogy. Nirdharane vibhakte yah: the rule for the ablative in *determination* — 'the best of cows' (nirdharana: singling out from a class, which takes the ablative or genitive). Bhi-tradinam ca yo vidhih: and that for '*bhi*' (fear) and '*tra*' (protect) and the rest — 'he fears the thief', where the thief is grammatically a source (3.7.136's upatta-visaya). Upatta-apeksita-apayah: its *separation* is *taken up* or *expected* (3.7.136's second and third kinds). Sah abudha-pratipattaye: *that* is *for the comprehension of the unlearned*. The extended ablative-rules are teaching-devices: they let the learner see that fearing has a source-structure, that singling-out has a departure-structure. The separations are constructed for instruction (2.13's divisions for the avidvan; 3.7.78's one apadana exhibited by way of difference). The samuddesa marks its own taxonomy as ladder.

3.7.148 unformalized

कर्तृकर्मव्यवहितामसाक्षाद्धारयत् क्रियाम् ।
उपकुर्वत् क्रियासिद्धौ शास्त्रेऽधिकरणं स्मृतम् ॥ १४८ ॥

kartṛkarmavyavahitāmasākṣāddhārayat kriyām |
upakurvat kriyāsiddhau śāstre'dhikaraṇaṃ smṛtam || 148 ||

Bearing, not directly, an action that is separated by agent and object; assisting in the accomplishing of the action — that is remembered in the science as the locus.

The adhikarana — the locus, the locative's karaka — defined, and the definition is by remoteness. Kartr-karma-vyavahitam asaksat dharayat kriyam: *bearing*, *not directly*, an action that is *separated* by *agent and object* — the locus holds the action at two removes: the agent and object stand between. Asaksat: not directly (3.7.94's arat, the agent assisting only from a distance — and the locus is further still). Upakurvat kriya-siddhau: *assisting* in the *accomplishing* of the action. Sastre adhikaranam smrtam: that is *remembered in the science* as the *locus*. So the adhikarana is the most remote karaka: it supports the whole configuration without touching any part of it. The mat is where Devadatta sits, and the mat does nothing to the sitting except be under it. Bharing-at-a-remove is the whole of its contribution — and it is a contribution (upakurvat), which is why it is a karaka at all.

3.7.149 unformalized

उपश्लेषस्य चाभेदस्तिलाकाशकटादिषु ।
उपकारास्तु भिद्यन्ते संयोगिसमवायिनाम् ॥ १४९ ॥

upaśleṣasya cābhedastilākāśakaṭādiṣu |
upakārāstu bhidyante saṃyogisamavāyinām || 149 ||

And the non-difference of contact is in sesame, ether, mat and the like; but the assistances of the conjoined and the inherent are differentiated.

The locus's kinds. Upaslesasya ca abhedah tila-akasa-katadisu: the *non-difference of contact* is in *sesame*, *ether*, *mat* and the like — three loci: the oil in the sesame (pervading), the thing in space (containing), the sitter on the mat (supporting). In all three the contact is undifferentiated: 'in' is 'in'. Upakarah tu bhidyante samyogi-samavayinam: but the *assistances* of the *conjoined* and the *inherent* are *differentiated*. The oil's being in the seed is inherence; the man's being on the mat is conjunction; and though the locative is the same, what the locus *does* differs. So the case-ending flattens a real distinction — one locative for pervasion, containment and support — and the karaka-analysis recovers it. This is 3.1.17-18's distinction (cognitions running along among the conjoined, difference withdrawing among the inherent) applied to the locus.

3.7.150 unformalized

अविनाशो गुरुत्वस्य प्रतिबन्धे स्वतन्त्रता ।
दिग्विशेषादवच्छेद इत्याद्या भेदहेतवः ॥ १५० ॥

avināśo gurutvasya pratibandhe svatantratā |
digviśeṣādavaccheda ityādyā bhedahetavaḥ || 150 ||

The non-destruction of weight; independence in obstruction; delimitation by a particular direction — such are the causes of difference.

Three marks by which loci differ. Avinasah gurutvasya: the *non-destruction of weight* — the mat bears the man, and his weight is not annulled: the locus supports without absorbing. Pratibandhe svatantrata: *independence in obstruction* — the locus can resist, and its resisting is its own (3.7.29's wall, whose capacity is in obstructing). Dig-visesad avacchedah: *delimitation by a particular direction* — the locus locates by direction (3.6.13's bhaga-prakalpana-sakti, the first capacity). Ity-adyah bheda-hetavah: such are the *causes of difference*. Three ways loci vary: how they bear weight, how they resist, how they orient. Each is a mode of the assisting-at-a-remove that 3.7.148 defined, and together they explain why the single locative case covers so heterogeneous a family.

3.7.151 unformalized

आकाशमेव केषांचिद्देशभेदप्रकल्पनात् ।
आधारशक्तिः प्रथमा सर्वसंयोगिनां मता ॥ १५१ ॥

ākāśameva keṣāṃciddeśabhedaprakalpanāt |
ādhāraśaktiḥ prathamā sarvasaṃyogināṃ matā || 151 ||

For some, ether alone, from its construction of a difference of places, is held to be the first locus-capacity of all conjoined things.

The locus reduced to ether. Akasam eva kesamcid: for *some*, *ether alone* — desa-bheda-prakalpanat: from its *construction of a difference of places* (3.1.15's places constructed by the properties of what is conjoined; 3.6.5's ether-portions serving as direction's adjunct). Adhara-saktih prathama sarva-samyoginam mata: is held to be the *first* *locus-capacity* of *all conjoined* things. Every locus is ether, ultimately: the mat is where the man is because the mat is somewhere, and being-somewhere is ether's construction. Prathama — first: the same word 3.6.13 gave direction (the *first* capacity, of constructing parts). So there are two claimants for primacy — direction constructing parts, ether constructing places — and they are the same construction differently named. The verse reports the reduction as one school's (kesamcid) and does not adopt it.

3.7.152 unformalized

इदमत्रेति भावानामभावान्न प्रकल्पते ।
व्यपदेशस्तमाकाशनिमित्तं संप्रचक्षते ॥ १५२ ॥

idamatreti bhāvānāmabhāvānna prakalpate |
vyapadeśastamākāśanimittaṃ saṃpracakṣate || 152 ||

'This is here' — the designation is not constructed, from the non-existence of things; they call that the occasioning ground which is ether.

The argument for 3.7.151's reduction. Idam atra iti: '*this is here*' — the plainest locative there is. Bhavanam abhavat na prakalpate vyapadesah: the *designation* is *not constructed*, from the *non-existence of things* — if there were no things, there would be no 'here': the locative needs a located thing and a locating one, and both are constructions (3.6.14's things without place; 3.7.104's buddhy-avastha). Tam akasa-nimittam sampracaksate: they call *that* — the ground of the 'here' — *the occasioning ground which is ether*. So what makes 'this is here' sayable is akasa, and akasa is what has no differentiation of its own (3.1.15). The locative rests on the undifferentiated, differentiated by what it locates — which is 3.1.16's structure (places belonging to what is not a place, conferred by relata) reaching the grammar's most basic spatial expression.

3.7.153 unformalized

कालात् क्रिया विभज्यन्ते आकाशात् सर्वमूर्तयः ।
एतावांश्चैव भेदोऽयमभेदोपनिबन्धनः ॥ १५३ ॥

kālāt kriyā vibhajyante ākāśāt sarvamūrtayaḥ |
etāvāṃścaiva bhedo'yamabhedopanibandhanaḥ || 153 ||

Actions are divided by time; all corporeal things by ether. And just this much is that difference, which rests upon non-difference.

The two great dividers, and their common ground. Kalat kriyah vibhajyante: *actions* are *divided by time* — one cooking from another by when (3.6.4's krama-rupe tu kalatah). Akasat sarva-murtayah: *all corporeal things* by *ether* — one pot from another by where. Two principles of individuation, and between them they carve everything: events by time, bodies by space. And then the deflation: etavan ca eva bhedah ayam abheda-upanibandhanah — *just this much* is that *difference*, which *rests upon non-difference*. Both dividers divide what is undivided (time and ether being themselves partless: 3.1.15, 3.2.8), and so all difference whatever — every event distinguished from every other, every thing from every other — rests on abheda. Etavan ca eva: just this much, and no more. The whole world's plurality is space and time carving the uncarved, and they are constructions.

3.7.154 unformalized

यद्यप्युपवसिर्देशविशेषमनुरुध्यते ।
शब्दप्रवृत्तिधर्मात्तु कालमेवावलंबते ॥ १५४ ॥

yadyapyupavasirdeśaviśeṣamanurudhyate |
śabdapravṛttidharmāttu kālamevāvalaṃbate || 154 ||

Although 'upavas' conforms to a particular place — from the property of the word's operation, it takes hold of time alone.

A case where the karaka follows the word rather than the situation. Yady api upavasir desa-visesam anurudhyate: although *'upavas'* (to fast, to dwell near) *conforms to a particular place* — the fast is kept somewhere; a place is unmistakably involved. Sabda-pravrtti-dharmat tu kalam eva avalambate: from the *property of the word's operation*, it *takes hold of time alone* — the verb's own behaviour makes its locus temporal ('he fasts for three nights'), not spatial, whatever the situation contains. So the word's usage overrides the situation's structure in fixing the karaka: the fast has a place, and 'upavas' will not take it as its adhikarana. This is 3.7.103's finding at its sharpest (the restriction is in the word, not in the thing) — and 3.7.66's two schools meet here, with the sabda-party winning: the action is where the words put it.

3.7.155 unformalized

वसतावप्रयुक्तेऽपि देशोऽधिकरणं ततः ।
अप्रयुक्तं त्रिरात्रादि कर्मं चोपवसौ स्मृतम् ॥ १५५ ॥

vasatāvaprayukte'pi deśo'dhikaraṇaṃ tataḥ |
aprayuktaṃ trirātrādi karmaṃ copavasau smṛtam || 155 ||

Even where 'vas' is not employed, place is therefore the locus; and 'three nights' and the rest, though unemployed, is remembered as the object in 'upavas'.

The consequence of 3.7.154, and it is a double displacement. Vasatau aprayukte 'pi desah adhikaranam tatah: even where *'vas'* (to dwell) is *not employed*, *place* is *therefore* the locus — the dwelling-verb is absent and its locus survives, supplied to the construction. Aprayuktam tri-ratradi karma ca upavasau smrtam: and *'three nights'* and the rest, *though unemployed*, is remembered as the *object* in *'upavas'* — the time-phrase is the object of an unstated verb. So 'he fasts three nights' has: a place-locus from an unuttered 'dwell', and a time-object from an unuttered action, and neither verb is there. The karaka-frame outlives the verbs that would license it, supplied by the buddhi that holds the situation (3.7.3). Aprayukta — unemployed, twice: the samuddesa ends its analysis of the locus by showing the case-frame running on words nobody said.

3.7.156 unformalized

संबन्धः कारकेभ्योऽन्यः क्रियाकारकपूर्वकः ।
श्रुतायामश्रुतायां वा क्रियायां सोऽभिधीयते ॥ १५६ ॥

saṃbandhaḥ kārakebhyo'nyaḥ kriyākārakapūrvakaḥ |
śrutāyāmaśrutāyāṃ vā kriyāyāṃ so'bhidhīyate || 156 ||

The relation is other than the karakas, and is preceded by action and karaka; whether the action is heard or unheard, it is expressed.

The genitive placed outside the karaka-system entirely. Sambandhah karakebhyo 'nyah: the *relation* is *other than* the karakas — the sixth case is not a karaka at all; it is the sesa, the remainder (3.7.44's sasthi sese). And its genesis: kriya-karaka-purvakah — it is *preceded by action and karaka*: a relation is what a completed karaka-configuration leaves behind. The king's man is the king's because some appointing happened. Srutayam asrutayam va kriyayam so 'bhidhiyate: whether the action is *heard* or *unheard*, it is *expressed* — the genitive stands whether or not its generating action is stated, and usually it is not (2.197's action that, having generated the connection, withdraws). So the genitive is the case of relations that outlive their actions, and its exclusion from the karakas is principled: karakas are participants *in* an action, and the genitive marks what is left when the action has gone.

3.7.157 unformalized

द्विष्ठोऽप्यसौ परार्थत्वाद् गुणेषु व्यतिरिच्यते ।
तत्राभिधीयमानः सन् प्रधानेऽप्युपभुज्यते ॥ १५७ ॥

dviṣṭho'pyasau parārthatvād guṇeṣu vyatiricyate |
tatrābhidhīyamānaḥ san pradhāne'pyupabhujyate || 157 ||

Though resting in two, it is set apart among the subordinates because of its being for another; being expressed there, it is enjoyed in the principal as well.

The genitive's peculiar double life. Dvistho 'py asau: though *resting in two* — a relation needs two terms; 'the king's man' holds both. Para-arthatvad gunesu vyatiricyate: it is *set apart* among the *subordinates* because of its *being for another* — the relation is parartha (2.113's term), serving something beyond itself, so it sits among the subordinate elements. Tatra abhidhiyamanah san pradhane 'py upabhujyate: being *expressed there*, it is *enjoyed in the principal* as well. Upabhujyate — enjoyed, consumed, made use of: the relation, though expressed in the subordinate term, does its work for the principal. So the genitive is expressed at one end and works at the other, which is what it is to be a relation — and 2.395's joint-property model (each owner having the whole, undiminished) is the right picture for how it can be wholly in both.

3.7.158 unformalized

निमित्तनियमः शब्दात् संबन्धस्य न गृह्यते ।
कर्मप्रवचनीयैस्तु स विशेषेऽवरुध्यते ॥ १५८ ॥

nimittaniyamaḥ śabdāt saṃbandhasya na gṛhyate |
karmapravacanīyaistu sa viśeṣe'varudhyate || 158 ||

A restriction of the occasioning ground for the relation is not grasped from the word; but by the karmapravacaniyas, it is confined to a particular.

The genitive's vagueness and its remedy. Nimitta-niyamah sabdat sambandhasya na grhyate: a *restriction of the occasioning ground* for the *relation* is *not grasped from the word* — 'the king's man' does not say *how* he is the king's: servant, son, subject, sculptor? The genitive marks relation and specifies none. Karma-pravacaniyaih tu sa visese 'varudhyate: but by the *karmapravacaniyas*, it is *confined to a particular* — the class of words (anu, prati, adhi) that Panini defines as speaking *about* the action, and whose office 2.204 identified exactly: sambandhasya bhedakah, *differentiators of the relation*. So the genitive opens a relation and the karmapravacaniya closes it. This is 2.199-204's analysis reappearing in the Padakanda, and the two treatments agree precisely: a relation left standing by a withdrawn action, and a particle that pins which relation it is.

3.7.159 unformalized

साधनैर्व्यपदिष्टे च श्रूयमाणक्रिये पुनः ।
प्रोक्ता प्रतिपदं षष्ठी समासस्य निवृत्तये ॥ १५९ ॥

sādhanairvyapadiṣṭe ca śrūyamāṇakriye punaḥ |
proktā pratipadaṃ ṣaṣṭhī samāsasya nivṛttaye || 159 ||

And when a thing is designated by the means, and the action is being heard again — the sixth case is stated word by word, for the exclusion of the compound.

Why the genitive is prescribed where a karaka would serve. Sadhanaih vyapadiste ca sruyamana-kriye punah: when a thing is *designated by the means*, and the *action is being heard* again — where the verb is present and the karaka-relation is live. Prokta pratipadam sasthi samasasya nivrttaye: the *sixth case* is stated *word by word*, *for the exclusion of the compound*. The genitive is prescribed individually so as to block compounding — keep 'rajnah purusah' from collapsing into 'rajapurusah', because the analysis must stay visible. This is 2.198 verbatim, and its recurrence here is not repetition but placement: the Vakyakanda gave the rule's motive, and the Padakanda now shows it operating inside the karaka-system. The genitive is posted where the science needs the relation *displayed* — an operational reason, not a semantic one (3.7.103's niyamah sabde na tu vastuni).

3.7.160 unformalized

निष्ठायां कर्मविषया षष्ठी न प्रतिषिध्यते ।
शेषलक्षणया षष्ठ्या समासस्तत्र नेष्यते ॥ १६० ॥

niṣṭhāyāṃ karmaviṣayā ṣaṣṭhī na pratiṣidhyate |
śeṣalakṣaṇayā ṣaṣṭhyā samāsastatra neṣyate || 160 ||

In the nistha-form, the sixth case whose domain is the object is not prohibited; the compound with a sixth case characterized as remainder is not desired there.

A technical case, and it discriminates two genitives. Nisthayam karma-visaya sasthi na pratisidhyate: in the *nistha*-form (the past participle in -ta) the *sixth case* whose *domain is the object* is *not prohibited* — 'krtam asya', 'done by him', where the genitive marks the object-relation and is allowed to compound. Sesa-laksanaya sasthya samasah tatra na isyate: the *compound* with a sixth case *characterized as remainder* is *not desired* there. So two genitives: the karaka-genitive (marking a role, compoundable) and the sesa-genitive (marking bare relation, not compoundable). The grammar distinguishes them by their compounding behaviour, and nothing in the form distinguishes them at all. This is 2.12's distributional method certifying a distinction the ear cannot hear — and it confirms 3.7.156: the genitive spans both the karaka-system and its remainder, and only its behaviour tells you which side it is on.

3.7.161 unformalized

अन्येन व्यपदिष्टस्य यस्यान्यत्रोपजायते ।
व्यतिरेकः स धर्मौ द्वौ लभते विषयान्तरे ॥ १६१ ॥

anyena vyapadiṣṭasya yasyānyatropajāyate |
vyatirekaḥ sa dharmau dvau labhate viṣayāntare || 161 ||

Of that which is designated by another, when a separation arises elsewhere — that one obtains two properties in another domain.

How a term can hold two relational roles. Anyena vyapadistasya: of that which is *designated by another* — something already standing in a relation. Yasya anyatra upajayate vyatirekah: when a *separation* arises *elsewhere* — it enters a second relation, in another context. Sa dharmau dvau labhate visayantare: *that one* obtains *two properties* in *another domain*. So a term can be the king's man and the village's headman at once, holding two relations without conflict, because the relations belong to different domains. This is 3.7.68's bhinna-kaksya (objecthood of a different level: the strata not colliding) applied to the genitive — and it explains why relational terms stack indefinitely where karaka-terms do not: karakas compete for one action's slots, and relations do not compete at all.

3.7.162 unformalized

प्राधान्यं स्वगुणे लब्ध्वा प्रधाने याति शेषताम् ।
सहयोगे स्वयोगेऽतः प्रधानत्वं न हीयते ॥ १६२ ॥

prādhānyaṃ svaguṇe labdhvā pradhāne yāti śeṣatām |
sahayoge svayoge'taḥ pradhānatvaṃ na hīyate || 162 ||

Having obtained principality in its own subordinate, it goes to remainder-hood in the principal; therefore, in joint connection and in its own connection, principality is not lost.

The relational hierarchy, and it mirrors the karaka one exactly. Pradhanyam sva-gune labdhva: having obtained *principality* in its *own subordinate* — a term is principal relative to what depends on it. Pradhane yati sesatam: it goes to *remainder-hood* in the *principal* — and subordinate relative to what it depends on. Saha-yoge sva-yoge atah pradhanatvam na hiyate: *therefore*, in *joint connection* and in its *own* connection, *principality* is *not lost*. Each term is principal below and remainder above, and neither status cancels the other. This is 3.7.20-23's nesting (the soldiers fighting, the king conquering; agency retained under subordination) transposed from karakas to relations — and the transposition is exact, which confirms that the samuddesa's structural insight is not about the karakas specifically but about how hierarchies of dependence work.

3.7.163 unformalized

सिद्धस्याभिमुखीभावमात्र संबोधनं विदुः ।
प्राप्ताभिमुख्यो ह्यर्थात्मा क्रियासु विनियुज्यते ॥ १६३ ॥

siddhasyābhimukhībhāvamātra saṃbodhanaṃ viduḥ |
prāptābhimukhyo hyarthātmā kriyāsu viniyujyate || 163 ||

They know the vocative to be the mere facing-toward of what is accomplished; for the meaning's self, having attained facing-toward, is applied to the actions.

The vocative defined, and it is barely a case at all. Siddhasya abhimukhi-bhava-matram sambodhanam viduh: they know the *vocative* to be the *mere facing-toward* of what is *accomplished* — matra: merely, nothing more. The vocative does not assign a role; it turns something to face. And siddhasya: of what is *accomplished*, already standing (3.7.79's siddha-rupa: the object presented as complete). Prapta-abhimukhyo hi artha-atma kriyasu viniyujyate: for the *meaning's self*, *having attained facing-toward*, is *applied to the actions*. So the vocative is prior to the karaka-system: it turns a thing toward the discourse, and *then* the actions can use it. 'Devadatta! Come!' — the calling makes him available, the imperative gives him a role. The vocative is the doorway into the sentence, and it is not a room in it.

3.7.164-165 unformalized

संबोधनं न वाक्यार्थ इति वृद्धेभ्य आगमः ।
उद्देशेन विभक्त्यर्था वाक्यार्थात् समपोद्धृताः ॥ १६४ ॥
विभक्त्यर्थेऽव्ययीभाववचनादवसीयताम् ।
अन्यो द्रव्याद् विभक्त्यर्थः सोऽव्ययेनाभिधीयते ॥ १६५ ॥

saṃbodhanaṃ na vākyārtha iti vṛddhebhya āgamaḥ |
uddeśena vibhaktyarthā vākyārthāt samapoddhṛtāḥ || 164 ||
vibhaktyarthe'vyayībhāvavacanādavasīyatām |
anyo dravyād vibhaktyarthaḥ so'vyayenābhidhīyate || 165 ||

(164) 'The vocative is not a sentence-meaning' — such is the tradition from the elders; the meanings of the case-endings, by way of indication, have been abstracted from the sentence-meaning. (165) In the meaning of a case-ending, let it be ascertained from the statement about the avyayibhava; the case-ending's meaning is other than the substance, and it is expressed by the indeclinable.

The vocative's exclusion, on authority, and then the case-endings' status. (164): sambodhanam na vakyarthah iti vrddhebhya agamah — '*the vocative is not a sentence-meaning*' — such is the *tradition from the elders* (vrddha: 3.2.7's vrddhebhya agamah; 2.485's counsel about waiting upon them). The vocative does not contribute to what the sentence says, because it is prior to the saying (3.7.163). And the general placement: uddesena vibhakty-arthah vakyarthat samapoddhrtah — the *meanings of the case-endings*, *by way of indication*, have been *abstracted from the sentence-meaning*. Apoddhrta: extracted (2.10, 3.1.1, 3.4.1 — the kanda's constant). So all the karaka-meanings this samuddesa has spent 165 verses anatomizing are extracts from the one sentence-meaning, indicated for the analysis's sake. (165) confirms it from the avyayibhava-compounds, where a case-meaning is carried by an indeclinable: anyo dravyad vibhakty-arthah — the case-ending's meaning is *other than the substance*, and it can be borne by a word that is not a noun at all. The role floats free of the thing.

3.7.166 unformalized

द्रव्यं तु यद् यथाभूतं तदत्यन्तं तथा भवेत् ।
क्रियाभेदेऽपि तस्यासौ द्रव्यात्मा नावहीयते ॥ १६६ ॥

dravyaṃ tu yad yathābhūtaṃ tadatyantaṃ tathā bhavet |
kriyābhede'pi tasyāsau dravyātmā nāvahīyate || 166 ||

But the substance, being such as it is, would be so utterly; even in a difference of action, that substance-self of it is not abandoned.

The samuddesa's penultimate verse, and it reasserts what all its constructivism did not touch. Dravyam tu yad yatha-bhutam tad atyantam tatha bhavet: the *substance*, being *such as it is*, would be so *utterly* — atyantam: absolutely, without qualification. Kriya-bhede 'pi tasya asau dravya-atma na avahiyate: *even in a difference of action*, that *substance-self* of it is *not abandoned*. The pot is a pot whether one makes it, breaks it, sees it or sits beside it; the karaka-role changes with every sentence and the thing does not. This is not a retraction of 3.7.104 (the roles arise from states of the intellect): the roles are constructed *and* the thing that wears them is not the roles. Dravya, as 3.2.11 established, is what stands when the configurations are withdrawn — and no amount of role-assignment reaches it. The verse marks exactly what the samuddesa's analysis has and has not dissolved.

3.7.167 unformalized

तस्माद् यत् करणं द्रव्यं तत् कर्म न पुनर्भवेत् ।
सर्वस्य चान्यथाभावस्तस्य द्रव्यात्मनो भवेत् ॥ १६७ ॥

tasmād yat karaṇaṃ dravyaṃ tat karma na punarbhavet |
sarvasya cānyathābhāvastasya dravyātmano bhavet || 167 ||

Therefore that substance which is the instrument would not again be the object; and there would be an otherwise-becoming of that whole substance-self.

The samuddesa's last verse, and it is a difficulty rather than a conclusion. Tasmat yat karanam dravyam tat karma na punar bhavet: *therefore* that *substance* which is the *instrument* would *not again* be the *object* — if the substance-self is unabandoned through every difference of action (3.7.166), then a thing that is an instrument cannot also be an object; the roles would have to be fixed in the things. Sarvasya ca anyatha-bhavah tasya dravya-atmano bhavet: and there would be an *otherwise-becoming* of that *whole substance-self* — for a thing to change roles, its very self would have to change. So the two halves of the samuddesa collide at its close: the roles are buddhi-constructed and shift freely (3.7.104), and the substance is utterly what it is and does not shift (3.7.166). If both, then role-change is substance-change, which 3.7.166 forbids. The samuddesa ends on the tension rather than resolving it — and the resolution is available from everything it has said: the roles are not *in* the substance to change, being postings on it (3.7.103's niyamah sabde na tu vastuni). But Bhartrhari leaves the reader to draw it, which is his habit at a movement's close.
contested: whether 167 states an objection to be dissolved (our reading, on the strength of 3.7.103-104) or a conclusion Bhartrhari endorses — fixing the karaka-roles in the things after all — is a genuine crux, and the samuddesa's last word is thereby left open