Brahma-kāṇḍa

1.1 · The Word-Brahman verified proof

अनादिनिधनं ब्रह्म शब्दतत्त्वं यदक्षरम् ।
विवर्ततेऽर्थभावेन प्रक्रिया जगतो यतः ॥ १ ॥

anādinidhanaṃ brahma śabdatattvaṃ yadakṣaram |
vivartate'rthabhāvena prakriyā jagato yataḥ || 1 ||

Brahman is without beginning or end, imperishable, and its essence is the Word. It appears in the mode of objects, and from it proceeds the unfolding of the world.

The opening karika states the sabdadvaita thesis in a single breath: the ultimate is not merely describable by language but is language — sabda-tattva, word-essence. Two terms carry the weight. First, aksara: its plain sense is 'imperishable', which is why it sits naturally beside anadinidhana, 'without beginning or end'. But aksara is also the ordinary grammarians' word for a phoneme, and Bhartrhari is a grammarian. The pun is not decorative; it is the argument in miniature. What is imperishable and what is phonic are, for him, the same thing. Second, vivartate. Bhartrhari does not say Brahman becomes the world (parinamate), which would make the world a real transformation of its substance, as Samkhya has it. Vivarta is appearance without loss of unity: the one presents itself as many while remaining one. This is the term on which the text's later reception turns, since Advaita Vedanta would make vivarta its central technical device. Note also prakriya, here 'unfolding' — in grammatical usage it is the term for a derivational procedure, the stepwise generation of a form from a root. The world issues from Brahman the way a word-form issues from its base. The choice of vocabulary is doing philosophical work.
contested: whether vivarta already carries the later Advaitic sense of *illusory* appearance, or only the weaker sense of manifestation, is disputed; the karika alone does not settle it

1.2 verified proof

एकमेव यदाम्नातं भिन्नशक्तिव्यपाश्रयात् ।
अपृथक्त्वेऽपि शक्तिभ्यः पृथक्त्वेनेव वर्तते ॥ २ ॥

ekameva yadāmnātaṃ bhinnaśaktivyapāśrayāt |
apṛthaktve'pi śaktibhyaḥ pṛthaktveneva vartate || 2 ||

That which tradition declares to be one alone, resting on its differentiated powers, functions as though it were divided — though it is in fact not separate from those powers.

Having asserted unity, Bhartrhari must account for plurality, and he does it through sakti, power. Brahman is one; its powers are many; the powers are not other than it. The syntax is careful: aprthaktve 'pi, 'even in non-separateness', concedes the metaphysical point, and then prthaktvena iva vartate, 'functions as if separate', concedes the phenomenal one. The whole position hangs on that iva, 'as if'. Remove it and you have either a real plurality (and the unity thesis collapses) or a bare denial of plurality (and the world becomes unaccountable). The sakti doctrine is Bhartrhari's device for holding both, and it places him close to what later doxography would call bhedabheda — difference-and-non-difference — rather than to a strict Advaita. Whether that proximity is genuine or an artefact of reading him backwards through his successors is a live question.

1.3 verified proof

अध्याहितकलां यस्य कालशक्तिमुपाश्रिताः ।
जन्मादयो विकाराः षड् भावभेदस्य योनयः ॥ ३ ॥

adhyāhitakalāṃ yasya kālaśaktimupāśritāḥ |
janmādayo vikārāḥ ṣaḍ bhāvabhedasya yonayaḥ || 3 ||

Resting upon its power of Time, on which divisions are superimposed, arise the six modifications beginning with birth; and these are the wombs of the differentiation of beings.

Time enters as a sakti of Brahman, not as an independent container. This is a commitment Bhartrhari develops at length much later, in the Kala-samuddesa of the third kanda, and it is worth noticing how early it is planted. The point of adhyahita-kala — 'having divisions superimposed' — is that Time in itself is undivided; the partition into moments is laid onto it, not found in it. Sequence is thus a feature of how the one appears, not of what it is, which is what the vivarta of 1.1 requires. The 'six modifications beginning with birth' are the sad-bhava-vikara that Yaska sets out in the Nirukta: a thing is born, exists, alters, grows, declines, perishes. Bhartrhari takes a piece of received grammatical-etymological analysis and makes it cosmological — the six become yoni, wombs, the generative sources of differentiated being. That move, appropriating the grammarians' apparatus as metaphysics, is characteristic of the whole work.

1.4 verified proof

एकस्य सर्वबीजस्य यस्य चेयमनेकधा ।
भोक्तृभोक्तव्यरूपेण भोगरूपेण च स्थितिः ॥ ४ ॥

ekasya sarvabījasya yasya ceyamanekadhā |
bhoktṛbhoktavyarūpeṇa bhogarūpeṇa ca sthitiḥ || 4 ||

It is one, the seed of all; and this abiding of it is manifold — in the form of the experiencer, in the form of what is to be experienced, and in the form of the experiencing.

The triad is subject, object, and act: bhoktr, bhoktavya, bhoga. Bhartrhari's claim is not that Brahman produces these three but that its sthiti — its abiding, its standing-as — simply is them. The one seed does not divide into three parts; it stands in three modes. The structure anticipates the tripuiti (knower, known, knowing) that later Vedanta and Buddhist epistemology both make central, and the resemblance is close enough that the direction of influence has been argued over. Read together with 1.2, the verse shows the sakti doctrine doing concrete work: experiencer and experienced are powers, appearing separate, not separate.

1.5 verified proof

प्राप्त्युपायोऽनुकारश्च तस्य वेदो महर्षिभिः ।
एकोऽप्यनेकवर्मेव समाम्नातः पृथक् पृथक् ॥ ५ ॥

prāptyupāyo'nukāraśca tasya vedo maharṣibhiḥ |
eko'pyanekavarmeva samāmnātaḥ pṛthak pṛthak || 5 ||

The Veda is both the means of attaining it and its likeness; and though one, it was handed down separately by the great seers, as though it had many paths.

Two words are doing the work: upaya and anukara. The Veda is the means of reaching Brahman, which is unremarkable; but it is also anukara, a likeness or imitation of it, which is not. If Brahman is word-essence, then the Veda is not merely a set of instructions pointing at the goal — it is the goal in linguistic image, and using it as a means is therefore continuous with attaining it. This is what licenses the grammarian's claim on soteriology, and the whole first kanda depends on it. The second line then handles an obvious objection: if the Veda is one, why are there many recensions? The answer reuses the pattern of 1.2 exactly — eko 'pi ... iva, one, yet as though many-pathed. The plurality of sakhas is phenomenal, like the plurality of saktis.

1.6 verified proof

भेदानां बहुमार्गत्वं कर्मण्येकत्र चाङ्गता ।
शब्दानां यतशक्तित्वं तस्य शाखासु दृश्यते ॥ ६ ॥

bhedānāṃ bahumārgatvaṃ karmaṇyekatra cāṅgatā |
śabdānāṃ yataśaktitvaṃ tasya śākhāsu dṛśyate || 6 ||

In its branches are seen the many courses taken by its divisions, the subordination of several elements to one rite, and the determinate power of words.

A continuation of 1.5, cataloguing what the recensions display. Angata is the technical Mimamsa relation of subordination — a subsidiary element standing to a principal rite — and its appearance here shows Bhartrhari assuming a ritual-exegetical background his readers would have supplied without effort. The third item, on the power of words, is the one that matters for his own project, and it is also the one whose text is least secure.
contested: yata-saktitvam is difficult; read as yata- ('restrained, fixed') it yields the determinate or fixed power of words, which is how we render it, but the compound is awkward and the line resists a confident construal on the karika alone

1.7 verified proof

स्मृतयो बहुरूपाश्च दृष्टादृष्टप्रयोजनाः ।
तमेवाश्रित्य लिङ्गेभ्यो वेदविद्भिः प्रकल्पिताः ॥ ७ ॥

smṛtayo bahurūpāśca dṛṣṭādṛṣṭaprayojanāḥ |
tamevāśritya liṅgebhyo vedavidbhiḥ prakalpitāḥ || 7 ||

The smrtis, various in form and having both seen and unseen purposes, were framed by those who know the Veda from indicatory marks, resting upon that alone.

The authority of the secondary literature is made derivative. Smrti texts are prakalpita, framed or constructed, by the veda-vid on the basis of linga — inferential marks within the Veda itself. They are not independent sources. The distinction between drsta and adrsta purposes, seen and unseen — that is, empirically evident aims versus those whose fruit is transcendent — is again standard Mimamsa currency. The verse is doing housekeeping: it subordinates the whole derivative tradition to the one Word before the argument proper resumes.

1.8 verified proof

तस्यार्थवादरूपाणि निश्रिताः स्वविकल्पजाः ।
एकत्विनां द्वैतिनां च प्रवादा बहुधा मताः ॥ ८ ॥

tasyārthavādarūpāṇi niśritāḥ svavikalpajāḥ |
ekatvināṃ dvaitināṃ ca pravādā bahudhā matāḥ || 8 ||

Resting on its arthavada portions, the manifold doctrines of the monists and the dualists are held — arising from their own conceptual constructions.

Sharp, and slightly deflationary. Arthavada is the Mimamsa category for the Veda's non-injunctive passages, the explanatory or laudatory material that does not directly command action. Bhartrhari's claim is that the rival philosophical systems — ekatvin and dvaitin, those who hold for one and those who hold for two — take their stand on precisely these portions, and that the resulting doctrines are svavikalpaja, born of their own vikalpa, their own conceptual construction. He is not quite calling them false. He is placing them: they are constructions raised on a base that was never injunctive to begin with, which is why they multiply and conflict.

1.9 verified proof

सत्या विशुद्धिस्तत्रोक्ता विद्यैवैकपदागमा ।
युक्ता प्रणवरूपेण सर्ववादाविरोधिना ॥ ९ ॥

satyā viśuddhistatroktā vidyaivaikapadāgamā |
yuktā praṇavarūpeṇa sarvavādāvirodhinā || 9 ||

The true purification is declared to be there: knowledge alone, reached through the single syllable, joined with the form of the pranava, which contradicts no doctrine.

Against the multiplying constructions of 1.8, a single point of convergence: the pranava, Om. The phrase eka-pada-agama — arrival by way of one word — recalls the aksara of 1.1, and the circle closes: the imperishable that is also the phoneme is now the syllable through which purification comes. The last quarter is the interesting one. Sarvavadavirodhina, 'not in conflict with any doctrine', makes a claim of a different order from 1.8: the rival systems contradict each other because each is built on a partial base, but Om contradicts none of them, because it is the base. Bhartrhari is not adjudicating between the schools; he is claiming a position underneath them all.
contested: vidya eva could be read either as 'knowledge alone' (excluding ritual action as a means) or more weakly as 'knowledge itself'; the first reading takes a definite stand in the knowledge-versus-action debate and may be reading later concerns into the verse

1.10 verified proof

विधातुस्तस्य लोकानामङ्गोपाङ्गनिबन्धनाः ।
विद्याभेदाः प्रतायन्ते ज्ञानसंस्कारहेतवः ॥ १० ॥

vidhātustasya lokānāmaṅgopāṅganibandhanāḥ |
vidyābhedāḥ pratāyante jñānasaṃskārahetavaḥ || 10 ||

Of that ordainer of the worlds, the divisions of knowledge — bound up with the limbs and sub-limbs — are extended, being the causes of the refinement of understanding.

The vidya-bhedas are the branches of learning, structured around the angas and upangas, the auxiliary disciplines of the Veda: phonetics, metre, grammar, etymology, ritual, astronomy, and their subsidiaries. Bhartrhari's point is functional. They are jnana-samskara-hetu, causes of the samskara — the refining, perfecting, preparing — of knowledge. They do not deliver the goal; they condition the instrument that reaches it. This sets up 1.11 directly, where one of these limbs will be singled out and elevated above the rest, and the reader who knows which discipline the author practises will already see it coming.

1.11 · Grammar as the Path verified proof

आसन्नं ब्रह्मणस्तस्य तपसामुत्तमं तपः ।
प्रथमं छन्दसामङ्गमाहुर्व्याकरणं बुधाः ॥ ११ ॥

āsannaṃ brahmaṇastasya tapasāmuttamaṃ tapaḥ |
prathamaṃ chandasāmaṅgamāhurvyākaraṇaṃ budhāḥ || 11 ||

The wise call grammar the closest to that Brahman, the highest austerity among austerities, the foremost limb of the Vedas.

The promised elevation, and it is unembarrassed: of all the auxiliary disciplines catalogued in 1.10, grammar is asanna, 'near', to Brahman itself. The claim would be mere professional self-regard were it not entailed by 1.1. If the ultimate is word-essence, then the science whose object is the word stands nearer to it than any other, not by dignity but by subject matter — proximity here is a structural fact, not an honour conferred. The other two predicates extend the point in different registers. Calling grammar tapas, austerity — and the highest of them — moves it out of the domain of technical skill and into the domain of ascesis: analysing speech is a spiritual discipline, not a preparation for one. And prathamam does double duty, meaning both 'first' in the received enumeration of the six Vedangas and 'foremost' in rank. Patanjali had already made a version of this claim for grammar in the Mahabhasya; what Bhartrhari adds is a metaphysics that makes it follow rather than merely assert it.

1.12 verified proof

प्राप्तरूपविभागाया यो वाचः परमो रसः ।
यत्तत् पुण्यतमं ज्योतिस्तस्य मार्गोऽयमाञ्जसः ॥ १२ ॥

prāptarūpavibhāgāyā yo vācaḥ paramo rasaḥ |
yattat puṇyatamaṃ jyotistasya mārgo'yamāñjasaḥ || 12 ||

Of speech that has come to be divided into forms, there is a highest savour; and of that most holy light, this is the direct path.

Prapta-rupa-vibhaga describes ordinary language — speech that has already arrived at its division into distinct forms, the differentiated medium we actually speak. Beneath or within it lies a paramo rasa, a highest essence; rasa is a word for savour, sap, the concentrated principle of a thing, and it carries a suggestion of what is tasted rather than merely known. The second line identifies that essence with a jyotis, a light, and grammar with the anjasa marga — the straight or direct road to it, from anjas, 'straightaway'. The image is precise: other approaches may reach the light, but they go around. Grammar goes at it. Note that the verse does not say ordinary differentiated speech is false; it says it has an essence, and that the essence is reachable through the discipline that studies the differentiation itself.

1.13 verified proof

अर्थप्रवृत्तितत्त्वानां शब्दा एव निबन्धनम् ।
तत्त्वावबोधः शब्दानां नास्ति व्याकरणादृते ॥ १३ ॥

arthapravṛttitattvānāṃ śabdā eva nibandhanam |
tattvāvabodhaḥ śabdānāṃ nāsti vyākaraṇādṛte || 13 ||

Words alone are the basis of the realities that govern activity toward objects; and there is no grasp of the reality of words apart from grammar.

The argument for 1.11 and 1.12, compressed into a syllogism. First premise: words are the nibandhana, the ground or fastening, of artha-pravrtti-tattva — the realities underlying our purposive activity toward objects. We do not act toward a bare world; we act toward a world already carved by language, and the carving is what our activity answers to. Second premise: the nature of words is not accessible without grammar. The conclusion, unstated because obvious: grammar grounds everything we do. The eva in the first line is doing real work — sabda eva, words *alone* — and it is the strongest form of the claim, ruling out any non-linguistic access to the structures that govern activity. This is the verse a critic would attack first, and it is where the sabdadvaita thesis stops being cosmology and starts making claims about ordinary cognition.

1.14 verified proof

तद् द्वारमपवर्गस्य वाङ्मलानां चिकित्सितम् ।
पवित्रं सर्वविद्यानामधिविद्यं प्रकाशते ॥ १४ ॥

tad dvāramapavargasya vāṅmalānāṃ cikitsitam |
pavitraṃ sarvavidyānāmadhividyaṃ prakāśate || 14 ||

That is the gateway to liberation, the therapy for the impurities of speech, the purifier of all the sciences; it shines as the science standing above the sciences.

Four predicates, escalating. Dvaram apavargasya makes grammar the door to liberation outright — not a preparation for the door, the door. Cikitsita is a medical term, 'treatment' or 'therapy', and casts corrupt speech as pathology and the grammarian as physician; the metaphor is inherited (Patanjali uses the register) but it is sharpened here by the metaphysics, since if speech is Brahman then disordered speech is disordered being, and correcting it is not pedantry but cure. Pavitram sarvavidyanam extends the therapeutic claim across the disciplines: grammar does not merely sit beside the other sciences but purifies them, since each of them is conducted in words. Adhividyam then names the resulting position — a science *above* sciences, a meta-discipline. The verse is the high-water mark of the first kanda's claims for vyakarana.

1.15 verified proof

यथार्थजातयः सर्वाः शब्दाकृतिनिबन्धनाः ।
तथैव लोके विद्यानामेषा विद्या परायणम् ॥ १५ ॥

yathārthajātayaḥ sarvāḥ śabdākṛtinibandhanāḥ |
tathaiva loke vidyānāmeṣā vidyā parāyaṇam || 15 ||

Just as all the classes of objects rest upon the forms of words, so in the world this science is the final resort of the sciences.

An analogy whose two halves are not merely parallel but causally linked. The first half asserts that artha-jati — the classes or universals under which objects fall — are nibandhana, grounded, in sabda-akrti, the forms of words. That is a substantive ontological claim: the world's joints are word-shaped, and the generality by which we group things is borrowed from language rather than found in things. The second half concludes that grammar is parayana, the final resort or culmination, of the sciences. The 'just as ... so' presents this as analogy, but the relation is stronger than analogy: because word-form grounds object-class, the science of word-form must ground the sciences that treat object-classes. Bhartrhari repeatedly builds arguments in this shape, where the rhetorical form is comparison and the logical form is entailment.

1.16-17 verified proof

इदमाद्यं पदस्थानं सिद्धिसोपानपर्वणाम् ।
इयं सा मोक्षमाणानामजिह्वा राजपद्धतिः ॥ १६ ॥
अत्रातीतविपर्यासः केवलामनुपश्यति ।
छन्दस्यश्छन्दसां योनिमात्मा छन्दोमयीं तनुम् ॥ १७ ॥

idamādyaṃ padasthānaṃ siddhisopānaparvaṇām |
iyaṃ sā mokṣamāṇānāmajihvā rājapaddhatiḥ || 16 ||
atrātītaviparyāsaḥ kevalāmanupaśyati |
chandasyaśchandasāṃ yonimātmā chandomayīṃ tanum || 17 ||

(16) This is the first foothold on the rungs of the ladder of attainment; this is the straight highroad for those seeking liberation. (17) Here one whose error has passed away, being versed in the Veda, beholds the sole body made of Veda — the womb of the Vedas, the Self.

The pair completes the sequence begun at 1.11 by naming first the road and then what stands at the end of it. In 16, two images for one point, and the first carries a pun that does not survive into English: pada-sthana is a 'place for the foot', a foothold, but pada is equally the grammarian's word for a word, so the first foothold on the ladder of siddhi is also the standing-place of the word. Ajihma rajapaddhati, the 'uncrooked royal road', then answers an objection — a road that is not crooked is not merely grand but shorter, and the claim, consistent with the anjasa marga of 1.12, is that grammar's superiority lies in going directly rather than in going somewhere the other disciplines cannot reach at all. Verse 17 supplies the vision itself, and it is reflexive: the Veda-knower beholds a body made of Veda which is at the same time the yoni, the womb, of the Vedas. What is seen is both the origin of the scriptures and their own form, and its identification with atman closes the circuit opened in 1.1, where Brahman was declared word-essence. Atita-viparyasa names the condition — viparyasa is inversion, taking one thing for another, and the vision opens only once it has fallen away.
contested: 1.17 resists confident construal: kevalam ... chandomayim tanum are feminine accusatives and plausibly form a single object ('the sole body made of Veda'), with chandasam yonim in apposition and atma as subject — but the alignment of atma with chandasyah, and whether kevalam qualifies the body or stands alone as 'the pure one', are both open; our rendering picks one reading among several defensible ones

1.18-22 verified proof

रत्यस्तमितभेदाया यद्वाचो रूपमुत्तमम् ।
यदस्मिन्नेव तमसि ज्योतिः शुद्धं विवर्तते ॥ १८ ॥
अइकृतं समतिक्रान्ता मूर्तिव्यापारदर्शनम् ।
व्यतीत्यालोकतमसी प्रकाशं यमुपासते ॥ १९ ॥
अत्र वाचो निमित्तानि चिह्नानीवाक्षरस्मृतेः ।
शब्दपूर्वेण योगेन भासन्ते प्रतिबिम्बवत् ॥ २० ॥
अथर्वणामङ्गिरसां साम्नामृग्यजुषस्य च ।
यस्मिन्नुच्चावचा वर्णाः पृथक्स्थितिपरिग्रहाः ॥ २१ ॥
अदेकं प्रक्रियाभेदैर्बहुधा प्रविभज्यते ।
तद् व्याकरणमागम्य परं ब्रह्माधिगम्यते ॥ २२ ॥

ratyastamitabhedāyā yadvāco rūpamuttamam |
yadasminneva tamasi jyotiḥ śuddhaṃ vivartate || 18 ||
aikṛtaṃ samatikrāntā mūrtivyāpāradarśanam |
vyatītyālokatamasī prakāśaṃ yamupāsate || 19 ||
atra vāco nimittāni cihnānīvākṣarasmṛteḥ |
śabdapūrveṇa yogena bhāsante pratibimbavat || 20 ||
atharvaṇāmaṅgirasāṃ sāmnāmṛgyajuṣasya ca |
yasminnuccāvacā varṇāḥ pṛthaksthitiparigrahāḥ || 21 ||
adekaṃ prakriyābhedairbahudhā pravibhajyate |
tad vyākaraṇamāgamya paraṃ brahmādhigamyate || 22 ||

(18) That highest form of speech, whose differentiations have set like the sun, the pure light that appears even within this darkness; (19) which those who have passed beyond the seeing of form and function, transcending light and darkness alike, worship as the radiance; (20) wherein the occasions of speech, like signs of the imperishable tradition, shine as in a reflection, through the yoga that is preceded by the word; (21) in which the high and low sounds of the Atharvans and Angirases, of the Saman, the Rk and the Yajus, hold their separate stations; (22) that which, though one, is divided manifold by the differences of derivational procedure — arriving at that through grammar, the supreme Brahman is attained.

A single five-verse sentence — the longest period in the kanda — building the object of the grammarian's worship clause by clause and landing, at its last word, on the discipline that reaches it. Verse 18 names the goal: the uttama rupa of vac, speech's highest form, in which bheda has 'set' (astamita, the sun-verb: differentiation gone down like daylight), a suddha jyotis that vivartate — the kanda's master-verb again — even asminn eva tamasi, within this very darkness of ordinary divided experience. 19 gives the worshippers: those past murti-vyapara-darsana, the seeing of shape and function (the whole apparatus of ordinary cognition), who have stepped beyond the pair aloka-tamasi, light *and* darkness — the highest form is not the bright member of an opposition but what precedes opposition. 20 turns to method, and its phrase is the charter of the entire tradition: sabda-purvena yogena, by the yoga preceded by the word — union pursued through language itself, in which the nimittas of speech (its occasioning apparatus: the phonemes, the aksara-smrti's signs) show like reflections, media become transparent toward their source (1.49's mirror already implicit). 21 anchors this in the concrete canon: in that highest speech the uccavaca varnas — the high and low sounds of all four Vedas, Atharvan and Angiras hymns, Saman, Rk, Yajus — hold prthak-sthiti, distinct stations: the entire audible revelation housed, differentiated, within the undifferentiated. And 22 closes the arc: that One, divided bahudha by prakriya-bhedas — differences of derivational procedure, the grammarian's own term (1.1 used prakriya of the world's unfolding; here it is declension and derivation that unfold the One) — is *arrived at through grammar*: tad vyakaranam agamya, param brahma adhigamyate. The sentence's architecture is its argument: five verses of theology suspended grammatically until vyakarana appears as the instrument of the final verb. This is the passage 1.11-14 promised and the whole kanda redeems — the discipline of speech-forms as the sabda-purva-yoga, the word-led path on which the differentiated Veda, the reflecting signs, and the settled darkness of ordinary seeing are traversed toward the light that speech itself is.
contested: the transcription is rough at two seams: 18a's ratyastamita- we read as (a)ty-astamita-bheda ('whose differentiation has utterly set'), and 19a's a-i-krtam samatikranta as atikranta with the preceding compound ('having passed beyond...'), following the passage's evident sense; 22a's adekam we read as tad ekam. More substantively, sabda-purvena yogena in 20 is a famous crux — 'yoga preceded by (grounded in) the word' names for the later tradition a whole spiritual method (sabdapurvayoga), and how technical its sense already is here is undecidable from the karika

1.23 · The Eternality of the Word-Meaning Relation verified proof

नित्याः शब्दार्थसंबन्धास्तत्रान्नाता महर्षिभिः ।
सूत्राणां सानुतन्त्राणां भाष्याणां च प्रणेतृभिः ॥ २३ ॥

nityāḥ śabdārthasaṃbandhāstatrānnātā maharṣibhiḥ |
sūtrāṇāṃ sānutantrāṇāṃ bhāṣyāṇāṃ ca praṇetṛbhiḥ || 23 ||

There the relations between word and meaning are handed down as eternal by the great seers — by the authors of the sutras with their ancillary treatises, and of the commentaries.

The nityatva thesis, stated as inherited rather than argued. That the sabda-artha-sambandha is eternal — that the bond between a word and what it means is not made — is the position the grammarians share with Mimamsa and hold against the Naiyayikas, for whom the relation is samaya, convention, established by agreement (divine or human). The stakes are not linguistic but scriptural: a conventional relation has an author and a beginning, and a Veda whose words mean by convention is a Veda someone composed. An eternal relation secures an authorless Veda. The second line names the lineage by their genres rather than their names — sutra, anutantra, bhasya — which for a grammarian points unmistakably at Panini, Katyayana and Patanjali, the munitraya whose authority this whole kanda presupposes.

1.24-26 verified proof

अपोद्धारपदार्था ये ये चार्थाः स्थितलक्षणाः ।
अन्वाख्येयाश्च ये शब्दा ये चापि प्रतिपादकाः ॥ २४ ॥
कार्यकारणभावेन योग्यभावेन च स्थिताः ।
धर्मे ये प्रत्यये चाङ्गं संबन्धाः साध्वसाधुषु ॥ २५ ॥
ते लिङ्गैश्च स्वशब्दैश्च शास्त्रेऽस्मिन्नुपवर्णिताः ।
स्मृत्यर्थमनुगम्यन्ते केचिदेव यथागमम् ॥ २६ ॥

apoddhārapadārthā ye ye cārthāḥ sthitalakṣaṇāḥ |
anvākhyeyāśca ye śabdā ye cāpi pratipādakāḥ || 24 ||
kāryakāraṇabhāvena yogyabhāvena ca sthitāḥ |
dharme ye pratyaye cāṅgaṃ saṃbandhāḥ sādhvasādhuṣu || 25 ||
te liṅgaiśca svaśabdaiśca śāstre'sminnupavarṇitāḥ |
smṛtyarthamanugamyante kecideva yathāgamam || 26 ||

Those word-meanings reached by abstraction, and those meanings whose defining marks are fixed; those words that are to be explained, and those that do the explaining; those relations obtaining by way of cause and effect and by way of fitness, which are ancillary to dharma and to cognition, holding among correct and incorrect words — these are set out in this science, by indicatory marks and in their own terms; and only some are followed, for the sake of what is to be remembered, according to tradition.

One long sentence spanning three verses, cataloguing the subject matter of the science. The technically loaded term is apoddhara, 'abstraction' or 'extraction', in apoddhara-padartha. It names what is, for Bhartrhari, a foundational concession: the word is not a natural unit given prior to the sentence but is *extracted* from it by analysis. The sentence is the real linguistic whole; words are what grammatical procedure abstracts out of it. Since the second kanda will argue this at length, its appearance here as a settled item in a list is worth marking — the doctrine is presupposed before it is defended. The rest of the catalogue is orderly: meanings with fixed marks, the anvakhyeya/pratipadaka distinction between what gets explained and what does the explaining, and relations of two kinds, causal and of yogyata or fitness. The final line is a piece of realism about the discipline: not everything true of these matters is transmitted, only kecid eva, some only, and that as tradition allows.

1.27 verified proof

शिष्टेभ्य आगमात् सिद्धाः साधवो धर्मसाधनम् ।
अर्थप्रत्यायनाभेदे विपरीतास्त्वसाधवः ॥ २७ ॥

śiṣṭebhya āgamāt siddhāḥ sādhavo dharmasādhanam |
arthapratyāyanābhede viparītāstvasādhavaḥ || 27 ||

Correct words, established from the cultivated by way of tradition, are the means of dharma; the incorrect are the reverse — though in conveying the meaning there is no difference between them.

One of the pivotal verses of the kanda, and its force lies entirely in the last quarter. Artha-pratyayana-abhede concedes, flatly, that incorrect forms communicate perfectly well: an ungrammatical word conveys its meaning no worse than a grammatical one. Bhartrhari does not try to argue that solecisms fail to signify — he grants that they succeed. The distinction between sadhu and asadhu therefore cannot be a semantic distinction at all. It is a distinction in dharma: correct forms are dharma-sadhana, a means of merit, and incorrect ones are viparita, the reverse. This is what preserves the grammarian's enterprise from the obvious objection that if people understand each other anyway, grammar is idle. The reply is that grammar was never in the business of securing communication; it is in the business of securing merit, and the two come apart. Note also that sadhu forms are established sistebhyah agamat — from the cultivated, by tradition — not by rule alone. Usage of the learned is the criterion, and the rules describe it.

1.28 verified proof

नित्यत्वे कृतकत्वे वा तेषामादिर्न विद्यते ।
प्राणिनामिव सा चैषा व्यवस्थानित्यतोच्यते ॥ २८ ॥

nityatve kṛtakatve vā teṣāmādirna vidyate |
prāṇināmiva sā caiṣā vyavasthānityatocyate || 28 ||

Whether they are eternal or made, no beginning of them is found — as with living beings; and this arrangement is what is called eternality.

A strikingly concessive move. Bhartrhari declines to settle whether words are nitya or krtaka, eternal or manufactured, and argues that the dispute does not matter for his purpose: on either horn, no adi, no beginning, is found. The analogy to praninam carries the argument — living beings are certainly produced, each from a parent, yet the series of them has no first term, and their beginninglessness is perfectly compatible with each member being made. Hence vyavastha-nityata, 'eternality as regular arrangement': what is eternal is not necessarily the item but the settled order in which items stand. This is a weaker and considerably more defensible thesis than the one 1.23 reports from tradition, and it lets the grammarian keep everything he needs — an authorless, beginningless linguistic order — without having to win the metaphysical argument about whether individual words are eternal substances.

1.29 verified proof

नानर्थिकामिमां कश्चिद् व्ववस्थां कर्तुमर्हति ।
तस्मान्निवध्यते शिष्टैः साधुत्वविषया स्मृतिः ॥ २९ ॥

nānarthikāmimāṃ kaścid vvavasthāṃ kartumarhati |
tasmānnivadhyate śiṣṭaiḥ sādhutvaviṣayā smṛtiḥ || 29 ||

No one would undertake this arrangement if it served no purpose; therefore the tradition concerning correctness is composed by the cultivated.

A brief argument from purpose, propping up 1.27. The reasoning runs backwards from the existence of the discipline to its point: nobody codifies an anarthika, purposeless, system, and the sistas plainly did codify one; so it must have a purpose, and by 1.27 that purpose is dharma rather than communication. The argument is frankly weak taken alone — the mere existence of a practice is poor evidence that it achieves what it claims — but it is not meant to stand alone. It is a bridge, converting the doctrinal claim of 1.27 into a reason to trust the smrti that follows, and setting up the far more interesting verses on the limits of reasoning that come next.

1.30 verified proof

नागमादृते धर्मस्तर्केण व्यवतिष्ठते ।
ऋषीणामपि यज्ज्ञानं तदप्यागमपूर्वकम् ॥ ३० ॥

nāgamādṛte dharmastarkeṇa vyavatiṣṭhate |
ṛṣīṇāmapi yajjñānaṃ tadapyāgamapūrvakam || 30 ||

Dharma is not established by reasoning apart from tradition; even the knowledge that the seers possess is itself preceded by tradition.

The most consequential verse in this stretch, and one of the most cited in the whole work. The first line subordinates tarka to agama in matters of dharma, which is orthodox enough. The second line is the radical one: even rsis, whose authority rests on direct vision, do not see without tradition — their jnana too is agama-purvaka, preceded by what was handed down. This closes the escape route by which one might appeal past the tradition to the unmediated insight of those who founded it. There is no such unmediated layer. Every seeing is already informed by what was received. Whether this is a humbling of reason or a quiet circularity — tradition validated by seers whose seeing is constituted by tradition — has been debated ever since, and the verse does not defend itself against the charge.

1.31 verified proof

धर्मस्य चाव्यवच्छिन्नाः पन्थानो ये व्यवस्थिताः ।
न तांल्लोकप्रसिद्धत्वात् कश्चित्तर्केण बाधते ॥ ३१ ॥

dharmasya cāvyavacchinnāḥ panthāno ye vyavasthitāḥ |
na tāṃllokaprasiddhatvāt kaścittarkeṇa bādhate || 31 ||

And those paths of dharma that stand unbroken — no one sets them aside by reasoning, because they are established in the world.

The complement to 1.30. Where that verse denied reasoning the power to *establish* dharma, this one denies it the power to *overturn* it. The operative term is avyavacchinna, 'unbroken', 'uninterrupted': what secures these paths is the sheer continuity of their transmission, and loka-prasiddhatva, their being established in the world, functions as the warrant. Bhartrhari is describing something like a burden of proof. Reasoning does not get to arraign an unbroken practice and demand its credentials; the practice's persistence is its credential, and tarka arrives too late to sit in judgement. The position is conservative, but it is not anti-rational in an interesting way — the claim is about what reasoning is *for*, not about whether it works.

1.32 · The Limits of Inference verified proof

अवस्थादेशकालानां भेदाद् भिन्नासु शक्तिषु ।
भावानामनुमानेन प्रसिद्धिरतिदुर्लभा ॥ ३२ ॥

avasthādeśakālānāṃ bhedād bhinnāsu śaktiṣu |
bhāvānāmanumānena prasiddhiratidurlabhā || 32 ||

Since the powers of things are differentiated by differences of state, place and time, the establishing of them by inference is exceedingly hard to attain.

The opening of a sustained critique of anumana, and it begins with the strongest available premise. Inference from observed cases requires that the power at work in the observed case be the power at work in the unobserved one. But sakti, Bhartrhari says, varies with avastha, desa and kala — with the condition of the thing, where it is, and when. If the causal powers of things are indexed to circumstances, then no finite survey of circumstances licenses a general conclusion, and prasiddhi by inference becomes ati-durlabha: not impossible, but exceedingly hard to come by. The force of the argument is that it does not need scepticism about the external world or about perception. It grants that things have powers and that we observe them, and locates the failure precisely at the inductive step. The next two verses will press the point from two different directions.

1.33 verified proof

निर्झतशक्तेर्द्रव्यस्य तां तामर्थक्रियां प्रति ।
विशिष्टद्रव्यसंबन्धे सा शक्तिः प्रतिबध्यते ॥ ३३ ॥

nirjhataśakterdravyasya tāṃ tāmarthakriyāṃ prati |
viśiṣṭadravyasaṃbandhe sā śaktiḥ pratibadhyate || 33 ||

Even in the case of a substance whose power toward this or that effect is well known, that power is blocked when it enters into connection with some particular substance.

The first pressure: defeasibility. It is not merely that we may be ignorant of a thing's power; it is that a power we know perfectly well can be pratibadhyate, obstructed, by the presence of something else. Fire burns — this is as well established as anything — and yet the standard examples of the tradition are gems and mantras that stop it burning. The point is structural rather than anecdotal. If any known causal power can be defeated by an unsurveyed viseṣa, a particular, then the inference from 'x has the power to produce y' to 'x will produce y here' is never secure, because securing it would require knowing that no defeater is present, and that is not something inference can deliver. Arthakriya, effective action, is the same term the Buddhist epistemologists make central to their account of the real, which sharpens the irony: the criterion of reality is exactly what turns out not to be reliably inferable.

1.34 verified proof

यत्नेनानुमितोऽप्यर्थः कुशलैरनुमातृभिः ।
अभियुक्ततरैरन्यैरन्यथैवोपपाद्यते ॥ ३४ ॥

yatnenānumito'pyarthaḥ kuśalairanumātṛbhiḥ |
abhiyuktatarairanyairanyathaivopapādyate || 34 ||

Even a matter inferred with great effort by skilled reasoners is established quite otherwise by others of still greater skill.

The most famous verse in the kanda, and the second and more damaging pressure. The previous verses argued that inference is hard because the world is complicated. This one argues that inference is unreliable because of something about inference itself. Take the best case available — kusala anumatr, skilled reasoners, proceeding yatnena, with effort — and it remains true that abhiyuktatara, those more expert still, overturn the result and demonstrate the opposite. The argument is a pessimistic induction on the history of reasoning: since the record shows that careful inferences are routinely reversed by more careful ones, the mere carefulness of an inference is no guarantee of its truth. What makes the verse bite is that there is no obvious reply. Adding more skill does not help, since the point is precisely that more skill is what overturns the previous result; there is no level of expertise at which the regress terminates. Bhartrhari deploys this to clear ground for agama, but the argument is detachable from that use and has been read on its own ever since.

1.35 verified proof

परेषामसमाख्येयमभ्यासादेव जायते ।
मणिरूप्यादिविज्ञानं तद्विदां नानुमानिकम् ॥ ३५ ॥

pareṣāmasamākhyeyamabhyāsādeva jāyate |
maṇirūpyādivijñānaṃ tadvidāṃ nānumānikam || 35 ||

The knowledge of gems, silver and such that experts possess arises from practice alone; it cannot be conveyed to others, and it is not inferential.

A quieter verse, and philosophically one of the most interesting in the work. The jeweller knows the stone is genuine. He did not infer it — asked for the linga, the mark from which he reasoned, he cannot produce one; asamakhyeyam, it cannot be told to others. The knowledge came from abhyasa, sheer repeated practice, and it is reliable. Bhartrhari's use of the example is local: he wants a wedge against the claim that anumana exhausts non-perceptual knowledge, and connoisseurship supplies one, since here is knowledge that is neither perception nor inference nor testimony. But the observation reaches further than the use he puts it to. What he has described is tacit knowledge — competence that outruns the agent's ability to articulate its grounds — and the description is exact. The verse also sets up 1.36-38: once it is granted that reliable knowledge need not be inferential or communicable, the seer's vision becomes harder to dismiss on the ground that he cannot show his reasoning.

1.36 verified proof

प्रत्यक्षमनुमानं च व्यतिक्रम्य व्यवस्थिताः ।
पितृरक्षःपिशाचानां कर्मजा एव सिद्धयः ॥ ३६ ॥

pratyakṣamanumānaṃ ca vyatikramya vyavasthitāḥ |
pitṛrakṣaḥpiśācānāṃ karmajā eva siddhayaḥ || 36 ||

The attainments of the ancestors, the raksases and the pisacas, standing beyond both perception and inference, are born of karma alone.

The wedge from 1.35 is widened by an appeal to what the tradition takes to exist. The siddhis of these beings are karmaja, born of karma, and they lie outside the reach of both pramanas that a critic would accept. Whether the modern reader grants the premise is beside the point of the argument's structure: Bhartrhari is establishing that his opponents' own commitments already include cases where perception and inference are jointly insufficient, so a principle restricting knowledge to those two is not one his interlocutors can consistently hold. The move is dialectical, and it is worth being clear that it is doing no independent work for a reader who does not share the commitment.

1.37 verified proof

आविर्भूतप्रकाशानामनुपप्लुतचेतसाम् ।
अतीतानागतज्ञानं प्रत्यक्षान्न विशिष्यते ॥ ३७ ॥

āvirbhūtaprakāśānāmanupaplutacetasām |
atītānāgatajñānaṃ pratyakṣānna viśiṣyate || 37 ||

For those whose inner light has come forth, whose minds are unclouded, knowledge of the past and the future is not different from perception.

The positive claim toward which the whole critique has been building. Two conditions are named: avirbhuta-prakasa, an inner light that has become manifest, and anupapluta-cetas, a mind not flooded or disturbed. Where they are met, knowledge of atita and anagata — what is gone and what has not yet come — is pratyaksat na visisyate, not distinguishable from perception. The claim is not that such knowledge is a superior inference. It is that it is not inference at all; it has perception's immediacy, and therefore inherits perception's freedom from the defeat 1.34 visited on reasoning. Structurally this is why the preceding critique had to come first. Having shown that inference cannot secure its conclusions, and that non-inferential expert knowledge exists (1.35), Bhartrhari can present yogic cognition as an extension of a category already conceded rather than as a special pleading.

1.38 verified proof

अतीन्द्रियानसंवेद्यान् पश्यन्त्यार्षेण चक्षुषा ।
ये भावान् वचनं तेषां नानुमानेन बाध्यते ॥ ३८ ॥

atīndriyānasaṃvedyān paśyantyārṣeṇa cakṣuṣā |
ye bhāvān vacanaṃ teṣāṃ nānumānena bādhyate || 38 ||

Those who see, with the seer's eye, things that lie beyond the senses and cannot be felt — their word is not overturned by inference.

The conclusion, and the join back to 1.30. If the rsis' knowledge is perceptual in character (1.37), and inference is defeasible in the way 1.34 showed, then vacana — their word, the agama — cannot be arraigned by tarka: the weaker instrument does not get to overrule the stronger. This completes the argument that 1.30 and 1.31 asserted without proof, and it does so without ever claiming that reasoning is worthless. The claim throughout has been comparative. Inference is a real pramana; it is simply not one that can stand against direct seeing, and the domain of dharma is exactly where direct seeing is available to some and inference to all. Whether the argument succeeds depends on a step it does not defend — that we can identify who the arsa-caksus belong to — and that gap is where its critics have always gone.

1.39 verified proof

यो यस्य स्वमिव ज्ञानं दर्शनं नातिशङ्कते ।
स्थितं प्रत्यक्षपक्षे तं कथमन्यो निवर्तयेत् ॥ ३९ ॥

yo yasya svamiva jñānaṃ darśanaṃ nātiśaṅkate |
sthitaṃ pratyakṣapakṣe taṃ kathamanyo nivartayet || 39 ||

When someone does not doubt a cognition — a seeing that is, as it were, his own possession — and it stands on the side of perception, how could another turn him back from it?

A phenomenological coda to the epistemology. The jeweller of 1.35 and the seer of 1.37 share a mark: their cognition arrives with the self-evidence of perception, svam iva, like one's own property, and na atisankate, without residual doubt. Bhartrhari's question — katham anyah nivartayet, how could another dislodge it? — is not rhetorical bluster; it records something true about the architecture of conviction. Argument gets its grip only where doubt has a foothold, and a cognition that presents itself perceptually offers none. The verse thereby also quietly marks the limit of the whole preceding polemic: if the opponent's conviction is likewise perception-shaped, no argument of Bhartrhari's will move him either. The appeal to what stands pratyaksa-pakse, on perception's side, cuts in every direction.

1.40 verified proof

इदं पुण्यमिदं पापमित्येतस्मिन् पदद्वये ।
आचण्डालं मनुष्याणामल्पं शास्त्रप्रयोजनम् ॥ ४० ॥

idaṃ puṇyamidaṃ pāpamityetasmin padadvaye |
ācaṇḍālaṃ manuṣyāṇāmalpaṃ śāstraprayojanam || 40 ||

'This is merit, this is sin' — in this pair of terms, for men down to the candala, the need for the treatise is small.

A concession that reframes what sastra is for. Everyone, acandalam — down to the outcaste, the expression marking the widest possible social range — already commands the pair punya and papa. Ordinary moral competence does not wait upon learning, and the treatise's prayojana there is alpa, slight. The verse prevents an overreading of 1.30-31: the dependence of dharma on agama does not mean the unlettered are morally blind, since tradition circulates in common life and not only in texts. What the sastra adds is precision at the margins — the hard cases, the exact forms — not the basic orientation, which is already universally distributed. The realism here about how little explicit doctrine ordinary practice requires is of a piece with 1.35's realism about how little explicit reasoning expert knowledge requires.

1.41 verified proof

चैतन्यमिव यश्चायमविच्छेदेन वर्तते ।
आगमस्तमुपासीनो हेतुवादैर्न बाध्यते ॥ ४१ ॥

caitanyamiva yaścāyamavicchedena vartate |
āgamastamupāsīno hetuvādairna bādhyate || 41 ||

And this tradition, which persists without interruption like consciousness itself, cannot be overturned by argumentation for the one who abides in it.

The strongest statement of the agama doctrine, made through its boldest simile. Tradition persists avicchedena, without break, caitanyam iva — the way consciousness does. The comparison is exact on two points. Consciousness has no observable first moment, and neither has the tradition (1.28's vyavastha-nityata); and consciousness is not the kind of thing one argues someone out of, since every argument presupposes it. The simile thus imports transcendental status: agama is presented not as a body of claims that might be true or false but as a standing condition of the upasina, the one who dwells in it. Hetuvada, reasoned refutation, arrives inside that condition and so cannot get behind it. The move is powerful and question-begging in equal measure — to one not already dwelling in the tradition, the analogy with consciousness is exactly what needs showing.

1.42 verified proof

हस्तस्पर्शादिवान्धेन विषमे पथि धावता ।
अनुमानप्रधानेन विनिपातो न दुर्लभः ॥ ४२ ॥

hastasparśādivāndhena viṣame pathi dhāvatā |
anumānapradhānena vinipāto na durlabhaḥ || 42 ||

Like a blind man running on rough ground, feeling his way by the touch of his hand, one who relies chiefly on inference does not find his fall hard to come by.

The polemic against anumana compressed into one image, and the image is chosen with care. The blind man's hand does give him real information — touch is a genuine pramana, as inference is — and on even ground, walking slowly, it suffices. The failure comes from the combination the verse specifies: visame pathi, on broken ground, and dhavata, running. Inference fails not because it delivers nothing but because it is local, serial and slow, while the terrain (1.32-33: powers indexed to circumstance, defeaters unsurveyed) is uneven and the reasoner in practice moves at speed. Vinipata, the headlong fall, is then na durlabha — 'not hard to obtain', the litotes doing dry work. The verse closes the epistemological digression; 1.43 draws the practical conclusion for grammar.

1.43 verified proof

तस्मादकृतकं शास्त्रं स्मृतिं च सनिबन्धनाम् ।
आश्रित्यारभ्यते शिष्टैः शब्दानामनुशासनम् ॥ ४३ ॥

tasmādakṛtakaṃ śāstraṃ smṛtiṃ ca sanibandhanām |
āśrityārabhyate śiṣṭaiḥ śabdānāmanuśāsanam || 43 ||

Therefore it is by resting on the uncreated teaching, and on the tradition with its grounds, that the cultivated undertake the regulation of words.

Tasmat gathers the whole arc since 1.27. Because correctness serves dharma and not mere communication (1.27), because dharma is not established or overturned by unaided reasoning (1.30-31, 1.42), the sistas' anusasana of words — the term echoes sabdanusasana, the traditional name of grammar itself — proceeds akrtakam sastram asritya, resting on a teaching that is not made. Sanibandhana, 'with its grounds' or 'bindings', concedes that the smrti is not arbitrary: it has nibandhana, attachments to its source, the lingas of 1.7. The verse ends the kanda's first movement. The ground of the discipline is now laid, and with 1.44 the text turns abruptly from why grammar matters to what a word is — the sphota doctrine, which is the reason the work is still read.

1.44 · The Sphota Doctrine verified proof

द्वावुपादानशब्देषु शब्दौ शब्दविदो विदुः ।
एको निमित्तं शब्दानामपरोऽर्थे प्रयुज्यते ॥ ४४ ॥

dvāvupādānaśabdeṣu śabdau śabdavido viduḥ |
eko nimittaṃ śabdānāmaparo'rthe prayujyate || 44 ||

In the words employed, those who know the word discern two words: one is the cause of words, the other is what is used in respect of meaning.

The kanda's central distinction, stated with deliberate paradox: in every uttered word there are dvau sabdau, two words. One is nimitta, the cause — the sound-complex, produced by the organs, serial, audible, the thing phonetics studies. The other is the one prayujyate arthe, actually deployed toward meaning — the word as meaning-bearer, which Bhartrhari will call sphota. The paradox dissolves once one sees that 'word' was doing two jobs all along. What conveys meaning cannot simply be the sound-series, since the series is never present as a whole (when the last phoneme sounds, the first is gone), yet meaning arrives whole. So the bearer of meaning and the vehicle of audition must be distinguished — the sphota being what is expressed, the dhvani or nada being what expresses it. Everything from here to 1.56 elaborates this pair, and the terminology (vaikhari, madhyama, pasyanti come later, at 1.142ff.) builds on it.
contested: which of the two is 'cause of words' admits debate: we take nimitta as the sound qua cause of the word's manifestation, but some construe the nimitta-sabda as the inner word that causes utterance, inverting the assignment; the karika's own syntax permits either

1.45 verified proof

आत्मभेदस्तयोः केचिदस्तीत्याहुः पुराणगाः ।
बुद्धिभेदादभिन्नस्य भेदमेके प्रचक्षते ॥ ४५ ॥

ātmabhedastayoḥ kecidastītyāhuḥ purāṇagāḥ |
buddhibhedādabhinnasya bhedameke pracakṣate || 45 ||

Some who follow the ancient way say there is a difference of essence between the two; others declare that the difference of what is in fact undivided arises from a difference in cognition.

Having drawn the distinction, Bhartrhari immediately reports two ways of taking its metaphysics, and declines — here — to arbitrate. The puranagas, followers of the old doctrine, hold atmabheda: sound and sphota are two in essence, really distinct items. The eke, the others, hold that the sphota is abhinna, undivided, and that duality is buddhibheda-ja, born of a difference in the apprehending cognition — one thing taken twice over. The second position is the one that coheres with the vivarta metaphysics of 1.1-2, where all plurality is appearance riding on the one, and the structure of this verse (real difference versus cognition-made difference) exactly reprises the sakti discussion of 1.2. Bhartrhari's own settled view emerges over the following verses as closer to the second, but the doxographic habit — stating rivals fairly before proceeding — is itself characteristic of the work.
contested: the identity of the puranagas is unrecoverable from the karika alone; whether they are early grammarians, Mimamsakas, or a lost school is guesswork, and we leave the term descriptive

1.46 verified proof

अरणिस्थं यथा ज्योतिः प्रकाशान्तरकारणम् ।
तद्वच्छब्दोऽपि बुद्धिस्थः श्रुतीनां कारणं पृथक् ॥ ४६ ॥

araṇisthaṃ yathā jyotiḥ prakāśāntarakāraṇam |
tadvacchabdo'pi buddhisthaḥ śrutīnāṃ kāraṇaṃ pṛthak || 46 ||

As the fire that resides in the kindling-wood is the cause of further illumination, so the word residing in the intellect is, separately, the cause of audible sounds.

The first of the great sphota similes. Fire is latent in the arani, the fire-stick, before any kindling; churned, it becomes flame, which then illuminates other things. So the word sits buddhistha, in the intellect, complete and unsounded; articulation churns it into audible sruti, and the sounds then do the manifesting. The simile is chosen for what it denies. The flame is not created by the churning ex nihilo — it was in the wood in potentia — and the sounds do not create the word; they release and transmit what was already whole within the speaker. Prthak, 'separately', guards the distinction of 1.44: the inner word is a cause of the sounds without being identical to them, as fire-in-wood is not the visible blaze. The direction of the simile is speaker-side; 1.47-49 will trace the same structure from utterance to hearer.

1.47 verified proof

वितर्कितः पुरा बुद्धया क्वचिदर्थं निवेशितः ।
करणेभ्यो विवृत्तेन ध्वनिना सोऽनुगृह्यते ॥ ४७ ॥

vitarkitaḥ purā buddhayā kvacidarthaṃ niveśitaḥ |
karaṇebhyo vivṛttena dhvaninā so'nugṛhyate || 47 ||

First conceived by the intellect and set upon some meaning, it is then assisted by the sound unrolled from the organs of speech.

The production story, told in the order that matters. The word is vitarkitah pura buddhya — deliberated first, in the mind — and nivesitah, installed, upon its meaning before a single sound occurs. Only then do the karanas, the articulatory organs, come into play, and their contribution is named with a precise and deflationary verb: the inner word anugrhyate, is *favoured* or assisted, by the dhvani. Sound is an auxiliary, a helper of manifestation, not the substance of the word. Vivrtta, 'unrolled', characterizes the sound as the serial, spread-out thing it is — what the organs produce must come out in sequence because the organs work in time. The whole verse is engineered to locate sequence in the machinery of expression and completeness in what is expressed, preparing 1.48's paradox directly.

1.48 verified proof

नादस्य क्रमजन्मत्वान्न पूर्वो न परश्च सः ।
अक्रमः क्रमरूपेण भेदवानिव जायते ॥ ४८ ॥

nādasya kramajanmatvānna pūrvo na paraśca saḥ |
akramaḥ kramarūpeṇa bhedavāniva jāyate || 48 ||

Since audible sound is born in sequence, the word is neither prior nor posterior; the sequenceless, taking on the form of sequence, is born as though divided.

The metaphysical core of the sphota doctrine, and the verse where its vocabulary locks onto the vocabulary of 1.1. Nada, realized sound, is krama-janman, sequence-born; that is simply what acoustic events are. But the word itself is na purvo na parah — it has no earlier and later parts, because the meaning-bearer must be present whole wherever it is present at all. The reconciliation is the vivarta pattern again: akramah kramarupena bhedavan iva jayate, the sequenceless is born *as if* divided, in the form of sequence. The iva carries the doctrine, exactly as it carried 1.2. And the wording deliberately mirrors the cosmology: as Brahman, undivided, appears as a world unfolding in time, so the word, undivided, appears as a string of phonemes. The microcosm-macrocosm parallel is not an ornament — for Bhartrhari the two cases are instances of the same structure, which is why a treatise on grammar could open with Brahman.

1.49 verified proof

प्रतिबिम्बं यथान्यत्र स्थितं तोयक्रियावशात् ।
तत्प्रवृत्तिमिवान्वेति स धर्मः स्फोटनादयोः ॥ ४९ ॥

pratibimbaṃ yathānyatra sthitaṃ toyakriyāvaśāt |
tatpravṛttimivānveti sa dharmaḥ sphoṭanādayoḥ || 49 ||

As a reflection standing elsewhere seems to follow the movements of the water — such is the relation, that dharma, between sphota and sound.

The subtlest of the similes, addressing a specific residual problem: why does the word seem to have the properties of its sounds — length, speed, accent — if it is sequenceless? The moon's reflection trembles when the water trembles; the trembling belongs to the water, but presents itself as belonging to the reflected moon, which stands anyatra, elsewhere, unmoved. So the sphota appears drawn out in slow speech and clipped in fast speech, appears to begin and end, while these are throughout properties of the dhvani, transferred in appearance only. The simile refines 1.46 and 1.47: manifestation is not neutral, it lends the manifested thing its own dress. Later grammarians systematized exactly this point by distinguishing prakrta from vaikrta dhvani — the sound that manifests versus the sound that varies with delivery — and this verse is the seed of that distinction.
contested: the compound sphotanadayoh with sa dharmah is elliptical: 'that property (relation) holds of sphota and nada' is our construal, but what exactly dharma picks out — the following-relation, the appearance of shared properties, or a property of the pair — is left to the reader

1.50 verified proof

आत्मरूपं यथा ज्ञाने ज्ञेयरूपं च दृश्यते ।
अर्थरूपं तथा शब्दे स्वरूपं च प्रकाशते ॥ ५० ॥

ātmarūpaṃ yathā jñāne jñeyarūpaṃ ca dṛśyate |
artharūpaṃ tathā śabde svarūpaṃ ca prakāśate || 50 ||

As in a cognition there is seen both its own form and the form of the object known, so in the word the form of the meaning and its own form both shine forth.

An analogy from the philosophy of mind, and one that assumes a definite position in it. A cognition, on this view, is self-revealing: in knowing the pot, awareness displays the pot's form and its own — no second cognition is needed to know that one knows. The word, Bhartrhari says, has the same double luminosity. It presents artha-rupa, the meaning, and simultaneously sva-rupa, itself; we hear the meaning through the word and the word along with the meaning. This matters for the grammarian's own practice: when grammar mentions a word (rather than using it), the sva-rupa is what is being talked about — a point Panini legislates for explicitly (svam rupam sabdasya, Astadhyayi 1.1.68), and which 1.44's two-word doctrine already gestured at. The self-luminosity model of cognition invoked here is the one the Prabhakara Mimamsakas and others would defend at length; Bhartrhari takes it as available without argument.

1.51 verified proof

आण्डभावमिवापन्नो यः क्रतुः शब्दसंज्ञकः ।
वृत्तिस्तस्य क्रियाभूता भागशो लभते क्रमम् ॥ ५१ ॥

āṇḍabhāvamivāpanno yaḥ kratuḥ śabdasaṃjñakaḥ |
vṛttistasya kriyābhūtā bhāgaśo labhate kramam || 51 ||

The energy called the word, having arrived as it were at the condition of an egg, its functioning becoming activity, attains sequence part by part.

The organic simile, completing the set. Kratu — a Vedic word for sacrifice, but here in its older sense of power, resolve, energy — names the word in its unmanifest state, anda-bhavam ivapannah, come as it were to egg-hood: everything the bird will be, folded and undisplayed. Its vrtti, functioning, kriyabhuta, turned into activity, labhate kramam bhagasah — takes on sequence limb by limb, as the embryo articulates itself. Where the fire simile (1.46) modelled latency and the reflection simile (1.49) modelled borrowed properties, the egg models internal complexity without sequence: the parts of the word are all 'there' in the sphota as the bird is in the egg — really recoverable, yet not laid out in time until manifestation unfolds them. It is the closest the similes come to answering how a partless whole can nonetheless ground the appearance of ordered parts.
contested: kratu is variously understood; taking it as 'energy/resolve' follows its Vedic usage, but a reader who takes the sacrificial sense will hear the word as a rite that unfolds in ordered stages, and the simile still works — which may be why the word was chosen

1.52 verified proof

यथैकबुद्धिविषया मूर्तिराक्रियते पटे ।
मूर्त्यन्तरस्य त्रितयमेवं शब्देऽपि दृश्यते ॥ ५२ ॥

yathaikabuddhiviṣayā mūrtirākriyate paṭe |
mūrtyantarasya tritayamevaṃ śabde'pi dṛśyate || 52 ||

As a figure, though the object of a single cognition, is executed on the canvas by way of the triad of another figure, so it is seen in the case of the word as well.

The painter sees the image whole — eka-buddhi-visaya, object of one cognition — but to put it on cloth he must work through parts and stages, sketch and outline and colour. The finished figure on the canvas is one; the procedure that realized it was many-staged. So with the word: unitary in the speaker's cognition, realized through the staged procedure of articulation. The simile adds something the others lack: agency. Fire, reflection and egg are natural processes, but painting is deliberate craft, and it models the *speaker's* situation — one who possesses the whole and must nonetheless produce it serially, constrained by his instruments as the painter is by brush and cloth.
contested: murtyantarasya tritayam, 'the triad of another figure', is obscure, and the two live construals diverge. We first read it as the painter's staged working procedure. The commentarial tradition (the Vrtti, glossed by Vrsabha) instead takes the triad as three stages of a recurring cycle that holds of painting and word alike: the thing is first apprehended in sequence, then held as the object of a single sequenceless cognition, then externalized in sequence again for another's benefit. On that reading the verse is not about craft constraints but about the word's permanent oscillation between unity and sequence — it 'goes again and again through three stages', each utterance re-enacting the cycle. The traditional reading integrates better with 1.53 (speaker's and hearer's cognitions landing on the same unity) and we now prefer it; the translation above is compatible with both

1.53 verified proof

यथा प्रयोक्तुः प्राग् बुद्धिः शब्देष्वेव प्रवर्तते ।
व्यवसायो ग्रहीतृणामेवं तेष्वेव जायते ॥ ५३ ॥

yathā prayoktuḥ prāg buddhiḥ śabdeṣveva pravartate |
vyavasāyo grahītṛṇāmevaṃ teṣveva jāyate || 53 ||

As the cognition of the speaker first operates upon the words themselves, so the determination of the hearers likewise arises upon those very words.

Symmetry of production and comprehension. The speaker's buddhi engages sabdesv eva, the words themselves, before utterance (as 1.47 said); the hearer's vyavasaya, determinate grasp, arises tesv eva, upon those same words. Two claims ride on the emphatic eva. First, comprehension is not a matter of receiving phonemes and computing a meaning from them; the hearer's cognition lands on the word — the same undivided sphota the speaker began with — and the meaning shines out (1.50) once it does. Second, communication is real: speaker and hearer converge on numerically the same word, not on private tokens that merely resemble each other. Sequence, on this picture, belongs only to the channel between them; at both ends stands the whole. That the last phoneme, aided by the traces of the preceding ones, triggers the hearer's grasp of the whole is developed in verses to come (1.84ff.); here the symmetry itself is the point.

1.54 verified proof

अर्थोपसर्जनीभूतानभिधेयेषु केषुचित् ।
चरितार्थान् परार्थत्वान्न लोकः प्रतिपद्यते ॥ ५४ ॥

arthopasarjanībhūtānabhidheyeṣu keṣucit |
caritārthān parārthatvānna lokaḥ pratipadyate || 54 ||

In the case of certain things-to-be-expressed, words become subordinate to the meaning and have discharged their purpose; being for the sake of another, the world does not attend to them.

A phenomenological observation that every reader can verify: in ordinary hearing we notice what is said and the words vanish into it. Bhartrhari's vocabulary places the observation precisely. The words become artha-upasarjani-bhuta — upasarjana is the grammarians' term for the subordinate member of a compound — demoted, in the economy of attention, below the meaning they serve; caritartha, their task discharged; pararthatvat, existing for another's sake, like an instrument. Hence lokah na pratipadyate, the world takes no cognizance of them. The verse explains why the sva-rupa of 1.50, though always co-presented, is ordinarily unnoticed — and, implicitly, what a grammarian is: one who reverses the ordinary direction of attention and looks at the instrument the world looks through. Kesucit, 'in certain cases', leaves room for the contrast case the next verses take up, where the word itself is the topic.

1.55 verified proof

ग्राह्यत्वं ग्राहकत्वं च द्वे शक्ती तेजसो यथा ।
तथैव सर्वशब्दानामेते पृथगवस्थिते ॥ ५५ ॥

grāhyatvaṃ grāhakatvaṃ ca dve śaktī tejaso yathā |
tathaiva sarvaśabdānāmete pṛthagavasthite || 55 ||

As light has two powers, that of being grasped and that of grasping, so all words, without exception, have these two powers standing distinct.

Light is the one thing in ordinary experience that reveals itself in revealing others: to see the lamp's light and to see by it are not two events. Words, Bhartrhari says, have the same double power — grahyatva, graspability (they are heard, they are objects), and grahakatva, graspingness (they seize their meanings and present them). The simile consolidates 1.50 and 1.54 into a doctrine: the word is a luminary, and the two powers are prthag avasthite, distinct in standing — which is why attention can fall on either, the world resting in the grasping use (1.54), the grammarian attending to the graspable face. Sarva-sabdanam, of *all* words: the doctrine tolerates no exceptions, and the next verse draws the consequence — nothing signifies without first being seized.

1.56 verified proof

विषयत्वमनापन्नैः शब्दैर्नार्थः प्रकाश्यते ।
न सत्तयैव तेऽर्थानामगृहीताः प्रकाशकाः ॥ ५६ ॥

viṣayatvamanāpannaiḥ śabdairnārthaḥ prakāśyate |
na sattayaiva te'rthānāmagṛhītāḥ prakāśakāḥ || 56 ||

Words that have not themselves become objects do not illuminate the meaning; not by their mere existence, ungrasped, are they revealers of meanings.

The consequence of 1.55, stated as a denial with an anti-Mimamsa edge. For meaning to be conveyed, the word must first itself be grasped — visayatvam apannaih, having become an object of cognition. Na sattaya eva: not by mere being. A word existing unperceived in the vicinity conveys nothing; the lamp must itself be lit within awareness before it lights anything else. The point sounds obvious but has teeth. Any theory on which words signify by their sheer existence — by an unmediated relation between real sound and real object that bypasses the hearer's apprehension — is ruled out; signification is routed, ineliminably, through cognition. This completes the light-analogy of 1.55 in the only consistent way: what is self-luminous must still be *seen* to illuminate, and the priority of grasping-the-word over grasping-the-meaning is the sphota doctrine's account of what hearing is.

1.57 verified proof

अतोऽनिर्ज्ञातरूपत्वात् किमाहेत्यभिधीयते ।
नेन्द्रियाणां प्रकाश्येऽर्थे स्वरूपं गृह्यते तथा ॥ ५७ ॥

ato'nirjñātarūpatvāt kimāhetyabhidhīyate |
nendriyāṇāṃ prakāśye'rthe svarūpaṃ gṛhyate tathā || 57 ||

Hence, when the form has not been discerned, one asks 'what did he say?'; in the case of an object illuminated by the senses, its own form is not grasped in that way.

Empirical confirmation of 1.56, drawn from a familiar failure. 'What did he say?' is what we ask when sounds reached the ear but the word's rupa was not seized — proof that hearing sounds and grasping the word are distinct events, and that meaning waits on the second. The verse then draws a sharp disanalogy with sense-organs. The eye illuminates the pot without itself being grasped in the act — one does not see one's seeing — whereas the word cannot mediate unless it is itself apprehended. This blocks the tempting assimilation of words to instruments of cognition generally (an assimilation the karana-comparison of 1.42-47 might have invited): the word is an instrument of a special kind, one that works only from inside awareness. The grammarians' later slogan that the word is both pratyayaka and pratyayya, conveyor and conveyed, is this verse's point condensed.

1.58 verified proof

भेदेनावगृहीतौ द्वौ शब्दधर्मावपोद्धृतौ ।
भेदकार्येषु हेतुत्वमविरोधेन गच्छतः ॥ ५८ ॥

bhedenāvagṛhītau dvau śabdadharmāvapoddhṛtau |
bhedakāryeṣu hetutvamavirodhena gacchataḥ || 58 ||

The two properties of the word, grasped separately by abstraction, function as causes with respect to their distinct effects without conflict.

The methodological resolution of the two-powers doctrine. Grahyatva and grahakatva (1.55) are apoddhrta — extracted by analysis, the same term 1.24 used for the abstraction of words from sentences — and bhedena avagrhita, apprehended as separate. In the word itself they are not two components but one thing doubly functioning; the separation is the analyst's, made for the sake of tracking which effects flow from which aspect. So partitioned, they operate avirodhena, without mutual interference: the word's being-heard grounds one class of effects (its availability as a topic, its citation in grammar), its meaning-conveying grounds another (comprehension), and neither obstructs the other. The verse licenses the entire grammatical practice that follows in 1.59-69: rules can treat the word's own form as an item precisely because analysis may legitimately isolate what reality does not divide.

1.59 verified proof

वृद्धयादयो यथा शब्दाः स्वरूपोपनिबन्धनाः ।
आदैच्प्रत्यायितैः शब्दैः संबन्धं यान्ति संज्ञिभिः ॥ ५९ ॥

vṛddhayādayo yathā śabdāḥ svarūpopanibandhanāḥ |
ādaicpratyāyitaiḥ śabdaiḥ saṃbandhaṃ yānti saṃjñibhiḥ || 59 ||

Words such as 'vrddhi', founded upon their own form, enter into relation with the words conveyed by 'at' and 'aic' as the things they name.

The discussion descends, characteristically, into the machine-room of Panini. The Astadhyayi's very first rule, vrddhir adaic, defines a technical term: 'vrddhi' names the vowels a, ai, au — given in the rule by the pratyahara 'adaic'. Bhartrhari's analysis: the word 'vrddhi' in that rule stands sva-rupa-upanibandhana, resting on its own form (this is Panini's own meta-rule svam rupam sabdasya, 1.1.68: a word in grammar denotes its own form unless it is a technical name), and it relates as samjna, name, to the samjnins, the named items that 'adaic' conveys. The point of dragging the reader into this is strategic. The sva-rupa that ordinary usage never notices (1.54) is exactly what grammar traffics in; the discipline's daily practice presupposes the double luminosity of 1.50. Grammar is thus the standing institutional witness to the two-word doctrine.

1.60 verified proof

अग्निशब्दस्तथैवायमग्निशब्दनिबन्धनः ।
अग्निश्रुत्यैति संबन्धमग्निशब्दाभिधेयया ॥ ६० ॥

agniśabdastathaivāyamagniśabdanibandhanaḥ |
agniśrutyaiti saṃbandhamagniśabdābhidheyayā || 60 ||

In just the same way this word 'agni', founded upon the word 'agni' itself, enters into relation with the word-form 'agni' as what is expressed by it.

The same structure exhibited in a harder case, Panini 4.2.33 (agner dhak) being the traditional locus. When grammar teaches a suffix after 'agni', the item the rule mentions is not fire and not any particular utterance — it is the word-form itself, and the rule's mention of it stands to it as name to named. The verse's deliberately vertiginous phrasing — the agni-word, founded on the agni-word, related to the agni-sruti — dramatizes the reflexivity involved: language here bends back on itself, one occurrence functioning as designator and the form itself as designatum. That such self-mention works, and works systematically enough for an entire science to be built on it, is for Bhartrhari evidence that self-presentation (1.50) belongs to the word's nature and is not an exotic special case.
contested: which Paninian context the verse targets is not stated in the karika; the vrddhi example of 1.59 is explicit, but 1.60 generalizes it, and readers differ over whether a specific sutra or the general practice of upadana is meant

1.61 verified proof

यो य उच्चार्यते शब्दो नियतं न स कार्यभाक् ।
अन्यप्रत्यायने शक्तिर्न तस्य प्रतिबध्यते ॥ ६१ ॥

yo ya uccāryate śabdo niyataṃ na sa kāryabhāk |
anyapratyāyane śaktirna tasya pratibadhyate || 61 ||

Whatever word is uttered does not thereby necessarily partake in the operation; its power to convey something else is not thereby obstructed.

A rule of grammatical hermeneutics falling out of the two-powers doctrine. When a rule utters a word, that occurrence is not niyatam karyabhak — not automatically the item to which the taught operation applies; and conversely, its power of standing for something beyond itself is not cancelled by its being mentioned. Use and mention, in modern terms, are not marked on the word's face: the same utterance may present its own form or convey another, and which it does in a given sutra is settled by the economy of the system, not by the phonetics. The verse thereby answers a pedantic but real objection to 1.59-60 — 'if the rule's word denotes its own form, how can the operation reach other tokens?' — by denying that mention exhausts the word's powers. The two saktis of 1.55 coexist in every occurrence; context selects.

1.62 verified proof

उच्चरन् परतन्त्रत्वाद्गुणः कार्यैर्न युज्यते ।
तस्मात्तदर्थैः कार्याणां संबन्धः परिकल्प्यते ॥ ६२ ॥

uccaran paratantratvādguṇaḥ kāryairna yujyate |
tasmāttadarthaiḥ kāryāṇāṃ saṃbandhaḥ parikalpyate || 62 ||

In being uttered it is dependent, and so the subordinate element is not joined to the operations; therefore the relation of the operations is construed with what those elements mean.

A refinement on 1.61 through the case of subordinate elements. A guna here is the dependent member — say, a constituent inside a compound or a bare stem inside a larger unit. Uttered, it is paratantra, not standing on its own, and grammatical operations do not attach to it directly in that dependent condition; the relation of karya to item is parikalpyate, construed, through the artha, the function or value of the element within the whole. The verse registers a genuinely deep feature of Paninian grammar: rules are stated over abstract positions and values — stems, bases, meanings — rather than over surface tokens, whose dependence makes them unfit targets. It extends the lesson of 1.61 from 'mention does not exhaust powers' to 'utterance does not even fix the operand'.
contested: guna is ambiguous here between 'subordinate element' generally and the specific sense of a quality-word; our rendering follows the grammatical context, but the verse is compressed enough that a quality-based reading (qualities, being dependent on substances, take no operations directly) can be sustained

1.63 verified proof

सामान्यमाश्रितं यद् यदुपमानोपमेययोः ।
तस्य तस्योपमानेषु धर्मोऽन्यो व्यतिरिच्चते ॥ ६३ ॥

sāmānyamāśritaṃ yad yadupamānopameyayoḥ |
tasya tasyopamāneṣu dharmo'nyo vyatiriccate || 63 ||

Whatever common property is relied upon between the standard of comparison and what is compared, in each case some further property distinguishes the standard from it.

An apparent digression into the logic of comparison, whose relevance is methodological. Every upamana-upameya pair rides on a samanya, a shared property — and yet, the verse notes, the standard always retains a dharma the compared item lacks, else the two would not be two and comparison would collapse into identity. Residual disanalogy is thus not a defect of similes but a condition of their possibility. Placed here, after a run of similes (fire, reflection, painting, egg, light) and before further technical argument, the verse is Bhartrhari auditing his own method: each simile for the sphota-dhvani relation was pressed for one structural point and will mislead if pressed elsewhere — the reflection really is unreal, the flame really is produced, and the word is neither. A text unusually dependent on analogy here states the rule for reading analogies.

1.64 verified proof

गुणः प्रकर्षहेतुर्यः स्वातन्त्र्येणोपदिश्यते ।
तस्याश्रिताद् गुणादेव प्रकृष्टत्वं प्रतीयते ॥ ६४ ॥

guṇaḥ prakarṣaheturyaḥ svātantryeṇopadiśyate |
tasyāśritād guṇādeva prakṛṣṭatvaṃ pratīyate || 64 ||

The quality that is the ground of pre-eminence is indicated as though independent; from that quality, itself resting on its substrate, the pre-eminence of the substrate is understood.

Companion to 1.63, on the logic of degree. In 'whiter cloth', the comparative reaches the cloth through its whiteness: the guna is the prakarsa-hetu, the ground of the 'more', and grammar treats it svatantryena, as if it were a free-standing item, though in reality it exists only asrita, resting on its substrate. The pattern — analysis grants provisional independence to what is ontologically dependent, in order to state generalizations — is by now familiar: it is how the two powers were separated (1.58) and how words were extracted from sentences (1.24). Bhartrhari is assembling instances of a single methodological theme: the objects grammar quantifies over are artifacts of legitimate abstraction, and their artifact-status does not impugn the science, so long as one does not mistake the analytical ledger for an inventory of the world. Panini 5.3.55ff., on comparative and superlative suffixes, is the technical shadow behind the verse.

1.65 verified proof

तस्याभिधेयभावेन यः शब्दः समवस्थितः ।
तस्याप्युच्चारणे रूपमन्यत्तस्माद्विविच्यते ॥ ६५ ॥

tasyābhidheyabhāvena yaḥ śabdaḥ samavasthitaḥ |
tasyāpyuccāraṇe rūpamanyattasmādvivicyate || 65 ||

The word that stands settled as the expressible of that — in the uttering of it too, a form distinct from it is discerned.

The regress inside citation, noticed and disarmed. When grammar names a word (1.59-60), the named word is the abhidheya, the expressible; but to name it one must utter something, and the uttered token is other than the settled type it presents. So even in mention, the duality of 1.44 recurs one level up: an expressing occurrence and an expressed form, viviycate, discerned apart. The observation blocks a naive picture in which citation collapses the two-word doctrine — as though in mention, at least, sound and word were simply identical. They are not; the token heard in the classroom is a dated acoustic event, the form taught is none of them. Every act of language, including language about language, exhibits the same two-sidedness, which is what one would expect if the duality belongs to the word's essence rather than to some uses of it.

1.66 verified proof

प्राक्संज्ञिनाभिसंबन्धात् संज्ञा रूपपदार्थिका ।
षष्ठयाश्च प्रथमायाश्च निमित्तत्वाय कल्पते ॥ ६६ ॥

prāksaṃjñinābhisaṃbandhāt saṃjñā rūpapadārthikā |
ṣaṣṭhayāśca prathamāyāśca nimittatvāya kalpate || 66 ||

Before connection with the named, the name has the form itself as its meaning; and so it becomes fit to be the occasion for both the sixth case and the first.

A scholastic knot, treated with complete seriousness. In a defining rule like vrddhir adaic, the samjna before its connection to the samjnin is rupa-padarthika — its meaning is just the form. This priority explains a grammatical puzzle: why the tradition can speak both in the nominative ('vrddhi is a-ai-au') and in the genitive ('of these, the name is vrddhi'), the first presenting identity, the second presenting the name as belonging to the named. Both constructions are licensed, nimittatvaya kalpate, because the name has a standing of its own prior to the assignment — there is a moment, logical rather than temporal, at which the name is only itself, and the definitional act then relates two items each with its own claim to case. Trivial as the grammar looks, the philosophical cargo is the same as 1.65's: designation is a relation between two reals, not an absorption of one into the other.

1.67 verified proof

तत्रार्थवत्त्वात् प्रथमा संज्ञाशब्दाद्विधीयते ।
अस्येति व्यतिरेकश्च तदर्थादेव जायते ॥ ६७ ॥

tatrārthavattvāt prathamā saṃjñāśabdādvidhīyate |
asyeti vyatirekaśca tadarthādeva jāyate || 67 ||

There, because it possesses the meaning, the first case-ending is prescribed after the name-word; and the differentiation 'of this' arises precisely from that meaning.

The resolution of 1.66's alternative. The nominative in the defining rule is justified arthavattvat — the name-word, at the moment of definition, has its meaning (its own form) and so takes the case of independent standing (prathamā, per the domain of Panini 2.3.46). The genitive construal, asya iti, 'of this', is not thereby lost: it arises tadarthat eva, out of that same meaning, once the assignment has related name to named and the name can be viewed as the named item's property. The two cases thus express two logical moments of one act — the name in itself, the name as belonging — and the grammar's tolerance of both is exact, not sloppy. The discussion is a miniature of the whole kanda's method: an apparently pedantic alternation of case-endings, pursued far enough, yields the structure of the naming relation itself.

1.68-69 verified doxographic proof

स्वं रूपमिति कैश्चित्तु व्यक्तिः संज्ञोपदिश्यते ।
जातेः कार्याणि संसृष्टा जातिस्तु प्रतिपद्यते ॥ ६८ ॥
संज्ञिनीं व्यक्तिमिच्छन्ति सूत्रे ग्राह्यामथापरे ।
जातिप्रत्यायिता व्यक्तिः प्रदेशेषूपतिष्ठते ॥ ६९ ॥

svaṃ rūpamiti kaiścittu vyaktiḥ saṃjñopadiśyate |
jāteḥ kāryāṇi saṃsṛṣṭā jātistu pratipadyate || 68 ||
saṃjñinīṃ vyaktimicchanti sūtre grāhyāmathāpare |
jātipratyāyitā vyaktiḥ pradeśeṣūpatiṣṭhate || 69 ||

(68) But by some, it is the individual that is taught as the name in 'its own form'; the class, being connected, receives the operations of the class. (69) Others hold that it is the individual that is the named, to be grasped in the sutra; the individual, conveyed through the class, presents itself in the particular contexts.

The svam rupam debate of the Paninian tradition, reported in its two classical positions. When Panini 1.1.68 says a word in grammar stands for 'its own form', is the rupa in question the individual utterance-form (vyakti) or the form-universal (jati)? On the first view (68), the mentioned individual is the name, and operations generalize because the class is samsrsta, connected — riding in on the individual. On the second (69), the named is indeed the individual, but it is jati-pratyayita, conveyed through the class, and so upatisthate, presents itself, at each pradesa, each site of application. The dispute mirrors, inside grammatical semantics, the great akrti/vyakti controversy over what words denote (which the Padakanda will treat at length), and Bhartrhari reports it here without adjudication because both positions confirm his working point: rule-stated language requires an apparatus of types and tokens, however the metaphysical credit is distributed between them.
contested: the attribution and even the division of labour between 68 and 69 vary with how samsrsta and grahya are construed; we give the standard doxographic reading — vyakti-as-name versus vyakti-as-named-via-jati — but the compressed syntax admits rearrangement

1.70 · Unity and Plurality: Phoneme, Word, Sentence verified doxographic proof

कार्यत्वे नित्यतायां वा केचिदेकत्ववादिनः ।
कार्यत्वे नित्यतायां वा केचिन्नानात्ववादिनः ॥ ७० ॥

kāryatve nityatāyāṃ vā kecidekatvavādinaḥ |
kāryatve nityatāyāṃ vā kecinnānātvavādinaḥ || 70 ||

Whether they are produced or eternal, some are proponents of unity; whether they are produced or eternal, some are proponents of plurality.

A doxographic grid opening a new movement. Two questions cross: are words karya or nitya, produced or eternal (the dispute 1.28 deflated), and is the linguistic unit ekatva or nanatva, one or many? Bhartrhari's careful repetition of 'whether produced or eternal' in both lines makes the second question orthogonal to the first — unity-theorists and plurality-theorists exist on both sides of the eternality dispute, so the questions must be settled independently. The unity in question is the one the sphota doctrine asserts: is the word one partless entity variously manifested, or a plurality (of phonemes, of tokens) unified only in cognition? The next five verses set out the data any answer must save — the same phoneme across words, the same word across sentences — and the two candidate explanations. It is the kanda's most dialectical stretch: positions are laid out with a fairness that leaves Bhartrhari's own verdict genuinely disputable.

1.71 verified proof

पदभेदेऽपि वर्णानामेकत्वं न निवर्तते ।
वाक्येषु पदमेकं च भिन्नेष्वप्युपलभ्यते ॥ ७१ ॥

padabhede'pi varṇānāmekatvaṃ na nivartate |
vākyeṣu padamekaṃ ca bhinneṣvapyupalabhyate || 71 ||

Even when words differ, the unity of the phonemes is not abolished; and even in different sentences, one and the same word is found.

The datum for any theory of linguistic units: recurrence. The 'a' in one word and in another is not two similar sounds but, as competent speakers unhesitatingly judge, the same phoneme; the word 'gauh' in two different sentences is one word, not twins. This identity-across-occurrence is the most elementary fact a science of language must save, and it is already fatal to a crude token-physicalism, since the acoustic events are numerically distinct and unrepeatable. Whatever phonemes and words are, they are repeatables — the type, not the dated token, is the grammarian's object. The verse states the fact at two levels (phoneme across words, word across sentences) so that the coming argument about parts and wholes can run at both.

1.72 verified proof

न वर्णव्यतिरेकेण पदमन्यच्च विद्यते ।
वाक्यं वर्णपदाभ्यां च व्यतिरिक्तं न किश्चन ॥ ७२ ॥

na varṇavyatirekeṇa padamanyacca vidyate |
vākyaṃ varṇapadābhyāṃ ca vyatiriktaṃ na kiścana || 72 ||

And there exists no word over and above the phonemes; nor is the sentence anything whatsoever apart from the phonemes and words.

The analyst's thesis, stated in its strongest form — and it is a thesis Bhartrhari reports, not (on the reading this edition takes) his own final voice. If phonemes recur identically (1.71), a partitive mind concludes that they are the reals: the word is nothing vyatirekena, over and above, its phonemes, and the sentence nothing over phonemes-and-words — mere collections, with no additional entity anywhere. This is the position of the vaiyakarana atomist and, in a different key, of the Mimamsaka for whom the varnas alone are eternal and the 'word' a convenient summary. Read in sequence with 1.73, the two verses form a deliberate antinomy: this one dissolves wholes into parts, the next dissolves parts into wholes, and 1.74 will say openly that practice can be run from either side. The arrangement is the argument — each thesis, stated alone, sounds compelling; stated together they cancel, clearing the ground for the sphota account on which partition is a device of analysis rather than a report of structure.
contested: whether 1.72 is pure purvapaksa (an opponent's view to be superseded) or a moment Bhartrhari endorses at the level of manifested sound is a standing question for the whole passage 1.70-77; we read it as reported doctrine, but the karika carries no explicit 'iti' marking it as another's voice

1.73 verified proof

पदे न वर्णा विद्यन्ते वर्णेष्ववयवा न च ।
वाक्यात् पदानामत्यन्तं प्रविवेको न कश्चन ॥ ७३ ॥

pade na varṇā vidyante varṇeṣvavayavā na ca |
vākyāt padānāmatyantaṃ praviveko na kaścana || 73 ||

In the word the phonemes do not exist, and in the phonemes there are no parts; from the sentence there is no absolute separation of the words.

The counter-thesis, dissolving in the opposite direction. In the word — the real, functioning word, the meaning-bearer — phonemes na vidyante, are not found: analysis finds them, hearing-with-understanding does not, and the sphota that conveys meaning has no phonemic grain (1.48). Phonemes in turn have no avayavas, no parts, so the partitive regress finds no floor. And words have no atyantam praviveka, no absolute separation, from the sentence — the doctrine of 1.24 (words as apoddhara, extracts) now stated frontally, and the thesis the entire second kanda will defend: the sentence is the primary unit, its 'words' being artifacts of the same abstraction that finds 'phonemes' in words. Against 1.72's 'nothing but the parts', this verse answers 'the parts are nowhere to be found in the functioning whole'. Both cannot be reports of structure; at most one is, and 1.74 will suggest that neither is — that each is a darsana, a way of taking the phenomenon.

1.74 verified proof

भिन्नं दर्शनमाश्रित्य व्यवहारोऽनुगम्यते ।
तत्र यन्मुख्यमेकेषां तत्रान्येषां विपर्ययः ॥ ७४ ॥

bhinnaṃ darśanamāśritya vyavahāro'nugamyate |
tatra yanmukhyamekeṣāṃ tatrānyeṣāṃ viparyayaḥ || 74 ||

Practice is carried on by resort to differing views; what is primary for some is for others the reverse.

The antinomy of 1.72-73 is resolved — or rather, declined. Vyavahara, the working practice of grammar and speech, anugamyate, proceeds, on either darsana: one school takes the parts as mukhya, primary, and constructs wholes; the other takes the whole as primary and extracts parts; and each can run the entire practice from its chosen starting point, the other's primitive appearing in its ledger as derived. Bhartrhari's equanimity here is methodological, not indifferent. That practice underdetermines metaphysics — that the phenomena of 1.71 can be saved from either end — is itself the philosophically significant discovery, and it relocates the question: the choice between part-primacy and whole-primacy will not be settled by grammatical data, which both fit, but by which starting point better coheres with what language *does*, namely convey undivided meaning. On that criterion the whole wins, and the sphota doctrine is the working-out; but the verse is careful to register that the loser is not refuted, only outweighed.
contested: viparyaya can be read as 'the reverse view' (others take the opposite item as primary) or more strongly as 'error' (what is primary for some is for others a mistake); the first, doxographic reading fits the verse's even tone and we adopt it, but the polemical reading is grammatically available

1.75 verified proof

स्फोटस्याभिन्नकालस्य ध्वनिकालानुपातिनः ।
ग्रहणोपाधिभेदेन वृत्तिभेदं प्रचक्षते ॥ ७५ ॥

sphoṭasyābhinnakālasya dhvanikālānupātinaḥ |
grahaṇopādhibhedena vṛttibhedaṃ pracakṣate || 75 ||

Of the sphota, whose time is undivided, yet which appears to follow the time of the sounds, they declare a difference of function according to the difference of the conditions of its apprehension.

The reflection-simile of 1.49 is now given its doctrinal statement. The sphota is abhinna-kala, without temporal division — it does not itself take time — and yet it is dhvani-kala-anupatin, appearing to run along the timeline of the sounds that manifest it. The resolution: what differs is not the sphota but its vrtti, its mode of functioning, and that differs grahana-upadhi-bhedena, according to the conditions under which it is grasped — the speed of utterance, the voice, the accent, all properties of the manifesting medium transferred in appearance to the manifested word. This is the seed of the later distinction between prakrta-dhvani (the sound-pattern that determines which word is manifested) and vaikrta-dhvani (the delivery that varies freely without changing the word) — a distinction that lets the grammarian say precisely which acoustic differences matter to identity and which do not. The upadhi vocabulary is significant: as the one Brahman appears many through limiting conditions in the Vedantic picture, so the one word appears extended and varied through the conditions of its apprehension. Microcosm mirrors macrocosm once more.

1.76 verified proof

स्वभावभेदान्नित्यत्वे ह्रस्वदीर्घप्लुतादिषु ।
प्राकृतस्य ध्वनेः कालः शब्दस्येत्युपचर्यते ॥ ७६ ॥

svabhāvabhedānnityatve hrasvadīrghaplutādiṣu |
prākṛtasya dhvaneḥ kālaḥ śabdasyetyupacaryate || 76 ||

On the view that the word is eternal, the time of the primary sound — in short, long, prolated and the rest, according to its natural differences — is attributed figuratively to the word.

The prakrta-dhvani half of the distinction 1.75 seeded, now with its own name. Vowel quantity — hrasva, dirgha, pluta — is linguistically contrastive: 'a' and 'a-drawn-out' can distinguish words, so this temporal difference cannot be waved away as free variation. Yet the eternal word itself has no duration. Bhartrhari's solution is the figure of speech made technical: the kala belongs to the prakrta dhvani, the primary sound whose pattern determines *which* word is manifested, and is upacaryate — transferred, predicated figuratively — of the word. Quantity is thus real and criterial, but it is a property of the manifesting pattern, not of the manifested sphota. The verse shows the theory meeting its hardest phonological datum head-on: even the acoustic differences that *matter to identity* are lodged in the dhvani, with the sphota borrowing them only in the loose way the moon borrows the water's trembling.

1.77 verified proof

शब्दस्योर्ध्वमभिव्यक्तेर्वृत्तिभेदं तु वैकृताः ।
ध्वनयः समुपोहन्ते स्फोटात्मा तैर्न भिद्यते ॥ ७७ ॥

śabdasyordhvamabhivyaktervṛttibhedaṃ tu vaikṛtāḥ |
dhvanayaḥ samupohante sphoṭātmā tairna bhidyate || 77 ||

But after the manifestation of the word, the modified sounds bring in the difference of speed of delivery; the sphota-self is not divided by them.

The vaikrta half. Once the word is manifested — urdhvam abhivyakteh, 'above' or after manifestation — a second class of sound-properties comes into play: the vaikrta dhvanayah, modified sounds, which carry vrtti-bheda, the differences of delivery (the tradition enumerates drutā, madhyamā, vilambitā — fast, medium, slow). These affect nothing criterial: the same word said quickly is the same word. The division of labour completed across 1.76-77 is elegant and does real explanatory work. Contrastive sound-properties (quantity, the phonemic pattern) belong to the prakrta dhvani and determine which sphota is manifested; non-contrastive ones (tempo, voice) belong to the vaikrta dhvani and determine only how long the manifesting takes. The sphota itself — tair na bhidyate — is divided by neither. A modern reader will recognize the shape of the phonemic/phonetic distinction, discovered here from the manifestation side.

1.78 verified proof

इन्द्रियस्यैव संस्कारः शब्दस्यैवोभयस्य वा ।
क्रियते ध्वनिभिर्वादास्त्रयोऽभिव्यक्तिवादिनाम् ॥ ७८ ॥

indriyasyaiva saṃskāraḥ śabdasyaivobhayasya vā |
kriyate dhvanibhirvādāstrayo'bhivyaktivādinām || 78 ||

The sounds effect a preparation of the sense-organ alone, or of the word alone, or of both: these are the three positions among the proponents of manifestation.

Having committed to abhivyakti — the word is manifested, not produced — Bhartrhari must say what the sounds actually do, and he opens with a clean doxography. Manifestation requires that something be made ready: either the dhvanis prepare the *ear* (as collyrium prepares the eye, sharpening the receiver until the ever-present word registers), or they prepare the *word* (bringing the object itself into manifestable condition, as grinding releases fragrance), or both at once (as light readies both the eye it strikes and the pot it falls on). The three options exhaust the logical space once manifestation is analyzed as a relation between a receiver and an object, and the next two verses supply the perceptual precedents for each. The care taken here is anti-dogmatic: the sphota doctrine is not hostage to one mechanism, and Bhartrhari characteristically secures the doctrine by showing every available mechanism leads to it.

1.79 verified proof

इन्द्रियस्यैव संस्कारः समाधानाञ्जनादिभिः ।
विषयस्य तु संस्कारस्तद्गन्धप्रतिपत्तये ॥ ७९ ॥

indriyasyaiva saṃskāraḥ samādhānāñjanādibhiḥ |
viṣayasya tu saṃskārastadgandhapratipattaye || 79 ||

There is preparation of the sense-organ alone by concentration, collyrium and the like; but preparation of the object for the apprehension of its smell.

Precedents for the first two positions, drawn from accepted perceptual lore. That the receiver alone can be prepared is shown by samadhana and anjana — the yogin's concentration and the doctor's eye-salve improve the eye while leaving the object untouched. That the object alone can be prepared is shown by scent: crushing sandalwood readies the wood's own gandha for apprehension while doing nothing to the nose. Both patterns are familiar, so neither samskara-model is ad hoc when applied to sound and word. The verse's method deserves notice: the manifestation-theorist is assembling an empirical taxonomy of how hidden things come to cognition, exactly as a naturalist would, and the word is being placed within a general theory of manifestation rather than treated as a unique mystery.

1.80 verified proof

चक्षुषः प्राप्यकारित्वे तेजसा तु द्वयोरपि ।
विषयेन्द्रिययोरिष्टः संस्कारः स क्रमो ध्वनेः ॥ ८० ॥

cakṣuṣaḥ prāpyakāritve tejasā tu dvayorapi |
viṣayendriyayoriṣṭaḥ saṃskāraḥ sa kramo dhvaneḥ || 80 ||

If the eye acts by reaching its object, then by light the preparation of both — object and organ — is accepted; that is the pattern for sound.

The precedent for the third position, and the one the tradition tends to prefer for the word. On the prapyakaritva theory of vision — the sense reaches out to its object, a live doctrine among Naiyayikas and others — light plays a double role, readying the eye that operates by it and bathing the object that becomes visible in it. Sa kramo dhvaneh: that, says the verse, is the sounds' way — they condition the hearer's faculty and bring the word to manifestable condition at once. The double model fits the phenomenology of 1.84 (repetition ripens the hearer's grasp: receiver-side) and the criterial role of the prakrta pattern (the sounds determine which word appears: object-side), which is presumably why it stands last, in the position of the favoured view. Bhartrhari still does not legislate; the argument continues into 1.81 with the question of whether the dhvani is separately perceived at all.
contested: whether 1.80 endorses the both-samskara view or merely completes the doxography is not marked in the karika; its final position and the analogy's aptness suggest quiet preference, and we read it so

1.81 verified doxographic proof

स्फोटरूपाविभागेन ध्वनेर्ग्रहणमिष्यते ।
कैश्चिद्ध्वनिरसंवेद्यः स्वतन्त्रोऽन्यैः प्रकल्पितः ॥ ८१ ॥

sphoṭarūpāvibhāgena dhvanergrahaṇamiṣyate |
kaiściddhvanirasaṃvedyaḥ svatantro'nyaiḥ prakalpitaḥ || 81 ||

By some, the apprehension of the sound is accepted as undivided from the form of the sphota; by others the sound is construed as never itself sensed, or as independent.

The question beneath the mechanism: what, in hearing, is actually *heard*? Three answers. On the first, the dhvani is grasped only aviBhagena, in non-separation from the sphota — we never hear bare sound, only word-clothed-in-sound, the two arriving as one phenomenal package (as one sees the lamp's light only in seeing the lit room). On the second, the dhvani is asamvedya, never itself an object of awareness at all — a pure vehicle, doing its manifesting work entirely offstage, inferred by theorists but absent from experience. On the third it is svatantra, independent — separately audible, as when one hears a voice too distant to make out words: sound without sphota, so sound in its own right. The phenomenological dispute matters because it fixes what the 'transfer' of 1.49 and 1.76 amounts to: only on the first and third views is there something *in experience* whose properties could be misattributed to the word. Bhartrhari again reports without sentencing; his own account of learning and error (1.85-86) will draw on the resources of all three.

1.82 verified proof

यथानुवाकः श्लोको वा सोढत्वमुपगच्छति ।
आवृत्त्या न तु स ग्रन्थः प्रत्यावृत्ति निरूप्यते ॥ ८२ ॥

yathānuvākaḥ śloko vā soḍhatvamupagacchati |
āvṛttyā na tu sa granthaḥ pratyāvṛtti nirūpyate || 82 ||

As a Vedic lesson or a verse comes to be mastered through repetition, and yet that text is not discerned anew at each repetition —

First half of an argument-pair completed by 1.83-84, introducing the phenomenon that most embarrasses part-assembly theories: learning by repetition. A student masters an anuvaka by reciting it again and again; sodhatva, the being-mastered, dawns gradually across repetitions. Yet the text itself — sa granthah — is not a different object at each pass, na pratyavrtti nirupyate: the tenth recitation does not present a tenth text, but the same text, better possessed. So repetition improves the *grasp* while its object stays fixed — which is exactly the structure manifestation-theory attributes to hearing a word: the sounds' repeated strokes ripen apprehension of one unchanging sphota. The example is chosen from the tradition's daily pedagogy, where every reader has felt the process from inside.

1.83 verified proof

प्रत्ययैरनुपाख्येयैर्ग्रहणानुगुणैस्तथा ।
ध्वनिप्रकाशिते शब्दे स्वरूपमवधार्यते ॥ ८३ ॥

pratyayairanupākhyeyairgrahaṇānuguṇaistathā |
dhvaniprakāśite śabde svarūpamavadhāryate || 83 ||

— so, by cognitions that cannot be separately reported, yet are conducive to the final grasp, the word's own form is ascertained once the sound has illuminated it.

The apodosis. The intermediate cognitions in learning — each pass through the verse, each stroke of sound — are anupakhyeya, not separately reportable: the student cannot say what exactly the seventh recitation added, and the hearer of a word cannot report the deliverance of each phoneme as it lands. But they are grahana-anuguna, conducive to the grasp; their work is real though silent, and it terminates in avadharana, the settled ascertainment of the word's svarupa. This gives the sphota theory its account of process without surrendering the unity of the object: the *ripening* is gradual and many-staged, the *ripened grasp* seizes one undivided thing. The phenomenology is precise — comprehension does in fact arrive as an all-at-once click after a subpersonal accumulation, not as a felt summation of parts — and the next verse states the mechanism in its classical form.

1.84 verified proof

नादैराहितबीजायामन्त्येन ध्वनिना सह ।
आवृत्तपरिपाकायां बुद्धौ शब्दोऽवधार्यते ॥ ८४ ॥

nādairāhitabījāyāmantyena dhvaninā saha |
āvṛttaparipākāyāṃ buddhau śabdo'vadhāryate || 84 ||

The seed having been sown by the sounds, the word is ascertained in a cognition brought to ripeness by repetition, together with the final sound.

The classical statement of how the partless word is grasped through serial sounds — the verse later doxographers quote for the sphota theory's answer to its standing objection. Each dhvani deposits a bija, a seed or trace, in the hearer's buddhi; the sequence is a sowing, avrtta-paripaka, ripening through the repeated strokes; and antyena dhvanina saha, together with the last sound, the whole word is ascertained at once. The final phoneme is not a final part completing a sum — it is the occasion on which a cognition prepared by all the traces seizes the entire sphota. The seed-and-ripening vocabulary does deliberate double duty: it is agricultural, and it is the yogacara-adjacent idiom of traces in awareness, and both registers say the same thing — what accumulates is a readiness in the cognizer, not an assembly of the object. The Mimamsaka has a rival account (the last phoneme aided by memory-traces of the preceding ones conveys the meaning) that is formally parallel while dispensing with the sphota; the argument between the two schools turns on whether the *object* of the final cognition is a mere phoneme-series remembered, or a unity over and above it — and 1.85 gives Bhartrhari's phenomenological evidence for the latter.

1.85 verified proof

असतश्चान्तराले याञ्च्छन्दानस्तीति मन्यते ।
प्रतिपत्तुरशक्तिः सा ग्रहणोपाय एव सः ॥ ८५ ॥

asataścāntarāle yāñcchandānastīti manyate |
pratipatturaśaktiḥ sā grahaṇopāya eva saḥ || 85 ||

And when one thinks that the parts of the word exist in the interval, and that what is non-existent exists — that is the apprehender's incapacity; that, in fact, is the very means of the grasping.

The theory's boldest inversion: the appearance of phonemic parts is itself diagnosed as error — but productive error. In the antarala, the interval before the final ascertainment, the hearer takes part-cognitions to be cognitions of real parts of the word; this, says the verse, is asaktih, an incapacity of the apprehender, taking the non-existent as existent. And then the turn: sa grahanopaya eva sah — that very incapacity is the means of grasping. The mistake is not an obstacle the theory must explain away; it is the mechanism, the ladder the mind climbs. One cannot seize the undivided word directly, so cognition proceeds through a staged fiction of parts that ripens (1.84) into the true unitary grasp, whereupon the fiction has discharged its office. The structure — truth reached only through a scaffolding of error that the attained truth then retires — recurs at the grandest scale in this philosophy: the world's diversity is to Brahman what the phonemes are to the word. That parallel is not decoration; 1.86 immediately generalizes it.
contested: the double sah in grahanopaya eva sah is emphatic but syntactically loose; some construe 'that incapacity is precisely his means of grasping', others make the error-cognition itself the subject — the doctrine is unaffected but the sentence resists tidy parsing

1.86 verified proof

भेदानुकारो ज्ञानस्य वाचश्चोपप्लवो ध्रुवः ।
क्रमोपसृष्टरूपा वाग् ज्ञानं ज्ञेयव्यपाश्रयम् ॥ ८६ ॥

bhedānukāro jñānasya vācaścopaplavo dhruvaḥ |
kramopasṛṣṭarūpā vāg jñānaṃ jñeyavyapāśrayam || 86 ||

The imitation of differentiation by cognition, and the disturbance of speech, are constant; speech has a form contaminated by sequence, and cognition rests upon its objects.

The generalization: the productive error of 1.85 is not an occasional mishap but the permanent condition of mind and speech. Jnana imitates bheda — cognition, whose nature is undivided illumination, perpetually mimics the differentiation of what it grasps; vak suffers upaplava, disturbance, dhruvam, constantly — speech never appears except krama-upasrsta, contaminated by sequence. The two clauses are chiastic: cognition, undivided, borrows division from its objects; speech, whose essence (the sphota, ultimately the sabda-tattva of 1.1) is undivided, borrows division from its manifesting medium. Neither ever presents its true face in ordinary transaction. This is the kanda's epistemological low point, deliberately placed — if distortion is constant, how is truth ever touched? — and the answer has already been given in miniature: as in 1.85, the distortion is the upaya. The doctrine that ordinary cognition is irreducibly word-involved (and so sequence-involved, and so distortion-involved) returns in force at 1.123ff., where it becomes the master-thesis of the kanda.

1.87 verified proof

यथाद्यसंख्याग्रहणमुपायः प्रतिपत्तये ।
संख्यान्तराणां भेदेऽपि तथा शब्दान्तरश्रुतिः ॥ ८७ ॥

yathādyasaṃkhyāgrahaṇamupāyaḥ pratipattaye |
saṃkhyāntarāṇāṃ bhede'pi tathā śabdāntaraśrutiḥ || 87 ||

As the grasp of the first number is a means to the cognition of further numbers even though they differ, so is the hearing of one word a means to the cognition of another.

An analogy from arithmetic for a fact of linguistic learning: counting to one is how one learns to count at all. The grasp of adya-samkhya, the first number, is the upaya through which the numbers beyond it — genuinely different items — become cognizable; no one grasps 'seven' who has not first mastered the schema that 'one' inaugurates. So too sabdantara-sruti: hearing one word opens cognitive access to other words not yet heard. The point serves the manifestation theory at a specific joint. If words were learned one-by-one as brute associations, each would be an independent conquest; in fact linguistic competence is systematic — mastering some words confers a capacity for the rest, as mastering early numbers confers the number-series. A theory on which all words are variously-manifested configurations of one underlying word-nature (and ultimately of the one sabda-tattva) predicts this systematicity; a theory of atomistic pairings does not.
contested: the verse is compressed and its target unstated; we read it as an argument from systematicity of acquisition, but it can also be construed more narrowly, as explaining how the phoneme-schema learned in one word transfers to others — the arithmetic simile supports either

1.88 verified proof

र्ध्न्यंअप्रत्येकं व्यञ्जका भिन्ना वर्णवाक्यपदेषु ये ।
तेषामत्यन्तभेदे'पि सङ्कीर्णा इव शक्तयः ॥ ८८ ॥

rdhnyaṃapratyekaṃ vyañjakā bhinnā varṇavākyapadeṣu ye |
teṣāmatyantabhede'pi saṅkīrṇā iva śaktayaḥ || 88 ||

The manifestors, distinct for each among phoneme, word and sentence — even though these differ absolutely, their powers appear as though commingled.

A precision about levels. The theory recognizes sphotas at three grains — varna, pada, vakya — and each has its own vyanjakas, its own manifesting sounds; and the manifested items are atyanta-bhinna, absolutely distinct: the sentence-sphota is not a sum of word-sphotas any more than the word was a sum of phonemes (1.73). Yet in the single stream of speech these manifesting powers appear sankirna iva, as if commingled — one and the same stretch of sound seems simultaneously to be manifesting phonemes, words and the sentence, and the hearer cannot assign portions of the acoustic flow to levels. The iva is again load-bearing: the commingling is apparent, an artifact of one medium serving three manifestations at once. The verse matters structurally because it prevents a regress-objection — 'if the word is partless, what are the phoneme-sphotas doing in it?' Answer: nothing *in* it; they are distinct manifestables sharing manifestors, and hierarchy among sphotas is a hierarchy of alternative unities, not of containment. The second kanda's argument that the sentence is the true unit stands on exactly this ground.
contested: the wisdomlib transcription prefixes this karika's first line with the sequence 'rdhnyam a-', which yields no construable Sanskrit and is evidently a transcription artifact; we translate the evident text (pratyekam vyanjaka bhinnah...) and note the corruption rather than silently repairing the source

1.89 verified proof

यथैव दर्शनैः पूर्वैर्दूरात् सन्तमसेऽपि वा ।
अन्यथाकृत्य विषयमन्यथैवाध्यवस्यति ॥ ८९ ॥

yathaiva darśanaiḥ pūrvairdūrāt santamase'pi vā |
anyathākṛtya viṣayamanyathaivādhyavasyati || 89 ||

Just as, by its earlier perceivings — from afar, or in deep gloom — sight makes the object other than it is and determines it quite otherwise,

Protasis of a pair completed in 1.90, supplying the perceptual model for staged misapprehension. A figure approaches from a distance or looms in darkness: sight does not withhold judgement until conditions improve — it delivers a series of determinate but wrong verdicts (a post, then a man, then finally the traveller it is), each earlier darsana anyathakrtya, making the object otherwise, and adhyavasyati, determining it with full assertoric confidence. The progression to truth passes through committed error, not through cautious suspension. The example is the stock-in-trade of Indian error-theory (the post/man, the shell/silver), but its use here is distinctive: where the epistemologists deploy it to classify illusion, Bhartrhari deploys it to show that serial correction is a *normal mode of successful cognition* — the endpoint is veridical, and the errors were its road. The application to sentence-hearing follows immediately.

1.90 verified proof

व्यज्यमाने तथा वाक्ये वाक्याभिव्यक्तिहेतुभिः ।
भागावग्रहरूपेण पूर्वं बुद्धिः प्रवर्तते ॥ ९० ॥

vyajyamāne tathā vākye vākyābhivyaktihetubhiḥ |
bhāgāvagraharūpeṇa pūrvaṃ buddhiḥ pravartate || 90 ||

— so, while the sentence is being manifested by the causes of its manifestation, cognition first proceeds in the form of a seizing of parts.

The apodosis, and the sentence-level restatement of 1.85. As the sounds do their manifesting work, the hearer's buddhi runs ahead in bhaga-avagraha-rupa, the form of part-seizure — it takes words, phrases, apparent constituents, as so many determinate grasps, exactly as the distant-viewer takes the post and then the man. These part-graspings are the anyatha-determinations of 1.89: committed, useful, and destined to be superseded when the vakya-sphota is seized whole. The two verses together complete the theory's phenomenology of comprehension at sentence scale: hearing is a controlled hallucination of structure that converges on an undivided meaning-whole, the intermediate 'words' having the same standing in the final grasp that the intermediate 'man at a distance' has once the friend is recognized — scaffolding, retrospectively cancelled but prospectively indispensable. The second kanda will inherit this result as its starting point.

1.91 verified proof

यथानुपूर्वीनियमो विकारे क्षीरबीजयोः ।
तथैव प्रतिपत्तॄणां नियतो बुद्धिषु क्रमः ॥ ९१ ॥

yathānupūrvīniyamo vikāre kṣīrabījayoḥ |
tathaiva pratipattṝṇāṃ niyato buddhiṣu kramaḥ || 91 ||

As there is a fixed order of succession in the transformation of milk or of a seed, so is the sequence in the cognitions of comprehenders fixed.

A determinism about the stages. Milk does not become cheese except through curd; the seed reaches ear-of-grain through sprout and stalk, aniyama excluded — natural transformation has anupurvi-niyama, fixed serial order. So too the hearer's cognitions: the part-seizures of 1.90 are not a random walk that happens to end at the whole but a niyata krama, a law-governed ripening (the paripaka of 1.84 now given its natural-kind analogues). The point defends the theory against a charge of arbitrariness — if the intermediate graspings are errors, why do all competent hearers commit the *same* ones, in the same order, and converge on the same whole? Because manifestation is a causal process with the same fixity as any other maturation: same sounds, same prepared faculty, same staged path. Error with lawlike trajectory is process, not accident. The organic imagery also quietly reasserts the egg of 1.51: language-understanding belongs with growth, not with construction.

1.92 verified proof

भागवत्स्वपि तेष्वेव रूपभेदो ध्वनेः क्रमात् ।
निर्भागेष्वभ्युपायो वा भागभेदप्रकल्पनम् ॥ ९२ ॥

bhāgavatsvapi teṣveva rūpabhedo dhvaneḥ kramāt |
nirbhāgeṣvabhyupāyo vā bhāgabhedaprakalpanam || 92 ||

Among those very units, if they have parts, the difference of form arises from the sequence of the sound; if they are partless, then the construing of a difference of parts is a device.

A dilemma posed and both horns absorbed. Take the units — phoneme, word, sentence (tesv eva, those of 1.88). Either they have parts or they do not. If bhagavat, then their apparent internal form-differences (the 'g' seeming to precede the 'au' in gauh) arise dhvaneh kramat, from the sound's sequence — the ordering belongs to the manifesting stream, projected onto the unit (1.49, 1.76). If nirbhaga, partless, then bhaga-bheda-prakalpanam, the very positing of parts, is abhyupaya — a device, the analyst's and the hearer's working fiction (1.85's upaya, now named as method). Either way, no real internal sequence belongs to the linguistic unit itself: the first horn locates sequence in the medium, the second concedes it only as fabrication. This is the argumentative pincer that lets Bhartrhari be serene about the part-whole antinomy of 1.72-74 — the dispute over which decomposition is 'real' loses its object once both decompositions are seen to be projections, differently sourced, onto undivided manifestables.

1.93 verified proof

अनेकव्यक्त्यभिव्यङ्ग्या जातिः स्फोट इति स्मृता ।
कैश्चिद् व्यक्तय एवास्या ध्वनित्वेन प्रकल्पिताः ॥ ९३ ॥

anekavyaktyabhivyaṅgyā jātiḥ sphoṭa iti smṛtā |
kaiścid vyaktaya evāsyā dhvanitvena prakalpitāḥ || 93 ||

The sphota is held by tradition to be the universal, manifested by its several individuals; by some, its individuals are construed as the sounds.

A rival systematization from within the school, reported with its refinement. On this view the sphota is a jati, a universal: gauh-said-by-Devadatta and gauh-said-by-Yajnadatta are vyaktis, individual instances, and the word itself is the type they instantiate — anekavyakti-abhivyangya, manifested by its many particulars, as cowhood is by cows. The refinement (kaiscit): the vyaktis in question just *are* the dhvanis, so that the sound-word relation becomes a special case of the particular-universal relation, and manifestation of the sphota by sounds becomes the familiar manifestation of a jati by its instances. The economy is attractive — one metaphysical relation instead of two — and this jati-sphota reading became a standing position in the later school (with vyakti-sphota views opposed to it). Bhartrhari records it as smrta, received doctrine, without making it his own statement; his own emphasis throughout has been on the sphota as *undivided meaning-bearer*, a functional characterization neutral between type-theoretic construals. The next two verses take up what the jati-reading must answer: how eternal universals can have transient, law-governed manifestation.
contested: whether 1.93 reports a view Bhartrhari endorses, tolerates, or merely catalogues is a classical crux; smrta suggests received authority, but the kaiscit of the second line marks at least the refinement as partisan, and we read the whole as doxography

1.94 verified proof

अविकारस्य शब्दस्य निमित्तैर्विकृतो ध्वनिः ।
उपलब्धौ निमित्तत्वमुपयाति प्रकाशवत् ॥ ९४ ॥

avikārasya śabdasya nimittairvikṛto dhvaniḥ |
upalabdhau nimittatvamupayāti prakāśavat || 94 ||

Of the unchanging word, the sound — modified by causes — comes to be the condition for apprehension, like a light.

The manifestation relation restated with its cleanest analogy, in answer to the standing worry: how can the changing reveal the changeless without infecting it? The dhvani is vikrta nimittaih — modified by its causes: the speaker's organs, effort, pace — while the sabda is avikara, unchanging. The sound's role is nimittatvam upalabdhau, being the condition of apprehension: prakasavat, like a lamp. A lamp may be bright or dim, steady or guttering; these vicissitudes affect how well the pot is seen, perhaps whether it is seen — they do not put flicker into the pot. So the sound's modifications condition the apprehending (speed, clarity, audibility) without transferring change to the word apprehended; where transfer *appears* to occur, 1.49 and 1.76 have already diagnosed it as figurative. The lamp analogy also silently answers 1.78's mechanism question in favour of the both-sides model of 1.80: light conditions eye and object at once. That the light-comparison recurs at every load-bearing joint of the kanda (1.46, 1.55, here) reflects its unique fitness: illumination is the one ordinary relation in which a mediator reveals without constituting.

1.95 verified proof

न चानित्येष्वभिव्यक्तिर्नियमेन व्यवस्थिता ।
आश्रयैरपि नित्यानां जातीनां व्यक्तिरिष्यते ॥ ९५ ॥

na cānityeṣvabhivyaktirniyamena vyavasthitā |
āśrayairapi nityānāṃ jātīnāṃ vyaktiriṣyate || 95 ||

Nor is manifestation settled by rule only for non-eternal things; manifestation of the eternal universals by their substrates is likewise accepted.

An objection dissolved by precedent. The objector holds that abhivyakti is at home only among anitya things — the pot manifested by the lamp is itself a transient item, and manifestation-relations, being causal and dated, seem to belong to the world of the produced; how can the *eternal* be manifested without being dragged into time? The reply is that the objector's own ontology already contains the counterexample: universals, nitya by everyone's admission, are manifested (vyaktir isyate) by their asrayas, the substrates in which they inhere — cowhood becomes apprehensible exactly where a cow stands, though cowhood neither arrives nor departs. Dated, located manifestation of the undated is thus a relation the realist schools (Nyaya-Vaisesika above all) are committed to independently. The sphota theorist claims nothing stranger: sounds stand to the eternal word as the cow stands to cowhood — occasioning its apprehension here and now without lending it their transience. The argument is dialectically tidy: it does not prove the word eternal, it removes the alleged incoherence in saying so.

1.96 verified proof

देशादिभिश्च संबन्धो दृष्टः कायवतामपि ।
देशभेदविकल्पेऽपि न भेदो ध्वनिशब्दयोः ॥ ९६ ॥

deśādibhiśca saṃbandho dṛṣṭaḥ kāyavatāmapi |
deśabhedavikalpe'pi na bhedo dhvaniśabdayoḥ || 96 ||

Connection with place and the rest is observed even for embodied things; even under the alternative of difference of place, there is no difference between sound and word.

The companion objection: manifestation seems to localize the word — the utterance happens *here*, so is the eternal word here rather than there? The first line defuses the assumption that location-talk tracks real division: even kayavat things, bodies, stand in place-connections that multiply without multiplying them (one body, many successive places; one fire, many hearths — the traditional illustrations run this way). The second line applies it: under desa-bheda-vikalpa, the alternative that utterances occur in different places, there is still no bheda between dhvani and sabda beyond what the theory already grants — the sounds are many and located; the word they manifest is not parcelled among the locations. As with time in 1.76, so with place: spatial predicates attach to the word upacarena, by transfer from its manifestors. The pair 1.95-96 thus closes the classical trio of worries — change, time, place — and with them the manifestation doctrine's defensive perimeter is complete.
contested: the illustrative force of kayavatam api is compressed; we take it as 'even bodies multiply place-connections without internal division', but some construe it as conceding location to embodied things precisely to contrast the word's non-location — the verse's conclusion is the same on either construal

1.97 verified proof

ग्रहणग्राह्ययोः सिद्धा योग्यता नियता यथा ।
व्यङ्ग्यव्यञ्जकभावेन तथैव स्फोटनादयोः ॥ ९७ ॥

grahaṇagrāhyayoḥ siddhā yogyatā niyatā yathā |
vyaṅgyavyañjakabhāvena tathaiva sphoṭanādayoḥ || 97 ||

As the fitness between an apprehension and its object is established and fixed, just so is that between sphota and sounds, in the relation of manifested and manifestor.

The relation itself, finally named as primitive. Why do *these* sounds manifest *this* word? The answer offered is yogyata, fitness — the same fixed, unexplained aptitude that pairs sight with colour and never with sound. No one asks in virtue of what the eye is fit for the visible; that fitness is siddha and niyata, established and fixed, a bedrock of the cognitive order. The vyanya-vyanjaka relation between word and its sounds has, Bhartrhari claims, the same standing: constitutive, regular, not further derivable. The move is philosophically self-aware — every theory must stop somewhere, and stopping at a fitness already admitted everywhere in epistemology is cheaper than inventing new machinery. It also quietly explains the *rule-boundness* of manifestation (only the right phoneme-pattern manifests gauh) without conceding that the word is assembled from that pattern: fit, not composition.

1.98 verified proof

सदृशग्रहणानां च गन्धादीनां प्रकाशकम् ।
निमित्तं नियतं लोके प्रतिद्रव्यमवस्थितम् ॥ ९८ ॥

sadṛśagrahaṇānāṃ ca gandhādīnāṃ prakāśakam |
nimittaṃ niyataṃ loke pratidravyamavasthitam || 98 ||

And in ordinary experience, the illuminating condition even of things similarly apprehended — smells and the rest — is fixed, standing distinct for each substance.

Support for 1.97 from the economy of perception. Smells resemble one another; yet the nimitta that releases each — the crushing of this sandal, the burning of that resin — is niyata and pratidravyam avasthita, fixed and lodged substance by substance. Similar manifestables, dedicated manifestors: nature is full of such assignments, and no one finds them mysterious enough to demand a compositional analysis (nobody thinks the fragrance is *made of* crushings). So likewise similar words each have their own dedicated sound-patterns, and the dedication is no argument that the pattern constitutes the word. The example continues the naturalist's method of 1.79-80: the word-sound relation is being normalized, placed in a family of manifestation-relations the opponent already accepts, so that scepticism about it becomes scepticism about perception at large.

1.99 verified proof

प्रकाशकानां भेदांश्च प्रकाश्योऽर्थोऽनुवर्तते ।
तैलोदकादिभेदे तत् प्रत्यक्षं प्रतिबिम्बके ॥ ९९ ॥

prakāśakānāṃ bhedāṃśca prakāśyo'rtho'nuvartate |
tailodakādibhede tat pratyakṣaṃ pratibimbake || 99 ||

And the illuminated object conforms to the differences of its illuminators — that is directly seen in a reflection when the oil, water and the like differ.

The transfer-doctrine of 1.49 receives its everyday proof. One face, reflected in water, in oil, in a sword-blade: the pratibimba differs — dim, dark, elongated — as the medium differs, and the differences are patently the medium's contribution, since the face has not changed. Prakasya anuvartate prakasakanam bhedan: the revealed conforms to its revealers. Pratyaksam: this is not speculation but something seen. Applied: the word heard through a child's voice, a shout, a drawl 'conforms' likewise — its apparent duration, pitch and colour are the sounds' vicissitudes worn borrowed. The verse arms the theory with a lived, checkable case where everyone already distinguishes the thing from its medium-conditioned appearance, and 1.100 will push the same example to its limit: media can even misrepresent *size*.

1.100 verified proof

विरुद्धपरिमाणेषु वज्रादर्शतलादिषु ।
पर्वतादिसरूपाणां भावानां नास्ति संभवः ॥ १०० ॥

viruddhaparimāṇeṣu vajrādarśatalādiṣu |
parvatādisarūpāṇāṃ bhāvānāṃ nāsti saṃbhavaḥ || 100 ||

In mirrors of contrary measure — a gem-facet, a blade-surface and the like — there is no possibility of entities really of like form with the mountain reflected.

The reductio that secures 1.99. A mountain shows in the facet of a jewel. If reflections were real duplicates — entities sarupa with their originals — the facet would have to contain a mountain-sized thing, which is absurd: viruddha-parimana, the measures contradict. So a reflection is not a second entity resembling the first; it is the original *as appearing under* an alien condition, and its size-in-the-mirror is a fact about the mirroring, not about anything's real dimensions. Applied to the word: the sphota 'in' a quick utterance is not a short word-duplicate, nor a long one in slow speech — there are no duplicates at all, only the one word under different manifesting conditions (1.75-77). The mirror-mountain is the strongest of the reflection-series because it makes quantitative transfer patently unreal, and duration was exactly the quantitative property whose transfer to the word (1.76) most needed the demonstration.

1.101 verified proof

तस्मादभिन्नकालेषु वर्णवाक्यपदादिषु ।
वृत्तिकालः स्वकालश्च नादभेदाद्विभज्यते ॥ १०१ ॥

tasmādabhinnakāleṣu varṇavākyapadādiṣu |
vṛttikālaḥ svakālaśca nādabhedādvibhajyate || 101 ||

Therefore, for phonemes, sentences, words and the rest, whose time is undivided, the time of functioning and their own time are distinguished according to the difference of the sound.

Tasmat — the conclusion of the whole movement since 1.75. All linguistic units, at every grain, are abhinna-kala, temporally undivided in themselves; the felt durations belong to nada. Hence a final clean distinction: vrtti-kala, the time of functioning — how long the manifesting takes, a real property of the sound-process — versus sva-kala, the unit's 'own time', which exists only as the figurative shadow the vrtti casts (the upacara of 1.76). Long and short vowels, quick and slow recitations, differ in vrtti-kala; no sphota differs from any other in sva-kala, since none has any. The verse closes the temporal dossier with the theory in equilibrium: sequence and duration fully acknowledged as data, fully assigned to the medium, the meaning-bearer standing outside time as 1.48 required. What remains for the kanda is doxography (1.102-113), the cosmological deepening (1.114-118), and the great theses on knowledge and word (1.123ff.).

1.102 · Rival Accounts of Sound and Word verified doxographic proof

यः संयोगविभागाभ्यां करणैरुपजन्यते ।
स स्फोटः शब्दजाः शब्दा ध्वनयोऽन्यैरुदाहृताः ॥ १०२ ॥

yaḥ saṃyogavibhāgābhyāṃ karaṇairupajanyate |
sa sphoṭaḥ śabdajāḥ śabdā dhvanayo'nyairudāhṛtāḥ || 102 ||

What is generated by the organs through conjunction and disjunction — that is the sphota; the sounds born from that word are what others call the dhvanis.

A terminological about-face, reported deadpan: for *some* theorists (the tradition identifies phonetician-grammarians whose view survives here and in the Mahabhasya's margins), 'sphota' names the *first-produced sound* — the one generated by the articulators' contacts and releases (samyoga-vibhaga, the standard physics of sound-production) — while 'dhvani' names the sabda-ja sabdah, the sounds born *from* that sound: the spreading, echoing continuum that carries it to distant ears. On this usage the sphota is itself acoustic and produced; what Bhartrhari has been calling sphota — the partless meaning-bearer — is not in this picture at all. He reports the usage without protest because half of his own apparatus survives translation into it: the first-sound fixes identity (prakrta), the propagated train varies freely (vaikrta, as 1.103 will say in this idiom). But the metaphysical core does not survive, and the sequence 1.102-113 is precisely a tour of what theories look like when the undivided word is left out.
contested: whose view 1.102 reports is uncertain; candidates run from early phoneticians to the audumbarayana-line of speculation, and the karika's anonymous 'anyaih' leaves it open — what is clear is that the terminology here inverts Bhartrhari's own

1.103 verified doxographic proof

अल्पे महति वा शब्दे स्फोटकालो न भिद्यते ।
परस्तु शब्दसन्तानः प्रचयापचयात्मकः ॥ १०३ ॥

alpe mahati vā śabde sphoṭakālo na bhidyate |
parastu śabdasantānaḥ pracayāpacayātmakaḥ || 103 ||

Whether the word is small or great, the time of the sphota does not differ; but the subsequent train of sounds has the nature of increase and decrease.

The produced-sphota theory of 1.102 developing its own version of the invariance thesis. Even with the sphota reduced to the first acoustic event, a distinction imposes itself: that first event's time 'does not differ' with loudness — alpa or mahat, whispered or bellowed, the initial articulated form is what it is — while the para sabda-santana, the after-train, waxes and wanes (pracaya-apacaya) with the force of utterance, which is why a shout carries and a whisper dies. Loudness, on this analysis, is a property of propagation, not of articulation. Bhartrhari's interest in reporting it is transparent: even theorists who deny the transcendent word are driven to split the invariant from the variable within speech — the structure of prakrta versus vaikrta (1.76-77) reappears inside pure acoustics. The rival theory thus testifies, despite itself, that no account of language survives without an invariance-core; the schools differ only over how deep the invariant lies.
contested: sphota-kala here must mean the duration of the initially-produced sound if 1.102's terminology is still in force, and we read the verse as continuing that reported view; a reader who takes 1.103 as Bhartrhari's own voice must instead take it as restating 1.75-76, which the 'para santana' clause fits less well

1.104 verified proof

दूरात् प्रभेव दीपस्य ध्वनिमात्रं तु लक्ष्यते ।
घण्टादीनां च शब्देषु व्यक्तो भेदः स दृश्यते ॥ १०४ ॥

dūrāt prabheva dīpasya dhvanimātraṃ tu lakṣyate |
ghaṇṭādīnāṃ ca śabdeṣu vyakto bhedaḥ sa dṛśyate || 104 ||

From afar, only the sound is discerned, like the glow of a lamp; and in the notes of bells and the like, that difference is plainly seen.

Empirical vindication of the sphota-dhvani split from two everyday observations. First: a voice heard across a valley — one hears *that* someone speaks without hearing *what*: dhvani-matram, bare sound, reaching the ear while the word does not, exactly as a lamp's prabha announces the lamp from a distance at which the flame itself cannot be made out. The manifestor, then, is separable in experience from what it manifests — settling against 1.81's second position (that the dhvani is never itself sensed). Second: the bell. Struck once, it gives its note and then the long after-hum; hearers effortlessly distinguish the strike-sound from its lingering resonance — vyakto bhedah, the difference is patent. That is the alpa/mahat train of 1.103 made audible in a single event: initial form, then propagating tail. Both observations are chosen for being no one's theory: they are data any account must accommodate, and they already contain the distinction the sphota-theorist needs.
contested: which distinction the bell 'plainly shows' — strike versus resonance (our reading), or the sphota/dhvani distinction proper carried into non-linguistic sound — turns on how far 1.102's reported terminology still governs; the example works either way, which may be why it was chosen

1.105 verified proof

द्रव्याभिघातात् प्रचितौ भिन्नौ दीर्घप्लुतावपि ।
कम्पे तूपरते जाता नादा वृत्तेर्विशेषकाः ॥ १०५ ॥

dravyābhighātāt pracitau bhinnau dīrghaplutāvapi |
kampe tūparate jātā nādā vṛtterviśeṣakāḥ || 105 ||

From the striking of the substance, in the amplification even the long and prolated arise distinct; but the sounds born when the vibration has ceased are the differentiators of delivery.

The physics pressed one step further, to apportion vowel-quantity itself. Dravya-abhighata, the striking of the sounding body — articulators on air, hammer on bell — yields the pracaya, the amplified initial sound, and within *that*, dirgha and pluta arise as distinct: quantity, being criterial (1.76), belongs to the primary production. But the nadas born kampe uparate, once the vibration has ceased — the after-resonances, the bell's hum, the voice's trail — are vrtter visesakah, differentiators only of delivery: they mark how the utterance decays, never which word it was. The verse thus locks the earlier division (1.76-77: prakrta bears contrast, vaikrta bears style) onto an explicit acoustic mechanism: contrastive properties ride the driven phase of the sound, stylistic ones ride its free decay. As physics of vibration this is remarkably observant — driven versus decaying oscillation is a real and fundamental distinction — and as linguistics it completes the assignment of every temporal property of speech to a definite layer of the medium.
contested: pracitau is difficult — we take it as locative of pracaya, 'in the amplification', following 1.103's vocabulary, but it has been construed as dual ('the two accumulated ones') with dirgha-pluta, changing the syntax though not the doctrine

1.106 verified doxographic proof

अनवस्थितकम्पेऽपि करणे ध्वनयोऽपरे ।
स्फोटादेवोपजायन्ते ज्वाला ज्वालान्तरादिव ॥ १०६ ॥

anavasthitakampe'pi karaṇe dhvanayo'pare |
sphoṭādevopajāyante jvālā jvālāntarādiva || 106 ||

Even while the vibration of the organ persists unsettled, further sounds are generated from the sphota itself, as flame from another flame.

The propagation mechanism, with a simile that repays pressing. While the karana still trembles — the articulation not yet settled — apare dhvanayah, further sounds, arise sphotad eva: from the first-produced sound itself (1.102's usage), not from fresh strokes of the organ. Each ring of sound begets the next outward ring, jvala jvalantarad iva, as one flame kindles another — the torch passed, not the torch stretched. The model is serial regeneration: sound crosses space by continually re-producing itself in the medium, each phase the cause of its successor. As acoustics, this is the wave-picture in the only idiom available — local excitation propagating by self-replication, the sounding source needed only to seed it. Doctrinally the verse explains how the word reaches far-off hearers without the original articulation reaching them: what arrives is a great-grandchild of the first sound, faithful in form (flame lights flame without loss of fire-nature), degraded only in the free properties — which is why the distant hearer of 1.104 gets dhvani-matra, the form's carrier worn thin.
contested: sphotad eva here reads naturally only under 1.102's reported terminology (sphota = first-produced sound); a reader taking sphota in Bhartrhari's own sense must treat the generation as manifestation loosely phrased, and the flame simile fits production better than manifestation — evidence that 1.102-106 form one reported block

1.107 verified doxographic proof

वायोरणूनां ज्ञानस्य शब्दत्वापत्तिरिष्यते ।
कैश्चिद्दर्शनभेदो हि प्रवादेष्वनवस्थितः ॥ १०७ ॥

vāyoraṇūnāṃ jñānasya śabdatvāpattiriṣyate |
kaiściddarśanabhedo hi pravādeṣvanavasthitaḥ || 107 ||

By some, it is air, by some the atoms, by some cognition that is held to attain the condition of the word; for the divergence of views among the doctrines is unsettled.

The doxographic frame widens to first principles: what *is* the stuff that 'becomes' word? Three candidate substrates, each with its school: vayu, air — the phonetician's answer, speech as wind shaped (1.108-109 will unfold it); anavah, atoms — an atomist refinement in which sound-particles accumulate (1.110-111); and jnana, cognition — the idealist answer, word as a modification of awareness itself (1.112-113, where Bhartrhari's own sympupathies will surface). The second line is a shrug with method in it: darsana-bhedo hi pravadesv anavasthitah — the divergence among doctrines settles nowhere. Bhartrhari's tolerance is again strategic. His own doctrine lives at a level these disputes do not touch: whatever the substrate-story, the manifested word's undividedness, its transfer-properties, its yogyata with sounds (1.97) stand unaffected — exactly as 1.28 declined the nitya/karya fight and 1.74 the part/whole fight. The catalogue that follows is thus not indecision but a demonstration that the sphota doctrine is compatible with every physics on offer, and hostage to none.

1.108-109 verified doxographic proof

लब्धक्रियः प्रयत्नेन वक्तुरिच्छानुवर्तिना ।
स्थानेष्वभिहतो वायुः शब्दत्वं प्रतिपद्यते ॥ १०८ ॥
तस्य कारणसामर्थ्याद्वेगप्रचयधर्मणः ।
सन्निपाताद्विभज्यन्ते सारवत्योऽपि मूर्तयः ॥ १०९ ॥

labdhakriyaḥ prayatnena vakturicchānuvartinā |
sthāneṣvabhihato vāyuḥ śabdatvaṃ pratipadyate || 108 ||
tasya kāraṇasāmarthyādvegapracayadharmaṇaḥ |
sannipātādvibhajyante sāravatyo'pi mūrtayaḥ || 109 ||

(108) Air, set in motion by the effort that follows the speaker's wish, striking upon the places of articulation, attains the condition of the word. (109) Through the potency of its cause, possessing the properties of velocity and accumulation, it divides by its impact even bodies that possess solidity.

The vayu doctrine given its full statement, and it is a mechanist's credo. The causal chain runs unbroken from mind to matter: iccha, the wish; prayatna anuvartin, effort conforming to it; air labdha-kriya, endowed with motion; abhihata sthanesu, striking the eight classical places of articulation — and the shaped collision *is* the word: sabdatvam pratipadyate, air attains wordhood. Nothing non-physical intervenes between intention and utterance except intention itself. 109 then insists the agent is robust enough for the role: this moving air carries vega and pracaya, velocity and mass-accumulation, from the samarthya of its cause, and at full sannipata it splits even solid bodies — the gale that breaks trees is the same stuff that, finely governed, becomes a vowel. The argument-by-escalation answers the obvious sneer (how can thin breath be the mighty word?) with observed pneumatic violence. Bhartrhari reports the view fairly and without adoption; its inadequacy for him lies not in the physics but in what the physics cannot reach — air shaped at the lips has still to be *grasped as one word* by a hearer, and the whole dossier of 1.82-92 waits on the far side of any collision.
contested: sāravatyo 'pi mūrtayah — 'even solid bodies' — is our construal of saravat as 'possessing pith/solidity'; the line has also been read as 'even the solid forms (of the articulators) are divided', making the wind shape itself by cutting against the organs, which fits the phonetic context more closely though it strains vibhajyante

1.110-111 verified doxographic proof

अणवः सर्वशक्तित्वाद्भेदसंसर्गवृत्तयः ।
छायातपतमःशब्दभावेन परिणामिनः ॥ ११० ॥
स्वशक्तौ व्यज्यमानायां प्रयत्नेन समीरिताः ।
अभ्राणीव प्रचीयन्ते शब्दाख्याः परमाणवः ॥ १११ ॥

aṇavaḥ sarvaśaktitvādbhedasaṃsargavṛttayaḥ |
chāyātapatamaḥśabdabhāvena pariṇāminaḥ || 110 ||
svaśaktau vyajyamānāyāṃ prayatnena samīritāḥ |
abhrāṇīva pracīyante śabdākhyāḥ paramāṇavaḥ || 111 ||

(110) The atoms, because they possess all powers, functioning by division and by combination, transform themselves as shadow, light, darkness and word. (111) When their own power is being manifested, impelled by effort, the atoms called 'word' accumulate like clouds.

The atomist refinement, stated in two strokes. Where the vayu theory made word a *state* of one continuous stuff, this view grinds the substrate finer: anavah, atoms, sarva-saktitvat — omnipotent in the small, each carrying every power — take on word-being as one parinama among siblings: chaya, atapa, tamas, sabda. The company the word keeps in that list is deliberate and strange: shadow, glare, darkness — the quasi-substances of Indian physics, borderline entities that are somehow *there* without ordinary thinghood. Word belongs, for this school, to precisely that twilight ontology. 111 gives the dynamics: under prayatna the sabda-akhyah paramanavah — 'the atoms called word', a name conceding that wordhood is a functional designation, not a natural kind — pracyante abhranive, mass like clouds: scattered potency condensing into audible weather. The cloud-simile earns its place: clouds are aggregates with edges only from a distance, form without unity — which is exactly the standing this theory can give the word, and exactly why it cannot be Bhartrhari's: a cloud of word-atoms has no principled oneness for the hearer's single grasp (1.84) to seize. The purvapaksa is thus allowed to paint its own limitation.
contested: bheda-samsarga-vrttayah we take as 'functioning by division and combination' (the atoms' two modes); it has also been construed as 'whose functioning involves difference-and-contact with one another', a physics of collision — the cosmology is unaffected

1.112-113 verified proof

अथायमान्तरो ज्ञाता सूक्ष्मे वागात्मनि स्थितः ।
व्यक्तये स्वस्य रूपस्य शब्दत्वेन विवर्तते ॥ ११२ ॥
स मनोभावमापद्य तेजसा पाकमागतः ।
वायुमाविशति प्राणमथासौ समुदीर्यते ॥ ११३ ॥

athāyamāntaro jñātā sūkṣme vāgātmani sthitaḥ |
vyaktaye svasya rūpasya śabdatvena vivartate || 112 ||
sa manobhāvamāpadya tejasā pākamāgataḥ |
vāyumāviśati prāṇamathāsau samudīryate || 113 ||

(112) Now this inner knower, standing in the subtle self-of-speech, turns itself, for the manifestation of its own form, into wordhood. (113) Attaining the condition of mind, ripened by the inner fire, it enters the breath-wind; and then it is uttered forth.

The third doctrine, and the temperature changes: the anonymous reportage of 1.108-111 gives way to vocabulary that is Bhartrhari's own. The subject now is the antarah jnata, the inner knower, already standing suksme vagatmani, in a subtle speech-self — language is not something the knower produces but a dimension it inhabits — and the verb is vivartate (112d): it *appears as* wordhood, the same appearance-without-transformation that 1.1 predicated of Brahman and 1.48 of the sphota. And the purpose: vyaktaye svasya rupasya, for the manifestation of its own form — speech as the self's self-disclosure, not as signal-engineering. 113 then supplies what neither rival could: a descent-path connecting this metaphysics to their physics. The word-nature attains mano-bhava (mind-condition), is ripened tejasa — by the inner heat, the old physiological fire of the Upanisads — enters prana, and is samudirya, uttered forth. Mind, fire, breath: the idealist keeps the whole pneumatic chain of 1.108 as the *last* stage of a process that begins in awareness. The three-stage descent — knower's subtle speech, mental word, breath-borne utterance — is the embryo of the pasyanti/madhyama/vaikhari hierarchy made explicit at 1.142, and from here to the kanda's end the reported view and the author's voice are one.
contested: whether 1.112-113 still belong to the doxography of 1.107 (as the 'jnana' candidate) or mark Bhartrhari's own resumption is a real question; the vivarta vocabulary and the seamless continuation into 1.114-118 argue for the latter, and we read the passage as the author speaking through the third position — the doxography ending by dissolving into the doctrine

1.114 verified proof

अन्तःकरणतत्त्वस्य वायुराश्रयतां गतः ।
तद्धर्मेण समाविष्टस्तेजसैव विवर्तते ॥ ११४ ॥

antaḥkaraṇatattvasya vāyurāśrayatāṃ gataḥ |
taddharmeṇa samāviṣṭastejasaiva vivartate || 114 ||

The breath, having come to rest upon the principle of the inner organ, pervaded by that principle's properties, appears — by the agency of the fire itself.

The descent of 1.113 examined at its middle joint, where mind and wind interlock. Prana, rising to carry the word out, is not raw air by the time it sounds: it has gone asrayatam, taken the antahkarana-tattva as its resort, and is tad-dharmena samavista — suffused, occupied, by the inner organ's properties. The breath that reaches the lips is mind-conditioned matter, wind already carrying form. And the connective is again tejasa eva vivartate: through the ripening fire, it *appears* — the vivarta verb now applied to the physiological process itself, so that even the pneumatic stage of speech is appearance-of-the-inner rather than fresh production. The verse answers, in passing, the question the vayu-doctrine of 1.108 could not: *why* shaped breath means anything. Because what shapes it from within is the mind's own structure riding down; the wind is expressive because it is invested, not because collisions are eloquent. The mechanist's whole chain is retained and re-read as the outermost garment of self-manifestation.
contested: tejasaiva vivartate is elliptical — we supply 'appears (as utterance) through the fire', continuing 1.113's cooking-fire; others read tejas here as the visual/luminous element into which breath transforms, making the verse a piece of element-physics; the vivarta verb tells against mere element-transformation

1.115 verified proof

विभजन् स्वात्मनो ग्रन्थीन् श्रुतिरूपैः पृथग्विधैः ।
प्राणो वर्णानभिव्यज्य वर्णेष्वेवोपलीयते ॥ ११५ ॥

vibhajan svātmano granthīn śrutirūpaiḥ pṛthagvidhaiḥ |
prāṇo varṇānabhivyajya varṇeṣvevopalīyate || 115 ||

Dividing the knots of itself by the differentiated forms of audible sound, the breath, having manifested the phonemes, dissolves into those very phonemes.

The last event of utterance, rendered with a violence and a pathos the doctrine has earned. The breath ascends carrying the undivided word; at the places of articulation it is svatmano granthin vibhajan — dividing the knots of itself. The knots (granthi, the Vrtti's own word for the involutions of 1.1) are the word's condensed unity as lodged in the moving prana; articulation unties them into sruti-rupaih prthagvidhaih, the differentiated audible forms. Then the extraordinary final clause: the breath, varnan abhivyajya, having manifested the phonemes, varnesv eva upaliyate — dissolves into those very phonemes. The carrier does not survive its delivery; prana spends itself utterly into what it manifests, as if the messenger became the message's letters. This is the production-side answer to 1.54's hearing-side vanishing (words unnoticed, meaning taken): at both ends of speech, the medium consummates itself by disappearing into what it serves. Cosmologically the verse is the microcosm of dissolution-in-manifestation — the one, descending into difference, is wholly *in* the difference, nothing held back — which is why the next verse must immediately insist that the source itself is nonetheless never exhausted.
contested: upaliyate admits both 'dissolves into' (our reading — the breath is used up in the phonemes) and 'merges/hides within' (the breath persists concealed in them); the first suits the physiology, the second the metaphysics of latency, and the word may be chosen to carry both

1.116 verified proof

अजस्रवृत्तिर्यः शब्दः सूक्ष्मत्वान्नोपलभ्यते ।
व्यजनाद्वायुरिव स स्वनिमित्तात् प्रचीयते ॥ ११६ ॥

ajasravṛttiryaḥ śabdaḥ sūkṣmatvānnopalabhyate |
vyajanādvāyuriva sa svanimittāt pracīyate || 116 ||

The word whose functioning is unceasing is not apprehended because of its subtlety; it is made manifest through its own occasioning cause, as air is by a fan.

The insistence 1.115 required: behind each spent utterance stands an unspendable source. The sabda meant here is ajasra-vrtti, of unceasing operation — the inner word-nature of 1.112, never silent, only suksma, too subtle for apprehension. Its non-appearance is thus an epistemic fact, not an ontological absence — and the simile makes the point with perfect economy: vyajanad vayur iva, as air by a fan. Air is always present, always in some motion; the fan does not create it but pracyate, concentrates and quickens it past the threshold of feeling. So utterance stands to the ever-functioning word: articulation is the fan, a sva-nimitta that raises the perpetual subtle operation to gross perceptibility for a moment, after which the word no more ceases than the air does when the fan stops. The verse retro-illuminates the whole manifestation dossier: this is why abhivyakti and not production was the right category from 1.44 onward — one does not produce what never pauses; one occasions its detection.

1.117 verified proof

तस्य प्राणे च या शक्तिर्या च बुद्धौ व्यवस्थिता ।
विवर्तमाना स्थानेषु सैषा भेदं प्रपद्यते ॥ ११७ ॥

tasya prāṇe ca yā śaktiryā ca buddhau vyavasthitā |
vivartamānā sthāneṣu saiṣā bhedaṃ prapadyate || 117 ||

The power of it that stands in the breath, and the power that stands settled in the intellect — that very power, turning itself about in the places of articulation, attains differentiation.

A unification, gathering the descent (1.112-115) into a single subject. What stands in prana and what stands in buddhi are not two agencies but ya ca ... ya ca ... saisa — one sakti doubly stationed, and it is *that* which, vivartamana sthanesu (the vivarta participle again, now in the very throat), bhedam prapadyate, arrives at differentiation. The verse forecloses a dualist misreading the descent-story invited: mind-word and breath-word are not successive entities handed a baton but one power viewed at two stations, its 'motion' from one to the other being appearance, not transit. Read against 1.2, the architecture is exact — as Brahman's many saktis are apthak, non-separate, from it, so the word-power's stations in intellect and breath are non-separate from the one power. Differentiation happens at the sthanas, the outermost edge, and even there it is something the power *presents*, not something it suffers. Everything since 1.44 is here in one sentence: one word-power, two inner stations, appearance of difference at the threshold of sound.

1.118 verified proof

शब्देष्वेवाश्रिता शक्तिर्विश्वस्यास्य निबन्धनी ।
यन्नेत्रः प्रतिभात्मायं भेदरूपः प्रतीयते ॥ ११८ ॥

śabdeṣvevāśritā śaktirviśvasyāsya nibandhanī |
yannetraḥ pratibhātmāyaṃ bhedarūpaḥ pratīyate || 118 ||

The power residing in words themselves is what binds this universe; and this self, whose essence is the flash of insight, whose eye is that power, is perceived in the form of difference.

The crescendo before the kanda's final theses, in which the linguistics is folded back into the cosmology without remainder. First line: visvasya asya nibandhani — the sakti lodged sabdesv eva, in words themselves, is the *binder of this universe*. Nibandhana has been this kanda's word for 'ground' since 1.13 (words ground the realities of practice) and 1.15 (word-forms ground the classes of things); now the grounding is total — the world hangs together wordwise, coherence itself being linguistic. Second line: the atman, pratibha-atma, whose essence is pratibha — the immediate flash of understanding, here named for the first time and destined to be the second kanda's answer to what sentence-meaning is — sees *through* that word-power (yan-netra, 'whose eye is it') and accordingly pratiyate bheda-rupah, is perceived in the form of difference. The construction is precise about where difference lives: an undivided seer, seeing through the instrument of the word-power, appears divided *because its eye is the differentiating instrument* — the eye colours the seen, and the world's plurality is the tint of the lens. The verse thus answers 1.1 across the whole arc of the kanda: how does the one Brahman-word appear as the processional world? Through its own power-of-words serving as the very organ of its self-perception.
contested: yannetrah is a bahuvrihi we resolve as 'whose eye/instrument is that (power)'; others take netra as 'leader' — 'led by which power' — softening the perceptual metaphor; and whether bheda-rupah pratiyate describes the atman as (falsely) perceived in difference or as (really) manifesting difference reopens, deliberately, the vivarta/parinama question of 1.1

1.119 verified proof

षड्जादिभेदः शब्देन व्याख्यातो रूप्यते यतः ।
तस्मादर्थविधाः सर्वाः शब्दमात्रासु निश्रिताः ॥ ११९ ॥

ṣaḍjādibhedaḥ śabdena vyākhyāto rūpyate yataḥ |
tasmādarthavidhāḥ sarvāḥ śabdamātrāsu niśritāḥ || 119 ||

Since the differences beginning with the note sadja are explained and given form by word, therefore all the varieties of objects rest upon the measures of word.

An argument from the hardest case: music. The notes of the scale — sadja and its fellows, the seven svaras — are pure audition, the least conceptual of objects, difference given to the ear with no help from thought. And yet, the verse observes, even these are vyakhyata and rupyate sabdena — explained, and *given their form*, through word: the notes are individuated as notes, held apart as a system of seven, only through their names and the theory that names them. The untrained ear hears continuous pitch; the named scale is what carves it into sadja-versus-rsabha. If even audible difference at its most immediate is word-fixed, then a fortiori artha-vidhah sarvah, all the varieties of objects, rest on sabda-matras, the measures of word. The verse converts 1.15's thesis (classes rest on word-forms) from assertion to argument by finding the a fortiori case, and it opens the final ascent: from here to 1.131 the kanda claims everything — cosmos, practice, cognition, salvation — for the word.

1.120 verified proof

शब्दस्य परिणापोऽयमित्याम्नायविदो विदुः ।
छन्दोभ्य एव प्रथममेतद्विश्वं व्यवर्तत ॥ १२० ॥

śabdasya pariṇāpo'yamityāmnāyavido viduḥ |
chandobhya eva prathamametadviśvaṃ vyavartata || 120 ||

This is the transformation of the word — so those who know the tradition understand; from the Vedic hymns first of all this universe unrolled.

The cosmological warrant, cited from scripture. Amnaya-vidah, the knowers of the received text, understand this world as sabdasya parinamah — and the second line gives their proof-text in paraphrase: chandobhya eva prathamam, from the metres, the hymns, first of all, this universe vyavartata, turned forth. The allusion gathers a genuine Vedic lineage — the cosmogonies in which creation proceeds from vac or through the sacrificial metres. Two cautions for the reader. First, parinama here sits in reported speech: it is the amnaya-knowers' formulation, and Bhartrhari's own considered verb has been vivartate since 1.1 — the juxtaposition quietly registers that scripture speaks the language of real transformation while the analysis has argued for appearance. Second, the verse is doing legitimating work: the extravagant claims of 1.119 and 1.121-123 are being shown to be not a grammarian's professional inflation but the oldest teaching of the tradition itself.
contested: the wisdomlib transcription reads parinapo, an evident typo for parinamah, and we translate accordingly; more substantively, whether Bhartrhari endorses the parinama idiom here or flags it as scriptural shorthand for what is strictly vivarta divides readers, and the kanda's own usage supports the latter

1.121 verified proof

इतिकर्तव्यता लोके सर्वा शब्दव्यपाश्रया ।
यां पूर्वाहितसंस्कारो बालोऽपि प्रतिपद्यते ॥ १२१ ॥

itikartavyatā loke sarvā śabdavyapāśrayā |
yāṃ pūrvāhitasaṃskāro bālo'pi pratipadyate || 121 ||

All procedure-knowledge in the world rests upon word — which even the child comprehends, through the residues deposited in it before.

From cosmos to cradle. Itikartavyata — 'the thus-it-is-to-be-done-ness' of things, know-how, the whole practical intelligence by which life is conducted — is sarva sabda-vyapasraya, wholly word-dependent. The counterexample any critic reaches for is the infant: the bala acts purposively — turns to the breast, grasps, avoids — before language. Bhartrhari's answer is purva-ahita-samskara: the child comprehends through residues laid down before — in prior lives, on the tradition's reckoning; in any case, antecedently to this learning. Infant know-how is not word-free cognition but the operation of word-formed traces not presently audible: an innate endowment that is itself linguistic in structure. The move is only as strong as one's tolerance for its premise, but its shape is exact and modern — practical capacities that precede overt language are explained by an inner, inherited language-likeness rather than admitted as non-linguistic. What 1.123 will call the interweaving is here traced beneath the threshold of speech acquisition.

1.122 verified proof

आद्यः करणविन्यासः प्राणस्योर्ध्वं समीरणम् ।
स्थानानामभिघातश्च न विना शब्दभावनाम् ॥ १२२ ॥

ādyaḥ karaṇavinyāsaḥ prāṇasyordhvaṃ samīraṇam |
sthānānāmabhighātaśca na vinā śabdabhāvanām || 122 ||

The first disposing of the organ, the driving of the breath upward, the striking upon the places of articulation — none occurs without a conceiving in word.

The dependence turned back on speech itself, completing a circle with deliberate care. Utterance is a motor act: the initial karana-vinyasa (tongue and lips taking position), the upward samirana of breath, the abhighata at the sthanas. Each of these, the verse notes, is itself a piece of itikartavyata — a procedure — and so by 1.121 requires sabda-bhavana, a conceiving or intending in word, to occur at all. One cannot *say* anything without inwardly word-forming the saying of it. The regress this threatens (does the inner wording itself need a prior wording?) is not an oversight but the doctrine: it terminates, non-viciously, in the ajasra-vrtti word of 1.116 — the unceasing inner operation that no act initiates because it never began. Speech does not start from silence; audible speaking is a local intensification (the fan of 1.116) of a wording that is always already under way. The verse thus welds the physiology of 1.108-115 to the epistemology about to be announced: even the body's speech-machinery runs on the very medium it outputs.

1.123 · All Cognition is Word-Interwoven verified proof

न सोऽस्ति प्रत्ययो लोके यः शब्दानुगमादृते ।
अनुविद्धमिव ज्ञानं सर्वं शब्देन भासते ॥ १२३ ॥

na so'sti pratyayo loke yaḥ śabdānugamādṛte |
anuviddhamiva jñānaṃ sarvaṃ śabdena bhāsate || 123 ||

There is no cognition in the world that does not follow upon word; all knowledge appears as if interpenetrated by word.

The most celebrated karika of the work, and the thesis for which Bhartrhari stands in the history of philosophy. Na so 'sti pratyayo loke yah sabda-anugamad rte: no cognition whatsoever without the accompaniment of word. Not: no *communicated* cognition, no *reflective* cognition — none. And the second line specifies the mode: sarvam jnanam appears anuviddham iva sabdena, as if interwoven, shot through, with word — anuviddha is the weaver's and jeweller's word, thread through cloth, thread through pearls. Cognition is not first accomplished wordlessly and then labelled; the wording is *in* the seeing, as the thread is in the fabric — remove it and no fabric remains. The iva registers phenomenological honesty (the interweaving is how it *appears*; introspection cannot get behind it to check) rather than hesitation, since by 1.124 the dependence is argued to be constitutive. Against this stands the great counter-tradition — the Buddhist doctrine that perception proper (nirvikalpaka pratyaksa) is precisely wordless, conception arriving after — and the battle line drawn here runs through a millennium of Indian epistemology, Dignaga and Dharmakirti on one side, the grammarians and (with qualifications) Bhatta Mimamsa and Nyaya distributing themselves around the other. Bhartrhari's ground: attend to any actual cognition and the attending is already articulate — the allegedly pure percept is a theoretical posit, never a datum. The kanda's arc closes onto its opening: if all knowing is word-woven and the world is what is known, then the sabda-tattva of 1.1 is not mysticism but the epistemology taken at full strength.

1.124 verified proof

वाग्रूपता चेदुत्क्रामेदवबोधस्य शाश्वती ।
न प्रकाशः प्रकाशेत सा हि प्रत्यवमर्शिनी ॥ १२४ ॥

vāgrūpatā cedutkrāmedavabodhasya śāśvatī |
na prakāśaḥ prakāśeta sā hi pratyavamarśinī || 124 ||

If that eternal word-formedness were to depart from awareness, the light would not shine — for it is that which makes recognition possible.

The transcendental argument for 1.123 — from assertion to necessity in one verse. Suppose the vag-rupata, the word-formedness, sasvati, standing, were utkramet, to withdraw from avabodha, awareness. Then na prakasah prakaseta: the light would not illuminate — cognition would lose precisely what makes it *cognition of* something. The reason: sa hi pratyavamarsini — it, the word-form, is the recognizer, the power of pratyavamarsa, that taking-as, that inner 'this is such-and-such', by which a bare impingement becomes an apprehension. Light that reveals nothing determinate is not light; awareness stripped of the articulating thread would be a shining that shines on nothing, which is no shining at all. So the interweaving of 1.123 is not an empirical regularity that might have exceptions but a condition of the possibility of cognition as such: to be aware is to be articulately aware, and articulation is the word's work. This verse is the philosophical hard core of the entire kanda — later Kashmir Saiva thinkers (Utpaladeva, Abhinavagupta) took pratyavamarsa/vimarsa from exactly here and made it the essence of consciousness itself, divinized: awareness's self-touching is speech. The debt is explicit in their citations.
contested: pratyavamarsini we render 'that which makes recognition possible', taking pratyavamarsa as the determining 'taking-as' within awareness; the term admits shades from 'reflective judgement' to 'introspection', and how intellectualist one makes it decides how strong a thesis the verse proves — the weakest reading (all *judgement* is verbal) is nearly trivial, the strongest (all *sentience* is verbal) is the one the Saiva tradition embraced. The Vrtti supports the strong reading: it extends the word-association into sleep and into indeterminate first-cognitions (which still grasp their objects 'as this or that'), leaving no wordless stratum of awareness at all

1.125 verified proof

सा सर्वविद्याशिल्पानां कलानां चोपबन्धनी ।
तद्वशादभिनिष्पन्नं सर्वं वस्तु विभज्यते ॥ १२५ ॥

sā sarvavidyāśilpānāṃ kalānāṃ copabandhanī |
tadvaśādabhiniṣpannaṃ sarvaṃ vastu vibhajyate || 125 ||

It is the binder of all sciences, crafts and arts; by its power everything that is brought forth is differentiated.

The scope of pratyavamarsini vak, spelled out horizontally. Sarva-vidya-silpa-kalanam upabandhani: sciences, crafts, arts — the theoretical, the technical, the aesthetic — all are 'bound together' by the word-power; upabandhani continues the nibandhana-line of 1.13, 1.15, 1.118 (word as the fastening of the world), now applied to the whole space of human competence. The second line adds the productive side: whatever is abhinispanna, brought to completion — the pot thrown, the hymn composed, the dance performed — vibhajyate tad-vasat, is differentiated, given its internal articulation, under the word's control. A craft-object has parts, stages, a right order; that articulation is word-borne (1.121's itikartavyata at work), and so is the finished thing's very identity as *this* kind of artifact. The verse quietly generalizes the kanda's linguistics of manufacture: as the sentence's parts were projections of analysis (1.73), the artifact's rationale is projected onto matter by the word-guided maker. Culture, in the widest sense, is language differentiating itself into media.

1.126 verified proof

सैषा संसारिणां संज्ञा बहिरन्तश्च वर्तते ।
तन्मात्रामव्यतिक्रान्तं चैतन्यं सर्वजातिषु ॥ १२६ ॥

saiṣā saṃsāriṇāṃ saṃjñā bahirantaśca vartate |
tanmātrāmavyatikrāntaṃ caitanyaṃ sarvajātiṣu || 126 ||

This it is that operates as the consciousness of transmigrating beings, without and within; sentience in all the kinds of beings does not step beyond its measure.

The widest claim of the sequence: the word-power is coextensive with sentience itself. Sa esa samsarinam samjna — it is the samjna, the consciousness or awareness-function, of all transmigrators — bahir antas ca, operating outward (in speech, in practice) and inward (in thought, in the silent taking-as of 1.124). And then the universal quantification: caitanyam sarva-jatisu, sentience in every order of beings — down through animals, to whatever stirs — tan-matram avyatikrantam, does not overstep its measure. Wherever there is consciousness at all, there is word-formedness; the beast's cognition too is anuviddha, at however dim a grade. This universalizing is required by the system: if word-interweaving held only for humans, it would look like an artifact of learned language, and 1.121 already blocked that reading for infants; extending it through all jatis makes vag-rupata a feature of caitanya as such, not of human acculturation. The verse is the bridge over which 1.1's metaphysics and 1.123's epistemology become one doctrine — consciousness, wherever found, is the word-essence functioning; samsara is the word talking itself through the forms of life.
contested: samjna we render as 'consciousness' (its upanisadic sense), though 'name/designation' — its grammatical sense used throughout 1.59-69 — is deliberately audible: beings' very awareness is their 'naming'; the pun resists translation and either English choice loses half

1.127 verified proof

प्रविभागे यथा कर्ता तया कार्ये प्रवर्तते ।
अविभागे तथा सैव कार्यत्वेनावतिष्ठते ॥ १२७ ॥

pravibhāge yathā kartā tayā kārye pravartate |
avibhāge tathā saiva kāryatvenāvatiṣṭhate || 127 ||

As, in the state of differentiation, the agent proceeds by means of it toward the deed, so, in non-differentiation, it is that very power which stands as the deed itself.

A compact metaphysics of action, closing the loop between word and work. Pravibhage, when differentiation reigns — the ordinary condition, agent here, act there — the karta proceeds toward his karya *by* the word-power (taya): intention, plan, procedure, all word-woven (1.121-122); language is the instrument through which agency reaches its deed. But avibhage, in non-differentiation — at the source, before the carving into agent and act — sa eva, that very power, karyatvena avatisthate: *stands as the deed*. Where there is no distance between doer and done, the word-power does not mediate the deed; it is it. Instrument in the divided state, identity in the undivided — the pattern is 1.5's exactly (the Veda as means *and* likeness of Brahman), now generalized to all action whatever. The verse also glosses the old Vedic intuition that the primal creation is a speaking (1.120): at the origin there is no creator wielding speech as tool; the speaking is the creating. Human action, mediated by word, is the divided image of that unmediated word-act.

1.128 verified proof

स्वमात्रा परमात्रा वा श्रुत्या प्रक्रम्यते यथा ।
तथैव रूढतामेति तया ह्यर्थो विधीयते ॥ १२८ ॥

svamātrā paramātrā vā śrutyā prakramyate yathā |
tathaiva rūḍhatāmeti tayā hyartho vidhīyate || 128 ||

Whether as its own measure or as the measure of another — however it is set going by the utterance, just so does it become established; for it is by utterance that the object is determined.

How word-dependence cashes out in semantic practice: things stabilize as they are said. Whatever is prakramyate srutya — launched, set going, by the utterance — tathaiva rudhatam eti, in just that form attains rudhi, established currency (the grammarians' term for conventionalized meaning). The opening alternative sva-matra va para-matra va we read as: whether the utterance presents a thing in its own measure or in the measure of another — directly or through transfer, literally or figuratively — either way, the presented determination is what hardens into the thing's standing identity, tayā hy artho vidhiyate, for the object is *ordained* by utterance. Vidhiyate is strong and chosen: the verb of injunction and ordination. Objects do not first possess determinate contours which speech then reports; the contour settles along the lines speech first draws, and repeated saying is what fixes it (rudhi). This supplies the mechanism for 1.119's claim that all object-kinds rest on word-measures — and 1.129 immediately supplies the limiting case, where saying determines a 'thing' with no thing under it at all.
contested: sva-matra para-matra is elliptical and our construal (own measure/another's measure = literal/transferred presentation) is one of several; the pair has also been taken as 'in its own form or that of another word' (phonological adaptation) — the verse's point, that establishment follows utterance, survives any of these

1.129 verified proof

अत्यन्तमतथाभूते निमित्ते श्रुत्युपाश्रयात् ।
दृश्यतेऽलातचक्रादौ वस्त्वाकार निरूपणा ॥ १२९ ॥

atyantamatathābhūte nimitte śrutyupāśrayāt |
dṛśyate'lātacakrādau vastvākāra nirūpaṇā || 129 ||

Even where the occasioning ground is utterly not so, by resort to the utterance one sees the determination of a thing-form — as in the fire-brand circle and the like.

The limiting case that proves 1.128: language determining an object where reality supplies none. The alata-cakra, the fire-brand circle, is the tradition's shared example of pure appearance — whirl a torch and there *is* no circle, atyantam atathabhute nimitte, the ground being utterly otherwise (a single moving point). Yet vastvakara-nirupana drsyate: a thing-form is determined, seen, held — and, the verse says, sruty-upasrayat, by resort to the word. We *say* 'circle', and the saying furnishes the form under which the non-thing is apprehended as a thing. The example is aimed with polemical precision: the Buddhist uses the alata-cakra to show that persistent objects are constructions; Bhartrhari accepts the analysis and identifies the constructing power — it is vac, the word, that builds the circle out of the whirl. Perception supplies moving light; the word supplies the cakra. And if the word can constitute the object *there*, where nothing underlies it, its form-giving role in ordinary cognition (1.123-124, 1.128) can hardly be denied — the illusion is not an embarrassment to the theory but its cleanest exhibition.

1.130 verified proof

अपि प्रयोक्तुरात्मानं शब्दमन्तरवस्थितम् ।
प्राहुर्महान्तमृषभं येन सायुज्यमिष्यते ॥ १३० ॥

api prayokturātmānaṃ śabdamantaravasthitam |
prāhurmahāntamṛṣabhaṃ yena sāyujyamiṣyate || 130 ||

Moreover, they declare the word standing within the speaker's self to be the great bull — union with which is what is sought.

The kanda's soteriology begins, and it begins with a citation. The 'great bull' is the mahan rsabha of the old texts — the allusion is to the Vedic figure of speech as the great bull (catvari srnga, 'four horns...', RV 4.58.3, is the passage the tradition connects here, whose bull the exegetes read as vac). What the seers so named, the verse says, is the sabda antar avasthita — the word stationed *within the speaker's self*: not the audible fourth part but the inner totality of speech, the ajasra-vrtti word of 1.116, the vagatman of 1.112. And with it sayujyam isyate — union, the technical term of Vedic soteriology (con-yoking, being joined-with), is what is sought. The religious end is thus declared to be union with the inner word — not with a deity beyond language but with language at its own root, which by 1.1 *is* the ultimate. The verse authorizes 1.131's conclusion: if union with the inner word is the goal, then the discipline of the word is the path, and grammar's claim to be dvaram apavargasya (1.14) stops being hyperbole and becomes the system's plain consequence.
contested: the rsabha allusion is certain in tenor but its exact source and the compound maha-antam-rsabham admit variant resolutions; we follow the standard identification with the Vedic bull-of-speech imagery, and note that the karika's api prahuh marks it explicitly as received teaching being adduced, not argued

1.131 verified proof

तस्माद्यः शब्दसंस्कारः सा सिद्धिः परमात्मनः ।
तस्य प्रवृत्तितत्त्वज्ञस्तद्ब्रह्मामृतमश्नुते ॥ १३१ ॥

tasmādyaḥ śabdasaṃskāraḥ sā siddhiḥ paramātmanaḥ |
tasya pravṛttitattvajñastadbrahmāmṛtamaśnute || 131 ||

Therefore that perfecting of the word is the attainment of the supreme self; one who knows the principle of its functioning enjoys the deathless Brahman.

The conclusion of the whole kanda's argument, marked tasmat. If the ultimate is word-essence (1.1), if all cognition is word-woven (1.123-124), if consciousness across beings is the word-power functioning (1.126), and if the sought union is with the inner word (1.130) — then sabda-samskara, the perfecting or refinement of the word (the grammarian's daily work: samskara is what grammar *does* to speech), simply is siddhih paramatmanah, the attainment of the supreme self. And the knower — pravrtti-tattva-jna, one who knows the principle of the word's functioning, its descent from undivided source through mind and breath to audible difference and its return — tad brahma-amrtam asnute, enjoys that deathless Brahman. Asnute, 'eats, enjoys': attainment as tasting, recalling the paramo rasa of 1.12. Every honorific claim of 1.11-22 is hereby redeemed as doctrine: grammar as highest tapas, as therapy, as the door of liberation — these follow, given the identifications the kanda has argued. The Brahmakanda ends its constructive arc here; what remains (1.132-155ish) is consolidation — the necessity of tradition, the divisions of speech, and the closing frame.

1.132 · The Necessity of Tradition verified proof

न जात्वकर्तृकं कश्चिदागमं प्रतिपद्यते ।
बीजं सर्वागमापाये त्रय्येवातो व्यवस्थिता ॥ १३२ ॥

na jātvakartṛkaṃ kaścidāgamaṃ pratipadyate |
bījaṃ sarvāgamāpāye trayyevāto vyavasthitā || 132 ||

Never does anyone accept a tradition to be without an author; when all traditions perish, the seed remains — hence the triple Veda stands established.

A return to the philosophy of tradition (1.27-43) at a deeper station, with a puzzle set on its face. First line: no agama circulates as ownerless — every current of teaching is received as *someone's*, akartrkam pratipadyate no one. Second line: sarvagamapaye bijam — when all traditions pass away (in the dissolution between world-ages, on the cosmology in play), a seed persists, from which the triple Veda re-establishes itself (trayy evato vyavasthita). The two lines look opposed — everything authored, yet something survives all authors — and the resolution is the doctrine of 1.28 writ cosmic: what is eternal is not any authored text-body but the vyavastha, the seed-order that re-flowers identically each age; the 'authors' each cycle are the rsis who *see* it again (1.37), not composers. The verse thereby threads between Mimamsa (authorless Veda) and Nyaya (authored, by God): authored in every manifestation, authorless as the seed. Bija completes the metaphor-chain of 1.84 and 1.111 — trace-seed in the hearer, cloud-seed in the atoms, now world-seed across dissolutions: at every scale, the word persists as latency between manifestations.
contested: the first line can be read as the opponent's premise (nobody accepts an authorless tradition — so how is the Veda authorless?) answered by the second, or as Bhartrhari's own concession en route to the seed-doctrine; we take the latter, but the dialectical cut differs between readers

1.133 verified proof

अस्तं यातेषु वादेषु कर्तृष्वन्येष्वसत्स्वपि ।
श्रुतिस्मृत्युदितं कर्म लोको न व्यतिवर्तते ॥ १३३ ॥

astaṃ yāteṣu vādeṣu kartṛṣvanyeṣvasatsvapi |
śrutismṛtyuditaṃ karma loko na vyativartate || 133 ||

When the doctrines have set, and other authors are no more, the world does not transgress the rite declared by sruti and smrti.

The empirical companion to 1.132's cosmology. Vadas — the philosophical systems, with their founders and their fortunes — asta yatesu, set like stars; their authors pass, their schools scatter. Yet the karma declared in sruti and smrti, the ritual and practical order, lokah na vyativartate — the world does not step across it; the agnihotra outlives every philosophy of the agnihotra. The observation is sociological and, within its horizon, accurate: doctrines are mortal in a way that rites are not, argument-driven allegiance decaying faster than practice-driven transmission. Its argumentative work: to display, within recorded history, the same asymmetry 1.132 asserted across cosmic dissolutions — teachings-with-authors perish, the practice-bearing core persists — so that the seed-doctrine is not exotic metaphysics but the long shape of ordinary cultural time. It also silently continues 1.30-31: the immunity of the unbroken practice to tarka is here shown, not argued — the vadas that would have judged it are the ones that have set.

1.134 verified proof

ज्ञाने स्वाभाविके नार्थः शास्त्रैः कश्चन विद्यते ।
धर्मो ज्ञानस्य हेतुश्वेत्तस्याम्नायो निवन्धनम् ॥ १३४ ॥

jñāne svābhāvike nārthaḥ śāstraiḥ kaścana vidyate |
dharmo jñānasya hetuśvettasyāmnāyo nivandhanam || 134 ||

If knowledge were natural, there would be no purpose whatever in the sastras; if dharma is the cause of knowledge, then tradition is its ground.

A dilemma posed to the naturalist about knowledge, each horn closing toward tradition. Suppose jnana svabhavika — cognition arises naturally, needing no instruction: then the sastras are pointless; yet no party to the debate, least of all the naturalist writing books, treats them so — reductio by performative fact. Alternatively, grant what the tradition holds: that dharma conditions knowing — the knower's moral-ritual formation is causally upstream of his cognitive reach (the yogic epistemology of 1.37 presupposed exactly this). Then, since dharma is not established by unaided reason (1.30), amnayo nibandhanam: tradition is knowledge's ground, at one remove. Either the sastras are idle (which practice refutes) or they are load-bearing via dharma (which makes agama foundational). The verse formalizes what the kanda has assumed since 1.14 called grammar a purifier: disciplines do not merely inform; they *qualify the knower* — and any epistemology that ignores the knower's formation has, on this view, mistaken its subject.
contested: the second pada's syntax (dharmo jnanasya hetus cet...) can be read as a genuine conditional the author endorses, or as continuing the opponent's alternatives; the wisdomlib transcription's nivandhanam for nibandhanam is the v/b alternation common in the source and we read nibandhanam

1.135 verified proof

वेदशास्त्राविरोधी च तर्कश्चक्षुरपश्यताम् ।
रूपमात्राद्धि वाक्यार्थः केवलान्नावतिष्ठते ॥ १३५ ॥

vedaśāstrāvirodhī ca tarkaścakṣurapaśyatām |
rūpamātrāddhi vākyārthaḥ kevalānnāvatiṣṭhate || 135 ||

And reasoning that does not conflict with Veda and sastra is an eye for those who do not see; for from the mere form alone the meaning of a sentence does not stand settled.

The rehabilitation of tarka — the kanda's last word on reason, and a deliberately generous one. After the demolition of 1.32-42, reasoning returns with a licensed office: veda-sastra-avirodhi tarkah, reasoning that keeps concord with Veda and sastra, is caksur apasyatam — an *eye for the sightless*, for those without the rsi's direct vision (1.37-38), which is to say for everyone now. And the second line gives the reason tradition itself requires this eye: rupa-matrat kevalat, from the bare form alone, vakyarthah na avatisthate — the meaning of a sentence does not settle. Texts do not interpret themselves; the received word, precisely because it is received as *form*, demands construal, and construal is reasoning's proper province. So the settlement: tarka is blind as a foundation (1.30) and indispensable as an instrument — sovereign nowhere, employed everywhere, its concord with the tradition being the condition of its safe use, as the frame of a lens is to its glass. That sentence-meaning outruns sentence-form is, moreover, no small aside: it is the thesis-seed of the entire second kanda, where what a vakya means will be argued to be an undivided pratibha that the parsed form merely occasions. The kanda's epistemological ledger thus closes balanced: vision above, tradition around, reason within.

1.136 verified proof

सतोऽविवक्षा पारार्थ्यं व्यक्तिरर्थस्य लैङ्गिकी ।
इति न्यायो बहुविधस्तर्केण प्रविभज्यते ॥ १३६ ॥

sato'vivakṣā pārārthyaṃ vyaktirarthasya laiṅgikī |
iti nyāyo bahuvidhastarkeṇa pravibhajyate || 136 ||

Non-intention of what is present, subservience to another, indication of a meaning through a mark — such manifold interpretive principle is articulated by reasoning.

What the licensed tarka of 1.135 actually does, shown by its instruments. The verse names three canons from the interpreter's working kit: sato 'vivaksa — a real feature of the situation may be *unintended*, present but not meant (the speaker mentions the man's staff without meaning staffness); pararthya — an element may stand wholly in service of another, its sense exhausted in subordination (the Mimamsa principle governing subsidiary and principal); vyaktir laingiki — a meaning may be delivered only through its linga, its indicatory mark, never said outright. Each canon adjudicates precisely the gap 1.135 identified between form and meaning: form alone cannot tell you what was not-intended, what is subordinate, what is marked-at — judgement must. Iti nyayo bahuvidhah: and such maxims are many; tarkena pravibhajyate, reasoning is what articulates and applies them. The verse is a window onto the actual exegetical culture — Mimamsa's interpretive jurisprudence — that the kanda's abstractions have presupposed throughout: 'tradition needs an eye' is no metaphor; here are the lens-grindings.
contested: the three items are compressed to the point of riddle and translators divide sharply; we take them as three named canons of interpretation (unintendedness of the actual, subservience, mark-borne indication), but each compound admits alternative resolution — e.g. sato 'vivaksa as 'non-assertion of the existent' in the ontological sense — and the verse may be quoting a list whose original context is lost

1.137 verified proof

शब्दानामेव सा शक्तिस्तर्को यः पुरुषाश्रयः ।
स शब्दानुगतो न्यायोऽनागमेष्वनिवन्धनः ॥ १३७ ॥

śabdānāmeva sā śaktistarko yaḥ puruṣāśrayaḥ |
sa śabdānugato nyāyo'nāgameṣvanivandhanaḥ || 137 ||

That power belongs to words themselves; the reasoning that rests on a person, when it follows the words, is interpretive principle — apart from the traditions it is without ground.

The settlement of reason's status in one careful sentence. The interpretive maxims of 1.136 work — but their sakti, their authority, sabdanam eva: belongs to the words themselves, to the structure of the received language they explicate. Tarka as it lives in a reasoner — purusasraya, person-based, the wit of Devadatta — is authoritative only sabda-anugata, when it follows the words: tracking, subserving, drawing out. Cut loose — anagamesu, where there is no received text to follow — it is anibandhana, groundless: the same activity that was an eye within tradition (1.135) is a wanderer without it. This resolves the apparent zigzag of the kanda's epistemology (reason demolished 1.32-42, licensed 1.135-136): there were never two doctrines, only one — reasoning is real as an *instrument of the word* and nothing as an independent source. The formulation is also quietly self-referential: the Vakyapadiya itself, bristling with arguments, claims authority not as one man's tarka but as reasoning anugata to the grammarians' agama — the book practices the epistemology it preaches.
contested: we read the verse as *subordinating* personal reasoning to word-grounded principle; it can also be read as *identifying* them ('the nyaya just is that word-power as refracted through persons'), a more charitable unity — the practical upshot (no freestanding authority for tarka) is the same

1.138 verified proof

रूपादयो यथा दृष्टाः प्रत्यर्थं यतशक्तयः ।
शब्दास्तथैव दृश्यन्ते विषापहरणादिषु ॥ १३८ ॥

rūpādayo yathā dṛṣṭāḥ pratyarthaṃ yataśaktayaḥ |
śabdāstathaiva dṛśyante viṣāpaharaṇādiṣu || 138 ||

As colour and the rest are seen to have their powers fixed with respect to each object, just so are words seen to have theirs — in the removal of poison and the like.

Evidence that words have real, object-fixed powers — from the mantra. The analogy's base: sensible qualities have yata-sakti, determinate powers indexed to their bearers (this herb's colour signals this virtue; fire's touch burns and water's cools, invariably). The claim: words exhibit the same fixity, and the evidence offered is visapaharana-adi — the removal of poison and such: the practice of mantra-healing, where *this* formula and no paraphrase of it counteracts *this* venom. Whatever a modern reader makes of the practice, its logic is the argument's point: mantra-efficacy is form-bound — substitute a synonym and the power is gone — which is inexplicable if words are arbitrary labels whose whole being is conventional reference, but exactly what one expects if particular word-forms carry particular saktis as natively as colours carry theirs. The verse recruits the tradition's applied linguistics (the whole culture of mantra-sastra) as empirical support for the ontology of 1.13 and 1.128: words are powers before they are signs.
contested: the wisdomlib transcription's yatasaktayah we read as yata-sakti ('fixed powers', as at 1.6); the sceptical reader will note the argument is only as strong as the mantra-datum, which Bhartrhari's audience granted and ours may not — the karika's own force is conditional on that concession

1.139 verified proof

यथैषां तत्र सामर्थ्यं धर्मेऽप्येवं प्रतीयताम् ।
साधूनां साधुभिस्तस्माद्वाच्यमभ्युदयार्थिनाम् ॥ १३९ ॥

yathaiṣāṃ tatra sāmarthyaṃ dharme'pyevaṃ pratīyatām |
sādhūnāṃ sādhubhistasmādvācyamabhyudayārthinām || 139 ||

As their capacity is acknowledged there, so let it be acknowledged with respect to dharma as well; therefore what is correct is to be spoken by the correct-speaking who seek felicity.

The inference from 1.138 drawn, and the kanda's practical ethics stated. If words have real form-bound powers in the visible domain (poison-cure), yatha esam tatra samarthyam, then evam dharme 'pi pratiyatam — let the same be acknowledged where the fruit is unseen: in dharma. The step is the standard extension from drsta to adrsta efficacy that ritual reasoning always makes (1.7's dual purposes), here applied to speech-forms themselves. Conclusion, marked tasmat: sadhunam sadhubhir vacyam abhyudaya-arthinam — correct words are to be spoken, by the cultivated, for those (or: by those) who seek abhyudaya, felicity, the rising good. The verse completes the arc opened at 1.27: there, correct and corrupt forms conveyed meaning equally and correctness was dharma, on authority; now the doctrine has its mechanism (words as powers, 1.138) and its imperative (speak the sadhu form — the power is in the form, so the form is the practice). Grammar's rules are hereby cashed as ethics: usage is conduct, and conjugation, rightly done, is a small continuous sacrifice.
contested: sadhubhih can be instrumental of agent ('by the cultivated') or of accompaniment/means, and abhyudaya-arthinam genitive plural can attach to the speakers or beneficiaries; the injunction's sense — right forms for the sake of felicity — is stable across the parsings

1.140 verified proof

सर्वोऽदृफलानर्थानागमात् प्रतिपद्यते ।
विपरीतं च सर्वत्र शक्यते वक्तुमागमे ॥ १४० ॥

sarvo'dṛphalānarthānāgamāt pratipadyate |
viparītaṃ ca sarvatra śakyate vaktumāgame || 140 ||

Everyone arrives at objects whose fruit is unseen through tradition; and about tradition, the contrary can everywhere be alleged.

A concession of disarming honesty, placed exactly where the argument is strongest. First line: adrsta-phala objects — dharma, the felicities of 1.139, everything whose fruit no one living has seen — are reached agamat, through tradition, by *everyone*: there is no other road (1.30 again, now as sociology). Second line: viparitam ca sarvatra sakyate vaktum agame — and about any tradition, the opposite can always be said. Counter-traditions exist; every agama has a rival alleging its inversion; no received teaching is immune to contradiction by another received teaching. Bhartrhari does not blink the consequence: tradition-dependence cannot be *demonstrated* into safety; the sceptic can always speak. What follows for him is not suspension but the practical resolution already given — one stands in the unbroken usage of one's sistas (1.31, 1.141 next) because standing somewhere is not optional and brokenness is the only disqualifier available. The verse is the kanda's moment of epistemic candour: the ground of practice is firm enough to live on and too contested to prove, and the philosophy of tradition must be written for exactly that terrain.
contested: the transcription sarvo 'drphalanarthan is corrupt at drpha-; context and the parallel constructions make adrsta-phalan arthan ('objects of unseen fruit') the evident reading, and we translate so while noting the source's fault

1.141 verified proof

साधुत्वज्ञानविषया सैषा व्याकरणस्मृतिः ।
अविच्छेदेन शिष्टानामिदं स्मृतिनिबन्धनम् ॥ १४१ ॥

sādhutvajñānaviṣayā saiṣā vyākaraṇasmṛtiḥ |
avicchedena śiṣṭānāmidaṃ smṛtinibandhanam || 141 ||

This grammatical tradition is that whose object is the knowledge of correctness; it is grounded in the unbroken remembrance of the cultivated.

The kanda's definition of its own discipline, quiet after the grand claims. Vyakarana-smrti — grammar as a *smrti*, a remembrance-tradition, not a sruti: the classification is itself doctrine (1.7 made smrtis derivative of the Veda via lingas; grammar is such a derivative, the sistas' codified memory). Its visaya: sadhutva-jnana, knowledge of which forms are correct — not meaning (which corrupt forms convey equally, 1.27), not communication, but correctness as such, the dharma-bearing property. Its ground: avicchedena sistanam — the *unbrokenness* of the cultivated line, idam smrti-nibandhanam, this is what fastens the remembrance. Every term has been earned: aviccheda from 1.31 and 1.41 (the immunity of the unbroken), sista from 1.27 (usage of the learned as criterion), nibandhana as the kanda's own word for grounding since 1.13. The verse is the discipline's charter written in its own vocabulary — and its modesty matters: after word-Brahman and cosmic seeds, grammar's proximate authority is simply that the chain of cultivated speakers never broke.

1.142 · The Levels of Speech verified proof

वैखर्या मध्यमायाश्च पश्यन्त्याश्चैतदद्भुतम् ।
अनेकतीर्थभेदायास्त्रय्या वाचः परं पदम् ॥ १४२ ॥

vaikharyā madhyamāyāśca paśyantyāścaitadadbhutam |
anekatīrthabhedāyāstrayyā vācaḥ paraṃ padam || 142 ||

This is the wonder: that beyond the threefold speech — vaikhari, madhyama and pasyanti, with their many distinct fords — there is the supreme station of the Word.

The kanda's celebrated hierarchy, named at last. The stages have been operating incognito since 1.112-117 (utterance; the word in mind and breath; the knower in the subtle speech-self); now the names: vaikhari, the elaborated — audible, articulated, sequence-ridden speech; madhyama, the middle — the inner word, mentally deployed, sequenced in thought though unsounded; pasyanti, the seeing — speech at the station where it *sees* rather than says: undivided, sequence-swallowed, intentionality without articulation. Aneka-tirtha-bheda: each level has many fords, many descents and crossings — dialects and deliveries at the gross end, modes of inner discourse and vision at the subtle. And the adbhuta, the declared wonder: trayyah vacah param padam — beyond even the triad, a supreme station of speech. The wording is a deliberate re-sounding of the oldest text in this lineage: the Rgveda's four quarters of vac, three hidden and the fourth what men speak (RV 1.164.45), and the padam of catvari padani. Whether that fourth is para vac as a distinct level (as the later tradition, especially Kashmir Saiva, systematized it — para as a fourth above pasyanti) or pasyanti itself in its absolute aspect is left genuinely open by the karika. The doctrinal work is done either way: speech is depth-structured, its audible face is its furthest emanation, and the sabda-tattva of 1.1 stands at (or as) the summit of an accessible gradient — which makes the return-path of 1.131 (union by the word's perfecting) a climb through one's own speaking, from voice through mind to vision.
contested: whether the karika teaches three levels with a 'supreme station' identical to pasyanti's summit, or implies the four-level scheme (with para vac) the commentators and the Saiva tradition made canonical, is the classical crux of the verse; the Vrtti already reads toward the higher systematization, but the karika's own syntax — 'the param padam of the threefold vac' — supports the minimal reading, and we translate minimally

1.143 verified proof

तद्विभागाविभागाभ्यां क्रियमाणमवस्थितम् ।
स्वभावज्ञैश्च भावानां दृश्यन्ते शब्दशक्तयः ॥ १४३ ॥

tadvibhāgāvibhāgābhyāṃ kriyamāṇamavasthitam |
svabhāvajñaiśca bhāvānāṃ dṛśyante śabdaśaktayaḥ || 143 ||

That, whether undergoing division or non-division, remains established; and the powers of words are seen by those who know the nature of things.

The invariance of the summit through its own descents. Tat — the supreme word-station of 1.142 — kriyamanam vibhaga-avibhagabhyam, being done into division and non-division, worked through both conditions (uttered into vaikhari's differentiation, gathered into pasyanti's unity), avasthitam: stands, unaltered. The doctrine is 1.75-77 and 1.101 lifted to the top of the hierarchy — as the sphota is undivided through fast and slow delivery, the param padam is unmoved through the entire cycle of speech's emanation and return; levels are its vrtti-bheda, never its own change. Second line: sabda-saktayah, the powers of words — the whole system of them, from mantra-efficacy (1.138) to world-binding (1.118) — drsyante svabhava-jnaih bhavanam: are *seen* by knowers of the natures of things. Drsyante is the verb of 1.37-38: this is vision-knowledge, the rsi's mode — the deep powers of speech are not inferred from usage but beheld by those whose cognition has climbed the very gradient 1.142 mapped. The verse thus locks hierarchy to epistemology: each level of speech is also a station of knowing, and the powers operative at a level are visible only from at or above it.
contested: the referent of tat and the syntax of kriyamanam avasthitam are elliptical; we take 'that' as the param padam persisting through division/non-division, but the pair has also been construed as smrti-work ('being made, through analysis and synthesis, it stands established'), continuing 1.141's charter — the transcription's ambiguity leaves both live

1.144 verified proof

अनादिमव्यवच्छिन्नां श्रुतिमाहुर कर्तृकाम् ।
शिष्टैर्निबध्यमाना तु न व्यवच्छिद्यते स्मृतिः ॥ १४४ ॥

anādimavyavacchinnāṃ śrutimāhura kartṛkām |
śiṣṭairnibadhyamānā tu na vyavacchidyate smṛtiḥ || 144 ||

They declare sruti to be beginningless, unbroken, without author; smrti, however, being composed by the cultivated, is yet not discontinued.

The two canons given their formal statuses, completing 1.132's threading of the needle. Sruti: anadi, avyavacchinna, akartrka — beginningless, uninterrupted, authorless: the seed-order of 1.132 in its manifestation, received by seers (1.37), never made. Smrti: sistair nibadhyamana — *presently being composed*, the participle notably continuous: the cultivated are always writing it, codifying usage and duty age by age (grammar itself among them, 1.141) — tu na vyavacchidyate, and yet it too is never severed: authored at every point, unbroken as a stream. The two eternities are thus of different types — sruti eternal in substance (or as the recurrent seed), smrti eternal only as continuity of composition, the praninam-analogy of 1.28 institutionalized. The pair resolves the tension 1.132 staged (nothing authorless received; something survives all authors): sruti is the surviving seed, smrti the perpetual re-authoring around it, and the authority of the second derives from unbroken contact with the first (1.7's lingas). The grammarian's own book — this book — is smrti in exactly this sense: made, dated, mortal, and a link in a chain that is none of those things.

1.145 verified proof

अविभागाद्विवृत्तानामभिख्या स्वप्नवच्छ्रुतौ ।
भावतत्त्वं तु विज्ञाय लिङ्गेभ्यो विहिता स्मृतिः ॥ १४५ ॥

avibhāgādvivṛttānāmabhikhyā svapnavacchrutau |
bhāvatattvaṃ tu vijñāya liṅgebhyo vihitā smṛtiḥ || 145 ||

Of the things unrolled out of the undivided, the naming in sruti is as in a dream; smrti, however, is framed from the indicatory marks, after knowing the reality of things.

A daring epistemology of revelation, run through the kanda's own cosmology. The bhavas, things, are vivrtta avibhagat — unrolled out of the undivided (the vivarta of 1.1, the descent of 1.112ff.). Their abhikhya, their naming-and-appearing in sruti, is svapnavat: *as in a dream*. The comparison is precise, not deflationary: dream-presentation is immediate, involuntary, whole — the dreamer does not compose the dream but undergoes its self-showing; so the rsi does not construct the Veda but receives the undivided's own self-unrolling into name (1.37's vision, now given its phenomenology — and note that 1.1's Vrtti had already reached for the dream-analogy for the world-appearance itself). Smrti by contrast is work: bhava-tattvam vijnaya, having *understood* the reality, lingebhyah vihita — framed from marks, inferentially, deliberately (1.7 verbatim). Revelation is thus dream-like showing; tradition-literature is waking construction upon it — two cognitive modes, not two grades of one. The pairing dignifies both: sruti's authority is its unconstructedness, smrti's is its method.
contested: svapnavat is the crux: we read it as characterizing the *mode* of sruti's reception (immediate self-presentation, like dream), not its truth-value; a reader who presses the simile toward illusion makes the verse subversive of sruti in a way the whole kanda forbids — yet the choice of so double-edged a comparison is itself surely deliberate

1.146 verified proof

कायवाग्बुद्धिविषया ये मलाः समवस्थिताः ।
चिकित्सालक्षणाध्यात्मशास्त्रैस्तेषां विशुद्धयः ॥ १४६ ॥

kāyavāgbuddhiviṣayā ye malāḥ samavasthitāḥ |
cikitsālakṣaṇādhyātmaśāstraisteṣāṃ viśuddhayaḥ || 146 ||

The impurities that stand settled in body, speech and mind — their purifications are through the sciences of medicine, grammar and the self.

The triad of disciplines, mapped to the triad of faults. Malas, impurities, station themselves in kaya, vac, buddhi — body, speech, intellect: the tradition's standard trio of act-channels (karma is done by body, word and mind). Each has its cathartic science: cikitsa, medicine, for the body's disorders; laksana, 'the science of characteristics' — grammar's own honorific title, the discipline of laksya and laksana, usage and rule — for speech; adhyatma-sastra, the science of the self, for the mind's. The verse institutionalizes 1.14's metaphor (grammar as cikitsita of speech-faults): grammar is not *like* medicine; it occupies the middle seat in a single therapeutic system of the person, flanked by medicine and metaphysics. And the ordering carries the kanda's conviction: speech stands between body and spirit as the hinge-channel — the vag-malas (the apabhramsas about to be discussed, 1.147ff.) are neither physical disease nor mental error but a third species of impurity, with its own pathology, its own materia medica (the sista usage), and its own physicians. Purity of speech is thereby made as real, and as consequential, as health and wisdom.

1.147 verified proof

शब्दः संस्कारहीनो यो गौरिति प्रयुयुक्षिते ।
तमपभ्रंशमिच्छन्ति विशिष्टार्थनिवेशिनम् ॥ १४७ ॥

śabdaḥ saṃskārahīno yo gauriti prayuyukṣite |
tamapabhraṃśamicchanti viśiṣṭārthaniveśinam || 147 ||

The word devoid of the refinement — when 'gauh' is what one intends to employ — that they call apabhramsa, though it settles on the distinguished meaning.

The formal definition of the corrupt form, with its crucial concession renewed. A speaker intends the correct gauh, 'cow'; what leaves him is gavi or goni — samskara-hina, lacking the refinement grammar names (1.131's sabda-samskara, here in its technical sense). That deviation is apabhramsa, the 'fallen-away' form. And the definition's honest clause: visistartha-nivesinam — the fallen form *settles on the distinguished meaning perfectly well*; the hearer gets the cow. Corruption is thus defined against intention and norm, not against communicative success — exactly 1.27's doctrine, now with the mechanism the intervening kanda supplied: the apabhramsa conveys, but conveys *through* the sadhu form it deviates from (as 1.152 will say), like a mispronounced name that reaches its bearer only because the right name is recoverable through it. The dossier 1.147-155 that this opens — the theory of linguistic corruption — is the kanda's most linguistically observant stretch: language change, register, and the sociolinguistics of prestige are all about to appear, inside a normative frame.
contested: prayuyuksite ('when one intends to employ') marks the definition as intention-relative: the same phonetic form is apabhramsa only against a meant sadhu target — which raises, without answering, the status of forms native to speakers who intend nothing else; 1.153-154 return to exactly this edge

1.148 verified proof

अस्वगोण्यादयः शब्दाः साधवो विषयान्तरे ।
निमित्तभेदात् सर्वत्र साधुत्वं च व्यवस्थितम् ॥ १४८ ॥

asvagoṇyādayaḥ śabdāḥ sādhavo viṣayāntare |
nimittabhedāt sarvatra sādhutvaṃ ca vyavasthitam || 148 ||

Words like asva and goni are correct in another domain; correctness everywhere is settled by difference of ground.

The relativity of correctness, stated with complete technical calm. Goni as 'cow' is apabhramsa; goni as 'sack' is sadhu — same phonetic form, different visaya, different verdict. Asva likewise (the traditional pair: as 'non-owner', a perfectly formed negative compound, versus dialectal deformation of asva, 'horse'). Sarvatra, everywhere, sadhutvam nimitta-bhedat vyavasthitam: correctness is fixed by the *ground* of use — the meaning-domain, the derivational warrant — not by the sound in itself. No form is corrupt as acoustics; corruption is a relation between form, meaning and derivational title. The verse quietly explodes any essentialist reading of the mala-doctrine (1.146): the impurity is not *in* the syllables, as a stain is in cloth — it is a failure of entitlement, a form appearing where it holds no warrant, like a true coin passed in the wrong currency. This sharpens rather than softens the norm: since correctness is relational, only the smrti's map of grounds (1.141) can adjudicate it, and the ear alone never suffices.
contested: the source's asvagonyadayah we resolve as asva-goni-adayah with the traditional examples; the exact pair matters little — the doctrine of nimitta-bheda is the verse's whole cargo

1.149 verified proof

ते साधुष्वनुमानेन प्रत्ययोत्पत्तिहेतवः ।
तादात्म्यमुपगम्येव शब्दार्थस्य प्रकाशकाः ॥ १४९ ॥

te sādhuṣvanumānena pratyayotpattihetavaḥ |
tādātmyamupagamyeva śabdārthasya prakāśakāḥ || 149 ||

Among correct words, those are causes of the arising of cognition by way of inference; they illuminate the word's meaning as if having attained identity with it.

How the corrupt form manages to mean — the semantics of deviation, first half. Te, the apabhramsas (continuing 1.147-148): amid the sadhus they cause understanding anumanena, by way of inference — the hearer, receiving gavi, recovers gauh as what this speaker, in this mouth-habit, must intend, and the meaning arrives through that recovery; the corrupt form is evidence of the correct one, not its equal. Yet phenomenologically — and the verse is careful to honour this — the deviant form illuminates tadatmyam upagamya iva, *as if* it had become identical with the meaning: the hearer of a familiar solecism feels no inference happening, the sense simply appears. The iva does its usual precise work: seeming-identity, actual mediation. The account is a fine piece of pragmatics — meaning through reconstruction-so-practiced-it-vanishes — and it preserves both halves of 1.27's data: communicative equality in *feel*, normative asymmetry in *mechanism*. The sadhu signifies directly; the apabhramsa signifies through the sadhu, and 1.150 will name the consequence.
contested: we take te as the apabhramsas working 'among the sadhus' (sadhusu locative of context); the verse is also construed with sadhus as subject — correct words as inferential causes — which inverts the doctrine and fits 1.150 poorly; the passage's flow decides for the first

1.150 verified proof

न शिष्टैरनुगम्यन्ते पर्याया इव साधवः ।
ते यतः स्मृतिशास्त्रेण तस्मात्साक्षादवाचकाः ॥ १५० ॥

na śiṣṭairanugamyante paryāyā iva sādhavaḥ |
te yataḥ smṛtiśāstreṇa tasmātsākṣādavācakāḥ || 150 ||

The correct forms are not followed by the cultivated as if synonyms of the corrupt; since that is so by the science of tradition, the corrupt are therefore not directly expressive.

The consequence drawn, against the leveller. If corrupt and correct forms both convey the meaning (1.147, 1.149), why not call them paryayas, synonyms — parallel names of one thing, as hasta and kara both name the hand? Because, the verse answers, the sistas do not so treat them: na anugamyante paryaya iva — in cultivated usage the pair never behaves as a synonym-pair does (either member freely chosen, neither corrective of the other); gavi is *corrected to* gauh, never the reverse, and the asymmetry is codified smrti-sastrena, by the received science (1.141). Therefore: saksat avacakah — the corrupt forms are not *directly* expressive; their conveying is the mediated, inferential kind of 1.149. The argument-form deserves notice: a semantic thesis (direct versus mediated expression) is established from a *sociological* datum (the direction of correction among the cultivated) certified by smrti — exactly the epistemology 1.27-31 and 1.137 licensed, where the practice of the sistas is the deep evidence and theory follows it. The synonym-test itself is genuinely sharp linguistics: substitutability-without-correction remains a working criterion for synonymy.
contested: the ablative/instrumental construction yatah smrti-sastrena is compressed; we read 'since (their non-parity) is so by the smrti', others read 'since they are not followed (as expressive) *because of* the smrti' — making the smrti the cause rather than the certifier of the asymmetry; the difference matters for how conventionalist one takes the doctrine to be

1.151 verified proof

अम्बाम्बेति यथा बालः शिक्षमाणः प्रभाषते ।
अव्यक्तं तद्विदां तेन व्यक्ते भवति निश्चयः ॥ १५१ ॥

ambāmbeti yathā bālaḥ śikṣamāṇaḥ prabhāṣate |
avyaktaṃ tadvidāṃ tena vyakte bhavati niścayaḥ || 151 ||

As a child learning to speak utters 'amba, amba' indistinctly, and by it those who know become certain about the distinct form —

The homeliest verse in the kanda, and the model for 1.152. The infant calls amba amba — 'mama' — avyaktam, indistinctly: the form is blurred, half-made. Yet tad-vidam, for those who know — the mother, the household — niscayah bhavati vyakte: certainty arises *about the distinct form*; the hearers pass through the blur to the word the child is reaching for, and understanding happens there, at the vyakta, not at the noise. The baby-talk case is perfect for its purpose because the mediation is undisguised: no one thinks the infant's syllables signify in their own right; everyone hears *through* them to amba — inference (1.149) with its workings still visible, before habit has hidden them. It also carries a quiet tenderness of doctrine: the child's fallen form is not censured but *completed* by the knowing hearer — which is what every hearer of apabhramsa does, and what the sista tradition does for the whole speech-community, holding the distinct forms on behalf of tongues still learning. Sanskrit itself, the verse implies, is the language's amba: what the blurred callings mean.
contested: none of substance; the verse is syntactically the clearest in the passage — its work is entirely in the analogy it prepares for 1.152

1.152 verified proof

एवं साधौ प्रयोक्तव्ये योऽपभ्रंशः प्रयुज्यते ।
तेन साधुव्यवहितः कश्चिदर्थोऽभिधीयते ॥ १५२ ॥

evaṃ sādhau prayoktavye yo'pabhraṃśaḥ prayujyate |
tena sādhuvyavahitaḥ kaścidartho'bhidhīyate || 152 ||

So, where the correct form is to be employed, when a corrupt one is employed, some meaning is expressed by it with the correct form intervening.

The apodosis, and the semantics of corruption completed. Evam — as with the child: sadhau prayoktavye, where the sadhu form is the one *to be* employed (the normative participle carrying 1.139's ethics), the apabhramsa that actually issues expresses its meaning sadhu-vyavahitah — with the correct form *intervening*, standing between the corrupt sound and the meaning as the relay through which expression passes. Vyavahita is the doctrine's exact word: not 'replaced by' nor 'associated with' but *interposed* — the hearer's route runs gavi → gauh → cow, the middle term supplied by competence (the mother's knowledge in 1.151, the sista-smrti at large). This is the final settlement of the problem opened at 1.27: corrupt forms convey (conceded throughout), but their conveying is parasitic — borrowed light, the sadhu form doing the expressing from behind the screen. The theory has real explanatory reach: it predicts that apabhramsa comprehension varies with the hearer's command of the target forms (as it does), and it makes the norm *internal* to the semantics rather than an etiquette laid over it. 1.153 immediately tests the theory at its breaking point: communities where no target form remains.
contested: kascid arthah, 'some meaning', is lightly hedged — either 'a certain meaning' (the intended one, reached mediately) or a genuine qualification ('some meaning or other', imperfectly determinate); the first fits the child-analogy, the second anticipates 1.153's degraded-transmission case, and the hedge may be deliberately open

1.153 verified proof

पारम्पर्यादपभ्रंशा विगुणेष्वभिधातृषु ।
प्रसिद्धिमागता येषु तेषां साधुरवाचकः ॥ १५३ ॥

pāramparyādapabhraṃśā viguṇeṣvabhidhātṛṣu |
prasiddhimāgatā yeṣu teṣāṃ sādhuravācakaḥ || 153 ||

When, through transmission among defective speakers, corrupt forms have attained establishment, for those speakers it is the correct form that is non-expressive.

The theory driven honestly into its hardest case — and yielding to it. Paramparyat: by transmission, generation upon generation, among vigunesu abhidhatrsu, speakers gone defective (communities severed from the sista line). There the apabhramsas prasiddhim agatah — have attained establishment, become the *current* forms, rudhi (1.128) now running entirely on the fallen side. And the verse does not flinch from the consequence: tesam, for those speakers, sadhur avacakah — it is the *correct* form that fails to express. The relay of 1.152 has inverted: where gavi is what everyone says and gauh is heard as strange noise, the mediation runs the other way or not at all; expressiveness follows establishment, and establishment follows transmission, wherever transmission goes. This is a remarkable concession for a normative grammarian — it grants the Prakrits their full semantic citizenship within their own communities and concedes that sadhutva confers no magical reach beyond the community that sustains it. What survives the concession is only, but exactly, the dharma-claim (1.27, 1.139): the established corrupt form expresses, in its own domain, as well as Sanskrit ever did — what it does not do is bear the unseen fruit. The norm has retreated, under pressure of honest linguistics, to the one ground (adrsta) where observation cannot follow to contest it.
contested: none syntactically; doctrinally, whether 1.153 describes a regrettable pathology or simply *how language communities work* is the live question — the verse's own even tone supports the descriptive reading, and 1.154 shows the school dividing over precisely this

1.154 verified proof

दैवी वाग्व्यतिकीर्णेयमशक्तैरभिधातृभिः ।
अनित्यदर्शिनां त्वस्मिन् वादे बुद्धिविपर्ययः ॥ १५४ ॥

daivī vāgvyatikīrṇeyamaśaktairabhidhātṛbhiḥ |
anityadarśināṃ tvasmin vāde buddhiviparyayaḥ || 154 ||

This divine speech has been overspread by incapable speakers; among the seers of non-eternality, in this matter, the judgement is the reverse.

The two ideologies of language change, set face to face. First the school's own telling: daivi vak — divine speech, the Sanskrit of the seed-tradition — vyatikirnā, overspread, scattered through, asaktaih abhidhatrbhih, by speakers without capacity (the vigunas of 1.153): change as corruption, history as the silting of a pure source. Then, with striking evenhandedness, the rival telling: anitya-darsinam — for the seers of non-eternality, those whose darsana makes impermanence fundamental (the Buddhists above all, though any anti-eternalist qualifies) — asmin vade buddhi-viparyayah: in this dispute their judgement is the exact reverse. On their view the current speech of peoples is the natural datum and 'divine Sanskrit' the artifact — a fixed, frozen elaboration raised out of living vernaculars; the direction of derivation flips with the metaphysics. Bhartrhari states the inversion without refuting it — consistent with 1.140's concession that about any agama the contrary can be alleged — and the honesty is structural: he has already granted (1.153) everything the rival needs descriptively; what separates the schools is not the linguistics but the eternality-metaphysics, and *that*, by his own epistemology, argument alone will not settle (1.34). The verse is perhaps the earliest clear statement that the Sanskrit-Prakrit priority dispute is theory-laden all the way down.
contested: vyatikirna we render 'overspread/scattered-through'; it admits 'mixed, adulterated' — the image of seed-crop overgrown; the identification of the anitya-darsins with Buddhists specifically is traditional and probable but the karika names no school

1.155 verified proof

उभयेषामविच्छेदादन्यशब्दविवक्षया ।
योन्यः प्रयुज्यते शब्दो न सोऽर्थस्याभिधायकः ॥ १५५ ॥

ubhayeṣāmavicchedādanyaśabdavivakṣayā |
yonyaḥ prayujyate śabdo na so'rthasyābhidhāyakaḥ || 155 ||

Because both streams are unbroken, when a word is employed with the intention of another word, that word is not the expressor of the meaning.

The kanda's final karika in this recension of the text, resolving 1.154's standoff at the level of practice. Ubhayesam avicchedat — both lines, sadhu and apabhramsa, sista-transmission and vernacular transmission, are *unbroken*: each has the continuity that 1.31 and 1.141 made the touchstone of standing; neither can be dismissed as a mere lapse of the other (this much is 1.153-154 consolidated). The discrimination therefore falls back — as at 1.147 — on intention within employment: anya-sabda-vivaksaya yo 'nyah prayujyate sabdah — when one word is employed *with the intention of another word* (gavi uttered reaching for gauh, or indeed the reverse), na so 'rthasya abhidhayakah: that uttered word is not the expressor; the intended word is, through it (1.152). The doctrine's last word is thus neither the triumph of Sanskrit nor its relativization, but a criterion indifferent to which stream is 'first': expression belongs to the form the speaker's vivaksa actually deploys, and forms reached-through are conduits, not expressors. That this closes the Brahmakanda is fitting beyond its topic: the kanda began with the undivided word appearing as many (1.1) and ends with the everyday case of one word appearing through another — the whole metaphysics recapitulated in a grammarian's scruple about who, in the utterance, is really doing the meaning.
contested: the verse generalizes 1.147's intention-relativity, and its scope is debated: read narrowly it restates the apabhramsa doctrine; read broadly it covers *any* employment-through (citation, metalepsis, even the dhvani-sphota relation itself) — the placement as kanda-close invites the broad reading, and we incline to it; note also that the karika-count of Kanda I varies between recensions, wisdomlib's text ending at 155 while other editions carry additional closing verses