The vṛttis (complex formations) are five, in the form of: kṛt [derivatives], taddhita [derivatives], samāsa, ekaśeṣa, and dhātus [ending in] san and the rest. "A vṛtti is the expression of another meaning (parārtha-abhidhāna)." "The vigraha (analytic paraphrase) is the sentence (vākya) that makes known the meaning of the vṛtti." It is of two kinds: laukika (belonging to ordinary usage) and alaukika (not belonging to ordinary usage). The laukika [vigraha] is correct (sādhu), because it is fully formed (pariniṣṭhita). The alaukika is incorrect (asādhu), unfit for use (prayoga). For example: rājñaḥ puruṣaḥ ("the king's man") [is laukika]; rājan as puruṣa su [i.e. the stems with their sup endings, the genitive singular as and the nominative singular su, is alaukika].
A nitya-samāsa (obligatory compound) is one that has no [laukika] vigraha, or one whose vigraha is not made with its own padas (asvapada-vigraha).
That samāsa is of four kinds is a statement of what generally holds (prāyovāda), because [samāsa] is prescribed by saha supā (SK 649) also for those [compounds] that lie outside the adhikāras of avyayībhāva, tatpuruṣa, bahuvrīhi and dvandva.
The ancients' saying (pravāda) as well — "the avyayībhāva has the meaning of the prior member (pūrvapada) as predominant; the tatpuruṣa has the meaning of the subsequent member (uttarapada) as predominant; the bahuvrīhi has the meaning of another word (anyapada) as predominant; the dvandva has the meaning of both members as predominant" — is intended [only] as holding generally (prāyo 'bhiprāyaḥ), because [this] is absent in avyayībhāvas such as sūpaprati and unmattagaṅgam, in tatpuruṣas such as atimāla, in bahuvrīhis such as dvitrāḥ, and in dvandvas such as dantoṣṭham.
The karmadhāraya is a particular [kind of] tatpuruṣa; the dvigu is a particular [kind] of that [karmadhāraya]. Having many padas (anekapadatva) belongs only to dvandva and bahuvrīhi; for the tatpuruṣa it has been stated [that this occurs] only in some cases.
Moreover:
(Verse) "Of sup with sup, with tiṅ, with a nāman (nominal stem), [and] with a dhātu; then of tiṅ with tiṅ, [and of tiṅ] with a subanta — thus samāsa is to be known by the wise as of six kinds."
[sup] with sup: rājapuruṣaḥ. [sup] with tiṅ: paryabhūṣayat. [sup] with a nāman: kumbhakāraḥ. [sup] with a dhātu: kaṭaprūḥ; ajasram. tiṅ with tiṅ: pibatakhādatā; khādatamodatā. tiṅ with sup: the [action] in which [one says] "kṛnta vicakṣaṇa" ("cut, O clever one!") is kṛntavicakṣaṇā. The samāsa [in this last] is because "ehīḍādayo 'nyapadārthe ca" ("and ehīḍa and the rest, in the sense of another word's meaning") is read in the mayūravyaṃsakādi [list] (the list headed by mayūravyaṃsaka).
- textual Several evident slips in the source are translated on the intended reading and not emended: 'parārārthābhidhānaṃ' for parārthābhidhānaṃ (a doubled syllable); 'vākyaḥ vigrahaḥ' for vākyaṃ vigrahaḥ; 'bahirbhūtānam api' for bahirbhūtānām api; 'giṅā' in the verse for tiṅā (the examples that follow have 'tiṅā'); 'danoṣṭham' for dantoṣṭham (the samāhāra dvandva 'teeth and lips').
- textual The source's 'kṛttaddhitasamāsaikaśeṣasanādyadhāturūpāḥ' has 'sanādya-dhātu', where sanādyanta-dhātu ('dhātus ending in san and the rest') is expected (compare sanādyantā dhātavaḥ, A 3.1.32). Translated as 'dhātus [ending in] san and the rest', with the ending supplied in brackets.
- translator 'saha supā' (SK 649) is A 2.1.4. The 'mayūravyaṃsakādi [list]' is the gaṇa of mayūravyaṃsakādayaś ca (A 2.1.72, SK 754), where the gaṇasūtra ehīḍādayo 'nyapadārthe is read. The source's own form of that gaṇasūtra here carries a final 'ca'.
- commentary BM and TB both open the section as taken up incidentally (prasaṅgāt), BM calling it a miscellany (prakīrṇaka) useful for all compounds. BM: the five are kṛdantas, formations containing a taddhita, samāsas, ekaśeṣas, and dhātus ending in san and the other affixes (sanādipratyayānta) — followed for the bracketed '[ending in]'. TB: 'vṛtti' is the earlier teachers' name for the five, adopted here.
- commentary On 'parārthābhidhānaṃ vṛttiḥ' the commentators construe 'para' differently. BM: the definition stands in the Bhāṣya on 'samarthaḥ padavidhiḥ' (A 2.1.1); abhidhāna is lyuṭ in the instrument sense; 'para' = other (anya) than the meanings of the padas of the vigraha, i.e. the single qualified meaning (viśiṣṭaikārtha) conveyed by the collective power (samudāyaśakti) of the padas. TB: para-artha = the qualified meaning arising through the inclusion of an affix or of another word's meaning; hence a tiṅanta is not a vṛtti; it quotes Kaiyaṭa on that sūtra: the meaning of another word is expressed by a different word, as in rājapuruṣa the meaning 'man' is expressed by rāja. Corrected 2026-10-05: the English formerly read 'a meaning [that belongs] to another', taking TB/Kaiyaṭa's side; it now reads neutrally 'another meaning'.
- commentary BM on the vigraha: although (per the Bhāṣya on the samartha-sūtra) only the vṛtti conveys the qualified single meaning, the vigraha sentence is the group of separately used padas bearing the vibhaktis fit for the compound. 'pariniṣṭhita' = refined by grammar (vyākaraṇa-saṃskṛta); 'prayogānarha' = because not so refined. rājñaḥ puruṣaḥ is the laukika and rājan as puruṣa su the alaukika vigraha sentence. Followed for 'fully formed' and the bracketed labels.
- commentary BM and TB: 'avigrahaḥ' = without a laukika vigraha (BM: the compound being obligatory) — followed for the bracketed '[laukika]', corrected 2026-10-05 from a bare 'no vigraha'. 'asvapada-': BM = a vigraha made with other words of the same meaning as the compounded padas; TB = one not containing all the compounded padas. Both fit 'not made with its own padas', left as it stands. BM adds that in name compounds (saṃjñā) a sentence is still used to show the member words, and the claim that such compounds are nitya is there secondary (gauṇa); recorded.
- commentary BM: 'prāyas' is the indeclinable for 'mostly', so the fourfold claim aims at the majority; 'bahirbhūtānām api' is completed by 'samāsānām' (followed in '[compounds]'). In the next sentence 'abhāvāt', heard after 'dvandve ca', construes with each of the four: sūpaprati is uttarapada-predominant and unmattagaṅgam / lohitagaṅgaṃ deśaḥ anyapada-predominant, so neither has the prior member predominant; atimāla etc. are pūrvapada-predominant; dvitrāḥ is ubhayapada-predominant; in dantoṣṭham only the samāhāra is predominant. Followed for '[this] is absent'.
- commentary BM: karmadhāraya is a kind of tatpuruṣa by 'tatpuruṣaḥ samānādhikaraṇaḥ karmadhārayaḥ', dvigu a kind of karmadhāraya by 'saṃkhyāpūrvo dviguḥ' (followed for '[karmadhāraya]'); 'anekapadatva' = having two, three, four or more padas, because the 'aneka' of 'anekam anyapadārthe' carries into the dvandva rule (TB same); 'kvacid eva' = in dvyahnajāta and the like (TB's OCR: brahmajāta); 'ity uktam' = in the Bhāṣya etc.
- commentary BM on the verse: subantas compound with a subanta, a tiṅanta, a prātipadika (TB: nāman = prātipadika) and a dhātu; 'atha' breaks off the preceding clause; tiṅantas with a tiṅanta and a subanta — six kinds. Followed for 'then'. Examples (BM; TB agrees where it comments): paryabhūṣayat — pari with the tiṅanta by splitting off 'saha' from 'saha supā' (yogavibhāga); kumbhakāraḥ — upapada compound by 'upapadam atiṅ' with the prātipadika kāra, since gati, kāraka and upapada compound with kṛts before the sup arises; kaṭaprūḥ — compound with the dhātu pru (gatau) made by nipātana in the vārttika 'kvib vacipracchyāyatastukaṭaprujuśrīṇāṃ dīrgho 'saṃprasāraṇaṃ ca', not an upapada compound; ajasram — nañ with the dhātu jas by nipātana in 'nami-kampi-smy-ajasa-kama-hiṃsa-dīpo raḥ'; pibatakhādatā — by the mayūravyaṃsakādi (TB: 'ākhyātam ākhyātena…' read there).
- commentary BM ties 'ehīḍādayo 'nyapadārthe ca' to kṛntavicakṣaṇā alone, as the answer to 'how does a tiṅanta compound with a subanta here?' (the tiṅ-with-tiṅ compounds having been accounted for by the mayūravyaṃsakādi already). Corrected 2026-10-05: the English formerly read 'The samāsa [in these]'; now '[in this last]'. BM closes by recording that the sixfold count is open to question (cintya) — fivefold is proper, sup-with-tiṅ and tiṅ-with-sup both being compounds of a sup and a tiṅ — while others distinguish them by which member stands first. Recorded.
- textual The Bālamanoramā on this prose is sk.p22.prose.b1, re-filed 2026-10-05 from sk.2628.b1 (p55), with its colophon 'iti bālamanoramāyāṃ sarvasamāsaśeṣaḥ'; the Tattvabodhinī is sk.p22.prose.t1, re-filed 2026-10-05 from sk.939.t1 (p21), with its closing 'iti sarvasamāsaśeṣaḥ'. The commentary notes above are made from both.