11. Halanta-puṃliṅga-prakaraṇa — Masculine nouns whose stem ends in a consonant

After this chapter you will be able to decline any masculine noun whose stem ends in a consonant: root-nouns such as lih "licker" and duh "milker", the ox anaḍuh, the numeral catur "four", the king rājan, the sky div, the pronouns kim, idam, adas and puṃs, the personal pronouns yuṣmad "you" and asmad "I", the road pathin, the numerals pañcan and aṣṭan, the priest ṛtvij, the directional words prāñc, pratyañc, udañc and tiryañc, the possessive dhīmat, the participle dadat, and the learned vidvas. A consonant-final stem is called halanta (hal "consonant" + anta "end"). Its declension is built from the same twenty-one endings as the vowel-stems of chapter 8, but the meeting of a stem-final consonant with an ending, or with nothing at all at the end of a word, sets off a chain of sound-substitutions that the vowel-stems never showed. This chapter teaches those substitutions rule by rule, always on a concrete word, and closes with the personal pronouns and their short unaccented forms te, me, vaḥ, naḥ.

Before you start

• Chapter 1 (Saṃjñā-prakaraṇa) is assumed: you can spot a marker letter (it), unpack a pratyāhāra such as ac, ik, hal, jhal, jaś, car, śar and khar, and you know what guṇa, vṛddhi, hrasva and dīrgha mean, and that a rule of the last three quarters of the Aṣṭādhyāyī (the Tripādī, A 8.2 onward) is asiddha, invisible, to every earlier rule.
• Chapter 2 (Paribhāṣā-prakaraṇa) is assumed: a substitute (ādeśa) replaces an original (sthānin); a substitute normally replaces only the final sound of what it is prescribed for; an augment (āgama) marked with ṭ goes in front, one marked with k goes at the end, and one marked with m goes after the last vowel; a substitute may be treated as if it were the original (sthānivadbhāva), except in a rule about a single sound (al-vidhi).
• Chapter 3 (Ac-sandhi) is assumed for the vowel joinings inside derivations: a + a → a (pararūpa, ato guṇe), a + ā → ā (savarṇa-dīrgha), a + i → e and a + u → o (guṇa, ād guṇaḥ), i → y and u → v before a vowel (yaṇ), and the vṛddhi au of ety-edhaty-ūṭhsu; also, from its end, jaśtva (SK 84: a stop or sibilant at the end of a pada becomes the voiced unaspirated stop of its class, ḍh → ḍ).
• Chapter 5 (Hal-sandhi) is assumed for the consonant changes that recur on every page here: cartva (a stop becomes the hard stop of its class before a hard sound, and optionally at a pause, ḍ → ṭ), ścutva (a dental becomes a palatal beside a palatal), ṣṭutva (a dental becomes a retroflex beside a retroflex), the anusvāra with its parasavarṇa (a nasal made homorganic with the following stop), and the doubling of consonants. Chapter 4 (Prakṛtibhāva) is not used.
• Chapters 6 and 7 (the visarga and the word-final s) are assumed: a final s becomes ru (an r with a marker u), and ru becomes visarga before a pause or a hard consonant.
• From chapter 1 you also have the accents udātta and anudātta; a pāda is a metrical quarter of a verse.
• Chapter 8 (Ajanta-puṃliṅga-prakaraṇa, the masculine vowel-stems) is assumed above all. From it you know the twenty-one case-endings su, au, jas; am, auṭ, śas; ṭā, bhyām, bhis; ṅe, bhyām, bhyas; ṅasi, bhyām, bhyas; ṅas, os, ām; ṅi, os, sup; the names aṅga (the stem an ending is being attached to), pada (a finished word, and also a stem before an ending that begins with a consonant), bha (a stem before an ending that begins with a vowel or y, from śas onward), sarvanāmasthāna (the five strong endings su, au, jas, am, auṭ), sambuddhi (the vocative singular) and ghi; the deletion of su after a consonant (hal-ṅyādi-lopa, SK 252); the lengthening of the vowel before the n of a strong ending (upadhā-dīrgha); the deletion of a final consonant-cluster's last member (saṃyogānta-lopa, SK 54); and the pronoun paradigm of sarva with its smai, smāt, smin, sām endings and its śī for jas. Every "the rest like rāma" or "like sarva" below sends you back there.
• Chapter 9 (the feminine vowel-stems) is assumed only for the change of n to ṇ after r or ṣ in a compound whose second member has one vowel (ṇatva, SK 307), and for the feminine affix ṅīp.

A. Stems in h — the ḍh, gh and bhaṣ substitutions at the end of a word — running example: duh → dhuk / dhug (also lih → liṭ, druh → dhruk)

A consonant-final stem meets its endings in two very different ways. Before a vowel-ending the final consonant simply stays: lih + au → lihau. Before an ending that begins with a consonant, or when the ending su has been deleted and the stem stands alone as a finished word, the final consonant is at the end of a pada and must obey the word-final rules of the Tripādī. For a stem in h those rules replace the h by ḍh (SK 324), or by gh if the root began with d (SK 325), and may also aspirate the first consonant of the root (SK 326). The result is then run through jaśtva and cartva. So duh "milker" appears as dhuk or dhug before a pause, but as duhau, duhaḥ before vowels.

SK 324 · हो ढः · ho ḍhaḥ (A 8.2.31)

Rule. An h is replaced by ḍh when a jhal (any stop or sibilant) follows, and at the end of a pada. The two conditions "before a jhal" and "at the end of a pada" are not in this rule's own words; they are carried forward into it by anuvṛtti, and they govern the whole group of rules that follows.

Example. The stem lih "licker", from the root lih.

lih + su → su is deleted after a consonant (hal-ṅyādi-lopa, SK 252) → lih

lih → h now stands at the end of a pada → h → ḍh → liḍh

liḍh → jaśtva (ḍh → ḍ) → liḍ → cartva, optional at a pause (ḍ → ṭ) → liṭ / liḍ.

Before a vowel the h is untouched: lihau, lihaḥ, liham, lihā. Before bhyām (a consonant, so the stem is a pada) the h becomes ḍh → ḍ: liḍbhyām. In the locative plural the ḍ before su is ṭ by cartva; the Kaumudī gives both liṭtsu and liṭsu.

Applies when / blocked by. Only before a jhal or at the end of a pada; never before a vowel. The stem nah is not covered: by a special rule, naho dhaḥ, its h becomes dh instead, giving upānat "sandal". Some h-stems have a second option, the gh of SK 325 and SK 327: uṣṇih gives both uṣṇiṭ and uṣṇik.

SK 325 · दादेर्धातोर्घः · dāderdhātorghaḥ (A 8.2.32)

Rule. The h of a root that begins with d in its upadeśa (its original statement in the root-list) is replaced by gh, before a jhal and at the end of a pada. This rule is an apavāda, an exception, to SK 324: for a d-initial root it gives gh where SK 324 would give ḍh.

Example. The root duh "to milk" begins with d, so its h becomes gh, not ḍh.

duh + su → su deleted (SK 252) → duh → h → gh at the end of a pada → dugh (continued under SK 326).

Applies when / blocked by. The words "in upadeśa" matter in both directions. The past form adhok "he milked" carries the augment a- in front, so at the moment the rule acts the root no longer begins with d; but it did begin with d when it was first stated, and that is enough, so adhok has gh (heard as k). Conversely, the noun dāmalih "licking the rope" can be turned into a root by a denominative affix, dāmalihyati "he wishes a rope-licker for himself"; that root, with the affix kvip, gives dāmaliṭ. Here gh must not apply, because dāmalih was never stated as a d-initial root; it became a root only later.

SK 326 · एकाचो बशो भष् झषन्तस्य स्ध्वोः · ekāco baśo bhaṣ jhaṣantasya sdhvoḥ (A 8.2.37)

Rule. Take a portion of a root that contains one vowel (eka-ac) and ends in a jhaṣ (a voiced aspirate: jh, bh, gh, ḍh, dh). The baś sound in that portion (a voiced unaspirated stop: b, g, ḍ, d) is replaced by the corresponding bhaṣ (voiced aspirate: bh, gh, ḍh, dh) when s follows, when the syllable dhv follows, and at the end of a pada. In short: the baś of the one-vowel, jhaṣ-final portion becomes bhaṣ before s, before dhv, or at the end of a pada.

Example. Continuing duh.

dugh (from SK 325) → the portion dugh is monosyllabic and ends in the jhaṣ gh; its baś d → dh → dhugh

dhugh → jaśtva (gh → g) → dhug → cartva, optional at a pause → dhuk / dhug.

Before a vowel nothing happens: duhau, duhaḥ. In the locative plural the s of su becomes ṣ after k (ṣatva) and the g becomes k before it (cartva): dhukṣu.

Applies when / blocked by. Bhaṭṭoji insists that the rule speaks of a portion of a root, not of the root as a whole. If "monosyllabic and jhaṣ-final" described the whole root, the rule would miss gardabh: from "he calls it a donkey", gardabhayati, with the affix kvip and loss of the causative ṇi, comes the root-noun gardabh; its nominative singular is gardhap, where the d of the portion dabh has become dh and the final has become p by cartva. That works only because the one-vowel portion dabh is enough. The condition "before a jhal", carried down from SK 324, stops here: the rule itself names s and dhv, and that mention shows that a jhal in general no longer counts. So in dugdham "milked" and dogdhā "milker", where the gh stands before a jhal that is neither s nor dhv, the d stays d. In the word duh the whole root is also its own portion; the grammar allows this by vyapadeśivadbhāva, treating the whole as if it were a part called by the same name.

SK 327 · वा द्रुहमुहष्णुहष्णिहाम् · vā druhamuhaṣṇuhaṣṇihām (A 8.2.33)

Rule. For the four roots druh "to hate", muh "to be confused", ṣṇuh "to vomit" and ṣṇih "to love", the h optionally (vā) becomes gh before a jhal and at the end of a pada. When the option is not taken, SK 324 gives ḍh.

Example. The stem druh "hater".

druh + su → su deleted → druh → h → gh (this rule) or h → ḍh (SK 324) → drugh / druḍh

→ the first consonant is aspirated by SK 326 (d → dh) → dhrugh / dhruḍh

→ jaśtva and optional cartva → dhruk / dhrug / dhruṭ / dhruḍ.

muh, ṣṇuh and ṣṇih go the same way. A stem in vāh, such as viśvavāh "all-carrying", takes only the ḍh of SK 324 in these positions: viśvavāṭ / viśvavāḍ; before vowels viśvavāhau, viśvavāhaḥ, viśvavāham, viśvavāhau. Its remaining forms need the next group.

B. The stem vāh — saṃprasāraṇa to ūṭh — running example: viśvavāh → viśvauhaḥ

The compound viśvavāh "carrying everything" is regular in its strong forms (viśvavāṭ, viśvavāhau) but changes shape before the weak vowel-endings: viśvauhaḥ, viśvauhā. The mechanism is saṃprasāraṇa, the replacement of a semivowel by its vowel: the v of vāh gives way to ū, the ā that followed the v is swallowed by that ū, and the a of viśva then combines with ū to give the vṛddhi au. Three rules do this: a name (SK 328), the substitute (SK 329) and the vowel-merger (SK 330).

SK 328 · इग्यणः सम्प्रसारणम् · igyaṇaḥ samprasāraṇam (A 1.1.45)

Rule. This is a saṃjñā, a naming rule. An ik vowel (i, u, ṛ, ḷ) that is put in the place of a yaṇ (a semivowel y, v, r, l) is called saṃprasāraṇa. The name belongs to the vowel itself.

Example. In viśva-vāh the v of vāh is a yaṇ; the ū that SK 329 puts in its place is an ik; that ū is therefore a saṃprasāraṇa. The same name is used again for śvan, yuvan, maghavan (SK 362–363) and for vidvas (SK 435).

SK 329 · वाह ऊठ् · vāha ūṭh (A 6.4.132)

Rule. For the vāh of a stem that is bha (before a vowel-initial weak ending), the saṃprasāraṇa substitute is ūṭh, that is ū; the ṭh is a marker. The ū replaces the v.

Example.

viśvavāh + śas → the ending is vowel-initial and weak, so the stem is bha → the v of vāh → ū → viśva-ū-āh + as (continued under SK 330).

SK 330 · सम्प्रसारणाच्च · samprasāraṇācca (A 6.1.108)

Rule. When a vowel follows a saṃprasāraṇa vowel, the single substitute (ekādeśa) for the two is the form of the earlier one (pūrvarūpa). So the vowel that used to follow the semivowel disappears into the saṃprasāraṇa.

Example.

viśva-ū-āh + as → ū is a saṃprasāraṇa and ā follows it → the pair ū + ā → ū → viśva-ūh + as

viśva-ūh → the a of viśva meets ū: by ety-edhaty-ūṭhsu (SK 73) the pair is replaced by vṛddhi au, not by guṇa o → viśvauh + as → viśvauhaḥ.

The other bha-forms follow viśvauhaḥ, viśvauhā. Scholars discuss how the classical word viśvavāh is formed at all: the affix ṇvi that forms it may be Vedic only, in which case the ordinary word is made with vic from a causative stem.

C. catur, anaḍuh, sah, div, and the stems in l and m — augments and substitutions before the strong endings — running example: catur → catvāraḥ, caturṇṇām, caturṣu

Four stems receive a vowel inserted into them before the strong endings. The numeral catur "four" and the ox anaḍuh take the augment ām (SK 331), anaḍuh also takes num in the nominative singular (SK 332), and both take am instead of ām in the vocative (SK 333). The stem sah changes s to ṣ when it has the shape sāḍ (SK 335), and div "sky" ends in au before su and in u elsewhere at the end of a word (SK 336–337). The genitive plural of catur takes an inserted n (SK 338), and its locative plural caturṣu shows why a final r does not become visarga and why the ṣ is not doubled (SK 339–340). Bhaṭṭoji closes the group with a stem in l (kamal) and a stem in m (praśām, SK 341). He interleaves the four stems; here catur is the thread and the others are handled beside it.

SK 331 · चतुरनडुहोरामुदात्तः · caturanaḍuhorāmudāttaḥ (A 7.1.98)

Rule. For catur and anaḍuh the augment ām is inserted when a sarvanāmasthāna (strong ending) follows, and that ām carries the udātta accent. An augment marked with m goes after the last vowel of the stem, so ām sits between the u and the final consonant.

Example.

catur + jas → jas is a strong ending → ām after the last vowel u → catu-ām-r + as → u before a vowel → v (yaṇ) → catvār + as → catvāraḥ "four" (nominative plural).

anaḍuh + au → ām after u → anaḍu-ām-h + au → u → v → anaḍvāhau (nominative dual); likewise anaḍvāhaḥ (nominative plural).

SK 332 · सावनडुहः · sāvanaḍuhaḥ (A 7.1.82)

Rule. Before su, anaḍuh takes the augment num (an n placed after the last vowel; u and m are markers). Bhaṭṭoji notes that the word āt, "after an a-vowel", is carried into this rule from an earlier one, so this num is placed after an a-vowel — which can only be the ā of the augment ām. That is why num, though the more specific rule, does not block ām: it presupposes it. And num is likewise not blocked by the vocative am of SK 333.

Example.

anaḍuh + su → ām (SK 331) → anaḍvāh + su → num after the ā → anaḍvānh + su

→ su deleted (SK 252) → anaḍvānh → the cluster nh loses its last member (saṃyogānta-lopa, SK 54) → anaḍvān.

Applies when / blocked by. Two rules that could spoil the form are kept off. The d of SK 334 (which replaces the final of anaḍuh at the end of a pada) does not touch the n of num: if it did, the augment would have been pointless, since anaḍvāh alone would already give anaḍvāt; the very prescription of num shows it is meant to be heard. And the rule deleting a final n of a nominal stem does not apply, because the deletion of h that made the n final is a Tripādī operation and so asiddha, invisible, to it.

SK 333 · अम् सम्बुद्धौ · am sambuddhau (A 7.1.99)

Rule. For catur and anaḍuh, in the vocative singular (sambuddhi) the augment is am, not ām. This is an apavāda to SK 331.

Example.

he anaḍuh + su (vocative) → am after u → anaḍu-am-h → u → v → anaḍvah + su → num (SK 332), su deleted, h dropped from the cluster → he anaḍvan.

Before a weak vowel-ending the stem is bare: anaḍuhā. Before bhyām see SK 334.

SK 334 · वसुस्रंसुध्वंस्वनडुहां दः · vasusraṃsudhvaṃsvanaḍuhāṃ daḥ (A 8.2.72)

Rule. At the end of a pada, d replaces the final of a word that ends in the affix vasu (and so in s), and the final of sraṃs, dhvaṃs and anaḍuh.

Example.

anaḍuh + bhyām → the stem is a pada before bhyām → h → d → anaḍudbhyām; so also before bhis, bhyas, su.

Applies when / blocked by. "Ending in s" excludes vidvān, where the vasu-stem ends in n at that moment. "At the end of a pada" excludes srastam and dhvastam, where sraṃs and dhvaṃs stand before a suffix inside the word.

SK 335 · सहेः साडः सः · saheḥ sāḍaḥ saḥ (A 8.3.56)

Rule. When the root sah "to overpower" has taken the shape sāḍ, its s is replaced by the retroflex ṣ. The rule stands under the heading "a retroflex for what is not at the end of a pada", the ṣatva heading.

Example. The compound turāsāh "overpowering the strong" (stem turāsāh, with long ā in both members).

turāsāh + su → su deleted → turāsāh → h → ḍh (SK 324) → jaśtva → turāsāḍ → the root now has the shape sāḍ → s → ṣ → turāṣāḍ → optional cartva → turāṣāṭ / turāṣāḍ.

Before a vowel the shape is sāh, not sāḍ, so the s stays: turāsāhau, turāsāhaḥ. The word is formed, in the Veda, with the affix ṇvi by chandasi sahaḥ (SK 3409); in ordinary usage with kvip on the causative sāhayati. The long ā of turā- is by anyeṣām api (SK 539).

SK 336 · दिव औत् · diva aut (A 7.1.84)

Rule. For the nominal stem div "sky", au replaces the final v when su follows.

Example. The compound sudiv "having a good sky".

sudiv + su → v → au → sudi-au + s → i → y before a vowel → sudyau + s → sudyauḥ.

Applies when / blocked by. One might expect su to be deleted, as after a consonant, on the ground that the substitute au should count as the v it replaced. It does not: the deletion of su is a rule about a single sound (an al-vidhi), and in such a rule a substitute is not treated as its original. So su stays and becomes visarga: sudyauḥ. Before the other strong endings the stem is unchanged: sudivau, sudivaḥ, sudivam, sudivau.

SK 337 · दिव उत् · diva ut (A 6.1.131)

Rule. For div, the final v is replaced by u at the end of a pada.

Example.

sudiv + bhyām → the stem is a pada → v → u → sudiu + bhyām → i → y → sudyubhyām; likewise sudyubhiḥ.

Bhaṭṭoji returns here to catur: nominative plural catvāraḥ, accusative plural caturaḥ (no ām, since śas is not a strong ending), instrumental plural caturbhiḥ, dative and ablative plural caturbhyaḥ.

SK 338 · षट्चतुर्भ्यश्च · ṣaṭcaturbhyaśca (A 7.1.55)

Rule. After a word that bears the name ṣaṭ (the numerals in ṣ and n, SK 369) and after catur, the genitive plural ending ām takes the augment nuṭ, an n placed at its front.

Example.

catur + ām → catur + nām → n after r → ṇ (ṇatva) → caturṇām → the ṇ after r is doubled → caturṇṇām.

The Kaumudī prints the genitive plural twice; the translation reads the second form as the undoubled caturṇām, the doubling after r being optional.

SK 339 · रोः सुपि · roḥ supi (A 8.3.16)

Rule. When a case-ending follows — in practice only the locative plural su, since the visarga rule needs a hard consonant after it — visarga replaces only ru (the r that stands for an original s), and not any other r. This is a restriction (niyama): it does not add anything, it only limits.

Example.

catur + su → the r is an original r, not a ru → it is not replaced by visarga → catur + su → the s after r becomes ṣ (ṣatva) → caturṣu → the ṣ after r is now due to be doubled (continued under SK 340).

SK 340 · शरोऽचि · śaro'ci (A 8.4.49)

Rule. A śar sound (ś, ṣ, s) is not doubled when a vowel follows it.

Example.

caturṣu → the ṣ after r would be doubled, but a vowel follows it → no doubling → caturṣu "in the four".

The bahuvrīhi priyacatur "whose four are dear" has a singular: priyacatvāḥ (nominative), he priyacatvaḥ (vocative), priyacatvārau, priyacatvāraḥ. When catur is a subordinate member of such a compound (gauṇa), the augment nuṭ is not wanted: genitive plural priyacaturām. When it is the principal member (paramacatur "the supreme four"), nuṭ does apply: paramacaturṇām.

A stem in l: kamal "one who says kamala or kamalā". Nominative kamal, kamalau, kamalaḥ; in the locative plural the s of su becomes ṣ after l: kamalṣu.

SK 341 · मो नो धातोः · mo no dhātoḥ (A 8.2.64)

Rule. The final m of a root is replaced by n at the end of a pada.

Example. The root-noun praśām "one who is calm", from praśāmyati.

praśām + su → su deleted → praśām → m → n at the end of a pada → praśān.

Applies when / blocked by. The rule deleting a stem-final n would now seem to apply to praśān. It does not, because this m → n is a Tripādī change and therefore asiddha, invisible, to the n-deletion rule which precedes it.

D. The pronouns kim and idam — running example: idam → ayam / imau / ime / anena / asmai / ebhiḥ / eṣām

Two pronouns end in m. kim "who? which?" is simple: before every ending it becomes ka and then declines like sarva (SK 342). idam "this" is the most irregular word of the chapter. Its nominative singular is ayam; before the other strong endings its d becomes m (imau, ime); before the instrumental and the endings after it, its first syllable id becomes an (anena) or vanishes altogether (asmai, ābhyām); and it refuses the instrumental plural ais (ebhiḥ, not aiḥ). When idam refers back to something already mentioned (an anvādeśa), it takes the light forms enam, enena, enayoḥ (SK 350–351). Bhaṭṭoji ends the group by starting the n-stem rājan.

SK 342 · किमः कः · kimaḥ kaḥ (A 7.2.103)

Rule. kim is replaced by ka when a case-ending (vibhakti) follows. The substitute applies even when kim carries the infix akac (the ak that gives the diminutive pronouns).

Example.

kim + su → kim → ka → ka + s → kaḥ; kau, ke (nominative); kam, kau, kān (accusative) — and so on, exactly like sarva (chapter 8), with the dative kasmai.

Scholars discuss one point of the dative, which some read as the ordinary kāya for a proper name ka; the pronoun itself has kasmai.

SK 343 · इदमो मः · idamo maḥ (A 7.2.108)

Rule. For idam, m replaces its m when su follows. This looks empty, but it is an apavāda: the pronouns of the tyad list (tyad, tad, etad, idam, adas, yad, kim …) normally replace their final by a before an ending (tyadādy-atva, A 7.2.102), and this rule keeps the m of idam in the nominative singular instead.

Example.

idam + su → the final m would become a → this rule replaces it by m → idam + su (continued under SK 344).

SK 344 · इदोऽय् पुंसि · ido'y puṃsi (A 7.2.111)

Rule. In the masculine, the segment id of idam is replaced by ay when su follows. Then su is deleted.

Example.

idam + su → id → ay → ayam + su → su deleted after a consonant (SK 252) → ayam "this (man)".

In all other cases the tyad-list a replaces the final m, and the a so produced merges with the a before it (pararūpa): idam → ida + a → ida.

SK 345 · दश्च · daśca (A 7.2.109)

Rule. And the d of idam is replaced by m when a case-ending follows.

Example.

idam + au → final m → a, merged → ida + au → d → m → ima + au → imau (nominative dual).

idam + jas → ima + jas → jas → śī after a pronoun (chapter 8) → ima + ī → guṇa → ime (nominative plural).

Watch out. The pronouns of the tyad list have no vocative; that is the general rule (utsarga), so do not look for he ima.

SK 346 · अनाप्यकः · anāpyakaḥ (A 7.2.112)

Rule. For the id of an idam that carries no k (that is, the plain pronoun, not the akac-form imaka), an is the substitute before an āp ending. āp is a pratyāhāra: the endings from ṭā up to the marker p of sup, that is, the instrumental singular and everything after it.

Example.

idam + ṭā → ida + ṭā (the tyad-list a) → id → an → ana + ṭā → ṭā → ina after an a-stem (chapter 8) → an-a + ina → anena (instrumental singular).

SK 347 · हलि लोपः · hali lopaḥ (A 7.2.113)

Rule. For the k-less idam, the segment id is deleted before an āp ending that begins with a consonant. Bhaṭṭoji's commentators stress that this is a single condition, "before a consonant-initial āp", not two: before the vowel-initial ṭā the an of SK 346 stands.

A vārttika (a supplementary statement) explains why the whole id goes, and not merely its final d: the principle that a substitute replaces only the final sound does not apply to a meaningless segment such as id, except when a reduplicated syllable is being modified.

Example.

idam + bhyām → ida + bhyām → id deleted before the consonant bh → a + bhyām (continued under SK 348).

SK 348 · आद्यन्तवदेकस्मिन् · ādyantavadekasmin (A 1.1.21)

Rule. This is a paribhāṣā: an operation performed on a single element applies as if that element stood at the beginning and as if it stood at the end. So a rule that acts on "the final of a stem" can act on a stem that is one sound long. Bhaṭṭoji's commentators also read the rule as the indication (jñāpaka) for vyapadeśivadbhāva, the treating of a whole as if it were the part (SK 326).

Example.

a + bhyām (from SK 347) → the lone a counts as the final of the stem → it is lengthened before bhyām like the a of rāma → ābhyām (instrumental, dative and ablative dual of idam).

SK 349 · नेदमदसोरकोः · nedamadasorakoḥ (A 7.1.11)

Rule. After idam and adas when they carry no k, ais is not substituted for the instrumental plural bhis. (After an ordinary a-stem, bhis → ais: rāmaiḥ.)

Example.

idam + bhis → ida + bhis → id deleted before the consonant (SK 347) → a + bhis → no ais → the a becomes e before bh (etva, as in rāmebhyaḥ) → ebhiḥ.

The dative singular: ida + ṅe → ṅe → smai after a pronoun; this substitution is obligatory (nitya), so it goes first → ida + smai → now the ending begins with a consonant → id deleted (SK 347) → asmai. The rest of the masculine paradigm:

Applies when / blocked by. With the infix akac the k is present, so SK 346, 347 and this rule are all off, and ais returns: ayakam, imakau, imake; imakam, imakau, imakān; imakena, imakābhyām, imakaiḥ.

SK 350 · इदमोऽन्वादेशेऽशनुदात्तस्तृतीयादौ · idamo'nvādeśe'śanudāttastṛtīyādau (A 2.4.32)

Rule. In an anvādeśa, idam is replaced by aś — an unaccented (anudātta) a — before the instrumental and the endings that follow it. An anvādeśa is a second mention: a thing was named in order to say one thing about it, and is now named again in order to say something further about it. The substitute is stated as aś with the marker ś, rather than as a bare a, so that the akac-form can be built on it.

Example. "anena vyākaraṇam adhītam, enaṃ chando 'dhyāpaya" — "grammar has been studied by him; now teach him the Veda": the second him is an anvādeśa (its form enam is by SK 351).

SK 351 · द्वितीयाटौस्स्वेनः · dvitīyāṭaussvenaḥ (A 2.4.34)

Rule. In an anvādeśa, idam and etad are replaced by ena before the accusative endings, before ṭā (instrumental singular), and before os (genitive-locative dual).

Example. "anayoḥ pavitraṃ kulam, enayoḥ prabhūtaṃ svam" — "the family of these two is pure; the wealth of these two is abundant."

Bhaṭṭoji now opens two further stems. A stem in ṇ: sugaṇ "counting well", from the root gaṇ (gaṇayati) with the affix vic: sugaṇ, sugaṇau, sugaṇaḥ; locative plural sugaṇṭhsu / sugaṇṭsu / sugaṇsu. With the affix kvip instead, the vowel is lengthened by anunāsikasya kvi-jhaloḥ (SK 2666): sugāṇ, sugāṇau, sugāṇaḥ; sugāṇṭhsu / sugāṇṭsu / sugāṇsu. Then the stem in n, rājan "king": rājan + su → the vowel before the n is lengthened before a strong ending (upadhā-dīrgha; it wins as the later rule) → rājān + su → su deleted (hal-ṅyādi-lopa) → the stem-final n is deleted → rājā. The rules on that n follow.

E. The stems in n — the deletion of the final n and its blocks — running example: rājan → rājā / he rājan / rājñe (also yajvan → yajvā, brahman → brahmaṇaḥ)

A nominal stem ending in n loses that n at the end of a pada: rājā, rājabhyām. This deletion (na-lopa, A 8.2.7) is a Tripādī rule, and much of this group is about when it is switched off (SK 352) and when its invisibility matters (SK 353). Before the weak vowel-endings the a of -an is deleted instead (rājñaḥ), unless a cluster ending in v or m precedes (SK 355: yajvanaḥ, brahmaṇaḥ).

SK 352 · न ङिसम्बुद्ध्योः · na ṅisambuddhyoḥ (A 8.2.8)

Rule. The deletion of a stem-final n does not take place before ṅi (locative singular) or in the sambuddhi (vocative singular).

Example.

he rājan + su (vocative) → su deleted → he rājan → the n would go, but this is a sambuddhi → n stays → he rājan "O king!"

For ṅi Bhaṭṭoji's example is Vedic: in parame vyoman "in the highest heaven" the ṅi is dropped by luk, and only the force of this prohibition lets the dropped ending still count (pratyaya-lakṣaṇa), keeping the n. In ordinary usage the locative is rājani or rājñi (SK 353).

A vārttika adds: when the ṅi stands before a second compound-member, the prohibition is itself cancelled, so the n does drop: carmatilaḥ "having sesame on the hide" (from carmaṇi), brahmaniṣṭhaḥ "devoted to brahman" (from brahmaṇi).

The strong forms: rājānau, rājānaḥ, rājānam, rājānau. Before a weak vowel-ending: rājan + śas → the a of -an is deleted (al-lopo 'naḥ, SK 234) → rājn + as → n beside j becomes ñ (ścutva) → rājñaḥ; likewise rājñā. Bhaṭṭoji explains why the deleted a does not stand in the way of the ścutva: in the Tripādī a substitute is not treated as its original, and the bahiraṅga principle (an outer operation is invisible to an inner one) does not apply here either; scholars discuss the second point.

SK 353 · नलोपः सुप्स्वरसंज्ञातुग्विधिषु कृति · nalopaḥ supsvarasaṃjñātugvidhiṣu kṛti (A 8.2.2)

Rule. The deletion of the final n is asiddha, treated as not having happened, only for four kinds of operation: one conditioned by a case-ending (sub-vidhi), an accent operation (svara-vidhi), a naming operation (saṃjñā-vidhi), and the insertion of the augment tuk before a kṛt affix. Everywhere else the deletion is fully real (siddha).

Example.

rājan + bhyām → n deleted at the end of a pada → rāja + bhyām → the lengthening of a final a before bhyām is a sub-vidhi, and for it the n is still there → no lengthening → rājabhyām (not rājābhyām).

So also rājabhiḥ (no ais), rājabhyaḥ (no e).

Applies when / blocked by. Outside the four kinds the deletion is real. In the compound rājāśva "king's horse", rāja + aśva merge by savarṇa-dīrgha as though the n had never been. — Bhaṭṭoji here adds the stem pratidivan "gambler", from pratidīvyati: pratidivā, pratidivānau, pratidivānaḥ; its bha-forms need the next rule.

SK 354 · हलि च · hali ca (A 8.2.77)

Rule. And before a consonant: the ik vowel (i, u, ṛ, ḷ) that is the penult of a root ending in r or v is lengthened before a consonant as well. The main rule, r̥vor upadhāyā dīrgha ikaḥ (SK 433), lengthens it at the end of a pada; this extends it.

Example.

pratidivan + śas → the a of -an is deleted (SK 234) → pratidivn + as → the root-portion div now ends in v with a consonant after it → its penult i → ī → pratidīvnaḥ; likewise pratidīvne.

Applies when / blocked by. The deleted a is not treated as still standing between v and n: a substitute does not count as its original in a lengthening rule. Nor does the bahiraṅga principle block the lengthening, for the reason already given under SK 352. — A root-noun in van: yajvan "sacrificer": yajvā, yajvānau, yajvānaḥ.

SK 355 · न संयोगाद्वमन्तात् · na saṃyogādvamantāt (A 6.4.137)

Rule. The a of -an is not deleted after a consonant-cluster that ends in v or m.

Example.

yajvan + śas → the a follows the cluster jv, which ends in v → a stays → yajvanaḥ; yajvanā; before a consonant the n is deleted: yajvabhyām.

brahman + śas → the cluster hm ends in m → brahmaṇaḥ (with n → ṇ after the r), brahmaṇā, brahmabhyām.

F. The stems in in, han, van — the han guttural, num for an ugit, and saṃprasāraṇa — running example: vṛtrahan → vṛtrahā / vṛtrahaṇau / vṛtraghnaḥ (also maghavan, yuvan, arvan)

The word vṛtrahan "slayer of Vṛtra" declines like rājan in its broad shape but with three special changes: its penult is lengthened only in the nominative singular (SK 356–357), its n becomes ṇ after the r of vṛtra (vṛtrahaṇau), and in the weak forms its h becomes gh before n (SK 358–359: vṛtraghnaḥ). Three stems in van follow. maghavan "Indra" may exchange its van for vat and then take the nasal augment num (SK 360–361: maghavān); in the weak forms its va becomes u by saṃprasāraṇa (SK 362: maghonaḥ), as does the va of śvan "dog" and yuvan "young man" (śunaḥ, yūnaḥ, SK 363). arvan "horse" takes tṛ everywhere except before su (SK 364).

SK 356 · इन्हन्पूषार्यम्णां शौ · inhanpūṣāryamṇāṃ śau (A 6.4.12)

Rule. For stems ending in in, for han, pūṣan and aryaman, the lengthening of the penult takes place only before śi (the neuter nominative-accusative plural), and not before the other strong endings. This is a restriction on the general lengthening before a strong ending that chapter 8 taught.

Example.

vṛtrahan + au → the penult a would be lengthened before the strong ending → this restriction forbids it → vṛtrahanau (finished as vṛtrahaṇau below).

The restriction covers the nominative singular too, so the lengthening there must be re-enjoined by the next rule.

SK 357 · सौ च · sau ca (A 6.4.13)

Rule. And before su, when it is not the vocative, the penult of these stems is lengthened.

Example.

vṛtrahan + su → penult a → ā → vṛtrahān + su → su deleted → n deleted at the end of a pada → vṛtrahā.

Vocative: he vṛtrahan (no lengthening, and the n stays in a sambuddhi, SK 352).

vṛtrahan + au → no lengthening (SK 356) → the n after the r of vṛtra becomes ṇ, since the second member han has one vowel (ekāj-uttarapade, SK 307) → vṛtrahaṇau; likewise vṛtrahaṇaḥ, vṛtrahaṇam, vṛtrahaṇau.

SK 358 · हो हन्तेर्ञ्णिन्नेषु · ho hanterñṇinneṣu (A 7.3.54)

Rule. The h of the root han "to strike, kill" is replaced by a guttural (ku, here gh) before an affix marked with ñ or ṇ, and before n. The change of a sound to its guttural is called kutva.

Example.

vṛtrahan + śas → the a of -an deleted (SK 234) → vṛtrahn + as → h stands before n → h → gh → vṛtraghn + as (finished under SK 359).

SK 359 · हन्तेरत्पूर्वस्य · hanteratpūrvasya (A 8.4.22)

Rule. Bhaṭṭoji reads this rule as two rules (yoga-vibhāga, splitting a rule). First, hanteḥ: the n of han becomes ṇ when an r or ṣ in a preverb precedes it — and the n of nothing else. Second, atpūrvasya: only the n of han that is preceded by a short a becomes ṇ, not otherwise.

Example.

pra + han + yāt → the r of the preverb pra precedes; the n follows short a → prahaṇyāt "may he kill".

pra + ghn + anti → the a of han has been deleted, so the n is no longer preceded by short a → no ṇ → praghnanti "they kill". The splitting of the rule is what lets it override the maxim that a rule or prohibition attaches to the nearest item; it also cancels, for praghnanti, the ṇ that SK 307 would give.

vṛtraghn + as (from SK 358) → the deleted a is not treated as still standing between h and n, since otherwise the guttural rule would never get to act before n → vṛtraghnaḥ; likewise vṛtraghnā.

Scholars discuss an optional ṇ in vṛtraghnaḥ; Bhaṭṭoji rejects it as contrary to the Bhāṣya and the vārttika. The same pattern serves śārṅgin, yaśasvin, aryaman, pūṣan. In yaśasvin the in is part of the affix vin and has no meaning of its own, yet it is caught by the "in" of SK 356, because a paribhāṣā says that the mentions of an, in, as, man cover stems ending in them whether the segment is meaningful or not. Locative singular: aryamṇi / aryamaṇi, pūṣṇi / pūṣaṇi.

SK 360 · मघवा बहुलम् · maghavā bahulam (A 6.4.128)

Rule. For the stem maghavan "Indra, the bountiful", tṛ — that is t, the ṛ being a marker — is variously (bahulam, in effect optionally) the substitute for the final. The marker ṛ makes the substitute an ugit, an item whose marker is a vowel of the uk group, and that matters for the next rule.

Example.

maghavan → final n → t → maghavat (continued under SK 361). When the substitution is not made, the nominative singular is maghavā (like rājā).

SK 361 · उगिदचां सर्वनामस्थानेऽधातोः · ugidacāṃ sarvanāmasthāne'dhātoḥ (A 7.1.70)

Rule. An ugit stem that is not a root, and the root añc when it has lost its n, take the augment num (an n after the last vowel) when a strong ending follows. Then the penult is lengthened (upadhā-dīrgha).

Example.

maghavat + su → the stem is ugit (SK 360) and not a root; su is a strong ending → num after the last vowel → maghavant + su → su deleted → the cluster nt loses its t (SK 54) → maghavan → penult lengthened → maghavān.

Bhaṭṭoji notes that the loss of the t would normally be invisible to the lengthening, being a Tripādī change; here the word bahulam of SK 360 lets the lengthening go through. Bhaṭṭi attests the form: "makheṣu maghavān asau" — "that Maghavān at the sacrifices".

Without the t the strong forms are those of rājan: maghavānau, maghavānaḥ. A vārttika adds that the form made with the affix vanip and accented on its middle syllable is Vedic only; the form accented on its last syllable occurs in ordinary speech too. Scholars discuss whether SK 360 is needed at all, since both maghavan and maghavat can be formed otherwise.

SK 362 · श्वयुवमघोनामतद्धिते · śvayuvamaghonāmataddhite (A 6.4.133)

Rule. For śvan "dog", yuvan "young" and maghavan — when they end in an, are bha (before a weak vowel-ending), and are not followed by a taddhita (secondary) affix — saṃprasāraṇa takes place: the v becomes u. Then SK 330 merges the following a into the u, and ād guṇaḥ (SK 69) turns a + u into o.

Example.

maghavan + śas → v → u → magha-u-an + as → u + a → u (pūrvarūpa, SK 330) → magha-un + as → a + u → o (SK 69) → maghonaḥ; likewise maghonā; before a consonant, maghavabhyām (the n is deleted, and no saṃprasāraṇa, the stem not being bha).

śvan + śas → śu-an + as → śunaḥ; śunā; śvabhyām.

Applies when / blocked by. "Ending in an" excludes the vat-forms maghavataḥ, maghavatā and the feminine maghavatī. "Not before a taddhita" excludes māghavanam "belonging to Maghavan". For yuvan the v → u is done, and then a question arises about its y (next rule).

SK 363 · न सम्प्रसारणे सम्प्रसारणम् · na samprasāraṇe samprasāraṇam (A 6.1.37)

Rule. When a saṃprasāraṇa vowel follows, the semivowel before it does not undergo saṃprasāraṇa. So once the v of yuvan has become u, the y before it does not become i. Bhaṭṭoji draws a further lesson from this rule: it indicates that, of two semivowels, the final one is turned into a vowel first.

Example.

yuvan + śas → the final semivowel v → u → yu-u-an + as → the y does not become i → u + a → u (SK 330) → yu-un + as → the two u merge → yūnaḥ; likewise yūnā; before a consonant yuvabhyām.

The stem arvan "horse": arvā (nominative singular), he arvan (vocative).

SK 364 · अर्वणस्त्रसावनञः · arvaṇastrasāvanañaḥ (A 6.4.127)

Rule. For a stem ending in arvan, unless the negative nañ (a-) precedes, tṛ replaces the final before every ending except su. Being ugit, the resulting arvat takes num before the strong endings (SK 361).

Example.

arvan + au → n → t → arvat + au → num → arvantau; arvantaḥ, arvantam, arvantau; weak arvataḥ, arvatā, arvadbhyām (t → d by jaśtva).

Applies when / blocked by. With nañ the rule is off: anarvan "unopposed" gives anarvā and then declines like yajvan (SK 355).

G. pathin, mathin, ṛbhukṣin — running example: pathin → panthāḥ / pathaḥ

Three stems in in — pathin "road", mathin "churning-stick", ṛbhukṣin "Indra" — build their strong forms by three substitutions at once: the final becomes ā before su (SK 365), the i becomes a before every strong ending (SK 366), and the th becomes nth (SK 367). So pathin gives panthāḥ, panthānau. In the weak forms the whole final syllable in is dropped (SK 368): pathaḥ, pathā, pathibhyām.

SK 365 · पथिमथ्यृभुक्षामात् · pathimathyṛbhukṣāmāt (A 7.1.85)

Rule. For pathin, mathin and ṛbhukṣin, ā replaces the final sound (antādeśa, a final-substitute) when su follows. Bhaṭṭoji reads the rule's āt as containing a second ā folded into it (praśleṣa), so that only the plain, non-nasal ā is prescribed.

Example.

pathin + su → final n → ā → pathi-ā + s (continued below).

SK 366 · इतोऽत् सर्वनामस्थाने · ito't sarvanāmasthāne (A 7.1.86)

Rule. For these three stems, the i is replaced by short a when a strong ending follows.

Example.

pathi-ā + s → i → a → patha-ā + s.

pathin + au → i → a → pathan + au.

SK 367 · थो न्थः · tho nthaḥ (A 7.1.87)

Rule. The th of pathin and mathin is replaced by nth when a strong ending follows.

Example.

patha-ā + s → th → nth → pantha-ā + s → a + ā → ā → panthā + s → panthāḥ "road" (nominative singular).

pathan + au → panthan + au → penult lengthened before a strong ending → panthānau.

SK 368 · भस्य टेर्लोपः · bhasya ṭerlopaḥ (A 7.1.88)

Rule. A stem such as pathin, when it is bha, loses its ṭi. The ṭi of a word is its last vowel together with whatever follows it; here the ṭi is in.

Example.

pathin + śas → the stem is bha → in deleted → path + as → pathaḥ; pathā; before a consonant the ordinary n-deletion gives pathibhyām.

Likewise manthāḥ from mathin, and ṛbhukṣāḥ from ṛbhukṣin. In the feminine the affix ṅīp (ī) is added because the stem ends in n; before it the stem is bha, so the ṭi goes: supathī "having good roads", sumathī, anṛbhukṣī senā. In the neuter there is no ā (SK 365), because the ending su is removed by luk, and na lumatāṅgasya (SK 263) forbids a luk-deleted ending to condition an operation on the stem: supathi vanam "a forest with good paths".

A vārttika adds that a neuter in the vocative singular optionally drops its n: he supathin / he supathi. Because the n-deletion is asiddha for a sub-vidhi (SK 353), the short i does not take guṇa before the vocative. The neuter dual drops the ṭi (bha): supathī; before śi, a strong ending, the full apparatus works: supanthāni. So supathi, supathī, supanthāni; supathā, supathe, supathibhyām.

H. The ṣaṭ numerals — pañcan, aṣṭan — running example: pañcan → pañca / pañcānām; aṣṭan → aṣṭau

The numerals from five to ten are pañcan, ṣaṣ, saptan, aṣṭan, navan, daśan: all end in n except ṣaṣ, which ends in ṣ. They are given the name ṣaṭ (SK 369), and a ṣaṭ word loses its nominative and accusative plural endings altogether: pañca. Its genitive plural takes the augment nuṭ and lengthens the vowel before the n: pañcānām (SK 370). aṣṭan has in addition the option of a final ā, which gives aṣṭau in the nominative-accusative and aṣṭābhiḥ, aṣṭāsu elsewhere (SK 371–372).

SK 369 · ष्णान्ता षट् · ṣṇāntā ṣaṭ (A 1.1.24)

Rule. A numeral (saṅkhyā) ending in ṣ or in n is named ṣaṭ. Then ṣaḍbhyo luk (SK 261): after a ṣaṭ word the endings jas and śas are deleted by luk.

Example.

pañcan + jas → pañcan is ṣaṭ → jas deleted by luk → pañcan → n deleted at the end of a pada → pañca "five" (nominative); likewise pañca (accusative).

pañcabhiḥ, pañcabhyaḥ, pañcabhyaḥ (the n deleted before the consonant); the genitive plural takes nuṭ by SK 338.

Applies when / blocked by. "Numeral" excludes vipruṣaḥ "drops" and pāmānaḥ "skin-diseases", which end in ṣ and n but count nothing. In śatāni "hundreds" and sahasrāṇi "thousands" the luk does not apply: the augment num that these neuters take depends on the strong ending śi, and an operation that depends on a condition does not destroy that condition (the sannipāta principle).

SK 370 · नोपधायाः · nopadhāyāḥ (A 6.4.7)

Rule. The penult of a stem ending in n is lengthened when nām (the genitive plural with nuṭ) follows. Then the stem-final n is deleted.

Example.

pañcan + ām → nuṭ (SK 338) → pañcan + nām → penult a → ā → pañcān + nām → final n deleted → pañcānām.

The compound paramapañcan "the supreme five" behaves the same: paramapañca, paramapañcānām. But when the numeral is a subordinate member (gauṇa), as in the bahuvrīhi priyapañcan "whose five are dear", there is no luk and no nuṭ: priyapañcā, priyapañcānau, priyapañcānaḥ, priyapañcām. saptan, navan, daśan decline like pañcan.

SK 371 · अष्टन आ विभक्तौ · aṣṭana ā vibhaktau (A 7.2.84)

Rule. For aṣṭan "eight", ā replaces the final when a case-ending beginning with a consonant follows.

Example.

aṣṭan + bhis → final n → ā → aṣṭā + bhis → aṣṭābhiḥ.

SK 372 · अष्टाभ्य औश् · aṣṭābhya auś (A 7.1.21)

Rule. After aṣṭan in its ā-form, auś replaces jas and śas. Being marked with ś, the substitute au replaces the whole ending (sarvādeśa), not just its first sound.

Example.

aṣṭan + jas → final → ā → aṣṭā + jas → jas → au → aṣṭau "eight" (nominative); likewise aṣṭau (accusative); paramāṣṭau.

Applies when / blocked by. Two things are read off the wording. The rule says "after aṣṭā" where "after aṣṭa" would have sufficed; that shows that the ā of SK 371 applies before jas and śas too, although they begin with a vowel. And the ā is optional: the rule aṣṭano dīrghāt (SK 3718) speaks of "aṣṭan with a long vowel", which would be idle if the long vowel were compulsory. Without the ā the word declines like pañcan: aṣṭa, aṣṭa. As a subordinate compound-member without the ā it declines like rājan: accusative plural priyāṣṭnaḥ — with no change of n to ṇ after ṣ, because the deleted a is treated as still there for this operation, or because its deletion is an outer operation and invisible; scholars discuss which. priyāṣṭā and so on. The ā before jas and śas holds only when aṣṭan is the principal member; for priyāṣṭan the optional ā appears only before a consonant: priyāṣṭābhyām, priyāṣṭābhiḥ, priyāṣṭābhyaḥ, priyāṣṭāsu.

Bhaṭṭoji ends with a stem in dh: budh "knower". By SK 326 the b becomes bh when the dh stands at the end of a pada, then jaśtva and cartva: bhut / bhud; before vowels budhau, budhau, budhaḥ, budhā; bhudbhyām (three times), bhutsu.

I. Root-nouns in kvin and kvip, and the guttural rules — then the d-final tyad pronouns — running example: yuj → yuṅ (alone) / suyuk (in a compound); ṛtvij → ṛtvik, rāj → rāṭ; then tyad → syaḥ

Many consonant-stems are bare roots turned into nouns by an affix that is deleted entirely: kvip or kvin, both reduced to nothing, leaving the root to take the case-endings. The difference between them shows up at the end of a word: a kvin-stem replaces its final by a guttural (kutva, SK 377: yuj → yuṅ, ṛtvij → ṛtvik), while any palatal at the end of a word becomes a guttural anyway (SK 378: suyuk), or, for rāj and a few others, becomes ṣ → ḍ → ṭ (rāṭ). A cluster's first s or k drops at the end of a word (SK 380: bhṛṭ). Bhaṭṭoji then begins the pronouns in d, whose t/d becomes s in the nominative singular (SK 381: tyad → syaḥ, tad → saḥ).

SK 373 · ऋत्विग्दधृक्स्रग्दिगुष्णिगञ्चुयुजिक्रुञ्चां च · ṛtvigdadhṛksragdiguṣṇigañcuyujikruñcāṃ ca (A 3.2.59)

Rule. The affix kvin comes after the roots that yield ṛtvij "priest" (from yaj with ṛtu), dadhṛś "bold", sraj "garland" (from sṛj), diś "direction", uṣṇih (a metre), añc "to go, bend", yuj "to join", and kruñc "curlew". The k and the n of kvin are markers. Because the words are listed ready-made (nipātana), any irregularity in them that no rule accounts for is licensed by the listing. yuj takes kvin only when it has no accompanying word (upapada).

Example.

yuj + kvin → the markers k and n fall away → yuj + vi (continued under SK 375).

SK 374 · कृदतिङ् · kṛdatiṅ (A 3.1.93)

Rule. A saṃjñā: an affix prescribed under the heading "after a root" (the dhātu-adhikāra), other than a verb-ending (tiṅ), is named kṛt. So kvin and kvip are kṛt affixes, and what they form is a nominal stem.

Example.

yuj + vi → vi was prescribed after a root and is not a tiṅ → vi is a kṛt affix.

SK 375 · वेरपृक्तस्य · verapṛktasya (A 6.1.67)

Rule. The affix vi, when it is apṛkta — reduced to a single sound, its i being a marker — is deleted. What remains, a root plus a kṛt affix, is a nominal stem (prātipadika) by kṛt-taddhita-samāsāś ca (SK 179), and so takes the case-endings su and the rest.

Example.

yuj + vi → v is apṛkta → deleted → yuj → a prātipadika → yuj + su (continued under SK 376).

SK 376 · युजेरसमासे · yujerasamāse (A 7.1.71)

Rule. The root yuj takes the augment num before a strong ending, provided it is not in a compound. Then su is deleted, and the final of the cluster is deleted (SK 54).

Example.

yuj + su → num after the last vowel → yunj + su → su deleted → yunj → the cluster nj loses its j → yun (continued under SK 377).

SK 377 · क्विन्प्रत्ययस्य कुः · kvinpratyayasya kuḥ (A 8.2.62)

Rule. For a stem after which kvin was added, the final sound at the end of a pada is replaced by the corresponding guttural (ku, the k-class). The nasal n of num becomes the guttural nasal ṅ.

Example.

yun → the stem is a kvin-stem; its final n is at the end of a pada → n → ṅ → yuṅ "joiner" (nominative singular).

yuj + au → num → yunj + au → the n before j becomes anusvāra (naś cāpadāntasya) and then the palatal nasal ñ (parasavarṇa) → yuñjau.

Applies when / blocked by. In yuñjau the ñ stands before j, a jhal, and SK 378 would turn a palatal before a jhal into a guttural. It does not, because the anusvāra and parasavarṇa are later Tripādī rules and so asiddha to SK 378. The accusative plural has no num, śas not being a strong ending: yujaḥ. Before bhyām the j becomes g (this rule): yugbhyām. Why "not in a compound"? See the next rule.

SK 378 · चोः कुः · coḥ kuḥ (A 8.2.30)

Rule. A palatal (cu: c, ch, j, jh, ñ) is replaced by the corresponding guttural (ku) before a jhal and at the end of a pada.

Example. The compound suyuj "joining well": in a compound there is no num (SK 376).

suyuj + su → su deleted → suyuj → j at the end of a pada → j → g → suyug → optional cartva → suyuk / suyug; suyujau, suyujaḥ.

The yuj of SK 376 is the root as read in the root-list with an i, yuji; the rule does not cover the other root yuj "to concentrate" (class 4, middle voice), so "one who concentrates" is yuk without num. A stem in ñj: khañj "limping": nominative khan (the cluster loses its j), khañjau, khañjaḥ.

Some palatal stems take ṣ instead of a guttural, by vraśca-bhrasja… (SK 294); the ṣ then becomes ḍ by jaśtva and ṭ by cartva. rāj + su → rāj → j → ṣ → rāṣ → rāḍ → rāṭ / rāḍ "king"; rājau, rājaḥ; locative plural rāṭsu / rāṭtsu. Likewise vibhrāṭ "shining"; deveṭ, devejau, devejaḥ "worshipping the gods"; viśvasṛṭ / viśvasṛḍ, viśvasṛjau, viśvasṛjaḥ "all-creator". (That the guttural does not apply to sṛj and yaj is stated in the neuter chapter.) parimṛṭ "wiping round". Of the two roots bhrāj, only ṭubhrājṛ "to shine" of the phaṇ group takes ṣ, because it is listed beside rāj; the bhrājṛ of the list ejṛ bhrejṛ bhrājṛ takes the guttural: vibhrāk / vibhrāg, vibhrāgbhyām. And parivrāj "wandering ascetic", formed by an Uṇādi rule (parau vrajeḥ ṣaḥ padānte) with kvip, a long vowel and ṣ at the end of a pada: parivrāṭ, parivrājau, parivrājaḥ.

SK 379 · विश्वस्य वसुराटोः · viśvasya vasurāṭoḥ (A 6.3.128)

Rule. The final a of viśva "all" is lengthened when vasu "wealth" or the word rāj (cited in its pada-final form rāṭ) follows.

Example.

viśva + vasu → a → ā → viśvāvasu "whose every wealth is such" (a name).

viśva + rāj + su → viśvārāṭ / viśvārāḍ "universal king"; but viśvarājau, viśvarājaḥ (no lengthening, since rāj is not in its pada-final form there); viśvārāḍbhyām.

The rāṭ of the rule only marks the pada-final shape of rāj; the ṭ of cartva is not meant literally, so viśvārāḍ is covered.

SK 380 · स्कोः संयोगाद्योरन्ते च · skoḥ saṃyogādyorante ca (A 8.2.29)

Rule. When a consonant-cluster stands at the end of a pada or before a jhal, an s or a k that begins the cluster is deleted. The word "deletion" is carried in from saṃyogāntasya lopaḥ.

Example. The root bhrasj "to fry", as the stem bhṛsj.

bhṛsj + su → su deleted → bhṛsj → the s beginning the final cluster is deleted → bhṛj → j → ṣ (SK 294) → jaśtva, cartva → bhṛṭ / bhṛḍ.

Before a vowel the cluster survives: bhṛsj + au → s beside j → ś (ścutva) → ś → j (jaśtva) → bhṛjjau, bhṛjjaḥ.

ṛtvij "priest": kvin after yaj with ṛtu as its accompanying word (SK 373); the j becomes g at the end of a pada (SK 377): ṛtvik / ṛtvig, ṛtvijau, ṛtvijaḥ. ūrj "strength": the restriction rāt sasya (SK 280) says that after r only an s is deleted from a cluster, so the j of ūrj survives and becomes a guttural: ūrk / ūrg, ūrjau, ūrjaḥ.

Now the stems in d, beginning with the pronoun tyad "that": before an ending its final d becomes a (tyadādy-atva), and that a merges with the a before it (pararūpa): tyad + su → tya + a + s → tya + s.

SK 381 · तदोः सः सावनन्त्ययोः · tadoḥ saḥ sāvanantyayoḥ (A 7.2.106)

Rule. For the pronouns of the tyad list, a t or d that is not the final sound is replaced by s when su follows.

Example.

tya + s (from above) → the t is not final → t → s → sya + s → syaḥ "that (man)".

tad + su → ta + s → saḥ; tau, te. etad + su → eṣaḥ (the s after e becomes ṣ); etau, ete. yad + su → yaḥ (no t or d to change); yau, ye.

Applies when / blocked by. In a compound with the pronoun as final member the rule still acts: paramasaḥ, paramatau, paramate. A vārttika, cited by Bhaṭṭoji, restricts the rule to the pronouns up to dvi in the list, so yuṣmad is not touched: tvam, not svam. Used as a proper name or as a subordinate member the tyad pronouns take neither the a nor the s: tyad, tyadau, tyadaḥ; atityad, atityadau, atityadaḥ. In an anvādeśa etad takes ena (SK 351): enam, enau, enān, enena, enayoḥ.

J. yuṣmad and asmad, I — the nominative and accusative substitutions — running example: yuṣmad → tvam / yuvām / yūyam / tvām / yuṣmān (asmad → aham / āvām / vayam / mām / asmān)

The pronouns yuṣmad "you" and asmad "I, we" are stems in d, and they are the same for all three genders. Their forms are built by replacing the beginning of the stem — everything up to and including its m (SK 383) — by short substitutes chosen by case and number (tva, aha, yuva, āva, yūya, vaya, ma), while the tail ad is either deleted (SK 385) or turned into ā (SK 387, 390) and the ending itself is often replaced by am (SK 382). The nominative and accusative are handled in this group; the other cases in the next.

SK 382 · ङे प्रथमयोरम् · ṅe prathamayoram (A 7.1.28)

Rule. After yuṣmad and asmad, am replaces the dative singular ṅe and every ending of the first two cases (the nominative su, au, jas and the accusative am, auṭ, śas).

Example.

yuṣmad + su → su → am → yuṣmad + am (continued under SK 384–385).

SK 383 · मपर्यन्तस्य · maparyantasya (A 7.2.91)

Rule. An adhikāra, a heading: "of the portion up to and including the m". The substitutions of the following rules replace that part of yuṣmad / asmad which ends in its m — yuṣm and asm — and leave the tail ad to other rules.

Example. In yuṣm-ad, the substitutes tva, yuva, yūya … replace yuṣm; in asm-ad the substitutes aha, āva, vaya, ma … replace asm.

SK 384 · त्वाहौ सौ · tvāhau sau (A 7.2.94)

Rule. Before su, tva replaces yuṣm and aha replaces asm.

Example.

yuṣm-ad + am → tva-ad + am; asm-ad + am → aha-ad + am (continued under SK 385).

SK 385 · शेषे लोपः · śeṣe lopaḥ (A 7.2.90)

Rule. In the remaining cases — before any ending that does not call for the ā or the y substitute of SK 390–393 — the final sound of yuṣmad and asmad is deleted. What is left is finished by ato guṇe (SK 191, a + a → a) and ami pūrvaḥ (SK 194, the a before am absorbs the a of am).

Example.

tva-ad + am → the final d deleted → tva-a + am → a + a → a (SK 191) → tva + am → the a absorbs the a of am (SK 194) → tvam "you".

aha-ad + am → aham "I".

Watch out. In tvaṃ strī "you, a woman", one might expect the feminine affix ṭāp to attach to the a-final tva before am is dealt with. It does not, and Bhaṭṭoji gives two answers: yuṣmad and asmad have no gender, so there is no femininity to mark; or the deletion strikes the whole remainder ad after the m, and only after ato guṇe has acted, so no a-final stem ever stands. Compounds: paramatvam, paramāham, atitvam, atyaham.

SK 386 · युवावौ द्विवचने · yuvāvau dvivacane (A 7.2.92)

Rule. When two are meant, yuva replaces yuṣm and āva replaces asm, before a case-ending.

Example.

yuṣm-ad + au → yuva-ad + au; asm-ad + au → āva-ad + au (continued under SK 387).

SK 387 · प्रथमायाश्च द्विवचने भाषायाम् · prathamāyāśca dvivacane bhāṣāyām (A 7.2.88)

Rule. And in the nominative dual, in the ordinary language (bhāṣā), the final of yuṣmad / asmad is replaced by ā.

Example.

yuva-ad + au → au → am (SK 382) → yuva-ad + am → final d → ā → yuva-a-ā + am → the vowels merge → yuvā + am → yuvām "you two"; likewise āvām "we two".

Applies when / blocked by. "In the ordinary language" excludes the Veda: yuvaṃ vastrāṇi has the short form. Bhaṭṭoji remarks that "before auṭ" would have been a simpler wording, since the accusative dual needs the same ā (SK 390). The heading "up to the m" (SK 383) is what makes the akac-forms come out right: yuvakām, āvakām, yuvakābhyām, āvakābhyām; and it keeps the instrumental as tvayā, mayā, not tvyā, myā.

SK 388 · यूयवयौ जसि · yūyavayau jasi (A 7.2.93)

Rule. Before jas, yūya replaces yuṣm and vaya replaces asm.

Example.

yuṣm-ad + jas → jas → am (SK 382) → yūya-ad + am → final d deleted (SK 385), a + a → a, am absorbed → yūyam "you (all)"; likewise vayam "we"; paramayūyam, paramavayam, atiyūyam, ativayam.

Watch out. After the deletion the stem ends in a and is a pronoun, so jas → śī (as in sarve) might seem due. It is not: once an operation on the stem has been done, no further operation on the stem is performed for the same ending (the aṅga-kārya paribhāṣā); or, on another reading of SK 382, the substitute is fixed as am with an extra m folded in and cannot be modified.

SK 389 · त्वमावेकवचने · tvamāvekavacane (A 7.2.97)

Rule. When one is meant, tva replaces yuṣm and ma replaces asm, before a case-ending.

Example.

yuṣm-ad + am (accusative singular) → tva-ad + am; asm-ad + am → ma-ad + am (continued under SK 390).

SK 390 · द्वितीयायां च · dvitīyāyāṃ ca (A 7.2.87)

Rule. And in the accusative, the final of yuṣmad / asmad is replaced by ā.

Example.

tva-ad + am → d → ā → tvā + am → tvām "you"; ma-ad + am → mām "me"; dual yuvām, āvām.

SK 391 · शसो न · śaso na (A 7.1.29)

Rule. After yuṣmad and asmad, na — that is, n — replaces the accusative plural śas; the na of the rule is cited without a case-ending. This is an apavāda to the am of SK 382. By ādeḥ parasya (SK 44) a substitute stated for a following element replaces only its first sound; then the final of the cluster is deleted (SK 54).

Example.

yuṣmad + śas → the n replaces the first sound of śas (as) → yuṣmad + ns → the stem-final becomes ā before the consonant, as in yuṣmābhiḥ (SK 393) → yuṣmā + ns → the cluster ns loses its s → yuṣmān "you (all)"; likewise asmān "us".

K. yuṣmad and asmad, II — the oblique substitutions — running example: tvayā / yuvābhyām / tubhyam / tvat / yuṣmat / tava / yuṣmākam

From the instrumental onward the tail ad of the stem becomes y before a vowel-initial ending (tvayā, SK 392) and ā before a consonant-initial one (yuvābhyām, SK 393), unless the ending itself has been replaced. Four endings are replaced: the dative singular by am with the stems tubhya / mahya (SK 394), the dative plural bhyas by bhyam (SK 395), the ablative singular and plural by at (SK 396–397), and the genitive singular by aś with the stems tava / mama (SK 398–399); the genitive plural becomes ākam (SK 400).

SK 392 · योऽचि · yo'ci (A 7.2.89)

Rule. For yuṣmad and asmad, y replaces the final before a vowel-initial ending that is not itself a substitute.

Example.

yuṣm-ad + ṭā → singular, so tva for yuṣm (SK 389) → tva-ad + ā → d → y → tvayā "by you"; mayā "by me".

SK 393 · युष्मदस्मदोरनादेशे · yuṣmadasmadoranādeśe (A 7.2.86)

Rule. For yuṣmad and asmad, ā replaces the final before a consonant-initial ending that is not a substitute.

Example.

yuva-ad + bhyām → d → ā → yuvābhyām; āvābhyām; yuṣmābhiḥ, asmābhiḥ (plural: no stem-substitute, so yuṣm and asm stand).

SK 394 · तुभ्यमह्यौ ङयि · tubhyamahyau ṅayi (A 7.2.95)

Rule. Before the dative singular ṅe, tubhya replaces yuṣm and mahya replaces asm. The ending is replaced by am (SK 382), and the remainder is deleted (SK 385).

Example.

yuṣm-ad + ṅe → tubhya-ad + am → tubhyam "to you"; mahyam "to me"; paramatubhyam, paramamahyam, atitubhyam, atimahyam. Dative dual: yuvābhyām, āvābhyām (SK 393).

SK 395 · भ्यसो भ्यम् · bhyaso bhyam (A 7.1.30)

Rule. After yuṣmad and asmad, bhyam — or abhyam — replaces the dative plural bhyas.

Example.

yuṣm-ad + bhyas → bhyas → bhyam → final d deleted (SK 385) → yuṣmabhyam "to you (all)"; asmabhyam.

Bhaṭṭoji explains the alternative. On the reading that SK 385 deletes only the final d, the stem ends in a, and bhyam is the substitute; the e that an a-stem takes before bhy (rāmebhyaḥ) does not appear, by the paribhāṣā that bars a second stem-operation. The substitute abhyam gives the right form on either reading of SK 385.

SK 396 · एकवचनस्य च · ekavacanasya ca (A 7.1.32)

Rule. And for the singular of the ablative (ṅasi) after yuṣmad and asmad, the substitute is at.

Example.

yuṣm-ad + ṅasi → tva-ad + at → tvat "from you"; mat "from me". Ablative dual: yuvābhyām, āvābhyām.

Bhaṭṭoji remarks that "ṅaseś ca" would have been a neater wording. Scholars discuss the exact steps by which the three a's of tva-ad + at reduce to one; the form is tvat.

SK 397 · पञ्चम्या अत् · pañcamyā at (A 7.1.31)

Rule. After yuṣmad and asmad, at replaces the ablative plural bhyas.

Example.

yuṣmad + bhyas (ablative) → bhyas → at → yuṣmat "from you (all)"; asmat "from us".

SK 398 · तवममौ ङसि · tavamamau ṅasi (A 7.2.96)

Rule. Before the genitive singular ṅas, tava replaces yuṣm and mama replaces asm.

Example.

yuṣm-ad + ṅas → tava-ad + ṅas; asm-ad + ṅas → mama-ad + ṅas (continued under SK 399).

SK 399 · युष्मदस्मद्भ्यां ङसोऽश् · yuṣmadasmadbhyāṃ ṅaso'ś (A 7.1.27)

Rule. After yuṣmad and asmad, aś — that is, a — replaces the genitive singular ṅas.

Example.

tava-ad + a → the vowels reduce to one → tava "your"; mama "my". Genitive-locative dual: yuvayoḥ, āvayoḥ (y before the vowel of os, SK 392).

SK 400 · साम आकम् · sāma ākam (A 7.1.33)

Rule. After yuṣmad and asmad, ākam replaces sām — the genitive plural ām together with the augment suṭ (s) that a pronoun's ām takes (sarveṣām). The rule names the ending with its suṭ so that no suṭ is added afterwards.

Example.

yuṣmad + ām → the pronoun's sām → ākam → final d deleted (SK 385) → yuṣmākam "of you (all)"; asmākam "our". Locative singular tvayi, mayi (SK 392); dual yuvayoḥ, āvayoḥ; plural yuṣmāsu, asmāsu (SK 393).

Bhaṭṭoji sums up the interplay of the substitutes in four verses. (1) When yuṣmad or asmad is compounded and, inside the compound, means two or one, then yuva / āva and tva / ma apply even if the compound as a whole has another number. (2) Before su, jas, ṅe and ṅas the substitutes tva / aha, yūya / vaya, tubhya / mahya and tava / mama always come. (3) Being the later rules, these block yuva / āva in their own domain, and they block tva / ma too, by the rule of prior conflict. (4) When the compound means two or one but yuṣmad / asmad inside it means many, neither yuva / āva nor tva / ma applies.

He then declines three compounds with ati "surpassing". Inner singular, "one who has surpassed you (tvām) / me (mām)":

Inner dual, "one who has surpassed you two (yuvām) / us two (āvām)": before su, jas, ṅe, ṅas as for the plain pronoun (atitvam, atiyūyam, atitubhyam, atitava; atyaham …); the rest:

Inner plural, "one who has surpassed you all (yuṣmān) / us all (asmān)": again su, jas, ṅe, ṅas as for the plain pronoun; the rest:

L. The accent headings and the unaccented short pronouns — running example: te / me / vām / nau / vaḥ / naḥ / tvā / mā

Sanskrit has short forms of "you" and "me" that lean on the word before them and carry no accent: te "to you, of you", me, vām, nau, vaḥ, naḥ, tvā, mā. Pāṇini prescribes them as substitutes for the full pronouns under three headings — "of a pada" (SK 401), "after a pada" (SK 402), "everything not at the start of a verse-quarter is unaccented" (SK 403) — so that the substitute is itself unaccented and only appears after another word and never at the start of a line. SK 404–407 give the substitutes case by case; SK 408–410 give the blocks and the options.

SK 401 · पदस्य · padasya (A 8.1.16)

Rule. An adhikāra: "of a pada". In the rules that follow, the operation enjoined applies to a finished word.

Example. The substitution te for tava is an operation on the pada tava, a finished word, not on the stem yuṣmad.

SK 402 · पदात् · padāt (A 8.1.17)

Rule. An adhikāra: "after a pada". The condition on the left for the rules that follow is a finished word.

Example. In dhātā te bhaktaḥ "the Creator is your devotee", te stands after the pada dhātā; at the start of a sentence, tava cannot become te (SK 407).

SK 403 · अनुदात्तं सर्वमपादादौ · anudāttaṃ sarvamapādādau (A 8.1.18)

Rule. An adhikāra: "everything that is not at the beginning of a pāda (a verse-quarter) is anudātta, unaccented". Read into the rules that follow, it makes each substitute unaccented and bars it from the head of a pāda.

Example. In vedair aśeṣaiḥ saṃvedyo 'smān kṛṣṇaḥ sarvadāvatu "may Kṛṣṇa, to be known through all the Vedas, always protect us", asmān opens a pāda and is therefore not replaced by naḥ (SK 407).

SK 404 · युष्मदस्मदोः षष्ठीचतुर्थीद्वितीयास्थयोर्वान्नावौ · yuṣmadasmadoḥ ṣaṣṭhīcaturthīdvitīyāsthayorvānnāvau (A 8.1.20)

Rule. For yuṣmad and asmad standing in the genitive, the dative or the accusative, after a pada and not at the beginning of a pāda, the substitutes are vām and nau respectively; both are unaccented. (The following rules narrow this: vām and nau are in effect the dual forms.)

Example. pātu vām api nau vibhuḥ "may the all-pervading one protect you two and us two": vām for yuvām, nau for āvām.

SK 405 · बहुवचनस्य वस्नसौ · bahuvacanasya vasnasau (A 8.1.21)

Rule. For the same two pronouns in the plural of the genitive, dative and accusative, the substitutes are vas and nas (heard as vaḥ, naḥ). This is an apavāda to vām / nau.

Example. so 'vyād vo naḥ "may he protect you all and us all": vaḥ for yuṣmān, naḥ for asmān.

SK 406 · तेमयावेकवचनस्य · temayāvekavacanasya (A 8.1.22)

Rule. For the same two pronouns in the genitive and dative singular, the substitutes are te and me.

Example. dattāt te me 'pi śarma saḥ "may he give happiness to you and to me": te for tubhyam, me for mahyam; svāmī te me 'pi sa hariḥ "that Hari is the master of you and of me": te for tava, me for mama.

SK 407 · त्वामौ द्वितीयायाः · tvāmau dvitīyāyāḥ (A 8.1.23)

Rule. For the same two pronouns in the accusative singular, the substitutes are tvā and mā.

Example. śrīśas tvāvatu mā pīha "may the Lord of Śrī protect you and me too here": tvā for tvām, mā for mām.

Bhaṭṭoji's two illustrative verses run through the whole set: "śrīśas tvāvatu mā pīha, dattāt te me 'pi śarma saḥ; svāmī te me 'pi sa hariḥ, pātu vām api nau vibhuḥ" and "sukhaṃ vāṃ nau dadātv īśaḥ, patir vām api nau hariḥ; so 'vyād vo naḥ śivaṃ vo no, dadyāt sevyo 'tra vaḥ sa naḥ".

Applies when / blocked by. Why "after a pada"? At the start of a sentence there is no substitute: tvāṃ pātu, māṃ pātu. Why "not at the beginning of a pāda"? vedair aśeṣaiḥ saṃvedyo 'smān kṛṣṇaḥ sarvadāvatu keeps asmān. The word stha "standing in" in SK 404 means that the case-ending must actually be heard: in iti yuṣmatputro bravīti "thus says your son", the ending inside the compound has been deleted, so there is no substitute. Three vārttikas add: the substitutes, like the loss of accent on a verb (nighāta), apply only within one sentence; a sentence is what has one finite verb, so odanaṃ paca, tava bhaviṣyati "cook the rice; it will be yours" keeps tava, while śālīnāṃ te odanaṃ dāsyāmi "I shall give you rice of śāli grain" has te; and where there is no anvādeśa the substitutes are optional (dhātā te bhakto 'sti or dhātā tava bhakto 'sti), but in an anvādeśa obligatory: tasmai te namaḥ "salutation to that same you", only te.

SK 408 · न चवाहाहैवयुक्ते · na cavāhāhaivayukte (A 8.1.24)

Rule. The substitutes do not occur when the pronoun is joined with one of the five particles ca, vā, ha, aha, eva.

Example. haris tvāṃ māṃ ca rakṣatu "may Hari protect you and me": ca keeps the full forms. kathaṃ tvāṃ māṃ vā na rakṣet "how would he not protect you or me?"

Applies when / blocked by. The word yukta "joined" limits the block to a direct connection. Where the particle joins something else and the pronoun is only indirectly related, the substitute does apply: haro hariś ca me svāmī "Hara and Hari are my lord".

SK 409 · पश्यार्थैश्चानालोचने · paśyārthaiścānālocane (A 8.1.25)

Rule. And the substitutes do not occur when the pronoun is construed with a verb meaning "see" used in a sense other than seeing with the eyes — mental perceiving, contemplating.

Example. cetasā tvāṃ samīkṣate "he beholds you with the mind": full tvām. bhaktas tava rūpaṃ dhyāyati "the devotee meditates on your form": full tava, even though the pronoun is related only indirectly. But with real sight, bhaktas tvā paśyati cakṣuṣā "the devotee sees you with the eye": tvā.

SK 410 · सपूर्वायाः प्रथमाया विभाषा · sapūrvāyāḥ prathamāyā vibhāṣā (A 8.1.26)

Rule. After a word in the nominative that itself has a word before it, the substitutes are optional — even in an anvādeśa, where they would otherwise be obligatory.

Example. bhaktas tvam apy ahaṃ tena hari tvāṃ trāyate sa mām "you too are a devotee, and I; therefore Hari protects you, and he protects me": after hari (preceded by tena) and after sa (preceded by trāyate), the full tvām and mām may stand, or tvā and mā.

M. The vocative and its treatment as non-existent — running example: agne naya / hare dayālo naḥ pāhi

A word of address (the vocative, named āmantrita, SK 411) that stands earlier in the sentence is treated as if it were not there for the purposes of the accent rules and of the short pronouns (SK 412): after agne "O Agni", tava does not become te, because there is no real pada before it. But an adjective-like word that refers to the same person as a following vocative is not treated as absent (SK 413: hare dayālo naḥ pāhi "O Hari, O merciful one, protect us"). Bhaṭṭoji then starts the stem supād "having good feet".

SK 411 · सामन्त्रितम् · sāmantritam (A 2.3.48)

Rule. A saṃjñā: the nominative used in address (sambodhana) is named āmantrita, and by the paribhāṣā of "that which ends in it" the word ending in that nominative bears the name too.

Example. In agne naya "O Agni, lead", the word agne is an āmantrita.

SK 412 · आमन्त्रितं पूर्वमविद्यमानवत् · āmantritaṃ pūrvamavidyamānavat (A 8.1.72)

Rule. A preceding āmantrita is treated as non-existent for the rules that follow (the accent rules, and the short pronouns which stand under them).

Example. agne tava …, devāsmān pāhi "O god, protect us", agne naya, agna indra varuṇa …: after the vocative, the pronoun is not replaced (tava stays, asmān stays), the verb does not lose its accent, and a second vocative does not lose its accent either — in each case because the vocative before it counts as absent.

Applies when / blocked by. Only the vocative is discounted; any real word before it still serves. In sarvadā rakṣa deva naḥ "always protect us, O god", deva is absent for the rule, but rakṣa before it is enough to give naḥ. In the mantra imaṃ me gaṅge yamune … the vocatives before yamune and the rest are absent, but the word me alone supports the loss of accent on all of them.

SK 413 · नामन्त्रिते समानाधिकरणे सामान्यवचनम् · nāmantrite samānādhikaraṇe sāmānyavacanam (A 8.1.73)

Rule. A word that names the thing qualified (the viśeṣya, the general term) is not treated as non-existent when an āmantrita that refers to the same thing (samānādhikaraṇa, co-referential) follows it.

Example. hare dayālo naḥ pāhi "O Hari, O merciful one, protect us": hare is the thing qualified and dayālo its co-referential vocative, so hare is not absent, and naḥ stands. agne tejasvin "O Agni, O radiant one".

Applies when / blocked by. For vibhāṣitaṃ viśeṣa-vacane (SK 3655) the Bhāṣya says "I shall say bahuvacanam": a qualified word in the plural, followed by a co-referential vocative, is optionally present. So yūyaṃ prabhavo devāḥ śaraṇyāḥ "you are mighty, O gods worthy of refuge"; and yuṣmān bhaje, devāḥ śaraṇyāḥ or vo bhaje "I worship you". Here even in an anvādeśa the substitutes are optional.

Bhaṭṭoji opens the stem in d, supād "having good feet": supāt / supād (d → t by cartva at a pause), supādau, supādaḥ, supādam, supādau. Its weak forms need the next group.

N. The bha-stem operations — pād → pad and the stems in añc — running example: prāñc (pra + añc) → prāṅ / prāñcau / prācaḥ / pratīcaḥ (also pratyañc, tiryañc, pād → pad)

Before the weak vowel-endings (the bha position) a few stems shrink. supād becomes supad- (SK 414). The direction-words made from the root añc "to bend, go" with kvin — prāñc "eastern, forward", pratyañc "western", udañc "northern", samyañc "right", sadhryañc "going together", tiryañc "across", and adas + añc — lose the n of añc before the kit affix (SK 415), get it back as num before the strong endings (SK 361: prāṅ, prāñcau), and in the bha position lose also the a of añc and lengthen the vowel before it (SK 416–417: prācaḥ, pratīcaḥ, udīcaḥ). The first members adas, sam, saha, tiras take special shapes before añc (SK 418–423). When añc means honour rather than direction, it keeps its n throughout (SK 424).

SK 414 · पादः पत् · pādaḥ pat (A 6.4.130)

Rule. For a stem that ends in the word pād "foot" and is bha, pad is substituted for that pād.

Example.

supād + śas → the stem is bha → pād → pad → supadaḥ; supadā; supadbhyām and so on.

A stem in th: agnimath "fire-kindler", from "he churns the fire": agnimat / agnimad (th → d by jaśtva, t by cartva), agnimathau, agnimathaḥ, agnimadbhyām. The rule SK 373 gives kvin to añc when a case-inflected word accompanies it; the stems of that root follow.

SK 415 · अनिदितां हल उपधायाः क्ङिति · aniditāṃ hala upadhāyāḥ kṅiti (A 6.4.24)

Rule. For a stem that ends in a consonant and is not marked with the marker i (anidit), the penultimate n is deleted before an affix marked with k or ṅ (kṅit). kvin is marked with k.

Example.

pra + añc + kvin → the penultimate n of añc deleted → pra + ac → prāc (the two a merge) — the base of all the weak forms.

prāc + su → añc has lost its n, so it takes num before a strong ending (SK 361) → prānc + su → su deleted → the cluster nc loses its c (SK 54) → prān → the n of num becomes ṅ at the end of a kvin-stem (SK 377) → prāṅ "eastern" (nominative singular).

prāc + au → num → prānc + au → n → anusvāra → ñ → prāñcau; prāñcaḥ, prāñcam, prāñcau.

SK 416 · अचः · acaḥ (A 6.4.138)

Rule. The a of añc — of the añc whose n has been deleted — is deleted when the stem is bha. The rule states only "of the a"; "deletion", "of a bha" and the reference to añc come from the surrounding rules.

Example.

prāc + śas → the stem is bha → the a of ac deleted → pra + c + as (continued under SK 417).

SK 417 · चौ · cau (A 6.3.138)

Rule. Before añc that has lost both its ā and its n — that is, before the bare c — the aṇ vowel (a, i, u) that precedes is lengthened.

Example.

pra + c + as → the a of pra lengthened → prācaḥ "the eastern ones"; prācā; before a consonant the c is a guttural: prāgbhyām (SK 378).

prati + añc + kvin → n deleted (SK 415), i → y → pratyac → strong pratyaṅ, pratyañcau, pratyañcaḥ, pratyañcam, pratyañcau; bha: prati + c + as → i lengthened → pratīcaḥ, pratīcā.

Watch out. In prati + ac one might turn the i into y before the a (yaṇ) — an inner operation, and inner operations usually go first. Here it must not, because the a is about to be deleted; the paribhāṣā akṛtavyūhāḥ pāṇinīyāḥ says that Pāṇini's followers do not perform an operation whose basis is about to disappear. Hence pratīcaḥ. Bhaṭṭoji then takes up adas + añc "going towards that": with adas + añc standing …

SK 418 · विष्वग्देवयोश्च टेरद्र्यञ्चतावप्रत्यये · viṣvagdevayośca ṭeradryañcatāvapratyaye (A 6.3.92)

Rule. The ṭi of viṣvac "all-round", of deva, and — by anuvṛtti — of a pronoun (sarvanāman) is replaced by adri when the root añc follows without a further affix on it (apratyaya).

Example.

adas + añc → adas is a pronoun; its ṭi is as → as → adri → ad-adri + añc → the i of adri before the vowel → y → adadry-añc (continued under SK 419).

SK 419 · अदसोऽसेर्दादु दो मः · adaso'serdādu do maḥ (A 8.2.80)

Rule. For adas in its as-ending portion — and not where an a stands for its s (asi) — the vowel after a d is replaced by u or ū, and the d by m. The u of the rule is a compound of short and long u; by closest resemblance a short vowel and a consonant get short u, a long vowel gets ū.

Example.

adadry-añc → the vowel after each d → u, each d → m → amumuy-añc → strong forms with num and the guttural: amumuyaṅ, amumuyañcau, amumuyañcaḥ, amumuyañcam, amumuyañcau; bha: amumuīcaḥ, amumuīcā; amumuyagbhyām.

Applies when / blocked by. In amumuīcaḥ the u before ī is not turned into v: the u is a Tripādī substitute, asiddha, invisible, to the yaṇ rule. Scholars discuss the shape: some make m only for the later d (adamuyaṅ), some only where an a stands for the s (adadyaṅ). — Why "viṣvac and deva"? Because aśvācī shows no adri. Why "añc"? viṣvagyuk has another root. Why "without an affix"? viṣvagañcanam has one. And the word "without an affix" indicates that elsewhere the mention of a root covers what begins with that root; this gives ayaskāraḥ "ironsmith", with its s by ataḥ kṛkami… (SK 160). udañc "northern": udaṅ, udañcau, udañcaḥ; the weak forms follow the deletion of a (SK 416) and the next rule.

SK 420 · उद ईत् · uda īt (A 6.4.139)

Rule. After the word ud, the a of the añc that has lost its n and is bha is replaced by ī.

Example.

ud + añc + śas → n deleted → ud + ac + as → bha → a → ī → udīcaḥ; udīcā; before a consonant udagbhyām.

SK 421 · समः समि · samaḥ sami (A 6.3.93)

Rule. sami replaces sam before añc ending in the affix v (that is, kvin reduced to its v, then to nothing). Scholars discuss whether the condition is instead "before añc with no further affix".

Example.

sam + añc → sam → sami → sami + añc → i → y → samyañc → samyaṅ "right, correct", samyañcau, samyañcaḥ; bha: sami + c → i lengthened (SK 417) → samīcaḥ, samīcā.

SK 422 · सहस्य सध्रिः · sahasya sadhriḥ (A 6.3.95)

Rule. sadhri replaces saha "together" before añc ending in the affix v.

Example.

saha + añc → sadhri + añc → i → y → sadhryaṅ "going together".

SK 423 · तिरसस्तिर्यलोपे · tirasastiryalope (A 6.3.94)

Rule. tiri replaces tiras "across" before añc ending in the affix v, when the a of añc is not deleted — that is, in the strong forms and before consonant-endings, not in the bha forms.

Example.

tiras + añc + su → tiras → tiri, i → y → tiryaṅ "going across, animal"; tiryañcau, tiryañcaḥ, tiryañcam, tiryañcau; tiryagbhyām.

tiras + añc + śas → bha, the a is deleted → tiras stays → tiras + c + as → s → ś before c (ścutva) → tiraścaḥ; tiraścā.

SK 424 · नाञ्चेः पूजायाम् · nāñceḥ pūjāyām (A 6.4.30)

Rule. The penultimate n of añc is not deleted when añc means honour (pūjā). Since the n is never lost, there is no num either (SK 361 gives num only to the añc that has lost its n), and no deletion of the a (SK 416 speaks of the n-less añc).

Example. prāñc "honoured, preceding" in the honorific sense.

pra + añc + su → the n stays → su deleted, c dropped from the cluster, n → ṅ → prāṅ; prāñcau, prāñcaḥ; bha with the a kept: prāñcaḥ, prāñcā; prāṅbhyām; locative plural prāṅkṣu / prāṅṣu.

So too pratyañc and the rest in the honorific sense. The root kruñc "to be crooked, to shrink" keeps its n by the listing in SK 373: kruṅ "curlew", kruñcau, kruñcaḥ, kruṅbhyām.

Bhaṭṭoji adds three more stems here. payomuc "cloud" (releasing water), with the guttural of SK 378: payomuk / payomug, payomucau, payomucaḥ. suvṛśc "cutting well", from vraśc with kvip: ṣ for the c (SK 294), deletion of the s that begins the cluster (SK 380), jaśtva, cartva: suvṛṭ / suvṛḍ; suvṛścau, suvṛścaḥ, suvṛścā; suvṛṭsu / suvṛṭtsu. And the stems in t: by the Uṇādi rule pṛṣan-mahad-bṛhaj-jagac-chatṛvac ca (Uṇādi 2.84) the words pṛṣat, mahat, bṛhat, jagat are formed ready-made in the present sense and behave like a stem in the participle affix śatṛ. śatṛ is ugit, so num comes before a strong ending (SK 361), and sāntamahataḥ (SK 317) lengthens the vowel: mahān "great", mahāntau, mahāntaḥ; weak mahataḥ, mahatā, mahadbhyām.

O. Stems in atu and as, and the reduplicated participles — lengthening and the num blocks — running example: dhīmat → dhīmān; bhavat → bhavān; dadat

The possessive stems in mat / vat (dhīmat "wise", gomat "owning cows") and the honorific bhavat "you, sir" are ugit stems, so they take num before the strong endings; in addition SK 425 lengthens their vowel in the nominative singular: dhīmān, bhavān. The present participle in śatṛ (bhavat "being") is also ugit and takes num, but not the lengthening: bhavan. A participle built on a reduplicated root (dadat "giving") takes no num at all (SK 426–428).

SK 425 · अत्वसन्तस्य चाधातोः · atvasantasya cādhātoḥ (A 6.4.14)

Rule. And the penult of a stem ending in atu, and of a stem ending in as that is not a root, is lengthened before su when it is not the vocative. Bhaṭṭoji notes the order: num is the later and the obligatory rule, yet the lengthening goes first, for otherwise the rule would have nothing to do; then num is added.

Example.

dhīmat + su → the stem ends in atu (the u of the affix matup being a marker) → penult a → ā → dhīmāt + su → num (SK 361) → dhīmānt + su → su deleted, t dropped from the cluster → dhīmān "wise" (nominative singular).

From śas onward dhīmat declines like mahat (SK 424). Even a root ending in atu takes the lengthening: from "he desires a cow-owner" or "he acts like a cow-owner", with the affixes kyac or the kvip of behaviour and then kvip, comes a root gomat, and its nominative is gomān, gomantau, gomantaḥ. Bhaṭṭoji explains the two "root" clauses. The word "ac" in SK 361 is a restriction: if an ugit operation is to touch a root, only añc gets it; so the ugit roots sraṃs and dhvaṃs give srat, dhvat without num. And "not a root" in SK 361 still provides num for what was formerly a non-root, like gomat. bhavat "your honour", formed by the Uṇādi affix ḍavatu from bhā: bhavān, bhavantau, bhavantaḥ. But the participle bhavat "being", formed with śatṛ, does not end in atu, so there is no lengthening: bhavan.

SK 426 · उभे अभ्यस्तम् · ubhe abhyastam (A 6.1.5)

Rule. A saṃjñā: the two members produced by the reduplication rules of the sixth book, taken together, are named abhyasta, "reduplicated".

Example. In dadat "giving" (from dā), the syllables da-da together are abhyasta.

SK 427 · नाभ्यस्ताच्छतुः · nābhyastācchatuḥ (A 7.1.78)

Rule. The śatṛ participle that follows an abhyasta does not take num.

Example.

dadat + su → no num → su deleted → dadat / dadad (cartva at a pause); dadatau, dadataḥ.

SK 428 · जक्षित्यादयः षट् · jakṣityādayaḥ ṣaṭ (A 6.1.6)

Rule. The root jakṣ "to eat" and six roots after it in the root-list — seven in all — are named abhyasta even without reduplication.

Example.

jakṣat + su → abhyasta, so no num → jakṣat / jakṣad; jakṣatau, jakṣataḥ. Likewise jāgrat "waking", daridrat "poor", śāsat "ruling", cakāsat "shining". The roots dīdhī and vevī are marked with ṅ and would take middle endings, but as that marking is Vedic they take active participles by interchange (vyatyaya): dīdhyat, vevyat.

A stem in p: gup "protector": gup / gub (jaśtva at the end), gupau, gupaḥ, gubbhyām.

P. The remaining root-stems — dṛś, naś, spṛś, and the ṣatva, lengthening and saṃprasāraṇa clean-up — running example: tādṛś → tādṛk; ghṛtaspṛś → ghṛtaspṛk; pipaṭhīḥ; cikīḥ; vidvas → vidvān / viduṣaḥ

The last root-stems are those in ś, ṣ, r and s. tādṛś "such" and viś "settler" end in ś, which becomes ṣ → ḍ → ṭ, or a guttural for a kvin-stem (SK 429–430: tādṛk); naś may take either (SK 431). spṛś "touching" takes kvin (SK 432: ghṛtaspṛk), and the numeral ṣaṣ "six" appears here as ṣaṭ, ṣaṇṇām. Stems in r or s from desiderative roots lengthen their vowel at the end of a word (SK 433: pipaṭhīḥ) and keep ṣ even across a nasal or a visarga (SK 434: pipaṭhīṣṣu). The perfect participle vidvas "knowing" undergoes saṃprasāraṇa in its weak forms (SK 435: viduṣaḥ).

SK 429 · त्यदादिषु दृशोऽनालोचने कञ् च · tyadādiṣu dṛśo'nālocane kañ ca (A 3.2.60)

Rule. When a pronoun of the tyad list is the accompanying word, the root dṛś "to see", in a sense other than seeing, takes the affix kañ; and, by the word "and", kvin as well.

Example.

tad + dṛś + kvin → the markers of kvin drop, the v is deleted (SK 375) → tad + dṛś "like that, such" (continued under SK 430); with kañ the stem is tādṛśa.

SK 430 · आ सर्वनाम्नः · ā sarvanāmnaḥ (A 6.3.91)

Rule. ā replaces the final of a pronoun before dṛś, dṛśa and vatu.

Example.

tad + dṛś → final d → ā → tādṛś + su → su deleted → tādṛś → the guttural of the kvin-stem (SK 377) is asiddha to the ṣ-rule, so first ś → ṣ (SK 294) → tādṛṣ → jaśtva → tādṛḍ → the guttural → tādṛg → optional cartva → tādṛk / tādṛg; tādṛśau, tādṛśaḥ.

Scholars discuss this derivation: on the view of Kaiyaṭa and Haradatta the guttural comes straight from ś as kh, and then only kh would be heard where cartva is not applied, not g; or the guttural may not be asiddha at all, as digādibhyo yat (SK 1429) suggests. The stem viś "settler, merchant": ś → ṣ (SK 294), jaśtva, cartva: viṭ / viḍ; viśau, viśaḥ, viśam.

SK 431 · नशेर्वा · naśervā (A 8.2.63)

Rule. For the root naś "to perish", a guttural optionally replaces the final at the end of a pada; otherwise ṣ → ḍ → ṭ as for viś.

Example.

naś + su → naś → ś → g or ś → ṣ → ḍ → nak / nag / naṭ / naḍ; naśau, naśaḥ; nagbhyām / naḍbhyām.

SK 432 · स्पृशोऽनुदके क्विन् · spṛśo'nudake kvin (A 3.2.58)

Rule. After the root spṛś "to touch", when a case-inflected word other than udaka "water" accompanies it, the affix kvin is added.

Example.

ghṛta + spṛś + kvin → ghṛtaspṛś + su → ghṛtaspṛk / ghṛtaspṛg "touching ghee" (the kvin guttural); ghṛtaspṛśau, ghṛtaspṛśaḥ.

Bhaṭṭoji reads "the stem from which kvin comes" of SK 377 as a bahuvrīhi, so the guttural also applies to spṛś with kvip: spṛk; the sounds ṣ, ḍ, g, k arise as in tādṛk. The root ñidhṛṣā "to be bold": kvin, reduplication and final accent are all licensed by the listing of SK 373; before the guttural, jaśtva gives ḍ, then g, then k: dadhṛk / dadhṛg "bold", dadhṛṣau, dadhṛṣaḥ, dadhṛgbhyām. ratnamuṣ "jewel-thief": ratnamuṭ / ratnamuḍ, ratnamuṣau, ratnamuṣaḥ.

The numeral ṣaṣ "six": jas and śas deleted by ṣaḍbhyo luk (SK 261) → ṣaṭ / ṣaḍ; ṣaḍbhiḥ, ṣaḍbhyaḥ (dative and ablative). In the genitive plural the augment nuṭ (SK 338) follows the ḍ; the n becomes ṇ by ṣṭutva (the exception "anām" in the rule that bars ṣṭutva at the end of a pada does not cover nām), and the ḍ before the nasal, optionally nasal by yaro 'nunāsike (SK 116), is obligatorily nasal here by the vārttika "before an affix, in the ordinary language, always": ṣaṇṇām. Locative plural ṣaṭtsu / ṣaṭsu. The name ṣaṭ extends to a compound ending in the numeral: paramaṣaṭ, paramaṣaṇṇām; but as a subordinate member the word declines like ratnamuṣ: priyaṣaṣaḥ, priyaṣaṣām. Since the change to ṣ is asiddha to the ru-rule, a final ṣ that stands for s still becomes ru by sasajuṣo ruḥ (SK 162) — as in pipaṭhīḥ below.

SK 433 · र्वोरुपधाया दीर्घ इकः · rvorupadhāyā dīrgha ikaḥ (A 8.2.76)

Rule. For a root ending in r or v, the ik vowel that is its penult is lengthened at the end of a pada.

Example. pipaṭhis "wishing to recite" (a desiderative root with kvip); its s is ṣ by ṣatva, then ru at the end of a word.

pipaṭhis + su → su deleted → the s → ru, an r → the root now ends in r at the end of a pada → penult i → ī → pipaṭhīr → visarga → pipaṭhīḥ; before a consonant pipaṭhīrbhyām.

Before a vowel the s stays and the i is short: pipaṭhiṣau, pipaṭhiṣaḥ. By vā śari (SK 151) the visarga before a sibilant is optional.

SK 434 · नुम्विसर्जनीयशर्व्यवायेऽपि · numvisarjanīyaśarvyavāye'pi (A 8.3.58)

Rule. The retroflex ṣ for an s after an iṇ vowel or a guttural (the ṣatva rule) applies even when a num, a visarga or a śar sibilant stands between — each one singly.

Example.

pipaṭhīr + su → r → visarga → pipaṭhīḥ + su → the s of su follows ī across the visarga → s → ṣ → pipaṭhīḥṣu; and by ṣṭutva the visarga's sibilant form may itself be ṣ: pipaṭhīṣṣu.

Applies when / blocked by. "Each singly" means that two intervening sounds block: in niṃssva and niṃsse an anusvāra and an s stand together between the vowel and the final s, so that s stays s. The "num" of the rule means only the anusvāra that stands in place of num. In suhinsu and puṃsu the nasal is not such an anusvāra, so there too the s of su stays s. For that same reason the mention of num is not made redundant by the mention of the sibilants. cikīrṣ "wishing to make": the s of the cluster after r is deleted by rāt sasya (SK 280) and the r becomes visarga: cikīḥ; cikīrṣau, cikīrṣaḥ; locative plural cikīrṣu, with no visarga, because SK 339 restricts the visarga before su to a ru. dos "arm", formed by damer ḍosi (SK 227): the ḍ-marker deletes the ṭi of the root; the ṣ is asiddha to the ru-rule, so ru and visarga: doḥ, doṣau, doṣaḥ; optionally doṣan by pad-dan… (SK 228): doṣṇaḥ, doṣṇā, beside doṣaḥ, doṣā. viviś "wishing to enter" (desiderative of viś with kvip): the cluster loses its final, ś → ṣ, jaśtva, cartva: viviṭ / viviḍ; vivikṣau, vivikṣaḥ. takṣ "carpenter": the k that begins the final cluster is deleted (SK 380): taṭ / taḍ; takṣau, takṣaḥ; gorakṣ "cowherd": goraṭ / goraḍ, gorakṣau, gorakṣaḥ.

For the causatives of takṣ and rakṣ with kvip, Bhaṭṭoji holds that SK 380 does not act, because the deleted causative ṇi still counts as present. A vārttika that would deny this ("in the Tripādī a deleted element does not count as present") does not apply here; but another vārttika finds a fault in that counting for the deletion of a cluster's first member, so the ordinary deletion of a cluster's final applies instead: tak / tag, gorak / gorag; likewise pipak / pipag, vivak, didhak. Scholars discuss the two routes. supis "moving well" (from pis "to move"): supīḥ, supisau, supisaḥ, supisā, supīrbhyām, supīḥṣu / supīṣṣu; likewise sutūḥ from tus "to break". The perfect participle vidvas "knowing": vidvān, vidvāṃsau, vidvāṃsaḥ; he vidvan; vidvāṃsam, vidvāṃsau.

SK 435 · वसोः सम्प्रसारणम् · vasoḥ samprasāraṇam (A 6.4.131)

Rule. A stem ending in the affix vas(u) undergoes saṃprasāraṇa when it is bha: its v becomes u. Then pūrvarūpa (SK 330) and ṣatva follow.

Example.

vidvas + śas → v → u → vid-u-as + as → u + a → u (SK 330) → vidus + as → s → ṣ after u → viduṣaḥ; viduṣā; before a consonant, d for the final of a vasu-stem (SK 334): vidvadbhyām.

sedivas "having sat" (from sad): sedivān, sedivāṃsau, sedivāṃsaḥ, sedivāṃsam; in the bha forms the augment iṭ that gave the i, though an inner operation, does not act in the domain of saṃprasāraṇa (the akṛtavyūha paribhāṣā, SK 417): seduṣaḥ, seduṣā; sedivadbhyām. Bhaṭṭoji adds that the s-final cluster of sāntamahataḥ (SK 317) means a nominal stem's cluster, not a root's, because the rule pairs it with mahat; so the root-stems in s below take no lengthening. suhiṃs "striking well": suhin (the s dropped from the cluster), suhiṃsau, suhiṃsaḥ, suhinbhyām, suhintsu. dhvaṃs "falling": dhvat / dhvad (the n deleted by SK 415, d by SK 334), dhvasau, dhvasaḥ, dhvādbhyām; likewise srat.

Q. The pronouns puṃs and adas — running example: puṃs → pumān; adas → asau / amum / amī / amunā / amuṣmai

Two stems in s close the chapter. puṃs "man" replaces its final by as before the strong endings (pumān, pumāṃsau) and keeps puṃs- elsewhere (SK 436). adas "that yonder" has a nominative singular asau (SK 437) and then, in every other form, turns its d into m and the vowel after it into u, ū or ī: amum, amū, amī, amunā, amuṣmai (SK 438–439).

SK 436 · पुंसोऽसुङ् · puṃso'suṅ (A 7.1.89)

Rule. When a strong ending is meant, asuṅ — that is as — replaces the final of puṃs. Being marked with ṅ, the substitute replaces only the final sound; the u is only for pronunciation. The stem puṃs itself is formed by the affix ḍumsun, whose u is a marker, so the stem is ugit and takes num (SK 361) — Bhaṭṭoji makes this point for the feminine bahupuṃsī, where the num is needed too.

Example.

puṃs + su → s → as → pum-as + su → the stem now ends in as and is not a root → penult lengthened before su (SK 425) → pumās + su → num → pumāns + su → su deleted, s dropped from the cluster → pumān "man".

Other stems in as: uśanas "Uśanas" takes anaṅ in the nominative singular by r̥d-uśanas… (SK 276): uśanā, uśanasau, uśanasaḥ. A vārttika makes the anaṅ and the n-deletion optional in its vocative: he uśanan / he uśana / he uśanaḥ; uśanobhyām. anehas "time": anehā, anehasau, anehasaḥ, he anehaḥ, anehobhyām. vedhas "creator": vedhāḥ, vedhasau, vedhasaḥ, he vedhaḥ, vedhobhyām. Because SK 425 says "not a root", root-stems in as take no lengthening: suvaḥ "clothing himself well" (from vas), suvasau, suvasaḥ; piṇḍagraḥ, piṇḍaglaḥ "swallowing a lump", from gras, glas "to eat".

SK 437 · अदस औ सुलोपश्च · adasa au sulopaśca (A 7.2.107)

Rule. For adas, au replaces the final when su follows, and su is deleted. The d becomes s by SK 381.

Example.

adas + su → final s → au, su deleted → adau → the non-final d → s (SK 381) → asau "that (man) yonder".

A vārttika adds that for adas with the infix akac the au is optionally forbidden, and then u replaces the vowel after the s; the u comes only together with that prohibition: asakau / asukaḥ. In the other cases the tyad-list a replaces the final, with pararūpa; for the dual, vṛddhi: ada + au → adau; then by SK 419 the d → m and the vowel after it → ū (long, matching au): amū (nominative dual). For the plural, jas → śī (SK 214) and ād guṇaḥ (SK 69): ada + ī → ade (continued under SK 438).

SK 438 · एत ईद्बहुवचने · eta īdbahuvacane (A 8.2.81)

Rule. For the e of adas that follows the d, ī is substituted, and the d becomes m, when the plural is meant.

Example.

ade (from SK 437) → e → ī, d → m → amī "those yonder" (nominative plural).

Because this rule and SK 419 belong to the Tripādī (pūrvatrāsiddham, SK 12), the operations on the ending come first and the u and m afterwards: ada + am → adam → amum (accusative singular), amū (accusative dual), ada + śas → adān → amūn (accusative plural). Once mu has been made, the stem ends in u and is ghi, so the instrumental ṭā becomes nā — but see the next rule.

SK 439 · न मु ने · na mu ne (A 8.2.3)

Rule. The mu substitution (the m and the u of adas) is not asiddha — not invisible — when the nā substitute for ṭā is to be made, nor after it has been made. Without this rule the ṭā could not see the u-stem and become nā.

Example.

ada + ṭā → d → m, a → u → amu + ṭā → the stem is a ghi in u, and the mu counts as done → ṭā → nā → amunā.

Terms used in this chapter

• saṃjñā — a technical name given by a rule to a sound, form or affix, so that later rules can refer to it (e.g. bha, ṣaṭ, saṃprasāraṇa).
• paribhāṣā — a rule of interpretation: it does no operation itself but says how other rules are to be read and applied.
• halanta — ending in a consonant (hal); the stems of this chapter.
• pada-final (padānta) — standing at the end of a pada: a stem before a consonant-initial ending, or alone after su has been deleted. The Tripādī's word-final rules act here.
• jhal — the pratyāhāra of all stops and sibilants; "before a jhal" is the second setting of the word-final rules.
• anuvṛtti — the carrying forward of a word from an earlier rule into the rules that follow it, so that the later rule is read with that word supplied (SK 324: "before a jhal", "at the end of a pada").
• su-lopa / hal-ṅyādi-lopa — the deletion of the ending su after a consonant (SK 252).
• jaśtva / cartva — a final stop becomes the voiced unaspirated stop of its class (SK 84); before a hard sound, and optionally at a pause, it becomes the hard unaspirated stop.
• apavāda — an exception: a narrower rule that takes precedence over a wider one in its own domain.
• upadeśa — the original statement of a form, as in the root-list; a root counts as "d-initial in upadeśa" even when an augment is later put before it.
• denominative — a verb made from a noun (gardabhayati "he calls it a donkey" from gardabha); a root-noun with kvip can be made from it in turn.
• baś / bhaṣ / jhaṣ — the pratyāhāras of the voiced unaspirated stops (b, g, ḍ, d), the voiced aspirates (bh, gh, ḍh, dh), and the same voiced aspirates counted from jh.
• eka-ac — containing one vowel; monosyllabic.
• vyapadeśivadbhāva — treating a whole as if it were the part of the same name, so that a rule about "a portion of a root" can act on a root that is all portion.
• saṃprasāraṇa — an ik vowel standing in place of a semivowel (yaṇ) (SK 328).
• bha — the name of a stem before a vowel-initial or y-initial ending from śas onward; the weak position.
• ūṭh — the substitute ū (with marker ṭh) for the v of vāh (SK 329).
• pūrvarūpa / ekādeśa — a single substitute for two adjoining sounds that has the form of the first (SK 330).
• āgama — an augment, a sound inserted into a form; ām, num, am, nuṭ, suṭ, tuk, iṭ are augments.
• sarvanāmasthāna — the five strong endings su, au, jas, am, auṭ (and, in the neuter, śi).
• udātta / anudātta — the high accent and the low (unaccented) pitch of a vowel.
• num — the augment n placed after a stem's last vowel (u and m are markers).
• mit — marked with m; an augment so marked goes after the last vowel.
• asiddha — treated as not having taken effect; a later Tripādī rule is asiddha to an earlier rule (SK 12).
• saṃyogānta-lopa — deletion of the last member of a word-final consonant-cluster (SK 54).
• na-lopa — deletion of a nominal stem's final n at the end of a pada (A 8.2.7).
• sambuddhi — the vocative singular.
• mūrdhanya / ṣatva — the retroflex sounds; the change of s to the retroflex ṣ.
• al-vidhi — a rule that concerns a single sound; in it a substitute is not treated as its original.
• sthānivadbhāva — the treating of a substitute as if it were the original it replaced.
• antādeśa — a substitute for the final sound of a form.
• nuṭ — the augment n placed at the front of the genitive plural ām (SK 338).
• ṇatva — the change of n to ṇ.
• dvitva — doubling of a consonant.
• ru — the substitute r (with marker u) for a final s; only it, not an original r, becomes visarga before su (SK 339).
• visarjanīya — the visarga.
• śar — the pratyāhāra of the sibilants ś, ṣ, s.
• niyama — a restriction: a rule that limits what another rule would otherwise allow.
• bahuvrīhi — a possessive compound, "whose X is Y".
• gauṇa / pradhāna — subordinate / principal: the role of a word inside a compound.
• vibhakti — a case-ending (or verb-ending); here the twenty-one case-endings.
• akac — the infix ak that forms diminutive or derogatory pronouns (imaka).
• tyad-ādi / tyadādy-atva — the list of pronouns beginning with tyad (tyad, tad, etad, yad, idam, adas among them); the replacement of their final by a before an ending (A 7.2.102).
• pararūpa — a single substitute for two adjoining sounds that has the form of the second; here a + a → a by ato guṇe (SK 191).
• āp — the pratyāhāra of the endings from ṭā to the marker p of sup: instrumental to locative.
• ais / etva — the substitute ais for the instrumental plural bhis after an a-stem; the change of a final a to e before bh and s.
• nitya — obligatory; a rule that applies whether or not another has applied, and therefore goes first.
• anvādeśa — a second mention of something already named, in order to say something further about it.
• vic / kvip / kvin — affixes that turn a root into a noun and are then deleted entirely; kvin stems take a guttural at the end of a word.
• upadhā / upadhā-dīrgha — the penultimate sound; the lengthening of the penult before a strong ending.
• prātipadika — a nominal stem, the base to which case-endings are added.
• vārttika — a supplementary statement on a sūtra.
• pratyaya-lakṣaṇa — the principle that a deleted affix still conditions operations, blocked by na lumatāṅgasya (SK 263) when the deletion was a luk.
• luk — deletion of an entire affix by a rule that says luk.
• uttarapada — the second member of a compound.
• sub-vidhi / svara-vidhi / saṃjñā-vidhi — an operation conditioned by a case-ending / an accent operation / a naming operation.
• kṛt — an affix, other than a verb-ending, added to a root (SK 374).
• tuk — the augment t.
• al-lopa — the deletion of the a of -an before a weak vowel-ending (SK 234).
• ścutva — a dental becomes the corresponding palatal beside a palatal (n → ñ before j).
• antaraṅga / bahiraṅga — inner / outer: an operation with fewer or nearer conditions is inner; an outer operation is treated as not done when an inner one is due.
• savarṇa-dīrgha — the merger of two like vowels into one long vowel.
• śi — the neuter nominative-accusative plural ending.
• ñit / ṇit — an affix marked with ñ / ṇ.
• ku / kutva — the guttural class; the replacement of a sound by its guttural.
• yoga-vibhāga — the splitting of one sūtra into two rules.
• preverb (upasarga) — a prefix such as pra, pari, ni placed before a root (pra-han); its r or ṣ can trigger ṇatva in the root.
• Bhāṣya — the Mahābhāṣya, the great commentary of Patañjali on the Aṣṭādhyāyī.
• ugit — an item whose marker is u, ṛ or ḷ (the vowels of uk); such a stem takes num before a strong ending (SK 361).
• taddhita — a secondary affix, added to a nominal stem.
• jñāpaka — an indicator: a wording that makes sense only if some further principle holds, and so establishes that principle.
• nañ — the negative prefix a- / an-.
• ṭi — the last vowel of a form together with whatever follows it (SK 368).
• ṅīp — the feminine affix ī.
• ṣaṭ — the name of a numeral ending in ṣ or n (SK 369).
• saṅkhyā — a numeral.
• sannipāta-paribhāṣā — the principle that an operation which depends on a condition does not destroy that condition.
• sarvādeśa / śit — a substitute that replaces the whole of its original; one marked with ś does so.
• nipātana — a ready-made listing of forms, which licenses whatever in them no rule accounts for.
• upapada — a word accompanying a root when an affix is added to it.
• tiṅ — a verb-ending.
• apṛkta — an affix reduced to a single sound (SK 375).
• anusvāra / parasavarṇa — the pure nasal ṃ; its replacement by the nasal of the following stop's class.
• cu — the palatal class c, ch, j, jh, ñ.
• adhikāra — a heading rule read into the rules that follow it.
• ma-paryanta — the portion of yuṣmad / asmad up to and including its m (SK 383).
• ṭāp — the feminine affix ā.
• śī — the substitute ī for jas after a pronoun in a.
• aṅga-kārya paribhāṣā — the principle that once an operation on the stem has been done for an ending, no second stem-operation is done for it.
• suṭ — the augment s at the front of a pronoun's genitive plural ām (sarveṣām).
• pāda — a metrical quarter of a verse.
• nighāta — the loss of accent on a finite verb.
• vākya — a sentence: what has one finite verb.
• āmantrita — the nominative of address; the vocative (SK 411).
• avidyamānavat — treated as non-existent.
• samānādhikaraṇa / viśeṣya — co-referential, referring to the same thing / the word naming the thing qualified.
• kit — an affix marked with k; before it the n of añc is deleted (SK 415).
• kṅit / anidit — marked with k or ṅ / not marked with i.
• aṇ — the vowels a, i, u (as used in SK 417).
• akṛtavyūha — the paribhāṣā that Pāṇini's followers do not perform an operation whose ground is about to be removed.
• adri — the substitute for the ṭi of a pronoun before añc (SK 418).
• āntaratamya — closest resemblance: a substitute matches its original in length or kind.
• va-pratyaya — the affix v, that is, kvin or kvip after the markers drop.
• pūjā — honour, reverence; the sense in which añc keeps its n.
• Uṇādi — the list of supplementary affixes that form ready-made nouns.
• śatṛ — the affix of the present participle, -at, marked with ṛ.
• atu — the affix -at of the possessives matup / vatup, marked with u.
• abhyasta — the two members of a reduplication taken together (SK 426).
• vyatyaya — interchange of endings licensed for the Veda.
• desiderative (sannanta) — a root extended by the affix san, meaning "wishing to …" (pipaṭhis "wishing to recite", cikīrṣ "wishing to make"); with kvip it yields a root-noun.
• kañ — the affix a (with markers k, ñ) after dṛś (SK 429).
• iṇ — the vowels other than a together with h, y, v, r, l; after them s → ṣ.
• vyavāya — intervention: a sound standing between the trigger and the target of a rule.
• iṭ — the augment i taken by certain affixes.
• asuṅ / anaṅ — the substitutes as / an for the final of puṃs / uśanas and the like, marked with ṅ so that only the final is replaced.
• ghi — the name of a stem in short i or u; after it ṭā becomes nā.
• matva / utva / mu — the change of d to m, of the following vowel to u, in adas; together, the mu substitution.

Summary