Vedic Metre in its Historical Development c12.s3: Distribution of Variant Final Vowels in the Rigveda

Vedic Metre in its Historical Development c12.s3

Language: English

Distribution of Variant Final Vowels in the Rigveda

c12.s3.1

| | | | Decisive positions | | | In 2nd place | | Before caesura |
| | | | A | B | D | E | 2TD | 2TD | 4A | 5B |
| | All long vowels | 1000 | 201 | 355 | 22 | 1 | 84 | 77 | 121 | 70 |
| | All short vowels | 1000 | 7 | 77 | 236 | 291 | 17 | 30 | 24 | 64 |
| § 158 | Perfects in -ā, etc. | 163 | 22 | 50 | 4 | 1 | 13 | 15 | 4 | 11 |
| | Aorist impv. -svā | 69 | 3 | 32 | 1 | 0 | 23 | 14 | 4 | 2 |
| | Gerunds in -tya, -yā | 85 | 8 | 18 | 2 | 4 | 1 | 2 | 15 | 6 |
| | Endings -tarī, -tī, etc. | 125 | 8 | 18 | 2 | 0 | 7 | 12 | 10 | 6 |
| | Adverbs | 538 | 35 | 249 | 15 | 4 | 114 | 121 | 43 | 18 |
| | Total | 980 | 76 | 367 | 24 | 9 | 158 | 164 | 76 | 43 |
| § 159 | Imperative in -ā | 1177 | 146 | 332 | 33 | 40 | 104 | 114 | 108 | 43 |
| | 2 pl. -thā, -tā | 707 | 141 | 140 | 40 | 43 | 17 | 14 | 61 | 38 |
| | ” -thanā, -tanā | 129 | 14 | 11 | 1 | 3 | 0 | 0 | 6 | 0 |
| | Neut. pl. -ā, -ī, -ū | 203 | 25 | 66 | 6 | 6 | 25 | 31 | 16 | 17 |
| | Total | 2216 | 326 | 549 | 80 | 92 | 146 | 159 | 191 | 98 |
| § 160 | tú, tú | 38 | 1 | 23 | 8 | 3 | 16 | 7 | 1 | 0 |
| | nú, nú | 215 | 22 | 37 | 30 | 13 | 6 | 14 | 15 | 7 |
| | ū, u | 387 | 38 | 47 | 171 | 44 | 13 | 30 | 10 | 0 |
| | sú, sú | 208 | 14 | 49 | 36 | 22 | 19 | 21 | 13 | 6 |
| § 161 | adyā, adyā | 184 | 11 | 31 | 8 | 7 | 10 | 14 | 14 | 39 |
| | ádhā, ádha | 182 | 5 | 68 | 78 | 18 | 40 | 42 | 0 | 0 |
| | yádi, yádi | 68 | 4 | 30 | 13 | 10 | 10 | 18 | 10 | 0 |
| | Total | 1282 | 95 | 285 | 344 | 117 | 114 | 146 | 63 | 52 |
| § 162 | Impv. in -uhí, -hí, -dhí¹ | 870 | 13 | 108 | 60 | 149 | 23 | 33 | 36 | 33 |
| § 163 | 1 pl. -ma | 533 | 28 | 34 | 19 | 67 | 2 | 6 | 28 | 85 |
| | Perfect -tha, -a | 539 | 18 | 45 | 16 | 52 | 6 | 12 | 31 | 91 |
| | Impv. -sva (thematic) | 387 | 9 | 22 | 6 | 54 | 0 | 0 | 20 | 59 |
| | Instr. -ena (nouns) | 505 | 25 | 67 | 38 | 109 | 0 | 0 | 64 | 72 |
| | Total | 1964 | 80 | 168 | 79 | 282 | 8 | 18 | 143 | 307 |
| | Adverbs in -tra | 224 | 3 | 93 | 9 | 15 | 35 | 59 | 25 | 51 |
| | Instr. -ena (pronouns) | 166 | 0 | 93 | 13 | 12 | 37 | 82 | 8 | 4 |
| | Total | 390 | 3 | 186 | 22 | 27 | 72 | 141 | 33 | 65 |
| § 171 | ápa, ápo | 167 | 4 | 11 | 101 | 35 | 2 | 19 | 1 | 0 |
| | utá, utó | 704 | 2 | 114 | 212 | 218 | 65 | 104 | 5 | 1 |
| | yásya, etc. | 263 | 0 | 44 | 77 | 72 | 13 | 54 | 2 | 23 |
| | Total | 1134 | 6 | 169 | 490 | 325 | 80 | 177 | 8 | 24 |
| § 172 | Hiatus after -ā | 218 | 16 | 71 | 8 | 30 | 9 | 14 | 48 | 48 |

c12.s3.2

¹ omitting kṛdhí, jahí, śṛṇudhí, śṛṇuhí and śrudhí (§ 162).

c12.s3.3

165. In the quantity of the final vowels of the prior elements of compounds and derivatives the Saṃhitā text is in general agreement with the metre, and doubtful cases are rare. Owing again to the complexity of the conditions, it is not easy to classify the occurrences or to treat them in a systematic way: but it may be recognized that the quantity of the final vowel depends partly upon the historical character of the prior element, and partly upon the phonetic character of the initial sound of the posterior element, besides any other influences that may be at work. It will be convenient to consider separately: (i) compounds in which the prior element is a noun or numeral; (ii) compounds with prepositions and particles; (iii) vowels preceding derivative suffixes; and (iv) the augment and the vowel of reduplication. The phonetic influence of *v* following will in each class claim special attention: in addition, it is possible that *y r l s* and *m* produce in certain groups of words some effect in the direction of lengthening.

c12.s3.4

166. In compounds of which the prior element is a noun (including adjectives, pronouns, and numerals), the final syllable of that element is long (i) if it is a case-form, as *akṣṇayādrúh*, *amājúr*, *tuávṛdha*; (ii) if it is a feminine stem in *-ā* or *-ī*, as *urvarā-jít*, *nadítama*, *senānī*; (iii) in most proper names, as *lópāmudrā*, *viśvámitra*; and occasionally (iv) in stems in *-u*, *ulúkhala*, *purū-táma*, and perhaps (v) in stems in *-an*, as *vṛṣāyúdh*. The prior element (vi) has always a long vowel before *-vṛdh* and perhaps in some other cases before *v*, and (vii) is occasionally found in some other words in positions favouring long quantity.

c12.s3.5

(i) With the compounds of case-forms are included those of adverbs which have flexional endings, as in *evāvadá*, *etādṛś*. Such compounds are often found side by side with others that shew the short final of a stem: as *sanājú* but *sanajá*.

c12.s3.6

Here must be included the *dvandva* duals, the prior element having usually the long vowel not only in the nom. voc. acc. form, but also in the oblique cases. The text gives regularly *índrāvíṣṇū*, *mitráváruṇā*, *súryāmásā*, and so forth: and *índrāvāyú* must in all instances be restored, as is clear from the fact that it never stands at the end of Triṣṭubh verse, or in any other position in which *a* is favoured. Other restorations favoured by the metre are *mítrā ... váruṇā* i 151 6b and iv 1 18d (2T, 7A), *mítrā- v* 62 3b (2T), 66 6b (2D).

c12.s3.7

On the other hand we always find the triplet *varuṇa mitra aryaman* with the singular vocative forms, and occasionally the pair *varuṇa mitra* in the same shape, as i 122 7a (7A, 9): cf. vii 66 9ab. *mitra* varuṇa should probably be restored in ii 41 4a, v 64 4a, 67 5ab, vii 66 19a, viii 25 4a, 72 17a, 101 3a: and see further § 174 ii.

c12.s3.8

(ii) The text probably gives dhārāvará correctly in ii 34 1a (2T), though other compounds give dhāra-. In i 43 4a gāthápati should be read: in ii 41 16ab (2D) ámbitame dévítame corresponding to nádítame, and vedíṣád i 140 1a (2T), iv 40 5b (4B). Cf. § 170 iii.

c12.s3.9

(iii) Words such as ugrádeva, jarābodha, nárāśáṁsa, vārṣágirá, viśvánara come under this heading, independently of their presumed derivation. In iii 18 4c the metre suggests viśvámitra, but correction is not justified in view of the general usage.

c12.s3.10

(iv) Of final -ū in composition there are many traces in the text, always confirmed in the metre, as urūṇasá, ulúkhala, purútáma, purūrávas, vibhúvasu, vasújū: whilst other words have -u, as iṣuhasta, gātuvíd, dasyuhán. The long vowels may be connected with adverbial forms, see § 160 above. Further there are many cases in which the metre suggests the restoration of ū regularly or occasionally, as follows:

c12.s3.11

uru-: perhaps urūkṛt in the single occurrence viii 75 11c (2D), and urúyuga in the single occurrence viii 98 9b (6).

c12.s3.12

purú-: the compounds have commonly u, sometimes ū. The words purutáma, purubhúj, are sometimes found at the beginning of the stanza, or the end of the dimeter of Jagatī verse, in which positions ū is favoured; sometimes after an early caesura, where u is favoured but is not certain. In purutáma the text follows the metre: if rightly, we should restore purubhúj in i 3 1c (2D), v 73 1c (6), viii 8 17b (6), 10 6a (10). But purúvāsu has always the long vowel, probably even in vi 22 4d, vii 38 1d (both 6A).

c12.s3.13

madhu-: the compounds have regularly u: the metre favours ū in madhupŕcam ii 10 6d, mádhumatī iv 57 3a and madhuvŕdham x 75 8d, for which see subsection vi a.

c12.s3.14

vasu-: most of the compounds have u; but ū should probably be restored in most (if not all) occurrences of vásudhiti, namely i 128 8a (10), 181 1d (2T), iv 8 2a (6), 48 3a (6); the other occurrences are iii 31 17a and vii 90 3d (both 6A). In vasurúc ū should probably be restored in the single occurrence ix 110 6b (2T), and in vasuvíd the long vowel should certainly be restored in the majority of instances, the following v being an accessory cause, namely in i 46 2c (6), viii 23 16a (6), 60 12d (6), 61 5c (10), ix 101 11d (6), 104 4a (6), x 42 3d (2T). The reading of the text is only supported by *i 164 49c, vii 41 6c, ix 96 10a, where the word follows an early caesura.

c12.s3.15

vibhū- should be restored in vibhúṣáh for vibhvásáh of the text v 10 7c (3D), ix 98 1d (5).

c12.s3.16

viṣu-: viṣūvát and viṣūvŕt are regular, but víṣuañcā should probably be restored for viṣūcīnā in *i 164 38c (6A), as viṣudríac for viṣvadríac in vii 25 1d (5A).

c12.s3.17

vīlu- is regular, and it is therefore hardly safe to restore vīlū- in vīlupaví v 58 6b (2T), viii 20 2a (2T), or in vīlúharas *x 109 1c (2T) these appear rather to be instances of protraction (§ 163 vii).

c12.s3.18

(v) The final syllable of a stem in -an seems to be represented by -a in vṛṣāyúdh: but more usually appears as -a, as in the compounds of brahma-. The metre however suggests the restoration of -ā in the following words: dāśabhuji i 52 11a (10); dhanvacará v 36 1c (2T); śatápavitra vii 47 3a (2T) and śatábhuji i 166 8a (2T), vii 15 14c (6); syúmagabhasti i 122 15d (2T), vii 71 3c (2T), syūmagŕbhe vi 36 2c (2T), and syúmaraśmi viii 52 2d (2D). See also § 174 i.

c12.s3.19

(vi) A long vowel appears more or less regularly when the second element is one of the words -vasu, -vŕdh, -vŕdha, -magha or -sah.

c12.s3.20

(a) -vasu. The preceding vowel is regularly long, often with other contributory causes: sahávasu ii 13 8a (6A) and puruvásu, when it also follows the caesura (see subsection iv), are only doubtful exceptions.

c12.s3.21

-vŕdh. A long vowel is found in annāvŕdh, ŕtāvŕdh, ŕduvŕdh, āhutívŕdh, girāvŕdh, ghŕtāvŕdh, tugriāvŕdh, parvatāvŕdh: but mahivŕdh, rayivŕdh are equally well attested. In x 75 8d (10) madhūvŕdh may be restored.

c12.s3.22

-vŕdha. The metre favours kavīvŕdhá viii 63 4a (6), but is unsupported.

c12.s3.23

(b) -magha. The preceding vowel is always lengthened, and therefore tuvimaghá v 33 6d (6A) should probably be corrected.

c12.s3.24

(c) -sah. The preceding vowel is regularly long, as in virāsáh (§ 178), yajñāsáh, carṣaṇīsáh: but short in nṛsáh, bhūriṣáh and a few other words.

c12.s3.25

Probably janasáh should be restored in i 54 11b (6A).

c12.s3.26

(vii) In the following compounds there is not sufficient evidence to establish the existence of a long final vowel, but the syllable in question appears with some regularity in ‘positions of protraction’:

c12.s3.27

devaví: the second syllable occurs six times in the position 2T, namely ix 74 5b, 97 26a, 104 2c, x 36 8b, 76 4d, 78 1b; six times in the positions 2T, 2D, and nine times in positions in which a short vowel is required. As to a possible reading devavíyam, etc., see § 172 i.

c12.s3.28

-ratha: tveṣáratha v 61 13b (2D), śrutáratha i 122 7c (2T), v 36 6d (2T), sarvarathá v 35 5c (2D), x 160 1b (2T), sukháratha v 30 1b (2T). In candráratha, surátha the short vowel is usually favoured.

c12.s3.29

vājasáni iii 51 2c (2T), ix 110 11c (2T), x 91 15c (2T).

c12.s3.30

-han: ahihánam i 117 9d (2T), 118 9b (2T), but ii 13 5b (6A), 19 3b (6A); vṛtrahánam i 121 12d (2T), iv 42 9d (2T), vi 16 14c (2D), 17 11d (2T), viii 38 2b (2D), but i 108 3b (6B). In dasyuhán the short vowel is usually favoured.

c12.s3.31

In the forms pávīru, pávīrava, pávīravat, tuvīráva, tuvīrávat there is no lengthening of i before r in a compound, but a stem development such as is found in gabhīrá, śárīra.

c12.s3.32

167. In compounds of which the prior element is a preposition or particle, with which may be considered certain combinations of prepositions with nouns and finite verbs, the phonetic character of the second element seems to be of chief importance. Thus (i) many prepositions appear to have long final syllables before -vṛta and some other words beginning with v: (ii) ápā is found in the combination ápā vṛdhi: (iii) abhí is occasionally supported by the metre in abhíyúj, and is found in the text in combinations such as abhí narāḥ, abhí navante, abhí ṣát: (iv) the negative a- is regularly long in āvṛta and frequently in áriṣṭa and cognate forms, and in ásat. There occur also (v) some other irregularities which cannot be classified.

c12.s3.33

All the prepositions and particles with which we are concerned have ordinarily short final vowels, so that the questions that arise have to do with occasional long quantity. Amongst the occurrences of prepositions are included such forms as ápo, úpo, sahó of the text, for the reasons given in § 171 iv.

c12.s3.34

(i) The compounds ápī-, abhí-, párī-, and prá-vṛta always occur where the long vowel is favoured, except vii 27 2d (6A); cf. ā-vṛta in subsection iv. In cognate words the value varies; thus nīvṛta, trivít and suvít are regular, but we find ápāvṛti viii 66 3c (10), abhívartá *x 174 1a (2D), ánapāvṛt vi 32 5c (8), x 89 3a (8), and should perhaps read ánīvṛta *iii 29 6c (10) and suvīvíta i 10 7a (2D).

c12.s3.35

Before derivatives of vṛj the long vowel is less certain: the text has apārrktá, dāsápravarga, prāvargá, suprāvargá; we should perhaps restore anapāvṛjyán i 146 3c (3T) and parīvíjam viii 24 24b (6).

c12.s3.36

The text has further adhīvásā, úpāvasu, pratīví, prāvaṇá, prāvís and prāvṛṣíṇa quite consistently with the metre: and upāvídā viii 23 3c (2T) and parīvíṣe x 62 10a (6) are restorations favoured by the metre.

c12.s3.37

(ii) Long quantity is given in the text and supported by the metre in the phrase ápā vṛdhi, which occurs i 7 6b (6), ii 2 7b (10), iv 31 13a (6), vii 27 2d (2T) and viii 23 29c (10). Otherwise the use is that of a short vowel, as shewn in the Table, § 164.

c12.s3.38

(iii) The metrical use of abhíyúj is the same as that of purubhúj (§ 166 iv), and it may be right to restore abhíyúj in iii 11 6a, viii 45 8a, ix 21 2a (all 6). On the other hand abhiṣáḥ must be restored in vii 4 8d (6A). The general use hardly justifies us in keeping the reading abhí where it stands in the text as a separate word, namely in abhí duá x 48 7b (2T), abhí narám ix 97 49c (2T), abhí narāḥ v 9 7a (6), ix 101 3a (6), abhí navante ix 100 1a (2D), abhí naḥ i 140 13a (2T), iv 31 4a (2D), ix 97 51a (2T), abhí ṣát ii 41 10b (2D), vii 32 24a (2D).

c12.s3.39

For abhí nú ii 33 7d (2T) and abhí ṣú iv 31 3a (2D), viii 93 21a (2D), *x 59 3a (2D with hiatus) abhí nú, abhí ṣú are probable corrections.

c12.s3.40

(iv) The metre everywhere supports the restoration āvṛta for ávṛta, for instance in i 133 7e (6): cf. subsection i. In derivatives of riṣ the value is optional: the negative is short in v 42 8a (9), vi 19 4d (9) and elsewhere, but long in i 63 5a (8), vi 24 9c (8), 25 2a (8) and regularly in *ariṣṭātāti* in the popular Rigveda, namely in *x 60 8e = *9e = *10d, *97 7d, *137 4b (all 3E). In *ásat* the metre agrees with the text in supporting the long vowel in v 12 4d (2T), *vii 104 12d (10) and perhaps *13c (7A).

c12.s3.41

(v) Amongst miscellaneous variations we may notice a long vowel in the text before *ruh*, *rudh* in *anūrúdh*, *upārúh*, *vīrúdh*, quite consistently with the metre; *apījú* ii 31 5b (10) and *nīhárā* *x 82 7c (init.), which may also be correct; and *sū-* in several compounds (cf. § 160 iv), as *súnāra*, *súnīta*, *súbhava*, *súmāya*, in agreement with the metre.

c12.s3.42

The metre favours *pratidhíṣe* viii 60 13c (10) 14a (10), and *víśvasúvīdaḥ* i 48 2a (8), which may be ranked amongst protracted vowels (§ 163 vii).

c12.s3.43

The words *prasáh*, *suyávasa* have in the text *prā-*, *sū-*, but the long quantity is inadmissible except in *prāsahám* i 129 4b (8) and *sūyávasa* i 42 8a (3D), vi 27 7a (8). Thus *prasáh* is required in vi 17 4d (10), and *suyávasa* i 190 6a, ii 27 13a, *vi 28 7a, vii 18 4a, x 106 10d (all 5A).

c12.s3.44

168. Before suffixes we find long vowels always before *-van*, but short vowels regularly before *-mant*, *-vant*, except where the ending is *-yāvant* (-*īāvant*): long and short vowels almost indifferently in derivatives in *-yánt*, *-yá*, *-yú* and the corresponding verbs in *-y*.

c12.s3.45

(i) Long vowels appear to be found regularly before the suffix *-van*, *-var*, as *ṛtávan*, *ṛṇávān*, *sumnávārī*: and the metre suggests forms such as *maghávanaḥ* as antecedent to *maghónaḥ* etc., the occasions being v 16 3a (6), 86 3b (6), vi 65 3c (2T), ix 32 1b (6). Accordingly *matsarávā* (from a stem *matsarávan*) is a probable correction in ix 97 32c (10).

c12.s3.46

(ii) On the other hand a short vowel is regular before *-vant*, unless the prior element contains itself a long vowel, as is the case in words like *távant*, *svadhávant*, *pátnīvant*, *viṣūvánt*, and even in *aṣṭhīvánt* from *asthán*.

c12.s3.47

Amongst derivatives of nouns in *-a* *sutávant* is an exception, for both text and metre favour the long vowel: *vayúnāvant* again is correctly given in iv 51 1b (8), and should be restored in vi 21 3b (7A). But *áśvavant*, *índravant*, *devávant* have regularly *a*: the occasional *ā* of the text needs correction in *áśvavant* i 30 17a (5), i 53 5d (6A), *índravant* iv 27 4a (6A), *devávant* (iv 26 6c, 10 but perhaps misplaced): the other occurrences are chiefly in the second place, and perhaps rank with the protracted vowels (§ 163 vii).

c12.s3.48

In the derivatives of feminine nouns in *-i* or *-ī* the short vowel must frequently be restored: thus *sáktivant* v 31 6c (2T) and *vi 75 9b (6A), śímivant* i 141 13a (7B), ii 25 3a (7B), v 56 3c (9), x 8 2b (7B), 78 3c (10 in irregular ending), 89 5b (4B), *sáptivant* vii 94 10c (2D), x 6 6b (7B), hṛṣivant ii 31 1d (6A): but hṛṣivant is justified in i 127 6f (10).

c12.s3.49

As to derivatives in -mant it need only be noticed that vāśimant must in all cases be restored: it occurs i 42 6b (5), 87 6c (3T), v 57 2a (2T) and x 20 6c (6 in Trochaic Gāyatrī). Cf. § 170 iii.

c12.s3.50

(iii) In the formations in -yánt, -yá, -yú both short and long vowels are found, and the restoration of a short vowel is required in ṛjuyá i 183 5c (7B: it is probably the adverb of ṛjú); ṛtayú perhaps in v 8 1a (6A) as found in the text in viii 70 10a (7); gātuyánti i 169 5d (6A); vasuyánt, vasuyú as optional forms in i 130 6a (7B), ii 32 1d (6A), iv 16 15a (7B), vi 51 12d (7B); śatruyánt in vii 20 3d (6A). Less certain is aghayánt iv 2 6d (6A); for v 24 3b see § 192. The text gives sumnāyánt, sumnāyú only where the vowel occurs in the second place: it may rank with the protracted vowels; so too a possible restoration of śramāyúvaḥ in i 72 2c (2T): see § 163 vii.

c12.s3.51

169. In reduplication a long vowel is found (i) in many perfects of words beginning with v, as vāvakre, vāvandhí, vāvaśúḥ, vāvasánā, vāvrje, vāvṛté, vāvṛdhúḥ; (ii) of words beginning with r, as rāraṇá, rārahāṇá; and (iii) of some other words, as tātrṣāṇá, nānāma. Also (iv) in disyllabic reduplication, and (v) in the re-duplication of nouns, and in the intensive verb-forms.

c12.s3.52

The augment is occasionally long before v, as āvar, ávidhat, ávidhyat and (more rarely) before y and r.

c12.s3.53

(i) The length of the vowel is irregular, but the text usually agrees with the metre. A long vowel is more common if a heavy syllable follows, as vāvṛté but vavṛtvā́ms.

c12.s3.54

(ii) From ruc ‘shine’ rūrucúḥ is suggested by the metre as an optional form, in iv 7 1c (10), x 122 5d (10).

c12.s3.55

(iii) From kan cākan- is regular: cākánanta should be restored i 169 4c (5B) (cf. v 31 13a) though not favoured by the metre.

c12.s3.56

From tard (tṛd) tātrḍánā should perhaps be restored in iv 28 5d (8) and v 53 7a (init.).

c12.s3.57

From tarṣ (tṛṣ) tātrṣúḥ tātrṣāṇá are given in the text in i 31 7c (2T), 173 11c (6B), *x 15 9a (2T); but tatrṣāṇá i 130 8f (6B), vi 15 5d (6B). The metre seems to support the latter form, with i 31 7c and *x 15 9a as instances of protraction.

c12.s3.58

From dhar (dhṛ) the text has dādhára, etc. in all strong forms: this is confirmed by the metre, for the word never stands in Triṣṭubh cadence, and only in one case, ix 74 2d (5A), is the short vowel metrically preferable.

c12.s3.59

From nam the text has nānāma i 48 8a (5A), ii 33 12b (3T), iv 25 2a (2T). The restoration of nanāma is probable in the first instance.

c12.s3.60

From pi, pī the text has pipáya, but pipáya must be restored throughout.

c12.s3.61

From *yudh* *yūyudhúḥ* is a probable restoration, though we have *yuyódha* vi 25 5b (9), *yuyudhāte* i 32 13c (5A): the other occurrences are iv 30 3b (6), v 59 5b (10), vii 83 7b (10).

c12.s3.62

From *sah* the text has *sāsáha* and so forth, but *sa-* is preferable in all decisive instances, namely in i 100 5b (5A), 132 1b (5A), ix 110 12c (5A), besides the two in which the text has *sa-*, namely viii 96 15d (9), *x 180 1a (2T). The forms *sāsáhaḥ*, *sāsáhat*, *sāsahiṣṭháḥ*, in which *sā-* has really the long vowel, must therefore be referred to a reduplicated aorist or intensive formation. Cf. *pipáya*, *pīpáyat* above.

c12.s3.63

(iv) In disyllabic reduplication *ī* is always long, and we should therefore read *dávīdhuat* viii 60 13b, on the exact analogy of *távituat* iv 40 4c.

c12.s3.64

(v) In nouns the vowel of reduplication is generally long: but *táturi* is endorsed by the metre in i 145 3c (5A), vi 22 2c (5A). In the remaining occurrences *tárutra* is a tempting correction, but not altogether necessary, namely in iv 39 2d, vi 24 2a, 68 7d.

c12.s3.65

(vi) The augment is long in the words *ávar*, *ávidhat* (text *ávidhat*), *ávidhyat*, *ávṛṇak* (text varying), *āvṛṇi*; as

c12.s3.66

*áceti citrá ví dúro na āvaḥ* i 113 4b
*tuám putró bhavasi yás ta ávidhat* ii 1 9c
*kuruśrávaṇam āvṛṇi* x 33 4a

c12.s3.67

But *avar* must be restored i 92 4d (9), 113 13b (7B). In ii 17 6d (6B) the text gives *ávṛṇak* with some support from the metre.

c12.s3.68

From many forms even of the same verbs the augment has its regular value, as *avṛñjan*, *avrjan*, *ávṛkta*.

c12.s3.69

The augment also appears as *ā-* before *y* in *āyukta* v 17 3b (2T) and *āyunak* *i 163 2b (10). Similarly before *r* we find *āriṇak* ii 13 5b (8), and *āraik* occurs several times in positions which prove the first syllable long.

c12.s3.70

170. With very few exceptions the text accurately represents the quantities of vowels in the final syllables of stems before flexional endings. In some cases however the metre suggests restorations, namely, (i) of *a* i for *ā* *e* *ī* in dual verb-forms before the endings *-the*, *-te*, *-thām*; (ii) of *ā* for *a* in the strong cases of *uṣás*, and of *ā* for *a* in the declension of compounds of *sáh*, some nouns in *-an* and *-man*, and the adverb *viśváhā* (text often *viśváhā*), and in the locative case, and perhaps in the vocative, of nouns in *-u*; (iii) of *i* for *ī*, or the opposite, in the declension of feminine nouns in *-i* *-ī*; and (iv) regularly of *ri-* *ru-* in the formation of verbs with the suffix *-nā*, *-nī*.

c12.s3.71

Nearly all these points admit of some question, as the number of instances is not very large.

c12.s3.72

(i) The occurrences which favour the short vowel are as follows: *āśathe* i 2 8c (7), 15 6c (7), 151 8d (11) 9a (11), v 67 1d (7); *āśate* i 25 6a (7), 136 3d (7), 144 6d (11), v 66 2b (7), viii 31 6b (7); *ásathe* i 182 3a (11); *ásate* ii 41 5c (7); *cakrate* viii 29 9a (6B); *cakṣathe* vii 70 5b (7B) (7); *dadhathe* i 151 9a (6B); *yuñjathe* i 151 4d (9), v 74 3b (5); *ucyate* (text *ucyete*) *x 90 11d; *vahathe* (-ethe) i 135 8a (6B), 182 2c (6B); *śobhate* (-ete) iv 32 23c (7); *yuñjathām* iv 45 3b (9); *rāsathām* i 46 6c (5); *anuṣatām* viii 8 12d (7); *trásithām* iv 55 1b (9), v 41 1c (9), vii 62 4a (9), 71 2d (9).

c12.s3.73

On the other hand the long quantity given in the text is quite admissible in *ásāte* v 68 4b (7G) 5c (7G); *ásāthe* v 62 5d (3T), and is required by the metre in numerous other passages, as *iyāte* vii 39 2b (10), *dadhāte* i 185 2b (10), and *vasāthe* i 152 1a (10).

c12.s3.74

The short vowel seems therefore to be an archaism which is disappearing.

c12.s3.75

(ii) (*a*) In the declension of *uṣás* the text usually gives *uṣásam*, *uṣásā*, *uṣásah* in accordance with the metre: but the forms *uṣāsam*, *uṣāsā*, *uṣāsah* are favoured by the metre, and usually appear in the text, in the following passages: A. 8, vii 99 4b, x 35 2c: 10, i 123 12d, 124 9d 13b, ii 28 9c, iii 31 4c, iv 3 11c, 5 13d, 12 2c, v 1 1b, vi 30 5d, 65 4b, 72 2a, vii 41 7a, 42 5c, 72 4a, 76 4d, 88 4d. B. 2D, iv 30 9c: 2T, iii 55 1a, viii 27 2b: 4B, i 92 2c, 134 4a, v 80 1c, vii 90 4a¹, x 39 1b, *88 18b: 4B, iii 20 1a¹, v 28 1b¹, vii 75 3b¹, viii 96 1a: 7A, ii 12 7c¹.

c12.s3.76

¹ The text has *uṣásam*, etc.

c12.s3.77

In addition the compounds *uṣásānáktā*, *náktosásā* always appear at the beginning of the verse: the text gives the long stem vowel in both cases: the metre favours it in the first compound, and is indifferent in the second.

c12.s3.78

(*b*) In the declension of *-sah* *viśvāsáham* is a probable correction in vi 44 4c (5) and *satrásāham* in viii 92 7a (7).

c12.s3.79

(*c*) Probable restorations, chiefly in the accusative singular, are: Stems in *-man*: *mahimánam* (or *mahitvanám*) viii 46 3a (7), 65 4a (7): *pārijmanam* viii 72 10b (7): *purutmánam* viii 2 38b (7). Stems in *-van*: *anarvánam* ii 6 5b (7), vii 97 5d (7A), viii 92 8a (7); *dhitávanam* iii 27 2c (7), 40 3a (7); *rathayávanā* viii 38 2a (7); *śubhrayāvanā* viii 26 19c (7). Stems in *-an*: *tigmámūrdhanaḥ*, vi 46 11d (7); perhaps *rājanām* v 54 7d (6A) 14d (6A), x 124 8c (7B), *rājanā* x 61 23a (7B?).

c12.s3.80

The regular form *vṛṣaṇam* should probably be restored in ix 34 3a (2D), and in x 89 9d (4B) in spite of the metre: the text has *vṛṣāṇam*, contrary to the general use. For neut. pl. *-ani* there is no sufficient evidence.

c12.s3.81

(*d*) A form *viśváhā* appears somewhat frequently in the text, chiefly at the beginning of the verse, by the side of *viśváhā*. Assuming that the final *ā* is justified (for which point see § 158) there is no metrical occasion, as there is no apparent historical justification, for *ā* in the stem-ending.

c12.s3.82

(*e*) The existence of the locative in *-av* from *-u* stems is only indirectly recognised in the text in the form *sáno* in the combination *sáno ávye* or *sáno avyáye* in bk ix. In the remaining instances which can be recognised by the metre the text regularly gives -*āv*, before consonants -*au*. The locative *vástav* however, in the phrase *vástav usrāḥ*, appears as *vástor* in defiance of the metre in some passages, and elsewhere more correctly as *vástav usrāḥ* or *usrāḥ*. If we disregard the varieties of spelling, we find that the form in -*av* is considerably more common than that in -*āv*: and it may so far be justified historically, that -*av* stands in just the same relation to -*avi* as -*an* to -*ani*.

c12.s3.83

The occurrences are as follows: A¹. 3D, v 52 7b: 5B, ii 27 16d: 6, viii 72 2b, ix 63 8b, 65 16b: 10, *x 167 4a. B¹. 2T, i 126 1b, x 116 9b: 2D, ix 11 5c. C². 5B, x 123 2c. 3D, viii 45 41a, ix 26 5a. E. 5, viii 6 46b⁴: 9, ii 39 3c⁵, iv 25 2b⁵, 45 5b⁵, v 49 3b⁵, vi 46 8a⁵, vii 69 5a⁵, viii 46 26a⁵, ix 86 3c, 91 1c, 92 4c, 96 13b, 97 3a 12d 16d 19b 40c, x 40 2a⁵. Similarly from a stem in -*i* v 41 14d (9).

c12.s3.84

¹ -*āv* is correct.

c12.s3.85

² From this point -*av* is probable.

c12.s3.86

³ *druhyáv*.

c12.s3.87

⁴ *párśav*.

c12.s3.88

⁵ *vástav*.

c12.s3.89

(*f*) A vocative in -*u* in place of -*o* of the text seems probable in i 135 9a (7A), ii 13 13a (6B), viii 25 12c (7B), 51 6a (6B), 52 6a (6B), 70 7b (5), ix 97 17d (9).

c12.s3.90

(iii) The confusion in Vedic Sanskrit between the feminine stems in -*i* and -*ī*, and the frequent transitions from the former class to the latter, prepare us for the occasional restoration of -*i*, which is probable in *vāśibhiḥ* viii 7 32c (7), *x 101 10b (2T): see also §§ 166 ii, 170 ii. On the other hand the restoration *nārībhiḥ* is required in i 43 6c to account for the form -*bhiaḥ* (§ 144).

c12.s3.91

The restoration of an instrumental in -*tī* for -*tyā* has already been discussed in § 139 iii; the examples are given on p. 29 (A 5). Similarly *yuvatéḥ* may be restored for *yuvatyāḥ* x 40 11b (6A), and *táviṣeḥ* for *táviṣyāḥ* v 29 14d (6A). In all these cases the rarity of a long syllable in the second place after the caesura is the justification of the change.

c12.s3.92

(iv) From *ri*, *rī* ‘flow,’ the text gives *riṇā-*, *riṇī-* correctly: and the short vowel must regularly be restored in all similar formations, as in *kriṇāti*, *druṇānā*, *priṇán*, *bhriṇánti*, *śriṇīhi*, in agreement with the history of the forms.

c12.s3.93

171. To complete the discussion above (§ 160) of the particle *ū u*, and as a preliminary to the study of the quantity of final long vowels before hiatus, it is necessary to investigate the relation of the endings -*āv* -*o* in the text of the Rigveda to final -*ā*, as well as the relation of the disyllabic ending -*ā u* to final -*ai*, -*au*, and *āḥ*. Final -*āv* regularly appears in place of a dual in -*ā* before hiatus (except before *u*-), even when the syllable is short. Final -*o* is of various origin: sometimes it must be resolved into -*a ū* or -*ā ū*, the latter element being the particle *ū* or *u*;

c12.s3.94

sometimes it represents a modification of final -ā or a in an adverb, which has the effect of making the final syllable exempt both from combination and from shortening where a vowel follows, this use being almost restricted to the later parts of the Rigveda: and lastly, it is used as a textual correction of -a or -ā in forms of all kinds, with the intention of disguising hiatus or consecutive short syllables in the second and third places. The combination -ā u represents occasionally final -ai, -au, or -āḥ, in all positions in the verse.

c12.s3.95

(i) In about 15 instances (§ 174 i) the final syllable of a dual in -av is shewn by the metre to be short, just as in the locatives in § 170 ii e. Only in one instance is such a dual written in combination, namely in vi 63 1a. This leads to the conclusion that in an earlier state of the text the duals in -ā were rightly distinguished according as hiatus or combination took place, and that a later and mechanical revision has substituted -āv for the ending before hiatus, without regard to the quantity. Of a dual in -au becoming automatically -āv before vowels there is no trace in the Rigveda proper, for final -au becomes regularly either -ā or -ā u, as shewn below (subsection v).

c12.s3.96

(ii) Final -o usually appears in the Pada-pātha as -o iti, but this is no satisfactory indication of its origin or value, for even vocatives in -o, as índo (índav) appear as índo iti. If however we exclude final -o as representing -aḥ, and also the vocatives, we find that in all other instances the vowel in the Rigveda is not capable either of combination or of shortening before vowels: and as it usually ends the first word in a clause, it may be conceived as resulting from the combination of final -ā or -a with the particle ū. This explanation however owes its attractiveness to the theory of the Pada-pātha, represented by its writing ūm íti, that the particle ū has a special character as pragṛhya or unconbinable. As this is hardly the case (see § 173 iv), there is no immediate reason to assume the presence of the particle except where the metre requires the restoration of an additional syllable, namely in ó (á ū) vii 40 1a, viii 7 33a, 22 1a; co (ca ū) vi 66 3b; nó (ná ū) *x 86 2c; pró (prá ū) i 37 2a, viii 62 1a, ix 89 1a; mó (má ū) i 38 6a; hánto (hánta ū) viii 80 5a. In i 186 10a, viii 62 1a, *x 86 2c, the ū thus obtained stands with hiatus before a vowel following (§ 124).

c12.s3.97

(iii) Final -o represents a phonetic developement of ā, incapable of combination or shortening, certainly in átho, probably also in mó and nó. The genuineness of the form átho is shewn first by its appearing only in the popular Rigveda and the Atharvaveda, whereas all revisions of the text of the Rigveda have been carried through the whole of it: and secondly by its use before vowels and consonants alike, although in the latter case no metrical advantage is obtained. Again átho does not stand for áthā ū, for in no instance is it equivalent to áthā ū with hiatus, and it cannot be equivalent to áthā ū combined in the Atharvaveda, for in that case there would be almost as many occurrences of the particle ū in this one combination as in all other possible positions, which is contrary to the principle of quantitative evidence. It follows that *átho* is a real phonetic developement of *áthā*; and, if so, it is impossible that the later tendency to shorten the word to *átha* can have been at work at this period.

c12.s3.98

The cases of *mó* and *nó* are not quite so clear, but are generally similar. In one instance *mó* is to be resolved into *mā ū* (subsection ii): in addition it is used twenty times initially, usually without metrical advantage, seven instances being in the popular Rigveda (*i 50 13d, *ix 114 4d, x *27 20b, *59 4a 8e = 9f = 10f) and only twice elsewhere in the verse, namely in v 31 13b (6B), *x 18 13b (8). So in one instance *nó* stands for *ná ū*: otherwise it occurs ten times in the popular Rigveda, and only twice outside it, namely in iv 21 9c (8), and vi 54 3d (init.). Both words occur occasionally before vowels, and then are neither combined nor shortened.

c12.s3.99

(iv) Much more frequently, it seems probable that final *-o* of the text represents the result of a metrical revision of final *-a* or *-ā*, and is betrayed as such by its systematic use in positions in which it is metrically convenient, that is, before hiatus, and (in the case of *-a*) in the positions 2T, 2D. Such a revision must have been the work of a later hand than that which recorded the value of the ‘protracted vowels,’ and has not as much historical justification. For in the case of the protracted vowels it appears that the words received special treatment from the bards (Table, § 164): but the words in which *-o* appears are used exactly as ordinary words with final short or long vowels respectively. As examples we may take the three words which appear most commonly, *ápo*, *utó*, and *úpo*.

c12.s3.100

Of these *utó* appears always initially, and before consonants 40 times; in 34 instances the next syllable is short. On the other hand *utá* appears 88 times initially, and in 79 of these the next syllable is long. It is clear then that *utó* stands according to a system before short third syllables. That this system was that of the bards themselves seems improbable: and the more so, because we find *utó nú* vi 47 1c, viii 25 23c, 72 6a 18a, 94 6a, where *nú* may be read, and *utó nú cit* viii 40 10c 11c, *utó ghā* vii 29 4a, indicating that the editor concerned reckoned *nú*, *gha* as short syllables, although *nú cit*, *ghā* are the regular forms. Further the general practice of the poets (§ 157) does not lead us to expect that *utá* would be followed by a long third syllable in so high a proportion of the occurrences as 90 per cent.

c12.s3.101

Further *ápo* occurs four times before consonants: in three cases the following third syllable is short, but in viii 67 15a the metrical difficulty is easily removed by reading *ápa sú*. *úpo* occurs thirteen times before consonants, in all cases but one in the position 2T or 2D, but the metre can be corrected by reading *úpa sú* in i 82 1a, viii 103 1c, ix 61 13a.

c12.s3.102

In other words ending in *-o* for *-a* the usage is the same, and though the particle *ū* may in some cases be present, the restoration of final *-a* seems generally most probable, with such consequent corrections as *ihá sú* ix 65 5c, *kéna nú* viii 66 9c, *tátra sú* i 37 14c, *bhūyáma sú* iv 32 6a, and *sahá sú* viii 7 32a. The chief exception (besides *nó* subs. iii above) is *pró*, which seems very frequently to stand for *prá ū*, except in *i 161 12d*, where *prá* should be restored.

c12.s3.103

Where *-o* stands for *-ā* (as in *etó*, *evó*, *kathó*, *tápo*) either a vowel follows or one of the particles *nú* or *sú*. In the latter alternative *ū* is probably contained: in the former it is more than doubtful.

c12.s3.104

(v) Since the particle *ū u* as such is regularly the second word in the clause, or at latest the third when the second word is a similar particle, it must be clearly distinguished from the *u* which attaches itself to certain verb and noun endings, and therefore regularly appears later in the sentence. This is the case with the ending *-tavai* (*-tavá u*), with the words *asaú* (*asá u*), *vaí* (*vá u*), and with the nom. sing. and other endings in *-āḥ* (*-ā u*).

c12.s3.105

The occurrences are *-tavá u* : *i 24 8b*, *164 5d* *28b*, iv 21 9d, *58 9a*, v 2 10b, 29 2d, 31 4d, *83 10b*, vii 4 8b, 44 5b. *vá u* : *i 162 21a*, ii 33 9d, v 73 9a, vii 85 2a, *104 13a*, viii 23 13a, 62 12a, *ix 112 1a*, *x 10 12a*, *27 5a*, *117 1a*, *137 6a*, *142 3a*. *asá u* : v 17 3a. *-ā u* (nom. sing. or pl.): i 156 1b (*evayá u*), 168 1b (*devayá u*), iv 23 10b (*turayá u*), vii 68 4a (*devayá u*), 88 2c (*adhipá u*), x 65 10d (*dhanasá u*). Also the neut. nom. sing. in *bhá u* i 46 10a and the gen. fem. sing. in *ánumatyā u* *x 167 3b.

c12.s3.106

The forms given above are more common in the popular Rigveda than elsewhere.

c12.s3.107

The reading *-sītāsa ū júvaḥ*, in i 140 4b, requires some similar explanation, but as the vowel is long in this case the analogy is not exact, and *ū* may very well belong to the word following. So *ū jananta* ii 18 2c.

c12.s3.108

172. The quantity of final *-ā* before hiatus, and of final *-ī* *-ū* with hiatus before similar vowels, is generally optional: but the dual forms usually retain long quantity.

c12.s3.109

Examples are: *aśvayá utá rathayá* viii 46 10b *vápurbhir á carata anyá-anyā* i 62 8d.

c12.s3.110

If we take into account only the instances in which *-ā* with hiatus is certainly to be restored, it will be seen from the list below that the short value on the whole predominates, but not to such extent as to justify the rule *vocalis ante vocalem corripitur*. In particular, there are numerous occurrences in the eighth place, and the number of occurrences before early and late caesura is equal. If further we include those cases in which the final *-o* of the text represents a slight phonetic variation of long final *-ā*, as explained in the last section, the proportion of instances with the full quantity will be greatly increased.

c12.s3.111

It may however be noticed that the short value very much predominates in the nom. sing. of nouns in *-ar*.

c12.s3.112

Instances of final *-ī* before a similar vowel are so rare and so indecisive that the rule affecting them can only be inferred from the analogy of final -ā. For -ū we have only the restored phrase nú ū (§ 124): and as nú is then always initial, the quantity cannot be determined. In internal combination we find su-uktá and other words, always with -u.

c12.s3.113

(i) The following are probable instances of -ā with hiatus: A. 3T, ii 19 3d, x 61 9c: 3D, viii 46 10b: 6, viii 17 1b, 34 11b: 8, i 48 7c 16c, 87 4c, 185 4a, ii 17 1b, iv 16 1a, v 29 15b, vi 20 8d, 24 9a, vii 1 7a, 40 4b. B. 2T, v 45 2d, vi 23 4a², *75 3d, viii 101 13c, x 30 2b, *129 5d: 2D, v 53 14b, x 26 1d 9b: 2T, 6 times: 4A, 48 times: 7A, ii 13 10a, iv 2 18b, 19 6c, vi 4 4b, viii 103 13a, x 61 18c, *101 3b, 132 4d. C. Initial, 15 times: 2D, 8 times: 4, 5 times: 5B, 48 times: 3TD, 17 times: also vi 16 27a (6 in Troch. Gāyatrī). D. 5A, i 189 4d: 6A, i 60 4b², 104 5a, 186 6a², iv 33 10a¹, vii 69 3a¹, x 49 10b², 65 15b¹. E. 5, i 133 6c, iii 13 1b, iv 52 2a, *v 51 15c, viii 20 17c, 46 29b, 70 12d, x 105 11b 11c: 7A, i 120 6b: 9, i 62 8d, 69 9a, 77 1a, 104 1c, 133 6b, 140 13c, 173 4a, iv 3 13b, 16 1a, v 41 16b 16c, vi 21 8d, 24 5d², vii 34 4a, 40 3d², viii 1 16c, 70 12c, 96 9d, x 5 †5d (reading pūṣā ásya), 61 14d².

c12.s3.114

¹ caesura after third syllable.

c12.s3.115

² nom. sing. of noun in -ar.

c12.s3.116

(ii) In the following cases the text has -o representing -ā before hiatus (§ 171 iii iv); átho i 28 6c (2D), 113 13b (2T), *191 2c (2D), ix 39 5b (2D), x *27 9d (2T), *60 8e = *9e = *10d (2D), *85 35b (2D), *96 13b (2T), *137 4b (2T); aryamó i 167 8b (7A); eṣó i 46 1a (2D); áviṣṭo vii 34 12a (3T); ímó vii 1 18a (2T); cattó *x 155 2a (2D); mó *i 50 13d (in.), v 65 6d (in.), *x 18 13b (8); rákṣo i 174 3c (2T); víṣo iii 27 14a (2D), viii 33 10d (2D).

c12.s3.117

(iii) Instances of hiatus of -ī before similar vowels are found vii 28 3a (5B), *ix 5 8b (4), *x 88 4c (4A). For the dual forms see § 174.

c12.s3.118

173. The vowels -ī, -ū are regularly shortened when followed by dissimilar vowels, but there are many exceptions. In particular, duals regularly, and the particle ū most commonly, retain the long quantity.

c12.s3.119

The diphthongs -e -o become short syllables when followed by any vowel, except that duals in -e, adverbs etc. in -o, and the locative tué tvé regularly retain the long quantity.

c12.s3.120

The diphthongs -ai -au are regularly represented by long vowels when followed by any vowel.

c12.s3.121

In all these cases there is authority even in the text of the Rigveda for considering that there is no real hiatus, but a resolution into a vowel and a semi-vowel, or in the case of -o for -ah, a retention of the original ending -as. Thus in the declension of dhí the text gives dhíyam for dhí-am, and in the vocative of -u stems -av for -o.

c12.s3.122

Consequently the regular use may be represented by such restorations as the following: dakṣiṇāvāḍ vājīnī práciy eti iii 6 1c píbā tú sómam gáv-rjīkam indra vi 23 7b yuvó rájāṃsi suyámāsas áśvāḥ i 180 1a.

c12.s3.123

Exceptions are probable as follows:

c12.s3.124

(i) final -ī is retained (a) in the nom. sing.; jānatí i 122 9a (3T); pṛthiví i 94 16d (8), iii 8 8b (7A), *x 10 5d (8); vājí x 56 3a (2T): (b) in the instrum. sing.; ūtí vi 29 6b (2T), viii 21 7b (2T), susámī *x 28 12a (8): (c) in the imperative in -hī: ihī i 80 3a (2D), jahī i 36 16a (8), pāhí viii 60 9b (2D), mimīhī vii 19 11c (8), śnathihī vii 25 2a (8); in -dhī, yuyodhī i 189 3a (8).

c12.s3.125

(ii) Final -ī of a noun-stem is regularly retained in the gen. loc. dual ródasīyoḥ, which is always found in Triṣṭubh cadence: and a suffix -īya is probable in yavīyā i 167 4a (10), 173 12c (10). Other instances are very uncertain, such as pṛthivīyám i 139 11b, x 73 9c, and yajñanīyam, etc., *x 88 17b, *107 6b (all 3T). The text reads devāyàm in six passages at the beginning of the verse, and the protraction of ā is probably correct, as pratīvíam regularly occurs in dimeter cadence.

c12.s3.126

(iii) The rule for forms in -ū depends largely upon analogy, since the only forms that occur rather favour the long quantity, namely tṛṣú i 58 2b (2T), sādhú (§ 160 i) ii 3 6a (2D), and tanú *x 183 2b (4A). The forms tú nú sú are frequently found before hiatus, as tú iii 30 12d (9), nú i 100 10b (9), vii 96 4a (5), sú i 111 2d (9): but in these words the short forms are also found before consonants. On the other hand the restorations proposed in this chapter frequently require that nú and sú should retain their long quantity even before vowels, namely nú *iv 18 4c¹, vi 47 1c², viii 3 13c¹, 72 6a² 18a², 94 6a² (all 3T or 3D); sú ix 65 5c² (3D), *x 59 3a³ (3T). So also tú vii 31 4c (3D)⁴.

c12.s3.127

¹ § 175 ii.

c12.s3.128

² § 171 iv.

c12.s3.129

³ § 167 iii.

c12.s3.130

⁴ § 162 iv.

c12.s3.131

(iv) In the case of ū u the long vowel is most favoured before hiatus, though both forms are common: instances in decisive positions are: ū ii 24 6d (3T), iii 5 2a (3T), vi 71 5a (2T), vii 81 1a (2D), viii 15 1a (2D), x *86 2c (2D), *88 10c (2T): u (chiefly in the popular Rigveda) *v 30 15d (6A), vi 27 1a (7B), * *27 7a (7B), *102 10b (6B), *149 3d (6B).

c12.s3.132

(v) The locative tué, tvé always retains the long vowel before hiatus, as in ii 5 8c (6): whereas asmé is always shortened, as in viii 22 16c (9), and yuṣmé presumably has the same value. Vocatives in -o are shortened, as indav ix 40 4b; for other words in -o see § 171. For the duals see the next section.

c12.s3.133

(vi) Final -ai, -au seem regularly to become -ā before a following vowel, as in classical Sanskrit. There is however in the Rigveda a marked distinction between the group asmai, tásmai, yásmai and such words as asaú, vaí, and the ending -tavaí. The former are perhaps occasionally shortened, as yásma v 7 8a (5), tásma *i 161 12d (7A), asma v 64 2b (5): whereas the latter develope into -ā u, as shewn in § 171 v. asaú appears as asáv x 132 4a.

c12.s3.134

174. All duals, both of nouns and verbs, are regularly long before hiatus: but to this rule there are many exceptions in the earlier parts of the Rigveda.

c12.s3.135

Examples of the regular usage are:

c12.s3.136

jrayasānā́ áram prthú v 66 5c
(text jrayasānā́v)
pāthó ná pāyúm jánasī ubhé ánu ii 2 4d
tá hí kṣatrám dhāráyethe ánu dyún vi 67 6a.

c12.s3.137

(i) Exceptions are relatively few, but still certain, namely:
(a) -a before hiatus: i 184 1c (9), iv 15 9a (5) 10a (5), v 74 1a (5), 86 5b (5) 5d (5), vi 63 1a (9), 67 8c (9), vii 65 2a (9), viii 5 31b (5), 25 4b (5), 26 7c (7A), 35 24b (5), x *90 11d (5), 132 1c (5), 143 3a (5). For the text see § 171 i.
(b) -i: vi 60 13a (7B), *x 94 9a (6A).
(c) -u: i 46 13c (5), ii 27 15d (9), v 43 4a (9).
(d) dual of a noun in -e: ubhá (ubhé) v 30 9c (7B).
(e) duals of verbs in -e: vii 93 6c (9), viii 26 13c (9), 29 9a (7A). In the dvandva compound indrā-agní, -ā is found wherever hiatus occurs: cf. § 166 i.

c12.s3.138

(ii) It will here be convenient to deal with the question of possible duals in -a, -i, -u before consonants. Occurrences in the first part of a dvandva compound have been dealt with in § 166 i. In the following instances the text has -a, but the metre favours or at least permits the restoration of -ā: i 15 6a (fin.), 151 4a (7A), v 64 6a (fin.), 66 6b (2d), vi 68 5b (7A), vii 60 12a (4A), 61 1a (7A), 85 4b (4A), viii 9 6b (4). Similarly the restoration of -ī is favoured by the metre for prthivī in ii 31 5c and iii 54 4d (both 7A): and that of -ū is admissible in vii 19 4d, 65 1d (both final). In four instances in i 17, namely 3b 7a 8a 9b, a dimeter verse begins with índrāvaruṇa, and the restoration of índra váruṇa on the analogy of varuṇa mitra in i 122 7a is not improbable: similarly mítra váruṇa may be restored in i 15 6b. Besides we have only deva v 67 1a (5), vīra vi 63 10c (7B) and hótāra v 5 7b (5), sáviṣṭha vi 68 2b (7A), mamhiṣṭha vi 68 2c (7A), vipanyu viii 8 19c (5) are favoured by the metre; but these seem insufficient as evidence of a fresh variation in so common a form. It seems therefore more than doubtful whether a dual form in -a exists before consonants: if so, it is of the vocative case only.

c12.s3.139

175. It appears from the whole course of the discussions in this chapter that the Saṃhitā text stands in need of substantial revision, not only in questions of external and internal Sandhi, but also in questions of quantity. It follows that isolated quantitative readings of the text, contrary to its general practice and the general tenour of the metrical evidence, must always be received with considerable hesitation: and for the purposes of metrical investigation it seems safer to assume in such cases that the words and endings have their ordinary value.

c12.s3.140

In the case of final vowels before hiatus, the text follows mechanical rules and entirely disregards the metre; so that in this point the Saṃhitā revision is precisely of the same character as in questions of Sandhi, and all modern criticism is agreed in requiring restoration. In the case of the variant final quantities, on the other hand, the Saṃhitā is guided partly by considerations of grammar, that is to say, by the recognized value of the endings in the editor's own time, and partly by the metre, so far as he was able to appreciate it. This work is therefore executed in a more intelligent spirit, and probably at an earlier date, than the Sandhi revision: but the more exact methods of modern literary criticism make it possible in the majority of forms to establish independently rules which are more uniform and at the same time more in harmony with the general character of the metre. Lastly in the case of final -o we can recognize in some words such as átho a state of the text so early that it recognizes the historical differences between various parts of the Rigveda, and in others like utó a restoration on purely metrical grounds, which produces an artificial regularity of metre going greatly beyond the regular practice of the bards.

c12.s3.141

That the text of the Rigveda should have been corrected by successive editors, and from different points of view, is in itself sufficiently probable: and under such circumstances no restoration of the original text can claim to be more than approximative. But whatever difficulties are involved, it is at least safe to say that it is better to face them than to treat as a text affording a basis for metrical investigation one that is so full of inconsistencies as the Saṃhitā. We constantly find in hymns belonging to the same group, and even in the same hymn and the same verse, differences of quantity for which no metrical or other cause can easily be imagined, and which it is therefore impossible to ascribe to the author. Such inconsistencies are illustrated in the passages quoted below in pairs, the number of which might be indefinitely increased:

c12.s3.142

{yénā pṛthivīám ní krívim śáyadhyai ii 17 6c
{yéna jánā ubháye bhuñjaté víśaḥ ii 24 10d
{kṛdhí no rāyá uśijo yaviṣṭha iii 15 3d
{kṛdhí rátnam susanitar dhánānām iii 18 5a
{āvíṣ karta mahitvaná i 86 9b
{jyótiṣ kartā yád uśmási i 86 10c
{píbā sómam mahatá indriyáya x 116 1a
{píba rāyé śávase hūyámānaḥ x 116 1c

c12.s3.143

The extent to which correction is required in each class of forms is discussed in the next section.

c12.s3.144

An isolated variation in the text may in some cases be a trace of an earlier recension, and may serve as a valuable confirmation of the indications of metre. But if it only has the effect of removing an occasional metrical irregularity, the most probable explanation is that it is due to excessive regard for the metre. In all the instances that follow the text gives long quantity to syllables which have ordinarily the short vowel, and which appear so rarely in positions that favour long syllables that there seems no reason to look for any other explanation than the usual metrical liberty of the Rigveda.

c12.s3.145

The occurrences are as follows:

c12.s3.146

(i) in flexional endings: *-á* nom. s. masc.; *sá* i 145 1b (8), vi 17 9a (8): *-a* vocative; i 61 16a (8), viii 45 22a (6) 38a (6): *-asya* genitive; *i* 162 19a (8), vii 79 4c (8): *-eṣu* loc. pl.; viii 31 9d (3D): *-ti* 3 sing.; *rákṣatī* *riṣáḥ* ii 26 4c (10) on the analogy of *rákṣatā* *riṣáḥ* ii 34 9b, but see § 167 iv: *-ta* 3 sing.; ii 33 13c (8), *vii 33 2d (8), viii 19 24a (8), x 25 7d (final syllable misunderstood for 8): *-ta* 3 pl. iv 5 5d (8), 34 1d (8). A slight doubt exists as to the neut. sing. in *-ā* from stems in *-an*, *-man*, *-van*. The form *bhúmā* is found in i 61 14b (5C), 62 8a (8), 173 6c (fin.) and appears to be a quasi-dual on the analogy of the preceding *dyávā*: other forms are found in the text in i 69 3b (8), iv 2 16d (2T), vii 28 1a (2T), x 45 4b (2D), *129 1b (8), *142 2a (8), 176 1c (2D); whilst *námā* is favoured by the metre in i 48 4d (2D), and *bráhmā* vii 24 4b (2T). Cf. § 166 v.

c12.s3.147

(ii) in particles: *áva*; vi 46 11b (6): *ihá*; iv 5 14c (8), 31 11a (4), x 178 1d (8): *ca*; i 77 2d (2T), iii 57 5d (8), viii 53 4a (10: the text has *cā*, i.e. *ca* *ā*): *caná*; i 84 20b (6): *ná*; i 147 3d (8)=iv 4 13d (the Saṃhitā has *náha*, Pp. *ná* *áha*, TS. correctly *ná* *ha*), *vii 104 13a (init.), *x 34 8c (8): prá-pra*; i 129 8a (2T): *ha*; iv 31 5b (2D), v 41 7d (2T). The text also has three times *nahí* *nú*, and twice *nahí* *nú* at the beginning of the verse: *nahí* *nú* is probably the correct reading throughout, namely in i 80 15a, 167 9a, vi 27 3a before consonants, and in *iv 18 4c, viii 3 13c before vowels (§ 173 iii). There are also occasional instances in compounds, as *iṣṭā-pūrtá* *x 14 8b.

First 147 of 213 passages. Read on

Cites (2)

Ṛgvedasaṃhitā — Sanskrit
Atharvavedasaṃhitā — Atharvan & Aṅgiras, Sanskrit