¹ I have to thank Prof. Franklin Edgerton for much help and advice in my present work in Sanskrit etymology (I have to thank him for more); I thank Prof. G. L. Trager and Dr. I. Dyen of Yale University for advice on English and Dr. H. I. Poleman of the Library of Congress for information.—

Citation:

B-R = O. Böhtlingk und R. Roth, Sanskrit-Wörterbuch, St. Petersburg, 1858–75.

Böht. = O. Böhtlingk, Sanskrit-Wörterbuch in kürzerer Fassung, St. Petersburg, 1879.

Dict.PTS = T. W. Rhys Davids and W. Stede, The Pali Text Society's Pali-Engl. Dict., Chipstead, Surrey, 1921.

Grierson, Vernaculars = G. A. Grierson, On the Modern Indo-Aryan Vernaculars, Indian Antiquary 1931–33.

Hc. = Hemacandra's Grammatik der Prākritsprachen, hgg. von R. Pischel, Halle, 1877. Sheth = Hargovind Das Sheth (Śeth), Pāia-sadda-mahannavo (Prakrit Hindi Dict.), Calcutta, 1923.

Turner = R. L. Turner, A Comparative and Etymological Dict. of the Nepali Language, London, 1931.

The New Indic sources listed by Turner p. xxiii are referred to by the sole surname; so are the following, which do not appear in Turner:

M. Bronson, A Dict. in Assamese and Engl., Sibsagor, 1867.

W. Carey, A Dict. of the Bengalee Language, Serampore, 1825.

Lodiana Mission, A Dict. of the Panjábí Language, Lodiana, 1854.

Bh. N. Mehta and Bh. Bh. Mehta, The Modern Gujarati-Engl. Dict., Baroda, 1925.

G. C. Praharaj, A Lexicon of the Oriya Language, Cuttack, 1931–36.

² Details below §16.

³ Discussed below §17.

⁴ ‘Diese im Epos und auch bei Kālidāsa, wie es scheint, noch nicht vorkommende Wurzel ist wohl aus mith- hervorgegangen.‘

⁵ The original Hindī: saṁgat honā, mel rakhnā; svīkār karnā.

⁶ Except a causative of saṁgacch- in Singhalese, Turner 646 b sub gachnu.

⁷ The original Hindī: ikaṭṭhā honā, saṁhat honā, sammukh ānā, āgaman karnā, ānā; prāpt karnā; jānnā.

⁸ Some of the meanings of saṁ-gam- and sam-i-, that do not yet appear with mil- in Sanskrit (‘to cohabit’, ‘to agree’, ‘to fit’), do so in New Indic, cf. below §§5 and 14; only ‘to know’ does not.

⁹ The root-inventories were unconcluded (aparisamāpta-) and grew (vardhate dhātugaṇaḥ) (Vāmana l.c.).

¹⁰ Cf. Turner sub milnu, milāunu (509 a) and mel (518 b).

¹¹ But the substantive mela ‘meeting, assembling’ also in Singhalese.

¹² In Oṛ., Guj., Mar. with ḷ.

¹³ Capt. G. Stack, A Dictionary, English and Sindhi, Bombay, 1849.

¹⁴ Grierson, Dict., and Manual 101.

¹⁵ J. T. Molesworth and T. Candy, A Dictionary English and Maráthí, 2d ed., Bombay, 1873.

¹⁶ The Beng. intransitive mil- is not given the meaning ‘to mix’ by Carey and Subal Chandra Mitra.

¹⁷ A Dictionary of the Bengalee Language (no author's name), vol. ii, English and Bengalee, 4th ed., Serampore, 1847.

¹⁸ The Nep. intransitive mil- is not given the meaning ‘to mix’ by Turner.

¹⁹ Hemacandra's other substitute for missaï, vīsālaï, which otherwise seems to be found only in Hemacandra's Deśīnāmamālā (Grierson, Dhātv-ādeśas 153), in view of Hc. 2.170 mīsāliaṁ = mīsaṁ, Apabhr. misala- and abs. misalivi (cf. §8), and Mar. misaḷ ‘mixed’, misaḷṇẽ ‘to mix’, may be a corruption for *mīsālaï; phonetic change m > v is less likely, because initial m otherwise seems to become v only by dissimilation with following nasal: AMāg. vīmaṁsā = mīm° etc., Pischel §251.

²⁰ Westergaard, Rad. 246 a; Monier-Williams 817 c.

²¹ Beside misseti ‘to mix’, Childers 248 a has also a missati ‘to be mixed’, ptc. pres. missa-māna- Mahāvaṁsa 234. Dict.PTS does not have this heading, and in fact for Turnour's (1837) missamānaṁ (madhuram iva visena missamānaṁ ‘like something sweet mixed with poison‘), Geiger's edition (1908), Cap. 36.133 c, reads missam annaṁ, with the Ṭīka (Commentary) and a former edition of 1883 (the mss. have only missamānaṁ, °pannaṁ, and °janaṁ).—In itself, a Pāli intransitive *missati would be quite possible, cf. below §12.

²² Probably a secondary alteration of earlier *missei.

²³ Grassmann, Wb., sub mih-, has ‘besprengen’ etc., Geldner, Rigveda in Auswahl, Glossar, ‘würzen’.

²⁴ Cf. Turner sub misnu, misi etc. (509 b).

²⁵ T. Grahame Bailey, Gram. of the Shina Lang. (1924) 153.

²⁶ Thus Turner 509, probably after Barua; Bronson has mihalā-.

²⁷ Thus, ‘v.n.’, Carey 1193 b; less precise Subal Chandra Mitra, but also he has miś-‘to mix, mingle; unite, come together’.

²⁸ Praharaj.

²⁹ J. Bloch 388 b gives also a simple Mar. misṇē ‘to mix’; I do not find this in Molesworth.

³⁰ The Bengālī and Oṛiyā versions are influenced by the Sanskrit one (or, rather, Oṛiyā directly by Bengālī).

³¹ Thus, mel- and bhel- with l, not ḷ, the text (Surat 1867), which, however, seems not to distinguish l and ḷ.

³² Thus, miḷ- with ḷ, the text.

³³ Stack 90 a gāḍruṛu ‘mixed’.

³⁴ Just as New Indic milā- is a synonym of miśr- ‘to mix’ trans., the intransitive mil-coincides with the Skt. intransitive miśrī-bhū- in the specific meaning ‘to cohabit’ in Kāshmīrī, Panjābī, Hindī, and Gujarātī (Robertson, Engl.-Guj. Dict. ‘to pair’ intr. maḷ-).

^34a This distribution: Rig-Veda preponderantly -miśla-: Atharva-Veda miśrá- is abnormal; otherwise, when Rig-Veda and Atharva-Veda differ, Rig-Veda has r and Atharva-Veda l: RV rap- ‘chatter’, rikh- ‘scratch’, aśrīrá- ‘ugly’: AV lap-, likh-, aślīlá- (Macdonell, Ved. Gram. 43).

³⁵ As Hc. lāhaï stands without meaning and is not attested otherwise, its definition is only based on the variant sāhaī (thus, ī) in mss. C. and E., and the gloss ślāghati (in C.? cf. p. vii). Pischel ad Hc. 1. 187 (1880) equates lāhaï with läghate (= räghate, both only Dhāt., ‘sāmarthye; to be able’), ‘not = ślāghate cf. He. 4.88’ (where as substitute of ślāgh-is given salahaï),— but Gram. §315 (1900), with ślāghate. (In fact, the existence of salahaï does not exclude a doublet lāhaï.)—New Indic has Hi., Panj., Si., Guj., Or. sarāh- ‘to praise’ (Pischel ad Hc. 4.88, Turner sub sarāunu, 590 a) (r-form; different from Old and Middle Indic); Sindhī also salāh- (precisely Sindhī has otherwise r for l in old words, cf. note 41).

³⁶ The treatment sal° of ślakṣṇá-, combined with the treatment kṣṇ > kk(h), appears, however, in Nep. salakka ‘smooth’ (erroneously Turner 592 a).— Another variant, again with kkh, is Asoka-Pāli sakhina- in sakhinālaṁbhe ‘of gentle actions’ Dhaulī Sep. Ed. 1.22.

³⁷ Thus correctly Wackernagel §221β (1896); Pischel §§268 and 330 (1900) wrongly equates lhasaï ‘to fall’ with Skt. hrasati S.+ ‘to become short, be diminished’, Dhāt. ‘to sound’, hlasati, only Dhāt., ‘to sound’ (’śabde’).— Wackernagel l.c. suspected another case of internal lh from sl in Pāli “usselheti ‘aufspringen’ ”, which he thought to be a reduplicated present *sisla- belonging with Skt. ucchala- Class. ‘to leap, jump, rise’, from *ut-sal-, and farther Gk. ἅλλομαι, Lat. saliō. However, this word, if identical with Dict.PTS 159 a ‘usseḷheti (?) Vin. 2.10 (for ussoḷh°?)‘, is a ἅπαξ of doubtful reading and meaning.

³⁸ But cf. note 21.

³⁹ with the typical North-Western change śr > ṣ (cf. recently T. Burrow, The Language of the Kharoṣṭhi Documents from Chinese Turkestan [1937] §38); Beng.-Oṛ. miś- with the Eastern one śr or śl > śś (cf. Māg. adj. miśśa- Pischel §64) (Oṛ. ś in the dialects bordering on Bengālī; pure Oṛiyā has s, Grierson, Vernaculars §333).

⁴⁰ l in -miśla- would be still older, in case this word had IE l (thus Wackernagel §192 b), and IE l were (with Wackernagel) directly continued in part of Proto-Indic.

⁴¹ Turner sub mil- 509 a; cf. Sindhī garo m. ‘throat’, biru m. ‘hole’ = Skt. gala-, bíla- etc., Turner, JRAS 1924, 576.— However, l might possibly be the regular outcome in Sindhī of original lh (*milh- from *miśl-).

⁴² Summarizing, the following forms are postulated: A. Phonetically: 1. Old Indic miśr-and *miśl-; 2. Middle Indic miss- (subtypes NW *miṣ-, East *miśś-) and mil-, the former from miśr-, and, except for *miṣ-, perhaps also from *miśl-, the latter from *miśl-. (The type *miśl- > miss-, *miśś-, therefore, is grouped in Old Indic with the type *miśl- > mil-, but in Middle Indic with the type miśr- > miss-, *miśś-.) B. Morphologically: causative miśrayati, *miślayati in all dialects; intransitive *miśrate, *miślate only in some: mil- in the type *miśl- > mil- and miś- in the type *miśr- > *miśś- or *miśl- > *miśś-.

⁴³ However, locative appears beside instrumental also in the predecessor saṁ-gam-, cf. kathaṁ . . . mayi saṁgatāḥ ‘woher haben sie sich an mich geschlossen’ Bhāg.Pur. (B-R 684); it might, therefore, stem from this.

⁴⁴ Just as the later mil-, the TS l-form is from an a-, not an aya-denominative.

⁴⁵ Grierson, Vernaculars §277.

⁴⁶ Cf. Turner sub bhelā 483 a (the words with other meanings to be set aside).— Sheth 815 (followed by Turner) gives bhel- ‘to mix’ also for Prakrit, quoting two forms: bhelaïttā from Pischel §206 (end), and bhelaviya- from a manuscript. But Pischel, who gives no meaning, refers ‘AMāg. bhelaïttā (Ṭhāṇaṅgasutta 421)’ to Dhāt. (Vopadeva) bhil- ( = Pāṇ. bil-) ‘bhedane’, which hardly speaks for a meaning ‘to mix’. (The Ardha-Māgadhī Dict. of Ratnacandraji does not have the word.)—

As Mar. bhel- has both ḷ and l, while meḷav- has only ḷ, and as bheḷ- (bhel-) has in Gujarātī optionally, in Marāṭhī exclusively, the short form concordant with Hi. melnā, whereas meḷ- in both languages has only the specific South-West causative with -av-, meḷav-, the suspicion arises that the bh-form may stem from a Hindī-dialect (Hindī itself has only the substantive bhel ‘mixture’). [One could also think of the possibility of Mar. (and Guj., see below) bhel- beside bheḷ- representing old *milh- beside mil- (cf. note 41 on Si. mil-), but this is hardly likely.]— Whereas Mehta, Gujarātī-Dictionary, and Tisdall, Gujarātī-Grammar, give only ḷ-forms for Gujarātī (Tisdall 73 maḷ- ‘to meet, be got’, caus. meḷav-, bhaḷ- ‘to be mixed’, caus. bheḷav-, moreover, p. 175, meḷvavũ ‘to mix, get’), Robertson, Eng-lish-Gujarātī Dictionary (1854), has also l-variants: bhelvũ and -vavũ ‘to mix’ beside bheḷvũ, meḷvavũ ‘to mingle’, malvũ ‘to pair intr.’ beside maḷvũ ‘to meet’ etc.; in view of Mar. bhel-beside bheḷ- these forms were perhaps not entirely without foundation.

⁴⁷ ‘Skt. bhela, root mil’ says Platts 200 b sub bhel f. ‘mixture’. (But what is ‘Skt. bhela‘?)

⁴⁸ E.g. in Jacobi's Jaina-Māhārāṣṭrā the normal form is sumaraï, ptc. sumariya-; bhar- in ptc. bhariya- (twice) and saṁbharei, -iya-; sar- in saraṇa- ‘remembrance’; with vi-: visumaraï and vissāriūṇa ‘to forget’; no *mar- or *-mhar-.— Hc. 4.74 for smárati: bharaï, bhalaï, sumaraï, saraï (besides other words); 4.75 for vismarati: vīsaraï, vimharaï (and pamhusaï).

⁴⁹ The meaning ‘to reckon, count’ of Si. sambhāraṇu and sāraṇu, which seems specifically Sindhī, again (cf. my remarks Bull. Soc. Ling. 23.116) strikingly agrees with Iranian: Here from Middle Iranian on means ‘to count’ (MPers. ušmār-, Pers. šumār- ‘to count’, Chr. Soghd. pčmrtÿt plur. ‘counted’), while for ‘to remember’ there emerge other words.

⁵⁰ Cf. Pischel ad Hemacandra 4.74; Turner sub samālnu 587 b, with unfounded hesitations, and sub sambhārnu 589 a.

⁵¹ Uhlenbeck 197 a and Turner 652 b compare bhāla- n. ‘lustre’ and ‘forehead’.

⁵² Traditional text: aomča sūkəm yim baraiti karō masyō upāpō urvaēsəm mārayeite.

⁵³ Traditional text: ašta rātayō vīspāhu paiti barəzāhu vīspāhu vaēδayanāhu spasō hišpōsdmna ave aipi daiδyantō ave aipi hišmarəntō yōi paurva miθrəm družinti.

⁵⁴ Thus Lommel with Bartholomae, Wb. 871; Wolff, Übersetzung: ‘die zuerst den Miθra betrügen’.

⁵⁵ Cf. Grierson, On the Phonology of the Modern Indo-Aryan Vernaculars, ZDMG 50.20, and Vernaculars §369 (who cites bhaīs and bheḍā).

⁵⁶ Not in Turner, but cf. Lodiana Dict. (1854) 369 c.

⁵⁷ On the pronunciation of majjh (majh) cf. Grierson, Vernaculars §363.

⁵⁸ Turner sub māsi 506 b.

⁵⁹ Beside Pers. mäγz; ‘marrow’ only mäγz. In Gujarātī the inverse.

⁶⁰ The bh-words are not mentioned by Turner l.c. But Platts 199 a: bhejā ‘Skt. medas or majjas’.— Panj. bhéjjā m. ‘brain’ is from Hindī.

⁶¹ Walde-Pokorny's explanation (2.309) of Pāli miñjā- as ‘redupliziertes *mi-mzgho-‘ is false; the correct interpretation, palatalization of a before jj and secondary nasalization, in W. Geiger, Pāli §9, note 1, and §18.2 (erroneously Pischel §§101 and 74).

⁶² Cf. Turner 518 a and 482 b resp.

⁶³ Zachariae, Die indischen Wörterbücher 23.

⁶⁴ ‘The common form’ Platts.

⁶⁵ Besides, mẽḍā, f. -ī ‘goat’.

⁶⁶ Without reference to sex.

⁶⁷ Except Nepālī, where meṛo only ‘ram for sacrifice’.

⁶⁸ Thus already Carey, Bengālī Dict. 1208 b, in 1825.

⁶⁹ An original *meṣṭra- or *miṣṭra- would probably cover all forms, but is difficult formationally.— Grierson's derivation from *mesa-ḍa- (with suffix -ḍa-), *mehaḍa- (ZDMG 50.20 and Vernaculars §369) contravenes the r-, ṭ-, and ṇ-forms (the latter two presupposing double consonance).

⁷⁰ Cf. Turner sub muṛo 513 and 653 b, and sub bhutte 479 b.

⁷¹ According to Przyluski, Bull. Soc. Ling. 30.196–201, the fluctuation m/bh in meṇḍha-/ bheḍa- (200), muṇḍa-fbhaṇḍu- (199), and other words (e.g. Skt. bhallūka- MBh., BhP. and mallu- Lex. ‘a bear‘), would be due to borrowing from Austro-Asiatic (e.g. Muṇḍa). This may possibly be true for single cases; but that the fluctuation m/bh essentially results from an intra-Indo-Aryan change mh > bh, is shown by its appearance in Proto-Indo-European words like smárati and majján-, and a Rig-Vedic one like mahiṣá-, cases not considered by Przyluski. Also muṇḍa-, as will be shown elsewhere, is doubtless Indo-European. The Sāntālī correspondents are loans.

⁷² Beside the change mh > bh, there exists also the inverse, bh > mh. Mar. mhaṇṇē ‘to speak, say’ is from Māh. bhaṇaï, same meaning. There is no reason to separate these two words because of the phonetic difference (Bloch, Langue Marathe 392 b, for mhaṇṇẽ: ‘Mot isolé, pour lequel la phonétique interdit tout rapprochement avec . . . bhaṇ-’); bh has been assimilated here to the following nasal.— If the common word for ‘to say’ is bhaṇ- in Māhā-rāṣṭrī and mhaṇ- in Marāṭhī, while there is no *mhaṇ- in Māhārāṣṭrī, nor *bhaṇ- in Marāṭhī, if, further, Mar. mhaṇ- means also ‘to recite, read, study’, exactly the meanings of bhaṇ- in the contiguous Gujarātī, and, finally, the two words are phonetically reconcilable,— then the words are identical.

⁷³ I would not posit a direct *melh-, because the guṇated causative probably arose only after the root-final had become simple l (deaspirated); lh (or the preceding hl) seems to have formed position, cf. Pkt. alhāda- with shortened a from Skt. āhlādá- (above §18).

⁷⁴ Marāṭhī has also a bhisaḷṇẽ, bhesaḷṇẽ ‘to mix’ (bhisaḷ, bhesaḷ f. ‘mixture’) beside misaḷṇẽ (misaḷ f.). This bh is analogical: bhisaḷ- arose beside misaḷ- after bheḷṇẽ beside meḷaviṇẽ, all ‘to mix’. (Also the e of bhesaḷ- beside bhisaḷ- is after meḷav- beside miḷav-; but there is no *mesaḷṇẽ; the older form is more resistant against analogical alterations.)

⁷⁵ What does *miṭ mean? Probably the nom. sg. of a root-noun *míṭ, *miśáḥ, of the type of viś- ‘house’, nom. sg. víṭ, gen. viśáḥ, instr. pl. viḍbhíḥ (Whitney §391 d). But there is no noun *miś-, nor could from *míṭ, *miśáḥ have developed a stem *miṭa-, whence, then, Middle Indic *miḍa- and *mila-.

⁷⁶ Stack 295 a and 297 b, Shirt 150 b and a, 158 b.

⁷⁷ Wackernagel §150 a, note, regards iḍ- (with Grassmann) as a variant of RV íṣ- ‘Labe-trunk’, arisen in the bh-cases (*iḍhíḥ < *iẓ-bhíṣ) (but no such form of iḍ- occurs), while Uhlenbeck 23 b (after Fröhde, cf. Wackernagel §146 a, note) connects it with Gk. ἀλδαίνω ‘to strengthen’, implying *ṛd-.