^{page 491 note 1} Beitrāge,II, 782 ff.

^{page 491 note 2} JQLS, NS, III, 2, 1909, 119.

^{page 491 note 3} Where not otherwise epecified, forms are from Welsh Gypsy as recorded by Sampson in DGW.

^{page 491 note 4} Sampson, DGW, IV, 40.

^{page 491 note 5} DGW,I, 34.

^{page 492 note 1} For an analogical formation in the reverse sense see p. 493, n. 4.

^{page 492 note 2} The explanation of this č = Skt. j- was first given by Wackernagel, JGLS,NS, III, 2, 1909, 119, as also for čangandčam (see below). His explanation of ič‘yesterday’ (beside As. Gy. hij) ∠ Pkt.hijjó, Skt. hyáfi) as due to a transference of initial h- to internal -jj- is very doubtful. Initial h- normally disappears in European Gypsy. Moreover a voiced consonant is recorded in Hung. Gy. īž, beside ič, overi'i ‘day before yesterday’ and It. Gy. yidč.In any case there appears to be some tendency to devoice final plosives: so Gk. Gy. panič ‘five’ beside panj),Germ. Hung. Gy. panč, Welsh Gy. panš beside panj. Beside Welsh Gy. etc. sigō, sig ‘quick’ we have Germ. Boh. Gy. sik ‘quickly’, whenoe sikdr-‘to hasten’, sikělo ‘quick, industrious’.

^{page 492 note 3} DGW, IV, 54, s.v. čamyer-. B. Gilliat-Smith, however, in JGLS, 3rd Ser., VI, 1, 1927, 25, points out that Bulgarian Gypsy only has čamh- ‘to chew’ as opposed to čham ‘cheek’ and suggests, without much probability, derivation from Skt. cčmya-: perhaps rather from carvayati in Hi. cābnā, etc, ‘to chew’, with onomatopoeic nasal.

^{page 492 note 4} Misled by Gk. Gy. khan which has lost its final -d (still preserved in the verb kand- ‘to stink’) Jules Bloch, LM, 326, wrongly derived this from Skt. ghrāná-.

^{page 492 note 5} See ND, 453 a 7, s.v. buro, where it is derived from *brddha-, replacing (pári)brdha- (for meaning cf. brmhita- ‘firm, grown’ MBh.). But it may perhaps belong to an immensely large group of words denoting some physical or mental defect and characterized by an initial voiced labial, the vowel u or o, and a dental or retroflex consonant-group, of which some are set out in ND, 479 b 30–50. In either case this buddha- was associated with vuddha- ∠vrddhá-.

^{page 493 note 1} cf. Pa. daḍḍhi-; Ksh. drüḍü f. ‘strength’, WPah. ḍaḍif. ‘a well-built goat’, Lah. dādh ‘strength’, ḍādhā ‘strong’, Pj. ḍāḍh f. ‘firmness of mind’, ḍāḍḍā ‘strong’, Hi. ḍhār(h)as f. ‘courage’, Ass. dári ‘seriously’, Or. dārha ‘strength, stubbornness’, Snh. daḍi‘hardness, adj. strong, firm’.

^{page 493 note 2} The loss of, or failure to record, aspiration is found in other words in Rumanian and Hungarian Gypsy, e.g. Rum. Gy. kos- beside Welsh Gy. khos-, Hung. Gy. tav-d- ‘to flow’ beside Welsh Gy. thab- ‘to flow, run’, Pal. Gy. dau- ∠Skt. dhavati.

^{page 493 note 3} ‘thūtsch-’ recorded by B. Liebieh and ‘tohsch-’ by F. Bischoff seem more accurate or earlier than the doš- of Miklosich, in which d- may represent unvoiced South German d: references in R. von Sowa, Wörterbuch des Dialekte der deutschen Zigeuner, 25, s.v. dóšâdva. There is the further difficulty that the expected j is represented by č or š.

^{page 493 note 4} cf. Pkt. dujjha- ‘fit to be milked’, Pj. dōjh m. ‘milking’ ∠Skt. dōhya-. Present stem dujjh- has been replaced by the analogical dubbh- (after the type labbhaï: laddha-) in Pkt. dubbhaï ‘ is milked’, Si. ḍubhāṇu, Lah. ḍubhaṇ, Mar. dubhnē.

^{page 493 note 5} ND, 220 b 16, s.v. jun ¹ and JOLS, 3rd Ser., VI, l, 1927, 26.

^{page 493 note 6} LM, 322, s.v. garhānē.

^{page 493 note 7} Sampson explains this a- as an exclamatory prefix; but his suggested derivation of khar-, from Skt. katháyati (DGW, IV, 167, s.v.) is impossible. But if ∠gárhati, this root entered Gypsy in both its -rh- and -lh- forms. For it is impossible to separate Pal. Gy. gáli ‘word’ from Pj. gall .‘word’, WPj. gall f. ’affair’, Old Hi. gālhanā ‘to speak’, Si. galhi f. ‘speech’, Mult. gālh f. ‘word’, pl. gālhī ‘abuse’, which despite the ā of Sindhi and Multani appear to be connected with Skt. garhā- ‘abuse’ and may be the origin of late Skt. gālayaḥ f. pl. ‘reviling speech’. If so, it is comparable with the two forms in Sindhi: garahanu and gālhi.

^{page 494 note 1} For tiknō ‘small’ (Miklosich's derivation of which, Beiträge, VIII, 84, from Skt. tiksná-‘hot, pungent’ Sampson, DGW, IV, 366, rejects on the grounds of semantic difficulty) there is no record of kh. The necessary assumption of a MI *tikkhina- is not supported either by Pa. tikhina-, or Pa. Pkt. tikkha-, linka-. But the conjunction in Nepali of tinu (possibly from this tinha-) with sānu ‘small’ in sānu-tinu ‘small, petty’ gives some support to Miklosich's theory. Sampson's derivation of phikō ‘shoulder’(DGW, IV, 294, s.v.) from Skt. prstká-, which would have become *pišt (as Pal. Gy. pišt ‘back’ ∠prstİ-: see ND, 380 a 43, s.v. pith), cannot be accepted. Like Ksh. phyok ‘shoulder’ it is a -kk- extension of Skt. sphİj- ‘buttock’ or sphyá- ‘a fiat wooden instrument’, see ND, 406 b 11, s.v. philo.

^{page 494 note 2} JGLS, 3rd Ser., VI, 1, 1927, 23.

^{page 494 note 3} Paspates, A. G., Études sur les Tchinghianés ou Bohémiens de l'Empire Ottoman, 29, kotch m.Google Scholar

^{page 494 note 4} ND, 106 a 35, s.v. kokh.

^{page 494 note 5} ND, 427 b 9, s.v. bahu ².

^{page 494 note 6} Chatterji, S. K., The origin and development of the Bengali language, 345.Google Scholar

^{page 494 note 7} ND, 138 b 13, s.v. gahū.

^{page 494 note 8} ND, 146 b 24, s.v. guhu.

^{page 494 note 9} ND, 310 a 41, s.v. däro.

For unexplained interchange of th-r and h-d eompare Skt. kuthāra- ‘axe’, Pa. kuthārī-, (possibly in Snh. keneri‘axe’, D. E. Hettiaratchi, The indeclinables in Sinhalese, 6) beside Pkt. kuhāda-, Si. kuhāro, etc.

^{page 495 note 1} LM, 305, s.v. kaḍhṇē.

^{page 495 note 2} Pischel, Beitrāge zur Kenntnis der deutschen Zigetiner, 37, rightly rejects von Sowa's derivation from the ghar- in Skt. gharmá- ‘hot’, etc., which occurs as a verb—jigharti, ghrṇōti—only in the Dhātupātha and does not survive elsewhere in Indo-Aryan. Sainpson's derivation from kēr- ‘to do’ is unacceptable on the score of meaning.

^{page 495 note 3} DGW, IV, 172.

^{page 495 note 4} ND, 103 a 17, s.v. kuhun ².

^{page 495 note 5} Sampson, DGW, IV, 180, probably rightly connects xev f. ‘hole ‘ with Skt. khan- (cf. khēya- ‘to be dug, n. ditch’) rather than gūhā- ’ cave’, though the gender makes difficulty. H. Berger's theory (Indo-Iranian Journ., III, I, 1959, 24) that it is a loan from Burushaski gam ‘hole’ (earlier *kham) is not very plausible.

^{page 495 note 6} ND, 427 b 16, s.v. bahut.

^{page 495 note 7} ND, 438 b 13, s.v. bie

^{page 495 note 8} Transactions of the Philogical Society, 1937, 11.

^{page 495 note 9} ND, 389 b 36, s.v. paicānnu. Miklosich, Mund., VIII, 53, and Bloch, LM, 370, s.v. paij, wrongly ∠pratijnā-.

^{page 495 note 10} ND, 459 b 37, s.v. baini.

^{page 495 note 11} JGLS, NS, II, 4, 1909, 326.

^{page 496 note 1} ib., 328.

^{page 496 note 2} The position of Romani in Indo-Aryan, 20.

^{page 496 note 3} ND, 611 b 35, s.v. sulco.

^{page 496 note 4} ND, 624 a 39, s.v. sohornu.

^{page 496 note 5} ND, 343 a 6, s.v. nikālnu.

^{page 497 note 1} ND, 620 b 25, s.v. sep.

^{page 497 note 2} DOW, I, 14.

^{page 497 note 3} Connexion with either Skt. raks- or lake- (as set out by Sampson, DGW, IV, 204) is highly improbable; but it is possible that Hung. Gy. alakh-, ’to find’ was due to contamination between (a)rakh-‘to keep’ and *lath- or lačh-‘to find’.