^{page 745 note 1} See Ching-ju, Wang,“Notes on Chinese and Tibetan Transcriptions of the Shishiah (Tangutan) Language” (in Chinese), in the Bulletin of the National Research Institute of History and Philology, vol. ii, part 2 (Peiping, 1930), pp. 171–184Google Scholar, and the same author's Hsi-Hsia Studies, parts 1 and 2 (in Chinese), in the Monographs of the above Institute, series A, No. 8 (Peiping, 1932)Google Scholar and No. 11 (Peiping, 1933). For brevity's sake these will be quoted in the following pages as Bulletin and Studies (1 and 2).

^{page 745 note 2} T'oung Pao, vol. xvii (1916), pp. 1–126Google Scholar.

^{page 745 note 3} See inter alia, Schmidt, Pater W., Die Sprachfamilien und Sprachenkreise der Erde, Heidelberg, 1926, p. 133Google Scholar; Linguistic Survey of India, vol. i, part 1 (1927), pp. 79–80Google Scholar; vol. i, part 2 (1928), p. 22; Kieckers, , Die Sprachstämme der Erde, Heidelberg, 1931, pp. 104–5 (where it is erroneously placed in the Siamese-Chinese Family)Google Scholar.

^{page 745 note 4} Unfortunately for early studies of thisspeech the wrong order of reading was followed. Consequently Laufer, accepting the method employed by Ivanov, (“ZurKenntniss derHsi-hsia Sprache”, Bulletin de I'Academie Imperiale des Sciences de St.-Petersbourg, No. 18, 1909, pp. 1221–1233Google Scholar, with one plate), transcribed as tsu-ni, as ko-ni, etc. This order, as first pointed out by Zach, E. von (Orientalische Literaturzeitung, Jhrg. 30, Nr. 1 (01, 1927), Sp. 4—5)Google Scholar, should be reversed, a fact which Ching-ju, Wang, apparentlyindependently, also noticed (Bulletin, p. 178)Google Scholar. As we shall see in the sequel, prefixes then begin to come to light where previously they were unsuspected (cf. Laufer, op. cit., p. 103), and, together with the Tibetan alphabetic transliterations, put the language in rather a different light from what was at first believed.

^{page 746 note 1} See below, in extr.

^{page 746 note 2} See especially Ching-ju, Wang, Bulletin, p. 183Google Scholar; Studies, 2, pp. 298–303.

^{page 746 note 3} See especially Devéria, , “L'Écriture du Royaume de Si-Hia ou Tangout” (Extrait des Mémoires présentés à l'Académie des Inscriptions et Belles-Lettres, Ire série, t. xi, Ire partie, Paris, 1898), p. 28Google Scholar, and Morisse, , “Contribution à 1'Étude de l'Écriture et de la Langue Si-Hia” (Mémoires, etc. (as above), Ire série, t. xi, 2me partie, Paris, 1904), pp. 37–8Google Scholar.

^{page 747 note 1} Op. cit., P. 103.

^{page 748 note 1} Ching-ju, Wang, Studies, 2, pp. xxviii and 301Google Scholar.

^{page 748 note 2} See Outlines of Tibeto-Burman Linguistic Morphology (quoted hereinafter as Morph.).

^{page 748 note 3} Morph., pp. 49–56 (§ 47).

^{page 748 note 4} Jya-rung words in the present article are from the author's own collectanea shortly to be published.

^{page 749 note 1} See Morph., pp. 26–30.

^{page 749 note 2} Their articulation has, in fact, so persisted that even in Central Tibet there are still such well-known survivals as čuk-či (= ču-gči) eleven, čup-dṳn (= ču-bdṳn) fourteen, zip=ču (= zi-bču) forty, and others.

^{page 749 note 8} See Ching-ju, Wang, Studies, 2, p. 300Google Scholar.

^{page 750 note 1} See “On the Tibetan Transcriptions of Si-Hia Words” by the present writer in JRAS., 1931, pp. 47–52.

^{page 750 note 2} We shall see below that these types of prefix interchange are exactly those found in Tibetan dialects, a consideration of which helps in great degree to clear up many puzzling features of Si-Hia.

^{page 750 note 3} Of this fact Ching-ju, Wang (Studies, 2, pp. 298–9)Google Scholar does not seem to be aware. The least familiarity with Tibetan dialects shows one the almost unlimited interchange among its prefixed elements. In earlier times the no less remarkable, and only slightly more restricted, scope of these interchanges is indeed evident without any knowledge of the more modern dialects.

^{page 750 note 4} Morph., p. 31.

^{page 751 note 1} Morph., p. 41.

^{page 751 note 2} For final ṅ and the nasalization (˜), see later below.

^{page 751 note 3} In the Chinese elements representing prefixes, it is moderately certain that the consonant only should be read and the following vowel discarded. I shall consequently follow this course in transcribing them in the ensuing pages. In the case of , and , it might seem that a palatal (ϓ-) was intended rather than γ-, were it not for its improbability with the Tibetan prefixes involved. There was naturally a real difficulty in reproducing the required sounds in Chinese, with the result that approximations only were achieved. I therefore transcribe in these cases, not so much the sounds which the Chinese elements actually contained (y, i) as those which they were intended to represent (γ-). For the material in the tables I am indebted to Nevsky's, Brief Manual of the Si-Hia Characters with Tibetan Transcriptions (= Research Review of the Osaka Asiatic Society, No. 4, 03, 1926)Google Scholar, a work of the utmost importance for any study of Si-Hia. As to the final ⁿ, and the nasalized vowels, see Nevsky, op. cit., pp. xxv–xxvi, and later in the present article.

^{page 756 note 1} In the last two sets of correspondencesthis is not yet certain. It appears, however, that the values ṅ- and ṙ- suggested for the Chinese prefixes, and ṙ- for that of Tibetan , are sufficiently probable to warrant the Tibetan entries being tentatively placed under column (a).

^{page 757 note 1} Ching-ju, Wang, Bulletin, pp. 178 and 184Google Scholar. Compare also Studies, 2, p. 303.

^{page 757 note 2} Ching-ju, v. Wang, Studies, 2, p. xxviiiGoogle Scholar. See also Ivanov, op. cit., pp. 1227–1231 (whose Chinese entries must be read from right to left), where , and in the Chinese transcriptions represent γ (< or ) as shown by the previous tables.

^{page 758 note 1} See Rockhill, , Land of the Lamas, p. 363Google Scholar. Whether or not this prefix (r-) tends to exhibit a velar value (ṙ-) we are not informed. Under similar circumstances in western Tibet it would do so.

^{page 759 note 1} The velar pronunciation of this prefix in western Tibet raises the question of the possibility of a similar manifestation in Si-Hia. Its interplay with γ- at least suggests it, though its presence cannot be affirmed as yet.

^{page 759 note 2} v. Rockhill, loc. cit.

^{page 759 note 3} ed. Francke, , in Asia Major, vol. i, pp. 243–346Google Scholar; vol. iii, pp. 321–339; vol. iv, pp. 161–239 and 481–540; vol. v, pp. 1–40 (Leipzig, 1924–8).

^{page 760 note 1} Nevsky, op cit., pp. xxiv–xxv.

^{page 760 note 2} Ching-ju, Wang, Bulletin, p. 183Google Scholar.

^{page 761 note 1} Even in Tibetan where prefixed presumably once carried its original sonant sound (g-) this was true, kṣa () is there transcribed kša. gša could never have reproduced the required sound.

^{page 761 note 2} For a further consideration of this point see below.

^{page 762 note 1} Ching-ju, Wang, Studies, 2, pp. xxviii and 301–3Google Scholar.

^{page 762 note 2} T'oung Pao, vol. xxviii (1932), p. 490Google Scholar. See also Nevsky, op. cit. supra, pp. xxv–xxvi.

^{page 763 note 1} Cf. Ching-ju, Wang, Studies, 2, p. xxviiGoogle Scholar.

^{page 763 note 2} Cf. Wang Ching-ju, op. cit., pp. 291–3.

^{page 764 note 1} JRAS., 1931, pp. 47–52. Three of the twenty-four equivalences there proposed seem to occupy a special position. They are: Si-Hia dmī, dmi “eye”, Jya-rung “te-mńok”; Si-Hia , dmu, dmi, “fire”, Jya-rung “te-mi, te-me”; Si-Hia dmeṅ “woman”, Jya-rung “te-mi, te-me”. These correspondences I believe to be basically true, although in Jya-rung the prefix (te-) is now pronominal as I have satisfied myself by working in Darjeeling with a speaker from the Jya-rung states. This could never have been the case with Si-Hia d-. However, I believe the Jya-rung prefix te- overlies in an undetermined number of cases(which early Tibetan forms may help us to unearth) a non-pronominal d- (5) which has been attracted only later into pronominal te- in line with the general usage of this dialect with substantives. This earlier non-pronominal Jya-rung d-may be represented in some of the forms spelt this way by Laufer, (T'oung Poo, vol. xv (1914), p. 107Google Scholar, n. 1), among which dmye “eye” occurs. This would then belong directly with Tibetan dmyig (Gzer Myig (as above quoted), folio 25a, 1. 6, folio 256, 1. 1, etc.), and the equivalence with Si-Hia would be exact. In one case, however, the proposed equation will have to be withdrawn, i.e. in that of Si-Hia , dgū “head”, Jya-rung “ta-ku, ta-ko”. Here ta- (tâ-) = tâ- in which â- is the non-pronominal Tibetan 'a-, and tě-, from its position before it, is obviously late and pronominal (< T. de). Si-Hia dgū, on the other hand, contains no pronominal prefix.

^{page 765 note 1} Examples of the use of the majority of these prefixes may be found on a preceding page. Of those not there represented the following will serve as illustrations: Jya-rung kě-něs, T. gńis “two”, Jy. kě=sŭm, T. gsum “three”, Jy. ûšât, ûšîât, T. bšad (perf. of acad-pa) “say, speak”, Jy. ŭtŭn, ŭtŭm, T. bduṅ-ba “beat”, Jy. wô-ryât, T. brgyad “eight”, Jy. kě-sčyît, T. dge-skyid “happy”, Jy. â-spó, T. spo “summit”.

^{page 765 note 2} I have found this pronunciation only once in Jya-rung, viz. in yě=šči “alone”, T. gčig.

^{page 766 note 1} JRAS., 1931, p. 51.

^{page 766 note 2} Ching-ju, Wang, Studies, 2, p. 297Google Scholar.

^{page 767 note 1} See Hänisch in Hediu's, SvenSouthern Tibet, vol. ix, p. 72Google Scholar. (For fuller quotation see next footnote.)

^{page 767 note 2} The most complete lists of these from Chinese sources are given by Rookhill, W. W., Land of the Lamas, pp. 344–354Google Scholar, and Hänisch, E., “Das Goldstromland im Chinesisch-Tibetischen Grenzgebiete, nach dem grossen Kriegswerk vom Jahre 1781 dargestellt,” in Hedin's, SvenSouthern Tibet, vol. ix, p. 72Google Scholar. See also Rosthorn, A. von, ZDMG., Bd. 51 (1897), pp. 524–6Google Scholar. These lists all differ from each other at various points, as does again one recently given the writer by a man from the principality of Jyâ-k'â. Nevertheless they all contain a large proportion of names in common— doubtless the most stable and powerful tribes. Cf. Hanisch, loc. cit.

^{page 767 note 3} One or two instances will suffice to show the need for further research here. In my list I have a form Taks-dê, also called Brag-sdê by the same speaker. This is given by Hanisch as Burakdi, noted as corresponding to Chinese Badi, which Rockhill writes Pati. Hänisch proposes the Tibetan reading Brag-di, Brag-adre () “Felsen-Dämon”. I gather, however, from my own informant that the correct original is Brag-sde, which is borne out by his further statement that the settlementconsists of small groups of houses, in threes and fours, situated in rocky surroundings. Then, again, I have a name cos-čyap, pronounced also by the same speaker as T'os-kyap. This is von Rosthorn's Chossftahiapu, which he and Rockhill believe to represent K'ro-slcyab. Hänisch has Cosgiyab, Rockhill a second form Tru-jyab, Tro-jyab, a third Cho-ssu-chia, and afourth ch'o-ch'i chia, while Baber (RGS., Suppl. Papers, vol. i, p. 94) gives Tchro-shiop. The correct reading here was given me as Cos-shyabs, a place where is situated a monastery known as T'u-jê cěm-pô with accommodation for some forty lamas. I have been through these lists many times, and am indebtedfor valuable suggestions to Lama Lobzang Mingyur of Darjeeling. The greater number of names, however, remain obscure.

^{page 768 note 1} Ching-ju, Wang, Studies, 2, pp. xxviii and 277Google Scholar.

^{page 768 note 2} Wang Ching-ju, op. cit., pp. xxviii, 276–7, 297.

^{page 768 note 3} JRAS., 1931, p. 51.

^{page 769 note 1} T'oung Pao, vol. xvii (1916), p. 108Google Scholar. “The missing links between Lo-lo and Mo-so on the one hand, and Si-Hia on the other, must have existed in the territory of Sze-ch'uan (or may still survive there).”

^{page 769 note 2} “Binoms of the type in the Tangut-Chinese Dictionary,” Comptes Rendus de I'Académie des Sciences de l'Union des Républiques Soviitiques Socialistes, Leningrad, 1929 (B, No. 8), pp. 145–8Google Scholar.