THE LOP-NOR REGION ON THE WU-TSCHANG-FU MAP. 283
while to is a Mongolian suffix indicating the adjectival form. If this interpretation is correct, the locality in question must lie somewhere near the mountains, for there are no steep patches of arable land in the lowlands. In fact, the name is placed quite close to the »pass» of Nukitu. Nuki too would be the name of a person, a chief, that being the only way by which we can account for the occurrence of the name both in the mountains and in the lowlands. The relation is one to which we find a parallel in Bedaulet's time, when in this same region there existed the names Jakub Baj-kuduk, Jakub Baj-karaul, Jakub Baj-kurghan, together with several similar place-names. Himly seems to doubt whether to is the Mongolian suffix.* Is it possible that to-kia is the same word that occurs in Takia-tagh, the little mountain-
ridge between Kurghan-bulak and Dschahan-saj ? This word I heard pronounced
as Tekia, and I was given to understand that it means a »cushion», because of the resemblance which the mountain bears to such an article. Generally speaking, geographical names beginning with N are extremely few in East Turkestan, as a glance at my alphabetical list in Peterm. Milted., Ergänzhft 131, will show at once. Nuki seems to be a person's name, probably a contracted form of two other names, such as Numet and Baki, both still in use in that same locality. For example, Numet Bek was bek of Jurt-tschapghan in 1900 and 1901, and at the same time Nias Baki Bek was chief of Kum-tschapghan. Hence it is easy to conceive the combination Numet Baki; though Numet itself is also a contracted form, and stands for Nur Mohamet. In a precisely similar way, it is easy to see how Nuki may be a contraction for Numet-Baki. Dake, deki, duke is a universally current genitive termination occurring in numerous names along the Tarim and its delta, e. g. Niasdake-uj = »the house of Nias», or as here Nukidake-tarim = »Nuke's river» or »arable land». The insertion of the a, converting the word into Nuki-dakia-tarim may be only a provincialism; just as in the Lop country an a is usually added to the title Bek, e. g. Islam Beka, Numet Beka. At all events, if my interpretation is correct, the original Nukidaketarim has been far less transmogrified in the Chinese transcription than many other words. For my own part, I do not believe that tarim does here mean »arable land»; it means »river», and when the materials for the map were collected an arm issued from the Tarim and flowed to the small southern lakes, an arm into which the entire river gradually transferred itself and so formed the Kara-koschun. In Nukitu-sekin and Nukitu-daban we have on the other hand the Mongolian suffix to (Turki = li, lik). According to the analogy already alluded to, the forms should have been Nukitukia-sekin and Nukitukia-daban; although the more familiar forms would be Nuki-bulak and Nuki-davan.
Turning west from Nukitu-daban we come to Nukitu-setsin (= sekin, which according to Himly means »spring»). On the Wu-tschang map we find this name at the south-east corner of the Lop-nor, close to the lake. Still farther west we find Nukitu-schan-k'ou; which, according to Himly, means the »mouth» or »valley-cpening» of the Nukitu mountain, or mountain-range. These two names, which plainly have to do with mountainous districts, occur nevertheless amongst the lakes, that is
* Nord-Tibet und Lop-nur Gebiet, by Wegener and Himly, in Zeitschrift d. Ges. f. Erdk. zu Berlin, vol. XXVIII (1893), p. 228.
** See for the position of this mountain Kosloff's account of his journey, p. loo.
IIIFカラー高解像度
