The World of the Huns. Chapter IX. Language
O. Maenchen-Helfen

6. Turkish names

Althias  —  Atakam  —  Basich  —  Berichos  —  Dengizich  —  Ellac  —  Elmingeir  —  Elmintzur  —  Emnetzur  —  Erekan  —  Eskam  —  Iliger  —  Kutilzis  —  Mundzucus  —  Sandil, Sandilchos  —  Zolbon

In his account of the war in Lazica in 556, [222] Agathias [223] mentions among the Byzantine officers of barbarian origin a Hun by the name of ; he was lochagos, commander of a lochos, a regiment. Agathias also mentions the name and the nationality of Elmingeiros' superior: He was the taxiarchos Dabragezas of the people of the Antes. In order to overcome the difficulties of transmitting orders, a formidable task in mercenary armies of as many different nationalities as the armies of Justinian and his successors, barbarians of the same regions were kept together in the same units. Dabragezas [224] must have come from those Antes who, according to Procopius, together with Huns and Sclaveni, "lived across the Danube or not far from it." [225] Elmingeiros was probably from the same region. The battle in which he distinguished himself took place in the spring of 556.

In the summer of the same year Justin, commander of the army in Phasis, sent one of the taxiarchoi, a Hun by the name of , with


two thousand horsemen to occupy the fortress Rhodopolis. [226] In the index of his edition of Agathias, Niebuhr listed Elminzur with the note, fortasse idem cum praecedente, i.e., Elmingeiro." [227] Stein identified Elmingeir and Elminzur. [228] It would be a strange coincidence indeed if in the same army and in the same months, there had been two Hun officers bearing names as similar as Elmingeir and Elminzur.

It is not necessary to know the exact foreign sounds represented by the Greek letters, [229] nor what the names mean, to recognize that the first is compounded of elmin and geir, the second of elmin and zur. If Elmingeir and Elminzur were actually two names of the same man, the change from -geir to -zur could correspond to his promotion from lochagos to taxiarchos, or, to use the Latin terms, from tribunus to dux. [230] This would support our assumption that čur means "captain, leader."

There are three more Hunnish names ending in -zur:

1. After the collapse of Attila's kingdom, his kinsmen Emnetzur and Ultzindur occupied Oescus, Utum, and Almus on the right bank of the Danube. [231] On the analogy with Elminzur, Emnetzur must be Emne-tzur.

2. Another name of this type is Ultzinzures, . [232] Together with other Hunnic tribes they followed Dengizich in the second war with the Goths.

3. Priscus' , and  appear in Jordanes as Alpidzuri, Alcildzuri, Itimari, Tuncarsi, and Boisci. [233] The explanation of the difference between Priscus and Jordanes was found by Krasheninnikov: [234] The archetype of the Jordanes manuscripts had alpidzuros, with the emendation alcildzuros written over it, which leads to *alpildzuros. Only this form is compatible with the name in Priscus which, therefore, must be emended to read .

In the Chinese annals, the titles of tribal leaders are sometimes used for the tribes themselves. In Han times the Chinese spoke of the Sai wang,


the "Saka kings," under the T'ang of the Hu-lu-wu chüeh, She-she-t'i tun, and Shu-ni-shi ch'u-pan. [235] This was not a misunderstanding on the part of the Chinese, as some scholars thought. [236] To the Tibetans the kingdom of the second dynasty of the northern Turks was known as Bug-čor = Mo ch'o. [237] Did they make the same mistake as the Chinese? Should we assume that Constantine Porphyrogenitus was also misinformed when he spoke of the *küärčičur? And before him Priscus about the *alpildzuri, or, as we now may say, the *alpilčur? This is most unlikely. Even today Kirghiz tribes, subtribes, and clans exist which call themselves čoro and xoro: Qara-coro (tribe), Čoro, Zol-čoro (subtribes), Boro-čoro, Ono-čoro (clans). The Kazakh have the clans Zhan-čura, Bai-čura, and Qara-čura. [238]

In an epitaph from Uibat in Tuva the deceased glories that he exerted himself for the people ilčur. [239] Whatever the origin of čur may be, in the inscription from Uibat ilčur is as Turkish as il qan and il baši. Hunnish *Alpilčur cannot be anything else but alp-il-čur, "hero-people-čur." [240]

The thesis that the Huns spoke a Turkish language has a long history behind it. Its earlier phase is no longer of interest. The later is still with us. Taking the identity of the Huns and the Hsiung-nu for granted, some scholars have no doubt and need no proof that the Huns spoke the same language as the "eastern" Huns, which they take to be Turkish. By the same reasoning the Norman conquerors of England should have spoken Old Norse.

That the Huns included Turkish-speaking tribes can be regarded as established only if a number of personal and tribal names of the Huns are undoubtedly as Turkish as orfèvre is French, goldsmith English, and Goldschmied German. One such name is *alp-il-čur.

The formal analysis of Turkish-sounding Hunnic names requires utmost caution. If English were as unknown as the language of the Huns, one could conjecture that fe- in female is a prefix and -dict in maledict a suffix to the root male.

-gir, like čur, occurs in both a Hunnic personal name (Elmingir) and the name of a tribe of the Pontic Huns, named twice in Jordanes, Getica 37. In the first passage, page 6311, all codices, except the inferior ones of the secundus ordo, have altziagiri or altziagri. In the next line, page


6312, the forms are: primus ordo altziagiri (H), ultziagiri, uultziagiri, autziagiri; secundos ordo alugiagiri, aulziagri; tertius ordo ultziagri, altziagri (Y).

Mommsen put Altziagiri in both passages in his text. Closs in his edition of the Getica, page 29, preferred Ultziagiri. He was right, in my opinion. In the second passage the name began with u. Three codices still have it; au obviously was u with superscribed a; the forms in H and Y were adapted to altziagiri in the first passage. We have, thus, altziagiri and ultziagiri. Although Altziagiri has no parallel in Hun tribal names, Ultziagiri can be compared with Ultzinzures. When we think of the personal name Uldin, and in particular of Elming(e)ir and Elminzur, the conclusion that  is but a slightly blundered , *Ultingir, seems inescapable. Gir, like čur, must be a rank or title. It seems to occur in , a Bulgarian genos, [241] and Yazghyr and Ürägir, two Oghuz tribes named by Kashghari. [242]

Five Hun names end in ic, and . Standard pronunciation treated c as aspirant in Byzantine Greek until the ninth century. [243] In the Greek transcription of Germanic names c corresponds to c in Latin forms. The same is true for Hunnic names.  and  appear in Latin sources as Denzic and Dintzic. Priscus wrote , Jordanes Hernac. There is no evidence that in fifth-century Greek transcription of foreign names c can reflect g or g. [244] Therefore, etymologies based on the equation ic = ig, ig or ac = ag, ag are inadmissible.

The name of an Utigur prince about 550-560 occurs in two forms. Agathias and Menander call him ; in Procopius his name is . [245] Sandilchos is Sandilk, Sandil-k.

is [246] the name of a Hun leader in 395. It could be Kurs-ik or Kur-sik. , the name of a barbarian officer in the Byzantine army about 578, [247] seems to indicate that Kursik is Kurs-ik.


was the qaghan of the western Turks about 580, [248] Tuldila a Hun leader in Majorian's army in 458. [249] -ila is evidently the same as -ila in Attila and Rugila, namely the Germanic diminutive suffix. It corresponds to the Turkish diminutive suffix +°q, +°k. [250] *Tuldiq would be in Turkish what Tuldila is in Germanic: "little Tuld." This *tuld can be compared with Ultinzur, Uldin, and Uldach, names which seem to be compounded of uld or ult and in/ach = *ïn/aq.

To maintain that all Hun and Turkish names ending in ic are diminutives would probably be wrong, but some of them apparently are. Take, for instance, . [251] Basich and Kursich are named together. If Kursich is Kurs-ich, *Kurs-iq, then Basich is probably Bas-ich, *Bas-iq, which can hardly be anything else but bašiq, "little captain."

It is almost generally agreed that  contains Turkish däŋiz. Dengizich cannot be Dengir-siq [252] because if it were, Priscus would have written , [253] nor can it be Dengis-sig (see above).  is a perfectly normal transcription of *däŋiz-iq, "little lake."

Another formant in Hun names is +l.  the name of a barbarian exarch, [254] stands in the same relation to [255] as  to . It evidently is Apsik-al.

The number of Hun names which are certainly or most probably Turkish is small. But in view of the wild speculations and irresponsible etymologies still being expounded, to lay a narrow but firm basis for studying all the names seems preferable to dreamily wandering through dictionaries. Some of the names in the following list have been etymologized before; instead of repeating the arguments brought forward, in particular the many parallels, I refer to Moravcsik, BT 2, where the literature is carefully listed.


Leader of Hun auxiliaries in the Byzantine army about 530. [256]  Altï, "six." In his study of names formed by numerals, Rásonyi (1961, 55-58) listed the Kazakh patronymic Altyev and a large number of personal and clan names having altï as the first element: Altybai, Altyortak, Altyate, and so forth. Compare also Alty bars (Sauvaget 38).


A Hun of noble birth, about 433. [257] The name could be compared with Iranian . [258] In an Iranian dialect spoken in South Russia the change from -rt- to -t- can be followed in the inscriptions: [259] cannot be separated from . [260] Some names beginning with ata are Iranian, for example,  and  [261] (*maza, "greatness") [262] or [263]. There exist dozens of Iranian names ending in kam, "wish," from [264] to Xudkām and Šadkām. [265] However, Eskam, another name ending in kam, has no similarity to any Iranian name and a most plausible Turkish etymology. Therefore, I accept Vámbery's etymology: ata, "father," and qam, "shaman." [266] Similar Turkish names, for example, Atabag, are [267] fairly common. [268]


Hun leader about 395. Basich is probably Bašīq.


Lord of many villages, [269] Berik, "strong." [270] The king under whom the Goths are said to have left Scandinavia had a similar name: Berig, Berg, Berigh, Berich, Berice, Berige; see Getica 2594. Although the Goths took over Hunnic names, they certainly did not rename one of their half-mythical rulers. Berig is probably *Bairika, the hypocoristic form of a name beginning with Bere-, like Beremod. [271]


A son of Attila. *Däŋiziq, "little lake." [272] Dengizich, as Priscus heard the name pronounced at Attila's court, [273] is the only authentic form. Denzic, [274] Dintzic, [275] apparently renders the Germanic pronunciation *Denitsik, with the frequent dropping of g, is assimilated to names like . [276]

The fact that täŋiz, däŋiz is not attested before the eleventh century is of little importance. [277] It occurs in all Turkish languages; besides, there is no language known from which the Turks could have borrowed the word. Mongol Tängiz is a Turkish loanword.



Attila's oldest son. [278] The scribes who made the excerpts from Priscus left the name out. It should be in EL 13036 and 18328. Jordanes' Ellac presupposes  in Priscus; compare  = Hernac. Ellac seems to be älik (ilik), "ruler, king." [279] To be sure, in Priscus' transcriptions of Germanic personal and Latin place names alpha always renders a, never i. [280] But a in the second syllable occurs also in Armenian, alphilaq > alp ilig. [281] Apparently Ellac was not the name but the title of the prince who was governor of the Acatziri. Latin and Greek authors often mistook foreign titles for names. [282]

*Elmingir. Tunguz elmin, "young horse," also the name of a Manchu tribe, [283] is probably a coincidental homophone; it would be the only Tun-


gus word in the language of the Huns. El seems to be el, al, il, [284] "realm"; -min- can be compared with -min in Bumin, Chinese T'u-men and Ch'i-men. [285]

*Elminčur, see p. 401.


*Emnečur, see p. 402.

Priscus mentions Attila's wife, Ellac's mother, in two passages. In the first, EL 13922, all codices have ; in the second, EL 1467, M and P have  B and E , C has . The copyists repeatedly dropped n at the end of personal names, but they never added it where it did not belong. [286] The name ended in -an. To choose between  and  would be impossible were it not for the Germanic names of Attila's wife: Herche, Helche, Hrekja, and Erka. [287] They prove that Priscus wrote . Bang's etymology is convincing:  is *arï(g)-qan, "the pure princess." [288] Aruvkhan (aruv, "pure") is a Qaraqalpak girl's name. [289, 290]


Eskam's daughter was one of Attila's many wives. [291] Eskam is most probably *as qam, as, "friend, companion", and qam, "shaman." [292] The



non-Tokharain name Yarkam in a Tokharian document [293] might be a hybrid name with the same meaning (Persian yar, "friend").


A Sabir, about 555. [294] Probably *Ilig-är. [295]


A Sabir, about 555. [296] When one thinks of the many Turkish names with qut, "majesty," it seams very likely that the name was *qut-il-či or *qut-elči.



The name of Attila's father occurs as  in Priscus, Mundzizcoabl in Jordanes, and gen in Theophanes. [297] The last one is so corrupt that it can be disregarded. [298] Cassiodorus undoubtedly wrote *Mundiucus, which Jordanes changed to Mundzucus as he changed Scandia to Scandza [299] and Burgundiones to Burgunzones. [300] In vulgar Latin d before i and e, followed by a vowel, became dz. [301] Jordanes pronounced Mundiucus as Mundzucus, and consequently wrote Mundzucus. But this does not necessarily prove that the Hunnic name was *Mundiuk. If Priscus should have heard a Pannonian Roman or a Latin-speaking Goth say "Mundzuk," he still could have written  on the assumption that his informant mispronounced the name in the same way he said dzaconus for diaconus. [302]



Nemeth and Rásonyi [303] take , Mundzucus, to be the transcription of Turkish munuq, bunčuq, Perle, Glasperle, Kügelchen oder Perlen, die man am Halse des Pferdes befestigt (Radlov). "Pearl" would indeed be an appropriate name for a prince. [304] I prefer Vämbery's etymology which took the name to mean Fahne, eigentlich Fahnenknauf, Koralle, die apfelartige Rundung, in welcher der Rossschweif, die primitive Fahne des Türkenvolkes befestigt wurde, und nach welcher das ganze militärische Abzeichen später den Namen erhielt. [305]

In his review of Moravcsik's Byzantinoturcica, Ligeti doubted the correctness of Németh's etymology. [306] At my request to state his reasons, he was so kind to write to me: L'exposé des raisons de ma réserve vis-à-vis de cette étymologie depasserait les cadres de cette lettre. Je me contenterai de vous indiquer qu'il m'est impossible de concilier cette etymologie avec ce que nous savons de l'histoire des langues turques. Ainsi, le j est caractéristique des langues oghouz, en face du č offert par les autres langues turques. En même temps l'initiale m caracterise les langues offrant un č, en face de l'initiale b qu'on attend dans les langues oghouz. [307]

To these objections of the eminent Hungarian scholar one could perhaps answer that to a Greek, in whose language  and č did not occur, the two must have sounded very much alike. More important is the known fact that b interchanges with m within a number of Turkish languages: bän in Osmanli in the eastern and man in the western Crimea, [308]mindi and bindi in Nogai; börü in the southern and mörü in the northern group of Altai Turkish. [309] One cannot even say that the Oghuz languages have the initial b, for although Osmanli, its Rumelian dialects, and Azerbaijan Turkish have it, the East Anatolian dialects have m. [310] Except the Auslaut, in the Osmanli dialect of Kars our word has the allegedly impossible form munuc. [311]

"Flag" as title or rank of the flagbearer occurs in many languages. Ensign, for instance, is both the insignium and the one who bears it: "hee is call's aunchient Pistoll," Henry V (aunchient, corrupt for ensign). It


is the same in the East. Tug, [312] "standard with a horse or yak tail," occurs by itself or with a suffix in early Turkish and Uyghur names: TugAšuq, Tuglug, "he who was the tug," Tug, "Tug bearer." [313]Munuq probably means the same. Qïzïl Mončuq, the name of a Mongol commander in Afghanistan about 1223 [314] means "Red Flag" rather than "Red Pearl."

In the eighth century the leaders of the ten arrows (tribes) of the Türgäš bore the standards. [315] The cauda equi was the signum militare of the proto-Bulgars. [316] It may have been that of the Huns, too.

The Germanic etymology of Mundzucus [317] is to be rejected. It is not only phonetically unsound. About 370, when Mundzucus was born, no Hun could have been given a Germanic name. [318]


Ruler of the Utigur, about 555. [319] Sandil cannot be separated from the Mamluk name Sandal, "boat." [320]

Commander of Hun auxiliaries in the Byzantine army, 491 A.D. [321]  Zolbon is "the star of the shepherd," the planet Venus, colban, colbon, solbon, and so forth. [322] Colpan is a Mamluk name. [323]

[Previous] [Next]
[Back to Index]

222. For the date, see Stein 1959, 2, 813.

223. III, 21, ed. Bonn, 186.

224. The name may contain Slavic dobry, "good."

225. Procopius V, 27, 2.

226. Agathias LV, 15, ed. Bonn, 236.

227. Ed. Bonn, 403.

228. Stein 1959, 2, 815.

229. Agathias took great care in transcribing foreign names as faithfully as the Greek alphabet permitted. His  (III, 2, 17) is closer to the Persian word (Christensen 1944, 21, n. 3) than Menander's , his  preferable to Procopius' , cf. Schönfeld 1911, 140.

230. On = dux, see Stein 1959, 2, 814-815.

231. Getica 266.

232. Ibid., 12822; Agathias V, 11, ed. Bonn, 300; Moravcsik BT 2, 230.

233. Getica 9010-11. To v. 11, add Alpizuros, Lizuros which Jiménez de Rada read in his copy of the Getica; cf. Alarcos 1935, 18.

234. Krasheninnikov 1915, 42, n. 1.

235. T'ang shu 225b, 8b.

236. The coexistence of Sai and Sai wang puzzled G. Haloun, ZDMG 1927, 252.

237. See footnote 173.

238. Abramzon 1946, 125, 127, 128, and 1960, 5, 31, 42, 45, 108, 111, 115, 126; Vinnikov 1956, 148, figs. 3, 6.

239. Radlov 340; Orkun 1941, 3, 144; Malov 1952, 62-63.

240. Tucked in an article on the Scythian name of the Maeotis, this etymology was suggested by J. Markwart as early as 1910 (Keleti Szemle 11, 1910, 13) and 1932, 108.

241. Beshevliev 1959, 289.

242. Brockelmann 1928, 244, 251; for Üräkir read Ürägir (Pelliot 1950, 190). The origin and the meaning of -gir in Tungus tribal names is obscure, cf. Kotwicz 1939, 185; Pelliot 1950, 229; Menges 1951a, 87; N. Poppe, UJb 24, 1952, 75. Whether it has anything to do with -gir in the adduced names is doubtful.

243. Sturtevant 1940, 85.

244. Of the five cases adduced by Moravcsik, BT 2, 36, four are of the tenth century and later. The spelling  for Čagatai in Laonicus Chalcocondyles, who flourished about 1485, has no bearing on the phonetic value of g in the writings of authors who lived a millennium before him.

245. Moravcsik, BT 2, 266.

246. Priscus, EL 14113; Moravcsik, BT 2, 169.

247. Moravcsik, ibid., Evagrius calls him a Scythian.

248. Theoph. Sim. 25923; Moravcsik, BT 2, 318.

249. Sidonius, Paneg. on Maiorian 488.

250. Gabain 1950a, § 57.

251. Priscus, EL 14113; Moravcsik, BT 2, 87.

252. As Pritsak 1956, 418, assumes.

253. He wrote .

254. Moravcsik BT 2, 82. Malalas has .

255. Theoph. Sim. 672, 7313; Moravcsik, BT 2, 83.

256. Moravcsik, BT 2, 62 (for 430, read 530).

257. Priscus, EL 12218; Moravcsik, BT 2, 76.

258. Xenophon, Cyrop. VIII, 6, 7; Anab. VII, 8, 25.

259. IOSPE 4, 423, 2; not later than the fourth century B.C.

260. Herodotus VII, 22, 117; 63, 8.

261. Mostly from Gorgippia (Vasmer 1923, 34, Zgusta 1955, § 596). Add  (Numismatika i epigrafika 1, 1960, 200).

262. Miller 1886, 257; he compared these names to Atakam.

263. Zgusta 1955, § 596; cf.  (Ctesias, Pers. 45, Justi 1895, 93).

264. Herodotus VII, 105; Justi 1895, 199, 498.

265. Justi 1895, 177, 271, 377, 498.

266. Vámbéry 1882, 40.

267. Moravcsik, BT 2, 77.

268. D. Pais, MNy 28, 1932, 275.

269. Priscus, EL 14325, 14710,21,28, 1481,8; Moravcsik, BT 2, 89-90.

270. L. Rásonyi, MNy 23, 1927, 280, and Archivum Europae Centro-Orientalis 1, 1935, 228. On bärk < bärig, cf. W. Bang. UJb 4, 1924, 17.

271. Müllenhoff in Jordanes, index 147; Schönfeld 1911, 50. One should expect Berica, but Berichus is also possible.

272. To the names adduced by L. Rásonyi, MNy 28, 1932, 102, add Sauvaget 1950, 45, nos. 78, 79. Markwart (1929, 83) recognized the diminutive suffix; he thought that dengi- might be the older form of yaŋi.

273. Priscus, EL 5886,24,28; Moravcsik, BT 2, 117.

274. Marcellinus Gomes, CM II, 907.

275. Getica 12021.

276. Chron. Pasch. (besides  and ).

277. It would be of interest to know at what time the Ossetes borrowed dengiz (Abaev 1958, 362) from the Turks. Incidentally, Tängiz, the youngest of the six sons of Oguz Qagan, is not the "oceanic" prince but Prince Ocean; his brothers are Sun, Moon, Star, Sky, and Mountain (W. Bang and G. Rachmati, SB Berlin 1932, 689, 691, 703; Abul Ghazi, Rodoslovnaia Turkmen, trans, by Kononov 1958 48, 50-52).

278. Getica 12526.

279. For the etymology, see W. Bang, UJb 10, 1936, 23.

280., and many more.

281. Pritsak 1953, 19, n.10, quoting Mehmed Fuad Koprülü.

282. Cf., e.g., Christensen 1944, 21, n. 3, on Ammianus' Nohodares.

283. Pritsak 1955, 68.

284. Rásonyi 1953, 333-336, listed numerous Turkish names and titles with el in the first syllable. For the Chinese transcription of el, see Pelliot, TP 1929, 226-228, and 1950, 182-183; Hamilton 1955, 151. Gf. also S. V. Kiselev 1948.

285. Chavannes 1903, 336. Cf. also Mo-yo-men, the name of two ambassadors from Maimargh and Samarkand (ibid., 135, n. 6, 145, n. 1).

286. For  all codices have  in EL 1427; C has three times the accusative , EL 12534, 1427, 14918 appears four times as , EL 1244, 12930, 1303.

287. Markwart 1929, 9, n. 1.

288. Bang 1916, 112, n. 2, accepted by Arnim 1936, 100, and Németh 1940, 223. On qan and arïg in names of women, see L. Rásonyi, UJb 34: 3-4, 1962, 233.

289. Bashakov 1951, 176, 403.

290. W. Tomaschek (SB Wien 117, 1889, 65) surmised in Kreka the ethnic name Qyrqyz; he had to work with the Bonn edition which had only Kreka. Why Haussig (1954, 361) still takes Kreka for the correct form is hard to understand; he maintains that the name is Gothic and means "the Greek woman." P. Poucha (CAJ 1, 1955, 291) takes Kreka or Hreka (sic) for Mongol gargai, "wife;" he repeats this etymology in 1956, 37, n. 39.

291. Priscus, EL 1312; Moravcsik, BT 2, 126.

292. Vámbéry 1882, 43.

293. W. Krause 1954, 327.

294. Agathias III, 17, ed. Bonn, 1775.

295. Moravcsik, BT 2, 138, following Németh and Rásonyi.

296. Agathias III, 17, ed. Bonn, 177; Moravcsik, BT 2, 170.

297. Moravcsik, BT 2, 194.

298. Codex B has . Although it is better than the codices which have  und  (C. de Boor, EL II, 516), it is still not good. The name was distorted at an early time; Anastasius in his Latin version left it out (C. de Boor, EL II, 10724); Nicephorus Callistus (PG 146,1269c) has the monstrous . Note that in the same passage and in all codices occurs , corrupt for .

Mundo (Moravcsik, BT 2, 194), the name of a Gepid of Attilanic descent (Getica 311), could be a variant of Theophanes' Mundios, provided that such a name existed. It has also been connected with Mundzucus; to the references in Moravcsik, BT 2, 194, add Pritsak 1955, 66. But Mundo's father  (Theophanes 21822), has a name with a Germanic ring (Diculescu 1922, 58) and Mundo itself may be Germanic; cf. Munderichus and Mundila (Schönfeld 1911, 169); for -o, see Schönfeld 1911, 52. Non liquet.

299. Getica 5519, 582,6,14.

300. Jordanes, index 158. Cassiodorus has of course Burgundiones (Variae 503).

301. Kent 1940, 46.

302. Zaconus in an inscription of 358 from Salona (Dessau 8254); zie for die (Detschev 1952, 1, 23.)

303. Moravcsik, BT 2, 194; cf. Brockelmann, Kāšgarī: "a precious stone, lion's claw, or amulet hung on a horse's neck."

304. "Pearl" was the title of the highest official of the Tibetans in the T'ang period (Demiéville 1952, 285).

305. Vámbéry 1882, 46. In Russian, Ukrainian, and Polish borrowed from the Tatars in the Crimea (M. Vasmer 1955, 1, 145).

306. AOH 10, 1960, 303.

307. Letter of September 10, 1962.

308. G. Doerfer, Fundamenta I, 379.

309. O. Pritsak, Fundamenta I, 579.

310. L. Bazin, Fundamenta I, 311.

311. A. Caferolu, Fundamenta I, 251.

312. Gabain (1955, 23) is inclined to derive Chinese tu < duok, "standard with a yak tail or pheasant feathers," listed in the Erh ya, from Turkish tug. It seems to me that tug is rather a Chinese loanword. Tu < duok < d'ok or tao < d'âu < d'og (GS 1016) is undoubtedly the same as tao < d'âu < d'ôg, "staff with feathers" (GS 1090z) and yu < iâu < diôg, "pendants of a banner" (GS 1080a, yu ; ancient diôg, "pennon" (GS 1080f), words which occur in the Book of Odes and the Tso-chuan, centuries before the first appearance of the Hsiung-nu, from whose allegedly Turkish language the Chinese are supposed to have borrowed duok.

313. Pelliot 1950, 69; Hamilton 1955,157,158. Proto-Bulgarian  means probably "the flag bearer", cf. Menges 1951a, 113.

314. J. A. Boyle, Islamic Studies 2:2 (Karachi, 1963), 241. The Mongols believed that Chinghiz Khan's soul went into his flag, tug-sülde, which became the patron saint of his clan and the whole Mongol people. Cf. Banzarov 1891, 24; Vladimirtsov 1934, 145.

315. Markwart 1920, 290-291.

316. Responsa Nicolai, Carm. XXXIII, p. 580. For the Kirghiz on the upper Yenisei, see Appelgren-Kivalo 1931, n. 93 (their flags are mentioned in T'ang shu, ch. 217b); for the Kurdykan, Okladnikov and Zaporozhskaia 1959, 121, 57; on the jug from Nagy Szen Miklos, see A. Alföldi, Cahiers archéologiques 1950, 132-133. On the flag of the Seljuk, see V. A. Gordlevskiĭ, Izbrannye sochineniia 1, 1960, 179; on Yak tail banners of Mongols in the time of Genghiz Khan, see Poucha 1956, 137-139.

317. Schönfeld 1911, 278.

318. This has been rightly stressed by G. Schramm 1960, 129-155. As an entirely tentative surmise, Schramm would derive Gundiok, the name of a Burgundian king, from Mundiuch, as he reconstructs the name of the prince. To judge the linguistic side of this derivation must be left to Germanic scholars. What we know about the relations of the Burgundians with the Huns in the 420's and 430's is not in its favor.

Whether  in an undatable epitaph from the northern Dobrogea (Moravcsik, BT 2, 311) has anything to do with Mundzuc is doubtful. , Tzeiuk's son, served in the corps of the Sagittarii. Cf. Diculescu 1923, 52; V. Parvan (Rendiconti della Pontifica Academia Romana di archeologia 2, 1924, 131) and Fiebiger (1939, 31-32) took Tzeiuk and Atala for Germanic, V. Beshevliev, Godishnik na Bulgarskiia Narodniia Muzeĭ 7 (1942), 1943, 232-234, and I. Stoian, Tomitana (Bucharest, 1962) 54, for proto-Bulgarian names.

319. Moravcsik, BT 2, 266.

320. Sauvaget 1950, 49, no. 120.

321. John of Antioch, EI 14222; Moravcsik, BT 2, 131.

322. Menges 1944, 264; Cf. Joki 1952, 294.

323. Sauvaget 1950, 47, no. 91. K. H. Menges (CAJ 8, 1, 1963, 56) surmises that Č'orpan in the Khazarian name Č'orpan T'arxan is colman.