The group consists of five almost complete letters and a number of fragments of similar letters. Each letter was folded several times and bore the names of the sender and addressee on the outside. Most were tied with string; one letter was wrapped in silk and enclosed in an envelope of coarse cloth addressed to Samarkand.


ANCIENT LETTERS, name conventionally applied to a group of Sogdian paper documents; discovered in 1907 by Sir Aurel Stein. They were found in the ruined watchtower T.XII.a (see Figure 1) on the Chinese frontier wall which formerly guarded the route between Tun-huang and Lou-lan. The group consists of five almost complete letters and a number of fragments of similar letters. Each letter was folded several times and bore the names of the sender and addressee on the outside. Most were tied with string; one letter was wrapped in silk and enclosed in an envelope of coarse cloth addressed to Samarkand, 2000 miles to the west. From the letters themselves it may be deduced that at least two were written in Tun-huang and one in Kutsang. The inference that they represent the contents of a “mailbag” lost or abandoned in transit from east to west accords well with the general tenor of the letters, which seem to consist largely of reports to wealthy Sogdian merchants by their representatives abroad.

At one time the Ancient Letters were thought to have been written within a few decades of the Chinese invention of paper in 105 A.D. and thus to be the oldest paper documents in existence. Although they are now attributed rather to the beginning of the 4th century (see below), the letters retain their importance both as primary sources for the early history of the silk route and as the most ancient monuments of the Sogdian language.

The date of the Ancient Letters has been much disputed. According to Stein they should have been written before the withdrawal of the garrison from the Chinese frontier posts in about the middle of the 2nd century A.D. There is in fact no archeological evidence for this assumption, since the letters were found, unassociated with any datable objects in a refuse heap, and there is nothing in the contents and nature of the documents (letters in transit) to link them with the period of settled occupation of the Chinese wall. The primitive technique of manufacture deduced by J. von Wiesner from his scientific analysis of the paper on which the letters are written would be consistent with the early dating proposed by Stein but cannot be said to provide independent confirmation of it, since paper may have continued to be produced by the same process for several centuries. Moreover the subsequent researches of A. Fujieda have led him to the entirely different conclusion that the paper of the letters cannot be earlier than the beginning of the 6th century. Their date must therefore be decided chiefly on the basis of internal evidence.

The only one of the Ancient Letters which provides precisely datable information is the second, which alludes (lines 10-19) to rumors that the Chinese emperor had fled from his capital Lo-yang, due to fighting between the Chinese and the Huns (Sogd. xwn), and to the destruction of the cities of Lo-yang and Yeh. According to W. B. Henning these details are compatible only with events at the beginning of the 4th century: the sack of Yeh in 307 A.D., the attempted flight of the emperor, and the destruction of Lo-yang by the Huns (Chin. Hsiung-nu) in 311 A.D. Henning’s conclusion that the letters were written in the years 312-13 A.D. has been generally accepted. A recent attempt by J. Harmatta to refer the letters to the period immediately following an earlier sack of Lo-yang (not by the Huns) in 190 A.D. appears less soundly based.

The references in this letter to political events in China are too brief to add significantly to the detailed accounts in Chinese annals. However, the equation of xwn with Hsiung-nu is of major historical importance, since it goes a long way towards establishing the sometimes disputed identity of the name of the Far Eastern Hsiung-nu with that of the Huns of Europe.

Other nationalities mentioned in the letters are the Chinese (cynt), Sogdians (swγ’ykt) and Indians (ʾyntkwt), i.e., the Indian inhabitants of Lou-lan, who evidently had close commercial and cultural relations with the Sogdians. Most of the personal names of the merchants and their dependants are Sogdian, but Indian names occur also, e.g., cʾtʾysh (= Kharoṣṭhī Catisa). The Iranian names, which include several compounded with the name of the goddess Nanai, provide information concerning the indigenous religion of Sogdiana.

All the places named in the Ancient Letters (insofar as they can be certainly identified) are located along the trade route, known as the silk route, linking central China with the West (see Figure 1). In some of these cities there were substantial colonies of Sogdian traders, as is shown by a reference to “100 freemen from Samarkand” (letter 2, lines 19-20). Since many of the letters are concerned with commercial matters, they are potentially sources of the first importance for the economic history of the silk route. Linguistic problems and the incomplete state of preservation of the letters unfortunately make it difficult to follow the details of the transactions involved.

The script of the Ancient Letters is less cursive than that found in later Sogdian documents (of the 8th century onwards) and represents an intermediate stage between this and the Aramaic script from which it ultimately derives. Both the shapes of the individual letters and the number of letters in use differ from those of standard Sogdian (as the later script and language may conveniently be designated). In particular one may note the preservation (in careful writing) of the distinction between gimel and ḥēth representingthe sounds γ and x respectively), which were largely confused in later times, and the survival of the letters qoph (as a symbol for the numeral “100”) and daleth. The latter occurs only in the ideogram ʿD (for Sogd. ’t “to”), one of about ten ideograms in the Ancient Letters which are no longer used in standard Sogdian. (For further details, see Sogdian script.)

The most striking feature of the vocabulary of the letters is the use of Indian loanwords for certain commercial terms: sʾrth “caravan”, prst- (a particular measure), mw’y “price”, ’ykh “letter” (corresponding to Sanskrit sārtha- , prastha- , mūlya- and lekhya respectively). The letters also contain a few words which appear to be of Western Iranian origin, such as xwtʾy “sir” (Middle Persian xwadāy “lord”), used as a mode of address. Apart from such loanwords, most of which do not occur in later documents, the vocabulary of the Ancient Letters differs surprisingly little from standard Sogdian, with only a few words which were not used later (e.g., xwtʾyṇβ “your lordship”; ʾzyy my’ “yesterday”) or which were due to undergo semantic changes (e.g., yʾtk “portion”, later meaning “meat”).

The grammar of the Ancient Letters, which has not yet received systematic study, displays many archaic features as well as a few peculiarities which may be due to non-chronological factors such as dialectal differences between standard Sogdian and the language of the letters. Some of these distinctive features of the Ancient Letters are briefly listed below.

Phonology. One of the most remarkable characteristics of the language of the Ancient Letters is the frequent occurrence of uninflected forms of words which in standard Sogdian could not occur without a vocalic inflection, e.g., krt “happened” (later ʾkrty, both forms deriving from Old Iranian *kṛtah), swγt “burned”, ʾ’yz “evil”, ’βz “hunger”. More to be expected, and possibly merely orthographic, is the preservation of some vocalic endings which were lost in later Sogdian, e.g., nwkry “now”, cntry “inside” (later nwkr, cntr).

Morphology. The declension of nouns includes some notably archaic forms such as ’rwtʾtwh (accusative of ’rwth “health”, OIr. *druwatāt-), ’wγ’ryh (ablative of ’wγth “daughter”, OIr. *duxtar-) and the nominative plural βγʾ “lords” from OIr. *bagā (replaced in the later language by βγyšt, originally a collective noun). Other distinctive features of the Ancient Letters which are rare or altogether unattested in standard Sogdian include the first person plural verbal ending -ʾymn, the enclitic third person pronoun -n and the preposition ʾc “from” (later ousted by extended forms such as cnn) from OIr. *hačā.

Syntax. The sentence ʾst iii sr’ ʾYKZYmnc pcβnt ʾβyʾrt “it is three years since I have received an answer from them” (letter 2, lines 32-33) illustrates two important syntactic traits. The first is the facility (much restricted in later Sogdian) to attach up to three enclitic particles to a single word (ʾYKZYmnc = ʾYKZY “since” + -m “I”, -n “them” and -c “from”). The second is the ergative (or passive) construction (-m . . . ʾβyʾrt, cf. Middle Persian u-m . . . padīrift), which is the normal means of expressing the perfect of transitive verbs (“I have received”), though the construction with auxiliary verb δʾr “to have”, which predominates in standard Sogdian, is also found (as in ʾβyrtw δʾrt “has received”).



Archeological details concerning the Ancient Letters and their discovery are given by A. Stein, Serindia, 4 vols., Oxford, 1921.

J. von Wiesner, Über die ältesten bis jetzt aufgefundenen Hadernpapiere . . . , Sb. Ak. Wiss. Phil.-Hist. Kl. 168/5, Vienna, 1911, presents a scientific analysis of the paper on which the letters are written. Early steps in their decipherment (A. Cowley, “Another Unknown Language from Eastern Turkestan,” JRAS, 1911, pp. 159-66.

R Gauthiot, “Note sur la langue et 1’écriture inconnues des documents Stein-Cowley,” ibid., pp. 497-507.

Idem, “A propos de la datation en sogdien,” JRAS, 1912, pp. 341-53) led finally to the full publication of the texts by H. Reichelt, Die soghdischen Handschriftenreste des Britischen Museums II, Heidelberg, 1931.

Reichelt’s edition, which includes facsimiles and a glossary, provided a sound basis for further work. Some useful corrections are to be found in the reviews by P. Pelliot in Tʾoung Pao 28, 1931, pp. 457-63 (which contains important remarks on the historical geography of the letters), and by F. Rosenberg (F. A. Rozenberg) in OLZ, 1932, cols. 758-63 (cf. also idem, “Sogdiĭskie " Starye pis’ma": K ranneĭ istorii sogdiĭskikh koloniĭ ts entral’noĭ Azii,” Izvestiya Akademii Nauk SSSR, Otdelenie obshchestvennykh nauk, 1932, pp. 445-69).

Major progress came with W. B. Henning’s article “The date of the Sogdian Ancient Letters,” BSOAS 12, 1948, pp. 601-15, which discusses not only the date but also many other historical and linguistic aspects of the letters. The date established by Henning (312-13 A.D.) has been disputed by A. Fujieda, Moji no bunkashi,Tokyo, 1971, pp. 140-43, and by J. Harmatta, “Eine neue Quelle zur Geschichte der Seidenstrasse,” Jahrbuch für Wirtschaffsgeschichte, 1971, no 2, pp. 135-43.

Idem, “The Archaeological Evidence for the Date of the Sogdian,” Acta Antiqua Academiae Scientiarum Hungaricae 24, 1976 (1979), pp. 71-86.

Idem, “Sogdian sources for the history of Pre-Islamic Central Asia,” Prolegomena to the Sources on the History of Pre-Islamic Central Asia, ed. J. Harmatta, Budapest, 1979, pp. 153-65.

Articles principally concerned with paleographical and linguistic problems of the Ancient Letters are: W. B. Henning, “Soghd. βγʾʾn-,” ZDMG 90, 1936, pp. 197-99.

N. Sims-Williams, “Notes on Sogdian paleography,” BSOAS 38, 1975, pp. 132-39.

P. O. Skjærvø, “Sogdian notes,” Acta Orientalia 37, 1976, pp. 111-16.

On Indian words in the letters see Sims-Williams, “Indian Elements in Parthian and Sogdian,” K. Rörhborn and W. Veenker, ed., Sprachen des Buddhismus in Zentralasien, Wiesbaden, 1983, p. 135.

(N. Sims-Williams)

Originally Published: December 15, 1985

Last Updated: August 3, 2011

This article is available in print.
Vol. II, Fasc. 1, pp. 7-9