Sacred Texts  Zoroastrianism  Index  Previous  Next 
Buy this Book at

Pahlavi Texts, Part I (SBE05), E.W. West, tr. [1880], at


The term 'Pahlavi,' in its widest extent, is applied to all the varying forms of the mediæval Persian language, from the time when the grammatical inflexions of ancient Persian were dropped, till the period when the modern alphabet was invented, and the language became corrupted into modern Persian by the adoption of numerous Arabic words and phrases. Some traces of Pahlavi words and phrases, written in old Semitic characters, have been found in the legends of coins struck by certain kings of Persian provinces, subordinate to the Greek successors of Alexander, as early As the third century B.C. 1 Further traces have been discovered in the legends on some provincial coins of the time of the Arsacidan dynasty. But, practically, our acquaintance with Pahlavi commences with the inscriptions, on rocks and coins, of Ardakhshîr-i Pâpakân (A.D. 226-140), the founder of the Sasanian dynasty, and ends with certain religious

p. xii

writings of priests and other devout Parsis of post-Muhammadan times, among the latest of which is one dated A.Y. 250 (A.D. 881). Any fragments of Pahlavi composition of later date than A.D. 1000, must be considered merely as modern imitations of a dead language, and cannot be quoted as authorities for the use of any particular Pahlavi words or construction.

With regard to the origin of the word Pahlavî, or language of Pahlav, many suggestions have been offered; but the most probable explanation 1 is that which connects it of with the Parthva, of the cuneiform inscriptions, the land the Parthians known to the Greeks and Romans, and of the Pahlavâs mentioned by Sanskrit writers; the change of Parthva into Pahlav being very similar to that of Av. Mithra into Pers. Mihr. No doubt the language of the Parthians themselves was not Pahlavi, but they were the actual rulers of Persia for some centuries at the time when the Pahlavi language was forming there; and, being formidable to their neighbours, it is not surprising that their name became identified with everything Persian, in the same way as the Roman name has been applied by the Persians, not only to the later Greek empire of Constantinople, but even to the earlier conqueror, Alexander the Great.

Strictly speaking, the mediæval Persian language is only called Pahlavi when it is written in one of the characters used before the invention of the modern Persian alphabet, and in the peculiarly enigmatical mode adopted in Pahlavi writings. Whenever it is transcribed, either in Avesta characters, or in those of the modern Persian alphabet, and freed from this peculiarity, it is called Pâzand.

The peculiar mode of writing Pahlavi, here alluded to long made the character of the language a standing puzzle for European scholars, and was first satisfactorily explained, by Professor Haug, of Munich, in his admirable Essay on the Pahlavi Language already cited.

Like the Assyrians of old, the Persians of Parthian times appear to have borrowed their writing from a foreign race.

p. xiii

[paragraph continues] But, whereas the Semitic Assyrians adopted a Turanian syllabary, these later Aryan Persians accepted a Semitic alphabet. Besides the alphabet, however, which they could use for spelling their own words, they also transferred a certain number of complete Semitic words to their writings, as representatives of the corresponding words in their own language. These Semitic representatives (the number of which might at any time be increased or diminished at the discretion of the writer) were probably never very numerous, and not more than four hundred of them are to be found in the Pahlavi, writings now extant; but, as they represent nearly all the commonest words in the language (excepting those specially relating to religious matters), they often constitute more than half the bulk of a Pahlavi text.

The use of such Semitic words, scattered about in Persian sentences, gives Pahlavi the motley appearance of a compound language; more especially as Persian terminations are often added to the Semitic words. But there are good reasons for supposing that the language was never spoken as it was written. The spoken language appears to have been purely Persian; the Semitic words being merely used, as written representatives, or logograms, of the Persian words which were spoken. Thus the Persians would write malkân malkâ, 'king of kings,' but they would read shâhân shâh. This is still the mode in which most Parsis read their Pahlavi literature; and it is only by assuming it to have been their universal practice, in former times, that we can account for the total and immediate disappearance of the Semitic portion of the Pahlavi, from their language, when the Persians adopted their modern alphabet. As the Semitic words were merely a Pahlavi mode of writing their Persian equivalents (just as 'viz.' is a mode of writing 'namely' in English), they disappeared with the Pahlavi writing, and the Persians began at once to write all their words, with their new alphabet just as they pronounced them.

In the meantime, the greater part of the nation had become Muhammadans, and a new influx of Semitic words commenced, but of a very different character. The Semitic

p. xiv

portion of the Pahlavi writing was nearly pure Chaldee, and was confined (as already stated) to the graphic representation of most of the simplest and commonest words unconnected with religion; but it seems to have formed no part of the spoken language, at all events in later times. Whereas the Semitic portion of modern Persian is borrowed from Arabic, and includes most words connected with religion, science, and literature; in fact, every class of words except that which was usually Semitic in Pahlavi writings; and these Arabic words form an essential part of the spoken language, being as indispensable to the modern Persian as words of Norman-French origin are to the English.

In Pahlavi writings, moreover, besides the four hundred Semitic logograms already mentioned, we also find about one hundred obsolete forms of Iranian words used as logograms; much in the same way as 'ye' may be used for 'the,' and 'Xmas' for 'Christmas' in English. The use of all these logograms was, however, quite optional, as their usual Persian equivalents might, be substituted for any of them at any time, according to each particular writer's taste and discretion. But whenever they are employed they form. what is called the Huzvâris portion of the Pahlavi; while the other words, intended to be pronounced as they are spelt, form the Pâzand portion.

Many attempts have been made to explain the word Huzvâris, but it cannot be said that any satisfactory etymology has yet been proposed. Like the word Pahlavî it seems hardly to occur in any old Pahlavi text, but only in colophons, chapter-headings, and similar notes of modern writers; it seems, therefore, more reasonable to trace it to modern Persian than direct to any more ancient source. Its Pahlavi form, hûzvâris or aûzvârisn, appears to represent the modern Persian uzvâris, which is rarely used; the usual Persian form of the word being zuvâris. Now zuvâris is precisely the form of an abstract noun derived from the crude form of a verb zuvârîdan, which has been admitted into some Persian dictionaries on the authority of Golius 1.

p. xv

with the meaning 'to grow old, to become thread-bare.' If such a verb really exists in Persian, although its meaning may imply 'decrepitude or decay' rather than 'antiquity or obsoleteness,' yet its abstract noun would not be altogether inapplicable to the logograms used in Pahlavi, which are, in fact, last remnants of older writings.

The word Pâzand is probably derived from Av. paitizanti, with the meaning 're-explanation,' that is, a further interpretation of the Pahlavi Zand in the Persian vernacular. This term is applied not only to the purely Persian words in Pahlavi texts, but also (as already noticed) to transliterations of the said texts, either in Avesta or modern Persian characters, in which all the Huzvâris words are replaced by their Pâzand equivalents. These transliterations form what are called Pâzand texts; they retain the exact idiom and construction of the Pahlavi original, and represent the mode in which it was read. It may be remarked, however, that all such Pâzand texts, as have been examined, seem to have been written in India, so that they may be suspected of representing some corrupt Gugarâti pronunciation of Persian, rather than the peculiar orthography of any period of the Persian language.

This theory of the origin and development of Pahlavi writing could hardly be upheld, unless we could trace the same artificial mixture of Huzvâris and Pâzand in all accessible Pahlavi records, from their earliest appearance to the present time. This we are able to do, even in the scanty materials afforded by the legends on the provincial Persian coins of the third century B.C. and second century A.D. already mentioned. But we can trace it with greater certainty not only in the coin legends, but also in the rock inscriptions of the earlier Sasanian kings (A.D. 226-388), in the latest of which we find the written language differing very slightly from that contained in the manuscripts pre served by the Parsis of the present, day, although the characters differ very much in form. And, finally, in the legends on the coins of the later Sasanian kings (A.D. 388-651) and on seals of their times, we find even this difference in. the shapes of the letters disappearing by degrees. In

p. xvi

fact, all the materials at our disposal tend to show that Huzvâris has been an essential constituent of all Pahlavi writings from the time of Alexander's successors to that of the disuse of Pahlavi characters; but we have no reason to suppose that the spoken language of the great mass of the Persian. people ever contained the Semitic words which they thus used as Huzvâris in their writings.

Although the use of Huzvâris, until explained recently, rendered the nature of the Pahlavi language very obscure, it added very little to the difficulty of understanding the Pahlavi texts, because the meaning of nearly every Huzvâris logogram was well known; being recorded in an old glossary preserved by the Parsis, in which every logogram is explained by its proper Pâzand equivalent. The extant copies of this old glossary generally contain the Huzvâris and Pâzand words written in the Pahlavi character, together with their traditional pronunciation, either in Avesta or modern Persian letters; there is, therefore, no particular difficulty in reading or translating the Huzvâris portion of a Pahlavi text, although doubts may often be entertained as to the accuracy of the traditional pronunciation.

The real difficulty of reading Pahlavi texts lies in the Pâzand portion (so far as it may be unexplained by existing vocabularies), and is chiefly occasioned by the ambiguity of some of the Pahlavi letters. The alphabet used in Pahlavi books contains only fourteen distinct letters, so that some letters represent several different sounds; and this ambiguity is increased by the letters being joined together, when a compound of two letters is sometimes exactly like some other single letter. The complication arising from these ambiguities may be understood from the following list of the sounds, simple and compound, represented by each of the fourteen letters of the Pahlavi alphabet respectively:—

a, â, h, kh. b. p, f, t, d. k, g, z, v. r, l. z. s, yî, yad, yag, yag, dî, dad, dag, dag, gî, gad, gag, gag, gî, gad, gag, gag. sh, s, yâ, yah, yakh, îh, îkh,

p. xvii

dâ, dah, dakh, gâ, gah, gakh, gâ, gah, gakh. gh. k. m. n. v, w, û, ô, r, l. y, î, ê, d, g, g.

From this list it is easy to see the confusion produced by the letter s being exactly like the letter y doubled, and by the letter sh being identical with a compound of y and â; and there are, in fact, some compounds of two letters which have from ten to fifteen sounds in common use, besides others which might possibly occur. If it be further considered that there are only three letters (which are also consonants, as in most Semitic languages) to represent five long vowels, and that there are probably five short vowels to be understood, the difficulty of reading Pahlavi correctly may be readily imagined.

When Pahlavi writing was in common use this difficulty was probably no more felt by the Persians, than the complexity of Chinese characters is felt as an evil by a Chinese mandarin, or the corrupt system of English orthography by an educated Englishman. It is only the foreigner, or learner, who fully appreciates the difficulty of understanding such cumbrous systems of writing.

With regard, however, to their Huzvâris logograms the Persians seem to have experienced more difficulty. As the actual sounds of these Semitic words were rarely pronounced, in consequence of their Pâzand equivalents being substituted in reading, there must have been some risk of their true pronunciation being forgotten. That this risk was understood by the Persians, or Parsis, is proved by the existence of the Huzvâris-Pâzand glossary already described, which was evidently compiled as a record both of the pronunciation and meaning of the Huzvâris logograms. But its compilation does not appear to have been undertaken until the true pronunciation of some of these logograms had been already lost. Thus, although the traditional readings of most of the Semitic portion of the Huzvâris can be readily traced to well known Chaldee words, there are yet many other such readings which are altogether inexplicable as Semitic

p. xviii

words. In most such cases, however, European scholars have found that the Huzvâris word can be easily read in some other way which at once connects it with some ordinary Chaldee equivalent. It may, therefore, be reasonably assumed that the compilers of the glossary had in some instances lost the correct pronunciation of these old Semitic words, and that, in such cases, they adopted (as a Parsi would probably do at the present day) the most obvious reading of the letters before them, which thenceforth became an artificial word to, be handed down to posterity, by successive generations of writers, with all the authority of old tradition.

In the same manner the artificial pronunciation of the Iranian portion of the Huzvâris may be explained. The compilers of the glossary found a number of words in the Pahlavi texts, which were written in some obsolete or contracted manner; they knew the meanings of these words, but could not trace the true readings in the altered letters; they, therefore, adopted the most obvious readings of the written characters, and thus produced another series of artificial words, such as anhômâ for aûharmazd, yahân for yazdân, madônad for maînôk, shatan for shatrô, &c.

Naturally enough the Parsis are loth to admit the possibility of any error in their traditional readings of Huzvâris, and very few of them have yet adopted the views of European scholars further than to admit that they are ingenious hypotheses, which still require satisfactory proof. They are quite right in demanding such proof, and they may reasonably argue that the conflicting opinions of various European scholars do not tend to increase the certainty of their explanations. But, on the other hand, they are bound to examine all proofs that may be offered, and to consider the arguments of scholars, before utterly rejecting them in favour of their own preconceived notions of traditional authority.

Fortunately, we possess some means of ascertaining the ancient pronunciation of a few Huzvâris words, independent of the opinions of comparative philologists, in the inscriptions

p. xix

already mentioned as having been engraved on rocks, and impressed on coins, by the earlier kings of the Sasanian dynasty in Persia. The earliest of these rock inscriptions records the name and titles of Artakhshatar son 1 of Pâpak, the first Sasanian monarch (A.D. 226-240); it is engraved in Greek and two kinds of old Pahlavi characters, which have been called Chaldæo-Pahlavi and Sasanian-Pahlavi, because the one bears more resemblance to Chaldee, both in its letters and the language they express, and the other is more frequently used by the subsequent Sasanian monarchs. A similar tri-lingual inscription records the names and titles of his son and successor Shahpûhar I (A.D. 240-271), who has also left a long bi-lingual inscription, in Chaldæo and Sasanian-Pahlavi, in a cave near Persepolis. Another long bi-lingual inscription, fragments of which have been found on stones among the ruins of Pâî Kûlî, is attributed to his early successors, who have also left us several uni-lingual inscriptions in Sasanian-Pahlavi, two of which are of great length, but none later than the end of the fourth century.

The language of the earlier of these inscriptions differs from that of the manuscripts preserved by the Parsis, chiefly in the use of several Semitic words unknown to the manuscript Huzvâris, the non-existence of Iranian Huzvâris (which is evidently a growth of later times), and the less frequent use of Persian terminations affixed to Semitic words. These differences, however, are hardly greater than those which distinguish the English of Chaucer from that of our own day. Moreover, they gradually disappear in process of time, as we find the later inscriptions of the fourth century approaching much closer, in language, to the manuscripts.

As the alphabets of these inscriptions are less imperfect and ambiguous than that of the Pahlavi manuscripts, they render the pronunciation of many words much more certain. They consist of eighteen letters, having the following sounds:—

p. xx

1. a, â. 2. b. 3. p, f. 4. t, d, 5. k, g, ‏צ‎. 6. kh, h. 7. d. 8. r, v, w, û, ô. 9. z. 10. s. 11. sh, s. 12. k. 13. g. 14. l, r. 15. m. 16. n. 17. y, î, ê. 18. doubtful, being equivalent to Chaldee , and to Pahl. MS. -man 1.

Comparing this list of sounds with that of the sounds of. the manuscript alphabet (pp. xvi, xvii) it is evident that the inscriptions must afford a means of distinguishing â from kh, s from any binary compound of y, d, g, or g, sh from any compound of y, d, g, or g with â, h, or kh, n from v, r, or l, and y, d, g from each other; all which letters and compounds are left in doubt by the manuscript alphabet. Unfortunately we do not possess trustworthy copies of some of the inscriptions which are evidently the most important from a linguistic point of view 2 but such copies as have been obtained supply corrections of traditional misreadings of about twenty-five Huzvâris logograms, and at the same time they confirm the correctness of three traditional readings which have been called in question by most European scholars. So far, therefore, the inscriptions would teach the Parsis that the decisions of comparative philologists are not likely to be right more than seven times out of eight, even when they are tolerably unanimous.

The Chaldæo-Pahlavi character appears to have soon

p. xxi

gone out of use, after the establishment of the Sasanian dynasty, as the latest known inscription, in which it occurs, is that of Pâî Kûlî, which contains the name of Aûharmazd I (A.D. 271-272); while the long inscriptions of Naqs-i Ragab and Naqs-i Rustam, which contain the name of Varahrân II (A.D. 275-283), are engraved only in Sasanian-Pahlavi. From these facts it seems probable that Chaldæo-Pahlavi went out of use about A.D. 275. The Sasanian characters continue to appear, with very little alteration, upon the coins until the end of the fifth century, when most of them begin to assume the cursive form of the manuscript Pahlavi, which appears to have altered very slightly since the eighth century.

The oldest Pahlavi manuscript known to be extant, consists of several fragments of papyrus recently found in a grave in the Fayûm district in Egypt, and now in the Royal Museum at Berlin; it is supposed to have been written in the eighth century. Next to this, after a long interval, come four manuscripts written on Indian paper, all by the same hand, in A.D. 1323-1324; they are two copies of the Yasna and two of the Vendidad, containing the Avesta with its Zand, or Pahlavi translation and commentary; two of these old MSS. are now preserved in Kopenhagen, one in London, and one in Bombay. Next to these in age are two MSS. of miscellaneous Pahlavi texts, written probably about fifty years later; one of these is now in Kopenhagen and one in Bombay. Another MS. of nearly the same age is also a miscellaneous collection of Pahlavi texts, written in A.D. 1397, and now in Munich; where there is also one of the oldest Pâzand-Sanskrit MSS., a copy of the Ardâ-Vîrâf-nâmak, written in A.D. 1410. Another Pâzand-Sanskrit MS., a copy of the Khurdah Avesta, of about the same age, exists in Bombay. Pahlavi and Pâzand manuscripts of the sixteenth century are rather more numerous.

Pahlavi literature reached the zenith of its prosperity about thirteen centuries ago, when it included the whole literature of Persia. Seventy years later its destruction commenced with the fall of the Sasanian dynasty (A.D.

p. xxii

[paragraph continues] 636-651); and the subsequent adoption of the modern Persian alphabet gave it its death-blow. The last remnants of Pahlavi writings are now contained in the few manuscripts, still preserved by the Parsis in Western India, and their almost-extinct brethren in Persia. A careful estimate of the length of these remnants, so far as they are known to, Europeans, has shown that the total extent of existing, Pahlavi literature is about thirty-six times that of the Bundahis, as translated in this volume. One-fifth of this, literature consists of translations accompanying Avesta texts, and the remaining four-fifths are purely Pahlavi works which are nearly all connected with religion. How much of this literature may have descended from Sasanian times can hardly be ascertained as yet; in fact, it is only very recently that any trustworthy data, for determining the age of a few Pahlavi writings, have been discovered, as will be explained hereafter, when considering the age of the Bundahis.


xi:1 See Levy's Beiträge zur aramäischen Münzkunde Eran's, und zur Kunde der ältern Pehlewi-Schrift, Zeitschrift der deutschen morgenländischen Gesellschaft, Leipzig, 1867; XXI, 421-465.

xii:1 See Haug's Essay on the Pahlavi language, Stuttgart, 1870; pp. 33-37.

xiv:1 See Castelli Lexicon Heptaglotton, Pars altera, London, 1669.

xix:1 So stated in the inscription, but Pahlavi MSS. call him the son of Pâpak's daughter and of Sâsân (see Bund. XXXI. 30).

xx:1 Whether the sound of this letter can ever be satisfactorily settled remains doubtful. Levy, in his Beiträge, cited on p. xi, considers it to be the Semitic ‏ה‎, on palæographical grounds; but there are serious objections to all the identifications that have been proposed.

xx:2 The Sasanian inscriptions, of which new and correct copies are most urgently wanted, are:—1. An inscription of thirty-one lines high up in the left side-compartment (behind the king) of the centre bas-relief of Naqs-i Ragab, near Persepolis. 2. Two inscriptions, of eleven and twelve lines respectively, on the stones of the edifice near the south-west corner of the great platform at Persepolis, south of the Hall of Columns (see Ouseley's Travels in Persia, vol. ii. p. 237 and plate 42). 3. All the fragments of the Pâî Kûlî inscription, of which probably not more than half have yet been copied.

Of the very long inscription behind the king's horse in the bas-relief of Naqs-i Rustam, containing more than seventy lines very much damaged, a copy taken by Westergaard in 1843 with his usual accuracy, probably gives nearly all that is legible. And of the Hâgîâbâd and shorter inscriptions, little or nothing remains doubtful.

Next: 3. The Bundahis