Then scholars formerly preferred to render by d, approximating it to the Persian, now we rather prefer to leave it in the more ancient form ; is probably a spirant (or a spirate), and we may have itg survival in our English third personal th, as in ‘doth’, ‘hath’ ete. I now write both ¢ and ¢, d(f).
Then the letter written 545 in the very ancient and important Zend-Pahlavi MS. J.? is transcribed by me s(t) solely to show its shape. Tt should be pronounced simply sh. In the Commentary and in the reprinted parts I put the & in parentheses to guard the beginner, thus 5(h).
The difficulty in meddling with transliterations is very great when one is preparing an extended work. The mechanical labour of producing the book cannot be completed before the fashions change. Some able scholars have changed their modes of transliteration even from year to year.
I shall not therefore reprint my entire work to replace é@ by e or é, w= yf by v, and v= »y by w. (This last is still somewhat du- bious in certain connections), nor shall I, on the other hand, in reprinting a large portion of the work now badly antiquated, for the sake of uniformity between the old and the new printing, use the antiquated characters, for wherever they are still left they were printed by me with great reluctance and aversion long after I had disapproved of © their use, and solely with the mistaken purpose of making the two parts of the book exactly correspond even as to this trifling mechanical peculiarity.
If this humble production were intended as a parade exposition without interior discussions, I might be tempted to reprint the very few antiquated transliterations which occur in a small portion of it, as it is, I rely on the ample notes to warn the beginner.
As to the Pahlavi, I have used Haug’s transliteration not because it is not susceptible of improvement, for I have slightly altered it at West’s suggestion, but because the glossaries for the most part cor- respond to it.
For Sanskrit, I have adopted the simplest possible modes, some- times at the sacrifice of pleasing effect. I have written ch because English ¢ suggests a different sound, and I have therefore necessarily written the clumsy chh, also m everywhere, correcting however its occurrence in the MSS. in such places as the third pl. °am#i for °anti,
As to my transliteration of the Parsi-persian, there is little that I would change at present, for no scholar will be at all likely to mistake it for an exercise in the latest dialect. Perhaps if I were not committed
to w for 4, 1 might write v everywhere now, but as I have used Ger- man w = English v for Zend yg this is only a further transgression.
Then I may also recall the somewhat dubious use of some Persian transliterators who considef the Persian v to be a sound verging upon w. This also induced me to adopt w in this mixed dialect. In a purely Persian text I should now prefer ».
Then I was perplexed at the Pazand w = Pahl. va, as occurring not in the middle of a sentence where it is unobjectionable, but at the beginning of a sentence, while the Persian has always wa (properly va) at the beginning and w in the interior, whereas we suppose the Pahlavi ) to be better transliterated as va throughout. In the chaos of usage I for-
merly wrote wa throughout after the Pahlavi, but in the reprinted portions I will adopt the Persian custom.
I have written this Parsi-persian hitherto well-nigh as unlike the latest Persian as possible, and intending to lead scholars to connect the words at once with their Zend, or Pahlavi originals (that is, where they are not Arabic). For instance I wrote ruwdn (ep. rubdn, urvan), “and not as usually in Persian rawan, jihin (ep. géhané) and not jahdn, and in my earlier treatment I even ventured upon wrand as more original than barand. For the differing letters I use h for xay © (Greek x), s, or th, for say Cs 2 for zal O, zh for jay (French j) fs § for sad US, z for zad US, x for zay i ‘for ‘ain & gh for ghain & k for kaf U9:
As to my alternative method in dealing with the translation of the Gatha, Imake no apology whatsoever. Some scholars in other branches of oriental study have been praised for the assurance of their statements as calculated to carry conviction home to their readers. This tone, while highly desirable for popular reproduction and short treatises, is in my opinion entirely to be discarded in exegetical works which go to the depths of a subject.
The Zend Avesta, while fully made out for all the purposes of com- parative religion, and history, yet presents in its ultimate detail diffi- culties so great, that more than one differing suggestion is a necessity in discussing with serious scholars.
New light, of what we hope to be new light, is sometimes got in these studies within a few months, not to say years, and I therefore present as much alternative opinion as is practicable, and naturally on those portions which have been the longest printed. And I do not always reprint for the simple reason that the older views are alternatively pos- sible, and may be (some of them) after all the best.
So likewise with the Pahlavi translation. Scholars in other depart- ments of oriental research may be surprised to hear of a translation of a translation, but Zendists will smile in their turn at such a feeling. As is the case in other important instances, the accompanying translation in a difficult, language helps indeed when partially deciphered and translated, but the full translation and explanation of this translation presents very much more difficulty than the now several times attempted
translation of its original. The first task is of course the decipherment of the Pahlavi charac-
ters which are at times so utterly indefinite that such words as a and nafsh°, for instance, are spelt with the same signs.
The next effort is to know how to begin to handle the question of its translation. No simple rendering of it as ordinary Pahlavi is at all safe. The reason of this is that the sequence of the words in pure Pah- lavi is of great importance to the syntax and the resulting meaning, but the consecutive order of the words in these translations is for the most* part controlled by that of the Gatha of which it is largely a word for word rendering, and the order of words in the Gatha lines differs very widely from this order in an ordinary Pahlavi sentence, the degree in which the two languages are inflected being also widely different. How then can we treat such a fettered rendering as if it were ordinary Pahlavi, the meaning of which depends so largely on the order of the sequence of the words?
Can we then abandon altogether the attempt at anything further than an indication of the roots present? This would indeed be more scientific than the foolish attempt to read these translations as ordinary Pahlavi. But here again we are headed off. The word for word sequence, while very largely that of the Gatha which it translates, is not wholly so. It is fitfully departed from to a greater or less extent in almost every strophe. We are also constrained to attempt a translation of the Pahlavi of the Gathas from the fact that the language as it stands offers a reasonable meaning, and that, when due allowance is made for the unusual circumstances, we can apply the laws which interpret the meanings of Pahlavi words from their positions in a sentence in a somewhat unusual manner. For instance, when a noun or pronoun stands at the beginning of a sentence in such a connection that we should naturally take it as a nominative while it yet translates a Gathic noun or pronoun in an oblique case, we are in such a case not entitled but obliged by critical laws to credit such a Pahlavi noun or pronoun with its priority in position, and accordingly we are both allowed and necessitated to render it as if in an oblique case to correspond to the original of which it is an attempted translation, or the relic of an attempted translation.
So also I think we should at times credit a Pahlavi word with the case of its original even when it possesses no claims to such a declension from prepositions, postpositions, or from its position in the sequence of words (I will not say in the sentence). We should not
forget that the last transcriber of the Pahlavi had just written the inflected Gathie word before its Pahlavi equivalent, and while this last transcriber was merely the reproducer of much more ancient materials in the Pahlavi translation which may haye translated a totally different Gathie text*, yet we cannot but suppose that the sight of the Gatha word written almost beside its supposed Pahlavi equivalent may have made the copyist less careful to reproduce the prepositions or postpositions which, beside the position of the Pahlavi word in the sentence, could alone determine its case.
So also in treating the glosses. No expert will now doubt that they are often originally of different age from the text. The text grew out of the Gatha itself, and the glosses have been added and varied from generation to generation. The text should be considered apart from them, and no treatment is complete without an alternative presented in that sense, while even in the most superficial rendering of the Pahlavi translation these glosses should be so managed as not to interrupt the flow of the sentences. One further element of difficulty is the undoubted fact that the Pahl. trir. reproduces at times alternative translations. Very often two (or more (?)) distinct translations of the same word have descended to the last translator from different predecessors. Occasionally he formally introduces them with the words ‘some say’, ait mim yemaleliméd, again he simply observes ‘it may be’, yehevindd. But more frequently the last worker-over of the Pahl. trl. betrays his ignorance of the fact that an alternative translation is before him; and works two widely differing and distinct translations for the same word (!) inherited from predecessors into the body of his text, or introduces them, if original with himself, in such a way as to give the reader no notice that they are alternative expressions for the same word.
Any competent philolog will acknowledge that we have here most harassing difficulties before us. I have accordingly first endeavoured to render the Pahlavi translation as being as closely faithful a repro- duction of its original as the facts will allow us to suppose. That is to say, I have pursued this course in the reprinted portions; formerly I was animated by a strong desire to make the Pahl. trl. appear as unlike its original as was possible; this from a mistaken conscientiousness. My first translation, even in my reprinted portions, I must offer of course
% Sometimes the Pahlavi translation is presented entirely without any Zend text, as in the Munich MSS, of Haug’s collection, 12 a, b,
as made in the light of the glosses, but I add copiously other Suggestions made as if with the marring glosses omitted. I need hardly say that this treatment is properly exposition rather than translation, and as such I desire it to be considered.
The same remarks apply with some modifications to the work of Neryosangh. Is it possible to translate his text? We must endeavour to explain it, if we cannot fully translate it, and we can only hope to do so by the exercise of unusual care.
In the first place it is a great step in advance that no respectable scholar will ever again criticise Neryosangh’s work as an immediate translation of the Yasna. Nor, may I hope, will any one again, suppose the meaning of Neryosangh to coincide with a true translation of the Pahlavi which was chiefly his original, if for no other reason, then because we do not know exactly what his Pahlavi text was. The best Pahlavi text which we can now produce evidently differs sometimes from the Pahlavi texts which he used, and Neryosangh’s rendering of those parts of the Pahlavi which we know to have been actually before him, while a noble attempt at his early day, and of great importance since as affording invaluable hints, still leaves, as might be expected, very much to be desired. No man writing at that early period could fail to err both as to transliteration and translation. Had I offered Neryosangh’s text without translation, or explanations so full as to be equivalent to translations, I should have left students a wide scope for error, and surely it is no very scholarly proceeding to cite Neryosangh quite astray.