period of the Indo-Iranic unity it would be necessary to suppose that it was lost in the neighbourhood of the Iranians as soon as these were separated from the Indians: a hypothesis certainly not absurd, but in the highest degree improbable and impossible to prove, because in all else the phonetic systems of the old Indian and the old Iranian dialects are closely cognate, and do not differ from each other in the total loss of primitive and common sounds, but only in developments and transformations of some among them; moreover languages, instead of losing their ancient sounds have a tendency to develope new modifications of them to be capable of expressing, by these, differences of meaning. Let us add that in the language of the Vedas, or the most ancient form of Indian known to us, the seems to be only at the commencement of its development, and many roots which later in Sanscrit have 7, in Vedic are still written with r. As it cannot be supposed that, from the very outset, there have existed double forms, one group with 1, the other with 7, for the same roots, and as it may be shown in every one of such forms that the 7 is a 19 transformation of r, the latter, and not the former, should be considered as the primitive sound in all of them. And hence Fick proceeds to note several words in which he thinks we must attribute to chance alone the agreement of Sanscrit and the Aryan dialects of Europe in the substitution of the sound for r, which he deems the primitive. He admits the existence of seven words which, on both Indian and European ground, agree in the 7 of the suffix without our being able to point to the more ancient r beside the 7: but, though we recognised them as primitive forms, we should still not be bound to consider them provided with the as early as the Proto-Aryan stage. We might with better reason suppose that, in these cases, only the less ancient Sanscrit forms with 7 have reached us, while the archaic forms with r were accidentally lost. No

one, concludes Fick, will deduce from these words a proof of the primitive nature of the sound 7. We may at the most allow that in the Proto-Aryan the r was pronounced not always uniformly, but in some cases with a sound approaching, especially at the end of a root and in the suffixes. But assuredly the 7, as a sound quite distinct from 7, cannot be assigned either to the great Aryan unity or to the Indo-Iranic unity: it was developed separately in Sanscrit, in the less ancient Iranic languages, in the fundamental European language. All the Aryan dialects of Europe agree in the change of r to 7; but the Greek and the Slavonic sometimes have where in the other languages the r is preserved unaltered: among the numerous examples quoted by Fick we will notice only laghu (light), li (Lat. linere), lik (to leave), ligh (to lick), lip (to anoint), luk (to shine), lug (to break), klu (to hear). Moreover it should be observed that the Europeans availed themselves of the change of r to l to denote new ideas, akin to those represented by the more ancient forms with r: or, if such forms had a widely extended sense, it was so distributed that part of it was left to the older forms with r, part 20 was derived from the later with 7. There follows a third series of roots which have 7 in the European languages, and to which there do not exist corresponding Indo-Iranic forms with r. Are they new roots which have arisen on European ground, or do they represent older roots with r which have been accidentally lost in the Indo-Iranian? The author does not venture to propose a solution of this problem, and contents himself with observing that without doubt also the languages of India and Irania lost a considerable part of their oldest store of roots, nor, perhaps, should we unconditionally deny to a linguistic period sq remote from us as that of the European unity the power of creating roots. But, continues Fick, roots of this kind, whatever be their origin, attest by their form a common European activity.

It is attested also by the l of several suffixes, since in the European languages new formations of words appear with derivative elements, the characteristic of which is the sound 7 and among these formations especial mention should be made of the diminutives, which, while very rare in IndoIranic, abound in the European dialects.1

2

§ 4. To the results of Fick's investigations into the history of the sound / stand in point blank opposition those which Heymann arrived at in his researches, and which he offered, furnished with as many proofs as he could collect, 21 in a recent monograph. He thinks that the agreement of Sanscrit with the European languages in the development of the in a series of examples cannot but lead us, as in similar cases, to admit the Proto-Aryan nature of this sound. In a large number of roots and words, undoubtedly primitive, 7 appears as the symbol of a well marked modification of the original sense, as opposed to older forms with r; and of this modification, no less than of the power of which expressed it, those who spoke the most ancient ProtoAryan tongue must have been conscious. Among the twenty-five examples quoted by Heymann it must suffice to mention ruk (to shine) and luk (to see), ri (to flow) and li (to adhere). Nor can Old Bactrian stand in forcible opposition to the claim of to be original, because, observes

1 Among the characteristics of the European mother-language, the existence of which he endeavours to prove, Fick enumerates also the development of the vowel sound e from α. Such development, he says, is common to all the European dialects, and was begun, and in great part completed, in the period of the unity of the European languages and peoples. This e, common to all the Aryan tongues of Europe, and ascending, therefore, in all probability to the primitive and fundamental Euro

pean, is found especially: (1) in a considerable series of old and important nominal forms (about 30); (2) in present-tense-stems (40 or more), the e of which sometimes pervades all the other forms of the verb. See ibid., v. 176-200. With respect to this argument we shall see later the opinion of J. Schmidt (see § 31).

Das 1 der indogermanischen sprachen gehört der indogermanischen grundsprache, Göttingen,

1873.

the author, whenever the forms with r and those with are distinct in meaning, Old Bactrian appears always to have rejected the latter, while the former appear for the most part abundant. It is known, and proved by examples, that a language can be so powerfully averse from certain sounds. as to lose them altogether. A grave objection to the existence of the Proto-Aryan / certainly cannot be derived from the Old Persian, in which, Heymann observes, with the exception of two proper names, no form with r corresponds to a primitive form with 7. Lastly, if the Sanscrit I had been developed from r, independently of the European 7, it is clear we ought to find examples of Sanscrit 7=European r, since it is evidently quite possible, and even probable, that the original was preserved, at least in some cases, in the European mother-language and became, on the contrary, in Old Indian, which does not always exhibit the primitive sound unchanged nor always preserve them intact more faithfully than the cognate languages. The lack of such examples is, in Heymann's view, a new proof of the change 22 ofr to leven in the primitive and fundamental language of the Aryans. But just as in many European roots the change of r to is complete, while the former sound remained unaltered in the corresponding Indo-Iranic roots, so the greater frequency in the development of the may be considered as a characteristic of the Aryan dialects of Europe: certainly not the change of r to l as a phenomenon arising in every case separately on Asiatic and on European ground.

The opinion of the author is supported by that most important prosecutor of Iranic studies, Spiegel,' who however acknowledges, on the other hand, the weight of the contrary arguments. That the ancient Iranians possessed a letter indicating the sound, and that it is a mere accident (as Lepsius and Oppert think) that 1 Beiträge, etc., viii. 121-28.

such a written symbol is not found in documents which have come down to us, seems to Spiegel, on account of the number and magnitude of such documents, by no means probable. Possibly the ancient Iranic dialects knew the not less than the Old Indian: but it may be that there was not a clear consciousness of the difference existing between 7 and r, and hence such difference was not represented by a written symbol. The experience of the readers compensated for the want of a special sign. Spiegel believes, however, that the absence of a letter expressing the in the ancient Iranic forms no grave obstacle to the admission of the Proto-Aryan existence of the . J. Schmidt, in the severe criticism which he passed upon Heymann's' brochure, throws upon him the reproach of having undertaken the investigation without sufficient preparation, of not having consistently followed the same method, of having sometimes arbitrarily derived meaning from meaning; while he affirms the only result of such labour to have been to prove that, as a rule, there do not appear in the Old Iranian those stems of words which in Sanscrit and in the European languages have 23 7. Schmidt thinks that in order to solve the proposed problem it is necessary to investigate the less ancient Iranic languages, all of which, he says, have the 7. Such a sound occurs also in Persian and Scythian names handed down to us by the Greeks, and in Zend alphabets. This fact Heymann should have brought forward, and he should have availed himself of the authority of Lepsius and of Oppert, who has made it appear, if not certain, at all events very probable, that a symbol used twice in proper names on the Cuneiform Persian Inscriptions has the power of 7.2 Bezzenberger, in his critical remarks on Heymann's monograph, declared himself less favourable

1 Jenaer literaturzeitung, 1874, pp. 204-5.

2 Revue de linguistique, iii. 459, sqq.; iv. 207, sqq.