NNER RELATION. [laying thus sketched the geography of these parts of the Indo-Enfo pean languages, we must next inquire whether any of these parts hold to each other especially close relations. Much that here seems obvious shows itself to he misleading. Thus Latin is often regarded as standing in a peculiarly close relationship to Greek. In point of fact, however, while the Greek and Latin literatures are closely connected, there is no special kinship of the two langnages. To detail all the groupings by sub divisions that hare been urged by different scholars would take too much space, hut the most prac tical historical differentiation is based on the varying treatment of the gutturals and vowels. In Hellenic, Italic, Celtic, and Teutonic, k and q represent the 4 and k of the Eastern languages. The Sanskrit tennis is preserved in the Western group. hut the media and aspirate are modified to j and h (confused with the palatal evolution). The Western group had a pure k instead of the series, and q-sounds (tending to become labials) instead of the k-series. Examples of the k-serie are Ilellenic 4-Kar6v, Italic ccatum, but Indic .4atom; 6xos, wagon, but Avestan vazaiti ; of the q-series, Italic gaud, Indic kad. In vocalism the Eastern group has a simpler series than the Western. Now between these extremes lies the Balto-Slavie (and in part the _.\natolie) system, which agrees with the Eastern group in having the .4 and k series (as against the k and q series), but with the Western group in its more varied vocalism. There are no divisions of the languages which are of so marked a character as these of the '.4ata»C and 'cent o»C divisions. Latin stands near to Celtic in some forms (future and pas sive), but near to Sanskrit in others (e.g. the ablative). So in other aspects of morphology, Balto-Slavic, Sanskrit, and Greek are akin, while in others they stand apart, and such resemblances and divergencies are found among the other lan guages as well. What the inner relation may have been cannot be more nearly established by a com parison of forms, still less of vocabulary, and it seems safest to establish as sub-groups only East ern (Middle) and Western divisions. That Indic and lranic lie so closely together may be due rather to their closer synchronism than to an originally greater similitude.
Before proceeding with the discussion of the inner relationship of these groups as shown by sounds, forms, and syntax, it is necessary to re vert to the question of the primitive language, which, as was shown above, occupied so large a share of the attention of Schleicher. Since his day the problem has assumed a new form which may be studied under two aspects, geographical and dialectical. It was the opinion of the older Sanskritists that the earliest home of the Inda Europeans was on the Pamir tablelands. Other sites have been assumed, the plains of Europe, Scandinavia, and many other centres. But on the basis of a comparison of the common lary of all the languages and the state of culture represented by it, present opinion, induced large ly by the results of Schrader's studies, inclines to the belief that the original home of the Indo Europeans was on the plains north of the Car pathian Mountains. A very important con
tribution to the question was made by Schmidt, who tried to show that the duo decimal system of the Indo-Europeans was derived from Semitic sources. The discovery in 1902 of the Arzawa Letters (see above) would strengthen the belief in an early intercommunica tion between Indo-Europeans and Semites. These data, if accepted, would show that the Indo-Euro peans at a very early period were near neighbors of the Semites, and perhaps in close communica tion with them. Further, it is known that trade communiention between the north of Europe and the southeast, possibly Semitic, took place at an early date, and, finally, it is scarcely to be doubted that Indo-European language and Judo European race are not terms implying an identity of race and language'. In short, lan guage proves nothing with regard to race. This truth leads to a clearer insight in re gard to the 'parent speech.' Instead of imagin ing that there was a racial unity once marked by linguistic unity, we must inquire whether, in giv ing up racial unity as shown by language, we must not also renounce all attempts to establish a former linguistic unity. The answer will un doubtedly be an affirmative, and it is rendered more probable by all that we know- of modern lan guages. The German dialects, for example, show no trace of ever having been identical in form. Still more striking is the illustration given by Slavic as compared with Baltic. In a word, there was no parent speech of the Indo-Europeans; but there were always related dialects. (See DLA i,Eers.) Thus the smallest dialect, geographical ly speaking, of Greece or Rome, may preserve forms as ancient as any other. The idea of a language-unity splitting up into dialects, and these again into sub-dialects is a false conception. Co-dialeetic forms is the only right expression. All efforts made to trace out a parent speech are based largely on doubtful data of the use, of words regarded as original, although we do not know whether the words may not have been borrowed. All recon structures of the parent speech, whether in the form of literature (such as Schleicher indulged in) or in the form of reconstructed hypothetical words, which are supposed to represent the sound of the Ursprache and are constantly employed in our comparative grammars, are pure fictions of the imagination, and are not to be regarded as representing an actual prehistoric condition of the language. We know the Indo-European lan guages only in dialectic form, as spoken by vari ous races over a wide area. The dialects are divarications from each other, not, so far as we know, from any unitary parent speech. A uni versal parent speech is an historical possibility, but an Indo-European parent speech is merely a eonvenient hypothesis, which, however, leaves more to be explained than it explains, and may best be discarded, however pleasing it be to the mind whieh prefers unity.