786 PHILOLOGY [ARYAX sound, seemed to indicate that the primitive Aryan stock had once been split into two halves, one of which remained in Asia and retained the primitive cr-sound, while the other half emigrated to Europe and there developed the new vowel-system, before any new divisions took place. The Aryan parent-speech would thus appear to have been split into a European and an Asiatic " base-language." Similar facts in the history of the single European languages then led to the further assumption of a southern European base as the parent of Greek, Italic, and Celtic, and a northern European base for Germanic, Baltic, and Slavonic, and, with further subdivision, an Italo-Celtic and a Litu-Slavic base for Italic and Celtic on the one hand and for Baltic and Slavonic on the other. The prehistoric development of Aryan, according to this genealogical theory (which makes division of language dependent on division of nations), may be illustrated by the following genealogical table. 1 Aryan. Asiatic. European. Indian. Iranian. Southern European. Northern European. Greek. Italo-Celtic. Germanic. Litu-Slavic. Italic. Celtic. criticized. r i Baltic. Slavonic. It may still be admitted that at least the mutual position of the ten families is not the same in all cases. It cannot be doubted that Indian and Iranian resemble each other more than either of them does any other family. The same may also be said of Baltic and Slavonic, and even of Italic and Celtic, Irr.vever different the latter two may appear to be at first sight.- But it is impossible to carry this system of genealogical grouping through. It will be observed that not all the ten families are represented in Genea- the genealogical tree given above ; Albanian and Armenian logical have not found a place in it, nor could they be introduced without disturbing the entire table. If we look at Ar menian, for instance, we find that its structure and phono logy on the whole follow the Asiatic type, and yet Armenian shares the European vowel-system alluded to before ; com pare, for instance, Armenian berem, "I bear," with Greek </>epw, Latin fero, Old Irish berimm (and dobiurior *do-beru), Gothic baira (pronounced bcra), Lith. berii, Slavonic bera, against Sanskrit bhdrdmi, Zend bardmi. Armenian, then, is half European, half Asiatic, and if such an intermediate idiom exists it is impossible to make a strict distinction between Asiatic and European. Let us take another in stance. All the Asiatic languages have changed the ori ginal palatal k into sibilants, and the same change we find again in Slavonic and Baltic, both of which otherwise clearly belong to the European type ; compare, for instance, Sanskrit and Zend da$an, "ten," Armenian tasn, Slavonic des$ti, Lith. deszimt, with Greek SeKa, Latin decem, Old 1 This pedigree is the one ultimately given by Schleicher. Others have assumed more or less different degrees of relationship. Greek and Itilic, for instance, were for a long time believed to be particularly near relation*. A totally contrary view would come nearer the truth. Greek and Latin are about as different, both in phonology and gram matical structure, as any two members of the Aryan family ; indeed there is nothing to recommend their combination but the intimate connexion in which the two nations and their literatures have stood within lii-torical times, and the custom derived therefrom of studying the two classical languages together from our schooldays. 2 Amongst the characteristics of these two groups the general resemblance in the declension, and in the verb the formation of a future in b or / (Latin amabo, Old Irish car/a, ruo charV) and of a passive in r (Latin fertur, Old Irish cnrthir, &c. ), are the most important. Irish deifh (for *dekvrn), Gothic taihun. In a similar way Litu-Slavic and Germanic are connected by the formation of a plural dative in m, as in Gothic undfam, Lith. vllkams, Slavonic vlukoinii, against the Sanskrit -bhyas, Latin -bus, Irish -b ; and so all round. The consequence is that every attempt at grouping the Aryan families of speech on the genealogical basis must fail, because it would have to cut asunder some of the natural ties that hold the single families together. It is true that some of the coincid ences falling under this head may be due to mere chance, especially those in phonology ; for we often see the same phonetic processes going on in languages which stand in no connexion whatever at the time. Yet in the case before us the number of the actual coincidences is too large to allow of such an explanation, and the fact of their existence is made all the more striking from the circumstance that it is each pair of neighbouring families which shows these connecting links. If they prove anything (and it cannot be doubted that they do), we must necessarily come to the conclusion that every such link is a witness for at least a temporary connexion between the two languages or families it holds together. To assume such temporary connexions in the time after a true division of nations had taken place (that is, to assume, for instance, that Slavonic had come into contact with the Asiatic languages after the Europeans had migrated from Asia to Europe, or the forefathers of the present Asiatic nations from Europe to Asia, as the case may be) seems impossible. It is likewise highly improbable that con nexions intimate enough to leave distinct marks in lan guage existed at a time when the original tribe had spread over the wide regions now covered by the Aryans, even sup posing this spreading to have been so gradual as not to cause any break in the continuity of the Aryan population. And, even if we concede this, how are we to account for the fact that we have no longer the supposed continuity of speech, but well-defined single languages, whose separa tion must, after all, be due to breaks in the continuity of intercourse between the respective speakers ? These and similar reasons point to the assumption that the origin of the phenomena alluded to must be sought in a remote period, when the Aryan tribe had an extension small enough to permit continuity of intercourse, and yet large enough to allow of dialectic variations in its different districts. In other words, when the actual break-up of the Aryan tribe into different nations came to pass the Aryan parent -speech was no longer a homogeneous idiom, but the development of dialects had begun. On their following wanderings, then, those tribes or clans would naturally cling together which had until then lived in the closest connexion both of intercourse and dialect (for community of intercourse and of speech always go together), or, as we might also say, the old unity would naturally be broken up into as many parts as there had been dialectic centres. Transition dialects, which might have been spoken in the outlying parts of the old dialectic districts, would also naturally be then reduced to a common level in consequence of the general mixture of speakers that could not but have taken place on wanderings so extensive as those of the Aryan tribes must have been. Such an assumption would indeed solve most of the difficulties mentioned above, especially the peculiar way in which the single families of Aryan are linked together. Each of these would then correspond to one of the main dialects of the parental language, and their mutual affini ties would therefore be of the same kind as those of neigh bouring dialects, say, of any living speech. And in these nothing is more common, nay even more characteristic, than the gradual transition from one to the other, so that each dialect of an intermediate position partakes of some
