RELATIONS OF COMPARATIVE GRAMMAR 49 



migratory movements such as are properly represented by a tree- 

 scheme, or that such movements often reflect themselves in dialect 

 relations in an unmistakable fashion. Let me illustrate from the 

 Greek dialects. If we survey the whole body of linguistic phenomena 

 we may divide the points of dialect agreement into three classes. 

 Some we regard as accidental. Others are significant of geographical 

 continuity in their historic positions, as probably the psilosis on 

 the coast of Asia Minor in which the vEolic, Ionic, and Doric dialects 

 of this region share. Others are obviously significant of geographical 

 continuity in a period preceding the great migrations, and there of 

 course are the points upon which are based all attempts to classify 

 the dialects and stems. No one can possibly doubt the historical 

 significance of the agreement in features not found elsewhere between 

 Arcadian and the remote Cyprian, between Asiatic JSolic and 

 Thessalian, or of the mixture of Doric and JEolic characteristics in 

 Thessalian and Boeotian. And I have no hesitation in asserting 

 that those historians, fortunately few, who regard the tradition of the 

 Doric migration as a pure myth, either have no first-hand knowledge 

 of the dialects or are absolutely impervious to linguistic evidence. 

 There is enough that is still obscure in the relations of the Greek 

 dialects, but there is also much that is as clear as day. 



It may be said of this or any other like case that it is arbitrary to 

 regard certain points of agreement as accidental and others as sig- 

 nificant, and that in combining the latter with vague traditional data 

 and then drawing historical conclusions we are guilty of reasoning 

 in a vicious circle. Perfectly true. But where is there a branch of 

 inquiry in which the so-called vicious circle is not employed, and jus- 

 tified too, if only the circle is completed without undue pressure? 

 When a number of linguistic facts fit together with one another and 

 with traditional data, which in itself may be of little weight, we are 

 entitled to regard them as significant. 



I can only allude to the historical significance of borrowed words 

 not due to any racial mixture, such as the Greek words in Latin which 

 bring before us the successive periods of Greek influence: first, the 

 remote period -when certain articles of commerce were brought to 

 Italy by Greek mariners, then the influence exerted by the Greek 

 colonists of Magna Grsecia, then the time when educated Romans 

 were familiar with Greek literature and sent their sons to Athens for 

 study, and lastly the period when Rome was filled with Greek- 

 speaking slaves. Or, to take an example of a totally different and 

 less usual character, the words which the Gypsies have adopted 

 from the various languages with which they have come in contact 

 since leaving their home in India, some of them, like the Armenian 

 and Modern Greek words, common to all dialects and so significant 



