48 COMPARATIVE LANGUAGE 



combinations of linguistic phenomena that the language of an in- 

 scription rarely leaves any doubt as to what part of Greece it comes 

 from, provided, of course, it antedates the wivy period? Or will it 

 be objected that we know these dialects only in written form and 

 that the relative uniformity within their limits may be artificial? 

 There may be something in this, and it is not unlikely that there was 

 more merging of one dialect into another near the boundaries than 

 the few inscriptional examples of this would indicate. But from the 

 varied character of the inscriptions, private as well as official, there 

 is no sufficient reason for doubting that we have in general a faith- 

 ful representation of what was actually spoken. And if evidence is 

 demanded of dialects which can be studied in their spoken form, it 

 may be pointed out that, as the whole discussion started with an 

 attack on certain groupings of French dialects, it has been shown 

 by minute investigation that well-defined French dialects do exist, 

 if only one recognize that the boundaries need not be mathematical 

 lines, but may be intermediary zones. 



There can be no doubt that it is the first necessity of dialect-study 

 to define precisely the area of each linguistic phenomenon, as is done 

 in Wencker's Sprachatlas des deutschen Reiches, or on a still more 

 elaborate scale in the Atlas linguistique de la France of Gillie'ron et 

 Edmont, which is to contain some eighteen hundred maps, each 

 showing the pronunciation of some word or phrase in upwards of 

 six hundred places. But it is the legitimate aim of the dialecto- 

 logist, with constant reference to available historical data, to classify 

 such material in larger groups and unfold their history. 



Since dialect relations reflect historical conditions, their evidence 

 may be used in turn to control and supplement imperfect historical 

 data. Nowadays one scarcely hears even echoes of the once lively 

 discussion of wave-theory versus Stammbaum-theory, for it is tacitly 

 recognized that there is truth in each. The difference is only one of 

 chronological emphasis, if I may so express it. There is no doubt 

 that points of agreement between dialects, so far as they are not 

 accidental, that is, due to independent development in each, are 

 significant of geographical continuity, at some time. But this 

 may be the geographical continuity of the historical period, and this 

 is what was emphasized by J. Schmidt in his famous work; or it may 

 be that of a prehistoric period, and this is what is emphasized by 

 a tree-scheme, which is intended to illustrate how dialects or lan- 

 guages have diverged from a common prehistoric source. One may 

 object to specific tree-schemes as arbitrary, and certainly the attack 

 on existing tree-schemes of the Indo-European languages was entirely 

 justified. One may dispute in each case as to how far it is possible 

 to go in such a scheme. But one cannot doubt the existence of 



