lOO JOURNAL AND PROCEEDINGS 



articulation. " In some districts " says Brugmann '"' " espec- 

 ially in Attica the sounds appear to have been spoken for a 

 long time with aspiration. Yet from the inexactness and 

 inconsistency of the written representation it is not possible 

 to determine accurately the extent of the tendency." In the 

 Umbrian and Oscan dialects in Italy kt and pf became ht 

 and//, as Umbr. rehte^ L. rede; Osc. Uhtavis, L. Odavius. 

 In Old Irish after vowels / and c became the spirants th, di^ 

 as niaihii'^ L. niader, E. viother^ etc. 



(b) In Armenian medi» became tenues, as tiv. Skt div, L. dies. 



(c) In Iranian, Old Irish and Balto-Slavic, for the most part in 



Armenian, and medially in Latin, aspirated mediae became 

 simple medise, as Lat. medius, O. Ir. ?nedon, Skt. madhya. 

 Similar shiftings in Greek have been noticed above (p. 95). 

 The Germanic shiftings may be due to such causes, whatever they 

 were, as produced the changes just noticed, only that the former be- 

 came general probably by the working of analogy. " Most sound- 

 changes," remarks Sweet '''' " seem to begin under special circumstan- 

 ces, and if they do extend themselves over the whole range of the 

 sound in question, it is only gradually. A change such as that of 

 d into / may begin at the end of a breath-group, and be then ex- 

 tended to the ends of words within a breath-group, as in German, 

 and finally to all the ^'s in the language, as when every Arian d be- 

 came a / in Germanic." In the Germanic, with its want of a litera- 

 ture, or, as Canon Taylor would rather say, because of the want of a 

 literature in the language adopted by the Germans, there was no 

 language-tradition strong enough to secure unshifted forms against 

 the influence of analogy ; while on the other hand the people of 

 India, Persia, Greece and Rome used a language which was in a 

 manner fixed in literature. 



The last word has not been said on Grimm's law. Enough 

 general statements have been made concerning it, and numberless 

 conclusions made regarding the early relations of the peoples with 

 whose languages it has to do ; but how few of these are trustworthy ! 

 Stricter methods, it is to be hoped, will give more reliable results. 

 Especially a better determination of the nature of the sounds in ques- 

 tion, and a careful study of the High-German shifting and of modern 

 German dialects may make clear many things that are yet obscure. 



(i) Grundriss I. Sect, 4S6. 



(2) History of English Sounds, p. 16. 



