94 JOURNAL AND PROCEEDINGS 



ing from Latin. I do not know of any unshifted forms, known to 

 be Germanic, from which we might learn of a time before the gene- 

 ral spread of the changes, and so determine roughly the date of its 

 beginning. As to the close of the period, there was no shifting 

 when the words apostol, candel, pawa were introduced into Anglo- 

 Saxon in the sixth or seventh century after Christ. 



The second shifting remains to be considered. A reference to 

 the illustrations given above will show that only in the dental series 

 is the shifting complete in Modern German ; / has become/; b, f, 

 g, k, h remain unchanged. Grimm cited Old High German forms 

 such as prechan, pirn, kans, chuni to show the completeness of the 

 circle of change ; but even in the earlier periods of the language 

 these are found alongside unshifted forms. The shiftings varied in 

 different parts, and were most thorough in Upper Germany. To 

 our partial statement of the law given on page 91 we may add gene- 

 rally that the subsequent shifting was similar, but varied in extent in 

 the several districts, modern literary German exhibiting a general 

 change only iti the de?ital series. 



We can very well determine the date of this movement. It took 

 place as Sayce''' points out, at a time subsequent to the wars between 

 the Teutons and Romans which led to the overthrow of the Roman 

 Empire ; thus {via^ strata and campus, our street and camp, became 

 in Old High German straza and kamph, chanipf. The year 600 of 

 our era is assumed to mark fairly well the setting in of the process, 

 but its progress was slow ; as late as 842 we meet dag (A. S. daeg, 

 G. tag), and it is interesting, in connection with Mr. Sweet's re- 

 marks quoted on page 100, to notice that godes (G. gottes) existed 

 then beside the nominative got. 



It is customary to speak of " exceptions " to Grimm's law, and 

 and here may be set down explanations of some forms which the 

 law as stated fails to account for. 



I. Onomatopoetic and imitative words at once suggest them- 

 selves. The sounds of the parent language remain practically un- 

 changed in Gr. klagge (gg = ngg), L. clangor, E. clang, G. klang, 

 and like words, for they are all imitations of like natural sounds. 

 With these may be grouped such words as Gr. pappa, L., E. and 

 G. papa, which are merely doublings of the earliest formed sounds 



(i) Introduction I., 307. 



