APPENDIX. 341 



The comparative absence, however, of the sound of 5, in 

 Australian, may be further refined on in another way ; and 

 it may be urged that it is absent, not because it has never 

 been developed, or called into existence, but because it has 

 ceased to exist. In the Latin of the Augustan age as 

 compared with that of the early Republic, we find the s of 

 words like arbos changed into r {arbor) . The old High 

 Grerman, also, and the Icelandic, as compared with the 

 Meso-Gothic, does the same. Still the change only affects 

 certain inflectional syllables, so that the original s being 

 only partially displaced, retains its place in the language, 

 although it occurs in fewer words. In Australian, where it 

 is wanting at all, it is wanting in toto: and this is a reason 

 for believing that its absence is referrible to non-develop- 

 ment rather than to displacement. For reasons too lengthy 

 to exhibit, I beheve that this latter view is not applicable 

 to Australian; the s, when wanting, being undeveloped. 

 In either case, however, the phonetic diff'erences between 

 particular dialects are the measures of but slight dif- 

 ferences. 



Now — with these preliminary cautions against the over- 

 valuation of apparent diff'erences — we may compare the 

 new data for the structure of the Kowrarega and Limba- 

 karajia Avith the received opinions respecting the Austra- 

 lian grammars in general. 



These refer them to the class of agglutinate tongues, 

 i. e. tongues wherein the inflections can be shewn to con- 

 sist of separate words more or less incorporated or amal- 

 gamated with the roots which they modify. It may be 

 said that this view is confirmed rather than impugned. 



Now, what applies to the Australian grammars apphes 

 also to Polynesian and the more highly- developed Malay 

 languages, — such as the Tagala of the Philippines, for in- 

 stance; and, if such being the case, no difference of 



