APPENDIX. 331 



manifest form. It is disguised in several ways. Sometimes, 

 as in the Indo-Em'opean tongues, there is one root for the 

 nominative and one for the oblique cases ; sometimes the 

 same form, as in the Finlandic, runs through the whole 

 declension; sometimes, as when we say you for tliou in 

 English, one number is substituted for another ; and some- 

 times, as when the German says sie for thou, a change of 

 the person is made as well. When languages are known 

 in detail, these compHcations can be guarded against ; but 

 where the tongue is but imperfectly exhibited a special 

 analysis becomes requisite. 



Generally, the first person is more constant than the 

 second, and the second than the thu'd ; indeed, the third is 

 frequently no true personal pronoun at all, but a demon- 

 strative employed to express the person or thing spoken of 

 as the agent or object to a verb. Now, as there are fre- 

 quently more demonstratives than one which can be used 

 in a personal sense, two languages may be, in reality, very 

 closely allied, though their personal pronouns of the third 

 person differ. Thus the Latin e^o=£7(o ; but the Latin 

 hie and ille by no means correspond in form with bq, avro, 

 and iKHvog. This must prepare us for not expecting a 

 greater amount of resemblance between the Australian 

 personal pronouns than really exists. 



Beginning with the most inconstant of the three pro- 

 nouns, viz., that of the third person, w.e find in the Kowi'a- 

 rega the following forms : — 



3. 



Singular, masculine nu-du=^he, him. 



feminine na-du^she, her. 



Dual, common . . pale^^they two, them two. 



Plural, . . tana^=^they , them. 



In the two first of these forms the da is no part of the 

 root, but an affix, since the Gudang gives us the simpler 



