SYSTEM. 145 



ture capable of retaining the language within its limits, and of 

 preserving it from all separation. 



Hence it results that anthropology must take its most pre- 

 cious authorities from the study of languages, in the language 

 of the islands, for instance, and in the idioms spoken at the 

 extremity of the continents : thus surrounded by the sea, in 

 relation by its less extent with the others, these idioms will be 

 preserved even more intact. We shall find here the real ex- 

 pression of the most ancient state of things which we can 

 directly recognise by philology. The click language, so pecu- 

 liar to one single race,* exists only in the most southern part 

 of Africa. They still speak the ancient Pali-f in the south . 

 of Asia and at Ceylon. The most ancient language of Europe, 

 so far as we know, namely the Celtic, still remains in Britanny , 

 and in Wales. 



From all which has gone before, we may then conclude that 

 in order to establish a rational classification of human species, 

 the first characteristics to be considered will be the external 

 aspect, and perhaps the moral characteristics ; the rest will come I 

 in the second rank : at first, language, then deep anatomical 

 varieties which do not strike us at the first glance, then physio- 

 logical and pathological varieties, etc. Such is, we think, the 

 only certain basis upon which anthropology can rest the true 

 distinctions between human species. We do not even yet 

 know exactly their number, and naturalists do not at all agree 

 on this subject ; the work is to be done over again, by follow- 

 ing a new route. 



Without troubling ourselves with enquiring into the whole 

 system of the genus homo, we must at first examine these well- 

 characterised centres of population which are entirely distinct 

 as regards aspect and physiognomy. We must mark these 

 centres with care, paying attention to all the physical, moral, and 

 philological varieties which we are able to notice. M. Flourens 

 has given some excellent principles for the study of animal 

 species ; he wishes simply to apply them to the study of human 



* See above, p. 78. 



f [Pali, the ordinary language of daily life in Hindoostan at the time when 

 Sanscrit was used in elevated literature alone. EDITOR.] 



