1898-99. | THE ORIGIN OF GENDER. 71 
does not mention, seems to me to have been one of primary importance, 
on which was founded the whole structure of Indo-European society. 
The head of the family, the free member of the tribe, was felt to be the 
persona, the only person who could act independently; and all the rest 
of the famzlia, his wife, children, slaves and cattle were simply his 
property. Primarily the names of fersone were the only masculines 
and all others were neuter. But presently his sons, his freedmen or 
favorite slaves, his horse, his sword, the heavenly bodies, the mountains, 
the grandest and most striking natural phenomena, were personified and 
the bounds of the masculine gender accordingly extended. 
It may be asked whether an examination of the usages found in 
languages not belonging to the Indo-European group, some of them 
very little developed in grammatical structure, throw some light on this 
question. I have no immediate knowledge of any of these languages 
and for details I venture to give here I am indebted to Winkler’s 
Grammatisches Geschlecht.* According to Winkler, many of the 
American languages do not distinguish gender at all. Iroquois dis- 
tinguishes two classes of objects, a higher and a lower; to the higher 
belong all supernatural beings and men ; to the lower, women, animals 
and lifeless objects. Of the languages of Asia, the Ural-Altaic groups 
have no genders. But the languages of Burmah and Thibet show a 
distinction into two classes, which in the Kassir language exactly 
resembles that found in Iroquois. Where nouns are thus divided into 
two classes, I may add, the only mark of distinction in the noun itself is, 
that nouns of the higher class form a plural, while those of the lower do 
not. Generally speaking, in the languages of America and Asia, to the 
stage of development which presents no distinction of gender, succeeds 
a stage where the distinction corresponds so closely with what I have 
assumed as an older division in Indo-European, that the distinction 
between fersona and res may be supposed to be essentially that which 
lies at the base of all primitive society. An examination of the 
languages of America and Asia, and of certain African languages, such 
as the Fulde, where grown men are distinguished from all other objects, 
seems to confirm very strongly the probability of a stage of development 
such as I have assumed in the Indo-European languages. 
For Winkler finds that the languages of Asia, Africa and America 
present three types as regards gender, indicating three stages of 
development. We have, first, languages that show no variation for 
gender ; second, languages that have a twofold division for men and 
lower objects ; third, languages that distinguish males and females, and 
*Berlin, 1889. 
