1898-99. | THE ORIGIN OF GENDER. 67 
That the highly artificial system of genders found in Greek and Latin 
was not mainly a development in each of these languages, but that. it 
can in great measure be traced back to the original Indo-European 
language, has been clearly shown by Delbriick in his Grundlagen der 
Griechischen Syntax, pp. 4and 5. But his further examination of the 
correspondence between the classes of words in Greek or Latin that are 
masculine or feminine by meaning with the corresponding classes in 
Sanskrit serves to show how little a comparison of these languages can 
teach us here. Names of trees, for example, which in Greek or Latin 
are mostly feminine, follow the analogy of %y and sz/va. But in 
Sanskrit they are mainly masculine. One curious correspondence he 
notes, is that, just as in Greek and Latin, while the names of trees are 
feminine, their fruit is denoted by a cognate substantive in the neuter, 
ex. gr. wyAid and pijhov, xepacta and zepdowy, Latin perus and pzrum, 
jicus and ficum, so in Sanskrit amra, m., is the mango-tree, and amram, 
n., its fruit. 
With regard to endings, we may adopt Brugmann’s rule: All stems 
in 9 are originally masculine, all in @ feminine ; others are primarily 
indifferent. Masculines in ¢s in Greek and a in Latin are derived from 
corresponding feminine abstracts in a, as in veatas and agricola. The 
Latin does not usually add the personal ending s here, though Festus 
gives the forms hostzcapas and parricidas. 
Now it is clear from what has been said already, that the artificial 
systems of gender found in Greek and Latin represent an advanced 
stage in the development of grammatical gender ; for very many names 
of lifeless objects in both these languages are masculine or feminine ; in 
both most abstract nouns are feminine; and according to both the 
theories already outlined, which make sex the basis of gender, all these 
must have been originally neuter. In Greek and Latin neuter nouns 
are in a minority, when compared with either masculines or feminines. 
Hermes in a programme Ueber das grammatische Genus* gives the 
following table of the comparative number of masculine, feminine and 
neuter substantives in Latin, Greek and Sanskrit : 
Masculine. Feminine. Neuter. 
EAT eg are Rte cote Aare eine te pe ee 10 1] 6 
(Cs eeved fe Bete peas boas Pade ake Pe a ae re 10 12 6 
ansSknit. ss wacenta case 2s cata: : ike) 5 8 
This table shows that in Sanskrit the number of feminines is much 
smaller and the number of neuters considerably larger than in Greek or 
Latin, and indicates a gradual change from neuters to feminines. But 
we have signs, too, in Greek and Latin that neuters are tending to 
“Berlin, 1851. 
