RELATIONS OF COMPARATIVE GRAMMAR 51 



unknown, since the old word may for various reasons have been lost 

 or replaced, just as the old words for brother and sister have been 

 replaced in Greek, those for son and daughter in Latin; further, 

 that agreement in a given word is not always proof of its existence 

 in the parent speech, since, aside from the possibility of independent 

 formation, this agreement may rest on a succession of borrowings, 

 as is the case with the word for wine; lastly, even where the existence 

 of the word in the parent speech is not open to question, its precise 

 meaning may be uncertain. From the series, Sanskrit ayas, Avestan 

 ayah-, Latin aes, Gothic aiz (English ore), which in different times 

 and places mean copper, bronze, iron, or metal in general, we can, 

 indeed, infer that the Indo-Europeans were acquainted with some 

 metal, but when we conclude that this was probably copper, we do 

 so on other than purely linguistic grounds. Furthermore there are 

 countless points upon which linguistic evidence is altogether silent. 

 But when the skepticism is carried so far as to assert that no value, 

 or at the most very slight value, is to be attached to linguistic evidence, 1 

 this can only be stamped as an unwarranted exaggeration. The 

 elimination of borrowed words from apparent cases of agreement 

 has long been recognized as an important corrective. But it is a 

 mere splitting of hairs to urge that all cases of agreement may rest 

 upon borrowing, only in the remote period when the later Indo- 

 European languages, though already somewhat differentiated, were 

 still spoken in contiguous territory. No exception need be taken to 

 such a statement if intended only as a warning that the conclusions 

 reached may not be applicable to precisely the same period and that 

 the combination of the various conclusions may not be truly homo- 

 geneous. The same is true of the reconstructed forms, and I would 

 emphasize again what was said in reference to the parent speech, 

 that we are concerned with it not so much for any intrinsic interest 

 it possesses for us as for its bearing on later development. If we are 

 able to trace a given institution back to a period before the bonds 

 between the Indo-European peoples were severed and antedating 

 the more individual development of each in the land of its per- 

 manent home, what more do we ask? We may deny the application 

 of linguistic evidence in individual cases, but not in principle. It 

 must be used with caution, but the danger of its abuse is not greater 

 than is the case with archaeological evidence. Often it fails us entirely, 

 but often it is, in the nature of things, the only available evidence. 

 What archaeological evidence can tell us how far the numeral system 

 was developed, or can throw such light on the family organization 



1 I refer especially to the' radical position taken by Krestcher, Einleitung in die 

 Geschichte der griechischen Sprache, p. 48 ff., in the criticism of which I am in 

 entire accord with the remarks of Schrader, Reallexikon der indogermanischen 

 Altertiimen, p. 8 ff. 



