ch. ix.] SPECIAL-CREATION OR DERIVATION 1 447 



change into the Latin lacru-ma, and thence into the Ficrau 

 larme ; the North Aryan form has undergone another kind of 

 change into the Old German tagr, and thence into the 

 English tear. 



Thus in general, as we go backward in time, we find the 

 lines of linguistic development drawing together. Between 

 the various Low-Dutch dialects spoken along the north coast 

 of German v, the differences are hardly great enough to inter- 

 fere with mutual intelligibility. Again, between Portuguese 

 and Spanish the differences are so small that one who is well 

 acquainted with Spanish can often get the sense of many 

 pages in a Portuguese book without having specially studied 

 the latter language. But German and Spanish have few 

 mutually intelligible words in common, and their differences 

 in idioms and in structure of sentences are no less con- 

 spicuous. While it might be possible to maintain that Dutch 

 and Platt-Deutsch, or that Portuguese and Spanish, are only 

 dialects of the same language, no one would hesitate about call- 

 ing Teutonic and Eomance quite different forms of language. 

 Yet we need only go back far enough to find the demar- 

 cation quite as obscure in the one case as in the other ; for 

 Teutonic and Romance began as the northern and southern 

 dialects of the same Old Aryan language. In similar wise 

 we may say that, even with the keenest linguistic instinct, it 

 would be difficult to decipher a line of modern Persian by 

 reason of its kinship with modern Greek ; while yet it is 

 undeniable that the Persian spoken by the officers of Xerxes 

 was strikingly similar to the Greek spoken by Demaratos 

 and Leonidas. 



In citing this example from the phenomena of language, I 

 do not cite it as direct testimony in favour of the theory of 

 derivation in biology. Because tear and larme can be traced 

 back to a common form, it does not follow that the pig and 

 the horse have a common ancestor. Yet, wdiile the linguistic 

 parallel is by no means available as direct testimony in a 



