﻿1 6 Darbishire, Mendelian and Biometric Theories. 



have these words from the champion of Mendelism in 

 this country * : — " In the Mendelian method of experi- 

 ment the one essential is that the posterity of each 

 individual should be traced separately. If individuals 

 from necessity are treated collectively, it must be proved 

 that their composition is identical. In direct contradiction 

 to the methods of current statistics, Mendel saw by sure 

 penetration that masses must be avoided." 



There is one direction in which my parallel may seem 

 at first sight to be incapable of being pushed very far ; it 

 may be urged that we never can have any knowledge of 

 the individual, but only of kinds of individuals, because 

 in single cases it is impossible to eliminate the attributes 

 which are due to chance : so that Mendelian methods are 

 more to be compared with chemistry, which tests the 

 property, not of units, but of masses of units which are 

 known to be all the same.f But this fact does not 

 in the least lessen the closeness of the parallel, for we 

 have no reason to believe that the demon, if his 

 attributes are as " essentially finite as our own," 

 would have or need any knowledge of the individual 

 molecules, but merely the ability to classify them into — 

 say — ten classes, ranging from very fast to very slow, and 

 to close his door according to their speed and direction. 

 This point does not seem to be of much importance, but I 

 did not wish, by not referring to it, to appear to have 

 overlooked it. 



One result, which seems to me to follow naturally 

 from the truth of my comparison, is that it is unreason- 

 able to apply, as has often been done, the criteria of 



* Presidential Address to Section D, Brit. Assoc, 1904, in Nature, 

 August 25, p. 409. 



t Cf. "The breeding-pen is to us what the test-tube is to the chemist" — 

 same Address, p. 409, 1st column; and cf. Reports to EvoL Coin. Royal Soc, 

 I., p. 159. 



