Manchester Memoirs, Vol. I. (1906), A*?. 11. 29 



metrician to test his own particular Law ; for the stringency 

 with which the mice were selected in the previous part of 

 the experiment rendered the results useless for anyone 

 who wished to test the Law of Ancestral Inheritance by 

 them. 



What was wanted was some device to ensure the 

 random mating of the mice, and, at the same time to 

 ensure the possibility of tracing all the ancestors and all 

 the offspring, in fact, all the relations of every degree of 

 every individual mouse ; the second condition had been 

 fulfilled in the previous part of my experiment ; but the 

 first had not, because the different kinds of mice had been 

 rigidly selected. 



The method by which I mated the mice at random 

 was very simple. I wrote the catalogue-name of each 

 mouse on a counter ; then I put the counters representing 

 female mice into one hat and those representing males 

 into another : all that remained to be done was to draw 

 out at random a counter from the ' female ' hat and 

 similarly one from the ' male ' hat, and to mate the actual 

 mice represented by these counters. As I have said, this 

 method enables one to test the Law of Ancestral Inheri- 

 tance and Mendel's Law. As far as the first is concerned 

 it is the most perfect conceivable ; but for the second it is 

 clumsy and involves unnecessary labour : because what is 

 aimed at in a Mendelian experiment is the discovery of 

 the properties of character-units, as far as they can be 

 discovered by determining the specific results of their 

 union with similar and dissimilar character-units. Now, 

 some particular combination of characters may turn up 

 very seldom by the method of random union ; and if one 

 wishes to discover the result of such a combination one 

 has to wait until the drawings from the hat give it. One is 

 in the position of an observer, and if one wishes to 



