VII. FAMILY RESEMBLANCES 



IN the preliminary report of these experiments it was found that 

 the mean variation of the entire group of 90 mice, including mice 

 from all the strains that had been tested up to that date, amounted 

 to 35.6. This means that any mouse picked at random from that 

 group would be likely to vary from the average by 35.6 seconds. In 

 order to find whether mice of the same litter vary less than unrelated 

 individuals, the mean variations for each of the eighteen families 

 was calculated, and these, when weighted for size of family, were 

 found to be 20.2. It was, therefore, stated that the resemblance in 

 behavior between mice belonging to the same litter was nearly twice 

 as great as between unrelated individuals, and that this corresponded 

 to a coefficient of correlation in the neighborhood of 0.5 for brothers, 

 as found by Pearson, Thorndike and others. It is probable, how- 

 ever, that in this case the mean variation was increased for the unre- 

 lated individuals, and the family resemblance correspondingly made 

 to appear greater, due to the fact that the mixed group was made 

 up of two fairly distinct strains of mice, a quick white group and a 

 slow yellow group. 



As previously stated the above calculation was made from the 

 results in only one task, namely, the initial learning period of fifteen 

 trials in the maze test. When the mean variations were calculated 

 for the four distinct tests that were later used, and in which con- 

 siderably more animals were tested, there was apparently no greater 

 resemblance, judged by this method of variation, between animals of 

 the same litter than between unrelated individuals. In order to dis- 

 count the inevitable reduction in mean variation that resulted from 

 the grouping of the individuals into family lots, in which case the 

 mean variations were calculated from the family averages, all the 

 mice that belonged to the family groups were rearranged according 

 to chance into other groups of exactly the same size. Then the mean 

 variations for the groups formed by chance arrangement were calcu- 

 lated in exactly the same manner as was previously done for the true 

 family groups. A comparison of the two results showed a strict 

 similarity in mean variation in each of the four tests; so that the 

 mean variation for the groups of related individuals was not less 

 than, but practically identical with the variation that obtained for 

 the same individuals grouped by a chance arrangement. Later in 

 this thesis, when a detailed study of the family histories is taken up, 

 the records for the various litters will give further light on this sub- 

 ject of family resemblances, and individual variations. 





