APPENDIX D. 



235 



can only make use of those in which at least two cases of temper 

 are recorded ; they are 146 in number. I have removed all the 

 cases of neutral temper, treating them as if they were non-existent, 

 and dealing only with the remainder that are good or bad. "VVe 

 have next to eliminate the haphazard element. Beginning with 

 Fraternities of tw;> persons only, either of whom is just as likely to 

 be good as bad tempered, there are, as we have already seen, four 

 possible combinations, resulting in the proportions of 1 case of both 

 good, 2 cases one good and one bad, and one case of both bad. I 

 have 42 such Fraternities, and the observed facts are that in 10 

 of them both are good tempered, in 20 one is good and one bad, 

 and in 12 both are bad tempered. Here only a trifling and un- 

 trustworthy difference is found between the observed and the 

 haphazard distribution, the other conditions having neutralised 

 each other. But when we proceed to larger Fraternities the test 

 becomes shrewder, and the trifling difference already observed 

 becomes more marked, and is at length unmistakable. Thus the 

 successive lines of Table III. show a continually increasing diverg- 

 ence between the observed and the haphazard distribution of 

 temper, as the Fraternities increase in size. A compendious com- 



TABLE 3. 



DISTRIBUTION OF TEMPER IN FRATERNITIES. 



