92 THE FEEBI^Y INHIBITED. 



having sharp tempers under strong but insecure control, so that they 

 'flare up' now and then." As a matter of fact, it is commonly recog- 

 nized that temperament is largely inborn and is prevailingly determined 

 by heredity. 



Third, admitting the hereditary factor in temperament, various 

 hypotheses are tenable. Galton's (1899, p. 235) statistical discussion 

 is not interpretive and can not be translated into modem biological 

 language. I sought originally to fit a simpler set of factors to the facts 

 than those here employed, but nothing simpler would meet the condi- 

 tions satisfactorily. But perhaps our hypothesis is too simple; there 

 may be more hereditary factors than I predicate. There is an a priori 

 probability in this hypothesis. Let us consider it. 



If there is only one gametic factor for excitation, B (its allelomorph 

 being e) , then when both parents are simplex for this factor the factors 

 should appear in the children in the proportion of i duplex, 2 simplex, 

 and I nuUiplex. An inspection of classes 15, 16, 17, 21, and 22 shows 

 that of the offspring there are 47 duplex, 115 simplex, and 76 nulliplex, 

 or nearly the ratio of i : 2 : i| or 68 per cent with the excitation factor 

 and 31.9 per cent without; this is a considerable departure from the 

 expected 3:1; being about 2:1. 



Similarly, if there is only one gametic factor for C (c being its allelo- 

 morph) in each parent, we should expect the condition of C2, Cc, and 

 C2 in the offspring to be in the ratio of i : 2 : i , or C to no C in the ratio 

 of 3:1. Actually we get 91:110:73, or 73.3 per cent to 26.6 per cent, 

 or close to the ratio of 3 : i . In some respects this is a better fit than 

 with the excitation factor but in other respects not so good. 



In the matings C2 X Cc we expect 50 per cent normal duplex and 50 

 per cent simplex; actually the numbers are 89 cheerful to 45 phleg- 

 matic; but doubtless many "cheerful" are reaUy simplex, but appear 

 as normal because of complete dominance. In the matings CcXc2 

 the ratio is 79 Cc: 68 C2, which is fairly close to the expected i : i ratio. 

 If we combine the "back-crosses" of simplex-excited on duplex-excited 

 and on nuUiplex-excited, we get an approach to the ratio of equality. 



Now, although in many of these tests the observed ratio is not close 

 to expectation, yet this ratio is more closely approached than any 

 other simple ratio. Thus the ratio e to E and c to C in F2 never 

 approaches i in 16. It is fairly close to i in 4 {i. e., 1:3). 



While I do not wish to insist that the hypothesis which fits so sur- 

 prisingly well tells the whole story of the hereditary factors involved 

 in temperament (and the whole story can be known only after much 

 more and more analytical data have been analyzed) , still it can not be 

 denied that it fits the facts approximately, and no other hypothesis 

 that has been tested serves anything like so well. 



