No. 603] MENDELIAN CLASS FREQUENCY 



151 



We have, by hypothesis, and from the statistics 



A distribution of 58 in a 1 : 2 : 1 ratio is 14.5 : 29 : 14.5. 

 Assume a first sample of 58 to show exactly this distribu- 

 tion 14.5 : 29 : 14.5, what will be the mean frequency of one 

 of the end classes, say white, expected in a second sample 

 of 58? 



We have 



Mean = 14.9833, 



Mode = 14, 

 d= .9833. 



By the approximate method we get 



Median = 14.656 approximately. 

 The standard deviation from (iii), is 

 <r = 4.6364, 



and, by (viii), 



Sk.= .2121. 



Actual tests with curves of a degree of skewness no 

 greater than this show that the approximate method gives 

 the quartile limits with sufficient accuracy for practical 

 purposes. We have for the approximate quartile limits, 

 .67449 X 4.6364 = 3.1272. This value, added to and sub- 

 tracted from the median 14.656, gives the results set down 

 in the last line of Table I. 



Exactly the same procedure, with different numbers, is 

 followed in the case of the blue column. 



6. Let us now consider a more completely worked out 

 illustration. Some time ago Mr. Alexander Weinstein, of 

 Columbia University, consulted the writer in regard to a 



whence, by (i), 

 and, by (ii), 



