482 



TRANSMISSION 



It is not likely that all characters behave precisely as does 

 stature, indeed, it is known that they do not ; but all studies 

 go to show that characters of every kind obey the same general 

 laws in descent, and all regression tables that have ever been 

 prepared exhibit the same general features. 1 This table, there- 

 fore, while primarily for the study of stature, may be considered 

 as typical of regression tables, and deductions made from it, 

 agreeing as they do with those made from all other similar 

 tables, whatever the race or the character, may be safely 

 accepted as % exhibiting fundamental laws in heredity. Because 

 all regression tables afford the same deductions, they may be 

 stated in the form of general principles as outlined in the 

 following sections. 



SECTION IV LIKE PARENTS BEGET UNLIKE OFFSPRING 



AND, CONVERSELY, LIKE OFFSPRING MAY BE 



BEGOTTEN BY UNLIKE PARENTS 



In this table the heights of the parents are in the rows b to m, 

 and those of the children (as adults) in the columns 2 to 15. It 

 will be seen at once that the offspring of parents of any given 



slight modifications are necessary on account of the parental ancestry, but such 

 corrections are not necessary for present purposes. 



Gallon calls attention to an error in the first row of children and mid-parents, 

 saying that an error was introduced somewhere in the original tables, which cannot 

 now be corrected. " It is obvious that four children cannot have five mid-parents," 

 he says, but the numbers are so small as to be generally discarded, and hence the 

 table is reproduced, error and all. He adds : " The bottom line (fourteen children 

 with one mid-parent), which looks suspicious, is correct " (Natural Inheritance, 

 p. 208). 



In calculations generally the extremes (above 72.5 or 73.2, or below 64.5 or 

 63.2) are discarded because the numbers are small and because exact measure- 

 ments are not given. In calculating the general mean, however, two values have 

 been determined, one without the extremes, the other by including the extremes, 

 assuming that measurements above 73.2 averaged 74.2, above 72.5 averaged 73.5, 

 below 64.5 averaged 63.5, and below 62.2 averaged 61.2. The assumption is 

 entirely gratuitous, but it affords a basis for using the extremes, although, as is 

 noticed, it makes but slight difference in the results. 



1 Regression tables may be prepared for any character that can be measured, 

 weighed, counted, or in any way accurately determined. It only happens that 

 studies in human stature have been the most complete of any, and are, therefore, 

 used here. 



