THE AMERICAN NATURALIST [Vol. XLIII 



As the diagram shows, individuals of the same size are 

 found in many different races. An individual of the size 

 shown at a might belong to any one of the eight races. 

 Thus from individuals of the same size we may get many 

 different results in breeding. Similarly, individuals of 

 very different size (as the extremes of any horizontal 

 series) may produce progeny of identically the same char- 

 acter. We can not tell by its external characters to what 

 race a specimen belongs ; breeding is the only test. In 

 higher organisms, as is well known, we often find a sim- 

 ilar state of affairs. 



How will selection act on such a complex species! ' As 

 we have seen, selection within a single race is without 

 effect. But if we make selections among the individuals 

 of a mixed collection of races, such as Fig. 5 shows, we 

 reach most instructive results. By making our selections 

 in the proper way, we for a time make steady progress 

 toward a certain goal. We will suppose that we do not 

 know of the existence of these races ; this is the case with 

 most experiments in selection. From the species as a 

 whole, as shown in Fig. 5, we will select for increased 

 size. Let us follow the old plan of selecting many indi- 

 viduals showing the desired character; we will preserve 

 all specimens above the mean size of the entire collection. 

 That is, we divide the collection at x — x rejecting all 

 those to the right. By so doing it is evident that we 

 exclude all specimens of the two smallest races c and i, 

 while preserving the majority of the specimens of the 

 larger races. Allowing these to propagate, we of course 

 get a mixture of the remaining larger races; hence the 

 mean size of the whole collection will be greater than at 

 first. Selecting again those above the mean size of this 

 lot, we drop out another small race, and the mean of the 

 collection as a whole again rises a little. We are making 

 good progress in the improvement of our species. By 

 taking successive steps of this character, dropping out 

 the smaller races, first partly, then completely, one after 

 another, we can for a long time continue to improve by 



