50 PHYLOGENY 



seen that it is hopeless to try to advance in the opposite direction, 

 from the bars forward to the checkered condition. No variations 

 will appear in that direction, but such as do appear will take the 

 opposite direction, tending to diminish the width of the bars and to 

 weaken their color. It is in this way that we must account for the 

 existence of some fancy breeds in which the bars have been wholly 

 obliterated. The direction of evolution can never be reversed. 



I have tried both experiments for eight years, and as both tell the 

 same story as to the direction of variation, I am satisfied that further 

 experiments will not essentially modify the results. 



With reduction traveling from before backward, in the manner 

 described, we get the bars in their typical number, form, and position, 

 as one of the necessary stages of the process, and without appealing to 

 de novo origin, incipient rudiments, etc. 



But if bars originated in such simple fashion, the direction of 

 evolution being precisely the same as that of embryological development, 

 if the theory of rudiments must be abandoned in this case, do we not 

 meet the same theory again in any attempt to account for the 

 checkers? 



What kind of rudiments could be imagined? We might assume 

 that minute flecks of pigment first appeared, one in each feather; 

 and then, further, imagine that these purely chance originations 

 happened to have some slight utility, and that natural selection did 

 the rest. But it is just as difficult to account for a small as a large 

 origin de novo, and the smaller it is, the more unfortunate it is for 

 the theory of natural selection. 



If we seek refuge in the doctrine of mutation, are we better off? 

 Mutation hides itself in the undiscoverable premutation, and so we 

 have all the difficulty of an incipient stage, and no means of advanc- 

 ing by ordinary variation. 



Fortunately we are not driven to either alternative, for the 

 checkers arise neither as mutations nor as rudiments, but by direct 

 and gradual modification of an earlier ancestral mark, which came 

 with the birth of the pigeon phylum, as a heritage from still more 

 distant avian ancestors. 



This ancestral mark is a dark spot filling the whole central part 

 of the feather, leaving only a narrow distal edge of a lighter color. 

 This mark is still well preserved in some of the old-world turtle- 

 doves best in the Oriental turtle-dove of China and Japan. The 

 checker of C. lima differs from the dark centre of T. orientalis 

 only in form and in having a lateral position. Typically this spot 

 appears in pairs, one on each side of the feather. The two spots 

 represent the two halves of the old central spot, which becomes 

 more or less deeply divided by the disappearance of pigment along 

 the shaft of the feather. This change begins at the tip of the feather 



