24 



THE AMERICAN NATURALIST [Vol. LII 



closer scrutiny shows, however, that the large feather is 

 ragged along one edge only (or on a part of one edge), 

 while the smaller piece has also on the same side (as can 

 be seen in some cases at least) a ragged edge with the 

 other vane more nearly complete and with a not-rough 

 edge. It seems, therefore, more reasonable to interpret 

 even these cases as extremes of the split-feather type in 

 which one piece has fared worse than the other (or in 

 which the original division was into unequal pieces). 



The Size of the Double Feathers 

 There is a graded difference between the outer and 

 inner vanes of the feather from the edge to the middle 

 of the tail Fig. 15. The outer half of the vane is rela- 

 tively smaller in the outermost feather, right or left, and 

 equality of the two sides is more and more reached, the 

 two middle feathers of the 12-feathered tails are about 

 symmetrical. In the multiple feather tail these relations 

 still hold, but are more difficult to trace than when the 

 tail is simpler. It would not be profitable to attempt to 

 analyze in detail these relations as applied to the double 

 feathers further than to compare their surface relations 

 with that of the feathers nearest to them or with their 

 symmetrical mates. In all cases of split feathers the 

 outer halves of the vanes are not so wide as is expected 

 from the nearest feathers (or their symmetrical mates as 

 seen in Figs. 14-16). The middle part is, as a rule, very 

 much less than a right or left vane. The total width of 

 the split feather is, as nearly as I can judge, about the 

 same as the expected feathers for that position. The im- 

 pression indicates that the sum of the four vanes is a little 

 greater than the sum of the two normal vanes, but there 

 can not be much difference as measurements show. The 

 looseness of the frayed inner edge makes it difficult to get 

 a very close estimate of the actual relations. 



The general conclusion is that we are dealing with a 

 single rudiment that has split at a very early stage into 

 two parts that have completed themselves as whole 



