9 7 J. " Clark, The Classification of Bitds. loct. 



ment of their own characteristic pattern. A similar illustration is 

 found in a comparison of the embryos of a Rail {Ralhis longiros- 

 tris) with those of a Heron {Ardea tricolor) which shows that the 

 two, just before hatching, have very similar pteryloses, which are 

 distinctly heronlike. A more important point shown by the 

 Heron embryos, is that the powder-down tracts are a more recent 

 acquirement than the pattern of the pterylosis, for only one of the 

 three pairs is indicated at all, and the presence of that pair is 

 shown only by the peculiar color and appearance of the skin. 

 The pair present is the femoral. I thought I could find indica- 

 tions of the ventral pair, but those near the furcula are entirely 

 wanting. 



These facts, though few in number, seem to me to indicate very 

 strongly that in the pattern of the pterylosis we have a character 

 which has changed but slowly, and is liable to little variation, and 

 is therefore of primary importance in seeking the proper classifi- 

 cation of birds. Indeed there is little reason why the general pat- 

 tern should change, for necessary differences in the thickness of 

 the feather-coat would naturally be brought about simply by 

 widening or narrowing the main tracts. That this is the case is 

 readily seen by comparing the tracts of a Goose and a Petrel, 

 where the pattern is essentially the same, but the tracts of the 

 more active bird are much narrower. In the taxonomy of the 

 Crustacea, a class characterized by a segmented external skeleton 

 and jointed biramose appendages, the division into orders is 

 based on the number and arrangement of segments and append- 

 ages ; and in other classes of animals the primary divisions are 

 based on variations in the principal character of the class. It is 

 both fitting and desirable, therefore, that the great class Aves 

 should have its orders based on variation in its striking character, 

 — the body covering of feathers. Such orders would be clearly 

 equivalent to the orders of other groups. 



But is it possible to adopt any such standard and arrive at any 

 definite results in the classification of birds ? After a careful study 

 of Nitzsch's work, and a review of my own in connection with it 

 I find there are eight distinct, and, in general, easily recognized 

 patterns of pterylosis in the class of birds. There are, besides, 

 two very distinct groups of birds which have no apteria, but have 



