THE PROBLEM OF THE ORIGIN OF SPECIES. 45 



Let us follow this series through, step by step, beginning with the uppermost 

 feather of the row. In this feather we find a well-defined half-element of a band 

 crossing the lower web a little obliquely. On the upper web only a mere shadow 

 of a counter half-element is discernible; and this is not discoverable in the juvenal 

 stage and is only slightly strengthened in the adult stage. 



The half-element of the lower web deserves closer examination, as it represents 

 a fully finished transformation. It has an even width of 1.5 mm. and sets off a 

 light apical margin of the feather 5 mm. wide. In so far it comes fully up to the 

 type. It falls a trifle short only in one particular, namely, its inclination from a 

 transverse line is about 23°, which exceeds by at least 4° or 5° the slant of any cor- 

 responding part in front of it. Slight as this deviation may appear, it is significant 

 in the last step of a series that approaches the type gradually in this as well as in 

 all other respects. 



This step is led up to, not only through the bar, at the upper limit of which it 

 stands, but also through all the bands in front of it. To understand its peculiarity, 

 we must therefore approach it from both directions — i.e., from the typical band as 

 well as from the typical element of the bar. 



On the fore part of the wing, where the type is best realized, the two half-elements 

 in each feather coalesce at the shaft-line and together form one straight mark 

 crossing the feather at right angles. This rectilinear form prevails back to the last 

 two or three bands, although a few elements may be found farther forward with a 

 slight curvature; and in such cases the convex side of the element looks forward, 

 and a minute notch is usually seen at the middle, making evident the double 

 nature of the element. 



As we pass backwards, the width of the bands increases very gradually, rising 

 from about 0.5 to 3 or 4 mm. In one out of a dozen cases examined a maximal 

 width of 5 mm. was reached in the fifth feather of the last row. It is in this row 

 that we get the widest elements, the strongest indications of their double nature, 

 and at the same time an inclination in the half-elements that almost reaches the 

 condition seen at the upper extremity of the bar. But these modifications in the 

 band are so small that, viewed as a whole, there is nothing, at first sight, to indicate 

 departure from the type. In the closed wing only the lower web of each feather is 

 visible, the upper web being covered always by the overlapping lower web of the 

 next feather above. The whole band, as thus presented, consists only of the lower 

 half-elements, which flow into each other with such regularity as to mask the slight 

 deviations from the type. For a clear view of all parts of the band the feathers 

 must be plucked and separately mounted in their proper sequence (as in pis. 14 to 

 17). Thus exposed, it will be seen that the width, shape, inequalities, etc., although 

 considerable, are yet so closely graded that the conditions in any one feather flow 

 into those of the next above or below, so that the series may be read in either 

 direction with equal facility. 



It is in the two upper feathers that the conditions approach more nearly those 

 of the typical band, and it is here that we again meet with a slight inclination of 

 the half-elements in the same sense, though not quite to the same degree as we 

 noted in the half-element at the upper end of the bar. This inclination runs through 

 the whole series and is most strongly expressed in the wider elements, which have 

 a crescentic shape, with the horns following the outline of the feather and so curv- 



