THE ORIGIN AND RELATIONSHIP OF THE ROCK-PIGEONS. 61 



pigeons, never appears in chequered dress. It is moving in the other direction, and 

 no reversal of course is now open to it. Moreover, the various conditions of chequers 

 and bars in rock-pigeons and domestic races are self-explanatory from the same 

 point of view. 



We could not explain how two bars could arise de novo in a clear gray wing- 

 surface; but Ave can see how a sweeping reduction process, antero-posterior in 

 direction, would leave two or more rows of chequers cut to dimensions that would 

 coalesce in transverse bars at the posterior end of the wing. The great variations 

 in the width of these bars, the serrate edge of the wider. bars, the more eventy cut 

 edge in narrow bars, the presence of one or two rudimentary bars, wholly concealed 

 from view, in short, all the peculiarities of the wing-pattern, become intelligible as 

 soon as we discover the nature and drift of variation. 



To attempt to explain all this as the work of natural selection would lead into 

 an endless tangle of conjecture that would leave even the simplest facts as unap- 

 proachable mysteries. Natural selection has probably had much to do with 

 the end-stages in the evolution of characters, but little or no direct influence in 

 originating them. 



The two-barred condition has been reached in the simplest possible way, not 

 by accidental variation of chance mutation, but by progressive modification of a 

 chequered condition previously established. The long coverts and the secondaries, 

 which are the larger of the chequered feathers, have the larger spots. In the course 

 of a progressive reduction of pigment, affecting all parts of the chequered area, but 

 advancing from before backward, the spots in the lesser and median coverts will 

 reach the point of disappearance at a time when the spots on the long coverts and 

 secondaries have only been cut down to a half or a third of the initial size. At that 

 time we should have left these two rows of spots, reduced in number at the lower 

 ends, where they are weak at the start, and shortened up by a more or less even 

 cutting off of their ends to square or roundish chequers. The two bars, as wholes, 

 would of course have the position and curvature prescribed in the arrangement of 

 the feathers bearing them. Standing alone on a pale-gray ground, these bars would 

 gain immensely in conspicuity and utility as ornamental recognition-marks. The 

 advantage of all this to the species, whatever it be, would be merely an accident 

 of the situation presented at this particular point in a progressive series of 

 modifications. It is conceivable that the utility of the bars might be great 

 enough to give natural selection a chance to step in and bar the way to further 

 reduction. But the process of obliteration has certainly gone much farther in 

 many other species. There may be stages in the process which suggest utility; but 

 when we consider the whole series of stages and note that the process runs on, 

 sweeping away the stages which we imagine to be most useful, we are left with 

 the conviction that some general principle underlying the course of events has not 

 yet been fathomed. 



The mutationist is compelled to take his stand on immutable unit-characters. 

 A character may fluctuate to and fro, but it never loses its balance, except by a 

 sudden transformation that makes it a wholly new character and its bearer a new 

 species. To the mutationist the bars of the Columba livia, both as a whole and in 

 their component elements, would be viewed as essentially fixed units. As we look 

 at the character in the mature state, it seems a perfect picture of immutability. To 



