42 ORTHOGENETIC EVOLUTION IN PIGEONS. 



that one of these characters differentiates from the other; and how far would such 

 differentiation have to advance before becoming specific? If carried to the extent 

 of turning the two bars into bands, would the specific rank of the bands then be 

 indisputable? If so, the whole point is conceded, for the transforming process 

 advances to the end by transitional stages, as is plainly to be seen in the few species 

 that have reached a higher level than Ocyphaps. 



Glancing now at the pattern as a whole, we notice first of all that the two charac- 

 ters are separately distributed, each in a field of its own. As the rows of feathers 

 succeed one another in regular order, without a break, how does it happen that 

 just two rows have been reserved for the older character, while all the others are 

 monopolized by the later character? There must be some definitely directed 

 process of transformation to account for such relations. Is it a process that operates 

 progressively and with continuity, or is every advance a clean-cut jump? But a 

 linear series of jumps running in one direction would leave no very certain distinc- 

 tion between mutation and orthogenetic variation. In the fore part of the wing 

 the bands are very narrow and pale, and they completely vanish a little behind the 

 front edge of the wing. As we pass backwards, they become wider from row to 

 row, the last band being about twice as wide as the one just before it, and about 

 one-third as wide as the bar that follows it. All this looks as if the transformation 

 had moved along quite gradually, following the law of reduction of pigment, which 

 I have demonstrated in other pigeons. If the nature and direction of the variation 

 are what they seem to be, then the regional distribution of the two characters and 

 the gradual increase of the width of the bands as they run backwards towards the 

 bars begin to clear up. 6 



However, before all the features and relation's in the pattern and the nature of 

 the variation can become fully intelligible, the mode of origin of the bands Ynust 

 be precisely ascertained. In this matter we come squarely to the main issue: Is 

 it mutation or transitional variation? At first sight the juvejial pattern suggests 

 mutation. Although the bands vary by increasing width, the difference between 

 the last band and the bar following it is manifestly greater than is the difference 

 between any two successive bands. But observe that in the upper part of the bar 

 the bridge of transitions is given in three spots. It is here that we see just how the 

 longitudinal elements of the bar are converted into the transverse elements of the 

 bands. This is done by reducing the length of the spot, chiefly from the proximal 

 end, until nothing remains but a narrow band-like remnant, with its longer axis 

 at right angles to that of the original spot. 



1 It would be very interesting to know whether the process is one of continuous development or one that moves 

 mutation-wise, jumping all at once from the Juvenal to the adolescent stage. As feathers once formed are fixed struc- 

 tures, we can not of course expect to see one of these long spots reduced in length from day to day, and thus gradually 

 turned into a transverse band. We can, however, demonstrate experimentally that the advance from the Juvenal to 

 the adolescent pattern is a progressive development that continues for at least several months and possibly for a year 

 or more. Successive plumages and color-patterns seem like discontinuous phenomena, because they measure off at 

 regular intervals the results of work that is done behind the scenes. The Juvenal feathers show only how far the trans- 

 forming process had advanced at the end of the first, week or ten days of life. The next plumage reveals the progress 

 made from that time to the time of the first molt. The proof that the process moves on gradually from one plumage 

 stage to the next is easily obtained by plucking a feather or two in the region of transitional change at any time or 

 times before the first molt. In this way we may make room for new feathers to appear at any desired ages intermediate 

 between the Juvenal and adolescent periods. In these new feathers we shall get the Juvenal pattern more or less 

 modified in the direction of the adolescent according to the ages represented by them. I have not made this test in 

 the crested pigeon, but have made it in the young of Geopelia cuneata with such perfect success (see Chapter X) 

 that I feel safe in assuming that similar results would follow the test in other cases. 



