XXVIU GENERAL INTKODOCTION. 



rock-pigeon. If there is any difference it is in the tail. 

 There are the typical number (twelve) of tail feathers, 

 and the outer margin of the outermost feather at each 

 side is edged with white, except near the tip, but the 

 feathers at each side of the middle line have their inner 

 edges very slightly tilted upwards. This is the only 

 hint of the highly specialised nearly upright fan-like tail 

 of the white sire. 



From these experiments with pigeons it is evident that 

 the reversion may be partial or all but complete in form 

 as well as in colour. In tlie case of the white fantails 

 tliere is partial reversion in colour onl^- ; in the case of the 

 white fantail and the blue pouter — two well-marked and 

 long-established breeds — the struggle amongst the parts 

 ended m the production of a bird taking after the one 

 parent in its colour and the other in its form ; with the 

 owl and archangel there was considerable reversion to- 

 wards the blue rock, but, perhaps because the archangel 

 is a comparatively recent production, none of its characters 

 could be detected in its cross-bred offspring ; in the case 

 of the white fantail and owl-archangel cross there is 

 practically complete reversion in form as well as in colour. 



Perhaps I may here say that reversion is more a 

 negative than a jjositive influence, that if I understand it 

 aright complete I'eversion is mainly due to the develop- 

 ment being abruptlj- ari-ested so as to reproduce a lost 

 ancestor. Sometimes several printings are required to 

 produce a coloured plate. Were one or more of the 

 printings omitted a kind of "reversion" would be the 

 result. When the owl and archangel are crossed the 

 latest colours added by the fancier are not reproduced, 

 and the older and simpler colours are again made visible. 

 When the white fantail and owl-archangel were crossed 

 all the recently acquired colours were lost, and the common 

 ancestor of all our domestic pigeons was once more re- 

 produced. How the ancestral units of germ-plasm over- 

 come during development the less stable units represent- 

 ing more recently acquired peculiarities can oulv be 

 ii'uessed at. 



