LATENCY AND RETROGRESSION 115 



plays the chief part. The same is the case with aphides, which 

 produce males only when the weather grows cold and nutriment 

 less abundant. In true retrogression, on the other hand, there is 

 not mere latency, but absolute loss. A hereditary tendency is 

 eliminated from the germ-plasm. If the character which was lost 

 was a new variation, it can recur in the race only through another 

 variation, of like nature ; if it has arisen through evolution, through 

 the piling up of progressive variations during a succession of 

 generations, it can recur only through a similar process of selection. 

 It is conceivable, of course, that no characters are ever completely 

 lost, but that all apparently lost characters become latent. But 

 this implies that, though the germ-plasm may gain, it cannot lose ; 

 that the only true kind of variations are progressive variations ; 

 and that while it is possible for the individual to vary so that 

 he completes the parental development and adds a step, it is 

 impossible for him to vary so as not to complete it, except by 

 making a character latent. It implies that all the variations of 

 the two parents, the four grandparents, the eight great-grandparents, 

 of all the billions of ancestors, are represented in the germ-plasm ; 

 that nothing, not even a minute variation in a hair, that ever 

 appeared during the immensely long and varied life-history, has 

 ever been lost ; and therefore, that a human embryo, for instance, 

 has latent not only all the physical and mental characters 

 of the many types which preceded him, but all the variations of 

 these characters which occurred in all his ancestors. The germ- 

 plasm in a microscopical germ-cell is doubtless a very complex 

 entity, but this hypothesis would endow it with a greater number 

 of hereditary tendencies than it has chemical atoms. In truth we 

 have no choice but to believe that hereditary tendencies may not 

 only become latent, but that they may be entirely lost. Indeed, 

 when we remember how numerous have been the variations and 

 the ancestors of every individual, and how immense the changes 

 during the life-history, it is evident that the proportion of characters 

 which have persisted, either in a patent or in a latent condition, 

 must be infinitely small when compared to those which have been 

 altogether lost which have retrogressed absolutely. 



189. A complete discussion of latent characters may be 

 postponed with advantage. We need note only that when a 

 character, for example the gorgeous plumage of Gallus banktva, 

 the wild ancestor of our domestic poultry, becomes latent, then 

 (since retrogression and reversion are identical) subsequent 

 retrogression, which eliminates the more sober colours of the 



