1^8 Retrograde Varieties 



descent is a requisite. By so doing, we exclude 

 most of the facts which were until now gener- 

 ally relied upon. For the roses, the hyacinths, 

 the tulips, the chrysanthemums always have 

 furnished the largest contributions to the dem- 

 onstrations of bud-variation. But they have 

 been crossed so often, that doubt as to the purity 

 of the descent of any single form may recur, 

 and may destroy the usefulness of their many 

 recorded cases of bud- variation for the demon- 

 stration of real atavism. The same assertion 

 holds good in many other cases, as with Azalea 

 and Camellia. And the striped varieties of 

 these genera belong to the group of ever-sport- 

 ing forms, and therefore will be considered 

 later on. So it is with carnations and 

 pinks, which occasionally vary hj layering, and 

 of which some kinds are so uncertain in char- 

 acter that they are called by floriculturists 

 '* catch-flowers." On the other hand there is 

 a larger group of cases of reversion by buds, 

 which is probably not of hybrid nature, nor due 

 to innate inconstancy of the variety, but must 

 be considered as pure atavism. I refer to the 

 bud-variations of so many of our cultivated 

 varieties of shrubs and trees. Many of them 

 are cultivated because of their foliage. They 

 are propagated by grafting, and in most cases 

 it is probable that all the numerous specimens 



