THE PKINCIPLES OF CLASSIFICATION. 319 



luxuriance or development, more or less under the control of 

 conditions, their origin being wholly unaccountable. They arise 

 in the seed-bed, or sometimes from buds, which as the gardeners 

 say " sport" l That is, some seedlings, or some shoots, are 

 unlike the rest in certain particulars. 2 



631. Most varieties originate in the seed, and therefore the 

 foundation for them, whatever it may be, is laid in sexual repro- 

 duction. But Bud-variation, or the u sporting" of certain buds 

 into characters in branch, flower, or fruit unlike those of the 

 stock, is known in a good number of plants. 3 It might also 

 o^cur in corals, hydras, and other compound animals propagated 

 by budding. Once originated, these varieties mostly persist, 

 like seedling varieties, through all the generations of budding 

 growth, but are not transmitted to the seed. 



632. Upon the general principle that progeny inherits or tends 

 to inherit the whole character of the parent, all varieties must 

 have a tendency to be reproduced by seed. But the inheritance 

 of the new features of the immediate parent will commonly be 

 overborne by atavism, i. e. the tendency to inherit from grand- 

 parents, great-grand-parents, &c. Atavism, acting through a long 

 line of ancestry, is generally more powerful than the heredity 

 of a single generation. But when the offspring does inherit the 

 peculiarities of the immediate parent, or a part of them, its off- 

 spring has a redoubled tendency to do the same, and the next 

 generation still more ; for the tendency to be like parent, grand- 

 parent, and great-grand-parent now all conspire to this result 

 and overpower the influence of remoter ancestry. Close-breed- 

 ing (398) is requisite to this result. In the natural wild state, 

 varieties many and conspicuous as they often are must be 

 much repressed by the prevalent cross-fertilization which takes 

 place among the individuals of almost all species. Cultivators and 

 breeders in fixing varieties are careful to secure close breeding 

 as far as this is possible. This has fixed the particular sorts of 

 Indian Corn, Rye, Cabbage, Lettuce, Radishes, Peas, &c., and 



1 Both the technical English term. Sport, and its Latin equivalent, Lusus, 

 are sometimes used for bud-variation only, yet as commonly for seedling 

 variation also. 



' 2 Darwin assumes that variation is of itself indefinite or vague, tending 

 in no particular direction, but that direction is wholly given by the elimina- 

 tion in the struggle for life of all but the fittest for the conditions. But 

 what we observe in the seed-bed does not suggest this view. Naageli, Braun, 

 and myself incline to the opinion that each plant has an inherent tendency 

 to variation in certain general directions. 



3 A list of known bud-varieties is given in Darwin's Variation of Animals 

 and Plants under Domestication, Chapter xi. 



