206 Plant-Breeding 



very likely may be recorded in the gametes and vitiate 

 the final results. (6) Variability itself may be a unit- 

 character and therefore pass over. There is probably 

 such a thing as a " tendency to vary/' wholly aside from 

 the fact of variation. (7) Many of the plants with which 

 we need most to work in plant-breeding are themselves 

 eminently variable, and the results, even if there is true 

 mendelism, may be so uncertain as to be wholly unpre- 

 dictable. (8) Many plants with which we must work 

 will not close-fertilize. Some of them are monoecious or 

 dioecious. Even if there is gametic purity in such plants, 

 the probability is that the fact can be discovered only by 

 a long line of scientific experimenting for that particular 

 purpose and not by the work of the man who desires only 

 to breed new plants. (9) A cultural variety, in any true 

 acceptation of the term, is a series of closely related 

 plants having a pedigree. It runs back to one individ- 

 ual plant, from which propagation has been made 

 by seeds or asexual parts. Now, one can never predict 

 just what combination of characters any plant will have, 

 even though it be strictly mendelian. A person might 

 have a thousand hybrids of which no one plant shows any 

 two characters in the proportion of 3 to 1 (both seed-char- 

 acters may appear in the same pod or in different pods) on 

 the same plant, let alone all the characters as 3 to 1 or in 

 other definite relation ; and yet the total average numeri- 

 cal results might conform exactly to the mendelian law. 

 Mendel's law is a law of averages. For example, in ten 

 plants of peas, Mendel found the following ratios in respect 

 to seed-shape and seed-color. (Similar ratios were found 

 for other characters.) 



