NILSSON'S DISCOVERY 85 



single specimen among the hundreds of thousands will sport, 

 deviating in some mark from the main type. As soon as such 

 an event is noticed, the sporting individual 13 usually isolated 

 and its grains are saved separately. 



This is practically nothing more than a new initial selec- 

 tion, a starting-point for a new race, which has to be isolated, 

 multiplied and tested in the same way as all the others, and 

 has at least an equal chance with them of yielding a valuable 

 contribution to agricultural practice. Some very good 

 novelties have been obtained in this way at Svalof. These 

 changes come on unexpectedly and all of a sudden, exactly as 

 ordinary mutations. No visible preparations and no groups 

 of intermediate forms accompany them. They are at once 

 quite distinct from the main type. But, like mutations, they 

 may be confused with the effects of accidental crosses, and 

 such, though very rare, are also seen to occur in the experi- 

 mental fields of pedigree-races at Svalof. Crosses may yield 

 hybrids which will split in succeeding generations as we have 

 seen and whenever the result of the isolation and separate 

 multiplication of a sporting individual is a heterogeneous 

 group, it must be assumed that the deviation was due to a 

 cross. 



On the other hand, such a cross does not necessarily 

 exclude a real mutation, since mutations are believed to occur 

 in the production of sexual cells and so may be limited either 

 to the pollen or to the ovaries of a mutating plant. But this 

 theoretical side of the question does not, strictly speaking, 

 belong to our present discussion, and so it may be considered 

 sufficient to have indicated it. My only aim was to point out 

 the difficulties which here, as everywhere, are found in the 1 

 way of distinguishing between mutants and accidental hy-j 

 brids. This, however, has reference only to the special cases, 

 but from a broad point of view the experimenters of the 

 staff of Svalof are satisfied that real mutations as well as 



