24 



K. S. LASHLEY 



TABLE 3. 



Mean number of tentacles of all buds produced during the time of cultivation by each 

 parent of the last selected generation (last column, table 1). Not less than ten 

 buds were obtained from each parent 



ence between the selected members of the two groups, remained 

 fairly constant throughout the experiment. 



DISCUSSION 



Clearly this experiment offers no positive evidence for evolu- 

 tion by the accumulation of continuous variations. It remains 

 to determine its value as evidence against such a method of evo- 

 lution. This may best be done by a consideration of the sort of 

 hereditary changes which such an experiment should reveal. 

 The figure, 1.321, determined as the average extent of selection 

 in each generation, makes it possible to compute roughly the 

 theoretical effect of selection through any number of generations 

 for any given strength of heredity, provided that selection of 

 somatic variates does not interfere during the experiment. The 

 number of buds obtained from the last selected generation is 

 great enough to give a probable error for the difference between 

 the groups of only 0.031 tentacles. From this, the probability 

 that any expected difference between the selected groups would 

 be revealed by the experiment may be computed. If we assume 

 that the offspring inherit one-tenth of the variation of their par- 

 ents, the theoretical result of 19 generations of selection at a 

 difference of 1.321 tentacles per generation is a difference of 1.176 

 tentacles between the offspring of the last selected generations of 

 the two groups. This is 38 times the probable error of the dif- 



