Abstracts. 303 



though the untreated Fi animals from alcohoHsts may appear 

 normal, they really are somewhat defective. 



f. If, on the other hand, the dosage, though absolutely the 

 same, be relatively lower in proportion to the mean absolute 

 resisting power of the germ cells it would be expected that all 

 three germ cell classes (a), (b) and (c) would be represented. 

 The zygotes actually formed would be chiefly produced by (c) 

 germ cells, and to a much smaller extent by (b) cells. Under 

 these circumstances it would necessarily follow that a random 

 sample of the zygotes produced after the action of the dele- 

 terious agent would, on the average, be superior in respect to 

 such qualities as growth, etc., which may be supposed to depend 

 in part at least upon germinal vigor, to a random sample of 

 zygotes formed before the action of the agent, because the germ 

 cells of class (c) are a selected superior portion of the total 

 gametic population. 



g. Essentially that proportionality between effective dosage 

 of the deleterious agent and absolute resisting power of the germ 

 cells outlined in the preceding paragraph (f) is believed to have 

 obtained in the present experiments with fowls, Nice's experi- 

 ments with mice, and nature's experiments with the working- 

 men's population studied statistically by Elderton and Pearson. 



THE PROBABLE ERROR OF A DIFFERENCE AND 

 THE SELECTION PROBLEM.* 



This paper deals with the results of Ackert in selection of 

 Paramecium. Due to his arithmatically wrong calculation of 

 the probable error of a difference Ackert deductions based on 

 this are wrong. Corrected his data is contradictory and when 

 compared with the other selection work on Paramecium neither 

 confirms nor refutes the previous results. 



*This paper is an abstract from a paper by Raymond Pearl, having 

 the same title and published in the Genetics, Vol. 2, pp. 78-81. 



