254 PEARL— EFFECT OF CERTAIN POISONS 



produced in the breeding season of 1915 from alcoholized parents 

 was no greater than the number produced from untreated parents. 

 In actual fact there was exactly one chicken out of 234 hatched from 

 alcoholized parents in 191 5 that was too weak to band, and was in 

 consequence killed at the time of hatching. None was deformed. 

 Out of 1,527 chicks from untreated parents, 16, or i.o per cent., 

 were weak or deformed or both. 



Discussion of Results. 



We have seen that out of 12 different characters for which we 

 have exact quantitative data, the offspring of treated parents taken 

 as a group are superior to the offspring of untreated parents in 8 

 characters. The offspring of untreated parents are superior to those 

 of the alcoholists in respect of but two characters, and these are 

 characters which are quite highly correlated with each other and 

 really should be counted as but a single character. Finally with 

 respect to two character groups there is no difference between the al- 

 coholists and the non-alcoholists. The character groups which have 

 been dealt with in this study, and for which definite quantitative 

 data are herein presented, seem to cover a much wider range of 

 physiological and genetic factors and phenomena than has ever been 

 included or even approached in any previous study regarding the 

 effects of parental alcoholism upon the progeny. They have the 

 further advantage of being characters which are measurable (either 

 statistically or otherwise) and on that account greatly reduce, if 

 they do not entirely eliminate, the possibility of personal bias or 

 prejudice influencing the results. 



The mutual accordance of the results from characters involving 

 such a manifold range of physiological factors is striking. This 

 fact in considerable degree offsets the fact that as yet our series of 

 experimental animals is statistically small, leading to such large 

 probable errors that the individual differences are not in every case 

 significant in comparison with their probable errors. The experi- 

 ments are of course being continued and expanded, and concurrently 

 the probable errors of differences are being reduced. If results in 

 the same sense as the present ones continue to appear (as seems to 



