150 CARNEGIE INSTITUTION OF WASHINGTON. 



comes to the increase over the controls in the numbers of Utters produced 

 by the untreated descendants of the alcohoUzed animals. It seems necessary 

 to assume that there are genetic factors influencing the number of Utters 

 that are produced; the alcohol prevents the reproduction of such females 

 as carry, genetically, a lower reproductive capacity, so that the litters pro- 

 duced come alone from females carrying higher Utter-producing capacity 

 and they in turn produce larger numbers of Utters than the unselected con- 

 trols. When these offspring of treated rats were themselves treated, they 

 produced fewer instead of more litters than the controls, but the genetic 

 superiority is shown by the fact that the alcohol reduced the numbers of 

 their litters by only 35 per cent, while it reduced the numbers in the earlier, 

 unselected generation by 65 per cent. This selective action of the alcohol 

 will account for the results from the number of litters, but not those from 

 litter-size; if this is correct, it indicates that the number of litters is influenced 

 by genetic factors that do not influence Utter-size. This is not difficult to 

 believe, since Utter-size is largely dependent upon the number and constitu- 

 tion of the germ-cells, while the somatic conditions of the parents play a 

 large part in determining whether or not a Utter will be produced. This 

 distinction between the effects upon the numbers of litters and upon their 

 size has not been made by previous investigators. The results from litter- 

 size agree strikingly with those of Stockard from similar studies with guinea- 

 pigs; the results from numbers of litters agree with Pearl's on fowl in so far 

 as both results demand the assumption of a selective action of alcohol. In 

 the fowl the alcohol appears to select between germ-cells, in the rats it 

 appears to select between females. 



'^Weight.— The weight data form an extensive series, consisting of weekly 

 weighings of practically all the rats raised in the various generations in the 

 second series of experiments (those with the heavy dosage, started in 1916). 

 The weights used in the summaries were read from the individual growth- 

 curves at certain ages. This procedure was necessary, since the weighings 

 for all rats were made on the same day each week, when the rats naturally 

 were of different ages. The results are based primarily on the males, since 

 the pregnancies of the females make their data less reliable. In the summaries, 

 the pregnancies have been arbitrarily smoothed out. Each of the four 

 strains shows that treated rats tend to grow more slowly than the controls. 

 This is an influence felt by the population as a whole, although there are some 

 males that remain as heavy as the heaviest controls. The females show a 

 similar retardation in growth, but this is not so marked as in the males. 

 The offspring of the treated rats tend to grow 7nore rapidly than the controls. 

 This result is not so clear as the opposite result in the preceding generation; 

 the differences are not so large and all strains do not show this in equal 

 measure. In comparison with the results from the numbers of litters this 

 shows a marked similarity, which is further borne out by the results from 

 treated rats from treated parents. Instead of causing still further reduction 

 in weight, the treatment of the offspring that came from treated parents 

 appears to leave the animals about equal to the controls. Just as upon the 

 numbers of litters alcohol works as a selective agent eliminating the Utters 

 that bear the genetic determiners for slower growth, so the offspring from 

 treated parents grow faster than the controls, and when they themselves are 

 treated the reducing effect of the alcohol makes them about equal the con- 

 trols instead of growing markedly slower, as did their parents. Very little 

 can be concluded from the weights of the grandchildren of treated animals. 

 The numbers are too small to determine whether or not the expected con- 

 tinued superiority of the tests is realized. Two of the strains show the tests 



