December 16, 1910] 



SCIENCE 



869 



Now the evidence whicli will presently be 

 offered shows that, contrary to the idea of 

 Cuenot as well as to the suggestion of Bate- 

 son and Punnett, the yellow egg which by 

 chance has met a yellow sperm has its career 

 ended thereby. It is not thereafter capable of 

 fertilization by a non-yellow spermatozoon. 

 So it seems probable that the homozygous 

 yellow zygote actually is formed and then 

 perishes, just as in the observations of Baur 

 on an " aurea" race of Antirrhinum, the homo- 

 zygous yellow seedling not only forms, but 

 may germinate, yet for lack of assimilating 

 power develops no further, so that all the sur- 

 viving " aureas " are heterozygous and these 

 are to the recessive green plants as 2:1. 



In November, 1907, the writers started with 

 a small number of yellow mice in an attempt 

 to obtain a homozygous yellow animal. This 

 quest was not successful, but as the numbers 

 of animals increased and the scope of the 

 experiment became greater, some results were 

 obtained of a striking nature and different 

 enough from previous results to make an 

 extensive study of the subject advisable. Such 

 a study has been carried on during the last 

 two years, and up to date the young in yellow 

 X yellow crosses have totalled 1,235. 



Of these young, as will be seen in the follow- 

 ing table, 800 have been yellow and 435 non- 

 yellow. This means that instead of the 75 per 

 cent, heterozygous yellows called for by Bate- 

 son and Punnett's hypothesis, or the 72.45 per 

 cent, obtained by Cuenot in 365 young, there 

 have been obtained 64.77 per cent, yellow, a 

 deviation of only 1.23 per cent, from the 66.66 

 per cent, that we should expect if the homo- 

 zygous yellow class is entirely absent. 



The result observed by us, 64.77 per cent, 

 yellow in 1,235 young, is a wide deviation 

 from 75 per cent., but close enough to 66.66 

 per cent, to enable us to say with considerable 

 certainty that the homozygous yellow class is 

 entirely lacking and is not replaced by hetero- 

 zygous animals of the same color. 



A still more striking result is obtained by 

 adding to the total of young obtained in the 

 experiment above mentioned the 363 young 

 of Cuenot's experiments. We then have 66.52 



per cent, yellow young in a total of 1,598, a 

 deviation of only 0.14 per cent, from the ex- 

 pected 66.66 per cent. 



A table showing the progress of the experi- 

 ment follows, to which is appended in similar 

 form a statement of Cuenot's results : 



To state in another way the closeness of 

 agreement between the expected and the ob- 

 served percentages of yellow young, we may 

 say that the ratio of 800 yellow to 435 non- 

 yellow obtained in our experiments equals 

 1.943 yellow to 1.057 non-yellow, the devia- 

 tion from the 2 : 1 ratio being 0.057. Now 

 the theoretical " average error " in the case 

 of a Mendelian 2 : 1 ratio based on the given 

 number of observations (1,235) as calculated 

 by Johannsen's formula (1909, p. 403) is 

 =t 0.013, which is slightly less than the ob- 

 served error. If, however, Cuenot's totals are 

 added to ours, the deviation from the 2 : 1 

 ratio is reduced to 0.005, while the theoretical 

 " average error " (for 1,598 observations) is 

 ± 0.011. The observed deviation is therefore 

 well within the limit of error and so points 

 strongly to the 2 : 1 ratio as the true ratio. 



Cuenot (1908) found that when yellow 

 mice are mated inter se, smaller litters of 

 young are obtained than when yellow mice are 

 mated with non-yellow ones. This observa- 

 tion we can confirm from a study of larger 

 numbers than were reported in Cuenot's ex- 

 periments. The averages reported by Cuenot 

 in the respective cases, based on a careful 

 count of 50 litters of either sort, are 3.38 and 

 3.74, respectively. From yellow X yellow ma- 

 tings we have obtained 277 litters including 

 1,305 young, an average of 4.71 young to a 



