252 Absence of Segregation . [ch. 



were bred showed only dubious traces of segregation. On 

 the other hand F^ was crossed with both parent types and 

 some large families raised. The general aspect of these 

 families might be described by the statement that they 

 came on the whole intermediate between F^^ and the pure 

 parents respectively. Pending a repetition of the experi- 

 ment, and especially before F^ has been properly investi- 

 gated, the negative cannot be asserted with complete 

 confidence, but I am disposed to regard the case as one in 

 which segregation is really absent. 



Experiments made by Bacot and Prout with Acidalia 

 virgularia (Geometrid moth) and a light variety from 

 Hyeres called canteneneraria showed that F^ was always 

 intermediate. In subsequent generations no segregation 

 could be detected. A good deal of fluctuation occurred, 

 but the evidence went to show that either extreme could 

 produce the other. The facts were described by Mr Bacot 

 in a lecture at the London Hospital in 1908, and I am 

 indebted to him for further information. A full account is 

 shortly to appear. 



2. Departures from Numerical Expectation, 



The material for a proper investigation of this important 

 question is still quite insufficient. It is the experience of 

 all who have made breeding experiments that from time to 

 time individual families show great aberration from the 

 numbers which are predicable by the simple rules. " Not 

 till a laree mass of otherwise homooeneous statistics are 

 available will it be possible to ascertain whether any of 

 these aberrations are really indicative of the eftects of 

 special causes of disturbance or are to be accepted as 

 random departures from normality. From the point of 

 view of the statistician it may seem a comparatively simple 

 matter to procure such material, but in practice there are 

 many difficulties to be overcome. The impression formed 

 on my own mind is that the output of allelomorphic gametes, 

 where equality is expected, does in certain types show in- 

 dividual departures which are not purely fortuitous, but in 

 any collection of statistics large enough to give reliable 

 conclusions these individual aberrations would be completely 



