E. R. Saunders 347 



larger i^o families composed entirely of singles with white plastids and 

 doubles with cream plastids would present no serious difficulty on the 

 present view. Segregation in these cases might be in accordance with 

 some higher term in the series. For with each successively higher 

 value for n the difference in the proportion of the two more frequent 

 forms (singles with white plastids and doubles with cream plastids) 

 would be so slight, that it would not be practically appreciable in 

 experiments on the present scale ; whereas the proportion of the two 

 rarer terms (doubles with white plastids and singles with cream plastids) 

 would be successively reduced by about one-half, and hence in small 

 families it would be unlikely that they would be recorded. If the value 

 for n were sufficiently high the coupling would appear to be complete, 

 and these two forms would then appear to be altogether wanting. Or 

 again a like result would follow if the no-d parent were by chance a 

 form in which W was linked with XY in some of the germs (see above 

 p. 33-5). In this case we should expect the F-^ plants derived from 

 the XTW germ cells to give F« families composed of only the two 

 forms — singles with white plastids and doubles with cream plastids — 

 while those derived from XYW gametes would yield the F„ families 

 which include all four types. 



The all-single F.2 families will be produced by the F^ plants derived 

 from the XYw ovules. These cross-breds will naturally breed true to 

 singleness. In all these F„ families we find either absence, or a marked 

 deficiency of individuals with cream plastids. This deficiency recalls 

 a similar result obtained in an earlier experiment where the parents, 

 both in this case eversporting, were of unlike plastid colour, d-cream 

 being used as ?, cZ-non-cream as </". In this latter case XYw ovules 

 were also fertilised by pollen carrying W, and here a single with cream 

 plastids was obtained in Fn when a descendant of plant K was used, 

 but not when other cream individuals were employed. We may 

 suppose that the distribution of W among the F^ pollen grains will 

 be the same in both crosses (see p. 332, where this point is discussed). 

 In the present case, among 10 families which included a total of 

 279 individuals none had cream plastids; in the two remaining families 

 7 plants with cream plastids were recorded in a total of 141, but these 

 7 are not beyond suspicion, since some doubt exists as to the genuine 

 eversporting nature of the cream parent (a descendant of plant K). 



To sum up ; we find that the union ci-cream $ x no-d-nou-crea,m </ 

 gave, in accordance with expectation, some all-single-families, and some 



