PROGRESS IN THE STUDY OF VARIATION. 567 



more from slugs — which attack the species greedily — than 

 the hairy ones did. We have no information on the point, 

 but it is just possible that some creature which elsewhere 

 eats off all the smooth seedlings is absent from the Val 

 Formazza. The eggs of Pieris napi are laid on both kinds, 

 and the larvae can eat both kinds, but whether they prefer 

 either I cannot say. 



These two cases must suffice as examples of those 

 varieties which, coexisting with the type, have the quality 

 of^ distinctness. Many such cases wull be known to those 

 who collect animals or plants. Of few indeed have any 

 satisfactory accounts, much less statistical records, been pre- 

 pared showing how or in what proportions the various 

 forms occur in nature ; and in still fewer has any cross- 

 breeding been systematically carried out. In most cases all 

 that we know is the bare fact that such a species has such a 

 variety commonly or occasionally found associated with it, 

 if so much. Rarely do we know for certain whether inter- 

 mediate forms exist or not. On such points the writings of 

 systematists are too often consulted in vain. 



In all these cases the course to be followed in the first 

 instance is simple, and consists in examining and, if pos- 

 sible, preserving a fairly numerous sample taken at random 

 from as many distinct localities as possible. In recording 

 the results of such collecting, the essential points to be 

 noted are these : — 



(i) That the variety occurs associated with the type; 

 that it is scarce or abundant. 



(2) Whether the variety is connected with the type by 



a series of intermediate forms or not. 



(3) If intermediates do occur, it is then to be deter- 



mined whether in each sample the specimens 

 can for the most part be readily sorted into type 

 and variety, or whether specimens that are so 

 intermediate as not to be thus referable are 

 common. 



(4) The above questions to be answered separately for 



each sample from a distinct locality. 

 With regard to the third point, it is obvious that when- 



