No. 510] VARIATION IN NUMBER OF SEEDS 355 



a character dependent not merely upon the interna] and 

 external factors determining the form of all plant organs, 

 but upon the environmental factors influencing the 

 chances for fertilization as well. Furthermore, it may he 

 noted that we are dealing not with a little variable char- 

 acter, but with one having a coefficient of variation of 

 about thirty five per cent. This is about two or more 

 times the value of the coefficient of variation of the ordi- 

 nary vegetative characters of plants, but not an extra- 

 ordinarily high value for seed numbers, as determined 

 from large series of unpublished data. 



Summarizing the results of this note, I think we may 

 say that so far as is shown by this one series of material 

 Cytisus is no more variable in the habitat to which it has 

 been very recently introduced than it is in one of its na- 

 tive habitats. As a matter of fact it is slightly, though 

 not significantly, less variable at Wood's Holl than at 

 Danby Dale. Possibly the sensible identity of the varia- 

 bility of the two samples may be due to the small number 

 of individuals which furnished my series of data. These 

 in their turn were probably derived from very few ances- 

 tors. I attach no great importance to the results. They 

 simply show that so far as the present material is con- 

 cerned there is no evidence in favor of the theorv that 

 the introduction of a species into a new habitat increases 

 its variability. Possibly some other character than num- 

 ber of seeds per pod would have shown an increased 

 variablity. I use this character merely because I had 

 the data on hand as a by-product of some other work and 

 felt that it would be better to let it contribute its bit of 

 evidence on this problem than to throw it away. Pos- 

 sibly as a fellow biologist remarked when I showed him 

 the closeness of agreement of the constants from the two 

 series the result is ' ' purely accidental. ' ' There are many 

 possibilities of error in work of this kind. But until we 

 have a large number of series of quantitative data it will 

 be quite impossible to say whether or not there is any- 

 thing of value in the theory of an increased variability of 



