PHYLOGENY AND CLASSIFICATION. 93 



definitions of a true species. Mere likeness counts for 

 little : males and females may often differ very consider- 

 ably from each other, and in polymorphic forms the off- 

 spring of the same parents may be of several different 

 types. Sterility has often been held to form a definite 

 barrier distinguishing true species from mere varieties 

 (p. 83) ; but there are all degrees of fertility, and crosses 

 between forms which no systematist would hesitate to 

 call species often yield offspring. Sterility is a character 

 variable like any other ; it occurs even between members 

 of undoubtedly the same species, as we know in the case 

 of our own. 



It is true that De Vries has recently tried to define 

 "elementary species" as forms produced with a new 

 progressive mutation, due to the acquisition of some new 

 factor. Such forms, if they really occur suddenly, could 

 certainly be considered as definite discontinuous steps in 

 evolution. Especially among plants there are many 

 widely distributed " species " containing a large number 

 of local races or subspecies, each breeding true and ap- 

 parently differing from the others by one or perhaps a 

 few "unit characters." Of the whitlow-grass (Drdba 

 verna) some two hundred constant races have been dis- 

 tinguished by Jordan ; and similar subdivisions of the 

 species of Viola, Heliantheinum, &c., are known. But the 

 evidence as to the sudden origin of the mutations is 

 weak ; and, at all events, it is by no means yet established 

 that new factors may not appear gradually in increasing 

 intensity, so that discontinuity becomes negligible, and 

 the new characters develop in steps small enough to 

 satisfy the most exacting selectionist (p. 76). The diver- 

 sity of variation and the smallness of the steps in muta- 

 tion are well illustrated in the case of the hawk weed 

 (Efieracium) and the dandelion (Taraxacum), which re- 

 produce parthenogenetically ; as in the Bacteria, there 

 may be differentiated as many " pure lines " as there are 

 parents, and a bewildering number of races results, 



