10 



MADRONO 



[Vol. 1, 



but is, on the whole, so variable that only slight importance can be 

 attached to it, except in one or two species. 



Search for technical characters for the separation of species 

 has been carried on by the writer, but has been to a large degree 

 unsatisfactory in results. Experience, both in field and in her- 

 barium, determines that habit, general aspect, and hue must be 

 taken as of first importance in segregation, although pubescence 

 and glandulosity would have practical value in the construction of 

 diagnoses. Now, if the Californian species be segregated on this 

 basis, one obtains about twenty species. These species, I may say, 

 fall into five or six fairly natural groups, the species in each group 

 being very closely related. In successive attempts to determine the 

 most satisfactory criteria for separating the species within each 

 group, all possible information regarding the life history was 

 sought. As a result, it was found that within the limits of a group 

 the differential mortality of two species very closely related is 

 markedly decisive. As this cleavage is largely between closely related 

 species rather than between groups, the difference in reaction to fire 

 is highly interesting, and is also, evidently, of taxonomic value. In 

 other words, the working conception of species above outlined 

 seems to be unexpectedly fortified by their habits in relation to fire 

 — that is, of root-crown sprouting or of failure to do so. 



Fig. 5. 



