PRESIDENTIAL ADDRESS. 655 
useful to the applied botanist as a professedly natural one. But our know- 
ledge is incomplete, and the accession and intercalation of new types renders 
any artificial, and most attempts at a natural, system sooner or later unwork- 
able. The more closely an arrangement approximates to the natural system, 
the less can the intercalation of new forms affect its stability. The more 
stable a system is, the more easily will its details be remembered and the more 
useful will it prove in practical reference work. Here, therefore, for once, 
self-interest and love of truth go hand in hand. 
Since the organographic systematist learns their characters from his 
groups, while the comparative morphologist defines groups by the characters 
he selects, their results, were knowledge complete, should be identical, and this 
identity should prove their accuracy. But knowledge is finite, and these 
results are not always uniform. The want of uniformity is, however, often 
exaggerated because the reasons are not always appreciated. 
One cause is the difference in personal equation, which affects alike the 
worker who deals with things and him who considers attributes. It would be 
contrary to expectation were every phylogenist to assign the same value to each 
. character, or every systematist to apply the same limitation to each type or 
group of types. The divergence of view on the part of two observers may 
show a small initial angle; it may nevertheless lead them to positions far 
apart. But while divergence of view is the most obvious explanation of the 
want of uniformity apparent in systematic results, it is the least effective 
cause. This inherent tendency to differ manifests itself in contrary direc- 
tions ; in the long run individual variations are apt to cancel each other. 
The nature of the work counts for more than the predisposition of the 
worker. The aggregations on organographic lines, which were the main 
guides to the composition of the higher groups until phylogenetic study was 
seriously undertaken, do not assist the comparative morphologist. The 
characters on which phylogenetic conclusions may be based increase in value 
in proportion to the width of their incidence, so that the greater their value 
for phylogenetic purposes the less do they aid the descriptive worker in dis- 
criminating between one plant-type and another. Often they are characters 
which for practical reasons the descriptive worker must avoid. Organo- 
graphy, then, may not give evidence as to characters whereof cognisance 
cannot be taken, while for another reason the comparative morphologist may 
not use characters derived from descriptive sources. The object of the 
phylogenist is to take his share in advancing our knowledge of taxonomy ; 
to seek from the systematist the evidence on which his results are based 
would be to vitiate the reasoning of both. All that the phylogenist can ask 
the descriptive worker to do is to supply the units that require classification. 
The comparative morphologist, relying mainly on anatomical and embryo- 
logical evidence, at first had a hope that his method of study might enable 
him to supply his own units and thereby render further taxonomic work 
based on organography unnecessary. This hope remains unfulfilled, and the 
phylogenist, as a rule, limits his efforts to a narrower field. The organo- 
graphic systematist realises that in the present state of our knowledge the 
study of the incidence of selected characters gives more satisfactory results as 
regards the composition of the higher phyla than repeated aggregation can 
attain, while the comparative morphologist recognises that, as matters stand, 
the approximations of organography in respect of types and kinds are more 
satisfactory than any he can yet offer. Since, however, the progress in one 
case is outwards, in the other the reverse, a zone of contact is inevitable. 
This zone, in which the influence of both methods of study is felt, is occupied 
by those groups immediately higher in value than the natural families of 
plants, and it is here that discrepancies in the results attained chiefly manifest 
themselves. These discrepancies take the form of unavoidable differences of 
opinion as regards the composition of collections of natural families. If a 
family A possesses ten characters of ordinal import, whereof it shares eight 
