PRELIMINARY TREATISE. 
27 
may be available in inferior subdivisions though very little 
reliance can be placed on them. The fundamental principles 
on which I build may be thus briefly recapitulated. 1. There 
is a general point of agreement between all vegetables, 
namely, the character which separates them from the other 
portions of the creation. 2. As individual vegetables differ 
in a variety of points, those points taken singly will separate 
the mass, step by step, from the point of general agreement 
to that of individuality. 3. The only mode of avoiding con- 
fusion is to work downwards from the point of agreement to 
that of individuality. 4. Great attention is requisite in de- 
ciding which points of separation should have the priority. 
5. That point should have the priority which shall be found 
not to separate (or, if that be impracticable, to separate the 
fewest) individuals connected either by a feature apparently 
more important, or by a prevailing weight of other features. 
6. Each point must be taken singly, and contrasted with that 
which differs from itself, and not with some other feature con- 
sistent with it. 7. No point of difficult investigation should 
be put foremost, if one of more easy access is found to coincide 
with it. 8. The differences between the absence and presence 
of a feature being in many cases not absolutely defined, but 
intermediate appearances being found to intrude, there are 
but two ways of dealing with them, either to place them in 
an intermediate division, or in that to which they appear most 
nearly allied, stating them to differ through superfluity or 
defect; for instance, Commelina is an hexandrous plant, trian- 
drous by defect ; Azalea is a decandrous Rhododendron, trian- 
drous by defect, and some species of Gethyllis and Vellosia 
are pleiandrous or more-stamened by superfluity. 9. Inter- 
mediate divisions, if adopted, are further separable by various 
modifications. 10. The Creator not having made all vege- 
tables to differ from each other by any fixed number of points, 
but in the process of separation by distinct features, some 
being detached from the whole mass at an early and others at 
a later step, any system of arrangement that shall group them 
by an equal and fixed number of subdivisions or steps from 
the point of universal agreement, must be inconsistent with 
nature and fundamentally untrue. Therefore, if it be thought 
advisable for the purpose of assisting the memory to adopt 
any given number of principal divisions, such as class, alli- 
ance, group, order, suborder, and section, it can only be done 
with truth by admitting in each of those divisions an unli- 
mited number of grades, and the determination of the 
