PRINCIPLES OF CLASSIFICATION. 125 
ture which to-day make a variety may continue to diverge 
- until in time they constitute a species, and specific characters 
in time come to be generic. 
If, following the same idea, we try to look backwards 
instead of forward, we may say that the species which exist 
now—let us call it epoch A—were only varieties in the pre- 
ceding epoch B; that the genera of to-day represented species 
in epoch B, and only varieties in epoch C; and that the 
orders of to-day were varieties perhaps as far back as epoch D 
or BK. ‘The outcome of such an idea is that a species is, aiter 
all, a unit of classification, possessing only a transient charac- 
ter; and, further—to come back to an idea already expressed 
—that if we could go far enough back in time we should find 
that all our existing plants were descended from a few (per- 
haps from one) primitive forms. The best (2.e., the most 
natural) classification, therefore, is one which will enable us 
to group plants exactly by their descent relationships. 
In the succeeding portion of this work it is only possible to 
state the characters of the more important Natural Orders, 
especial prominence being given to those characteristic of New 
Zealand. ‘The arrangement followed among the Dicotyledons 
is not the best, which would probably be to place the Gamo- 
petale first as the most highly-developed group, then the 
Polypetalz, and lastly the Incomplete; but, as the main 
systematic works dealing with New Zealand, Australian, and 
Huropean botany (those of Hooker and Bentham) commence 
with the Polypetale, the same order has been followed here. 
