ARTIFICIAL SYSTEMS 307 
other genera that might be mentioned. Crude as this ar- 
rangement was, it afforded for many generations the best 
general classification of plants that anyone had to offer; and 
it was not until after the revival of learning in Europe, dur- 
ing the sixteenth century, that any important efforts were 
made to find a better way. 
84. Artificial systems. An attempt was made to over- 
come the above objection regarding unnatural separation 
of sorts much alike, by calling the larger shrubs, trees, and 
the smaller ones, herbs, thus doing away altogether with 
the intermediate division. This, of course, lessened the 
difficulty in a way, but can hardly be said to have removed 
it. To make smaller groups, these two were again subdivided 
according to differences observed in this or that part. Thus, 
some writers made subdivisions according to the shape or 
arrangement of the leaves; others according to the form of 
the fruit or seeds; others still, according to peculiarities of 
some part of the flower; and so on, each writer basing his 
system upon characters taken from one or two parts. Many 
attempts of this sort were made during the next two centuries. 
Some of these systems were decided improvements over 
the earlier classification, but even the most elaborate of 
them had the same fundamental weakness already pointed 
out in the arrangement according to size. We know that 
plants which differ a good deal as regards a single part may 
be very much alike in all other respects, while plants much 
alike in a certain part may be otherwise very different from 
one another. For example, the fruits of the almond and the 
peach differ much in appearance when ripe, but otherwise 
' an almond-tree and a peach-tree are almost exactly alike. 
On the other hand, the root of the beet and of the turnip are 
often of exactly the same shape, while the plants are strik- 
ingly different in all other respects. It is plain, therefore, 
that any arrangement of plants based upon a single character 
or very limited set of peculiarities, is bound to be unsatis- 
factory, because it cannot accomplish the chief purpose of 
a classification, namely, to group nearest together the sorts 
that are most alike. In a word, these systems failed chiefly 
because they are artificial, and so not well calculated to ex- 
