i8o The Use of Anatomical Characters . 
of the stem-anatomy, that is most to be condemned, for this, as 
being less subject to variation owing to external conditions, may be 
expected to afford generic characters in many cases, whilst the 
structural features of leaf and petiole are commonly only of specific 
value. Finally from certain observations it seems very probable to 
me that the anatomical structure of the fruit will not rarely be 
found to be of considerable importance in systematic diagnosis. 
An examination of the structure of the root is of course mostly 
impracticable from this point of view, besides being of a very uniform 
nature in most cases. The necessity of collecting spirit-material of 
new species wherever possible cannot be sufficiently emphasised; 
and it might further be suggested that the large botanical establish¬ 
ments, concerned with systematic botany, should keep slides 
illustrative of the anatomy of species examined, in order to facilitate 
future reference. 
The anatomical characters of a plant have either been acquired 
in the remote past and handed down from generation to generation 
or are relatively recently acquired. The former are toagreat extent 
constant and characteristic of larger groups of affinity, which are 
all derived from one ancestor, possessing those characters, whilst 
the latter are more subject to variation, and are frequently only of 
specific value; the former indicate the affinity, whilst the latter give 
rise to the structural differences in an allied group of plants. The 
former are naturally also characters which are little dependent on 
the external conditions. We must imagine the ancestor of each 
allied group of plants to have possessed the constant characters of 
this group and in addition to these, certain characters, more subject 
to variation—either internal or owing to the influence of external 
conditions—which have led to the formation of new species, etc., 
within the group. 
Turning first to the structure of the axis, which seems most 
important for the characterisation of larger groups of affinity, pith, 
wood and cortex all yield excellent characters. In the case of the 
pith we have to distinguish between a homogeneous one, consisting 
of starch-containing cells alone, or of dead and empty cells alone, 
and a heterogeneous one, composed of both kinds of cells; these 
three features are not rarely characteristic of entire genera or even 
larger systematic groups, but will generally be found to be accom¬ 
panied by more pronounced characters in wood or cortex, so that 
they are mostly only of supplementary value. The wood, as was first 
convincingly shown by Solereder (1885), affords a considerable 
