PRELIMINARY TREATISE. 
31 
difference in the attachment of the anther, and exactly what 
variability of anther is consistent and what is inconsistent 
with identity of kind, and I believe it to be the most difficult 
point that botanists have to investigate. The insertion of 
the filaments appears to be of the highest importance, their 
more or less continued adhesion to the tube, when not abso- 
lutely inserted, to be of very little weight. Whatever may 
be the truth in these respects, it is evident that a distinct view 
of the relative importance of the several differences that exist 
between vegetables, in ascertaining their generic identity and 
in the disposition of natural groups, is the great desideratum 
to give stability and consistence to the labours of botanists. 
On minute examination of the riches of the several her- 
bariums which have been opened and entrusted to me by the 
liberality of their possessors, the difficulty of deciding whe- 
ther or not to consider individual specimens as local varieties 
of one species, is greater than when a more perfect know- 
ledge of their peculiarities is obtained from the sight of living 
plants. The mode in which I propose to surmount this 
difficulty is very simple, and I think its advantage will be 
felt, and that it will be universally adopted. I suggest that 
in every botanical work varieties should be specified as be- 
longing to one of four characters, viz. 1. local; 2. acci- 
dental; 3. cultivated ; 4. hybrid. 1. Varietas, i. e. var. loci; 
2. var. fortuita ; 3. var. hortensis; 4. var. hybrida. That 
the system of marking varieties by letters of the Greek alpha- 
bet (which the unlearned cannot read and pronounce, and 
when spoken give the very unsatisfactory names alpha, beta, 
gamma, &c.) should be abandoned as unmeaning and incon- 
venient, and each local permanent variety distinguished by 
a Latin name not found in that genus or in any other genus 
so nearly allied as to make their union not quite impossible, 
and that such name should agree grammatically with the 
generic name; and, further, that closely allied species should 
be placed in groups headed by their most striking features 
of agreement. It will then appear that it makes little diffe- 
rence whether they are looked upon as varieties of one or as 
several allied species, as their affinity will be equally shown 
and the peculiar name would stand the same in either case. 
Thus I find a race of Crinum, occupying the coast of South 
America, distinguished by a short column, a stoloniferousbulb, 
diverging leaves with a rough margin, flowers from 4 to 8, 
oftener 6, and cinerascent anthers, which were the features of 
