METHODS OF STUDY _ 13 
under consideration. The resulting numbers are then 
ranged in order so that a precise view of the numerical 
characters of a large number of specimens can be 
readily obtained. By the use of such methods valu- 
able information is often to be arrived at. But the 
same limitation affects them as in the preceding case. 
So that the only way in which we can come to a 
definite decision as to whether a given putative 
species does or does not represent a definite and con- 
stant type is by resorting to our third method, that 
of sowing its seeds and actually rearing its progeny. 
And this is not so simple a matter as might appear 
at first sight, for a great many precautions have to 
be taken. Thus we must separately sow the seeds of 
many different individuals of the species which we 
are examining, so as not to base our conclusions upon 
a few experiments only. But in many cases, even 
when this has been done, we should only know one of 
the parents of our seedlings—that is to say, in cases 
where the pollen for fertilization may possibly have 
been conveyed by natural agencies from a different 
plant. In such a case we must either ensure self- 
pollination by isolating our plants, or we must arti- 
ficially provide pollen from a separate known parent. 
If under these circumstances a particular group of 
plants preserves the characteristic differences which 
distinguish it from another group which has also been 
grown for a number of generations under the same 
conditions, we have at last reasonable grounds upon 
which to base the opinion that we are dealing with 
two distinct physiological species; even though the 
