COLLECTORS’ NUMBERS. 137 
that it is useless for any object to occupy space by citing the 
umber. But it is still more disheartening, after getting together 
numbered) to ag included in the binominal symbol of that list. It 
is troublesome to find the list used; difficult (and rarely worth 
while) to aie what were the supposed limits of the species in 
that list 
The outcome is that, in the case of Europe, I oe find it 
os to decide what is the geographic area of some comm 
-defined species, say, Eleocharis seuistrrode Smith, even to 
within a ie error of 200-400 miles. I can of course determine 
the area by the arm-chair-and-coffee method—by SOP, compiling 
the authorities. But the best authority is not to be trusted one 
inch in such a matter. 1 should not wish to state that " ies 
multicaulis grew in the Atlas unless I had seen a specimen collected 
there, and I should wish then to cite that specimen with the 
collector’s name and genuine field-number. As a matter of fact, a 
very considerable peroentage of FE rat multicaulis was (six years 
ago) named wrongly in Kew and South Kensington, and I need 
therefore add no further priitaer cor it was named elsewhere. 
But, says Mr. J. G. Baker, ‘‘ you work on critical bese e sedges, 
xed numbers 
is exceptional.” To which I reply, are your Crocus pre Tris less 
critical, and are your ferns better numbered ? The South American 
and Indian ferns are much better numbered ieomge the European. 
There are not many plants in herb. H. C. Watson that have a field- 
from one tree, the sheets are numbered 2383 A, 2383 B. And 
small blasts, collected at one time and place (I always 4:8 he 
my own hands), are similarly numbered 2384 B 
I felt morally sure that : had exactly the same for rm. But aa a 
male tree was supposed to belong to a female hard by, I always 
gave them different aes and added a note that t ends 
them one species. I do not know that I could, even from lon 
experience, greatly improve on this plan; I believe it wou uld be 
better, after reaching 9999, to begin at 1 over again; five digits 
cause a sensible delay, in transfer and gr over four digits; as 
explained above, the having two numbers 2773, referring one to 
we the other to something widely different, would lead to no 
sok. 
I have generally cut out from my present work all citations of 
my own sees ge numbers; for, as I issue my p pe nam 
way, no person who gets one can ever be assisted by the 
citation of the number. I am led thus to the curious -eGhintison 
