78 UNIVERSITY OF WISCONSIN STUDIES 
elassifivation. The hopelessness of establishing a stakle classi- 
fication by using only the fermentations of various substances 
is shown by the large number of proposed schemes, some of 
which permit over one hundred possible varieties. The one 
which has been most generally followed was proposed by 
MacConk+y and is merely the possible combinations of the 
plus and minus signs under two test substances. The four 
species permitted by this arrangement may, by applying 
similar methods, be further divided into varieties, limited 
only by the number of test substances used. A classification 
of this kind may be convenient in that it adapts itself readily 
to a workable key but there is no evidence to show that it has 
any relation to the evolution of the group or that it serves 
in any way to indicate the origin of the culture. 
It should be possible to discover some combination of char- 
acters which separates the group into natural species. Since 
the evolution of a bacterial species is largely a question of 
environment, the species so established should be coincident 
with more or less well-defined habitats. For taxonomic pur- 
poses it may safely be assumed that the nature of the fer- 
mentation is of more significance than the particular material 
fermented. 
THE CHARACTERS STUDIED 
The characters of a group of bacteria can be properly 
established only by studying a considerable number of cul- 
tures. The value of this method of study has been so com- 
pletely demonstrated by some of the recent work on systematic 
bacteriology, particularly by the work of the Winslows on the 
Coceaceae, that its use should need no further justification. 
In attacking this problem we have so far as possible used 
exact chemical methods in determining the reactions produced 
hy the bacteria in various culture media. This has limited 
the number of cultures which could be studied but has per- 
mitted distinctions on the basis of quantitative measurements 
which have been of the greatest value. The details of the 
methods followed have been given in the papers cited and 
for the sake of brevity will not be repeated here. The enl- 
