82 UNIVERSITY OF WISCONSIN STUDIES 
This figure shows these 615 culturcs in three distinct groups 
based on the relation of the volume of carbon dioxide and 
hydrogen produced. By far the larger number produced 
carbon dioxide and hydrogen in almost equal amounts, thus 
giving a ratio of approximately 1:1. The average was 
CO,H.=1.06, with only an occasional culture giving less 
CO, than H, and comparatively few varying appreciably on 
the other side of the mean. It was observed that as the 
technique of the gas determinations became more refined, the 
variation in the gas ratio became less, and it is probable that 
the greater part of the variation shown for this group is in 
the determinations rather than in the fermentation. The 
volume of gas produced by this group is relatively small. 
From 10 ¢. «. of 1 per cent dextrose broth about 14 ¢. ¢. of 
gas was produced. Of the 350 cultures giving the 1.06 ratio 
only 3 liquefied gelatin. These will be considered later with 
the liquefying cultures. 
Distinetly separated from this group so far as the gas 
ratio is concerned is a second type characterized by the pro- 
duction of considerably more CO, than H,, by a wider range 
of ratio, and by a greater variation in the gas ratio of in- 
dividual cultures. Nearly all of the liquefying cultures be- 
long in this group. A study of the tables and figures im some 
of our previous papers will show that the actual amount of 
hydrogen produced is nearly constant throughout these two 
groups and that the increase in the ratio is brought about 
almost entirely by a larger produetion of carbon dioxide. 
In our first paper’ we suggested, as a possible explanation 
of the remarkable constancy of the B. coli ratio of 
CO,,/H,— 1.06 and the variation found in the aerogenes ratio, 
that the equal volumes of carbon dioxide and hydrogen were 
produced by a fermentation common to both groups while 
in the aerogenes group there occurs an additional and in- 
dependent fermentation producing carbon dioxide only. 
Clark has shown that while an increase of sugar in the 
medium has no appreciable effect on the relation of CO, to H, 
© Loc, cil, 
