New YorK AGRICULTURAL EXPERIMENT STATION. y | Ast 
corded as 10 per ct. or less. Owing to the rounded form of 
the upper end of the fermentation tube it is difficult to esti- 
mate quantities under 10 per ct. with any considerable accu- 
racy. Since practically all of the determinations were under 
this amount an accurate measurement was not attempted at 
the New York laboratory, but fortunately this was practically 
always done at the Vermont laboratory and the resulting 
measurements are given in the table. 
A striking fact brought out by this table is the frequent 
failure to form gas which occurred in fermentation tests made 
with the same organism at different times. So marked was 
this tendency to vary that contradictory results were obtained 
with 21 of the 43 strains which were studied. This number 
would undoubtedly have been even larger if all of the strains 
had received an equal amount of study. Those with which the 
action on each of the sugars was tested on but two occasions 
make up 13 of the 22 strains among which variations were 
not noted. Taking the results as they stand there were 91 
out of a total of 550 tests which did not show gas where it 
was found at other tests. Accordingly, when making a fer- 
mentation test in triplicate with one of these cultures the 
chances were over 16 per ct. that the conclusion reached was 
diametrically opposed to the truth, provided we accept the 
contention that a positive result is of more value than a nega- 
tive result, in fermentation test, proper care being used to 
prevent outside contamination and to provide uniform condi- 
tions. If this chance of error is considered as restricted to 
the strains with which variation was actually observed the 
above record does not express the actual gas-forming ability 
of the organism in 91 out of 347 observations, or 26 per ct. 
The influence of the revivifying process on the certainty of a 
correct result is shown by comparing the error noted when 
this preliminary treatment was given with the error when it 
was omitted. Where the cultures had not been previously 
revivified, but inoculations were made into the fermentation 
tubes from young, actively-growing agar slopes, there was an 
error in 63 out of 189 observations, or 33 per ct.. Where the 
cultures were given a preliminary cultivation as already ex- 
