350 
boiled and cooled distilled water. [It was again examined 
on April 3, when it contained no organic forms whatever.] 
February 7.—A pyzmic abscess of the elbow joint was 
opened ; a fullstream of pus issued from the incision. Several 
large but still capillary tubes were then filled by inserting 
their open ends into the track of the bistoury. The tubes 
were immediately sealed, and the contents used the same day 
to impregnate a charged eprouvette. After a few days the 
test liquid was teeming with bacteria. In this case the 
knife was not previously immersed in boiling water, but the 
discharge of pus from the wound was so copious that I do 
not think there is the slightest doubt that the quantity used 
was collected without any contamination, whether arising 
from this source or from the surface of the skin. 
XX VIII.—The collection of blister fluid is attended with 
much greater difficulties than that of pus, for it is almost 
impossible to abstract it from the vesicles without risk of 
contact with the surface of the skin. It can be best obtained 
by opening both ends of a collecting tube, and then intro- 
ducing the capillary end into a vesicle after first snipping the 
epidermis. This done, the liquid must be drawn into the 
tube by suction. Liquid thus collected was used as 
follows :— 
January 10.—Blister fluid was added in the usual propor- 
tion (one drop to one cubic centimetre) to a charged eprou- 
vette. For a long time the liquid remained clear, but even- 
tually bacteria appeared in small numbers. 
February 13.—Blister fluid from another source was used 
in a similar manner and with a similar result. 
March 27.—The same experiment was repeated with dif- 
ferent fluid, but in this case the eprouvette was kept in the 
incubator. The development of bacteria was much more 
rapid. On the same day another quantity of the same liquid 
was diluted with boiled and cooled distilled water in a super- 
heated eprouvette and also placed in the incubator. [In a 
few days it became turbid and swarmed with bacteria.] 
The equivocal results of these experiments are to be 
attributed entirely to the difficulty of obtaining blister fluid 
pure, that is, to accidental contamination in the process of 
collection. 
1 This important experiment could not be repeated, for an attendant who 
entered the laboratory in my absence carelessly destroyed all the tubes 
excepting the one which had been already used. The single result was so 
satisfactory that I myself entertained no doubt of its significance. 
