68 
tinued long enough. In this case the organisms are believed to grow 
through the pores of the filter. 
It is also evident that the passage of small particles through the 
pores of a filter depends directly upon the pressure used, and in all 
filtration experiments the exact pressure, whether positive or negative, 
should be stated. 
It is further necessary to call attention to the very great discrep¬ 
ancy in filters. We have made a careful study of various filters found 
upon the market and find that there is no satisfactory method by 
which they can be accurately graded, although we find in certain 
makes an attempt to graduate the power of the filter. Filters of 
course should always be tested under water with air pressure for pin 
holes and cracks, and also with small bacteria for permeability. It is 
only in this way that we may determine approximately what a par¬ 
ticular filter is capable of doing. 
The Berkefeld filters, made of diatomaceous earth, are more porous 
and variable than the Pasteur-Chamberland bougies, made of 
unglazed porcelain, which have finer pores- and are more constant in 
their ability to filter micro-organisms. 
After filters have been tested they must be dried and sterilized with 
the greatest care in order to prevent cracking, and should always be 
tested for porosity with microbes after filtration in order to insure 
this point. 
We have found in testing various filters that the weakest part is apt 
to be the joint, and that in any mechanical arrangement of the filter 
and flask there is the greatest danger of contamination and untrust¬ 
worthy results if either the liquid that is being filtered or any other 
fluid comes in contact with a joint. 
It will be noted that in the arrangement which we had for filtration 
by pressure (fig. 2), the fluid was simply passed into the Pasteur- 
Chamberland candle and withdrawn by means of a pipette in such a 
manner that contact between the two fluids was eliminated, and that 
no dependence was placed upon the security of any joints except those 
necessary to retain the air pressure. In the case of the small Berke¬ 
feld filter (fig. 3), in which the filtration was done by the pressure 
of the atmosphere produced by a vacuum, the joint between the filter¬ 
ing candle and the metal top was kept well out of the liquid, so that 
here again was avoided the possibility of error from this source. 
TESTING- OP FILTERS WITH OBJECTS OE “ULTRAMICROSCOPIC” SIZE, 
Four filters of the Pasteur-Chamberland system, letter B, were 
tested to determine whether they would allow particles of micro¬ 
scopic size to pass into the filtrate. 
At the outset we were confronted with the difficulty of finding a 
