192 
by sweeping the floors. On the other days it was requested 
that everything should be kept still. This of course brings 
in a practical difficulty and limits the use of the test to 
cases where care can be used and thoughtful observation, 
since there are many ways by which dust may be made to 
interfere, even although the act of sweeping should not take 
place. The house experiments gave similar gradations. 
The result seems to be that a piece of glass, of a definite 
size, hung up in any place, will receive deposits of ammonia, 
or substances containing ammonia, in a short time, and by 
washing the ammonia off with pure water, and testing it 
with a Nessler solution, it may be seen whether there is too 
much or not. It is the simplest test for ammonia yet found. 
Its discoverer deserves great thanks. It must not be forgotten 
that we may have ammonia in very different conditions ; 
it may be pure, or it may be connected with organic matter. 
This mode of enquiry is better suited as a negative test to 
show that ammonia is absent, than to show what is present. 
When ammonia is present there may be decomposing matter, 
when absent there is not. I am hoping to make this a ready 
popular test for air — a test for sewer gases*, for over-crowd- 
ing, for cleanliness of habitations, and even of furniture — as 
well as for smoke and all the sources of ammonia. Of course 
it must be used with consideration, and the conclusions 
must not be drawn by an ignorant person. 
How far it may be used as a test of climate is a matter to 
be considered also. 
After this I made another series of trials with air, 
Nesslerising the washings at once and not after laborious 
distillings, as in former cases; the results are very valu- 
able, showing that we obtain comparative quantities equally 
in this way. 
The amount of ammonia obtained in this ready way does 
not give exactly the same results as the more laborious 
methods which I have used, but it may be taken as the 
* For sewage also to a larger extent than it has yet been used. 
