extracted from different Kinds of Waters. 445 
corrected, and all the caufes of error hitherto unknown 
or neglected by the moft diligent obfervers, are avoided 
(which caufes fenfibly alter this kind of experiments) it 
is abfolutely neceffary to follow always a conftant and 
equal method, not only in the a£t of introducing the 
various kinds of air into the tube, but alfo after the mix- 
ing of the two kinds of air. The leaft variation of cir- 
cumftances caufes very great variations in the refults of 
the experiments, and thefe variations of circumftances 
are fo minute that I never faw any of the perlbns that 
obferved my experiments who could difcover them, al- 
though apprized of my delign. The negledt alone of this 
uniformity of operation may occalion an error of from 20 
to 50 parts and upwards in the experiment with common 
air; but with dephlogifticated air the error is incompa- 
rably greater, fo much that the fame quality of air a 
moment after may decompofe even a double quantity of 
nitrous air; fo that the pureft common air would appear 
to be noxious, and phlogifticated air; and the dephlogifti- 
cated air would appear lefs good, and even noxious : for, 
by the teft of nitrous air, it might appear little different 
from a mixture of dephlogifticated and phlogifticated 
air. 
I fhall take another opportunity to fpeak of all the 
particulars relative to my method ; but for the prefent I 
Vol. LXIX, N n n muft 
