124 Mr. cavendish's Account of 
From the refult of thefe experiments I am perfuaded, that 
my method of mixing the airs is really rather more accurate 
than fontana’s, as in trying the fame bottle of air fix or 
feven times in my method the different trials would not often 
differ more than „4-_.dth part, and very feldom more than 
^4-^dth ; whereas in his there would commonly be a difference 
of .j^dth, and frequently near twice that quantity, though I 
endeavoured to be as regular as 1 could in my maimer of trying 
the experiment. My method alfo certainly requires lefs dex- 
terity in the operator than his. 
It is of much importance towards forming a right judgement 
of the degree of accuracy to be expected in the nitrous tefl, to 
know how much it is affected by a difference in the nitrous air 
employed. Now it muff be obferved, that nitrous air may 
differ in two refpects ; firft, it may vary in purity, that is, 
in being more or Ids mixed with phlogiflicated or other air ; 
and, fecondly, it is poffible, that out of two parcels equally 
pure one may contain more phlogifton than the other. If it 
differs in the fecond refpeCt, it will evidently caufe an error in 
the tefl, in whatever proportion it is mixed with the refpirable 
air ; but if it differs only in the fir ft refpedt, it will hardly 
caufe any fenlible error, unlefs it is more than ufually impure, 
provided care is taken to ufe fuch a quantity as is fufficient to 
produce the full diminution. This has been obferved by the 
Abbe fontana, and agrees with my own experiments; for 
the tefl: of common air tried in my ufual method, w r ith fome 
nitrous air which had been debafed by the mixture of common 
air, came out only 18 thoufandths lefs than when tried with 
air of the befl quality, though this air was fo much debafed 
that the diminution, on mixing two parts of this with five of 
common, was one-bxth part lefs than when good nitrous air 
was 
