a new Eudiometer. 1 27 
always taken to obferve the diminution which nitrous air buf- 
fered by being fhaken in the water, as mentioned in p. 115. 
The heat of the water in the tub alfo was commonly let down. 
Mod. of the bottles were tried only in the firft method ; but 
l'ome of them were alfo tried by the fecond, and by the method 
juft deferibed in the manner of fontana. 
The refult was, that the teft of the different bottles tried on 
the lame day never differed more than .013, and in general not 
more than half that quantity. The teft, indeed, of thole 
tried on different days differed rather more ; for taking a mean 
between the tefts of the bottles tried on the fame day, there 
were two of thole means which differed .025 from each other ; 
but, except thofe two, there were none which differed more 
than .013. Though this difference is but fmall, yet as each of 
thefe means is the mean of feven or eight trials, it is greater 
than can be expended to proceed from the ufual errors of the 
experiment. This difference alfo is not much diminilhed by 
eorreffing the oblervations on account of the heat and abforb- 
ing power of the water, according to the rule in p. 1 18. This 
might incline one to think, that the parcels of air examined 
on lome of thofe days of trial were really more dephlogifti- 
eated than the reft ; but yet, I believe, that they were not : 
for whenever there was any conliderable difference between the 
means of two fucceftive days of trial, there was nearly the 
fame difference between the tefts of the two bottles of the 
very fame air tried on thofe two days. For example, the 
mean of the trials on fifty 7. was .016 Ids than that of thofe 
on the 15th of the fame month; but then the teft of the air 
caught and tried on the 7th was equally lefs than that of the 
air of the fame day tried on the 1 5th ; which lhews, that this 
difference between the means of thofe two days was not owing 
to 
