[354 a ible 
XXVI. Sequel to the Thoughts on the conflituent Parts of Wated 
and Dephlogifiicated Air. In a fubfequent Letter from Mr. 
James Watt, Eagincer, te Mr. De'Lucy B. Ros: : 
Read May 6, 1784. 
Fi : Birmingharsy 
pt AR oS TR, April 30, 1754. 
N re-confidering the fubje& of my letter to you of the 
26th of November laft, I think it neceflary to. refume 
the fubject, in order to mention fome neceflary cautions tothofe 
who may chufe to repeat the experiments mentioned there, and 
to point out fome circumftances that may caufe variations in 
the refults. 
In experiments where the dephlogifticated air is to be dif- 
tilled from common or cubic nitre, thefe falts thould be puri- 
fied as perfectly as poflible, both from other falts and from 
phlogiftic matter of any kind; otherwife they will produce 
fome nitrous air, or yellow fumes, which will leflen the quan- 
tity, and, perhaps, debafe the quality of the dephlogrfticated 
air. If the nitre is perfectly pure, no yellow fumes are per- 
ceptible, until the alkaline part begins to aét upon the glafs of 
the retort, and even then they are very flightly yellow. 
When earthen retorts are ufed, and a large quantity of air is 
drawn from the nitre, it aéts very much upon the retort, dif- 
folves a great part of it, and becomes very alkaline, retaining 
only a {mall part of its acid, at leaft only a {mall part which 
6 can 
