Brass . 
101 
mixture again, and immediately pour it into cold water* 
Then in the second melting (with as little heat as will suf- 
fice) take for every 16 parts of the composition 1 part of 
white arsenic, wrap it in paper, thrust it to the bottom of 
the fluid metal, and stir with a wooden rod till no more 
arsenical fumes escape, immediately after which cast the 
metal in a sand- mould* Then while it is still red-hot lay 
it in a pot full of very hot embers and cool it very slowly : 
unless this precaution is particularly observed, the metal 
will fly in pieces when cold, sometimes even long after 
all such danger is thought to be past ; or it will split in 
the polishing* For the particular manner of constructing 
the mould, and the whole of the nice and laborious opera- 
tions of grinding and polishing to a perfectly true figure, 
the reader is referred to the two papers above mentioned,* 
Both the brass, and the silver, and the arsenic appear to 
have their distinct use. The brass makes the mixture 
tougher and not so excessively hard and brittle. The sil- 
ver improves both the texture and colour, but is not 
an essential though really an useful addition. The arse- 
nic is found by actual comparison to make the metal fi- 
ner, and particularly closer in texture, and therefore less 
liable to be porous. It sensibly increases the specific gra- 
vity, which, before the arsenic is added, is 8.78, but af- 
terwards, 8.89. It is added in the second melting, that as 
little as possible may be dissipated in vapour. A greater 
proportion would make the metal liable to tarnish. An 
alloy containing 6 of copper, 2 of tin, and 1 of arsenic, is 
nearly the proportion of Sir I. Newton’s specula, which 
is very good, but polishes somewhat yellow. 
The separation of copper from bell- metal and all the tin 
alloys in the large way, happened to be an object of consi- 
derable importance in France in the midst of the revolu- 
tionary war, when the importation of copper was nearly 
impracticable, and a large quantity of it was required for 
f And to a paper in part extracted a few pages hence. T C, 
