FOR MEASURING THE EXPANSION OF SOLIDS. 
453 
reduced almost to a point, and was very considerably thickened at its lower 
end, and moulded to the bottom of the register, as if it had been partially 
fused. It was hard and brittle. The vapour of the mercury had probably 
combined with it at some temperature below the boiling point ; the amalgam 
so formed had flowed down to the bottom of the bar, and the mercury was 
afterwards expelled by the boiling temperature. 
I may here observe, as not unworthy of attention, that in no instance have 
I seen a metal acted upon by the vapour of mercury at its full boiling tempe- 
rature ; — even gold, which has so strong an affinity for it, comes out of it with 
its yellow colour perfectly unstained ; but when the mercury is in the fluid 
form at the same temperature, the gold is immediately dissolved by it. 
Under these circumstances there certainly may exist some doubt whether 
the full amount of expansion in zinc to the boiling point of mercury was pro- 
perly registered. 
On the other hand, in confirmation of the result so recorded, it may be seen, 
in Table XIV. of the expansion of the alloys, that a composition of half cop- 
per and half zinc presents the same anomaly ; the expansion for the 600° to 
boiling mercury is not quite four times that of the 150° to boiling water. In 
the alloy of three fourths copper to one fourth zinc, the rate of expansion in- 
creases in a small degree; and in common brass, where the proportion of zinc 
is still less, it increases still more rapidly. 
My purpose in instituting these experiments upon the alloys, was to observe 
the relation which might exist between the expansions of the pure metals and 
those of their mixtures : and the better to illustrate any such, I made alloys of 
copper with known multiple proportions of zinc and tin. I shall here present, 
in a tabular form, the temperatures of their melting points, as derived from 
their expansions to the boiling points of water and mercury ; as, although I 
am not able to compare them with results directly obtained by immersion, we 
can judge, by comparison with the similar calculation of the pure metals, within 
what limits any error is probably confined. 
