250 Proceedings of the Royal Society 
heat would prove to be the agent which caused the curious motions 
which he had observed. I therefore placed the coin outside the 
window to be cooled in the east wind, and the rain which was 
falling plentifully. I found that the cold coin caused no sensible 
motion of the bubbles. I next heated the coin in a spirit lamp 
flame as hot as I could conveniently handle it. Its energy in 
moving the bubbles was now so greatly increased, that in some 
trials rapid motions were observed while the coin was still out of the 
field of view. 
Seeing that the bubbles thus moved towards the heated side of 
their cavities, I concluded that they ought to be repelled from a side 
which was cooled: and to try if such were the case, I cooled a florin 
piece in a freezing mixture of nitre, sal ammoniac, and water, to a 
temperature below 0° C. I had now the satisfaction to find that 
the cold coin, resting on the spar and brought up towards a cavity, 
sent the bubble away to the remote side of the cavity, just as the 
hot coin had brought it to the near side. 
It is clear, then, that the phenomenon is not due to a repulsion of 
the liquid in the cavity by the piece of metal, but is a consequence 
of the passage of a heat current through the liquid, the bubble 
always moving in a direction opposite to that in which heat is 
flowing. 
I found that metals possess no specific property in causing these 
motions. The bubbles moved on the approach of a silver coin or a 
copper wire. But similar motions were readity obtained when, in- 
stead of these, were substituted a heated rod of glass, a slender thin 
test tube containing hot water, or a piece of shellac moulded into a 
pencil shape, and still hot from the flame employed to soften it. 
All these substances — glass, water in the thin tube, and shell-lac — 
when cooled in the freezing mixture, repelled, or seemed to repel, 
the bubble, just as when heated they had attracted, or seemed to 
attract it. The temperatures in these experiments were not accurately 
observed, but they must have been as follows : — Coins taken from 
my pocket would be hotter than the air of the room, which was 
10° C. or 50° F. The coins being of silver, an excellent conductor 
of heat, would, when held in the hand, be hotter than the spar 
lying on the microscope stage in air at 10° C. The coins taken 
from the flame were as hot at first as could well be handled, and 
