624 
Proceedings of the Royal Society 
The substances to be experimented on were obtained in sheets, 
from which eight or nine circular discs 3 inches in diameter were 
cut, and piled one on another so as to form a cylinder. Between 
the first four discs thermo-electric junctions of fine copper and 
iron wires were inserted. A |-in. copper plate, with three small 
hooks on the circumference, was placed on either end, and the 
whole, having been slightly compressed in a Bramah press, was 
tied together with strings stretching from hook to hook. The 
cylinder thus prepared was surrounded with cotton wool, and placed 
horizontally in a wooden frame, with one end projecting about a 
quarter of an inch. Over this end a sheet of tinned iron was then 
drawn so as to screen all except the copper plate from the heat. 
The hitherto free ends of the thin copper and iron wires were 
attached to similar pieces of thick copper wire, and to insure 
equality of temperature were immersed in small vessels of water, 
placed in a larger vessel also containing water; the other ends of 
the thicker wires were then carried to the mercury pools of a com- 
mutator, so arranged that the junctions could be thrown singly, 
and in rapid succession, into the circuit of a Thomson’s dead beat 
mirror galvanometer of about 24 ohms resistance. A further resist- 
ance of about 30 ohms was also placed in the circuit. The source 
of heat was a large vessel of boiling water. From one side of this 
vessel, which was placed on a movable retort stand, a cylinder, with a 
flat end 3J inches in diameter, projected for about an inch and a 
half. 
The method of observation was as follows. The water being kept 
boiling, the vessel was applied for ten minutes with its flat surface 
in contact with the copper plate, then removed to a distance for ten 
minutes, then again applied for ten minutes, and so on during the 
whole of the experiment. After this had been continued for about 
two hours observations were begun. The galvanometer deflections 
for each of the three junctions were read every minute, the readings 
being taken from the coldest to the hottest; 15 s were taken to read 
the three. These readings were continued till two or three com- 
plete periods had been observed after the steady periodic state had 
been arrived at. The deflections thus obtained were plotted in terms 
of the time; and from the curves so obtained the necessary calcula- 
tions were made. 
