of Edinburgh , Session 1881-82. 
807 
2. On the Change in the Peltier Effect due to Variation in 
Temperature. By Albert Campbell. Communicated by 
Dr. Knott. 
The series of experiments here described had for their object the 
measurement of the variation in the Peltier effect due to change of 
temperature ; they were made during the last six weeks. The 
arrangement which I used in these experiments was suggested by 
a form of apparatus used in Professor Tait’s laboratory ; the follow- 
ing is a description of it. A plate of sheet lead (say) is bent 
into the shape of an arch (fl-shaped), and to its lower edges are 
soldered two plates of sheet-iron, to the opposite edges of which 
are soldered three or four copper wires proceeding to mercury-pools, 
which can be put into connection with a battery. In the trenches 
between the iron and lead are inserted the opposite ends of an iron- 
German silver thermopile of twelve to twenty junctions ; these 
junctions are placed so that their points may nearly touch the 
junctions of the iron and lead plates, from which they are insulated 
by a thickness or two of thin paper. The whole is then wrapped 
tightly in cotton wool and pushed into a tin box, which is rolled 
round with flannel and placed in a larger tin box containing boiling 
water, an arrangement which gives a sufficiently uniform tempera- 
ture in the inner box. The thermopile of iron and German silver 
wires has two advantages ; first, these two metals give, for a given 
difference of temperature, a greater deflection than any other pair 
of easily obtainable metals ; secondly, the deflection is very nearly 
proportional to the difference of temperatures. 
The following is my method of working : I connect the thermo- 
pile with the galvanometer, turn on the battery current for thirty 
seconds (say), and note the galvanometer deflection at the end of 
this time; for the next thirty seconds the current is allowed to 
remain off, and at the end of this period the deflection is again noted ; 
the current is then turned on for thirty seconds in the opposite 
direction, the deflection taken, the battery left off for another thirty 
seconds, the deflection again taken, and so on as at the beginning. 
This cycle of operations is gone through a number of times in order 
to get an average result. 
