59-1 
Proceedings of the Royal Society 
of contact between the intervening sheet of tinfoil and tlie surface 
of the ice may not have been the same during two successive 
periods. 
Jl J 2 
Ji J 2 
Ji J 2 
Ji J 2 
277 
42 7 
40*8 
3-00 
27*4 
29-6 
34-4 
28-9 
28-3 
451 
38-8 
30-0 
27-4 
30-5 
337 
28-8 
29-6 
481 
371 
29'4 
27-3 
31-5 
331 
28-2 
31-5 
50-6 
35 7 
29-2 
27-3 
32-4 
32*4 
27-85 
337 
51-9 
33-5 
28-8 
27-5 
33-6 
31-2 
27-6 
35-8 
507 
327 
28-65 
27-9 
34-4 
30-5 
27-3 
38-0 
471 
31-8 
28-5 
28-5 
351 
30'0 
27-1 
40-2 
43-5 
311 
28-5 
29'0 
1 
34-8 
29*4 
27-05 
Galvanometer zero, 29 '0. 
Detailed above are a set of readings taken during one period — 
one of the few in which the changes were practically periodic. 
The junctions are numbered in order, beginning with that nearer 
the surface. The interval between two readings of the same junction 
is 20 seconds, and they were taken alternately. 
Plotting these readings (which express differences of temperature) 
against time, a curve is obtained, the equation to which is 
f(0) = A 0 + Aj cos 6 -r A 2 cos 2 6 + A 3 cos 36 + . . . . 
+ Bj sin 0 + B 2 sin 2 0 + B 3 sin 3 0 + . . . . 
where 0 is an angle which increases in proportion to the time, and 
where A 0 , A x , A 2 , B 1? B 2 , . . . &c., are unknown coefficients. By 
taking six values of f(Q) we may determine the coefficients A 0 , A ]? 
A 2 , A 3 , B x , B 2 , B 3 . Subtracting from each value of / ( 0 ), A 0 , the 
mean value of the function, we get 
7] = Aj_ cos 6 + . . . 
+ B : sin 6 + ... Then putting P x = jAf + Bf, 
this may be written 
7] = P : cos 0 4- sin o) + &c. 
\Pi B 1 J 
