390 
Proceedings of the Royal Society 
■ 
Fence, for 0, we have the differential equation, 
A© ~ 4- i [i sin 0 + A 2B E € sin 2$J =0 . (16.) 
which shows that the ring oscillates rotation ally according to the 
law of a horizontal magnetic needle carrying a bar of soft iron 
rigidly attached to it parallel to its magnetic axis. 
When 6 is and remains infinitely small, 0, y, and y are each 
infinitely small, x remains infinitely nearly constant, and the ring 
experiences an oscillatory motion in period 
2 ,_ b@ 
V [I + (A - + Ax) ' 
compounded of translation along OY and rotation round the dia¬ 
meter D. This result is curiously comparable with the well-known 
gyroscopic vibrations. 
3. Laboratory Notes. By Professor Tait. 
1 . On Thermo-electricity. 
Messrs J. Murray and J. C. Young have been carrying out 
experimentally the idea mentioned in my former note on this 
subject. ( Proc . Dec. 1870.) Their first sets of observations, of 
the results of which I subjoin a specimen, were made with an 
iron-silver and an iron-platinum, circuit working opposite ways on 
a differential galvanometer. The resistances (including the galva¬ 
nometer coils) were in this particular experiment 53 - l and 25*9 
B.A. units respectively, so that hut very slight percentage changes 
could he produced in them by the elevation of temperature of the 
junctions. As one of a number of closely agreeing preliminary 
trials the result is extremely satisfactory, though the exact adjust¬ 
ment has not yet been arrived at. To show the parabolas due to 
the separate circuits, and thus exhibit the advantage of the method, 
1 have requested the experimenters to break the circuits alter¬ 
nately after taking each reading of the complex arrangement, and 
take a rough reading. The last four columns of the table give 
the results; but, as the temperatures were probably slightly different 
from those in the first columns, no very direct comparison can be 
instituted. A glance at the 4th, 6th, and 8th columns, however, 
shows how nearly a linear relation between temperature-difference 
of junctions and galvanometer deflection has been arrived at in the 
