OF ARTS AND SCIENCES. lo3 



In order that the electromotive force of the current, and tlierefore 

 the deflections of the galvanometer, should be directly proportional to 

 the excess of temperature of the point touched over that of the air, 

 a pair of metals must be taken whose neutral point is at a very high 

 or vei-y low temperature, and whose relation to each other varies the 

 least possible with the temperature. Iron and German-silver satisfy 

 these conditions remarkably well. Their neutral point is at 1354'^, 

 as deduced from Prof. Tait's table. Some rude experiments of ours 

 on the relation between the temperature and the deflections gave quite 

 ragged curves ; but they were evidently in every case straight lines. 

 The three lines obtained by heating and cooling water were nearly 

 parallel. The mean of these experiments showed that one degree of 

 tempei'ature corresponds to 4.3 scale-divisions. One microvolt corre- 

 sponds to 2.6 scale-divisions, or one scale-division to .39 microvolts. 

 We thus find that one degree of change in temperature produces an 

 electromotive force of 18.3 microvolts. The value as calculated by 

 Prof. Tait's formula is 21.8. 



Our first experiments were with two thermopiles. The method of 

 work was this. Each thermopile consisted of one joint, which was 

 fitted to a piston moving vertically 3 cm. in a brass cylinder. The 

 piston was depressed by a weight resting on a lever, whose fulcrum 

 was on the frame around the table, and its end at the thermopile. 

 The average pressure applied was 9 kilos. This was oj)posed by the 

 reaction of springs, which raised the piston when the weight was re- 

 ' moved. The cylinders were placed on carriages, running on wooden 

 strips, which crossed the table transversely just above tlie infusorial. 

 These strips and the sides of the table were graduated, and thus we 

 got the (cc, y) co-ordinates of the points touched. 



One thermopile was kept in a fixed position, the other was moved 

 about to various points where equal deflections were obtained. All 

 these last points lie on an isothermal. Great care is necessary that 

 the contact with the plate be good, and that the piston move in an 

 exactly vertical direction. The results thus found were unsatisfactory, 

 because, as before pointed out, an inappreciable error in observation 

 produces a very perceptible change in the equation of the curve. An- 

 other objection is that the method requires a great expenditure of 

 time. 



A second method is more satisfactory. One thermopile was moved 

 lengthwise along the plate, and the deflections laid off in a curve. 

 The same may also be done transversely, but the curves obtained are 

 of little value. 



