344 CHARLES G. ROGERS AND KENNETH S. COLE. 



filled with naphthalene and the lag of the couple was less than 

 twelve seconds. The auxiliary or secondary thermopile (b) had 

 five pairs of junctions and gave the temperature difference be- 

 tween flask II. and the surrounding bath. The e.m.f. developed 

 by the couples was measured by a "White" potentiometer. 

 This instrument, manufactured by Leeds and Northrup, gives 

 dial readings by single microvolts and is so arranged that the 

 resistance in the galvanometer circuit remains constant. This 

 fact makes it possible to read fractions of a microvolt with the 

 galvanometer. In this work one microvolt gave a deflection of 

 18.5 mm. on the galvanometer scale, so there were about 13,550 

 mm. per degree. With the high magnification telescope used it 

 was easily possible to estimate fifths of a millimeter on the 

 galvanometer scale. The galvanometer was of the D'Arsonval 

 type, also manufactured by Leeds and Northrup. Its resistance 

 was 13 ohms, period 5 seconds, and sensitivity io~ 8 amperes 

 per cm. with the scale a little over three meters distant. The 

 total resistance of the galvanometer circuit was 53 ohms, which 

 was an ohm more than the critical damping resistance. In spite 

 of this fact there were never any oscillations, even when the 

 sperm were introduced into the experimental flask, only what 

 seemed to be a very steady and somewhat rapid rise. The 

 galvanometer circuit was shielded as far as possible, as suggested 

 by White (10) to prevent the entrance of leakage currents into 

 the circuit. It was, however, impossible to shield the thermo- 

 piles effectively since their leads were carried through the running 

 sea water and very strange and erratic e.m.f.s were introduced 

 into the circuit if the shielding system for the potentiometer and 

 galvanometer had anything to do with the salt water or any of 

 the piping in the room. Care had to be taken not to use for 

 connections any wire that had been unduly bent or in any other 

 way maltreated so as to destroy its homogeneity. On the whole 

 the electrical apparatus worked very satisfactorily. 



It was our original plan to use Hill's ingenious application of 

 vacuum flasks to the twin calorimetric method (2) but it was 

 found that for our small flasks the heat loss did not remain 

 even approximately constant as the volume of the contents was 

 varied. It was also impossible to obtain accurate values for the 



