COMPARISON OF PLATINUM AND C;.\s THKUMMMKTKR8. 47 



gliuss. Provision was mode in the outer space of the tank for a regulator and heating 

 arrangement underneath. This was, however, not used during the present experi- 

 ments. Suitable thermometers indicated the temperature of the water in the outer 

 tank, and two sensitive ones, divided to tenths of a degree with thin hulbs and their 

 stems bent at right angles, indicated the temperature of the internal coil-space. 



IX. GALVANOMETER SHUNTS AND BATTERY RESISTANCES. 



The resistance-box was provided, as in the Callendar-Griffiths type, with a set of 

 galvanometer shunts and a series of battery resistances of 20, 50, 100, and 500 ohms. 

 It was afterwards found a great advantage to have a more exact adjustment of the 

 battery current, and for this purpose a subsidiary three-dial box, working up to 

 1,000 ohms, was provided. 



Previous experience at Kew had shown the occurrence of differences in the point of 

 balance, according as the battery-current was in one direction or the other ; it was 

 found that the difference between the readings with the current in the two directions 

 generally increased gradually during the first quarter of an hour on commencing the 

 observations, and was greater the greater the intensity of the battery current. In 

 order to be able to study, and if possible to eliminate this cause of uncertainty, we 

 placed in the battery circuit of the new apparatus a high-insulation reversing switch. 

 The working of this switch was at first unsatisfactory, but was subsequently perfected 

 by short-circuiting the rubbing contacts at the pivots by flexible brass strips, and 

 covering the five contact studs with thin platinum plates. 



X. BATTERY POWER EMPLOYED. 



The battery used throughout the experiments consisted of two dry-cells of the 

 Olwich type, obtained from Messrs. SIEMENS; we ascertained that the E.M.F. of the 

 two cells was practically constant throughout and about 2 '8 volts, and that the 

 internal resistance of the two in series did not rise to more than 1 ohm, changing by 

 a (jiiaiitity quite negligible in comparison with the large resistance always added in 

 the battery circuit. 



XI. THERMOELECTRIC KEY. 



For the completion of the different circuits a Griffiths' thermoelectric key was 

 employed, as in the first Kew apparatus. The essential feature of this key consists 

 in the addition to the ordinary form of double bridge-key of a lever so arranged that 

 when the key is released the galvanometer circuit remains made. Thus a simple 

 depression of the key first breaks the circuit of the galvanometer, then makes that of 

 the liattei v. and finally remakes the circuit of the galvanometer. 



The key we used was somewhat modified from the original pattern, which, being 



