as a Source of small Constant Currents. 359 



southern aspect, and therefore not subject to any sudden 

 change of temperature. Light was admitted through a 

 sheet of paper forming a window in otherwise closed shutters. 

 The temperature lay between 14° and 16° C. during the 

 winter months, in which the experiments were made. At the 

 close advantage was taken of a hot day to repeat the investi- 

 gation at 23° 0. 



The galvanometer generally used had a resistance of about 

 six thousand ohms, and a sensitiveness of about four scale- 

 divisions for 10~ u amperes by direct test. This was some- 

 times supplemented by another galvanometer of nearly six- 

 teen thousand ohms resistance, and whose sensitiveness could 

 be brought up to several scale-divisions for 10 -13 amperes. 



The smallest change of P.D. usually noted was that com- 

 pensated by altering the resistance in the compensator by 

 0*1 ohm corresponding to 0*000045 volt. 



In addition to the apparatus requisite for finding the 

 terminal P.D. of any cell at rest, resistances were required for 

 the purpose of short-circuiting the terminals of the cell. For 

 this purpose a very large box of German-silver resistance- 

 coils was used. All the coils had been previously compared 

 with standards by Carey Foster's method, and were in fact 

 adjusted by this method in the laboratory. The corrections 

 are in no case large enough to require taking into account. 

 The insulation of the various portions of the apparatus was 

 carefully attended to by the use of blocks of paraffin and was 

 very necessary. 



For the determination of the most probable value of the 

 E.M.F. of any cell, comparisons of all the cells were made by 

 the method mentioned, and the mean value was taken as 

 identical in absolute measure with Lord Eayleigh's mean 

 value, i. e. 1*43500 Eayleigh volt at 15° C. Many complete 

 comparisons were made during three months, partly to find 

 the mean and partly to check the action of each cell. The 

 large cell and hospital cell were compared with the small ones, 

 sometimes directly and sometimes through the intermediary 

 of the master cell. The changes of temperature occurring 

 were too small to affect the difference between the cells within 

 our limits of observation. One hundred and sixty com- 

 parisons were used in computing the probable values of the 

 E.M.F. of the large cell and of the hospital cell at rest. 



Nothing could have been more satisfactory than the be- 

 haviour of all the cells, though they were badly used between 

 whiles. Suffice it to say that the result was almost the same 

 whatever cell was used to oppose the others ; and throughout 

 the months of June, July, and August the mean E.M.F. of 



