174 THE SCIENTIFIC PAPERS OF 



sistances to be measured, viz. : For resistances not exceeding 100 

 units, from 5 to 6 Daniell or Leclanche elements, which cause an 

 active decomposition without sensibly heating the coils or effecting 

 a partial insulation of the electrodes by excessive generation of 

 gases; for resistances of from 100 to 1,000 units, the number of 

 elements may be increased with advantage to 15 or 20, and a still 

 greater number of elements may be employed in measuring resist- 

 ances exceeding 1,000 units. 



It is not advisable under any circumstances to use less than five 

 Daniell's elements, although active decomposition may be obtained 

 with a less number, for the reason that the voltameter itself exer- 

 cises an opposing electro-motive force by polarisation, which may 

 vary under certain conditions from 1*1 to 1'3 Daniell's elements, 

 and that these variations would exercise a sensible difference in 

 the result if the electromotive force of the battery did not very 

 decidedly predominate. 



In using large battery power the heating of the coils has to be 

 guarded against, which may, however, be easily clone by arresting 

 the current, in reversing it, from time to time, whilst allowing 

 the gases in the tubes to accumulate until a sufficiently precise 

 reading can be obtained. From two to four minutes duration of 

 current will, under general circumstances, suffice to fill the tubes. 



4. The india-rubber pads should from time to time be smeared 

 with a waxy substance, to prevent escape of gas between them and 

 the edge of the glass tube, and I find that paraffin answers well for 

 this purpose. 



5. The state of the barometer has no influence upon the reading- 

 of this instrument, because fluctuations of the atmospheric pressure 

 affect both branches equally. A slight error through difference of 

 pressure would, however, arise if the reading of the instrument 

 were taken after the current had ceased to acfc, and the reservoirs 

 were to remain in their elevated position opposite the zero point 

 of the scale, exercising a hydrostatic pressure equal to the depression 

 of the liquids in the tubes. In order to eliminate this source of 

 error, the two moveable reservoirs must be lowered until a balance 

 of levels is established on each side between the tube and its reser- 

 voir before the reading is taken. This being done, the weighted 

 lever is raised from each tube for the discharge of the gases, and 

 the moveable reservoirs are raised back to their zero position. 



