produced by the Electric Current. 473 



arbitrary units, the divisions of the scale, which were observed 

 by reflection in the mirror of the instrument. 

 Calculation, therefore, gives : — 



Nitrate of silver. 



Tube No. 1, height 1-6 metre ... S 1= = 8-32 

 „ „ 2, „ 3-6 „ ... S 2 = 16-22 



Iodide of cadmium. 

 Tube No. 2, $ 2 = 43\L0 [£8 2 = 21-55.] 



On passing an external current through my tubes, I soon 

 perceived that the polarization which it produced in the elec- 

 trodes was so intense and so variable that it was impossible to 

 obtain any result by this method. I therefore resolved to 

 employ instead the current given by the tube itself in virtue 

 of the inequality of the electrodes, their polarization, &c. By 

 precautions, some of which have been described above, this 

 current could be rendered constant and of the same order of 

 magnitude as the current due to the electromotive force e. 



In the series of experiments given in the following Tables 

 (pp. 474, 475), I observed alternately the intensity of the pro- 

 per current of the tube when ascending and when descending 

 through the liquid. Tables I., II., and III. refer to nitrate 

 of silver. Columns 1 and 2 of these Tables contain the elon- 

 gations of the galvanometer on the current in the latter being- 

 changed in direction by a commutator ; column 3, their dif- 

 ference, which measures the intensity of the current. A de- 

 signates the ascending, D the descending current. 



Table IY. refers to iodide of cadmium. Considering the 

 intensity of the proper current of the tube in this case, it was 

 inconvenient to change its direction. I therefore observed 

 the excursions only on one side of the position of equilibrium 

 (which was at division 54 for the series of experiments to 

 which this Table belongs). The tube being in a determined 

 position, and the galvanometer having attained a constant 

 deviation, I turned the tube without touching the commutator. 

 I then observed, sometimes a strengthening, sometimes a 

 weakening of the current. It is evident that this method has 

 only half the delicacy of the preceding, and that the difference 

 of the deviations due to the ascending and the descending 

 current is \h f 2 . At the foot of each Table are recorded the 

 mean differences D— A or A — D according to the case. 



