112 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



numbers calculated from electromotive force with those determined by 

 the Hittorf method. In his first gravitational experiment he again 

 made use of cadmium iodide solution which owing to complex ion for- 

 mation apparently has a transference number greater than unity, and 

 hence gives a large electromotive force. In the experiment the solu- 

 tion was placed in a glass tube with the electrodes at the two ends 

 about 91 cm. apart. The gravitational electromotive force was deduced 

 from a series of readings of galvanometer deflection, the positions of 

 the upper and lower electrodes being interchanged between each two 

 measurements. The transference number of the solution was found 

 by calculation to be 1.236, differing by at least five per cent from that 

 determined in the analytical way. 



In order to test the theory on more simple solutions such as the 

 chlorides of the alkali metals, which do not have abnormal values of 

 the transference number, it was necessary to compensate for the de- 

 creased size of the effect by the construction of an apparatus with 

 greater difference in height between the electrodes. The electrodes 

 were placed in glass vessels and connected together by a rubber tube 

 full of the solution. The electrode containers could be raised and low- 

 ered at will, permitting a difference in level of 377 cm. 



The following table gives a summary of his results, including a com- 

 parison between the value Ta for the transference number of the anion 

 as calculated from his results and as determined by the Hittorf 

 method. The measurements were made with calomel electrodes (in 

 the case of CdCl2 with Cd amalgam electrodes as well). The e.m.f. is 

 expressed in microvolts per centimeter difference in height. 



Des Coudres estimates that the uncertainty in the values of the elec- 

 tromotive forces given is not greater than ±().()09 microvolts. 



