278 MORRIS LOEB 



liquid into two nearly equal portions. It frequently happened 

 that, during the electrolysis, the electrolyzed silver would 

 grow out from the bulb along the wall of the horizontal 

 tube. In such cases the circuit was broken before the 

 silver had reached the main tube; otherwise, [110] we dis- 

 connected whenever we estimated that a sufficient change 

 of concentration had taken place to give the most reliable 

 results. 



The analyses were performed, as already intimated, by 

 titration with a solution of ammonic sulphocyanate, the 

 standard of which (about j^ normal) was ascertained by fre- 

 quent comparison with an argentic nitrate solution of known 

 strength, occasional gravimetric determinations of which 

 gave absolutely constant results. We claim for the titrations 

 an error well within ^ cc. (= 0.038 milligrams Ag), each 

 end-reaction being observed twice, as 1 cc. of a ^ normal 

 solution of argentic nitrate was added from a pipette, after 

 the first appearance of the characteristic tint, and the titra- 

 tion repeated. 



The ratio we are seeking to determine requires, besides the 

 data of the change in concentration, the weight of the silver 

 which has been actually deposited upon the cathode, in the 

 same interval of time. This would most naturally be deter- 

 mined by a silver voltameter placed in the same circuit with 

 the apparatus. But as the examination of dilute solutions 

 often necessitated the deposit of less than 20 milligrams in 

 all, and the silver is deposited in such a form upon the vol- 

 tameter that it is difficult to weigh it without a slight loss, we 

 found it to be more advantageous to determine the quantity 

 of electricity which traversed the apparatus by galvanometric 

 measurement. A resistance-box was introduced into the cir- 

 cuit for this purpose (see figure), to the extremities of which 

 a second circuit was connected, containing a Clark's cell and 



