The usual form does not show nearly as constant results as the other three. The last large difference may, however, be due to loss of silver in one bowl.

Tables 2 to 8 allow a comparison between the different types, when the same quantity of electricity is sent through them. In the first column of each table the current is given and in the last the percentage difference between the two forms. The asterisks denote that the deposit is on silver, i. e., that it is a second deposit.

From these tables it can be seen that there are two distinct classes of voltameters, on one side the usual, Leduc's, and the silver oxide type, on the other the Richards and the large anode type, the latter class yielding deposits about 0.05 per cent smaller than the former.

The difference between the usual form and Richards's agrees very well with the final average of 0.050 per cent given by Richards." A voltameter in which the electrolyte was always kept saturated with Ag,O has not been used before. Patterson and Guthe employed a solution saturated before the experiment was made, but in all other respects adhered to the legal specifications. Richards's comparison between his type and Patterson and Guthe's showed for the latter a

a Richards and Heimrod: Proc. Am. Acad., 37, p. 439; 1902.

deposit 0.112 per cent larger than for the former, besides a considerable variation in the percentage difference (from 0.039 to 0.230 per cent). We see from Table 3 that the form of Ag,O voltameter used in the present investigation gives somewhat more concordant results and a much smaller difference between the two types. Possibly the presence of filter paper in the Patterson and Guthe voltameter may explain the greater value. A porous cup has doubtless the effect of decreasing the weight of the deposit.

a

It must be confessed that the results obtained with Leduc's voltameter were not expected. He has repeatedly asserted that no trouble arises when small anodic current density is employed. As will be seen. from Table 4, in these comparisons small currents were used, but the deposit was in each case 0.057 per cent larger than that obtained with the Richards type. A comparison of Tables 3 and 4 verifies Leduc's statement that a solution saturated with the oxide and a normal solution will give identical results. It was also noticed that even with these small currents the heavy anode liquid sinking to the bottom of the platinum bowl and there spreading produced a starlike figure of the deposit. This peculiar arrangement of the crystals, very characteristic for the usual type, has been studied by Behn," who showed it to be due to the motion of the heavy anode liquid along the surface of the kathode.

With a porous cup around the anode the crystals are distributed perfectly evenly over the kathode without showing the striæ spoken of. Filter paper allows an easy passage to the anode solution, and should therefore be avoided, even if the chemical action upon neutral nitrate solution, to which attention was drawn above, will not affect the electro-chemical equivalent.

All previous researches on the subject have led to the conclusion that the source of trouble is to be looked for at the anode. The solution becomes acid during electrolysis, and therefore the presence of acid has been suggested as the reason for the inconstancy of the voltameter. Leduc has shown, however, that a fresh solution, originally neutral, but afterwards slightly acidified, gives smaller deposits than a neutral solution. According to this observation, the presence of acid alone has not the effect of increasing the weight of the deposit. Another interesting observation by the same author shows that if the amount of acid surpasses a certain limit in the original solution, its density will decrease during electrolysis, a fact observed before by

a Leduc: Rapp. Congr. intern., 2, p. 444; Paris, 1900.
Behn: Wied. Ann., 51, p. 105; 1894.
Leduc: J. de Phys., 1, p. 573; 1902.

Rodger and Watson." The latter seem to be the first to have pointed out that during electrolysis a complex silver ion may be formed and that this may produce too heavy deposits. The thorough work of Richards and his students have strengthened this view and my own results seem all to point to the same explanation. The existence of such a complex ion, whatever it may be,' would easily enough explain the formation of the acid at the anode. It is well known that, during electrolysis of silver nitrate solution, crystals are formed at a platinum anode, which according to Sule and Mulder and Heringad have the formula Ag,NO11. I have been able to obtain these bluish-black crystals mixed with silver at the cathode from a solution saturated with Ag,O, using a large current density. In this case the deposit was abnormally heavy, 0.23 and 0.32 per cent greater than that obtained in a large silver anode voltameter. Complex ions are therefore present in such a solution and it seems reasonable to suppose that if they reach the kathode, as they will in voltameters in which the anode is surrounded by filter paper or in the Ag,O type, the deposit will be too heavy, resulting in a wrong value for the electro-chemical equivalent of silver.

11

Richards's method of removing the heavy anode liquid from the bottom of the porous cup and preventing its diffusion toward the kathode by keeping the level inside lower than the outside is therefore correct, and will not give too small an equivalent, as Leduc asserts.

e

The large silver anode voltameter gives the same results as Richards's form, in spite of the fact that the solution is at the same level inside and outside. (See Table 7.) How can we explain this? A few experiments by Richards and Heimrod show that a crucible on which silver had been deposited previously will increase in weight when filled with the anode solution. Apparently a chemical reaction takes place, resulting in a breaking up of the complex ion and a deposition of silver. In the large anode type the porous cup prevents the anode solution from diffusing rapidly, and the silver at the bottom of the cup breaks up the troublesome ion. This secondary chemical reaction, I believe, is the reason why the anode is always covered by a layer of finely divided silver-the "anode dust"-the anode itself acting, to a certain extent, chemically on the ion produced by electrolysis. The

a Rodger and Watson: Phil. Trans., 186, p. 631; 1895.

Richards suggests Ag,O. See also the theoretical discussion between Revay and Küster and von Steinwehr: ZS. f. Electroch., 4, pp. 313, 451; 1898.

Sulc: ZS. f. anorg. Chem., 12, p. 90; 1896.

d Mulder and Heringa: Verh. Kon. Ak. Wet., 3, p. 37; 1896. Richards and Heimrod: loc. cit., p. 431.