The phenomena to which it is desired to call attention were first observed in the course of an investigation into the electrolytic methods of determining nickel, carried out in connection with the work of the British Association Committee on the Electrolytic Methods of Quantitative Analysis. The results of the investigation are given in the Report of last year's meeting (Bristol, 1898, p. 300), but the present matter is not referred to there, as it was first noticed only during the preparation of the report.

The apparatus employed for most of the determinations consisted of the usual Classen form of cathode basins, with disc anodes. The current was drawn from storage batteries at 12 volts, variable resistances being used in the circuits. A Davies' ammeter could be introduced easily into any circuit at will, and a Davies' voltmeter was also used to indicate the difference of potential at the electrodes. These measuring instruments were so arranged and connected that the operation which introduced the ammeter into a circuit simultaneously brought the voltmeter into action; hence it happened that voltmeter readings were always taken conjointly with ammeter ones, although attention was paid almost entirely to the effects of varying the current density, and to any changes which might occur during the course of the experiment, with no special consideration of the potential involved.

For the determination of quantities of nickel varying between 0·l and 0·5 gram, the conditions adopted as most generally suitable when employing the usual method were those given below:—

Volume of solution, about 135 c.c. (giving about 100 sq. cm. of cathode area), containing 5 grams each of ammonium sulphate and of ammonia; current density at cathode (amp. per sq. decim.), 0·6–0·7; ordinary temperature; time necessary, about 4 hours.