In previous notes I have established, first, that the galvanic depositions experience a change of volume, from which there results a pressure exercised on the mould which receives them; second, that the Peltier phenomenon is produced at the surface of contact of an electrode and of an electrolyte. Fresh observations have caused me to believe that the two phenomena are connected, and that the first is a consequence of the second. The Peltier effect can clearly be proved when the electrolysis is not interfered with by energetic secondary actions, and particularly with the sulphate and nitrate of copper, the sulphate and chloride of zinc, and the sulphate and chloride of cadmium. For any one of these salts it is possible to determine a value, I, of the intensity of the current which produces the metallic deposit such that, for all the higher intensities the electrode becomes heated, and such that it becomes cold for less intensities. I will designate this intensity, I, under the name of _neutral point of temperatures_. The new fact which I have observed is, that in the electrolysis of the same salts it is always possible to lower the intensity of the current below a limit, I', such that the compression produced by the deposit changes its direction, that is to say, instead of contracting the metal dilates in solidifying. This change, although unquestionable, is sufficiently difficult to produce with sulphate of copper. It is necessary to employ as a negative electrode a thermometer sensitive to 1/200 of a degree, and to take most careful precautions to avoid accidental deformations of the deposit; but the phenomenon can be observed very easily with nitrate of copper, the sulphate of zinc, and the chloride of cadmium. There is, therefore, a _neutral point of compression_ in the same cases where there is a neutral point of temperatures. With the salts of iron, nickel, etc., for which the neutral point of temperatures cannot be arrived at, there is also no neutral point of compression; and the negative electrode always becomes heated, and the deposit obtained is always a compressing deposit. I have determined, by the help of observations made with ten different current strengths, the constants of the formulæ which I have explained elsewhere, and which gives the apparent excess, y, of the thermometer electrode compressed by the metallic deposit in terms of the time, t, during which the metal was depositing: A t (1) y = ------- B + t The constant, A, is proportional to the variation of volume of the unit of volume of the metal. The values of A, without being exactly regular, are sufficiently well represented within practical limits by the formula: (2) A = - a'i + b'i², of the same form as the expression E: E = - ai + bi²,