The internal resistance of a voltameter and the cost of platinum electrodes of a few grammes should not arrest the physicist in an experiment; but, in a production on a large scale, it is necessary to decrease the resistance of the liquid column to as great a degree as possible--that is to say, to increase its section and diminish its thickness. The first condition leads to a suppression of the platinum, and the second necessitates the use of new principles in the construction of the voltameter. A laboratory voltameter consists either of a U-shaped tube or of a trough in which the electrodes are covered by bell glasses (Fig. 1, A and B). In either case, the electric current must follow a tortuous and narrow path, in order to pass from one electrode to the other, while, if the electrodes be left entirely free in the bath, the gases, rising in a spreading form, will mix at a certain height. It is necessary to separate them by a partition (Fig. 1, C). If this is isolating and impermeable, there will be no interest in raising the electrodes sensibly above its lower edge. Now, the nearer together the electrodes are, the more it is necessary to lower the partition. The extension of the electrodes and the bringing of them together is the knotty part of the question. This will be shown by a very simple calculation. [Illustration: FIG. 1.--A, B, COMMONEST FORMS OF LABORATORY VOLTAMETERS. C, DIAGRAM SHOWING ASCENT OF BUBBLES IN A VOLTAMETER.] The visible electrolysis of water begins at an E.M.F. of about 1.7 V. Below this there is no disengagement of bubbles. If the E.M.F. be increased at the terminals of the voltameter, the current (and consequently the production of gas) will become proportional to the excess of the value over 1.7 V; but, at the same time, the current will heat the circuit--that is to say, will produce a superfluous work, and there will be waste. At 1.7 V the rendering is at its maximum, but the useful effect is _nil_. In order to make an advantageous use of the instruments, it is necessary to admit a certain loss of energy, so much the less, moreover, in proportion as the voltameters cost less; and as the saving is to be effected in the current, rather than in the apparatus, we may admit the use of three volts as a good proportion--that is to say, a loss of about half the disposable energy. Under such conditions, a voltameter having an internal resistance of 1 ohm produces 0.65 liter of hydrogen per hour, while it will disengage 6.500 liters if its resistance be but 0.0001 of an ohm. It is true that, in this case, the current would be in the neighborhood of 15,000 amperes. Laboratory voltameters frequently have a resistance of a hundred ohms; it would require a million in derivation to produce the same effect. The specific resistance of the solutions that can be employed in the production of gases by electrolysis is, in round numbers, twenty thousand times greater than that of mercury.