The composition of the ions of a salt can be determined experimentally[123] by devices of which the U-tube experiment (p. 45) may be considered to be a simple type. For instance, if we wish to determine the composition of the ions of sodium nitrate, we could cover a solution of sodium nitrate with a solution, say, of hydrochloric acid, pass a current through the liquids, and determine the composition of the components that have moved to the negative and positive poles, respectively. In practice, the device could be elaborated for the sake of convenience. Stopcocks, for instance, might be placed in the U-tube, at the points of separation of the nitrate solution and the hydrochloric acid (see Fig. 13), the stopcocks being opened only during the passage of the current. Or porous plates or cells might be used, in place of stopcocks, at these [p071] points. Now, if we assume sodium nitrate to be dissociated, not into Na^{+} and NO_{3}^{−}, but let us say into positive ions NaO^{+} and negative ions NO_{2}^{−}, the changes which would result from the passage of a current would be as follows: starting with the action at the positive pole, we should find chloride ions discharged and chlorine evolved at the pole (the evolution of chlorine could be avoided, if considered desirable, by the use of a silver anode, which would absorb the liberated chloride-ion to form insoluble silver chloride on the electrode). At the same time, hydrogen ions would move out of the space ‹P›, being repelled by the positive pole, and attracted by the negative. At the boundary between the sodium nitrate solution and the hydrochloric acid, the negative ions of sodium nitrate, which we are supposing to have the composition NO_{2}^{−}, would move up from ‹B› toward the positive pole (‹cf.› exp., p. 45), being attracted by its charge; at any moment we should have in any part of ‹P› as many negative ions (Cl^{−} and NO_{2}^{−}), as there are hydrogen ions, the solution showing no excess of free electricity at any point. Now, if NO_{2}^{−} were the ion that moved up into the space ‹P›, then we should presently find ‹nitrous› acid ‹around› the ‹positive pole› in space ‹P›, H^{+} and NO_{2}^{−} combining to form nitrous acid, HNO_{2}. But, as a matter of experiment, although the tests for nitrous acid belong to the most sensitive ones in chemistry, no trace of this acid is found there; what we do find is ‹nitric acid›, HNO_{3}, resulting obviously from the presence in space ‹P› of both hydrogen ions and ‹nitrate ions›, NO_{3}^{−}, which have moved up from space ‹B›. It is clear, that the presence of nitric acid in the region around the positive pole means that the nitrogen atoms must have carried with them all three of the oxygen atoms of the nitrate—in a word, that the composition of the negative ion of sodium nitrate is NO_{3}^{−} and not, say, NO_{2}^{−}. Similarly, considering what happens in space ‹N›, round the negative pole, we have here an evolution of hydrogen, a migration