But the main interest in this paper lies in the indications which it gives that Cavendish was aware of the laws which regulate the flow of electricity through multiple conductors, and in the comparisons of electrical resistance which are introduced. It had been formerly believed that electricity would always select the shortest or best path, and that the whole of the discharge would take place along that route. Franklin seems to have assumed this in the passage quoted[4] respecting the discharge of the lightning down the uninsulated conductor instead of through the building. The truth, however, is that, when a number of paths are open to an electric current, it will divide itself between them in the inverse ratios of their resistances, or directly as their conductivities, so that, however great the resistance of one of the conductors, some portion, though it may be a very small fraction, of the discharge will take place through it. But this law does not hold in the case of insulators like the air, through which electricity passes only by disruptive discharges, and which completely prevent its passage unless the electro-motive force is sufficient to break through their substance. In the case of the lightning-conductor, however, its resistance is generally so small in comparison with that of the building it is used to protect, that Franklin's conclusion is practically correct. [Footnote 4: Page 96.] In his paper on the torpedo Cavendish stated that some experiments had shown that iron wire conducted 400,000,000 times better than rain or distilled water, sea-water 100 times, and saturated solution of sea-salt about 720 times, better than rain-water. Maxwell pointed out that this comparison of iron wire with sea-water would agree almost precisely with the measurements of Matthiesen and Kohlrausch at 11 deg.C. The records of the experiments which led to these results were found among Cavendish's unpublished papers, and the experiments also showed that the conductivity of saline solutions was very nearly proportional to the percentage of salt contained, when this was not very large--a result also obtained long afterwards by Kohlrausch. In making these measurements Cavendish was his own galvanometer. The solutions were contained in glass tubes more than three feet long, and a wire inserted to different distances into the solution; thus the discharge could be made to pass through any length of the liquid column less than that of the tube itself. From the Leyden battery of forty-nine jars, six jars of nearly equal capacity were selected and charged together, and the charge of one jar only was employed for each shock. The discharge passed through the column of liquid contained in the tube, from a wire inserted at the further end, until it reached the sliding wire, when nearly the whole current betook itself to the wire on account of its smaller resistance, and thence passed through the galvanometer, which was Cavendish himself.