But when the progress of the electrolysis is carefully followed, by consulting the indications of the amperemeter placed in circuit, it is found to be by no means as regular as the preceding formulae would indicate. With the new apparatus, the decomposition is quite irregular at first, and does not attain regularity until it has been proceeding for upward of two hours. Upon stopping the current and unmounting the apparatus, the platinum rod upon which the fluorine was liberated is found to be largely corroded, and at the bottom of the U-tube a quantity of a black, finely divided substance is observed. This black substance, which was taken at first to be metallic platinum, is a complex compound containing one equivalent of potassium to one equivalent of platinum, together with a considerable proportion of fluorine. Moreover, the hydrofluoric acid is found to contain a small quantity of platinum fluoride in solution. The electrolytic reaction is probably therefore much more complicated than was at first considered to be the case. The mixture of acid and alkaline fluoride furnishes fluorine at the positive terminal rod, but this intensely active gas, in its nascent state, attacks the platinum and produces platinum tetrafluoride, PtF_{4}; this probably unites with the potassium fluoride to form a double salt, possibly 2Kl.PtF_{4}, analogous to the well known platinochloride 2KCl.PtCl_{4}; and it is only when the liquid contains this double salt that the electrolysis proceeds in a regular manner, yielding free fluorine at the positive pole, and hydrogen and the complex black compound at the negative pole. PHYSICAL PROPERTIES OF FLUORINE. Fluorine possesses an odor which M. Moissan compares to a mixture of hypochlorous acid and nitrogen peroxide, but this odor is usually masked by that of the ozone which it always produces in moist air, owing to its decomposition of the water vapor. It produces most serious irritation of the bronchial tubes and mucous membrane of the nasal cavities, the effects of which are persistent for quite a fortnight. When examined in a thickness of one meter, it is seen to possess a greenish yellow color, but paler, and containing more of yellow, than that of chlorine. In such a layer, fluorine does not present any absorption bands. Its spectrum exhibits thirteen bright, lines in the red, between wave lengths 744 and 623. Their positions and relative intensities are as follows: [lambda] = 744 very feeble. | [lambda] = 685.5 feeble 740 " | 683.5 " 734 " | 677 strong 714 feeble. | 640.5 " 704 " | 634 " 691 " | 623 " 687.5 " | At a temperature of -95 deg. at ordinary atmospheric pressure, fluorine remains gaseous, no sign of liquefaction having been observed.