1. 2. 5. cms. 30... 37 47 49 for the brass oscillator, and numerous other readings accentuated this result. Afterwards, however, the platinum shell (No. 6) was constructed, and proved the most reliable and efficient of all. Now this shell was thicker than the calculated "skin" (Table I), and further, on wrapping two layers of heavy Distance of from TABLE IV. No. of Oscillator; scale-reading in mm. Detector. No. No. No. No. No. No. No. No. No. No. No. No. No. No. 6. 7. 8. 9. 10. 11. 12. 13. 14 42 15. 16. 46 47 46 37 39 39 52 40 48 49 46 57 44 155 47 33 30 33 46 26 32 39 35, 30 19 33 30 32 41 40 40 27 ེཚ 21 25 33 22 30 18 26 27 25 26 31 30 35 20 38 23 35 22 tin-foil (thickness 0-005cms) about the mantle and binding it closely there, no alteration whatever was made in its efficiency. The proper conclusion, therefore, seemed to be that, within the limits of the experiments, the thickness of the metal had no effect. If one could obtain a thin continuous carbon shell, possibly more decisive experiments might be made, as the "skin" for carbon is comparatively thick, but such was not available. The magnetic nature of the iron made no difference whatever; iron and brass were indistinguishable in their behavior. Silver did not appear equal to platinum as a spark-surface. As the experiments with cylindrical shells failed to exhibit any trace of the "skin"-effect, it was decided to try with spheres, and Righi oscillators with spheres 2.5, 4, and 10cms, respectively, in diameter were constructed. By these solid brass spheres were compared with gold-leaf spherical shells of the same diameter. The latter were made by covering with gold-leaf accurately turned and beautifully polished lignumvitae spheres of the required sizes. In the oscillators composed of solid spheres the brass surface was used as sparking faces. Observation showed that, in air, newly polished brass is at least equal and possibly superior to sheet platinum, and the oil dielectric prevents the rapid deterioration of the surface. For the gold-leaf shells platinum faces were necessary. Two AM. JOUR. SCI.-FOURTH SERIES, VOL. XIII, No. 73-JANUARY, 1902. were fastened on each sphere, at the extremities of a diameter, by electro-deposition of copper. The results obtained with the three oscillators are given in Tables V, VI and VII, which explain themselves. With the solid spheres the first reading was usually the highest, but with the shells there were usually two or three indifferent ones to start with. This was observed with platinum faces generally. The readings recorded are: for the solid spheres, the first 8 taken; for the 4cm shells, the first 8; for the 2.5cm shells, the last 8 of a series of 14; and for the 10cm shells, all the satisfactory readings obtained. The 4cm spheres seem to show evidence, at the distance of 30cms, of the reversal effect mentioned above, but it disappeared before a distance of 50cms was reached. With the other spheres no trace of the phenomenon can be found. No difficulty at all was experienced in handling the smaller spheres, but the larger ones gave some trouble. But a single reading had been secured with the first pair of 10cm shells before one of them failed. Veins appeared in the gold-leaf encircling one platinum cap. On replacing this sphere by a new one only three or four readings were secured before the other sphere of the first pair failed in the same way. This was removed and a new one put in its place, and with the pair thus formed the readings in Table VII were obtained. If curves be drawn with the mean readings of Tables V, VI and VII as ordinates and the distances from the oscillator as abscissae, the curve for a shell will be found to practically coincide with that for the solid sphere of the same diameter. The experiments, therefore, indicate that, both in the case of the cylindrical and of the spherical doublets, the excessively thin gold-leaf shells were quite as efficient as the solid metal bodies. This investigation was made in the Jefferson Physical Laboratory, Harvard University; and while under especial obligation to Professor Trowbridge for his never-failing consideration and encouragement, I would express my gratitude to every member of his staff for innumerable kindnesses. Jefferson Physical Laboratory, Harvard University. ART. II. The Influence of Hydrochloric Acid on the Precipitation of Cuprous Sulphocyanide; by R. G. VAN ΝΑΜΕ. [Contributions from the Kent Chemical Laboratory of Yale University.—CV.] In a previous contribution to this Journal* the writer gave the results of a series of experiments upon the estimation of copper by precipitating and weighing as cuprous sulphocyanide. It was concluded from this work that the process did not lose greatly in accuracy even when quite large amounts of sulphuric or hydrochloric acid were present. In discussing the results obtained, however, no distinction was made, as should have been done, between free acid originally present in the solution and that remaining in the free state after the addition of the precipitants, ammonium bisulphite and ammonium sulphocyanide. With hydrochloric acid the effect of the bisulphite, as will be shown, is to convert a definite quantity of acid into ammonium chloride, so that by increasing the bisulphite the free acid present when the copper is precipitated may be diminished to any desired extent. The results obtained by precipitating copper in solutions containing various amounts of hydrochloric acid, weighing as cuprous sulphocyanide, and determining the copper in the filtrate, are given in Table I. In neutral or faintly acid solutions cuprous sulphocyanide is precipitated in a very finely divided state and often shows a tendency to pass through the filter with the wash water as soon as the latter becomes nearly free from dissolved salts. This may be prevented by washing with a salt solution. For this reason the precipitates were washed with decinormal ammonium sulphocyanide, a medium in which cuprous sulphocyanide is even less soluble than in pure water. An electrolytically standardized copper sulphate solution of about normal strength was measured out in portions of 50cm3 for the first seven determinations. For the last four, like quantities of approximately normal copper sulphate were taken. The required amount of strong hydrochloric acid of specific gravity 1·17-1∙18 was then added, followed, after diluting, by ammonium bisulphite and ammonium sulphocyanide in the order named, the last. in measured portions of a standardized decinormal solution. The ammonium bisulphite solution was prepared by saturating aqueous ammonia with sulphur dioxide. After precipitation the mixture was allowed to stand for * Vol. x, 451. |