automatic regulation may be obtained by connecting the shuntcoils to the third brush and the forward main brush. Table I gives results of a test of a 120-volt 300-ampere machine with the shunt coupled in the ordinary way. Table II. gives results of a test with the shunt coupled in parallel, and one end coupled to the third brush as described; this Table shows a rise of pressure on the shunt between no load and full load of 22 per cent. The arrangement of the third brush can easily be applied to any existing machine having more than one shunt-coil in series, the coils being connected in parallel so as to adapt them for excitation with half the machine voltage. W. G. R. Electric Traction on Railways. P. LANINO. (Elettricità, Milan, vol. xvii., 1898, pp. 164-165 and 183-185.) Improvements in train service take the form of greater speed and more commodious carriages, the latter necessitating greater weight. Both of these require a more powerful locomotive; but, as the permanent way limits the size and weight of the locomotive, improvement ultimately consists in reduction of weight of the motor per HP. The Author considers that for steam-locomotives the limit has now been reached, the best types developing 1,800 HP., and weighing 154 lbs. per HP. For progress, therefore, he looks to the electric motor fed from a central station, this being the lightest motor known; he believes that the weight can be reduced to 22 lbs. per HP., and that the absence of any reciprocating parts removes all limit to the ultimate speed attainable. He describes the scheme of Davis and Williamson for an electric railway between New York and Philadelphia, to run at a speed of 187 miles per hour, and of Behr for a permanent way consisting of five rails supported on inverted V trestles. The latter, when tried at the Brussels Exhibition of last year, failed on account of the instability of the trestles. Criticising the Heilmann locomotive, the Author expresses his opinion that this will never prove superior to the ordinary steam one, as the great disadvantage of extra weight (220 lbs. per HP.) will outweigh the advantages of independent driving of each axle, triple expansion engines and greater regularity in the working of the engines. He believes that the present field of electric traction is that of local lines requiring frequent journeys and light trains, and cites as a successful example a three-mile trolley line opened at Baltimore in 1895. The locomotives weigh 90 tons, and have motors of a total power of 1,600 HP. distributed on six independent driving axles. The works cost per ton-mile, when one, two, or three locomotives are running, are, respectively, 0.186, 0.123 and 0.1015 penny. G. H. BA. Efficiency of Glow-Lamps. J. E. RANDALL. (Mechanical Engineer, vol. i., 1898, pp. 636-638.) The Author briefly reviews the various advances made in lamp manufacture within recent years, and in illustration of the superiority of the modern cellulose filament over the older bamboo one gives the following Table : Duration of Run, in Hours power Average watts per candle Average candle (110-volt cellulose. 110-volt untreated) 16 14.1 12.9 11.8 11.0 10.4 bamboo 50-volt treated) 9.99 16 15 8 15 86 15 68 15 41 15.17 14.96 14.74 9.6 16 15.8 15.3 15 0 14.6 14.2 14.0 13.7 110-volt cellulose. 3.16 3.26 3.13 3.37 3.53 3.51 3.54 3.74 3-20 3·28 3-37 3·45 3.53 3.61 3.67 3-76 A. H. Oxygen in Helium Stars. F. McCLEAN, M. Inst. C.E. (Proceedings of the Royal Society, vol. lxii., 1898, pp. 417-423.) The Author has recently returned from the Cape, where he has obtained the spectra of 116 stars with the aid of his large objective prism attached to the Astrographic Chart telescope at the Observatory there. In his discussion of the spectra he finds in many of the stars what he calls "extra lines, outstanding from the usual helium and hydrogen lines; and to show the probability of these being due to oxygen, he maps the spectrum of oxygen alongside the stella spectra. The evidence afforded by the many apparent coincidences seems to him to be sufficient to warrant their being accepted as oxygen lines. C. P. B. Celestial Phenomena and Kathode Rays. H. DESLANDRES. (Comptes Rendus de l'Académie des Sciences, Paris, vol. cxxvi., 1898, pp. 1323–1326.) The Author, profiting by his observations of the total eclipse of the sun in 1893 to investigate the chromosphere and corona, traces a connection between these phenomena and the appearances exhibited by various combinations of kathode rays. His special experiments with kathode rays have been described in a former Paper (Comptes Rendus, vol. cxxiv., pp. 678, 945, 1,297, and vol. cxxv., p. 373). The two points to which he draws special attention are: (1) the attraction exerted by an anode on a kathode ray; (2) the non-repulsion by a kathode when a solid body is interposed in the path. În the eclipse of 1893, the Author says he found certain regions of the solar surroundings to emit kathode rays. He regards the base of the solar atmosphere as being the seat of this emission, the rays being most intense where the chromosphere is brightest, viz., above the spots and faculæ. By considering the combined result of the solar attraction and the kathode ray repulsion on particles of matter surrounding the sun, he explains the production of the corona and comets. He regards the light of these objects to be produced by the heat and phosphorescence induced by the impact of the kathode rays upon the matter surrounding the sun. The fact of the tails of comets being always turned from the sun is also explained. The particles of matter are assumed very small. Then the attraction being proportional to the mass, and the kathode repulsion proportional to the surface, the latter may eventually overcome the former; hence the outward direction of the cometory appendages. Perrin is cited as authority for stating that the kathode ray carries a negative charge, which will modify the electric and magnetic state of the sun, thereby producing such phenomena as aurora and terrestrial magnetic storms. In conclusion, the Author points to a possible proof of his view. The varying periods of brightness of comets should correspond to the passage of some large sun-spot or facula near the line joining the comet to the sun's centre. C. P. B. Electrodynamic Slit Action. M. LATRILLE. (Annalen der Physik und Chemie, vol. lxv., 1898, pp. 408-430.) A narrow slit transmits polarized light to a different extent according to the angle between the slit and the plane of polarization of the light. In electro-magnetic waves a similar observation is made. The greatest transmission is obtained when the slit is perpendicular to the direction of the vector of electric force. The Author investigates the dependence of the transmitted energy upon the width of the slit. The indicator used is a coheror, and it is found that the direction of the axis of the latter also influences the result, since it reacts more strongly upon electric vibrations along its axis than across it. When the slit is increased in length, the energy transmitted increases rapidly at first, and then more slowly. When the width is increased, the energy transmitted increases slowly at first, and then at a greater rate up to a certain maximum. The absorption is due to the resonance of the edges of the slit. E. E. F. Ketteler-Helmholtz Dispersion-Formulas. A. Pflüger. (Annalen der Physik und Chemie, vol. lxv., 1898, pp. 173-213, and pp. 225-228.) To test them experimentally, let corresponding values of A, n, K, be observed; then, by a tentative process, a set of values (λm, g) are found, so that the graphs for agree. Now (1) may be written in the approximate form where a, b, c are constants, and the values of (Am, g) on the right hand are the same as those calculated from (3). Three separate determinations of n, κ fix the constants a, b, c, and if, for these values, the equation (4) is approximately satisfied for a wide range of A, this is, so far, a confirmation of the original formulas (1) and (2). The object of this research is to test the theory by observations upon solid substances showing anomalous dispersion, and for this purpose cyanin and fuchsin were selected. To determine κ, the index of absorption, thin films of the pigment were deposited on glass plates, and their thickness found by Wernicke's modification of Weiner's method; × was then deduced from observations with König's spectral photometer. The values of n (the index of refraction) for fuchsin in the ultra-violet were found by a method suggested by Kayser. An iron spectrum of wide dispersion was obtained by means of a narrow slit and a Rowland concave grating of 6.5 metres focal length. The rays of a portion of this fell upon a double prism of fuchsin deposited on a quartz plate, and after this upon a photographic plate, where two images of each spectral line were produced. Let d' be the distance between them; then, if the double prism is sufficiently far from the photographic plate, where B is the sum of the very small angles of the prism in seconds, and 8 is the (calculated) value of d' for an angular deviation of one second. The values of n for cyanin in the visible part of the spectrum were also carefully determined directly by means of a spectrometer: the results are tabulated and compared with the Author's previous results. The final conclusion is that, so far as the visible spectrum is concerned, the Ketteler-Helmholtz dispersion formulas applied to media whose optical constants are exceedingly variable within a very small range of wave-length, are (allowing for unavoidable experimental errors) so far consistent with observation that they account for the greater part of the two graphs which are deduced from experiment. discrepancy for the red part of the spectrum. In the supplementary paper it is remarked that the optical There is, however, a |