methods show, however, that if the electron is a rigid ring whose plane is invariable, the scattered energy follows equation (2) rather closely, and is symmetrical on the incident and the emergent sides. If the electron is a flexible ring, or one capable of rotation about any axis, the scattering is more nearly that given by equation (4), but should be somewhat greater for large values of a/λ. The ratio of the incident to the emergent scattered radiation should also be appreciably larger than that given by expression (5). It seems probable, therefore, that the scattering of y-rays and X-rays may be completely explained on the hypothesis that the electron is a ring of electricity of radius about 2 x 10-10 cm., if the ring is capable of rotation about any axis. This hypothesis makes it possible to explain also the effect noticed by A. H. Forman that the absorption coefficient of iron for a beam of X-rays is greater when the iron is magnetized parallel with the transmitted beam than when the iron is unmagnetized or magnetized perpendicular to the X-ray beam. Using an effective potential of 27,000 volts the effect was about 0.4 per cent, and with a potential of 81,000 volts it was 0.6 per cent. From X-ray spectra obtained under similar circumstances it can be shown that the effective wave-length used in the two cases was about 1.0 × 10-8 and 0.5 × 10-8 cm. respectively. If the ring electron acts as a tiny magnet, as suggested by Parson,1 it may be turned by the magnetic field until its plane is perpendicular to the incident beam of X-rays This will make the rays scattered by the different parts of the electron more nearly in the same phase, so that the absorption due to scattered radiation will be increased. Moreover, since the incident rays can get a better hold on the electron in this position, its displacement will be greater than when unorientated, and absorption due to transformation of the X-rays into other types of energy will be greater. For the relatively long waves used by Forman the ratio of the absorption coefficient when 16 FORMAN. Phys. Rev. 7: 119. 1916. 17 PARSON, A. L. Smithsonian Misc. Collections, Nov. 1915. Parson estimates his "magnet on," or ring electron, to have a radius of 1.5 X 10-9 cm. magnetized to that when unmagnetized should be approximately k (2x4), sin2 (2x) where a is the radius of the ring electron and k is the fraction of the electrons which are oriented by the magnetic field. Using the value a = 2.3 × 10-10 cm., this means that the change in the absorption due to magnetization for λ = 1.0 × 10-8 cm. is 0.7 k per cent, and for λ = 0.5 × 10-8 cm. is 2.8 k per cent. From the observed values of this difference we find that the fraction of the electrons oriented by the magnetic field is 0.6 and 0.26. The experimental basis of the latter value is much the more certain. Taking the number of electrons in the iron atom to be 26, this means that in order to explain Forman's effect in terms of ring electrons a number 0.26 × 26 = 7 of the electrons must be capable of being oriented by the magnetic field. This is what would be expected if it is the 8 valence electrons of iron which are responsible for its ferro-magnetic properties. Our hypothesis of a ring electron of radius 2.3 × 10-10 cm. is therefore capable of explaining satisfactorily Forman's effect. It should be noted that Forman explains his effect as being due to an orientation of the molecules in the iron. The experiments of Rognley and the writer's on the effect of magnetizing a crystal on the intensity of the beam of X-rays reflected by it have shown that any orientation of the molecules, if it occurs at all, must be extremely small. It was found further that unless it is very nearly isotropic the atom also is not rotated by magnetization. Thus Forman's explanation of his effect is inadequate. The fact that his experiments can be explained in terms of an orientation of the electrons must be taken as a confirmation of the conclusion arrived at by Rognley and the writer that it is not the atom as a whole, but the electron itself that is the ultimate magnetic particle. 18 COMPTON and ROGNLEY. Science (N. S.) 46: 415. 1917. Summary. Ishino's experiments, showing that the scattering of hard y-rays by different materials is strictly proportional to the number of electrons and is not proportional to the masses, proves that the electrons are responsible for practically all of the scattering, and that for these wave-lengths they act independently of each other. According to the classical electrodynamical theory, this means that if the electrons are sensibly point charges of electricity, the absorption coefficient due to scattering for these rays must be given by equation (1). Since this equation does not hold for these wave-lengths, we cannot consider the electron to be a point charge. In order to account for the small absorption coefficient of y-rays the electron must have an effective radius of about 2.3 × 10-10 ст. In order to explain the fact that the emergent scattered radiation is more intense than the incident radiation, it is necessary to assume further that the different parts of the charge of the electron can possess certain motions independently of each other. It appears that these phenomena, together with the electromagnetic mass of the electron, can be quantitatively explained on the hypothesis that the electron consists of a ring of electricity subject to rotation about any axis and of radius about 2.3 × 10-10 cm. This hypothesis is confirmed by the fact that it explains satisfactorily Forman's effect of magnetization of iron upon its absorption coefficient, for which there is no other apparent explanation. CHEMISTRY. A silica-glass mercury still. J. C. HOSTETTER and R. B. SOSMAN, Geophysical Laboratory. Although numerous electrically heated mercury stills have been described and are doubtless being used with satisfaction, nevertheless there is one undesirable feature that is common to all vacuum mercury stills and that is avoided in the one about to be described: namely, that a still made of ordinary glass or even of combustion glass will, when slightly overheated, collapse under the pressure of the atmosphere.' 1 The form of failure of such a tube is of some interest in itself. One of our ordinary glass stills that collapsed one night when the voltage on the power line became too high yielded symmetrically around its vertical axis, instead of flattening out, producing a figure with three cusps separated by angles of 120 degrees. After several such exasperating experiences with glass stillsthe overheating being caused by rising voltage on the power line we had our still remade of silica glass (fused quartz, or "quartz glass"), and it has been in use more or less continuously for several years. The additional cost of the silica-glass still is well expended in the insurance thereby secured against interruption of the distillation. We have been requested to put on record a brief description of the still, having had a number of inquiries for information concerning its design. Design of the still. The construction is shown in detail in figure 1. An effort has been made to simplify the still as much as possible in order to minimize the difficulties connected with the working of silica glass. No originality is claimed for the design: an inspection of the diagram will show that the essential points of any continuously acting still have been embodied, some having been taken from one still and some from another. The distillation chamber (D) has an inside diameter of 30 mm. and a length of 110 mm. The condenser (C) has an inside diameter of 13 mm. Both of these parts of the apparatus are made of the semitransparent inexpensive variety of silica glass. The smaller vertical tubes, A and B, however, are made of the transparent variety, thus enabling one to observe readily the rate of condensation and to see whether the vacuum is being maintained. At V the still is connected to a suction pump that gives a pressure of 1 cm. of mercury or less. This joint is made with hard De Khotinsky cement and is located where it will be air-cooled as thoroughly as possible. The lengths of the small vertical tubes must be such that at ordinary atmospheric pressure the distillation chamber is about half filled: the length of A is 720 mm. to the bottom of the distillation chamber, while B has a length of 840 mm. The inside diameters of A and B should not be too small; the tubes on our still have a bore of 3 mm., but 5 mm. might be better. The furnace. The furnace is wound so as to go directly on the 110-volt circuit (alternating or direct current) without any external resistance. The power consumption is about 60 watts. Since the temperature required is comparatively low, various |