This JOURNAL, the official organ of the Washington Academy of Sciences, aims to present a brief record of current scientific work in Washington. To this end it publishes: (1) short original papers, written or communicated by members of the Academy; (2) a complete list of references to current scientific articles published in or emanating from Washington; (3) short abstracts of certain of these articles; (4) proceedings and programs of meetings of the Academy and affiliated Societies; (5) notes of events connected with the scientific life of Washington. The JOURNAL is issued semi-monthly, on the fourth and nineteenth of each month, except during the summer when it appears on the nineteenth only. Volumes correspond to calendar years. Prompt publication is an essential feature; a manuscript reaching the editors on the second or the seventeenth of the month will ordinarily appear, on request from the author, in the next issue of the JOURNAL. Manuscripts may be sent to any member of the Board of Editors; they should be clearly typewritten and in suitable form for printing without essential changes. The editors cannot undertake to do more than correct obvious minor errors. References should appear only as footnotes and should include year of publication. Illustrations will be used only when necessary and will be confined tc text figures or diagrams of simple character. The editors, at their discretion, may call upon an author to defray the cost of his illustrations, although no charge will be made for printing from a suitable cut supplied with the manuscript. Proof-In order to facilitate prompt publication no proof will be sent to authors unless requested. It is urged that manuscript be submitted in final form; the editors will exercise due care in seeing that copy is followed. Authors' Copies and Reprints.-On request the author of an original article will receive gratis ten copies of the number containing his contribution and as many additional copies as he may desire at five cents each. Reprints will be furnished at the following schedule of prices: Covers bearing the name of the author and title of the article, with inclusive pagination and date of issue, will be $2.00 for the first 100. Additional covers $1.00 per 100. As an author may not see proof, his request for extra copies or reprints should Invariably be attached to the first page of his manuscript. The rate of Subscription per volume is.......... Semi-monthly numbers.. Monthly numbers..... $6.00* .25 .50 Remittances should be made payable to "Washington Academy of Sciences," and addressed to the Treasurer, William Bowie, Coast and Geodetic Survey, Washington, D. C., to Williams & Wilkins Company, 2419-2421 Greenmount Ave., Baltimore, Md., or to the European Agents. European Agents: William Wesley & Son, 28 Essex St., Strand, London, and Mayer and Müller, Prinz Louis-Ferdinand Str., Berlin. Exchanges.-The JOURNAL does not exchange with other publications. Missing Numbers will be replaced without charge, provided that claim is made within thirty days after date of the following issue. Volume I, however, from July 19, 1911, to December 19, 1911, will be sent for $3.00. Special rates are given to members of scientific societies affiliated with the Academy. THE WAVERLY PRESS PHYSICS.-The size and shape of the electron. ARTHUR H. COMPTON, Research Laboratory, Westinghouse Lamp Company. (Communicated by G. K. Burgess.) The radius of the electron is usually deduced from the energy of the electron in motion, assuming its magnetic energy to be identical with its kinetic energy. If the electron is a sphere, its radius must be, according to this assumption, about 1 × 10-13 cm. It is thus sufficiently small to act as a point charge of electricity even with the shortest y-rays. Calculating on the basis of such an electron, J. J. Thomson1 has shown that the fraction of the energy of an electromagnetic wave incident upon an electron which is scattered by it is given by the expression This corresponds to a mass absorption coefficient due to a scattering of the primary beam equal to where N is the number of electrons which contribute to the scattering in a gram of the absorbing medium, C is the velocity of light, and e and m'have their usual significance. As Barkla has pointed out, there may be absorption due to other causes, 1 THOMSON, J. J. Conduction of Electricity through Gases, 2d ed., p. 321. such as the production of secondary photoelectrons or beta rays, and for other than waves of short length the rays scattered by the different electrons in an atom are nearly enough in the same phase to produce the phenomenon of "excess scattering," so that the absorption coefficient is in most cases considerably greater than the value given by this expression. If the electron acts as a point charge there is, however, no possible grouping of the electrons which can, according to classical theory, produce a smaller absorption than that calculated according to Thomson's, formula. Barkla and Dunlop2 have shown that for a considerable range of wave-lengths of X-rays the mass scattering coefficients of the lighter elements are given accurately by equation (1) if the number of electrons in the atom is taken to be approximately half the atomic weight. For elements of high atomic weight the phenomenon of excess scattering occurs, except with the very shortest wave-lengths, and the absorption coefficient due to scattering becomes much greater than this value. For wavelengths less than 2 X 10-9 cm., however, the absorption coefficient becomes very appreciably less than that theoretically calculated, falling as low as one-fifth as great for the shortest y-rays. Soddy and Russell' and Ishino' have shown that for these shortest rays the amount of energy scattered by the different elements is accurately proportional to their atomic numbers, so that all the electrons outside the nucleus are effective in producing absorption. It is therefore impossible to account for this very considerable decrease in the absorption coefficient for very short electromagnetic waves if the electron is considered to be a point charge of electricity. If, however, the diameter of the electron is comparable in magnitude with the wave-length of the incident wave, the radiation scattered by different parts of the electron will be so different in phase that the energy of the scattered rays will be materially reduced. If, for example, the charge on an electron 2 BARKLA and DUNLOP. Phil. Mag., March, 1916. 3 SODDY and RUSSELL. Phil. Mag. 18: 620. 1910; 19: 725. 1910. 4 ISHINO. Phil. Mag. 33: 129. 1917. is supposed to be in the form of rigid spherical shell, incapable of rotation, a simple calculation shows that the mass absorption coefficient due to scattering is given by where a is the radius of the spherical shell and x is the wavelength of the incident beam. For long waves this becomes identical with equation (1), but it decreases rapidly as the wavelength approaches the diameter of the electron, as is shown in curve I, figure 1. Such an assumption is therefore able to explain at least qualitatively the decrease in the absorption for electromagnetic waves of very high frequency. It would appear more reasonable to imagine the spherical shell electron to be subject to rotational as well as translational displacements when traversed by a y-ray. The scattering due to such an electron is difficult to calculate, but an approximate expression can be obtained if the electron is considered to be perfectly flexible, so that each part of it can be moved independently of the other parts. On this hypothesis it can be shown that the intensity of the beam scattered by an electron at an angle with an unpolarized beam of y-rays is given by the expression λ 2 I1 = 1 e'(1+cos') { (^)' sin2 (4ra sin )/sin' } 2r2m2C4 Απα 2 (3) Here I is the intensity of the incident beam, r is the distance at which the intensity of the scattered beam is measured, and the other quantities have the same meaning as before. The mass absorption coefficient due to scattering by such an electron is therefore This integral may be evaluated graphically or by expansion into a series. The values of a/p in the case of aluminium, taking the numbers of electrons per atom to be 13, are plotted in curve II, figure 1, for different values of a/λ. The values for a |