Eperiments on Diffraction of Cathode Rays. 609 relation to the complete theory that geometrical optics does to the wave theory. However difficult it may seem to accept such a sweeping generalisation,it seems impossible to explain the results obtained except by the assumption of some kind of diffraction,and the numerical agreement with the wave-length given by the theory is striking.It should be emphasised that there are no adjustable con- stants;the agreement is direct except for a 5 per cent.error.It is an important question whether this error is an experimental one in the measurement of the speed of the rays,as suggested above,or whether it represents some correction on the simple theory.*Further experiments on this point are in progress.It is also hoped to extend Dymond's results by obtaining the diffraction pattern for the molecules of a gas with these fast rays.My sincere thanks are due to Mr.C.G.Fraser and to Mr.J.D.McKay for their valuable assistance in pre- paring the films and making the experiments. Summary. 1.Experiments are described giving the patterns formed by cathode rays scattered by thin films of aluminium,gold,celluloid,and an unknown substance. 2.These patterns are closely similar to those obtained with X-rays in the “powder method..” 3.The sizes of the patterns agree to 5 per cent.with those predicted on the de Broglie theory of wave mechanics,regarding the phenomenon as one of diffrac- tion of the phase waves associated with the electrons. [Nole added in Proof.-I have now traced the error to a loss of potential in the high tension leads,ammeter,etc.If the spark gap is measured directly across the discharge tube instead of,as in the above experiments, near the coil,the value found for "a"in the case of platinum is within 1 per cent.of that found by X-rays,while with the measurements taken in the old way there is a discrepancy of 6 per cent.as in the case of the other metals,the results with which are therefore also explained.] It may be remarked that Davissons and Germers'results appear to show an error of about the same magnitude in the same direction