direction using the 180-Mev bremsstrahlung. Backgrounds due to cosmic rays varied from about 25% at 6=0° to about 2% at 6=90°. Other backgrounds, mostly due to accidental coincidences, were determined by measurements below counting threshold at 150 Mev. At 0=90° these backgrounds were about 25%. At all other angles they were 10% or less. Several runs were taken at each angle. An attempt was made to alternate the runs on each side of 0=45° to reduce systematic errors. These data have been corrected for the expected geometrical asymmetry discussed above.The angular distribution of Fig. 4 indicates no marked asymmetry. A least squares fit of the form f(6) = 1+a cos 4 0 gives a value of a = 0.025±0.090.As a consequence of our inability to observe directly the origin of a cosmic-ray particle, we begin the development with a discussion of the limitations within which we can construct a cosmic-ray accelerator mechanism. We find that we are allowed only the betatron effect and the Fermi mechanism. We review some of the many variations of these mechanisms which are to be found in the literature. Then it is shown that trains of oppositely moving hydromagnetic waves of large amplitude and with sharp crests can accomplish large and continued particle accelerations which are adequate to maintain the observed galactic cosmic-ray held. The large acceleration arises as a consequence of the simple fact that each wave tends to sweep up the cosmic-ray particles before it, so that head-on collisions of particles with waves are much more common than overtaking collisions. It is pointed out that the sharp crests of the waves are a natural consequence of the observed supersonic mass motions. Therefore, the acceleration by oppositely moving waves does not depend upon any special wave form, and we suggest that it is the naturally occuring acceleration process. * Assisted in part by the Office