Intermittently, for some months, we have been working on the analysis of the spectrum of OH produced in water vapor containing 35 percent of the hydrogen as H 2 , from which we have obtained good photographs. 1 We have been successful in identifying and assigning nearly 500 new lines, in the OH 2 spectrum, which are distributed through eight bands, including the (3,1) and (3,2) bands. Assignments extend beyond K = 30 in the (0,0) band.In order to calculate isotope shifts between corresponding lines of 0 16 H X and 0 16 H 2 we find that it is necessary to introduce corrections for isotope effects in spin coupling 1 and in A-doubling in the ground 2 n level and to include y e terms in the expansion of the rotational constant B v in both the upper and the lower levels. The latter is easily accomplished from the rotational constants now in the literature 2 which give: 5/=l7.375 « e ' = 0.838 y/= -0.0075 £ e "= 19.009 a/'= 0.686 Y«" = -0.0125The well-known measurements of the doublet separations in the Balmer series by Kent, Taylor and Pearson, 1 and by W. V. Houston 2 indicate that H/3 is about three percent narrower than predicted by the relativistic fine structure theory, whereas the separations for Ha and H7 agree with the theory within the limits of experimental error. The more recent observations of Spedding, Shane and Grace 3 indicate, if we interpret their preliminary report correctly, that the YL l a and H 2 a doublets are both somewhat less than one percent narrower than expected from the theory. Finally, the new interferometer measurements of Houston and Hsieh 4 increase the discrepancy to about three percent for all the doublets of light hydrogen (H 1 ). These last experiments were conducted under such conditions that the intensity ratios of the components were those expected theoretically, and the discrepancy is accordingly attributed to a deficiency in the theory. A simple computation shows that this discrepancy can be accounted for by a deviation from the Coulomb law at small distances from the nucleus. Such a deviation is known to occur for heavier nuclei and there exists evidence pointing to the same conclusion for hydrogen. 5 It might be due in the present case to the finite size of the electron and proton suggested by Born, 6 or to an eventual composite structure of the proton (neutron plus positron).We suppose that the field is Coulomb down to a radius a where a change occurs into some other type of law. A potential attractive as l/r n , where n is greater than or equal to 2, must be rejected on the basis of mathematical considerations. 7 Since the exact shape of the potential curve is not important for such very small distances, we have represented the case where the potential approaches a finite limit by the constant potential e 2 /a and the case where the potential becomes positively infinite by a law repulsive asThe results of the empirical analysis will be published in the near future.Foundation while this work was carried out. 4 DuPont Fellow in Chemistry 1932-3. e 2 /r. As both types were fo...