air, per unit time, i.e. the flux weighted by the photon energy and mass energy transport coefficient for air. The flux was calculated down to a low energy cutoff of To + 13.2 (16 = 2211), this low energy cutoff has been found to include over 99% of the energy except for the highest energies and for very large interface to detector distances. For these latter cases a correction was made by interpolating the differential energy spectrum down to lower energies. For all calculations, annihilation radiation, bremsstrahlung, and coherent scattering were neglected. The polynomial expansion method can be expected to provide exposure rate estimates for scattered y-rays of accuracy better than 5%'7) In many cases the scattered component is smaller than the unscattered component. The latter was calculated exactly and any error in it is due only to errors in the cross section data. Thus, the error in the total exposure rate is smaller than the error in the scattered component. To improve the calculational procedure, the values of the differential scattered flux and the exposure rate at the source energy were calculated directly using an exact expression. Inasmuch as the y-ray cross section data are felt to be quite accurate (better than 2%) for the source energies and media used in our calculations, we conclude that the error in expoSure rate values as well as in the differential energy spectra and the integral exposure rate spectra is always less than +5%. The angular distributions are not as accurate and the error here may be as much as +10% or more for the scattered component when the detector is near the interface‘*?.,