ee and considerably greater speed. Litner way, of course, this approximation contributes to the overall error, the errors becoming progressively creater toward the lower energies. (2) If two or more energies are present too close together to be resolved, one can draw two or more Gaussians of the proper width beneath the broad peak representing these unresolved energies, but it is difficult if not impossible to determine which of the enert gies is preponderant. ° more than estimate. v4 In this case one can do little The enersies present in this way . therefore have larger estimated errors associated with tnem, (3) Because of statistical fluctuations due to the ' : low count rate on many samples, the curves drawn by the X-Y recorder are not smooth and the size of the peaks is somewnat questionable. rate exists, therefore, cn records where low count the associated errors are esti- mated to be larser. __ naan ene ee For comvarison betwem , areas determined by superimposition of genereted curves and areas measured vith the plenimeter, SevV- eral total absorption neaks were measured by both methods. The dispsrity between those peaks cent. | 4 so compsred ranged from 0 to = ver- This error is in genorel lass then the ststisticel error.