Coherent and incoherent scattering processes are commonly ignored, or at best estimated only crudely, in X-ray fluorescence experiments designed either to measure physical constants or to effect nondestructive chemical analysis of mixtures and alloys. The influence of this scattering is of second order, but is nevertheless large enough (generally 2 4 % ) to overshadow inherent errors in experimental measurement. In this paper, we develop a formalism for calculating scattering corrections applicable to experiments involving either thin polycrystalline or amorphous samples studied in transmission or thick samples of a similar kind examined using reflection geometry. Our principal approximations are physical; we assume (i) that crystalline structure and texture in the sample serve merely to modulate the angular distribution of coherent scattering without changing its total integrated intensity, (ii) that scattering can be described adequately by smooth angular distribution functions chosen for convenience in calculation, and (iii) that multiple scattering can be ignored. Mathematical development of the model is essentially exact. Adequacy of approximation (i) is supported by previously published experimental and theoretical work; as for approximation (ii), we have found that results do not depend crucially upon the particular angular distribution function assumed for scattering. We believe, at a conservative estimate, that overall errors in our scattering corrections are in almost all cases of interest less than 25%; data to which they are applied, therefore, are potentially accurate to better than 0.5-1.25%. Three important effects have to be considered: (a) the generation of fluorescence by exciting radiation which has been scattered'within the sample, (b) the scattering of fluorescent radiation away from the acceptance aperture of the detector, and (c) the scattering of fluorescent radiation into this aperture. In our formalism, only (a) and (c) need be calculated explicitly. Corrections are expressed simply as products of quotients u/p with certain enhancement terms (p being a total attenuation constant and u an attenuation constant for scattering). In transmission geometry, the

Superoxide dismutase has been found to be widely distributed and of approximately the same specific activity in all regions of human brain examined. It is not reduced during degeneration of the basal ganglia in Huntington's Chorea. After subcellular fractionation of human and guinea-pig cerebral cortex, the highest specific activity of the enzyme was found in the soluble fraction.

A detailed study of the performance of a silicon detector system is reported, with emphasis on the feasibility of using such a system for accurate quantitative analysis of X-ray spectra covering a wide range of photon energies (2.5-50 keV). Calibration of efficiency, with an accuracy of 2 0.1% over most of this range, is described. Second order effects in the detection system have been characterized quantitatively, including escape peaks, pulses due to Compton recoil electrons, pulses whose energies are reduced because of incomplete charge collection and sum pulses. This information has been used to devise programs that permit accurate correction of spectra as recorded in a multichannel analyzer. Corrections are made for the following factors in turn; sum pulses, the second order effects enumerated above and detector efficiency. An analysis of the formation and distribution of sum pulses that represent close coincidences not resolved and eliminated by the pulse pile-up rejector is given in an Appendix. Methods are described for removing sum counts, including 'mixed' sum counts, from spectra. The removal of these counts reduces the overall count but results in essentially undistorted representations of spectra as they would be recorded in a system having infinitesimal dead-time. Some recommendations are made for improving the design of silicon detectors for quantitative spectral analysis.