analytic level. To fail in this entails the fallacy of Particular reductions stand or fall according to the test of evidence, and the behavioral sciences have seen many premature reductions fail. With respect for Parsons' early concerns, it is only fair to mention their persistence in the apparent differences between animate and inanimate nature and within the latter between symbolic and infra-symbolic behavior, with which any comprehensive naturalistic reduction will have to come to terms. But there is no general fallacy of reductionism. In his opposition to the reduction of norms, Parsons has consistently given an argument which, though offered as an account of the conditions necessary and sufficient if norms are to be variables in an explanatory system, is actually an argument for according them a certain ontological status-the status of the eternally independent variable. This is not required for norms to be explanatory variables. Consider norms as factor B. If B explains C, but A explains B, then A explains C; but this hardly implies, as the opponents of reduction so often assert, that all theories of the effect of B on C are otiose, or that B becomes epiphenomenal. The explanation of B does not cause the effect of B on C to cease; that effect becomes instead a special case of the effect of A. Sociologists already entertain a partial explanation of norms in the argument that, whatever their specific source, they have evolved in response to the requirements of viable social organization. This hypothesis, a very general one, does not immediately account for much normative variation. Some day the behavioral sciences may be able to explain this remaining variation-the conditions under which men hold to one value rather than another and under which norms respecting these values prevail in particular groups. But such explanation does not obviate accounts of the effect of particular values on social institutions. Rather, it increases the range and precision of such accounts by subsuming them under a truly more general theory.