This paper undertakes a critical examination of the concept of 'centre of gravity' as adapted by economics from classical mechanics, relating it to the idea of 'long-run' profits, prices and quantities, as presented in the work of the post-Sraffians.(1) It will also address the origin of this concept of 'long-run' in Marshall's distinction between long-run and short-run determinations of economic magnitudes. It shows that economists have generally conceived of centre of gravity as a theoretical magnitude which is not observed, but around which observed magnitudes oscillate either randomly or in some deterministic manner; this much is generally agreed. This idea has, however, been interpreted in two distinct ways in the history of economic thought: (1) as an attractor dynamically determined at each point in time by path-dependent historical processes which have led the economy to be in its present state. (2) as a hypothetical static equilibrium state of the economy determined independent of history by its current exogenous parameters (utility, technical capacity, etc) It demonstrates that these two ideas are necessarily distinct and that both must be taken into account in any pluralistic research programme. Mathematically the attractor of a variable is not in general equal to its hypothetical static equilibrium, except in highly restricted circumstances such as the absence of technical change. Moreover, again outside of exceptional circumstances, the divergence between the predictions of observed magnitudes given by the two approaches increases over time, so that it cannot even be accepted that one converges on the other. Error will therefore result if it is assumed a priori that (1) is identical to (2). The fact that the two conceptions lead to different predictions does not decide that either one is correct. This should be determined empirically and therefore, an agreed empirical test should be established by the community of social scientists or, better still, society. The paper will argue that, empirically, the 'test variable' against which both conceptions should be checked is the time average of the variables in question. This is not a distinct concept of 'centre of gravity' but an empirical observable. In a pluralistic programme, the predictions of both conceptions should be evaluated against this proposed test variable. The second part of the paper examines the common basis for the critical stance taken by both Keynes and Marx to the second conception, which is rooted in a common attitude to the relation between substance and accident, and a correspondingly similar conception of uncertainty. It will relate this to the work of Quetelet and the development of the statistical method in sociology which, it will argue, is rooted in an ontologically distinct conception of social magnitudes to that found in economics, closer to the concept which Keynes and Marx shared. It argues that the post-Sraffian conception of long-run is based on a fallacious identification of these two distinct concepts, rendering the post-Sraffian approach equally incompatible with Keynes's and Marx's theories. It argues that the post-Sraffian conception of centre of gravity is 'intrinsically antipluralistic' in that it depends absolutely on the conflation of two concepts which are in fact necessarily distinct, leading to the suppression of the non-equilibrium concept as an alternative to the scientific procedure of testing the predictions of both concepts against an observable.