§ 2. The main problems confronting us in this examination will then be—(1) How far is calculable uniformity of sequence really incompatible with the presence of purpose and intelligence? (2) Have we any real ground for ascribing _such_ uniformity to the actual sequences of physical nature? (3) if not, What is the real logical character of the principle of the so-called uniformity of nature? (4) and What amount of truth is contained in the conception of the physical order as a _mechanism_? Into the problem suggested by the popular contrast between the _necessity_ of mechanical sequence and the _freedom_ of purposive action, it will be needless to enter at any length. For, as we saw in dealing with the popular view of necessary causal relation, the necessity of a mechanical sequence is a purely subjective and logical one. The sequence is necessary only in the sense that _we_ are constrained, so long as we adhere to the purpose of thinking logically, to affirm the consequent when we affirm the antecedent. True _necessity_ is always compulsion, and therefore, so far from being opposed to purposive action, can only exist where an actual purpose is overruled or thwarted.[127] So long as we are dealing solely with phenomenal sequence in the physical order, necessity is a mere anthropomorphic name for routine undeviating uniformity of sequence. (1) _Calculable Uniformity and Intelligent Purpose._ It is sometimes assumed that all successful _prediction_ of a thing’s behaviour is incompatible with the ascription of intelligence or purpose to the thing. Thus it has been argued, and continues to be argued in moral philosophy of a popular type, that if we are intelligent beings with purposes of our own, it must always be impossible for an onlooker to predict how we shall behave in circumstances which have not yet arisen. This extreme view of the incompatibility of calculability with intelligent purpose, however, manifestly rests on a double confusion. To begin with, those who assert this view commonly make the mistake of supposing that _prediction_ of the future stands somehow on a different logical level from _calculation_ of the past from present data. Prediction of my future behaviour is supposed somehow to conflict with my character as a purposive being in a way in which inference as to my past behaviour does not. This is, of course, an elementary fallacy in Logic. The conditions required for the successful inference of the absent from the present are identical in the two cases, as we have already seen in dealing with the problems of Causality. Precisely the same kind of insight is requisite to judge how a given man must have behaved in a certain situation in his past history as are needed to determine how he will behave in a situation which is yet to arise. We may thus dismiss from consideration the special case of _prediction_, and confine ourselves to the general question, how far the general