=98.= In this way, then, it may be possible to prove the _general_ utility, for the present, of those actions, which in our society are _both_ generally recognised as duties _and_ generally practised; but it seems very doubtful whether a conclusive case can be established for any proposed change in social custom, without an independent investigation of what things are good or bad in themselves. 159 =99.= And (_d_) if we consider the distinct question of how a single individual should decide to act (α) in cases where the _general_ utility of the action in question is certain, (β) in other cases: there seems reason for thinking that, with regard to (α), where the generally useful rule is also generally observed, he should _always_ conform to it; but these reasons are not conclusive, if _either_ the general observance _or_ the general utility is wanting: 162 =100.= and that (β) in all other cases, _rules of action_ should not be followed at all, but the individual should consider what positive goods, _he_, in his particular circumstances, seems likely to be able to effect, and what evils to avoid. 164 =101.= (4) It follows further that the distinction denoted by the terms ‘duty’ and ‘expediency’ is not primarily ethical: when we ask ‘Is this really expedient?’ we are asking precisely the same question as when we ask ‘Is this my duty?,’ viz. ‘Is this a means to the best possible?.’ ‘Duties’ are mainly distinguished by the non-ethical marks (1) that many people are often tempted to avoid them, (2) that their most prominent effects are on others than the agent, (3) that they excite the moral sentiments: so far as they _are_ distinguished by an ethical peculiarity, this is not that they are peculiarly useful to perform, but that they are peculiarly useful to sanction. 167 =102.= The distinction between ‘duty’ and ‘interest’ is also, in the main, the same non-ethical distinction: but the term ‘interested’ does also refer to a distinct ethical predicate--that an action is to ‘my interest’ asserts only that it will have the best possible effects of one particular kind, not that its total effects will be the best possible. 170 =103.= (5) We may further see that ‘virtues’ are not to be defined as dispositions that are good in themselves: they are not necessarily more than dispositions to perform actions generally good as means, and of these, for the most part, only those classed as ‘duties’ in accordance with section (4). It follows that to decide whether a disposition is or is not ‘virtuous’ involves the difficult causal investigation discussed in section (3); and that what is a virtue in one state of society may not be so in another. 171