Mr Ross is very ready to agree that the use of this principle can never allow us to attain certainty as to what in particular is our duty. And he mentions two main causes of uncertainty. In the first place, each act obviously originates countless effects, many of which are unforeseeable, and of these unforeseeable effects some at least have moral significance (i.e. make a contribution to the balance of prima facie rightness and wrongness). We cannot, then, attain certainty that we are ever choosing the act which really promotes the greatest balance of prima facie Tightness over prima facie wrongness. The most that we can do is to reflect as carefully and as comprehensively as we can upon the consequences of our possible acts, in the confidence that we are 'more likely' to do our duty thus than by acting without thought. In the second place - to give Mr Ross's own words - 'for the estimation of the comparative stringency of these prima facie obligations no general rules can, so far as I can see, be laid down. We can only say that a great deal of stringency belongs to the duties of "perfect obligation" - the duties of keeping our promises, of repairing wrongs we have done, and of returning the equivalent of services we have received. For the rest, έv çή àίåθήåîι ή κρίåις' (pp. 41, 42).

Now the first of these 'reasons for uncertainty' raises no special difficulty. A similarly conditioned uncertainty of particular moral judgments is admitted by most ethical theories, and the validity of the ethical theory is not impugned thereby. An ethical theory must supply us with a principle for moral guidance, but there is no a priori necessity that the principle should be such that the conditions of its practical application are simple. But the second reason for uncertainty indicated by Mr Ross appears to me to belong to a different category altogether. The first reason for uncertainty was just that we never have complete information as to the facts. But the second reason - our lack of ' general rules' for ' estimating the comparative stringency of these prima facie obligations' - means that we don't know how to appraise the information that we have: that even if our information as to the fact was perfect - even if the ' countless effects' were fully known in respect of their contributions to the different prima facie Tightnesses and wrongnesses - we still should not be able to tell what our 'actual' duty is. For, being ignorant of the 'comparative stringency' of the different prima facie obligations, we have not the knowledge by which we can even begin to calculate which act promotes the greatest balance of prima facie Tightness over prima facie wrongness. And if this is so, we have here a defect which, unlike the former, does impugn the validity of the ethical theory.

And surely it is so. 'Calculation' implies the reducibility to a common measure of the things to be calculated. But it is just this reducibility to a common measure which seems here to be clearly impossible. Our 'common measure' should, I suppose, be units of prima facie Tightness on the positive side, and units of prima facie wrongness on the negative side. But on the theory we do not, and apparently can not, know the relative number of units which belong to each prima facie obligation. According to Mr Ross, we know (doubtless by intuition) that certain characteristics of acts (e.g. the characteristic of promise-keeping) possess more units of prima facie Tightness than certain other characteristics (e.g. the characteristic of relieving distress). Such, at any rate, seems to be implied in the known 'greater stringency' of the 'duties of perfect obligation.' But we do not know how many more units the one possesses than the other. And such knowledge is indispensable to moral calculation. It may not seem to be required if we have a simple case where we have to choose between keeping a promise on the one hand and relieving distress on the other, no further moral factors being relevant (though no case is really quite so simple, and certain difficulties which we shall notice later are ignored in the very postulation of such a case). It might seem that here our knowledge that promise-keeping has more prima facie Tightness than relieving distress is all that we need, since our ignorance of how much more Tightness belongs to the former will not introduce any difficulty into the determination of which act is our 'actual' duty. But whatever we may say of these 'simple' cases, our ignorance matters vitally in those more complicated cases in which we have to choose, let us say, between keeping a promise, on the one hand, and, on the other hand, relieving distress plus manifesting gratitude to benefactors. In cases like this, it is absolutely impossible to discover what is our 'actual' duty, what act will produce the greatest balance of prima facie Tightness over prima facie wrongness, unless we can assess the values of the different obligations relatively to one another in terms of the number of units of prima facie Tightness which each possesses. It is not that the calculation is, otherwise, a difficult one. The point is that it is in principle impossible. We are asked to do a sum which we do not, and can not, know how to do.

It is all-important to appreciate in its proper character the 'uncertainty' which belongs to our particular moral judgments in terms of this aspect of the theory under review. It is quite other than the uncertainty which is due to our inadequate information as the effects of our action - an uncertainty which we can progressively diminish (thereby advancing towards 'more probable' opinions) by widening the scope of our reflections. The present 'uncertainty' is one which no amount of reflection will even begin to diminish. In respect of this cause of uncertainty, we have no means of advancing towards 'more probable' opinions.1 The impasse which confronts the moral self is absolute. Even were the self's information as to 'effects' ideally complete, it would not know what to do with its information in order to determine what act is its duty. And if the admission of this is not tantamount to ethical scepticism, the difference is not easy to detect.