So compromises have to be made to get at a common 'truth.' It must be recognized that not everything which is believed to be 'knowledge' is knowledge. In fact, it is safer to assume that none have knowledge, though all think they have; to say fact, men only have 'opinions,' which may be nearer to or farther from 'the truth,' but are not of necessity as unquestionable as they seem to be. Out of this concession to the social life arise three problems. How are 'opinions' to be compared with each other, and how is the extent of their 'truth' or 'error' to be determined? How is the belief in absolute truth to be interpreted and discounted? How is the penitent dogmatist, once he has allowed doubt to corrupt his self-confidence, to be stopped from doubting all things and turning sceptic? As regards the first problem, the first question is whether we shall try to _test_ opinions and to arrive at a standard of value by which to measure them by comparing the opinions themselves with one another, or shall presume that there must be some absolute standard which alone is truly true, whether we are aware of it or not. The former view is _relativism_, the latter is _absolutism_, in the matter of truth. Now, there can be no doubt that absolutism is more congenial to our natural prejudices. Accordingly it is the method tried first; but it soon conducts dogmatism to an awkward series of dilemmas. 1. If there is absolute truth, who has it? and who can use the absolute criterion of opinions it is supposed to form? Not, surely, everyone who _thinks_ he has. It will never do to let every dogmatist vote for himself and condemn all others. That way war and madness lie. Until there is absolute agreement, there cannot be absolute truth. 2. But absolute truth may still be reverenced as an ideal, to save us from the scepticism to which a complete relativity of truth would lead. But would it save us? If it is admitted that no one can arrogate to himself its possession, what use is it to believe that it is an ideal? For if no one can assume that he has it, all _human_ truth is, in fact, such as the relativist asserted, and scepticism is just as inevitable as before. It makes no difference to the sceptical inference whether there is no absolute truth, or whether it is unattained by man, and human unattainable. 3. It was a mistake, therefore, to admit that opinions cannot be compared together. Some are much more certain than others, and, indeed, 'self-evident' and 'intuitive.' Let us therefore take these to be 'truer.' If so, the thinker who feels most certain he is right is most likely to be right.