a story book true and those of a text of physics false, the coherence view fails utterly. Something more, that is, must be added to coherence, namely, a principle in terms of which the compatibility is to be established, and this would alone then be the actual criterion. If I am given a set of statements, among which are found some that contradict each other, I can establish consistency in a number of ways, by, for example, on one occasion selecting certain statements and abandoning or altering them and on another occasion doing the same with the other statements that contradict the first. 

Thus the coherence theory is shown to be logically impossible; it fails altogether to give an unambiguous criterion of truth, for by means of it I can arrive at any number of consistent systems of statements which are incompatible with one another. 

The only way to avoid this absurdity is not to allow any statements whatever to be abandoned or altered, but rather to specify those that are to be maintained, to which the remainder have to be accommodated. 

IV 
The coherence theory is thus disposed of, and we have in the meantime arrived at the second point of our critical considerations, namely, at the question whether all statements are corrigible, or whether there are also those that cannot be shaken. These latter would of course constitute the "basis" of all knowledge which we have been seeking, without so far being able to take any step towards it. 

By what mark, then, are we to distinguish these statements which themselves remain unaltered, while all others must be brought into agreement with them? We shall in what follows call them not "protocol statements," but, "basic statements" for it is quite dubious whether they occur at all among the protocols of science.
 
The most obvious recourse would doubtless be to find the rule for which we are searching in some kind of economy principle, namely, to say: we are to choose those as basic statements whose retention requires a minimum of alteration in the whole system of statements in order to rid it of all contradictions. 

It is worth noticing that such an economy principle would not enable us to pick out certain statements as being basic once and for all, for it might happen that with the progress of science the basic statements that served as such up to a given moment would be again degraded, if it appeared more economical to abandon them in favor of newly found statements which from that time on ?- until further notice -- would play the basic role. This would, of course, no longer be the pure coherence viewpoint,