THE DISJUNCTIVE JUDGMENT. 499 The knowledge that we have discovered two classes within our genus is dependent on the proposition that ' No B is C,' but not only on this but on the fact that B and C carry with them peculiar properties, so that in virtue of this statement as premiss we can go on to deny DEF, etc., which are pro- perties of c, of B. It is obvious that ' No B is C ' must be indemonstrable, e.g., that no lake fish, belonging to genus A, inhabit the streams ( = F), or though we may appear to prove it through the fact that they feed on a species of insect found only in lakes, we must have previously explained this fact by their being in the lakes. Or, to put the matter in another light, we may prove that No B is C by assuming that No B is D, but this negation has to be itself derived from ' No B is C '. We must select one of those negations as an ulti- mate premiss, and we should take as the terms of our ultimate those properties which seem to be the cause of the subsequent differentiations. The negation No B is C is indemonstrable (just as the definition, B is B&, is indemon- strable). Thus we see that when we have two terms B and C each involving along with it a number of attributes in such a way that by denying B and C of each other we are able to deny of the other the attributes involved in each, we are entitled to raise B and C into the distinguishing marks of species. When once this has been done it is merely going over old ground again to deny the properties of one class of the other. Now it is when B and C are regarded as specific differences that the modus ponendo tollens is valid, but when valid it is valueless. " Ked fleshed trout are lake-dwellers, therefore they are not to be found in the streams." This is true, but is based upon the mutual exclusion of classes, whose distinc- tion had already been justified by the difference of their properties. There are cases of disjunction in which we are not specify- ing the various classes which fall within a genus and in which it would yet appear that we might draw a valid conclusion from them by the modus ponendo tollens. But in all such cases the mutually exclusive predicates must be known to be incompatible with each other in virtue of some principle antecedent and superior to the particular science which deals with the subject of predication. The predicates must be specific differentiations of a wider reality. Thus we may say, ' Planetary orbits fall either wholly inside or wholly outside the earth's orbit '. We can therefore infer, if we care to take the trouble for such a trifling purpose, that Jupiter's orbit, lying without that of the earth, cannot lie
