The effect of thus contra-distinguishing 'collocation'from causation,and admitting that 'uncaused coexistences'cover a large part of all observable phenomena,is to make the uniformity of nature exceedingly precarious.Indeed,Mill denies it to be conclusively proved.The chapter in which he sums up 'the evidence of the law of universal causation'leads to remarkable results.No one,he thinks,with a properly trained imagination will find any difficulty in conceiving that in remote parts of the universe 'events may succeed one another at random without any fixed law'.He concludes by asserting that it would be 'folly to affirm confidently'that the law does prevail in 'distant parts of the stellar regions.'(85)A truth which depends upon locality might,for anything one sees,break down in Australia and even at Paris as easily as at Sirius.Mill,accordingly,reaches a thoroughly sceptical conclusion,(86)and reduces the evidence for universal causation to an induction per enumerationem simplicem.The wider the generalisation,the greater the efficacy of such induction,upon which depends not only the law of causation but the principles of number and geometry.If the 'subject-matter of any generalisation'be so widely diffused that it can be tested at every time and place,and if it 'be never found otherwise than true'its truth cannot depend on any collocations,unless such as exist at all times and places,nor can it be frustrated by any counteracting agencies except such as never actually occur.'(87)Now no exception to the 'law of causation'has ever been found,and apparent exceptions have only confirmed it.It is no doubt true that if a law be universal,it will be confirmed by all our experiments;but it hardly follows that,because all our experiments have failed to detect an exception,it is true universally.All our experiments have covered but a small fragment of nature,and they do not justify us,as he expressly asserts,in reasoning about the stellar regions.It is difficult,moreover,to see how an 'exception'could ever be proved,since,wherever we do not see a 'cause,'we can always suppose,and do in fact suppose,an invisible cause.Finally,the theory of 'natural kinds,'as it has now been interpreted,seems to fail us in our need.He takes it to indicate,indeed,that there are connections in nature,which,if known,would justify certain general inferences;but it does not appear that we can know what are these connections,and as,moreover,we have been carefully told that they are ultimate or not 'derivative,'we have no right to be certain that they will recur.We do not know,for example,whether blackness be a property of kind.If we found a black crow among white ones,the property would be casual,and therefore 'caused.'If we found a race of white crows in Australia,we should simply say that there was a kind hitherto overlooked.(88)Such a discovery,he says,is not at all incredible.It might be proved by the evidence of a single credible witness.It merely supposes that there is a kind with a different set of attributes,and as the attributes are in no way 'derivative,'there is no improbability in this.The more general the rule,however,the greater the probability of its holding,because the greater the improbability of the exception being overlooked.We should easily believe in white crows,but not so easily in crows with a property 'at variance with any generally recognised universal property of birds,'and still less,if it were 'at variance with such a property of animals.'(89)We could hardly,that is,believe in crows with the stomachs of wolves,or in crows without stomachs at all.But the difficulty appears to depend upon nothing else than the improbability that such animals,had they existed,would have been unnoticed.
Without trying fully to unravel this logic,we may notice one characteristic.Mill,trying to refer everything to 'experience,'has gone far to make experience impossible.What has dropped out of this theory of knowledge is the constructive part.He substitutes for organisation combinations of disparate 'things.'
He will admit of no logic,except that.of an external connection of radically different objects.Attributes must be stuck together without any reciprocal relations.All causation becomes in his phrase 'collocation,'though he declares causation and collocation to be not only different but mutually exclusive.His one logical formula is the nota notae est nota rei ipsius.(90)Things are marks of each other,not implied by each other.He forces this language even upon mathematics.Even a geometrical 'kind,'if we may use the word,an ellipse,or a curve of the second order,is treated by the formula applicable to purely empirical conjunctions.The equality of two straight lines,it seems,is simply a 'mark'that if applied to each other they would coincide;the fact that two things are sums of equals is a 'mark'that they are equal,and so forth.(91)He would apparently be inclined to say that a thing's existence is a mark of its not being nothing.Thus,even the 'natural kind'becomes merely a permanent combination.When the properties of a curve are merely connected by 'marks,'it is no wonder that the properties of crows should be mere bundles.If it is only on such terms that we can thoroughly get rid of 'intuitions,'the advantage is doubtful.
I will venture to say another word upon the uniformity of Nature difficulty.It is easy,says Mill,to conceive of things happening at random.It is,indeed,in one sense perfectly easy.
