530 



FALLACIES. 



CHAPTER VII. 



FALLACIES OF CONFUSION. 



§ I. Under this fifth and last class 

 it is convenient to arrange all those 

 fallacies in which the source of error 

 is not so much a false estimate of 

 the probative force of known evi- 

 dence, as an indistinct, indefinite, and 

 fluctuating conception of what the 

 evidence is. 



At the head of these stands that 

 multitudinous body of fallacious rea- 

 sonings in which the source of error 

 is the ambiguity of terms : when 

 something which is true, if a word be 

 used in a particular sense, is reasoned 

 on as if it were true in another sense. 

 In such a case there is not a mal- 

 estimation of evidence, because there 

 is not properly any evidence to the 

 point at all ; there is evidence, but to 

 a different point, which, from a con- 

 fused apprehension of the meaning of 

 the terms used, is supposed to be the 

 same. This error will naturally be 

 oftener committed in our ratiocina- 

 tions than in our direct inductions, 

 because in the former we are decipher- 

 ing our own or other people's notes, 

 while in the latter we have the things 

 themselves present, either to the 

 senses or to the memory. Except, 

 indeed, when the induction is not 

 from individual cases to a generality, 

 but from generalities to a still higher 

 generalisation ; in that case the fal- 

 lacy of ambiguity may affect the in- 

 ductive process as well as the ratioci- 

 native. It occurs in ratiocination in 

 two ways : when the middle term is 

 ambiguous, or when one of the terms 

 of the syllogism is taken in one sense 

 in the premises and in another sense 

 in the conclusion. 



Some good exemplifications of this 

 fallacy are given by Archbishop 

 Whately. "One case," says he, " which 

 may be regarded as coming under the 

 head of Ambiguous Middle, is (what 

 I believe logical writers mean by 'Fal- 

 lacia Figxrce Dictionis ') the fallacy 

 built on the grammatical structure of 



language, from men's usually taking 

 for granted that paronymous (or con- 

 jugate) words, i.e. those belonging to 

 each other, as the substantive, adjec- 

 tive, verb, &c., of the same root, have 

 a precisely corresponding meaning ; 

 which is by no means universally the 

 case. Such a fallacy could not indeed 

 be even exhibited in strict logical form, 

 which would preclude even the at- 

 tempt at it, since it has two middle 

 terms in sound as well as sense. But 

 nothing is more common in practice 

 than to vary continually the terms 

 employed, with a view to grammatical 

 convenience ; nor is there anything 

 unfair in such a practice, as long as 

 the meaning is preserved unaltered ; 

 e.g. * murder should be punished with 

 death ; this man is a murderer, there- 

 fore he deserves to die,' &c. Here we 

 proceed on the assumption (in this 

 case just) that to commit murder, 

 and to be a murderer, — to deserve 

 death, and to be one who ought to 

 die, are, respectively, equivalent ex- 

 pressions ; and it would frequently 

 prove a heavy inconvenience to be 

 debarred this kind of liberty ; but 

 the abuse of it gives rise to the fal- 

 lacy in question ; e.g. projectors are \ 

 unfit to be trusted ; this man has] 

 formed a project, therefore he is unfit 

 to be trusted : here the sophist pro- 

 ceeds on the hypothesis that he whojj 

 forms a project must be a projector ; 

 whereas the bad sense that commonlyj 

 attaches to the latter word, is not at j 

 all implied in the former. This fal-, 

 lacy may often be considered as lying j 

 not in the Middle, but in one of thej 

 terms of the Conclusion ; so that! 

 the conclusion drawn shall not be, in| 

 reality, at all warranted by the pre- 

 mises, though it will appear to be so, 

 by means of the grammatical affinity! 

 of the words : e.g. to be acquainted 

 with the guilty is a presumption ofj 

 guilt ; this man is so acquaintec" 

 therefore we may presume that he ia 

 guilty : this argument proceeds on' 

 the supposition of an exact corre- 

 spondence between presume and pre- 

 sumption, which, however, does not 



