PART I
Material Fallacies

In terms of the analogy of a manufacturing process (p. 4), the final section of this book will examine the misfunctioning of the formal machinery of argument, where the product is so poorly fitted together that the conclusion dangles unsupported in reason. The second source of error is the emotional tricks and appeals which distract from the process of reasonable discussion and tempt disputants to hold all sorts of untruths and half-truths to be self-evident. These, the vulgar sources of error and befuddlement, the claptrap of argument, will have their turn in the next section. There remain the errors that come from raw material of poor quality., When the material is bad, it can be small wonder if the product is also unsatisfactory, however carefully processed, however careful the manufacturer . This failure in the materials is described in the present section, where we take up some of the more common ways in which the materials of argument fall below standard.

To find good standards, the best thing to do is to examine a model product. The field of science abounds in clear examples of sound reasoning. The very model of a convincing argument built on reliable evidence is a scientific argument. Let us look at these model arguments, the demonstrations of science, and we may see more clearly how some other arguments fall short.

The various physical sciences establish their general laws by working backwards, as it were, from observational sentences, sentences known to bo true in experience. The laws can be considered as premises that lead to the observations as conclusions: if the laws arc true, the observations have to be true, too. This is the backwards effect, since, after all, it is the observations that are known to be true -- the laws are inferred from them. But in the demonstrations, the arguments in scientific writing, the observations are "derived" from the laws by a vigorous process of logical or mathematical proof.

The laws are so designed as to be perfectly inclusive; that is, no known observations contradict them., Moreover, no other plausible premises are known from which the observations could follow. The laws are economically drafted: each has as wide a scope as the facts allow, and two laws never stand where one would do.^Finally, all scientific laws are consistent with one another.

In all cases new observations can be predicted, not mere duplications of previous data (such as laboratory experiments in school), but actually new experiments. When Einstein published his theory, in addition to taking into account all relevant past observations in a way no other theory had succeeded in doing, he was also able to predict further observations that would be logical consequences of the laws he had discovered. The observations were made as soon as feasible, for example those connected with the bending of light rays. They "confirmed" the laws, since they could not have been predicted as a consequence of any other intelligible hypothesis.

Why, in their arguments, do the scientists treat the laws as "premises"? Why, that is, do they want to express the observations as conclusions derivable from the laws? They want to know exactly where they stand. If only one predicted observation should contradict the "law," then the latter would become a discarded hypothesis. Moreover, though the observations are derived from the law, they do not prove the law true in Jurn, no matter how numerous they are. The so-called la^s remairyhypotheses.

What would it mean to "prove" a law? Until the time of David Hume it had been thought that there was a "necessary connection" between the law and the observations, the sort of relation that there is in geometry between the theorem and the postulates and axioms. Hume showed that in the case of empirical laws, one can always imagine the sun rising in the west, gravitation working in reverse, water freezing at 100° C. But where there is a necessary connection, the contrary case is inconceivable. I cannot conceive of a prime number between 7 and 11. I cannot conceive of my being both present and absent at the same time, in the literal sense of these words. I can say these things, but I cannot say them without contradiction. I can say that 8, or 9, or 10 is a prime number, but I cannot say so and mean by "prime number" or by "8," "9," "10" what mathematicians mean by them. On the other hand, there is no logical contradiction at all in speaking of water running uphill.

There is no mathematical certainty, but the probability that the laws of science hold is enormous . They may be regarded as generalizations with no exceptions. Moreover, these generalizations are much stronger than the sort of naive generalizations traditionally discussed by philosophers of science, such ajJlA]J,.i;ri^^jireJilacJsil" or J. S. Mill's "All hyacinths are blue.". The generalizations of science never stand isolated. Rather, they are interwoven into more and more complex statements, like a web drawn together at certain points. The web endures as a whole, the strength of each strand contributing to the strength of the others. The "induction" for a generalization of physical science is thus not a simple leap from positive instances, free from the occurrence of a negative instance, to an "all" statement. It is a moving from strand to strand in a tightly woven lattice, Everything known about the world, or nearly everything, supports what is known about any small part of it.

Though there cannot be certainty that every occurrence will con-form to the laws, throughout all time and space, there is still perfect certainty that every known instance is derivable from them and, moreover, that a counter instance can not occur if the laws arc true. The absence of a counter or negative instance is a necessary condition for the truth of a given law. The presence of such an instance would be a sufficient condition for the falsification of the Iaw. ln this respect, it would seem that the given law is no better off than the generalization about hyacinths. Logically it is not, but practically it is, even apart from the matter of cross-inference lattices just discussed. I The experimental conditions are so well defined that scientists know exactly how to test the laws. This is to say, they characteristically know what experimental or observational procedures to set up for finding the negative instance if it has the remotest probability of occurring. They don't have to sit around waiting for it to show itself,.

With the model of scientific law before us, what can we say about the principles and generalizations by which we must, in our every-day problems, attempt understanding and venture deeds?

A step-by-step comparison is hardly necessary. It is all too evident that ordinary life wisdom is a tissue .of vague categories. where truth is relative to ignorance, which is vast, where procedures are clumsy and blind. Far from weaving a tight lattice of systematic investigations, individuals make isolated observations. Since man must understand so that he may move and act, he leaps to some hasty generalization and "induces" some broad principles from scraps of evidence.

The common man's hypotheses sometimes fail to survive their first test. As to the so-called laws, such as the laws of human conduct, they are often incapable of confirmation with the means at hand ("Democracy is the most efficient form of government," "A world state is the only check to world war"). Responsible persons wish to act on principle, so they affirm their generalizations on faith or pretend to believe in them while the crisis of action lasts.

This is the human predicament.ylt has been well expressed by I. A. Richards. (From Practical Criticism, New York: Harcourt, Brace & Company, 1929.)

There are subjects -- mathematics, physics and the descriptive sciences supply some of them -- which can be discussed in terms of verifiable fact and precise hypothesis. There are other subjects -- the concrete affairs of commerce, law, organization, and police work -- which can be handled by^ rule of thumb and generally accepted convention. But in between is the vast corpus of problems, assumptions, adumbrations, fictions, prejudices, tenets; the sphere of random belief and hopeful guesses; the whole world, in brief, of abstract opinion and disputation about matters of feeling. To this world belongs everything about which civilized man cares most.

It is not always the case, of course, that people must understand and act at once, getting on with the evidence at hand. Sometimes they can wait for better evidence and continue to gather it. Sometimes, when action is forced on them, they act well and meet with signal success. What a man can always do is act in humility. He can learn to regard his hypotheses as tentative aids to understanding, rather than as eternal principles or absolute dogmas. When the pressure of events forces a man to take sides, to do what he can, he should recognize that he is engaging in a trial-and-error process, which, though he fail, may still afford rich experience for future guidance.

Even in the "sphere of random beliefs and hopeful guesses," there are some reasons better than others. All of the materials out of which the common man builds his arguments may be far from scientific standards, but some are considerably further than others. There are many occasions of error in the gathering and arranging of the evidence from which to build arguments.

In an argument this evidence must be marshalled in a way to lead somewhere. Where it leads is called a "conclusion.". Evidence cannot lead to a Conclusion unless it jfi rarpfiilly nr^nr\J7^e\ in fl, ^ngfipfr way,

that is, unless it can be cast into the form called "premises. . The premises not only must lead to a conclusion, but they must be true to the evidence also.yMaking sure that the premises are true to the evidence is a problem of meaning: premises must mean no more than the evidence supporting them. Moreover, they must be intelligible and clear, for the conclusion will be vague or haphazard if the premises are -- you cannot get out more at the end of an argument than goes in at the beginning. Casting the evidence into the form of cogent premises, with the language clear and intelligible, is also a problem of meaning. The whole process, then, of setting up true premises for an argument is a matter of meaning, of what is called '^semantics.'. There are two fairly distinct problems:

a. is the evidence truly and fairly represented in the premises?

b. are the premises which represent the evidence clear and in-rtefligible?

A simple example may make these necessary distinctions clear. If one wished to make an argument to show that certain forms of advertising are good for the consumer as well as for the business man, he would want to be sure that the general propositions he was advancing as premises represented the facts as they are, and that the classification of the kinds of advertising treated was clear and consistent. Could he assert (£j> that all advertising must, say, appeal to some real need of the consumer? Is the distinction (b) between, say, prestige sponsorship of a symphony orchestra and direct product pushing clearly drawn and easily applicable?

We shall first treat the problems of stating the facts right, of correctly representing the known situation in the premises. Here occur the familiar fallacies of hasty generalization, post hoc arguments, faulty analogies.^After treating of these and related fallacies, we shall turn to the problems of vague classifications, word magic, and the rest of the nightmare horde that haunt the semanticistr