Aristotle distinguishes (as has been already stated) three purposes in the dialogue:— (1) Teaching and Learning; (2) Contention, where both questioner and respondent strive only for victory; (3) Investigating and Testing the consequences of some given doctrine.402 The first two of these three are dismissed rapidly. In the first, the teaching questioner has no intention of deceiving, and the pupil respondent has only to answer by granting all that appears to him true.403 In the second, Aristotle tells us only that the questioner must always appear as if he were making some point of his own; while the respondent, on his side, must always appear as if no point were made against him.404 But in regard to the third head — dialogues of Search, Testing, Exercise — he is more copious in suggestions: he considers these as the proper field of Dialectic, and, as we saw, claims to have been the first who treated them apart from the didactic dialogues on one side, and the contentious on the other.405

The thesis which the respondent undertakes to defend (in a dialogue of Search or Testing) must be either probable, or improbable, or neither one nor the other. The probability or improbability may be either simple and absolute, or special and relative — in the estimation of the respondent himself or of some one or more persons. Now, if the thesis be improbable, the opposite thereof, which you the questioner try to prove, must be probable; if the thesis be probable, the opposite thereof must be improbable; if the thesis be neither, its opposite will also be neither. Suppose, first, that the thesis is improbable absolutely. In that case, its opposite, which you the questioner must fish for premisses to prove, will be probable; the respondent therefore ought not to grant you any demand which is either simply improbable or less probable than the conclusion which you aim at proving; for no such concessions can really serve your purpose, since you are bound to prove your conclusion from premisses more probable than itself.406 Suppose, next, that the thesis is probable absolutely. In that case, the opposite conclusion, which you have to make out, will be improbable absolutely. Accordingly, whenever you ask concessions that are probable, the respondent ought to grant them; whenever you ask for concessions that are less improbable than your intended conclusion, he ought to grant these also; but, if you ask for any thing more improbable than your intended conclusion, he ought to refuse it.407 Suppose, thirdly, that the thesis is neither probable nor improbable. Here, too, the respondent ought to grant all concessions that appear to him probable, as well as all that he thinks more probable than the opposite conclusion which you are seeking to arrive at; but no others. This is sufficient for the purpose of Dialectic, and for keeping open the lines of probable argument.408

When the probability or improbability of the thesis is considered simply and absolutely, the respondent ought to measure his concessions by the standard of opinion received usually.409 When the probability or improbability of the thesis is considered as referable to the respondent himself, he has only to consult his own judgment and estimation in granting or refusing what is asked. When he undertakes to defend a thesis avowedly as the doctrine of some known philosopher, such as Herakleitus, he must, in giving his answers, measure probability and improbability according to what Herakleitus would determine.410

Since all the questions that you ask must be either probable, improbable, or neuter, and either relevant411 or not relevant to your purpose of refuting the thesis, let us first suppose that you ask for a concession which is in itself probable, but not relevant. The respondent ought to grant it, adding that he thinks it probable. If what you ask is neither probable nor relevant, he ought even then to grant it; but annexing a notification that he is aware of its improbability, in order to save his own credit for intelligence.412 If it be both probable and relevant, he ought to say that he is aware of its probability, but that it is too closely connected with the thesis, and that, if he grants it, the thesis will stand refuted. If it be relevant, yet at the same time very improbable, he must reply that, if he grants it, the thesis will be refuted, but that it is too silly to be propounded. If, being neutral, it is also not relevant, he ought to grant it without comment; but if, being neutral, it is relevant, he ought to notify that he is aware that by granting it his thesis will be refuted.413

In this way of proceeding, the march of the dialogue on both sides will be creditable. The respondent, signifying plainly that he understands the full consequences of his own concessions, will not appear to be worsted through any short-comings of his own, but only through what is inherent in his thesis; while you the questioner, having asked for such premisses as are really more probable than the conclusion to be established, and having had them granted, will have made out your point. It must be understood that you ought not to try to prove your conclusion from premisses less probable than itself; and that, if you put questions of this sort, you transgress the rules of dialectical procedure.414

If you ask a dialectical question in plain and univocal language, the respondent is bound to answer Yes or No. But if you ask it in terms obscure or equivocal, he is not obliged to answer thus directly. He is at liberty to tell you that he does not understand the question; he ought to have no scruple in telling you so, if such is really the fact. Suppose the terms of your question to be familiar, but equivocal; the answer to it may perhaps be either true or false, alike in all the different senses of the terms. In that case, the respondent ought to answer Yes or No directly. But, if the answer would be an affirmation in one sense of the terms and a negation in another, he must take care to signify that he is aware of the equivocation, and to distinguish at once the two-fold meaning; for, if the distinction is not noticed till afterwards, he cannot clearly show that he was aware of it from the first. If he really was not at first aware of the equivocation, and gave an affirmative answer looking only to one among the several distinct meanings, you will try to convict him of error by pushing him on the other meaning. The best thing that he can then do will be to confess his oversight, and to excuse himself by saying that misconception is easy where the same term or the same proposition may mean several different things.415

Suppose you put several particular questions (or several analogous questions) with the view of arriving ultimately by induction at the concession of an universal, comprising them all. If they are all both true and probable, the respondent must concede them all severally; yet he may still intend to answer No, when the universal is tendered to him after them. He has no right to answer thus, however, unless he can produce some contradictory particular instance, real or apparent, to justify him; and, if he does so without such justification, he is a perverse dialectician.416 Perhaps he may try to sustain his denegation of the universal, after having conceded many particulars, by a counter-attack founded on some chain of paradoxical reasoning such as that of Zeno against motion; there being many such paradoxes contradictory of probabilities, yet hard to refute. But this is no sufficient justification for refusing to admit the universal, when, after having admitted many particulars, he can produce no particular adverse to them. The case will be still worse, if he refuses to admit the universal, having neither any adverse instance, nor any counter-ratiocinative attack. It is then the extreme of perverse Dialectic.417

Before the respondent undertakes to defend any thesis or definition, he ought to have previously studied the various modes attacking it, and to have prepared himself for meeting them.418 He must also be cautious of taking up improbable theses, in either of the senses of improbable. For a thesis is so called when it involves strange and paradoxical developments, as if a man lays down either that every thing is in motion or that nothing is in motion; and also, when it implies a discreditable character and is contrary to that which men wish to be thought to hold, as, for example, the doctrine that pleasure is the good, or that it is better to do wrong than to suffer wrong. If a man defends such theses as these, people hate him because they presume that he is not merely propounding them as matter for dialectical argument, but advocating them as convictions of his own.419

The respondent must farther be able, if you bring against him a false syllogistic reasoning, to distinguish upon which among your premisses the false conclusion really turns, and to refute that one. Your reasoning may have more than one false premiss; but he must not content himself with refuting any one or any other: he must single out that one which is the chief determining cause of the falsehood. Thus, if your syllogism be:— Every man in a sitting position is writing, Sokrates is a man in a sitting position; therefore, Sokrates is writing, — it will not suffice that the respondent should refute your minor premiss, though this may be false;420 because such a refutation will not apply to the number of other cases in which men are sitting but not writing; and therefore it will not expose the full bearing of the falsehood. Your major premiss is that upon which the full bearing of the falsehood depends; and the respondent must show that he is aware of this by refuting your major.421

This last-mentioned proceeding — refutation of that premiss upon which your false conclusion in its full bearing really turns — is the only regular, valid, and complete objection whereby the respondent can stop out your syllogistic approaches. There are indeed three other modes of objection to which he may resort; but these are all either inconclusive or unfair. He may turn his objection against you personally; and, without refuting any of your premisses, he may thus perplex and confuse you, so that you are disqualified from pursuing the thread of your questions. Or he may turn his objections against portions of your questions; not refuting any one of your premisses, but showing that, as they stand, they are insufficient to warrant the conclusion which you seek to establish; when, if you are master of your subject, and retain your calmness, you will at once supply the deficiency by putting additional questions, so that his objection thus vanishes. Or, lastly, he may multiply irrelevant objections against time, for the purpose of prolonging the discussion and tiring you out.422 Of these four modes of objection open to the respondent the first is the only one truly valid and conclusive; the three others are obstructions either surmountable or unfair, and the last is the most discreditable of all.423

To blame the argumentative procedure and to blame the questioner are two distinct things. Perhaps your manner of conducting the interrogation, preparatory to your final syllogism, may be open to censure; yet nevertheless you the questioner may deserve no censure; for it may be the respondent’s fault, not yours. He may refuse to grant the very premisses which are essential to the good conduct of your case; he may resort to perverse evasions and contradictions for the mere purpose of thwarting you; so that you are forced to adapt yourself to his unworthy manœuvres rather than to aim at the thesis itself. Dialectic cannot be well conducted unless both the partners do their duty to the common purpose; the bad conduct of your respondent puts you out, and the dialectic presently degenerates on both sides into angry contention.424 Apart from this, too, it must be remembered that the express purpose of Dialectic is not to teach, but to search and test consequences and to exercise the intellect of both parties. Accordingly you are not always restricted to true syllogistic premisses and conclusions. You are allowed to resort occasionally to false premisses and false conclusions; for, if what the respondent advances be true, you have no means of refuting it except by falsehood; and, if what he advances be false, the best way of refuting it may be through some other falsehood.425 You render service to him by doing so; for, since his beliefs are contrary to truth, if the dialogue is confined to his beliefs, the result may perhaps contribute to persuade him, but it will not instruct or profit him.426 It is your business to bring him round and emancipate him from these erroneous beliefs; but you must accomplish this in a manner truly dialectical, and not contentious; whether you proceed by true or by false conclusions.427 If you on your side, indeed, put questions in a contentious spirit, it is you that are to blame. But often the respondent is most to blame, when he refuses to grant what he thinks probable, and when he does not apprehend what you really intend to ask.428 He is sometimes also to blame for granting what he ought to refuse; such as Petitio Principii or Affirmation of Contraries. It is often difficult to distinguish what questions involve Petitio Principii or Affirmation of Contraries: they are asked and granted without either party being aware, and the like mistake is committed by men in private talk, not merely in formal dialogue. When this happens, the argument will inevitably be a bad one; but the fault is with the respondent who, having before refused what he ought to have granted, now grants what he ought to refuse.429

Such then are the cases in which the conduct of the dialogue is open to censure, without any fault on your part as questioner. But there are other cases in which the fault is really yours. These are five in number:— (1) When all or most of your questions are so framed as to elicit premisses either false or improbable, so that neither the conclusion which you seek to obtain, nor any other conclusion at all, follows from them; (2) When, from similar defects, the proper conclusion that you seek to obtain cannot be drawn from your premisses; (3) When the proper conclusion would follow, if certain additions were made to your premisses, but such additions are of a character worse than the premisses already obtained, and are even less probable than the conclusion itself; (4) When you have accumulated a superfluous multitude of premisses, so that the proper conclusion does not follow from all of them but from a part of them only (5) When your premisses are more improbable and less trustworthy than the proper conclusion, or when, though true, they are harder and more troublesome to prove than the problem itself.430

In regard to the last item, however, the fault may sometimes be in the problem itself rather than in you as questioner. Some problems, being in their own nature hard and not to be settled from probable or plausible data, ought not to be admitted into Dialectic. All that can be required from you as questioner is that you shall know and obtain the most probable premisses that the problem admits: your procedure may be thus in itself blameable, yet it may even deserve praise, having regard to the problem, if this last be very intractable; or it may be in itself praiseworthy, yet blameable in regard to the problem, if the problem admit of being settled by premisses still more probable.431 You may even be more blameable, if you obtain your conclusion but obtain it from improbable premisses, than if you failed to obtain it; the premisses required to make it complete being true and probable and not of capital importance, but being refused by the respondent.432 However, you ought not to be blamed if you obtain your true and proper conclusion but obtain it through premisses in themselves false; for this is recognized in analytical theory as possible: if the conclusion is false, the premisses (one or both) must be false, but a true conclusion may be drawn from false premisses.433

When you have obtained your premisses and proved a conclusion, these same premisses will not serve as proof of any other proposition separate and independent of the conclusion; such may sometimes seem to be the case, but it is a mere sophistical delusion. If your premisses are both of them probable, your conclusion may in some cases be more probable than either.434

One other matter yet remains in which your procedure as questioner may be blameable. The premisses through which you prove your conclusion may be long and unnecessarily multiplied; the conclusion may be such that you ought to have obtained it through fewer, yet equally pertinent premisses.435

The cases in which your argument will carry the clearest evidence, impressing itself even on the most vulgar minds, are those in which you obtain such premisses as will enable you to draw your final conclusion without asking any farther concessions. But this will rarely happen. Even after you have obtained all the premisses substantially necessary to your final conclusion, you will generally be forced to draw out two or more prosyllogisms or preliminary syllogisms, and to ask the assent of the respondent to these, before you can venture to enunciate the final conclusion. This second grade of evidence is however sufficient, even if the premisses fall short of the highest probability.436