Published in Philosophical Review, vol. XXXIV, No. 6, 1925. Republished as Chapter 10 in Truth, Knowledge and Causation, 1968, pp.132-149.


In two previous articles1 the writer has endeavoured to defend what may be called a liberalistic interpretation of the predicates Real, and Right. The present paper attempts a similar task with regard to the predicate True. Attention must first of all be called to some of the characters of the situations where questions of truth or falsity arise.

1. Datum and Dubitatum

When actual doubt exists concerning the truth of a proposition, it attaches in most cases only to some one element of the situation which the proposition sets forth. Thus, if, wishing to avail myself of the mails to send a gift to a friend, I present myself at the post-office window and formulate a doubt concerning the truth of the proposition 'This package weighs 1 lb.', there is no doubt either in the postmaster's mind or in mine, of the concrete presence of something, nor of its being a package, nor of its having weight. These are data, i.e., facts assumed or already admitted as self-evident to observation then and there. What is doubted, on the other hand, is whether the weight of the object concretely present (and sufficiently distinguished from the rest of the environment by the characterization 'package'), is 1 lb., or not. That is the dubitatum, i.e., the assertion ventured.2 Sometimes indeed, more than one point is at issue at once; but what we then have is really not one but several doubtful propositions, with some common datum and diverse dubitata (which latter may or may not be logically independent of one another). But in every case of a proposition actually doubted there is distinguishable something given and something ventured;3 and the two are best separated by stating explicitly the question to which the proposition is intended to be an answer. For we can then say that such elements of the proposition as figure in the question indicate the given; while the element which figures in the proposition but not in the question, constitutes the ventured -- the dubitatum. Thus if the proposition 'This package weighs 1 lb.' is intended as an answer to the question, 'How much does this package weigh?', the given is then a package and its weight in concreto; and the dubitatum is the particular weight 1 lb.

2. Truth a junction of three factors

The writer being firmly convinced that the proper method of philosophical investigation is inductive, the present inquiry will begin with the analysis of a concrete case of a proposition, e.g., that already used as an example, 'This package weighs 1 lb.'

1. It is evident that the question whether or not that proposition is true is insoluble unless some one definite meaning is attached to the term '1 lb.', for otherwise no question has really been asked. Obviously, then, the truth or falsity of the proposition is functionally dependent upon, among other things, a definition of the term '1 lb.'4 And any definition of it that can be offered is essentially of the nature of a verbal convention, and therefore theoretically quite arbitrary; that is to say, theoretically, any one such definition is as good as any other, if it be but clear and free from contradictions. What is theoretically important is that there be such a convention, but not what particular one it shall be, provided it be known.5 The meaning, which, in these United States, belongs 'officially' to the term '1 lb.' was fixed by Act of Congress, and may be changed by Act of Congress. This is to say that an Act of Congress can make false tomorrow a proposition which was true yesterday, since the truth of falsity of the proposition is dependent, among other things, upon the meaning of its predicate, and Congress can change that meaning. But, obviously, it is not to say that Congress can, by an Act passed today, bring it about that yesterday's true proposition shall no longer have been true yesterday.

The first of the factors, then, of which the truth of the proposition considered is a function, is some definition of the term 1 lb.' e.g., '1 lb. means the weight of this piece of metal'; this definition specifying what we may call the standard with which the package is to be compared, and agreement or disagreement with which determines the truth or falsity of the proposition.

2. At once, however, the question arises as to just what 'agreeing with' is to be taken to mean; and to this question no such (sweeping and at the same time determinate answer as 'correspondence' theories of truth have often attempted, is possible. For the particular meaning of 'agreeing with' is, once more, a matter of convention in each case or sort ot case. What can be said in general terms is that 'agreeing with' the standard means being 'equal to' it in the desired respect, i.e., having to it in that respect -- here that of weight -- a symmetrical and transitive relation. But whether such a relation obtains, or does not, between the object and the standard, depends on the manner of comparison selected; that is, in the case of weight, on the particular scales used, e.g., on whether the chemist's or the postmaster's. For there is no such thing as equality or its absence apart from some at least definable test. The test may not be practicable -- in which case the truth or falsity of the proposition remains unknown; but unless a test is definable the assertion or denial of equality as between the object and the standard is meaningless; and until one is defined it remains ambiguous.6

But which scales are used is, once more, something theoretically quite arbitrary, and a matter chiefly of the purposes of the occasion. This means, in the instance, that the weight of the package is not more truly ascertained by the use of the very sensitive scales at the jeweller's or the chemist's, than by the less sensitive scales at the postmaster's; but only more usefully for certain purposes, e.g., those of physics or chemistry. For when we weigh the package on the post-office scales, we give one meaning to the question we ask, and when we weigh it on the jeweller's scales we give it another, even though the same standard pound has been used, and the same words have formulated the question, in each case. For each meaning of the question there is one definite answer; not, one of them, truer than the other, but one better for such purposes as may be ours, than the other. On these purposes depends whether the more_delicate or the coarser instrument shall be used: an axe is worthless in a barber shop, and so is a razor in a lumber mill. Thus, the use of the jeweller's scales does not in the least yield a 'correction' of the weight obtained by the post-office scales, nor at all a nearer approach to the 'true' weight of the package, if our purpose is to mail the package. Correction, and nearer approach to 'the truth', are expressions which have applicability only to cases where two methods of comparison are available, both of which are relevant to the purpose which actuates us. In such cases we can speak of one as being better than the other in the proper, quantitative sense of the term 'better': e.g., if rapidity (within certain limits of accuracy) is part of our purpose, then, of two methods both of which yield results that fall within the desired limits of accuracy, one may be better than the other in the sense of more rapid. But there is a different sense of the word 'better', which is illustrated by the case of the man whose purpose is to mail the package. It is the sense in which A is said to be better than B because A (e.g., the use of the post-office scales) is perfectly satisfactory7 and B is not merely less satisfactory, but not satisfactory at all. The use of the jeweller's scales is perhaps not a very good illustration of the latter; a better illustration would be the use of scales (and such exist) sensitive enough to detect the weight of the graphite deposited by a pencil on a sheet of paper in writing one's name. It is safe to say that the answers returned by such a balance to the question, 'Does this weigh 1 lb. ?' would not correlate with the postmaster's answers any more closely than would answers based on the toss of a coin -- and would therefore be as wholly worthless as the latter for the purpose of mailing the package.

It is, however, a great and insidious temptation when -- as now -- scientific purposes are those which animate us, to ascribe to them a superior validity in all cases -- even in cases where not they but others actually rule. Utmost accuracy is a god to whom others than the scientist need not always sacrifice, for it is a god whose commands, to them, are often impertinent, i.e., irrelevant to their purposes.

It is apparent, then, that the meaning of the proposition 'This package weighs the same as the standard pound' is not unambiguously fixed until the balance to be used, and the environmental conditions under which it is to be used, have been specified. And therefore such specifications constitute another factor on which the truth or falsity of the proposition is no less functionally dependent than on the standard definition of the dubitatum. And that other factor is, as stated, again one that is theoretically arbitrary.

3. Whatever the manner of comparison specified, the comparison must actually be made and its outcome observed, if the truth or falsity of the proposition is to be not merely defined, but ascertained. It must be observed whether or not the pointer on the beam of the balance is at the zero mark on the quadrant -- whether or not a difference is felt between the colour of an object and some standard colour which it was asserted to match, etc. And these are matters of immediate, individual, sensory intujtion, which no instrument can ultimately eliminate. What instruments can do is only to translate, or magnify, differences not directly observable, into differences that are so. But the instrument furnishes no theoretical guarantee that the ultimately direct sensory intuitions of various observers, i.e., the several readings of the instrument (if an instrument has been specified) by these various observers, will agree. And if they do not agree, then there is no one answer to the question whether, for instance, the package does or does not weigh 1 lb. It is true that it weighs 1 lb. if equality of weight is defined in terms of one man's intuition of the place of the pointer on the quadrant, and false if it is defined in terms of some other man's. That is to say, once more, the meaning of the proposition has not yet been unambiguously fixed. To remove the ambiguity, one more convention is needed, which will specify the particular observer whose intuition is finally to define the sameness or difference of the weights. And the specification of such a standard observer, including his circumstances, is the third factor on which the truth of a proposition is functionally dependent.

Even the greater sensitiveness of a particular observer, which enables him to perceive differences not detectable by others, and on the basis of them perhaps to predict facts that the others will eventually perceive -- even this is a theoretically arbitrary ground for selecting him as standard observer. For this respect of superiority which he has may not be relevant to the purposes back of a given problem any more than, in the case of the instrument of comparison, was the greater sensitiveness of the jeweller's scales relevant to the purpose of mailing the package. This, however, obviously does not mean that the truth of a proposition is a matter of the success of our purposes. Whether the purposes that animate us succeed or fail is wholly irrelevant to the truth of the proposition. What is relevant to it is, not their success or failure, but their nature. That nature determines the test in terms of which equality to standard shall be defined in the instance, i.e., what the proposition considered shall mean; but not whether the proposition shall, by that test, be true rather than false. That is something to be settled only by observation.

It may seem far-fetched to suppose that two observers, looking at the position of the pointer of a balance through the same peephole in turn, should perceive that position diversely. Every person whose eyes are normal, we are convinced, must necessarily perceive it alike. But obviously this is only to say that everybody will perceive it alike, except people whose eyes are such that they perceive it otherwise. And all such terms as colour-blind, hard of hearing, abnormal, hypersensitive, crazy, stupid, feebleminded, prejudiced, astigmatic, etc., but designate classes of observers whose reports are ruled out by other people in certain cases. A man who perceives no differences of colour where the rest of us do, is called by us colour-blind. But in a world of persons like him, persons like us would be charged with 'seeing things', and our reports ruled out for many purposes -- although indeed, for certain other purposes our reports might, in such a world, be acknowledged as the only relevant ones.

4. Since purposes or interests determine not indeed the outcome of any proposed test,8 but the relevancy of it to the question of the truth of a given proposition, the problem arises, what bearing contlicts of purposes would have on the truth of the proposition. And the answer obviously is, no bearing at all in theory. For as soon as a meaning has been unambiguously assigned to the proposition, there is an absolute answer to the question whether the proposition is true. But, in practice, other considerations enter. Our social system functions not so much on the basis of contents (meanings), as on the basis of labels (words). It is to words that men mainly react in society, and such reactions to labels are what confer or remove the values that they and others seek: What is of immediate practical importance to Tom Jones is whether it is commonly pronounced 'true' that Tom Jones is a sharper; and not what exactly 'sharper' means. So far as definitions come to be considered at all in such cases, then, the practical issue will for the most part be only whether any proposed definition is such as to make the proposition true, or false. Tom Jones will insist on one definition, and the man with whom he did business on another. And the outcome of such a conflict of definitions is purely a matter of the relative might possessed by, or at the command of, the parties to the dispute.9 But since definitions themselves are neither true nor false, but only used or not used, the outcome of such a conflict has no theoretical bearing on the truth of the proposition. For all that victory in the conflict confers, is the power to attach penalties to the consequences in behaviour, of refusing to define a term in the particular way insisted upon by the victor.

5. Before we leave the example which we have been examining, close attention must be given to a point so far barely mentioned, but of great significance, and to illustrate which that example is particularly appropriate. It is this: The meaning of the proposition 'This package weighs 1 lb.' is not unambiguously fixed, as we have seen, without the specification of three things -- standard, method, and observer. But ultimately, these factors are not themselves unambiguously specifiable otherwise than by being concretely denoted: this piece of metal, this balance, this observer, each of them under the conditions present here now. But then the truth or falsity of the proposition, although rendered absolute by such completely unambiguous specification, is also at the same time rendered useless. That is, the proposition loses all scientific or practical significance, and becomes of purely historical interest. On the other hand, however, what we sought was, in the instance, a proposition true not merely at the post-office where the package is mailed, but also at the post-office which is to deliver it. And this character of 'community' or 'over-individuality', in the truth we seek, is exactly what is involved by saying that the relation of 'equality' (e.g., of the package and the standard, as to weight) is not only a symmetrical but also a transitive relation.10 That the verdict of the particular balance gives us 'equality' means, so far as symmetry is concerned, that if the object and the standard pound are each shifted to the opposite pan of the balance, the pointer remains at the same place. But so far as transitivity is concerned, it means that the outcome of the particular test at the mailing post-office will be the same as, e.g., the outcome of the test to be performed at the delivering post-office. And the only condition that will guarantee this is that the two tests shall be the same test -- which is impossible as soon as either of them has been specified in the completely concrete, i.e., unique, manner mentioned above as needed to render the proposition wholly unambiguous. And we therefore find ourselves in this dilemma: If the test has been uniquely specified, then the other test will necessarily be a different test, and it then remains doubtful whether the two will have the same outcome. And if, on the contrary, the test specified is not an individual and unique one, but only a test of a certain kind, it then remains doubtful whether the two tests, although both of them of the prescribed kind, will not, owing to accidental circumstances, have different outcomes. Thus in either case the sameness of outcome, i.e., the transitivity of the relation, i.e., the 'equality' of the weights, i.e., the (other than purely historical) truth of the proposition, remains ultimately a postulate, that is, a fond hope only. That any generalization of the subject of a proposition rests upon some such postulate as that of the so-called 'Uniformity of Nature', is a commonplace. And the above remarks only point the fact -- by no means novel, but not usually described in these terms -- that the generalization of the predicate is no less of a sheer gamble than that of the subject. This, however, furnishes no occasion to throw up our epistemological hands in despair. To say that all knowledge rests upon certain postulates, i.e., fond hopes, does not imply that to weigh the package means the same thing as to guess at its weight. To know does not mean the same as to guess. And the distinction between the two amounts to a distinction between the guesses that we cannot avoid and the guesses that we can avoid. The guesses that we cannot avoid (the 'postulates') condition the ultimate meaning of the terms of the propositions about which we can avoid guessing (e.g., the proposition that the package weighs 1 lb.) no less if such propositions be put forward as the result of a guess or as articles of sheer faith, than if they be put forward as the result of a test actually performed. To guess, in the sense in which to guess can be contrasted with to know, is thus to guess where one could test, to guess before one needs to begin.

3. The truth of Existential propositions

So much for the analysis of the concrete example with which we began. The question may now well be asked whether the view of the nature of Truth formulated on the basis of the analysis of that example fits propositions of other sorts also. And if the inquiry is to continue to proceed inductively, that question can be answered only on the basis of an examination of, at least, cases, of such sorts of propositions as might most easily appear not to fit in with the view of truth defined.

1. Propositions, or, if one prefers, interpretations of their imports, may be distinguished as Descriptive and Existential. In the Descriptive, the concrete existence of something is assumed given and the problem is, What is it ?, i.e., does a certain proposed concept (the dubitatum) fit it or not ? (in the sense of fitting, or agreeing with, already indicated). The proposition analysed was of this sort. In Existential propositions, on the other hand, a certain nature, i.e., a certain description, a certain 'what', is assumed understood, and the problem is, Is it there ?, i.e., is it discernible in concreto among the characters of any of the objects of a specified field? In Descriptive propositions the existent is the datum and the description is the dubitatum; in Existential propositions it is the description which is the datum, and the existence of something agreeing with it, which is the dubitatum.11

The word 'existence', however, is used in all sorts of ways, and it will therefore be conducive to clearness if, here again, we examine a concrete case. Thus, we may ask, 'Is there a copy of Hume's Treatise in this room ?' Now this problem, at first sight, appears to differ from such problems as 'Does a God exist?', 'Are there mermaids?', 'Is there a number the square of which is 9?', etc., in being not so much a problem of existence as one of location. That is, one might say that the existence of copies of Hume's Treatise is, in the question, not doubted, but only the location of one of them in this room; while in the other cases suggested it is the very existence of the sorts of things mentioned, that is doubted. But I do not believe any essential distinction can be maintained between the two sets of cases. For in each of the other questions also, a locus of existence is essentially involved. If it is not in each case explicitly mentioned, it is either because it is assumed to be already clearly understood, or else because the proponent of the question is not himself clear as to just what he wants to know, i.e., the question is ambiguous. Thus, in the case of the existence of God, the question is ambiguous so long as it is not specified whether the existence which God is declared to have is that of physical objects, that of mathematical entities, that of beliefs without which men cannot live righteously, etc. And the matter is plainer still in the case of the other two examples: the intended locus, in the case of the mermaids, is the sea; and in the case of the number the square of which is 9, it is integers, etc. Thus, in every question of existence there is reference to a certain realm or class of entities to which 'objective' status is given to start with; and the question is whether, among the individuals of that group one (or more) is to be found possessing, in addition to the characters which make it member of that group of 'objects', the other characters desired, viz. being a God, a copy of Hume's Treatise, a mermaid, a number the square of which is 9, etc.

And by speaking of 'objective status' I mean simply that the entity considered as having it, is, at the time, functioning for us as the terminus or limit of our thought, i.e., as meant but not as itself meaning; this being, I submit, the sole essential difference between an 'object' and a 'notion' -- the latter being definable as that which not only is meant (by a name or by another notion) but also itself means (another notion or an object).12 However, it is of the greatest importance to note here that to ask whether something exists, and to ask whether that thing is real, are two wholly distinct questions. Most anything one may be pleased to mention exists in syme realm; but what we wish to know in every case where we raise the question of existence is whether what we mention has existence in the realm of real existence, i.e., in the realm of existence which is important to us at the time. The distinction between existence and reality is quite fundamental here if confusion is to be avoided, but it cannot, for lack of space, be now more than indicated.13

2. There is another consideration to which it is necessary to advert before a fair comparison can be made between Descriptive and Existential propositions, in respect of the elements upon which Truth was asserted to be functionally dependent. It is that of the distincdon between problems of Proof and problems of Discovery, which, in the case of Descriptive propositions would be sufficiently illustrated by the two questions 'Does this package weigh 1 lb.?', and 'How many lbs. does this package weigh?', respectively. It is obvious that the second question cannot be solved except by the repeated formulation and solution of questions such as the first, each with the hypothesis of a different weight, until the correct hypothesis happens to be hit upon. Thus the direct problem is alwavs one of proof. Now the same thing is true, but not so evident, in the case of questions of Existence. Such questions as 'Is there something weighing 1 lb. in this room?', 'Is there a copy of Hume's Treatise on this shelf?', cannot be answered directly in most cases, but only on the basis of the repeated formulation and solution of questions such as 'Is a copy of Hume's Treatise this, that, the other . . . object on this shelf?'. In other words, to find out whether or not there is one, we have to examine in turn the various entities jointly constituting the realm of existence specified, and observe whether or not, in any of them, the characters understood by 'copy of Hume's Treatise' are concretely discernible. The direct existential problem also, is therefore always again a problem of proof -- here one of the form 'Is X-ness present in this?'.

And now it is evident enough that, in a proposition of the form 'X-ness is present in this' the factors of which the truth of the proposition is a function are essentially the same as those already brought to light in the case of a proposition of the form 'This has X-ness'. The differences are only that, as already noted, what in the latter is a datum is in the former the dubitatum, and vice versa; and that what, in Descriptive propositions is naturally spoken of as a method of comparison for determining equality is, although not concretely a different thing at all, more naturally described in Existential propositions as a method of inspection for determining presence.

3. Some of the bearings of the analysis of Existential propositions just set forth may be illustrated by some remarks on the relation between mathematics and physics -- a question which, since the advent of the theory of Relativity, is of greater interest than ever. That theory, it is sometimes said, gives us one all-embracing mathematical formula into which fit all the events of the physical world, and one from which the observer has been entirely eliminated; and this easily becomes an argument in support of a radical Realism.

However, as a distinguished exponent of the Theory of Relativity14 has pointed out, to eliminate the observer is one thing, and to generalize him is another and a very different thing. And it is the latter, he maintains, which Relativity has really done -- the observer, in his metrical status, being the reference system. Now, I could not agree that the observer can, even metrically, be identified with the reference system, for this, it seems to me, would be to identify, e.g., the clock with the person who reads the clock. But aside from this, I would agree both that it is very important to distinguish generalization from elimination, and that the observer has not been and cannot be eliminated.

However, a misunderstanding as to the sort of observation involved when this is said would be very easy, and, to avoid it, it is indispensable to distinguish sharply between two things that are easy to confuse and in practice often confused, viz., the mathematical and the physical import of equations or formulae. In the present instance the distinction becomes one between what may respectively be referred to as the Principle of Relativity, or of Invariance, and the Theory of Relativity. The Principle is a proposition of pure mathematics: the truth which it claims -- and which I am told it has rigidly been proved to have -- is mathematical truth, i.e., truth in the sense of agreement of the assertion which it ventures with data belonging to the same sort of a realm as do such objects as integers, roots, powers, etc., viz., to the realm of entities that are strictly and solely creatures of definition. Thus, the Principle does not even so much as whisper or hint anything concerning the entities of the physical world.

But the Theory of Relativity, on the contrary, is essentially a proposition concerning the physical world, and it is made such by introducing the supposition that the quantities entering into the statement of the Principle represent measurements. Now, that the physical world and the processes of actual or conceivable measurement are such as to yield none but measurements that can be fitted into the formula of the Principle, is a physical hypothesis, to be tested like any other hypothesis about the physical world, on the basis of appropriate observations of that world, e.g., the now famous eclipse observations. By the outcome of such observations, the Theory stands or falls; while on the contrary their outcome has not even the remotest bearing upon the truth or falsity of the Principle. This, of course, is not to say that the truth of the Principle is definable independently of observation, for it is not; but the sort of observation relevant to it is not the observation of physical objects, but that of pure mathematical entities as such -- it is the same sort of observation as that involved in establishing the truth of such a proposition, for instance, as that the cube of 4 is 64, namely the observation of the result of a mathematical computation.15

The above remarks, when generalized, turn out to mean something which, in the abstract, is obvious and commonplace enough, namely, that the truth of a proposition concerning entities (whether mathematical or logical) that are creatures of definition, is one thing; and the truth of a proposition concerning entities of some other realm, e.g., the physical, alleged to be cases of the former entities, is a very different thing and one that does not automatically follow from the former, but must on the contrary be certified by testing the truth of that allegation. But the overlooking of this in practice is easy and frequent, owing to the fact that the very same words (or symbols) are very often used to denote on the one hand pure mathematical or logical entities, and, on the other hand, physical entities.

4. The meaning, the criterion, and the knowledge of truth

We are now in a better position to formulate a general proposition concerning the meaning of truth. We may say that, in every case, the truth of a proposition means the agreement, according to some definable test, between the notion and the object which the terms of the proposition stand for. I say 'object' rather than 'fact', for 'a fact' means nothing but the import of a true proposition. And I use 'object' and 'notion' in the senses specified above. Also, I must here recall and stress the distinction already indicated between definable, defined, and performed; for that distinction marks the difference between the meaning, the criterion, and the knowledge of truth, of which much is often made.16 But it should now be apparent that that difference does not avail to eliminate reference to a test from the account of the nature of truth.

5. Analysis of cases of propositions of other sorts

The case for the view of the nature of truth now set forth will be strengthened by the examination of some propositions differing in various respects from those already considered. We may take first the consecrated 'All men are mortal'.

1. It must be noted at the outset that this proposition might be intended either as real or as verbal. Let us examine it first in its verbal interpretation. It would then declare that 'Man' means, among other things, being mortal. And the assertion that 'Man' possesses such a meaning might be made either a priori, viz., as a (partial) definition laid down by the speaker; or a posteriori, as an inductive generalization from observations of the apparent intention of persons, on occasions where they used the term 'Man'. Since definitions, are neither true nor false, but (if free from contradictions) only adopted or rejected, we need consider here only the second alternative. Now it is clear that the utterances of people, in which a given term is used17 can be observed no less than can physical things; that hypotheses concerning the intension of the term in such actual good usage can be formulated no less than can hypotheses concerning the properties of things; and that such hypotheses concerning the meaning of terms as used can be strengthened or disproved, by checking against additional cases of actual good usage, no less than can hypotheses as to the properties of things by checking against additional cases of those things. And the agreement of the hypothesis ventured with the data about which it is, can, in the verbal proposition also, be precisely defined in terms of an appropriate test; and the test can be performed and its outcome observed. The matter is but slightly complicated in the example we are considering by the fact that two hypotheses are there ventured at once: one, that when the term 'Man' was used in this, that, and the other actual utterance, 'being mortal' was part of what the speaker intended by it; and the other, that these utterances constituted a 'fair sample' of utterances in which the term 'Man' is used, and therefore warrant a probable conclusion concerning all such utterances.

2. However, the proposition 'All men are mortal' will hardly in fact be regarded by anyone as a verbal proposition. And the analysis just made is therefore useful chiefly to show that the view of truth set forth applies to verbal no less than to real propositions. That it applies to the proposition 'All men are mortal' when it is regarded as a real proposition, is obvious: That proposition, under the assumption that there are men, again involves two hypotheses, viz., first that the men actually examined were mortal -- which is a hypothesis of exactly the same type as that 'this package weighs 1 lb.'; and second that the men examined were a 'fair sample' of Men and therefore that what was true of them is probably true of all men.

3. We may next consider briefly a typically relational proposition, e.g., 'This man is brother of that man'. The truth or falsity of this proposition is a function of, first, a standard definition of the relation hypothetically asserted to exist between the two men, e.g., 'brother of, means child of the same parents as'; second, a method of comparison between the relation as defined and the relation concretely present between the men, e.g., 'a person A shall be said to be child of the same parents as a person B if the parents of B, on being asked whether A is their child, both answer affirmatively'; third, the immediate perception by a specified observer (e.g., ourselves) of what the answer has been.18

4. Lastly, we may examine a proposition similar to that just discussed, but concerned with mathematical instead of physical entities, e.g., the proposition '64 is the cube of 4'. Here again are involved, first, a definition of the relation 'cube of; second, a method of comparing the relation as defined with the relation concretely present between 64 and 4, e.g., putting the question to a trusted computer, possibly, oneself, or to a calculating machine; and thirdly the immediate observation of the result of the performance of the comparison, e.g., the observation of the computer's 'yes' or 'no', if he be other than oneself; or, if oneself, of the identity of the number calculated with the number given, etc.

6. 'The trite view of Truth'

According to the considerations that have now been set forth, Truth might, like Right, be said either to be something absolutely relative, or something relatively absolute. That is, the truth of a proposition is relative to the specification of the three factors already described -- but how they shall be specified is something which is theoretically arbitrary, i.e., absolute in the same sense as verbal definitions. On the other hand, relatively to any one such arbitrary set of specifications, the truth of a proposition is absolute and final.

Such a view of Truth is perhaps best characterized as a liberalistic one. And it is obvious what would be involved if, instead of contenting oneself with describing or naming this view, one were to claim it to be the true view of Truth! The claim could of course be made without involving theoretical inconsistencies, but so far as settling it goes, the most that on that very view of Truth could be done, would be to settle it in a manner consistent, for instance with the use of terms, the interests, and the intuitive observations, of most of us. But the others, with their uses of terms, their interests, and their immediate intuitions, would continue to exist in spite of our calling them names meaning that they do not 'count' (when we do the counting). No such procedure will bully them into nonentity, and, in our very own sense of the word 'true', propositions true for us would not necessarily be true for them. The liberalistic position thus involves that if any man is to be brought to a conviction of his errors -- whether in his view of Truth or of anything else -- it needs must be by words out of his own mouth.


1 'A Defence of Ontological Liberalism,' Journal of Philosophy, vol. XXI, 13; 'Liberalism in Ethics', International Journal of Ethics, April, 1925.

2 To say that what the proposition asserts is that the object present is a member jointly of the two classes 'packages' and 'things weighing 1 lb.', would be wholly to misrepresent its import. The membership of the object in the class 'packages' is not asserted but assumed, not doubted but admitted. An important instance of a misrepresentation of this sort is furnished by the statement often heard nowadays, that the contradictory of 'All men are mortal' is not the traditional 'Some men are not mortal', but instead, 'Either there are no men, or some men are not mortal.' This statement constitutes a grave error, which rests on the failure to distinguish between assertion and assumption. In this instance, the usual assumption is that there are men, and under this assumption, the traditional contradictory is the correct one. Of course, we do not have to assume that there are men. But not to assume that, does not mean that no assumption is made. For to say that 'Either there are no men, or some men are not mortal' is the contradictory of 'All men are mortal' is correct only under the assumption that the existence of men is not taken for granted. This assumption, indeed, is an assumption about assumptions about men, instead of about men, but it is an assumption just the same, and one which is no less arbitrary than the other. It is very desirable, of course, to make our assumptions as clear and explicit as possible, but this can be done without traversing the fundamentally important distinction between assumption and assertion.

3 Such a proposition as 'The nightingale now perched on the end of my pen is singing', appears to constitute an exception to the assertion just made. However, the form 'The... which now ...', is relevant only if what is presupposed by it, viz., that there is a nightingale there, is admitted. But if on the contrary this is not admitted, then the use of that form gives rise to a self-irrelevant (not self-contradictory) subject; just as, in the proposition 'You have not quit beating your mother', we have (if the person referred to is not admitted to have been doing it), a self-irrelevant predicate. And if, in either case, the presupposition is not only not admitted, but false, then we have a fallacy of false insinuation in subject or predicate. In such a case the proposition about the nightingale would really then be a false existential proposition, viz., there is a singing nightingale now perched on my pen -- in which datum and dubitatum can be distinguished. See below, section 3.

4 Were it objected that what we have with each such definition is a different proposition, the reply would be that this is to confuse a proposition with the import of a proposition. There is between them a useful and well established distinction, and no occasion to traverse it arises. What we do have is one proposition, but an ambiguous one, cf. 'A Criticism of Scepticism and Relativism' by R. M. Blake, Journal of Philosophy, vol. XXI, 10, pp. 169-70; and the writer's article already referred to, in the International Journal of Ethics.

5 What the convention shall be is, as Whewell has pointed out (Novum Organon Renovatum, pp. 36-7), often a matter of certain propositions which we desire to make true by the definition. However, such definitions ad hoc are permissible only when, or so far as, the terms defined do not already possess an established meaning in the language, for otherwise nothing but confusion results. The meaning of most terms is obviously a much more ancient affair than that of the term '1 lb.', but it is not on that account any less essentially of the nature of a verbal convention. The convention in their case is tacit and passively inherited, instead of explicit and actively entered into; it is of those which most essentially constitute the particular tongue to which the word belongs. And to accept it, i.e., to speak that tongue as already made, is the most convenient thing for us to do. But no such convention is logically binding, since, if we but specify another, we shall be equally intelligible.

6 An illustration of a proposition in the case of which a test is, if not perhaps undefinable, at least difficult to define, would be the proposition that in the 'Bohr steady state' of the hydrogen atom, the electron is in motion. The 'Bohr steady state' is a state of perfect equilibrium of energy in the atom, and this means that the atom does not then radiate and is therefore not merely too small to be seen, but essentially unseeable, untouchable, etc.; in short, unobservable even supposititiously, because the supposition that it is observed contradicts the supposition of the perfect equilibrium of energy in it. But how else than in terms of a supposititious observation can a meaning be given to the question whether or not the electron is at the same place at differeqt instants? Yet, unless it can be done, the proposition is meaningless; and until it is done, unintelligible, cf. W. S. Franklin, 'The Quantum Puzzlej and Time', Science, Sept. 19, 1924. Also 'The Non-Existence of Time', by the writer, in the Journal of Philosophy, vol. XXII, 1.

7 In such a case, the Absolute could not know the true weight of the package more perfectly than we do.

8 Including in the meaning of 'test' both a method of comparison and an observer.

9 cf. 'Liberalism in Ethics', loc cit.

10 i.e. As the reader will recall, a relation such that if it exists between A and B, and also between B and C, then it exists also between A and C.

11 cf. Royce, The World and the Individual, vol. I, p. 49; Coffey, The Science of Logic, vol. I, p. 101, apud Satolli; Aristotle, Anal. Post., bk. II, chapters VII, VIII.

12 I have elsewhere endeavoured to defend in detail these definitions of 'object' and 'notion'. See Causation and the Types of Necessity, Univ. of Washington Press, pp. 103-10.

13 I have argued it at length in the paper on 'A Defence of Ontological Liberalism' already mentioned. Some comments on the views of Russell and Johnson on reality and existence will be found on pp. 345-6.

14 Lyndon Bolton, in a lecture on 'Materialism and the Theory of Relativity', read at Sion College, London, Nov. 25, 1924, of which he has very kindly allowed me to see the manuscript, cf. His Introduction to the Theory of Relativity, ch. XXI.

15 See below section 5, art. 4.

16 In a very rough way it might perhaps be said that insistence upon either the first, or the second, or the third, wellnigh to the point of blindness to the others, is what most nearly characterizes the essential difference between the coherence, the pragmatic, and the self-evidence theories of truth. But each of these has been so variously stated, and withal usually with such ambiguity, that any unambiguous assertion concerning them would probably be foredoomed to futility.

17 To observe how people use a word is a very different thing from asking them to say what they mean by it. The definition they then attempt is their theory of what they mean by the word, but actual cases of their manner of using it are the facts which the theory must fit; and there is no a priori reason to believe that their theory is more likely to be correct than that of someone else. And by their actual usage of a word, is not meant the nature that the object to which they apply the word, has, but the nature which, all unanalysed, they believe that object to have.

18 The temptation is strong here to say that even the supposition of an affirmative answer does not really define the truth of the proposition. The parents of B might be lying, one objects, and that would not change the fact that A either is, or is not, their child. But this objection only means that the particular test suggested above might have an outcome different from that of some other test, which other is perhaps the only one relevant to our purposes on the occasion; it does not mean that the truth of the proposition is independent of any test whatever. The test suggested above might, e.g., be relevant for legal purposes; but it would not be directly relevant to the purposes, e.g., of an investigation into the transmissibility of acquired characters; and vice versa. This amounts to saying that 'child of' may be interpreted to mean 'legally child of' or, no less, 'physiologically child of'.