sort of way that other authority is transmissible. Hence, if I say it lightly, I may be responsible for getting you into
trouble.
If you say you know something, the most immediate challenge takes the form of asking 'Are you in a
position to know?': that is, you must undertake to show, not merely that you are sure of it, but that it is within your
cognisance. There is a similar form of challenge in the case of promising: fully intending is not enough—you must
also undertake to show that 'you are in a position to promise', that is, that it is within your power. Over these points
in the two cases parallel series of doubts are apt to infect philosophers, on the ground that I cannot foresee the
future. Some begin to hold that I should never, or practically never, say I know anything—perhaps only what I am
sensing at this moment: others, that I should never, or practically never, say I promise—perhaps only what is
actually within
end p.100
my power at this moment. In both cases there is an obsession: if I know I can't be wrong, so I can't have
the right to say I know, and if I promise I can't fail, so I can't have the right to say I promise. And in both cases this
obsession fastens on my inability to make predictions as the root of the matter, meaning by predictions claims to
know the future. But this is doubly mistaken in both cases. As has been seen, we may be perfectly justified in
saying we know or we promise, in spite of the fact that things 'may' turn out badly, and it's a more or less serious
matter for us if they do. And further, it is overlooked that the conditions which must be satisfied if I am to show
that a thing is within my cognisance or within my power are conditions, not about the future, but about the present
and the past: it is not demanded that I do more than believe about the future.1
We feel, however, an objection to saying that 'I know' performs the same sort of function as 'I promise'. It
is this. Supposing that things turn out badly, then we say, on the one hand 'You're proved wrong, so you didn't
know', but on the other hand 'You've failed to perform, although you did promise'. I believe that this contrast is
more apparent than real. The sense in which you 'did promise' is that you did say you promised (did say 'I
promise'): and you did say you knew. That is the gravamen of the charge against you when you let us down, after
we have taken your word. But it may well transpire that you never fully intended to do it, or that you had concrete
reason to suppose that you wouldn't be able to do it (it might even be manifestly impossible), and in another 'sense'
of promise you can't then have promised to do it, so that you didn't promise.
Consider the use of other phrases analogous to 'I know' and 'I promise'. Suppose, instead of 'I know', I had
said 'I swear': in that case, upon the opposite appearing, we should say, exactly as in the promising case, 'You did
swear, but you were
end p.101
wrong'. Suppose again that, instead of 'I promise', I had said 'I guarantee' (e.g. to protect you from attack):
in that case, upon my letting you down, you can say, exactly as in the knowing case 'You said you guaranteed it,
but you didn't guarantee it'.1 Can the situation perhaps be summed up as follows? In these 'ritual' cases, the
approved case is one where in the appropriate circumstances, I say a certain formula: e.g. 'I do' when standing,
unmarried or a widower, beside woman, unmarried or a widow and not within the prohibited degrees of
relationship, before a clergyman, registrar, &c., or 'I give' when it is mine to give, &c., or 'I order' when I have the
authority to, &c. But now, if the situation transpires to have been in some way not orthodox (I was already married:
it wasn't mine to give: I had no authority to order), then we tend to be rather hesitant about how to put it, as heaven
was when the saint blessed the penguins. We call the man a bigamist, but his second marriage was not a marriage,
is null and void (a useful formula in many cases for avoiding saying either 'he did' or 'he didn't'): he did 'order' me
to do it, but, having no authority over me, he couldn't 'order' me: he did warn me it was going to charge, but it
wasn't or anyway I knew much more about it than he did, so in a way he couldn't warn me, didn't warn me.2 We
hesitate between 'He didn't order me', 'He had no right to order me', 'He oughtn't to have said he ordered me', just as
we do between 'You didn't know', 'You can't have known', 'You had no right to say you knew' (these perhaps
having slightly different nuances, according to what precisely it is that has gone wrong). But the essential factors
end p.102
are (a) You said you knew: you said you promised (b) You were mistaken: you didn't perform. The
hesitancy concerns only the precise way in which we are to round on the original 'I know' or 'I promise'.
To suppose that 'I know' is a descriptive phrase, is only one example of the descriptive fallacy, so common
in philosophy. Even if some language is now purely descriptive, language was not in origin so, and much of it is
still not so. Utterance of obvious ritual phrases, in the appropriate circumstances, is not describing the action we
are doing, but doing it ('I do'): in other cases it functions, like tone and expression, or again like punctuation and
mood, as an intimation that we are employing language in some special way ('I warn', 'I ask', 'I define'). Such
phrases cannot, strictly, be lies, though they can 'imply' lies, as 'I promise' implies that I fully intend, which may be
untrue.
If these are the main and multifarious points that arise in familiar cases where we ask 'How do you know
that this is a case of so-and-so?', they may be expected to arise likewise in cases where we say 'I know he is angry'.

There is a good deal of freedom in 'structuring' the history of someone's activities by means of words like
'intention', just as when we consider a whole war we can divide it into campaigns, operations, actions, and the like;
but this is fairly arbitrary except in so far as it is based upon the plans of the contestants. So with human activities;
we can assess them in terms of intentions, purposes, ultimate objectives, and the like, but there is much that is
arbitrary about this unless we take the way the agent himself did actually structure it in his mind before the event.
Now the word 'intention' has from this point of view a most important bracketing effect: when the till-dipper claims
that he intended all along to put the money back, what he is claiming is that his action—the action that he was
engaged upon—is to be judged as a whole, not just a part of it carved out of the whole. Nearly always, of course,
such a contention as this will carry with it a contention that his action (as a whole) is not to be described by the
term chosen to describe (only a part of) it: for example, here, it was not 'robbing' the till, because the action taken
as a whole would not result in the absence of any money from the till. Reculer pour mieux sauter is not to retreat.
Quite distinct is the use of the word 'purpose'. Certainly, when I am doing something for a purpose, this
will be known to me, like my intentions, and will guide my conduct. Indeed, like an objective, a purpose will
influence the forming of intentions. But my purpose is something to be achieved or effected as a result of what I'm
doing, like the death of my aunt, or the sickness of the penguins if I did indeed feed them peanuts on
end p.285
purpose. (Very commonly my purpose is to put myself into a position to be able to go on with the next
action, the next operation in the campaign.) I need not, however, have any purpose in acting (even intentionally);1
just as I need not take care or thought. I act for or on (a) purpose, I achieve it; I act with the intention, I carry it out,
realize it.
I act deliberately when I have deliberated—which means when I have stopped to ask myself, 'Shall I or
shan't I?' and then decided to do X, which I did. That is to say, I weighed up, in however rudimentary (sometimes
almost notional or fictional) a fashion, the pros and cons. And it is understood that there must be some cons, even
when what I do deliberately is something unexceptionable, such as paying my taxes. The pros and cons are not
confined to moral pros and cons, nor need I decide in favour of what I think best or what has the most reasons in
favour of it. (Nor, of course, when I have decided to do it, must I necessarily carry it out.) Deliberation is not just
any kind of thinking prior to action: to act with forethought or with premeditation, or to think about ways and
means—all these show that we took thought, perhaps over a period of time, but none of these shows that we were
acting deliberately, and indeed are quite distinct matters from deliberation. Ways and means are a matter for the
planning staff; decision is a matter for the commander. That there should be slowness in moving into action or
conducting it (so much relied on by lawyers) is the merest symptom.
I will close by adding a general word of warning: there are overriding considerations, which may be
operative in any situation in which I act, which may put all three words out of joint, in spite of the other standard
conditions for their use being satisfied. For instance, I may be acting under a threat: however much I weigh up the
pros and cons, if I act under the influence of a threat I do not do that act deliberately. This
end p.286
sort of overriding consideration must always be allowed for in any case.1
end p.287
13 The Line and the Cave in Plato's Republic

This reconstruction of Austin's views is based on three sources. There are notes dating from the 1930s for a reply to a
paper (so far as I know unpublished) by W. D. Ross on the metaphysics of the Republic and the Phaedo. These notes are very
full and the first half of the printed paper is very faithful to them; alterations are mainly excisions of comments on Ross which
cannot be read profitably without Ross's paper and which do not advance Austin's argument. The later parts of these notes rely
heavily on the view that Plato, like Aristotle, always used the word 'hypothesis' to mean an existential postulate; Austin came
to doubt this at a later date. For the second half of the paper I have therefore made considerable use of Austin's own notes for a
class held at Oxford in the late 1940s, and of notes taken at this class by Professor Hugh Lloyd-Jones, with a modified thesis
about the nature of hypothesis. I am grateful to Professor Lloyd-Jones for allowing me to see his notes. I transliterate and
translate what Austin left in Greek; I, not he, am responsible for the English names given to the segments of the line.
j. o. urmson

Too much has already been written on the interpretation of the Line and Cave in Plato's Republic (509-18).
In Britain in the present century, to omit other references, there have been elaborate discussions in Adam's edition
of the Republic; in Ferguson's articles in the Classical Quarterly of 1921, 1922, and 1934; by Stocks in the
Classical Quarterly of 1911; by Murphy in the Classical Quarterly of 1934; by Paton in the Proceedings of the
Aristotelian Society of 1921-2; and by Hardie in his Study in Plato of 1936. In this paper I shall assume that the
reader is
end p.288