gs look and what they are, it would seem that the object of
perception cannot be the thing, but only an appearance produced by
it. Moreover, the doctrine gains in plausibility from the existence of
certain illusions in the case of which the reality to which the
illusion relates seems non-existent. For instance, if we look steadily
at the flame of a candle, and then press one eyeball with a finger, we
see, as we say, two candles;[16] but since _ex hypothesi_ there is
only one candle, it seems that what we see must be, not the candle,
but two images or appearances produced by it.

[16] Cf. Dr. Stout, on 'Things and Sensations' (_Proceedings
of the British Academy_, vol. ii).

This difficulty is raised in order to draw attention to the fact that,
in the case of the railway lines, where it can be met on its own
ground[17], this is because, and only because, we believe space to be
'real', i. e. to be a characteristic of reality, and because we
understand its nature. The distinction between the actual and the
apparent angle made by two straight lines presupposes a limiting case
in which they coincide. If the line of sight along which we observe
the point of intersection of two lines is known to be at right angles
to both lines, we expect, and rightly expect, to see the angle of
intersection as it is. Again, if we look at a short portion of two
railway lines from a point known to be directly above them, and so
distant that the effects of perspective are imperceptible, we can say
that the lines look what they are, viz. parallel. Thus, from the point
of view of the difficulty which has been raised, there is this
justification in general for saying that two lines _look_ parallel or
_look_ at right angles, that we know that in certain cases what they
look is identical with what they are. In the same way, assertions of
the type that the moon _looks_ as large as the sun receive
justification from our knowledge that two bodies of equal size and
equally distant from the observer _are_ what they look, viz. of the
same size. And in both cases the justification presupposes knowledge
of the reality of space and also such insight into its nature as
enables us to see that in certain cases there must be an identity
between what things look and what they are in respect of certain
spatial relations. Again, in such cases we see that so far is it from
being necessary to think that a thing must be perceived as it is, that
it is not only