ceptions) is _possible_'. Again, if we know that
the object of a conception is possible, how are we to determine
whether it is also actual? It is clear that, since we cannot advance
from the mere conception, objectively real though it may be, to the
reality of the corresponding object, we need perception. The case,
however, where the corresponding object is directly perceived may be
ignored, for it involves no inference or process of thought; the
appeal is to experience alone. Therefore the question to be considered
is, 'How do we determine the actuality of the object of a conception
comparatively _a priori_, i. e. without direct experience of it[10]?'
The answer must be that we do so by finding it to be 'connected
with an actual perception in accordance with the analogies of
experience'[11]. For instance, we must establish the actuality of an
object corresponding to the conception of a volcanic eruption by
showing it to be involved, in accordance with the analogies (and with
particular empirical laws), in the state of a place which we are now
perceiving. In other words, we can say that 'that which is connected
with the material conditions of existence (sensation) is _actual_'.
Finally, since we cannot learn the existence of any object of
experience wholly _a priori_, but only relatively to another existence
already given, the necessity of the existence of an object can never
be known from conceptions, but only from its connexion with what is
perceived; this necessity, however, is not the necessity of the
existence of a substance, but only the necessity of connexion of an
unobserved state of a substance with some observed state of a
substance. Therefore we can (and indeed must) say of an unobserved
object corresponding to a conception, not only that it is real, but
also that it is necessary, when we know it to be connected with a
perceived reality 'according to universal conditions of experience';
but the necessity can be attributed only to states of substances and
not to substances themselves.'

[10] Cf. B. 279, M. 169 and p. 4, note 1.

[11] B. 273, M. 165.

Throughout this account there runs one fatal mistake, that of
supposing that we can separate our knowledge of things as possible, as
actual, and as necessary. Even if this supposition be tenable in
certain cases,[12] it is not tenable in respect of the objects of a
complex conception, with which Kant is dealing. If we know the object
of a complex conce