21 
it does, as we admit, raise a probable expectation of recurrenve 
in the future. So far as the observed presence of a given ante- 
cedent, seemingly next before the consequent, raises the probability 
that we see in that antecedent the true, efficient cause ; just so 
far have we probable evidence that the consequent will follow 
it in future. 
But ordinarily the observed sequences can only raise a pro- 
bability that we} have found in the antecedent the true 
cause; for this reason: that we know there are often such 
things as unobserved, or latent, or invisible causes. For 
instance, the old empirical chemists knew that something 
turned the metal, when sufficiently heated, into the calx. 
They talked of an imponderable agent which they named 
phlogiston. They had not suspected that oxygen gas was the 
cause ; for this gas is transparent, invisible, and its presence 
in the atmosphere had not been clearly ascertained. Had the 
frequently observed sequence, then, led them to the conclu- 
sion that heat was the efficient and sufficient cause of calcina- 
tion, they would have concluded wrong. Further experiment 
has taught us this error: some metals, as potassium, calcine 
rapidly in the midst of intense cold, if atmosphere and water 
be present. None of the metals calcine under heat, if atmo- 
sphere and: water are both excluded, as well as all other 
oxygen-yielding compounds. Here, then, is the weakness 
of the induction by the mere enumeration of agreeing 
instances: We have not yet found out but that an unobserved 
cause comes between the seeming antecedent and the effect, the 
law of whose rise we wish to ascertain. 
And here is the practical object of all the canons of induc- 
tive logic, and of all the observations and experiments by 
which we make application of them; to settle that question, 
whether between this seeming antecedent and that effect, another 
hitherto undetected antecedent does not intervene? Just so 
soon as we are sure there is no other, whether it be by many 
observations or few, we know that the observed antecedent 1s 
the true efficient cause; and that we have a law of nature 
which will hold true always, unless new conditions arise, over- 
powering the causation. Not only is it possible that we may 
be assured of the absence of any undetected cause between 
the parts of the observed sequence by a few observations ; we 
may sometimes reach the certainty, and thus the permanent 
natural law, by a single one. To do so, what we need is, to 
be in circumstances which authorise us to know certainly that 
no other antecedent than the observed one can have intruded 
unobserved. Such authority may sometimes be given by the 
testimony of consciousness. For instance, a party of explorers 
