On Scientific Method. 
489 
1877.] 
conservation of energy. For if we do not do so we are very 
probably assuming the incorrectness of that great body of 
facts upon which the theory rests. It would in most cases 
be almost better to distrust our personally-observed facts 
than to distrust so well-founded a generalisation. This is 
one view to take of the question. But, on the other hand, 
the theory of the conservation of energy is a theory only : 
it is probably true ; we do not, and cannot, know whether it 
is or is not certainly true. If the observed faCts, after the 
most careful observations, still remain unmoved, and if they 
are apparently opposed to the generally-accepted theory, 
the better method will doubtless be complete reservation of 
judgment until further experimental data is forthcoming. 
If the observed faCts are, however, absolutely opposed to the 
theory, and if these faCts cannot be gainsaid, then the theory 
must go ; it has done its work, and must be supplanted by 
a wider generalisation. The scientific investigator must 
therefore cling to theory, and yet be ready to abandon theory 
at the call of facft. 
It is, it seems to me, of the utmost importance to insist on 
this view of the work of the student of Nature ; to declare 
that he trusts Nature altogether, but he distrusts his own 
powers of comprehending the workings of Nature ; that he 
feels that all things are changing, but he nevertheless clings 
to what he can grasp of the changeless. The frame of 
mind of the man of Science is, then, at once opposed to 
those who would have us believe that “ victorious analysis ” 
has now at last reduced all things under her feet, and to 
those who would have us accept the teaching of authority 
in place of the teaching of faCts. Both alike assume a 
vast amount of knowledge which neither is possessed of. 
But the last characteristic of a good scientific hypothesis is 
its readiness to submit to have its predictions proved by 
strictly experimental methods. Every newly-discovered fadt 
which is capable of explanation in terms of an accepted 
hypothesis adds something to the probable truth of that 
hypothesis. Every newly-discovered fadt which cannot be 
explained in terms of the hypothesis takes away something 
from the probable truth of the hypothesis. We may ob- 
serve fadts which are apparently opposed to the hypothesis 
which we have provisionally accepted, and yet we may not 
be justified in condemning the hypothesis, because these 
fadts may either be but partially examined by 11s, or the 
hypothesis may not have been fully grasped in all its 
bearings. But if the hypothesis is to hold its ground there 
must be no experimentally demonstrated faCt, the existence 
