50 THE METHOD OF SCIENCE 



observation. 1 In any case, since the inductions reached through 

 the laboratory are so few and widely separated ; since, in effect, 

 there is no system of reality to which appeal can be made, or at 

 most only a very fragmentary system, every induction must remain 

 isolated and untested. Under such conditions, the only kind of 

 deduction possible is one which expands the induction without 

 testing it or linking it up with any other, which cannot be tested 

 itself, and which, therefore, is valueless and illegitimate. 2 We 

 shall see that, almost without exception, every important biological 

 induction based on laboratory work important in the sense of having 

 attracted wide attention has been expanded in this way, and that it 

 is the deduction, not the induction^ on which stress is laid and which is 

 regarded as the Theory. Moreover, not only have untested induc- 

 tions often been expanded by untested deductions, but, as often, 

 the latter have been regarded as proved simply and solely because 

 the phenomena on which they are remotely based were, like 

 those of physics and chemistry, observed in the laboratory. 3 

 It is true that we are exhorted to use experiment as a test, 

 but, especially when the larger problems of heredity are under 

 consideration, the advice is rarely heeded even by those who 

 utter it. In effect, biologists have used the laboratory only as a 

 means of discovery. 



80. The theories (laws) of physicists and chemists have usually 

 stood the test of time, not because their facts were experimentally 

 observed, nor even because their data were capable of being 

 exactly measured, but chiefly because, though the facts were 

 difficult to gather, the thinking founded on them was in many 

 cases relatively simple, and, more especially, because it was in all 

 cases very rigorously tested. The formulation of a physical or 

 chemical law often consists in little more than a bare summary ot 

 the facts discovered in an experiment or series of experiments, a 

 summary almost as easy to make as a classification of the facts of 

 anatomy. In other words the relations of the facts are comparatively 

 easy to discern. Compare, for example, Boyle's law of the 



1 See 225 et seq. 



z " The deductions which Bacon abolished were from premises hastily 

 snatched up or arbitrarily assumed. The principles were neither established by 

 legitimate canons of experimental inquiry, nor the results tested by that indis- 

 pensable element of a rational Deductive Method, verification by specific experience. 

 Between the primitive method of deduction and that which I have attempted 

 to characterize, there is all the difference which exists between the Aristotelian 

 physics and the Newtonian theory of the heavens." (J. S. Mill, Logic, III. xiii. 7.) 



3 See 208 et seq., see also chapters vii. and viii. 



