366 FRAGMENTS OF SCIENCE. 



it in fact ; your thoughts wander amid the very atoms of 

 your steel, and you conclude that each atom is a magnet, 

 and that the force exerted by the strip of steel is the mere 

 summation or resultant of the forces of its ultimate par- 

 ticles. 



Here, then, is an exhibition of power which we can call 

 forth or cause to disappear at pleasure. "We magnetize our 

 strip of steel by drawing it along the pole of a magnet ; 

 we can demagnetize it, or reverse its magnetism, by prop- 

 erly drawing it along the same pole in the opposite direc- 

 tion. What, then, is the real nature of this wondrous 

 change ? What is it that takes place among the atoms of 

 the steel when the substance is magnetized ? The question 

 leads us beyond the region of sense, and into that of imagi- 

 nation. This faculty, indeed, is the divining-rod of 'the man 

 of science. Not, however, an imagination which catches 

 its creations from the air, but one informed and inspired by 

 facts, capable of seizing firmly on a physical image as a 

 principle, of discerning its consequences, and of devising 

 means whereby these forecasts of thought may be brought 

 to an experimental test. If such a principle be adequate to 

 account for all the phenomena, if from an assumed cause 

 the observed facts necessarily follow, we call the assump- 

 tion a theory, and, once possessing it, we can not only re- 

 vive at pleasure facts already known, but we can predict 

 others which we have never seen. Thus, then, in the prose- 

 cution of physical science, our powers of observation, mem- 

 ory, imagination, and inference, are all drawn upon. We 

 observe facts and store them up ; imagination broods upon 

 these memories, and by the aid of reason tries to discern 

 their interdependence. The theoretic principle flashes, or 

 slowly dawns upon the mind, and then the deductive fac- 

 ulty interposes to carry out the principle to its logical con- 

 sequences. A perfect theory gives dominion over natural 

 facts; and even an assumption which can only partially 



