EXCEPTIONAL PHENOMENA. 329 



nothing to do but calculate as exactly as possible the 

 amount of interference, and make allowance for it; the 

 apparent failure of the law under examination should 

 then disappear. But in the second case the results may 

 be much more important. A phenomenon which entirely 

 fails to be explained by any known laws may indicate the 

 interference of some wholly new series of natural forces. 

 The ancients could not help perceiving that the general 

 tendency of bodies downwards failed in the case of the 

 loadstone, nor would the doctrine of essential lightness 

 explain the exception, since the substance drawn upwards 

 by the loadstone is a heavy metal. We now see clearly 

 that there was no breach in the perfect generality of the 

 law of gravity, but that a new form of energy manifested 

 itself in a conspicuous form in the loadstone for the first 

 time. In this case the forces concerned, those of gravity 

 and electrical attraction, have never yet been brought 

 into correlation with each other. 



Other sciences show us that laws of nature, rigorously 

 true and exact, may often be developed by those who are 

 ignorant of far more complex phenomena involved in their 

 application. Newton's comprehension of geometrical 

 optics was sufficient to explain all the ordinary refractions 

 and reflections of light. The simple laws of the bending 

 of rays apply to all rays, whatever the character of the 

 undulations composing them. Newton suspected the 

 existence of other classes of phenomena when he spoke of 

 rays as having sides; but it remained for later experi- 

 mentalists to show that light is a transverse undulation, 

 like the bending of a rod or cord. 



Dalton's atomic theory is doubtless true of all chemical 

 compounds, and the essence of it is that the same com- 

 pound will always be found to contain the same elements 

 in certain definite proportions. Pure calcium carbonate 

 contains 48 parts by weight of oxygen to 40 of calcium, 



