320 THE PRINCIPLES OF SCIENCE. [chap. 



likely to possess. Nearly the same considerations apply 

 to the atomic weight as the standard of mass. It is im- 

 possible to prove that all atoms of tlie same substance are 

 of equal mass, and some physicists think that they differ, so 

 that the fixity of combining proportions may be due only 

 to the approximate constancy of the mean of countless 

 millions of discrepant weights. But in any case the de- 

 tection of difference is probably beyond our powers. In a 

 theoretical point of view, then, the magnitudes suggested 

 by Professor Maxwell seem to be the most fixed ones of 

 which we have any knowledge^ so that they necessarily 

 become the natural units. 



In a practical poiut of view, as Professor Maxwell would 

 be the first to point out, they are of little or no value, be- 

 cause in the present state of science we cannot measure a 

 vibration or weigh an atom with any approach to the 

 accuracy which is attainaWe in the comparison of standard 

 metres and kilograms. The velocity of light is not known 

 probably witliin a thousandth part, and as we progress in 

 the knowledge of light, so we shall progress in the acru- 

 rate fixation of other standai'ds. All that caa be said tht n, 

 is that it is very desirable to determine the wave-lengths 

 and periods of the principal lines of the solar spectrum, 

 and the absolute atomic weiLjhts of the elements, with all 

 attainable accuracy, in terms of our existing standards. 

 The numbers thus obtained would admit of the reproduc- 

 tion of our standards in some future age of the world lo a 

 corresponding degree of accuracy, were there need of such 

 reference ; but so far as we can see at present, there is no 

 considerable probability that this mode of reproduction 

 would ever be the best mode. 



Subsidiary Units. 



Having once established the standard units of time, 

 space, and density or mass, we might employ them for the 

 expression of aU quantities of such nature. But it is often 

 convenient in particular branches of science to use nml- 

 tiples or submultiples of the original units, for the ex- 

 pression of quantities in a simple manner. We use the 

 mile rather than the yard when treating of the magnitude 

 of the globe, and the mean distance of the earth and 



