SCIENCE 



[N. S. Vol. XXXIV. No. 887 



make precise measurements of its amount, 

 requires a higher resolving power than was 

 furnished by the most powerful gratings 

 then in existence. 



As a final illustration, let us consider the 

 structure of the spectral "lines" them- 

 selves. Rowland's exquisite maps had 

 shown many of these which were then 

 thought simple, to be double, triple or 

 multiple, and there are clear indications 

 that even the simpler lines showed differ- 

 ences in width, in sharpness and in sym- 

 metry. But the general problem of the 

 distribution of light within spectral lines 

 had scarcely been touched. Here also the 

 total "width" of the line is of the order of 

 one one-hundredth of the distance between 

 the sodium lines and it is evident that 

 without more powerful appliances further 

 progress in this direction was hopeless. 



Enough has been said to show clearly 

 that these modern problems were such as to 

 tax to the utmost the powers of the best 

 spectroscopes and the experimental skill of 

 the most experienced investigators. 



Some twenty years ago a method was de- 

 vised which, though somewhat laborious 

 and indirect, gave promise of furnishing a 

 method of attack for all these problems, far 

 more powerful than that of the diffraction 

 grating. 



Essentially, the extremely simple appa- 

 ratus which is called the interferometer 

 consists of two plane glass plates. These 

 can be made accurately parallel and their 

 distance apart can be varied at will. 

 When light is reflected from the surfaces 

 which face each other, the two reflected 

 beams of light waves "interfere" in such a 

 way as to add to each other, giving bright 

 maxima, or to annul each other's effect, 

 producing dark spaces between. 



The alternations of light and darkness 

 which occur when the eye observes in the 

 direction of the normal are very marked 



so long as the plates are very near together 

 — but as this distance increases, the inter- 

 ferences become less and less distinct until 

 at a distance which depends on the char- 

 acter of the incident light they vanish com- 

 pletely. A perfectly definite relation 

 holds between the "visibility curve" and 

 the character of the radiation so that the 

 one can be deduced from the other. 



Now the "resolving power" of such an 

 apparatus is measured by the number of 

 light waves in the doubled distance be- 

 tween the surfaces. This is about 100,00Q 

 for a distance of one inch ; but the distance 

 is in fact unlimited and as the instrument 

 itself is practically free from errors of any 

 sort, its resolving power is practically 

 unlimited. 



The use of this method of light wave 

 analysis is attended with certain difScul- 

 ties, and the results obtained are not 

 always free from uncertainties ; but in view 

 of the fact that at this time no other 

 methods of this power had been devised, it 

 has amply proved its usefulness. Among 

 the results achieved by it may be men- 

 tioned: the resolution of many lines sup- 

 posed single into doublets, quadruplets, 

 etc. ; the measurement of their distances 

 apart ; the distribution of light in the com- 

 ponents; the measurement of their width 

 and the changes produced in them by tem- 

 perature, pressure, and presence of a mag- 

 netic field. 



Among the radiations thus examined one 

 proved to be so nearly homogeneous that 

 over two hundred thousand interference 

 bands could still be observed. Otherwise 

 expressed, the exact number of light waves 

 in a given distance, say ten centimeters, 

 could always be determined ; and by a com- 

 parison with the standard meter the abso- 

 lute wave-length of this radiation could be 

 measured and made to serve as a basis for 

 all wave-lengths. 



