December 29, 1911] 



SCIENCE 



899 



The standard of length itself, the 

 standard meter, is defined as the distance 

 between two lines on a metal bar ; and not- 

 withstanding all the care taken in its 

 manufacture and preservation, there is no 

 assurance that it is not undergoing a con- 

 stant slow change, doubtless very small, 

 but appreciable by the refinements of 

 modern metrological methods, if there were 

 any fundamental unchangeable stand- 

 ard with which it could be compared. The 

 earth's circumference was supposed to be 

 such a standard and the meter was origi- 

 nally defined as the millionth part of an 

 earth-quadrant; but the various measure- 

 ments of this quadrant varied so much that 

 the idea was abandoned. The attempt to 

 base the standard on the length of a 

 seconds-pendulum was no more successful. 



But we have now the means of compar- 

 ing the standard meter with the length of 

 a light wave (the standard meter contains 

 1,553,163 waves of the red radiation from 

 cadmium vapor) so that should the present 

 standard be lost or destroyed, or should it 

 vary in length in the course of years, its 

 original value can be recovered so ac- 

 curately that no microscope could detect 

 the difference. True it is that in the 

 course of millions of years the properties 

 of the atoms which emit these radiations 

 and the medium which propagates them 

 may change — but probably by that time the 

 human race will have lost interest in the 

 problem. 



The difficulties in the application of the 

 interferometer method of investigating the 

 problems of spectroscopy, it must be ad- 

 mitted, were so serious that it was highly 

 desirable that other instruments should be 

 devised in which these difficulties were 

 avoided. This need was supplied by the 

 "echelon," an instrument based on the 

 same principle as the diffraction grating, 

 but consisting of a pile of glass plates of 



exactly equal thickness and forming a kind 

 of stairs, whence its name. 



The grating acts by assembling light- 

 waves whose successive wave trains are re- 

 tarded by some small whole number of 

 waves (usually less than six, the distance 

 between the grating spaces being about six 

 light- waves), whereas this retardation in 

 the echelon is many thousand. 



But the resolving power depends on the 

 total retardation of the extreme rays, and 

 this may be made very large, either by hav- 

 ing an enormous number of elements with 

 small retardations — or by a comparatively 

 small number of elements with large re- 

 tardations. For example, an echelon of 

 thirty plates of glass one inch thick, each 

 producing a retardation of 25,000 waves, 

 would have a resolving power 750,000 — 

 about seven times that of the grating ; and 

 this high value has actually been realized 

 in practise. 



Simultaneously Perot and Fabry showed 

 that by the repeated reflections between 

 two silvered surfaces' a very high resolving 

 power is obtained, and a few years later 

 Lunimer devised the plate interferometer 

 which embodies practically the same idea. 



The resolving power of all of these newer 

 devices is clearly many times as great as 

 that of the grating — but all equally share 

 the objection which holds (but to a far less 

 extent) for the grating, that the different 

 succeeding spectra overlap. It is true that 

 this difficulty may be overcome (though 

 with some loss of simplicity and consider- 

 able loss of light) by employing auxiliary 

 prisms, gratings, echelons, etc., and in this 

 form all these modern instruments have 

 contributed results of far reaching im- 

 portance, and which would have been im- 

 possible with the older instruments. 



' Boulouch, 1893, had observed that Na rings 

 were doubled both by reflection (grazing inci- 

 dence) and transmission (normal incidence) witlj 

 a light silver film. 



