8S 



tempt to carry into practice the method of counting the alterna- 

 tions of hght and darkness in the interference method, is the de- 

 fect in homogeneity of the light employed. This causes indis- 

 tinctness of the interference rings when the distance is greater 

 than a few centimeters. The Hght emitted by various kinds of 

 gases and metallic vapors, when made luminous by the electric 

 discharge, differs enormously in this respect. A systematic search 

 showed that among some forty, or more, radiations nearly all were 

 defective, some being represented by a spectrum of broad hazy 

 "lines," others being double, triple, or even more highly com- 

 plex. But the red light emitted by luminous vapor of metallic 

 cadmium was found to be almost ideally adapted for the purpose. 

 Accordingly this was employed; and the results of three inde- 

 pendent measurements, made by different observers and at dif- 

 ferent tim.es, of the number of light-waves of red cadmium light 

 in the standard meter are as follows : 

 I- I553392-4- 

 n. 1553393-2. 

 ni. 1 5533934- 



It will be seen that the diff'erences are less than half a mil- 

 lionth part, and this is about the limit of accuracy of the com- 

 parative measurements of the material standards. Within the 

 last year a similar determination has been carried out by Perot 

 and Fabry, widi a result not to be distinguished from the above. 

 It fellows that we now have a natural standard of length — the 

 length of a light-wave of incandescent cadmium vapor — by means 

 of which a material standard can be realized, whose length can 

 not be distinguished from the actual standard meter, — so that 

 if, through accident or in time, the actual standard meter should 

 alter, or if it were lost or destroyed, it could be replaced so ac- 

 curately that the difference could not be observed. 



In the search for a radiation sufficiently homogeneous for 



