Wave-length of Light. 259 



measurements of Muller and Kempf, and Kaulbaum had 

 been referred. From this comparison was derived the rela- 

 tion : — 



S a ,-l-68/i(±0-15/.)=^A . 



On returning to Baltimore, the first step was to rede- 

 termine the length of S a 2 . A series of comparisons was there- 

 fore instituted between it and the steel yard and metre A 4 , 

 the relation of which to K 2 was accurately known, A 4 having 

 been traced and determined by Prof. Eogers, and furnished 

 by him to the Johns Hopkins University. Only half of this 

 standard is subdivided to decimetres; but a series of compari- 

 sons with the various pairs of decimetres gave the relation 



S a 3 + l-3/*=£Ao. 



This result taken together with the relations found between 

 S a ! and S a 2 made it tolerably clear that a change had taken 

 place in the speculum-metal standards; and to obtain a 

 further confirmation, Prof. Rogers kindly consented to give 

 them a rigid examination and again compare them with 

 all attainable accuracy to Pi 2 . Hi s results for S a 2 were as 

 follows : — 



DimS^ = Dm 2 S a 2 + 1-70/*, 

 S^+l-O/^i-Ao. 



There is no escape from the conclusion, therefore, that the 

 speculum-metal bars S : and S 2 have changed both in absolute 

 length and the relative lengths of their parts. Here are two 

 bars of the same shape, mass, material, and constant of ex- 

 pansion. Each had the relation between its halves deter- 

 mined in the early part of 1885. Two years later these 

 relations are found to have changed by at least 1/ju; and an inde- 

 pendent determination by the original observer confirms this 

 result in the most unequivocal way. Further, the original 

 observer recompares one of these standards with the standard 

 from which it was originally determined, and finds a change 

 of \p. 



It should be borne in mind that with the comparator used 

 by me in this work, l,u is completely outside of any possible 

 errors of observation. The microscope used was especially 

 made for micrometric work, and has a power of two hundred 

 and fifty diameters, while one division of the micrometer 

 equals 0"28yu,. The average error of a single comparison 

 between two decimetres is rarely greater than 0'1/a, while the 

 temperature of the observing vault can be kept for several 

 days constant within o- 5 C, and during a day's observations 



