Wave-length of Light. 249 



before him, he used Robert's gratings ; and in spite of the fact 

 that, like all Robert's gratings, they gave very imperfect 

 definition and showed numerous " ghosts," his results were 

 more than usually consistent. But, in spite of all Angstrom's 

 care, the event has shown that his wave-lengths are in error 

 by as much as one part in seven or eight thousand, mainly 

 through an o error in the assumed values of his standards of 

 length. Angstrom measured his gratings by means of a 

 dividing-engine, the screw of which was very exactly deter- 

 mined by comparisons resting on the Upsala metre, which, in 

 turn, had been compared by M. Tresca with the prototype of 

 the Conservatoire des arts et metiers. Had this comparison 

 given the correct value of the Upsala metre, Angstrom's 

 wave-lengths would have been very nearly exact except for 

 corrections due to errors of ruling in the gratings. 



After Angstrom's research the question of absolute wave- 

 length was not seriously raised for ten years ; when Mr. C. 8. 

 Peirce, under the auspices of the United States Coast Survey, 

 again attacked the problem, armed with Rutherfurd gratings 

 far superior to those used in any previous research. No 

 official report of his very elaborate and exhaustive experiments 

 has ever been published, save a very brief preliminary report 

 in the American Journal of Science in 1879. Such of his 

 results as have been made in any way public will be discussed 

 in the experimental part of the present paper. 



Meanwhile Thalen, who so efficiently aided Angstrom in 

 his work, had taken up the part of it left uncompleted by the 

 latter' s death ; and in his paper, Sur le Spectre du Fer, 

 published at Upsala in 1885, discussed the corrections which 

 must be applied to Angstrom's values by reason of the error 

 in the Upsala metre. It seems that, through the experiments 

 of Professor Lindhagen, Angstrom became aware, as early as 

 1872, that the assumed value of his standard was considerably 

 too small. His death prevented his verification of M. Lind- 

 hagen's results, and nothing further was done till Thalen 

 took up the work. Tresca' s comparisons had obtained for 

 the length of the Upsala metre at 0° 999*81 millim. But 

 the very exact experiments of M. Lindhagen have shown the 

 above to be somewhat too small, and that the correct value 

 is 999'94. This difference makes, of course, a marked error 

 in the wave-lengths based on Tresca' s results. Applying the 

 appropriate correction, the wave-length of E, the line most 

 carefully determined by Angstrom, becomes 



5269-80, 

 instead of the original 



5269-12. 



