TABLE OF STANDARD WAVE-LENGTHS OF THE SPECTRAL LIXES 553 



to a degree of accuracy hitherto unknown, and thus of solving the great 

 problem of the mathematical distribution of these lines. 



But for the comparison of spectra, as measured by different observers, 

 some absolute scale is needed. Hitherto Angstrom has been used. 

 But it is now very well known that his standard measure was wrong. 

 As his relative measures are also very wrong, I have concluded that the 

 time has come to change not only the relative measures, but the abso- 

 lute also. To this end Dr. Louis Bell worked in my laboratory for 

 several years with the best apparatus of modern science, using two 

 glass and two speculum metal gratings, ruled on two dividing engines 

 with four varieties of spacing, three of which were incommensurable 

 or nearly so, with two spectrometers of entirely different form, with a 

 variety of standard bars compared in this country and in Europe, and 

 with a special comparator made for the measure of gratings. His result 

 agrees very well with the next best determination, that of Mr. C. S. 

 Peirce of the U. S. Coast Survey. His final result agrees within 1 in 

 50,000 with his preliminary value. 7 This most recent value, combined 

 with those of Peirce, Miiller and Kempf, Kurlbaum and Angstrom, I 

 have adopted to reduce my final results to, although the calculations are 

 made according to Bell's preliminary value. See Appendix A. 



But it rests with scientific men at large to adopt some absolute 

 standard. The absolute standard is, of course, not so important as the 

 relative, and possibly the average of Angstrom might be adopted. But 

 for myself I do not believe in continuing an error of this sort indefi- 

 nitely. All the results obtained before the concave grating came into 

 use were so imperfect, that they must be replaced by others very soon. 

 With a good concave grating one man in a few years could obtain the 

 wave-lengths of the elements with far greater accuracy than now 

 known. 



As an aid to this work, I have constructed the table of wave-lengths 

 given in this paper, which have already been adopted by the British 

 Association and by the most noted writers of Germany and other 

 countries, and sincerely hope that it will aid in the work of making 

 the wave-length of a spectrum line a definite quantity within a few 

 hundredths of a division of Angstrom. 



ABSOLUTE WAVE-LENGTH OF D 



The following is an estimate of the absolute wave-length of the D line 

 from the best determinations. First, I shall recalculate the portion of 



1 American Journal of Science, 1887. 



