October 20, 1911] 



SCIENCE 



533 



same Southern Mills spectrograph but with \ 4,340 

 Central and Fe-spark eomparison spectra. These 

 tables are preliminary to the more extensive in- 

 vestigation which has been made possible by the 

 generous loan from observatories in different parts 

 of the earth of suitable sets of spectrograms. 



The relative accuracy attained, though naturally 

 not quite equal to that of the International sec- 

 ondary and Kayser's tertiary standards, appar- 

 ently approximates these in each set of measures 

 separately for a large number of the lines which 

 are best suited for accurate measurement. Differ- 

 ences of a systematic nature were encountered 

 which might easily have escaped notice if the 

 work had been based upon only one instead of 

 three practically different sets of spectrograms. 



A large number of lines has been found, in 

 addition to those already published, whose wave- 

 lengths vary progressively from type to type. 



Inasmuch as the accuracy attainable in the rela- 

 tive wave-lengths of lines in stellar spectra, with 

 a good modern three-prism spectrograph, is not 

 much inferior to the accuracy of the available 

 standards of wave-length, it is important for the 

 future progress in this work that results show 

 clearly on what line or group of lines any newly 

 determined wave-lengths are based. Designations 

 like "On the Rowland system" are not sufficient. 

 Results should be published in such form as to be 

 readily convertible to any standard system. This 

 should be done with due regard to the increasingly 

 necessary precaution to condense published results 

 as much as possible without impairing their use- 

 fulness. 



Wave-lengths of the Silicon Lines \ i,B52.7, 

 X 4,567.9 and X 4,674.9 in Stellar Spectra and 

 m laboratory Spectra: S. Albrecht. (Read 

 by E. J. McDiarmid.) 



Among the lines whose wave-lengths were de- 

 determined as part of the extensive program re- 

 ferred to in my preceding note are the three 

 silicon lines X 4,552, X 4,567 and X 4,574, which 

 were first identified with silicon, in stellar spectra, 

 by Lunt. It is not within the scope of this paper 

 to go into much detail. The principal object, 

 besides giving the wave-lengths in stellar spectra, 

 is to call attention to the need of more extensive 

 study of these lines in the laboratory. This wiU 

 be evident from the tables given below. In table 

 I. are given the wave-lengths of these lines as 

 determined from stellar spectra. For purposes of 

 comparison with the wave-lengths obtained from 

 the Lick spectrograms, I have computed the wave- 



lengths from the measures of Frost and Adams, 

 published in the Publications of the Yerkes Ob- 

 servatory, Vol. 11. Table II. gives a summary 

 of the determ in ations of the wave-lengths in stars 

 and in the laboratory. 



Lick spectrograms. 

 Mean.... 



45.52.752 



.798 



4552.759 



4552.770 



.717 



4552.765 



Lick spectrograms. ..!.#567.970 

 Yerkes spectrograms. .(f567.9e5 



Lick spectrograms... U57.^.9i6 

 Yerkes spectrogramsl.<!i57.ii.9S0 



Yerkes spectrograms 

 Mean 



Determined from 



B toB2 

 B3 to B5 

 B toB5 



B toB3 42 

 B5toA 5 



B toA 47 



B toB5 



B toA 



B toB3 



B toA 



Wave-length in Stars 



.^5^.765 4552.79,4552.6 

 .#•567.967 4567.90 4567.5 

 .#57.^.975 4574.68 4574.5 



Laboratory Values 



With High Dis- With Low Dis- 

 persion persion 



Exner & Frost & Lock- 

 Haschek Brown yer 



4552.75 4559.. 64 '^^^'i:^ ^ahl-lh 

 4567.95 4567.90 45R8.0j4567.82 

 4574.9 \4574.79 4o74.9\'i574.86 



The values of the wave-lengths in stars, table I., 

 as determined from the Lick spectrograms and 

 from the Yerkes spectrograms are in very close 

 agreement. The early determinations by GUI and 

 by McClean, which are added in columns 2 and 3 

 for the sake of completeness, were, I believe, 

 determined with low dispersion. The laboratory 

 values, table II., by Exner and Haschek (Aph. 

 Jour., 12, 49, 1900) and by Frost and Brown 

 (Hid., 22, 159, 1905) were both determined with 

 high dispersion. The wave-lengths by Exner and 

 Haschek are slightly smaller, 0.016 A, on the av- 

 erage, while the values of Frost and Brown are 

 0.122 I, 0.067 A and 0.128 1 smaller, respectively, 

 for the three lines than the wave-lengths in stars. 

 The two laboratory determinations differ by 

 0.11 A 0.05 A and 0.11 a, respectively, for the 

 three lines, which is equal to an average sys- 

 tematic difference of 0.090 1. Measures by Frost 

 of another plate, taken by Mr. Fidcher, of the 



