LYMAN. — FALSE SPECTRA FROM THE ROWLAND GRATING. 251 



Johns Hopkins in the year 1894. The mounting was the same as that 

 of grating No. 1. 



The character of the photographs obtained with grating No. 4 was 

 quite different from those obtained with grating No. 1. Lines A 1? C 1? 

 and A 4 . C 4 were present, and with about the same apparent wave length 

 as before. They were, however, very much less distinct than when ob- 

 tained with grating No. 1. The fluting effect was not very sharp ; but, 

 on the other hand, the space between positions 1878 /a and 924 /x was 

 filled with reproductions of groups A x and C l5 of such clearness as to be 

 cpiite unmistakable. The wave lengths of some of these phantom lines 

 was determined in the same manner as before. Groups corresponding in 

 appearance to the family marked A u A 2 , A 3 , etc., were found at 1970 /z, 

 1020 /x, 993 /x, and 97u^. The groups called C l5 C 2 , etc., were found at 

 1878^ and 924 1/. very cleanly, and at more than eight other positions 

 in a less distinct form between 1878 and 924. In fact, the spectrum in 

 this region is so tilled with groups of the A, C, type that the lines of one 

 family overlap lines of the other in many cases. 



Such, then, are the experimental data connected with this phenomenon. 

 Before coming to a statement of conclusions, however, the following facts 

 are of interest : — 



Some of the reproductions belonging to the set of lines A 2 and E 2 in 

 aluminum occur under the real spectrum, and may be observed if proper 

 care is used. 



Gratings of large radius do not show the false lines with as much clear- 

 ness as gratings of small radius. So feeble, indeed, are the false lines 

 obtained with the 21 ft. grating, that it is improbable that they would 

 be noticeable in ordinary spectrum work. With a six-foot grating, 

 however, the false lines are so strong that they might easily cause errors 

 in series spectrum investigations. We have seen, moreover, that the 

 false lines from grating No. 1 are much stronger than those from grating 

 No. 4, and we have reason to suspect that gratings exist which pro- 

 duce false lines even stronger in character than those obtained with 

 grating No. 1. Gratings 1 and 4 differ little in radius, and have exactly 

 the same number of lines to the inch, yet their false spectra are entirely 

 different in character. We cannot apparently predict the nature of the 

 false spectra of one grating from the false spectra produced by another 

 of the same ruling and radius. 



The conclusions arrived at in this paper may be stated as follows : — 



1. The first spectrum produced from the Rowland concave grating 

 is not pure, but is complicated not only by underlying spectra of 



