442 PROCEEDINGS OF THE AMERICAN ACADEMY. 



open for a short time ami then covered with red glass, the other half 

 being so covered throughout. In this way, images of unknown red 

 lines would form \ydvt of the same spectrum witli known second order 

 lines: no shift could occur to alter their relative positions, as the 

 colored glasses were supported independently and could be changed 

 without affecting the position of the slit in the least. In all cases the 

 beam of light from the condensing (([uartz) lens, passing through the 

 slit, filled the mirror and grating completely. The condensing lens was 

 rigidly fixed throughout. The writer, for these reasons, feels certain 

 that the photographs obtained can be relied upon to show the true 

 positions of the lines. 



The measurement of the films was accomplished by means of a 

 Gaertner micrometer microscope with a run of 5 cm. graduated to read 

 to 0.005 mm. Its screw was investigated and found to possess no error 

 large enough to be worth considering. A magnif^'ing power of about 

 15 diameters was used. In measuring an unknown line, in every case 

 measurements were taken on several standard lines lying on both sides 

 of it, and its position was calculated from each of these; 10 settings 

 were made on each line. As a rule, the wave-length of any line, as 

 given, is the average of several such sets of measurements taken from 

 different photographs. 



In the following tables of the complete arc spectra of the alkali 

 elements, the writer has given in the first column the series to which 

 the line belongs (P for principal, I for first subordinate, &c.). In the 

 second column are placed various values for the wave-lengths, and, 

 opposite each, in the third and fourth columns, the error as estimated by 

 each observer and the observer's initial letter. The following are the 

 observers quoted : L, Lehmann * ; Ld, Lenard ; K & R, Kayser and 

 Runge ; K & H, Konen and Hagenbach ; H, Ilagenbach f ; E & H, 

 Exner and Haschek | ; L & D, Liveing and Dewar; B, Lecoq de 

 Boisbaudran, and S for the writer. The custom of stating errors seems 

 to vary with different observers. The writer believes that the errors of 

 measurement proper are usually small compared with errors due to 

 wrong interpretation of the photographic image. In his own experience, 

 several settings on a diffuse line may have agreed with one another to 

 less than O.l t.-m., while a different observer has made equally con- 

 cordant measurements leading to a result 0.2 or more away. AVhere so 

 many lines are broadened or diffuse, as in these spectra, the inq)ortance 



* Ann. der. Pliys 5, 608 (1901). t Ann. der Tliys., 9, 729 (I'JOli). 



t Wellcnlangen-TubcUen (1002). 



