SAUNDERS. — ARC SPECTRA OF THE ALKALI METALS. 4-13 



of this class of error will be easily seen. The writer's own estimates of 

 error are uot based on variations in his microscope readings ; if they 

 were, they would be half or a third as large. He has tried to fix the 

 error at such a value that the chances are extremely small that the 

 measured wave-length will differ from the true one by more than 

 the amount given. These estimates have been formed by the help of 

 test measurements taken on accurately known hues, following the same 

 method as with unknown ones. 



It is usual to give an estimate of the intensity of each line along with 

 its wave-length. This has uot been done in the following tables, as such 

 estimates have usually, especially for tlie greater wave-lengths, depended 

 more on the sen.sitiveness of the photographic plate for each vibration 

 than on the real intensity in the source of light. The lines of a series, 

 of course, decrease in intensity with decrease in wave-length, those of 

 the principal series more rapidly than the others. The first subordinate 

 series is stronger than the second, and the new series lines are the faintest 

 of all. Quantitative measurements of the real intensities of spectrum 

 lines are much to be desired, but the writer does liot know of any that 

 are applicable to these spectra. 



Lithium. 



The writer was not aware until after this work was done that Koiien 

 and Hagenbach had already found the lines at 6240 and 4148, which 

 form a new series (with 4636) in the lithium spectrum ; he gives his 

 values for the wave-lengths of these and a few other lines in the hope 

 that they may be of value, especially as some of them dilfl-r from the 

 values alreatly given by considerable amounts. The line at 4148, such 

 as it is, is visible in Figure 6, Plate 1, of Kayser's Handhuch, Vol. II, 

 immediately to the right of 4132.44 (which by a misprint is numbered 

 4273 in the figure). 



The " line " at 4602 deserves especial mention. Kayser * regards 

 this as a line heavily reversed and much broadened toward the red. 

 Hagenbach regards it as a pair of lines, the weaker one being constantly 

 reversed, the heavier occasionally; the separation of the two amounts to 

 over a tenth-meter. He is, apparently, ready to believe that all the lines 

 in this spectrum are pairs with so great a separation. It is easy to obtain 

 photographs of the strong line at 6708, and others, in which the lines are 



* Haudbuch, 2, 366 ; also Plate II., Figure 5. 



