SAUNDERS. — ARC SPECTRA OF THE ALKALI METALS, 441 



The plate should be batlied in this for two minutes and washed for one ; 

 it may be used at once, even before drying. Such plates do not keep 

 for many days as a rule, though the writer has used some that were kept 

 under very favorable conditions for a couple of mouths and found them 

 still fairly good. 



The source of light for the present work was usually the carbon arc, 

 using 10 to 15 amperes, direct current. The writer is indebted to the 

 International Achesou Graphite Co. of Niagara Falls, N. Y., for some 

 unusually pure graphite rods with which all these spectra were taken. 

 This graphite by itself gave only half a dozen lines (mostly Ca ; no 

 iron) outside of the baud spectrum of carbou (which showed the " tails *' 

 beautifully), but when it was saturated with a salt solution, several lines 

 of titanium came out, evidently from the graphite. These were not 

 unwelcome, as they were always sharp, and, as their wave-lengths are 

 given in Rowland's table of solar lines, they made excellent standards of 

 measurement. The differences in wave-length between these lines in the 

 sun and in the arc are too small to be wo'rth considering in the present 

 set of measurements. Eye observations showed that the alkali metal 

 spectra were particularly well developed in the arc when the graphites 

 were well saturated with salt solution and were separated by only 2 or 

 3 mm. "With such a source the most successful photographs were taken ; 

 the well-known lines of the elements were then very much strengthened 

 and broadened and the newer lines made their appearance. Fairly good 

 photographs were, however, obtained with the arc longer, so that the 

 "flames" were fully formed, if the light were taken from near the 

 terminal. Exposures of three hours' duration were taken in the effort to 

 pick up new lines in the deepest red. If the source of light could have 

 been maintained in its most efficient condition during the whole time, the 

 results might have been more complete, but the task of keeping the image 

 of such lively flames as form these arcs constantly on the slit proved 

 impossible. 



A few photogr:iphs were also taken of the spark spectra of some of 

 these elements, both with and without self-induction in the spark circuit, 

 but no differences were detected in the relative intensities of any of the 

 lines of the spectra of the spark or arc. 



The photographs were usually taken with half the slit exposed directly 

 to the light, and the other half covered witii colored glass. In the re- 

 sulting spectrum, lines in the first and second orders could ensily be 

 picked out where these overlap, as tlie ultra-violet lines were half the 

 length of the others. lu the deep red photographs, half the slit was 



