8 ABSORPTION SPECTRA OF SOLUTIONS. 



with time much more rapidly with yellow light than with blue light. The 

 action in the minimum also increases more rapidly with time than is the 

 case with the X 5050 minimum of the Wratten plate. With a full exposure 

 (1 minute) blackening of the Seed film begins to shade off at X 5900, but 

 may be seen as far as X 6200. 



The developer used throughout was a concentrated hydrochinone 

 solution made up according to Jewell's formula (Astrophys. Journ., 1900, 

 pp. 240-243). 



SOURCES OF LIGHT. 



The most satisfactory source for the region of the spectrum lying 

 between the extreme red and the beginning of the ultra-violet is the Nernst 

 lamp, as it is brilliant enough to bring the time exposure down to about a 

 minute, and is, of course, perfectly continuous and steady. In the ultra- 

 violet the spectrogram on Plate 1 shows that its action decreases rapidly 

 with the wave-length, ceasing practically at about X 3200. For this region, 

 then, some spark-spectrum must be used. The cadmium zinc spark used 

 by Jones and Uhler is fairly satisfactory, being especially strong in the 

 extreme ultra-violet, but it has the disadvantage of having a limited 

 number of very intense lines on a rather faint, continuous background. 

 It was hoped that some spark-spectrum could be found having a very large 

 number of lines, but without any lines of very great intensity. 



A reference to published tables of the spectra of the elements showed 

 that tungsten, molybdenum, and uranium all satisfied this requirement. 

 Each of these has so many lines, and these so closely packed, that with 

 an instrument of moderate dispersion the spectrum ought to be nearly 

 continuous. The problem was to make spark terminals of these sub- 

 stances which could be used satisfactorily. The metals not being easily 

 obtainable, the following plan was tried: Sheet carbon about 3 mm. thick 

 was cut into pieces about 1 cm. by 4 cm. and dipped into concentrated 

 solutions of ammonium molybdate, or uranium nitrate; a suitable solu- 

 tion of a tungstate was not tried. These pieces of carbon were then heated 

 to redness in a Bunsen flame and again quickly immersed in the solution, 

 the process being repeated two or three times. Some pieces were also 

 treated with solutions of salts of iron, copper, and cobalt, and others 

 were treated with two or more of the solutions in succession. The spectra 

 of these carbon terminals showed the lines of the metals with which they 

 were treated almost as well as terminals of the metals themselves; but in 

 the case of iron, copper, or cobalt the cyanogen bands were also present. 

 With molybdenum and uranium, however, the cyanogen bands were 

 absent, the merest trace of the X 3883 band appearing. The uranium 

 spectrum is almost continuous with the dispersion employed, but its 

 intensity falls off very rapidly from X 3000 towards the ultra-violet. The 

 molybdenum spectrum, although not so nearly continuous, is much richer 

 in ultra-violet, being quite strong as far as X 2300. 



Curiously enough, if a pair of carbon terminals is treated with molyb- 

 denum and also with one or more of the other metals, very little except 

 the molybdenum spectrum is seen. Uranium, however, seems to increase 

 the intensity of the continuous background somewhat, and hence the termi- 



