xi.] SPECTRUM OF MAGNESIUM. 149 



We then carefully examined the disappearance of the b lines 

 and found that they had behaved exactly as they do on the sun. Of 

 the three lines the most refrangible was the shortest. In fact, 

 had the experiment been made in hydrogen instead of in air, the 

 phenomena presented in the telescope would have been almost 

 perfectly reproduced ; for each increase in the temperature of 

 the spark caused the magnesium vapour to extend further from 

 the pole, and increased the disparity in the relative lengths of 

 the lines. 



I have given the above as the first experiment which indi- 

 cated that this method of observation seemed likely to bring 

 us much new knowledge, and indeed it will not be difficult 

 to show that the results obtained by this method have a very 

 important bearing upon every question connected with solar 

 chemistry. When these spectra were observed the spectra 

 of the longs and shorts of course we had a perfectly new 

 set of phenomena to deal with. In all spectra observed 

 on the old method all the lines had been practically seen 

 of the same length, or else lengths shown in maps had 

 represented intensities. But now we soon found that we 

 had, in the case of each chemical substance, to deal with 

 the remarkable fact that when the spectrum of that chemical 

 substance was examined in this way, some of the lines 

 were long and some of them short ; and the question 

 naturally arose, How is it that some of the lines are long 

 and some of them short ? That question was an exceedingly 

 difficult one to answer. I do not know that it has been 

 thoroughly answered yet. More than this, we wanted to know 

 ,how the long and short lines behaved when the spectrum of any 

 particular chemical substance was seen in the chromosphere. 



We shall see that while researches were being made for the 

 answer to these questions, certain general statements soon became 

 possible which are of very considerable importance to us in our 

 inquiry. 



