200 



Prof. J. N. Lockyer. 



years; "unknown" means a line not so far traced by me in any 

 metal with, the exception of Cerium. This exception is necessitated 

 by the fact that the spectrum of that metal contains practically as 

 many lines as appear in the solar spectrum. The wave-length map 

 of Rowland's second series has been taken as a standard. 



In the present communication I confine myself to submitting pro- 

 visional curves based upon a preliminary inquiry into the number of 

 times the lines of both categories have been observed to be widened 

 in spots. Some slight corrections will, doubtless, be ultimately 

 required when some uncertainties connected with some of the earlier 

 observations, made before Rowland's maps were available, have been 

 cleared up. The highest points of the curves represent the maximum 

 frequency of iron lines in one case and of unknown lines in the other. 



The period embraced by the observations practically enables us to 

 study what has taken place at two successive sun-spot minima and 

 two maxima. Ifc will be seen that the phenomena which followed the 

 minimum of 1879 have been exactly reproduced after the minimum 

 of 1890. At the minima the iron lines are prominent among the 

 most widened lines'; at the maxima we only find lines about which 

 nothing is known. Since the discussion indicates that the iron lines 

 involved, which ultimately disappear, are almost invariably those 

 seen most prominent in the spark, the view put forward in my paper 

 of 1886 that the change observed is due to the dissociation of iron in 

 the spots as a sun-spot maximum is approached, is corroborated, and, 

 so far, I have heard of no other simple and sufficient explanation. 



It will be noted that the maxima and minima of solar temperature 

 thus revealed to us, if my hypothesis be confirmed, lag behind the 

 spot maxima and minima. This may explain the lag observed in 



