258 Professors Liming and Dewar, On the circumstances [Oct. 30, 



The reversals thus observed may be classed under several 

 heads. 



I. Reversals by the expansion of the line observed or as they 

 may be called self-reversals. These are the reversals most gene- 

 rally known. The sodium lines are frequently so reversed, indeed 

 always when a sufficiently volatile salt of sodium such as the car- 

 bonate or chloride is held in a gas flame. If sufficient sodium is 

 in the flame the lines are widened, and the less dense and less hot 

 sodium vapour outside the flame produces a narrow absorption line 

 down the middle of the bright yellow band. In this way the 

 metal itself gives the background against which the reversed line 

 is seen ; and in order to produce the result the line must be more 

 expanded at the higher temperature than at the lower. That does 

 not always happen, but it is so common that many reversals may 

 be seen in this way. The magnesium line of wave length 2852 has 

 great power of expansion and is always seen reversed in this way 

 when the arc is taken in a magnesia crucible as in Plate I. fig. 1. 



One set of the photographs exhibited are original negatives of 

 the arc spectrum shewing reversals of this class in the case of lines 

 of magnesium, thallium, indium, tin, aluminium, antimony, lead 

 and zinc. They are only a selection out of a very large number 

 taken by us but sufficient to shew the characters of this class of 

 reversals. 



Professor Hartley has lately (Proc. Roy. Soc. xxxiv. 84) called 

 attention to the pseudo-reversals of this class which may be pro- 

 duced in the case of a strong line by over-exposure. It is well 

 known that over exposure (solarization as we used to call it for- 

 merly) produces such an alteration in the sensitive preparation of 

 the photographic plate that the over-exposed parts cease to be de- 



velopable, so that a very strong line may appear white in the 

 negative where it ought to be black, but with a dark border, 

 and so give the appearance of a reversed line. Professor Hartley 

 finds it difficult to distinguish real reversals of the class we are 

 now discussing from these pseudo-reversals. His difficulty has 

 not occurred to us, first because we have always been in the habit 

 of taking our photographs in series with varying exposure, in 

 order to get impressions both of the feeble lines in some and of 

 strong lines in others ; and secondly because we almost always 

 close part of the slit of the spectroscope with a shutter so that the 



