Arc and Spark Spectra of Nickel under Pressure, 393 



errors of measurements, though taken in conjunction with 

 other evidence it is significant. We might go further and 

 attribute the good general agreement which obtains between 

 the arc and spark displacements of unreversed lines to the 

 equality in the number of reversals in the two spectra, which 

 indicates that there is a measure of similarity in their density 

 gradients, while the discordant behaviour of reversed lines 

 may be referred to differences in their temperature gradients, 

 since the temperature conditions determine at which layers 

 absorption shall begin and end. An experimental investi- 

 gation of this point might prove profitable. 



We may approach the matter from another standpoint : — 

 I have pointed out elsewhere* that the presence of layers 

 of varying density and pressure above the solar photosphere 

 produces a composite spectrum line, and prevents the pres- 

 sure upon the deepest layers of the sun's atmosphere from 

 being estimated; and the same argument may be urged in 

 the case of the arc and spark, whose spectrum lines, whether 

 reversed or bright, are the summation of the effects of the 

 several layers in the source of light which differ in density, 

 instead of in pressure and density as in the case of the >un; 

 since part at least of the displacement is due to the amount 

 of material presentf, the resultant displacement must be 



N is the normal undisplaced position of the line. 



influenced to a corresponding extent by the density gradient 

 of the source of light. For example, we may regard the 

 intensity curves a, b, c as due to successive layers of equal 



* Dufh'eld, "Meteorology of the Sun," Royal Meteorological Society, 

 f Duflield, Phil. Trans. Roy. Soc. A. ccxv. p. 205 (1915). 



