( 350 ) 



If e. g. the opening in the diaphragm has the shape 

 of tig. 4, then the spectrum s is obtained. When the 

 density gradient is diminished the figure shrinks, ?; if 

 now the density gradient is made to change its sign 

 and to increase, the spectrum proceeds through the 

 stages a (gradient exactly zero) and ^j to 0. 

 Fig. 4. The relation between the shape of the opening 



in the diaj^hragm and that of the bright spots in the spectrum might 

 easily liave been foretold from the shape of the 

 dispersion curve. Having, however, experimentally 

 found the relation " between the two figures for a 

 simple case as the one above, it is not difficult to 

 design for any desired distribution of light the shape 

 of the required opening in the diaphragm. The 

 Fig. 5. flower I and its inversion n required the diaphragm, 



represented in fig. 5. By reversing the gradient the image i passes into x. 

 So in this way one may also arbitrarily produce duplications, 

 reversals, bright or dark ramifications of spectral lines and it would 

 e. g. be possible faithfully to reproduce all phenomena observed in 

 this respect in the spectra of sun-spots, faculae or prominences. On 

 Plate II a number of arbitrary distributions of light have been 

 collected. They w^ere all produced in sodium vapour of 390^ on the 

 average; «' is again the spectrum with equal temperatures of the 

 tubes A and B. In v on the dark dispersion band D^ a bright 

 double line is seen, reminding us of the spectrum of the calcium 

 flocculi of Hale. In the same negative D^ also shows a line double 

 line, which however is no longer visible in the reproduction. The 

 spectra <p, /, if' imitate the origin of a sun spot and prominence 

 spectrum ; tp namely represents the spectrum of the quiet solar limb 

 with radially placed slit ; in x ^ prominence appears and a spot with 

 phenomena of reversal ; if' shows all this in a stronger degree. If 

 now the density gradient is made to change sign, the image first 

 shrinks again lo (p after which it expands to w, in a certain sense 

 the inversion of if'. The remarkable aspect of these gradual changes, 

 admitting of perfect regulation, is only imperfectly rendered by the 

 photographs. 



The 1'elation between the curvature of the rays a7id 

 the density gradient. 



The (|uestion arises whether it is prohahle thai circumstances as 

 were realised in our experiments are also met with in nature, or in 



