358 THE CHEMISTRY OF THE SUN. [CHAP. 



these higher strata are gradually added, the spectrum must end 

 by becoming very rich indeed in lines of gradually increasing 

 faintness and length. 



The point here raised is so important that it will be well 

 to dwell on some details. If the short lines seen at first are 

 limited to a low stratum no reduction of the illumination will 

 make them seem higher than they did at first. And speaking 

 generally, all the lines existing in the lower stratum alone 

 should be seen at first, because the molecules which give rise 

 to them are at the same transcendental temperature. The 

 difference, therefore, between the short lines seen at first 

 and the longer dim lines gradually revealed each degree of 

 unveiling being accompanied by the appearance of a set of 

 longer lines will become more marked as totality is ap- 

 proached. We cannot make these observations by the ordinary 

 method so well as in an eclipse, because the lines are brightest 

 when the conditions are most disturbed, and when we cannot 

 tell, for reasons I shall show afterwards, whether we are dealing 

 with ascending or descending currents, or both combined ; 

 but during the approach to. totality, and naturally, therefore, 

 the recession after it, we can watch the phenomena on an 

 undisturbed part of the sun through a long range of varying 

 illumination. 



This is the place to point out that the new hypothesis has 

 to explain everything by appealing to changes of tempera- 

 ture alone, and that if such a consideration of temperature 

 fails to accord it a power of prediction, the hypothesis 

 must fall to the ground. Others who have written on this 

 subject have suggested that -the various phenomena which they 

 cannot explain may be dependent upon variations of pressure 

 and density, about which at the sun at present we know 

 nothing or next to nothing, or upon other more obscure 

 couditions which are not even defined. It is in this sense that 

 I have said that success here is crucial for the hypothesis. 



