COMPOSITION OF THE SUN RUSSELL 209 



Here new diiRculties beset us. We may ask, what is the total 

 amount of iron vapor in the sun's atmosphere per square mile (or 

 square centimeter) of its surface ? But the sun has no definite surface. 

 At the top, of course, its atmosphere thins out gradually into space, 

 like the earth's. But at the bottom it is not limited by a solid surface, 

 or even by a cloud layer, but becomes gradually more and more hazy, 

 so that we can not see down very far. This increasing opacity is 

 due to the presence of free electrons and ions in the gas, as was first 

 shown by my colleague Stewart, of Princeton, and later worked 

 out in more detail by Milne, of Oxford. Milne's calculations indi- 

 cate that the change from extremely low density to practical opacity 

 takes place in a layer only about 20 miles thick, which explains the 

 sharpness of the sun's limb, even as seen with the largest telescopes. 



By far the greatest part of the absorption which produces the 

 Fraunhofer lines takes place in this thin reversing layer. In the 

 upper part of it the atoms are few, but get in their full effect, while 

 the more numerous atoms lower down are " blanketed " to an increas- 

 ing degree by the general opacity of the layers above them. Milne 

 has shown that the net effect of this partly obscured absorption is 

 very nearly the same as we would get if the atmosphere were perfect- 

 ly clear — except for the specific line absorption — down to a certain 

 depth and quite opaque below this, and that we may introduce this 

 imaginary surface at a level in the actual atmosphere such that the 

 gas above it blocks one-third of the escaping light in the regions 

 between the spectral lines. 



The amount of material above this fictitious surface is surprisingly 

 small. This was first pointed out by Lockyer, who showed by 

 direct comparison that a Bunsen flame an inch or so thick, charged 

 with a very small proportion of sodium vapor by a bit of salt held on 

 a platinum wire, absorbed the ultimate lines of the metal more 

 strongly than the whole thickness of the sun's atmosphere. This 

 very important conclusion was almost forgotten for 40 years, and 

 revived only in our own day, with so powerful a support from 

 atomic physics as to be irresistible. 



Stewart and Unsold have shown theoretically that the width of 

 a dark spectral line, produced in a rarefied gas, should be propor- 

 tional to the square root of the number of atoms per unit area in 

 the absorbing layer (no matter what its thickness) which are active 

 in producing this particular line. The strongest solar lines are 

 wide enough to make measures of their contours practicable — though 

 not easy — and the corresponding numbers of atoms per unit area 

 " above the photosphere " can thus be deduced. For the strongest 

 lines of all (the H and K lines of Ca-f, taken together) this comes 

 out 2 X 10 ^° atoms per square centimeter, which equals the number 

 of molecules in a layer of ordinary air one-third of an inch thick. 



