196 Dr. Tycho E:son Aur£n on 



From the table it appears that the values calculated by 

 means o£ the formula (9) undoubtedly agree much better 

 with the observed values than those obtained by means of 

 the formula (10). The formula (9), as Owen using the 

 values in my first paper also shows, approximately gives 

 the relation between the relative atomic absorption coefficient 

 and the atomic number of a great number of elements rather 

 satisfactorily, but we cannot pretend that it has proved to be 

 the proper expression of the real facts over the whole range 

 of the section examined. For the elements Li-F they will 

 thus turn out to be too low, for Ca-M all of the calculated 

 values are lower than the observed, and for Sr-Ag all 

 the calculated values are too high. On the whole, the 

 docker's formula (13) describes the relation of k ,q 



and Z in a better way than that just mentioned ; but here, 

 too, marked deviations occur, for instance at N, P, CI, K, 

 and for Ca-Ni as well as for Mo-Ag the same view will be 

 true as for the formula (9), though the deviations here 

 are considerably lets. 



Thus, although the atomic absorption coefficient in a 

 great; number of elements may be pronounced to be pro- 

 portional to a power of the atomic number, such a formula, 

 however, cannot render the real relation. From the above 

 experiments it is clear that in certain elements there are 



some very marked discontinuities in the values of lc tri , 



J a/Ou' 



which in the lighter elements are still more marked by 

 the change by sudden leaps of the absorption effected by 

 scattering (fig. 4). As shown by Table XT. and fig. 2, 

 the different elements may be arranged into groups within 

 which lc ,« according to formula (7) is expressed as a 



linear function of the atomic number. Great deviations 

 do not occur except at Mg and Si. In respect to the latter 

 element, the probable cause of the deviation has already 

 been discussed. As to the former element, the deviation 

 presumably depends on the fact that in the examined salts 

 absorption for about 80 per cent, is determined by the 

 negative radicle, and that therefore the errors in deter- 

 mining the absorption of these radicles will proportionally 

 too much influence the calculation of ^ Mff / Cu - 



IY. Outer Electrons and Scattered Radiation. 



It was shown previously that the part of the relative 

 absorption coefficient which was thought to be due to 

 the outer electrons is in a high degree dependent on the 



