184 Dr. Tycho E:son Auren on 



obtained for calculating *^y /c in the mean wave-lengths- 

 used by the author. The calculated values are given below r 



, AUn :— I. II. III. IV. 



K Al/Cu 



according to Hull and Rice . . 4*77 4'88 4*99 5*34 

 „ Barkla and White . 4'89 5' 52 6*05 5'85 

 „ Table VIII. . . . 5-17 5*40 5*61 5-76 



There is, as appears from the table, a fairly good agree- 

 ment between my values and those of Barkla and White, 

 whereas those of Hull and Rice are more deviating. The 

 reason why, especial \y for the lighter elements, there are so 

 great differences between the values of the absorption co- 

 efficients recorded by different observers may probably first 

 of all be ascribed to the experimental method. As for AJ, 

 the observers may have disregarded that a source of errors 

 lies in the fact that the metal plates existing on the market 

 are seldom perfectly pure. The impurities are always Fe 

 and Si, often Cu and also other elements. From a chemical- 

 technical point of view, common aluminium metal, according 

 to Haber, is regarded as satisfactorily pure if it does not 

 contain more than 1 per cent. Fe and On and no more 

 than 0*8 per cent. Si. Supposing that in the metal there 

 is 1 per cent. Fe, this would enhance absorption by 6 or 

 7 per cent., and it we also admit that there are other 

 impurities the error seems not unlikely to amount to as 

 much as 10 per cent, or more. When analysing the 

 metal used by me, it has been found to contain 021 per 

 cent. Fe, 0*29 per cent. Si, and only traces of Cu. Hence 

 the metal is comparatively pure. The error caused by the 

 impurities cannot exceed 2 or 3 per cent., and as this is 

 within the limits of experimental errors no correction has 

 been made in this respect. The circumstance pointed out 

 here as to the quality of the common aluminium metal 

 will suffice to show that it is not quite appropriate to be 

 a standard metal for absorption measures. Besides, Barkla 

 has pointed out that Al in regard to absorption with a 

 diminishing wave-length shows a slight, certainly, but very 

 unmistikable increase at \ = 0'37, which Barkla thinks must 

 depend on the metal sending out a characteristic so-called 

 J-radiation of that wave-length. Barkla, therefore, prefers 

 copper as a standard metal, and I have made use of it also 

 and found it to be quite appropriate for this purpose. The 

 difference observed by different authors in regard to Al 

 absorption may at least partly be explained hj difference 

 in quality of the experimental material. In a still higher 



