168 L. H. Martin: 



The values W^a = 208.60 X 10-n cm., W^a' = 213.41 X lO-Ucm. 

 are those of W. Diiane and W. Stenstromc, which lead to a value for 

 the erbium K absorption edge. 



AEi-K = 215.9 X 10-iicm. 

 PI. 2. Oct. 24. X = 216.09 X units. Weight 1. 



PL 4. Nov. 16 A = 215.85 X units. Weight 2. 



PI. 5. Nov. 21. X = 215.87 X units. Weight 2. 



Weighted mean A = 215.9 X lO-n cms. 

 Term value ('/R units) = 4222. 



Discussion. 



In the determination of the wave lengths of such penetrating 

 rays as the above, we are beset with the difficulty, that, owing to crystal 

 penetration the effective plane of reflection lies below the surface of 

 the crystal. If Q is calculated from the geometrical properties of the 

 apparatus, and the distance of " edge " from direct ray impression, 

 an error is introduced amounting to nearly 3% in short wave length 

 determinations. Errors of this magnitude are observed in the K ab- 

 sorption edge determinations of de Broglie' for elements Hg to U. 



In this method, which might be called a method of coincidences, 

 the reference rays and the erbium K absorption edge have sensibly the 

 same wave length, and the result is free from this objection. 



Determinations have been made by Duane and Blakes, using an 

 ionisation chamber method in which was read directly from an 

 accurately divided circle, and by Siegbahn and Jonssons, using a 

 photographic method, both of which are free from the above source 

 of error. Siegbahn and Jonsson used their crystal as a transmission 

 grating, a method first devised by Rutherfordio for his determinations 

 of the wave length of the penetrating y rays of RaC, the rays being 

 constrained by slits to meet the plane of reflection passing through 

 the axis of rotation. 



Although both methods are free from objection, the values obtained 

 by Sieghahn and Jonsson are systematically smaller than those ob- 

 tained by Blake and Duane for the absorption edge wave lengths of 

 the K series. 



Extrapolating the two series of values for the value expected for 

 erbium we find — 



Siegbahn and Jonsson A = 215 x lO-i^ cm. 

 Blake and Duane a = 216 x lO-ii cm. 



The value found here 215.9 X lO-^i cm. agrees better with the 

 results of Blake and Duane. Duane has suggested that the discrepancy- 

 may be found in that Siegbahn probably measured from the point at 

 which darkening began, instead of the point corresponding to the centre 

 of the slit. Here that settings were made for those rays which cor- 

 respond to the centre of the slit. 



I wish to thank Professor Laby for his valuable advice and in- 

 terest during the execution of this work. I am also very much indebted 



