of Refraction of the Metals. 9 



Discussion and Control-experiments. 



I must refrain from discussing here at length all the pos- 

 sible sources of error of the methods of observation ; I only 

 mention some objections which might be made. It may be 

 asked whether the refracting angle can be as correctly deter- 

 mined by observations on reflexion in our small and very 

 thin prisms as with thicker prisms with larger faces. 



It is known that if the thickness of a metallic film on glass 

 increases continuously from zero, the change of phase upon 

 reflexion at first changes with increasing thickness of metal, 

 and only becomes constant after a certain thickness has been 

 attained*. 



If the prisms used were so thin near the edge that reflexion 

 at different parts of the prism would be accompanied by dif- 

 ferent change of phase, then the whole plane wave reflected 

 from the entire face of the prism might thus assume a direction 

 differing from that required by the ordinary law of reflexion. 

 In reply to this it is to be remarked that all the prisms used 

 were so thick that the change of phase was the same at all 

 parts. I have, however, employed two other methods of 

 proving that the refracting angles were 'measured correctly. 

 Dr. Wiener has, in fact, determined the angle of one of the 

 prisms used above by his method f. 



The double prism of silver No. 5, which unfortunately 

 was one of those with faces not perfectly plane, gave by ob- 

 servations on reflexion a sum of angles of a mean value of 

 14*38 divisions, corresponding to 27*98 seconds of arc. Dr. 

 Wiener determined the sum of angles of the prism by his 

 method at about 30". Converted into iodide of silver — the 

 angle was determined by reflexion to be 67*59 divisions, 

 or lSl^'SO. Wiener's measurement by the method of in- 

 terference gave 128". The differences lie well within the 

 limits of errors of observation. A film of silver increases 

 upon conversion into iodide to about four times its thickness ; 

 consequently the angle of the silver-iodide prism must be four 

 times that of the original silver prism. 



From the value 67*59 divisions the angle of the silver was 

 calculated to be 16*89, whereas 14*38 was observed directly. 

 The difference is indeed somewhat large, but if it is remem- 

 bered that this prism had faces not perfectly plane, the dif- 

 ference still lies within the limits of error. 



* Compare Wiener. 



t AVied. Ann. lid. xxxi. pp. 629-G72. 



