244 MR. J. C. MAXWELL GARNETT 



and (8), when the quantities n and UK have been determined for the light in question 

 by direct experiment on the metal in the standard amorphous state. R. S. MINOR* 

 has made such experimental determinations, for various kinds of monochromatic light, 

 from the polished surfaces of silver and copper. His values of n and n/c for silver 

 and copper, together with the numerical values of a, ft, a!, ft', and of certain other 

 functions as calculated for various values of v, are shown in Tables II. and III. The 

 values of n and n/c, for X = '630 and X = '589 only, have also been determined by 

 DRUDEf from the polished surface of gold ; but values of n and HK for other values of 

 X have been obtained by HAGEN and RUBENS J from gold prisms deposited on glass. 

 Since however the state of the metal in the prisms is not known, these latter values 

 cannot be depended upon for our purpose ; but as a rough estimate of the values of 

 a, ft, a', ft', &c., may be formed by their means, the numerical values of these quantities 

 have been calculated ; the results are shown, together with all the observed values of 

 n and UK, in Table I. The wave-length X of the light will throughout be measured 

 in thousandths of a millimetre. 



All the calculated numbers given in Tables I., II., and III. have been carefully 

 checked with a " Brunsviga " machine. I believe that in no case does an error 

 amounting to 1 per cent, survive in these Tables, which must accordingly supersede 

 those -given in the former communication. || 



* R. S. MINOR, 'Ann. der Phys.,' vol. X., 1903. 



t 'P. DRUDE, 'Physik. Zeitschr.,' January, 1900. 



\ RUBENS, ' Wied. Ann.,' vol. XXXVIL, 1889. 



It will appear in the sequel that on this account the optical properties of silver glasses and of colloidal 

 solutions of silver are much more accurately represented by our calculations than is the case with gold 

 ruby glasses and colloidal solutions of gold. 



|| Lot. tit., p. 396. 



