1887.] On Radiation from Dull and Bright Surfaces. 435 



The filament C gives confirmatory results, but unfortunately it 

 seems to have broken down early in the condition CCO. I find, 

 however, — 



Table IV. 









c. 



CC. 



CCC. 



To pass from — 

 4 candles to 10 candles ■ 

 4 „ 20 „ 



" increase of con- ' 

 ductance is 



>■ from 1 to 



1-047 

 1-081 



1*028 

 1-041 



1-043 



Comparing these numbers we are led to the result that, if we admit 

 the assumptions I have made, the temperature to which the carbon 

 must be raised in order that it may give out light of a definite candle- 

 power is higher when the surface is in the dull condition than when 

 it is in a brilliant metallic-looking state. 



This result was to me so unexpected that I proceeded to test it 

 directly by the following experiments : — Two glass tubes, similar in 

 every respect, were constructed, containing two precisely similar 

 platinum wires cut from the same hank, which had been specially 

 drawn for me some months before by Messrs. Matthey and Johnson. 

 One of the wires was in its natural bright condition, while the other 

 was covered with the thinnest possible coating of lampblack, which 

 was put on by passing the wire quickly and steadily through the 

 flame of a paraffin lamp. The construction of these tubes is shown 

 in fig. 1. The platinum wire db is kept stretched by two spiral 



Fig. 1. 



springs of copper, being silver-soldered to two extremities of these 

 springs. Two loops I at the other extremities of the spirals pass over 

 two pieces of glass rod gg, gg, which are passed in by side tubes, 

 blown on to the main glass tube ; and the spirals pull on the glass 

 rods. The ends of the side tubes are sealed up after the glass rods 

 are in their places, with the exception of one, which is used for con- 

 necting to the Sprengel pump, and is finally sealed when a complete 



