OF ARTS AND SCIENCES. 151 



If the index of refraction = n = 1.54, from Fresnel's formulae, we 

 have the intensity of the image about ^-J^ of the whole prismatic 

 image. 



If now the area of the curve be greatly larger than the width of the 

 thermopile slit X maximum ordinate, this superimposed image may be 

 a considerable cause of error. 



The measured spectrum extended over about 140 mm- , but the 

 ordinates were comparatively very small, excepting over about 50 mm - ; 

 so that the curve may be considered very nearly equivalent to a trian- 

 gle, with altitude h (maximum ordinate), base 50 mm- , and the area 



The width of the slit was about 5 mm -, and we have 5 h mm - to com- 

 pare with 25 h mm -, or a quotient of £ which is contained in the -gfo 

 T ^ times, i. e. there would be an error of about lef . 



This, however, is within the limits of probable error, and as the 

 image always fell near the maximum ordinate, it was never noticed in 

 the experiments. Besides these causes of error there was, at first, 

 trouble on account of diffuse radiation from the source, heating, and 

 consequent radiation from the prism, radiation from surrounding 

 objects, change of temperature of pile, variations of temperature of 

 the source due to air currents, heating of bodies near the source, 

 moisture of the air, and many disturbances of the galvanometer used 

 with the pile, due to change of sensitiveness of the needles and to 

 changes in the earth's magnetism. But these and other errors have 

 been avoided in the apparatus to be described. 



The source of radiation was a strip of heavy platinum foil, about 

 15 mm - long and l mm- wide, heated by a current of electricity. The ex- 

 periments on the radiation from platinum were made from this wire 

 direct, and those from the various other substances by coating the foil 

 with the substances to be experimented upon ; the heat generated 

 by the current in the foil being sufficient to raise the substance to 

 incandescence. The radiations from this wire were focused by means 

 of a rock-salt lens, and dispersed by a rock-salt prism. A delicate 

 thermopile was moved through the spectrum thus formed. 



On the long table in the centre of the plate was a wooden box 

 about seven feet long, and of the shape shown in the figure. This 

 was intended to shield the experiment from outside radiations, and, 

 besides being closed, was wrapped with woollen cloths during an 

 experiment. 



At S was placed the platinum wire, which was mounted between 

 two brass rods, W (Fig. 2). These rods were themselves mounted 



