198 Mr. S. P. Langley's Experimental Determination of 



d=(<f)\ representing the required relation to any degree of 

 exactness. 



In the invisible spectrum the difficulties are immensely 

 greater, and demand special means, not only on account of 

 this invisibility, but owing to the absorption by the prism 

 and to its compressing the rays. 



The prism here used was made by Adam Hilger, of London, 

 and its optical properties pre in every way satisfactory. It is 

 of a white flint, which has proved singularly transparent to 

 the longest solar waves. Its principal constants have been 

 given in this Journal, vol. xv. p. 173. 



Apparatus for measuring Obscure Wave-Lengths. 



In 1882 an apparatus was employed in which invisible 

 rays, after passing through the Hilger prism, at a known 

 deviation, fell on a Rutherfurd reflecting grating (either of 

 681 lines to the millimetre, or half that number), from which 

 the diffracted invisible ray fell on the bolometer, at a mea- 

 sured angle with the grating. By the use of the known for- 

 mula (ns\= sin i + sin r) connecting the angle of diffraction 

 with the wave-length, the wave-length was then found. 



Several determinations were thus made of wave-lengths in 

 the upper part of the infra-red, where the heat is relatively 

 great; but, though the definition of the Rutherfurd grating 

 was admirable, it was not large enough to supply sufficient 

 heat to enable measures in the lower infra-red to be made 

 with confidence. 



In May 1882 I had the good fortune to secure one of the 

 very large concave gratings, then newly constructed by 

 Professor Rowland, and which he was kind enough to make 

 for me of a very short focus, so as to give a specially hot 

 spectrum. After many essays, during which a great number 

 of mechanical and optical arrangements for getting rid of the 

 superposed spectra were tried with unsatisfactory results, it 

 became clear that, for this large and concave grating, it was 

 necessary to let the ray fall first on it, and then on the prism, 

 thus making the wave-length the known and the deviation 

 the unknown quantity. 



In the use of this form of grating, the slit is placed in the 

 circumference of a circle whose diameter is equal to the 

 radius of curvature of the grating, and which touches its 

 surface. The spectra are then formed, without the need of 

 collimator, observing telescope, or any further apparatus, all 

 lying upon the circumference of the circle which contains the 

 slit. The grating which was employed contains 18,050 lines, 

 142 to the millimetre, ruled on the surface of a concave 



