408 



SCIENCE. 



[N. S. Vol. XIV. No. 350. 



these Mines' — that is, to the accurate deter- 

 mination of the details of the infra-red energy 

 curve. It is noticeable that most of the effort 

 has been put upon the measurement of the 

 abscissae of the curve {%. e., wave-lengths), the 

 ordinates being considered for the present of 

 secondary importance ; and no determination 

 of the radiant energy in absolute measure has 

 been attempted. 



The problem can be considered as made up 

 of tvro distinct parts : first, the production of 

 a suflBciently intense, pure and well-defined 

 solar spectrum by means which permit of the 

 direct or indirect measurement of wave lengths ; 

 second, the detection of absorption lines in a 

 region of this spectrum quite beyond the range 

 of visual or even special photographic processes, 

 and in a manner which will result in the precise 

 location of these lines in the spectrum and in 

 the approximate determination of their intensity 

 and character. As a means of producing the 

 spectrum the grating, on account of loss of 

 energy, was necessarily discarded in favor of the 

 prism ; of glass for the shorter wave lengths, of 

 rock salt for thelonger. The observatory possess- 

 es undoubtedly the finest specimens known of 

 the latter substance ; one in particular, from the 

 salt mines of Eussia, in the form of a 60° prism 

 with faces 13 cm. wide and 19 cm. high, having 

 been used in the present research. The spec- 

 troscope as finally used was of the extremely 

 convenient fixed-arm type described by Wads- 

 worth {Phil. Mag., '94) combined with a very 

 ingenious collimating system of cylindric mir- 

 rors, due to Mr. Abbot, which enabled a slit of 

 suflBciently small angular width to be used, 

 while at the same time the linear width could 

 be so great (over 0.2 mm.) that the loss of 

 energy by diffraction at the slit was avoided. 

 The refinement of the apparatus is shown by 

 the fact that in the region of wave-lengths 

 A<2.5/i, a slit image of angular width 1^^. 5 could 

 be used with a bolometer of angular width 1^^.2, 

 while for longer wave-length the values 4^^.5 

 or 9^^ and 3^^. 4 respectively were necessary to 

 obtain proper deflections. The Rayleigh form- 

 ula for visual spectroscopic resolving power in- 

 dicates that with the prism used some 1,300 

 ■ lines might be detected in the region between 

 ^ = .76," and ^ = 5. 3^. It is here shown that by 



the use of instruments of detection other than 

 the eye the resolving power might be different 

 and possibly greater than that given by Ray- 

 leigh' s formula ; and apparently this limit has 

 been exceeded in practice. With the prism 

 spectroscope, in order that the wave-length 

 corresponding to a particular position in the 

 spectrum may be known, the angle of the prism 

 and its dispersion curve must be known ; the 

 former was determined with great care from time 

 to time during the progress of this work, and a 

 full discussion is given of the methods used, and 

 of the precautions necessary when — as seems 

 usually to be the case with rock-salt — the prism 

 faces are curved ; while the determination of the 

 dispersion curve formed the principal subsidiary 

 research. 



As regards the method of detecting the lines, 

 the instrument chosen was, naturally, the bo- 

 lometer which for such purposes, where a linear 

 form is essential, possesses decided advantages 

 over any other form of radiation measurer yet 

 devised. Instead of the usual visual observa- 

 tion of the motion of the spot of light reflected 

 from the galvanometer mirror, produced by 

 changes in the radiant energy falling on the ex- 

 posed bolometer strip, a method of photographic 

 registration was used. The spot of light fell 

 upon a vertical photographic plate which was 

 given vertical motion so connected mechanically 

 with the motion of the spectrometer that each 

 position of the plate corresponded to a definite 

 position of the bolometer in the spectrum. The 

 spot of light therefore occupied at any instant 

 a position on the plate such that its one coordi- 

 nate was proportional to the radiation striking 

 the bolometer, and its other to the position of 

 the bolometer in the spectrum ; it therefore 

 traced out the energy curve above referred to, 

 except that the abscissae were proportional not 

 to the wave-length, but to the deviation pro- 

 duced by the particular prism used. In the 

 form of fixed-arm spectroscope here used, the 

 only moving part is the central table carrying 

 the prism and a pane mirror suitably placed ; 

 if the table is given a uniform rotation, the 

 angular motion of the spectrum past the fixed 

 bolometer strip will be uniform also, and twice 

 as rapid, and the particular wave-length in 

 focus on the bolometer at any instant will have 



