November i, 1894J 



NA TURE 



ou, differs widely from the old. In the old, two ob- 

 servers at least are engaged : one, who notes that reading 

 of the micrometer or of the vernier, which fixes in angular 

 measure the exact part of the spectral region whence 

 (though nothing is visible) a thermo-electric disturbance 

 has proceeded; and another, who simultaneously notes 

 through how many divisions of the scale the spot of light 

 from the galvanometer mirror is deflected by the same 

 electric disturbance. The process may be compared to 

 a groping in the dark, and it was only by these means 

 that the considerable inflections of the energy curve 

 much below the region about l/x were then fixed by 

 the bolometer, by being gone over again and again, with 

 what seemed almost interminable repetition, and which 

 did in fact call for over a thousand galvanometer readings 

 to obtain the position and amount of each single inflection j 

 of the energy curve, with the degree of accuracy which i 

 was then obtained, and which was shown in the former 

 memoir. 



If it took two years to fix the position of twenty lines 

 Ijy this process, it would take two hundred years to fix 

 two thousand, supposing they existed, and it became 

 evident that if the bolometer continued to be the only 

 means available, new and more effective methods of 

 •ising it must be found. 



New Methods. 



.About ten years ago a plan was first studied, which 

 has ever since been maturing, by means of which this 

 work could be carried on, not only with far greater 

 rapidity, but with greater certainty, and by an automatic 

 process. The idea in its original simplicity is very easily 

 understood. 



In the old process, just described, the deflection of a 

 spot of light upon a scale was read by one observer, 

 while another read simultaneously the position in the 

 spectrum of the cold band, or line, which caused the 

 thermo-electric disturbance. 



Now, in imagination, let us take away both the 

 observer at the circle and the one at the galvanometer, 

 and, in the latter case, remove the scale also, and put in 

 its stead a photographically sensitive plate. As the 

 needle swings to the right or left, the spot of light will 

 trace upon the plate a black horizontal lire whose length 

 will show how far the needle moves, and how great the 

 heat is which originated the impulse. If this be all, 

 when (under an impulse originated by the movement of 

 the spectrum over the bolometer thread) the needle 

 swings a second time, it will go over the same place ; 

 but if the plate have a uniform vertical movement, pro- 

 portional to the horizontal movement of the spectrum, 

 the combination of the two motions of the needle and the 

 plate will write upon the latter a sinuous curve which will 

 be, in theory at least, the same as the curve formerly 

 deducible only with such pains from thousands of such 

 galvanometer readings. 



If we suppose that the movements of the invisible 

 spectrum over the bolometer thread are controlled by 

 clockwork so that this spectrum is caused to move 

 uniformly, and that three movements are, by accurate 

 mechanism, rendered absolutely synchronous with those 

 of the moving plate, it is clear that we shall be able to 

 readily deduce from the photographic curve traced on 

 the latter, not merely the amount of the heat, but each 

 particular position in the spectrum, of the thread of the 

 bolometer, which alone can correspond with any given 

 inflection of the curve. 



Thus simple is the theory ; but no one had better 

 occasion to know how difficult the practice would be, 

 than myself 



The researches by the old method, and the early 

 attempts to improve them, were interrupted by my 

 acceptance in 1SS7 of a position which implies the 

 administrative charge of different branches of the public 



NO. 1305, VOL. 51] 



scientific service and of duties largely incompatible with 

 original research. What time could be spared from these 

 was, however, partly employed in elaborating the plan 

 of investigation just referred to. An appropriation had 

 been asked of Government for the establishment, on a 

 modest scale, of an astro- physical observatory in 

 Washington, whose first work should be the investiga- 

 tion of the whole infra-red solar spectrum, by some 

 means which would open that great region to knowledge. 

 It had been asked of Government because it seemed that 

 such knowledge, if attained, might teach us facts about 

 the sun and the absorption of its rays by the terrestrial 

 atmosphere, which there was ground to hope would 

 ultimately lead to results of such importance as to justify 

 this national aid. 



These observations were resumed in 1890 on the new 

 system, with the aid of the Smithsonian Institution, 

 which provided larger and more efficient apparatus, 

 whose design embodied the results of nearly fifteen 

 years' study of these subjects. 



Pending the provision of a suitable observatory build- 

 ing, an inadequate and temporary one was erected in 

 the Smithsonian Park in Washington, to shelter the 

 apparatus, presently to be mentioned, with which it was 

 designed to commence work while making provision for 

 more permanent scientific quarters (which I may add are 

 still lacking). 



Apparatus. 



The Foucault siderostat, perhaps the most powerful 

 instrument of the kind existing, was originally made by 

 Sir Howard Grubb, of Dublin, from my indications ; but 

 its dispositions have since been considerably modified. 

 A beam from its 20-inch mirror is conveyed through 

 the slit of a horizontal coUimating telescope having 

 a rock-salt objective of nearly seventeen centimetres 

 aperture, and of ten metres focal length, to the prism 

 or grating. The prism is of rock salt of correspond- 

 ing dimensions, worked (by Brashear) with the pre- 

 cision of, and presenting all the external appearance of 

 one of flint glass. It is mounted on a massive spectro- 

 bolometer (as the instrument which supports the prism or 

 grating used in producing the spectrum is called). This 

 instrument includes a large azimuth circle, over the 

 centre of which the prism is placed, and it also carries the 

 bolometer, which registers the spectral heat. The focal 

 lengch of the image-forming lens, or mirror, is in this 

 instrument much greater than in the first one used, and 

 all parts of the apparatus are correspondingly increased 

 in size and stability. The most important and novel 

 feature is, however, the mechanical connection of the 

 large azimuthal circle carrying the prism, with a distant 

 photographic plate, susceptible of vertical motion, and 

 which latter takes the place of the scale formerly in front 

 of the remote galvanometer, both circle and plate being 

 moved by the same clockwork, which is of such steadi- 

 ness and precision as to make the two movements as far 

 as possible perfectly synchronous. 



To fix our ideas, let us suppose that the slow-moving 

 azimuthal circle carrying the prism revolves through one 

 minute of arc in one minute of time ; in which case the 

 spectrum will move horizontally across the vertical bolo- 

 meter thread at a proportional rate. Now, if the same 

 mechanism which causes this circular motion of the prism, 

 and of the spectrum, of one minute of arc in one minute 

 of time, causes the photographic plate to move vertically 

 before the galvanometer mirror at the rate of one centi- 

 metre of space in one minute of time ; if there be no 

 allowance to make for changes of temperature in the prism 

 or for like corrections, if the mechanician has done 

 his part in such perfection that everything works as it 

 should, it obviously follows that, under such conditions, 

 during every second of this minute a portion of the 

 spectrum represented by the small quantity of one second 

 of arc, will have glided before the bolometer thread, and 



