690 THE NEW SPECTRUM. 



part of the spectrum in which it occurred. For rea.sons into which he 

 does not enter, this old plan was tedious in the extreme and required, 

 as has been said, hundreds of observations to fix with appropriate 

 accuracy the position in wave-length of one invisible line. It has 

 been stated that onl}' about twenty such lines had been mapped out 

 in neai'ly two years of assiduous work prior to 1881, and if a thousand 

 such lines existed, it was apparent that fifty years would Ix' rcipiired 

 to denote them. 



The writer then devised a second apparatus to he used in connection 

 with the })()l()meter. This apparatus was simple in theory, thouoh it has 

 taken a dozen years to make it work well in practice, but it is working 

 at last, and with this the maps in this volume of the "Annals*' and that 

 before us have been chietiy made. It is almost entirely automatic, and 

 as it is now used, a thousand inflections can be delineated in a single 

 hour, nuich l)etter than this could have been done in the half cen- 

 tury of work just I'eferred to. 



Briefly, the method is this: A great rock-salt pi'ism (for a glass one 

 would not transmit these lower rays nor could they easily })e detected 

 in the overlapping spectra of the grating) is ol)tained of such purity 

 and accuracy of figure, and so well sheltered from moisture, that its 

 clearness and its indications compare favorably, even in the visil)le 

 spectrum, with those of the most perfect prism of glass, with the 

 additional advantage that it is permeable to the extreme infra-red rays 

 in ((uestion. This prism rests on a large azimuth circle turned by 

 clockwork of the extremest precision, which causes the spectrum 

 to move slowly along, and in one minute of time, for example, 

 to move exactly on(> miiuite of arc of its length ))efore the 

 strip of the bolometer. l)ringing this successively in contact with 

 one iinisihic line and another. Since what is blackness to the 

 eye is cold to the bolometer, the contact of the black lines chills 

 the strip and increases the electric current. The bolometer is con- 

 nected by a cable with the galvanometer, whose consequent swing to 

 the right or the left is photographically registered on a plate which 

 the same clockwork causes to move synchronously and uniformly up 

 or down by exactl}^ one centimeter of space for the corresponding- 

 minute. B}' this means the energy curve of an invisible region, which 

 directly is wholly inaccessible to photography, is photographed upon 

 the plate. 



Let it be noted that whatever the relation of the movement of the 

 spectrum to that of the plate is, (and ditt'erent ones might be adopted), 

 it is absolutely sjnichronous — at least to such a degree that an error in 

 the position of one of these invisible lines can be determined, as has 

 been stated, with the order of precision of the astronomical measure- 

 ment of visible things. 



The results were before them in the energy curves and the linear 

 infra-red spectrum containing over seven hundred invisible lines. This 



