WAVE-LENGTHS IN THE INVISIBLE PEISMATIC SPECTRUM. _ 159 



(0.91.) Iiicouspicuous; possibly ii part of Dkai'ER's fi (0.8!) and 0.01 from part of a group 

 called Tt by Auney). 



(0.94 to ().9S.) Very heavy band ; marks the extreme limit of Draper's iuvestigatious, according 

 to his own statement; possibly identifiable with a gap in Lamansky's curve, and corresponding 

 to the group called "pfr" by Abney. (Allegheny ob.servations make it probably telluric.) 



(1.13 to 1.18.) Still colder than preceding; possibly ideiitifiably with a gap on Lamansky's 

 curve, and with Abnky's " </^". (Allegheny observations make it probably of telluric origin.) 



(1.27.) ln(;onsi)icu(>ns line. 



(1.3G to 1.4G.) \'ery remarkable band. Almost absolutely cold and black. So broad and diffuse 



that it is diflicult to mark its limits, but coldest part seems to have a wave-length of 1.36 to 1.37. 



(Allegheny observations make it i)robably of telluric origin.) Possibly f of Abney's chart, and 



identifiable with the last gap of Lamansky's curve. It seems to be the " ultima Thtde^' of previous 



iuvestigatious. 



newly-discovered lines and cold bands.. 



(1.52 and 1.59.) Inconspicuous lines. 



(1.81 to 1.87.) Great Cold Band, first discovered on Mount Whitney. Probably of telluric 

 origiu. It is not the furthest line, but is here called H ou account of its being the last compicuous 

 break iu the energy curve. 



"(1.98 and 2.04.) Small but definite lines. The last discovered by the bolometer. But the 

 observable solar spectrum certainly extends to a wave-length of over 2(^.70. 



DISTRIBUTION OF ENERGY' IN THE NORMAL SPECTRUM. 



The curve d = qA giveu iu Eig. 3 enables us to mark off a wave-length scale upon the map of 

 the prismatic s|)ectrum, without any extrapolation, between our present ])oints of observation, a 

 deviation of 50° 58' (corresponding to A =:0//.344), and a deviation of 44^ 25' (corresponding to 

 A = 2//. 350), and also to construct a map in which -the wave-length scale is an ordinary scale of 

 equal parts, but iu which the degrees of deviation, if represented, would be uuequally spaced. 

 Such a chart of the nonnal .spectrum has, as we have already remarked, the advantage of being 

 entirely independent of any particular prism or grating, and consequently of being directly com- 

 parable with all other maps of the same kind. 



If, besides making a map of the normal spectrum, we wish to construct a curve representing the 

 corresponding distribution of energy, a further consideration of the relations existing between the 

 two charts is necessary. The law of dispersion of the prism causes the distribution of energy in 

 its spectrum to be quite different from what would have been observed with a diffraction grating.* 

 Disregarding the absorbing' action of the apparatus, the amount of heat between two definite wave- 

 lengths, as between the A and B lines, should be the same iu both spectra, provided the total 

 (piantity of heat is the same iu both. The area between any two ordinates of the curve may be 

 considered to represent the amount of heat iu the part of the spectrum included between them, and 

 the total area of the curve represents the total amount of heat. If, then, we suppose the area of 

 the normal curve required, to be the same as that of the prismatic one, the condition to be ful. 

 tilled by the former curve is that the area included between the ordinates at any two wave-lengths 

 shall be equal to that included between the same wave-lengths in the latter, and from this condi- 

 tion we can deduce a rule for effecting the required transformation, t 



Lay off upon a line, AB (Fig. 3), any couveuient distance, and divide it into equal spaces to 



* J. W. Dkapeu, Phil. Mag., vol. xliv, p. 104, 1872. 



t See J. Muller, Pogg. Aiin.aleu, vol. cv ; Luudqiiist, Pogg. Auiialeu, vol. civ, p. 140 ; Moutou, Comptes Keudus, 

 vol. Ixxxix, p. 206. 



