of tha Solar Spectrum. 249 



in the undalai' length, then the whole length H S haij to be 

 divided into fifteen equal parts, as is done in fig. 4. The corre- 



Fiff. 4. 



spending divisions of the refraction- and deflection-spectra are 

 accordingly of unequal width. The same quantity of rays is 

 distributed in the one over a greater or less space than in the 

 other ; whence for corresponding places of the two spectra the 

 heating must appear unequal. 



The space from the violet to the boundary of the blue and 

 green, between the verticals O'OOOl' and 0*0005, occupies a 

 width of 22 millims. in the prismatic spectrum (fig. 3, p. 248), 

 and only a width of 6 millims. in the grid-spectrum. Hence in 

 the grid-spectrum the violet, indigo, and blue rays are compressed 

 into ^ of the space which they occupy in the prismatic spec- 

 trum. At the place in question, therefore, the intensity of the 

 heat must be -y- times as great in the diffraction-spectrum as in 

 the refraction-spectrum (fig. 3), and accordingly the ordinate in 

 the middle, between the verticals 0"0004 and 0"0005 in fig. 4, 

 is made -— times as great as the height of the ordinate in the 

 middle between 0-OOOi and 0-0005 in fig. 3. 



The division between the verticals O'OOOo and 0"0006 (con- 

 taining chiefly green and yellow) is 12 millims. broad in fig. 3, 

 and G millims. in fig. 4, that is, only half as broad in the latter. 

 As the ordinate in the middle between 0*0005 and 0*0006 for 

 fig. 3 is 16 millims. in height, the corresponding ordinate iu 

 fig. 4 must be twice as high, that is, 32 millims., if it is propor- 

 tional to the heat which occurs at this point. 



Proceeding in the same manner, points have been determined 

 through which the curve of intensity (fig. 4) has been drawn. 

 This has been done by making the middle of each of the divi- 

 sions in fig. 4 as much higher or lower than the corresponding 

 ordinate in fig. 3, as tlie breadth of the division in fig. 4 is less 

 or greater than the breadth of the corresponding division in 

 ■fig. 3. 



Although the curve (fig. 4) cannot lay claim to great accu- 

 Phil. Mar/. S. 4. Vol. 17. No. 114. April 1859. S 



