422 MR WILLIAM SWAN ON THE PRISMATIC SPECTRA OF THE 



of the readings then gave double the angle of the prism. For, if DGI, FHK re- 

 present the course of the reflected rays, since the telescope has been adjusted to 

 sidereal focus, GI and HK, must be parallel ; and the angle DEF, will obviously 

 be double the angle BAC. Now, since DE and EF, are at the times of ob- 

 servation successively in the same direction, namely, that of the parallel rays emerg- 

 ing from the collimator, it follows that the telescope must have been turned 

 through the angle DEF. Hence the difference of the readings is DEF, or twice BAC. 



In order to test the adjustment of the collimator to sidereal focus, I made two 

 series of observations of the angle of the prism given in Table I. ; Series I. having 

 been made by means of the collimator, and Series II. on a definite point of the 

 tower of St Stephen's Church, distant about 2240 feet, where the parallax due to 

 any difference in the directions of the rays incident on the two faces of the prism 

 could not have caused an error exceeding 4" in the measured angle. 



These results agree so closely as to show that any want of parallelism in the 

 rays emerging from the collimator, arising from want of perfect adjustment to 

 sidereal focus, could not have appreciably affected the observations of the absolute 

 deviations of the refracted rays. I may also observe, that since, during the ob- 

 servations of the carbohydrogen and solar spectra, the whole apparatus remained 

 unaltered, any want of parallelism in the rays incident on the prism, whether aris- 

 ing accidentally from imperfect adjustment of the collimator, or necessarily from 

 the unavoidable want of perfect achromatism in its lens, — for either cause might 

 modify the apparent direction of the observed object, if the pencil of rays incident 

 on the prism were not accurately central, — would affect the observed deviations in 

 the two spectra alike. The accuracy of the observations, viewed merely as afford- 

 ing a comparative view of the relative positions in the scale of refrangibility occu- 

 pied by the lines in the two spectra, would thus remain entirely unimpaired. 



I have ascertained, however, by actual experiment, that the observations of 

 absolute deviation cannot have been sensibly affected by any want of achromatism 

 in the lens of the collimator. Having caused the telescope wires to coincide ac- 

 curately with the image of the collimator slit, I illuminated the slit alternately 

 with the extreme red and the extreme violet rays of the solar spectrum formed 

 by a flint-glass prism. I then found that the image of the slit did not in the 

 slightest perceptible degree alter its apparent position ; so that, while the illumi- 

 nation was changed from red to violet light, the wires continued to bisect the slit 

 with perfect accuracy. 



As the spectrum of the Bunsen lamp is so faint that the telescope wires, when 

 projected on all but its brightest lines, are invisible, it became necessary to illu- 

 minate the wires ; but I speedily found that, from the feeble luminosity of the 

 spectrum, observations with an illuminated field were nearly impracticable, 

 and I was therefore obliged to observe with illuminated wires on a dark field. 

 The arrangement for illuminating the wires which I devised is so simple, and 



