TRANSACTIONS OF SECTION A. 575 



and the experimeut which resulted has most satisfactorily confirmed the prediction 

 of theory. 



The light from the sun was reflected from a heliostat furnished with a 4-inch 

 optically flat mirror, worked by Sir Howard Grubb, F.R.S. The mirror is silvered 

 on the front, and may be relied on to furnish reflected light capable of forming a 

 good image. The reflected beam was received by a horizontal telescope furnished 

 with a 2-inch objective by Cook and an eyepiece by Watson. By this apparatus 

 an image of the sun was formed in a vertical plane at a distance of a little more 

 than a metre from the telescope, and of a size somewhat larger than the Rowland 

 grating. Whenever there happened to be minute spots on the sun at the time of 

 observation, the image was good enough to show them satisfactorily. 



The surface of the grating was made to coincide with this image, so that the 

 light reaching different parts of the grating came from diflferent parts of the sun. 

 At the same time, in consequence of the arrangements described above, all light 

 reached the grating from nearly ^ the same direction, viz., from the direction in 

 which the eye-stop of the telescope was seen from the grating. 



"WTieu the apparatus was set up in this way, the same full series of bright 

 impure spectra were produced as are seen when tlie portions of light reaching the 

 several reflecting strips come from identical sources. 



Still further to test the predicted result, a spectroscope slit was placed near 

 the telescope, in the position of the eye-stop of the telescope. This reduced the 

 light forming the image of the sun and impaired its definition, but still left the 

 image good enough to ensure that the light reaching reflecting strips of the 

 grating which are somewhat distant from one another came from different 

 parts of the sun. The spectrum of the second order on one side was then 

 viewed through the telescope of the spectrometer, when the Fraunhofer lines 

 were well seen in large numbers. The E group in the green was carefully 

 examined, and the definition was so good that all but one- of the 30 lines in 

 Rowland's great map were seen. The closest doubles that were observed to be 

 resolved were at 5265"8 in the E group, and the corona line with the iron line 

 adjoining it at 5316'9. The spacing of these doubles is about -J of an Angstrom 

 unit, which in that part of the spectrum would, according to Lord Rayleigh's 

 formula (X/SX = 2w), require a grating of 16,000 lines to resolve them in the 

 second spectrum if the grating and the adjustments were perfect. 



The performance as seen was regarded as good, considering the impossibility in 

 some respects, and the difficulty in others, of getting the adjustments more tliau 

 approximately made : 16,000 lines occupy 28 mm. on the grating, which is more 

 than an inch. It therefore extended over a considerable part of the image of the 

 sun which illuminated the grating. Moreover, having regard to the fact that 

 the brightness of the light reaching the different reflecting strips was not quite the 

 same, and to the other shortcomings mentioned above, it seems not unlikely that 

 the whole of the 26,000 reflecting strips of the grating were actually in operation 

 to produce such definition as was observed. If so, light was made use of from 

 parts of the image of the sun as far asunder as IJ inch. 



\_Note added October 1901. — The experiment is very much improved by 

 introducing a collimating lens between the slit and thegi'ating. The lens employed 

 is a lens of 73 cm. focus, and was set up at a distance of about 12 cm. in front of 

 the grating. It does not sensibly impair the image of the sun formed on the 

 grating, and it enables the adjustments to be fidbj made which had to be left 

 imperfect before. When the adjustments were carefully made the spectrum of 

 the sun in the second spectrum did not appreciably fall short in either definition, 



' The light reaches all parts of the grating from exactly, and not only nearly, the 

 same directions when the collimating lens described in Note above is added to the 

 apparatus. 



- The line not seen is the faint chromium line of wave length .5275'34 and of 

 intensity 00 on Rowland's scale. It is between two stronger lines, the nearer of 

 which is of intensity 1 and at a distance of about a fifteenth of an Angstrom unit. 

 This is too close for resolution by a grating of 26,000 lines in its second spectrum. 

 The pair are, however, widely separated by the grating that was used in its fifth 

 spectrum. 



