366 Transactions of the Society, 



light was projected on to a large right-angled internal reflexion 

 prism. By this the light was directed vertically upwards through 

 the instrument. The mercury lamp was capable of being run for 

 long periods without attention, but a special switch was provided 

 by which the intensity of the light from the lamp could be, for a 

 short time, very greatly increased. 



The parallel beam of light from the lamp, having suffered 

 internal reflexion by the glass prism, was converged by means of 

 a condensing lens on to a small aperture A, which formed part of 

 the collimator ; but the light in so doing had to traverse the slits 

 by which the directions of the parallel pencils incident to the 

 grating were ultimately controlled. The number and positions of 

 the slits depended on the type of measurement being made, and 

 will therefore be considered later. The collimator was similar in 

 principle to that used in the ordinary spectroscope ; but a more 

 complex lens-system was found to be necessary because, in order 

 to avoid reflexions in the superstage lens-system, the light had to 

 be limited to a portion of one diatom valve. To do this the image 

 of the small aperture A was projected by means of an immersion 

 condenser by Conrady, and supplementary lens, on to the diatom. 

 The supplementary lens was used to shorten and at the same time 

 increase the aperture of the substage lens-system ; it had the 

 further property of allowing a more perfect correction of the 

 spherical aberration for the chosen colour to be obtained for the 

 particular cone of the condenser employed. Now for this system 

 to ^ project beams of parallel light on to the diatom, it was found 

 that the slits would have to occupy a position within the triple 

 back lens of the Conrady condenser. The difficulty was, how- 

 ever, avoided by placing the slits below the aperture A and 

 throwing their image instead into the correct position by means of 

 a lens placed immediately beneath the aperture. In this way the 

 beams of light incident on the diatom were rendered strictly 

 parallel. 



The reason for the selection of the diatom Pleurosigma angu- 

 latnm for the present research lay in the regularity of its markings. 

 For in order to obtain a record of the positions of the diffracted 

 pencils set up by each individual portion of the diatom valve, it 

 would be essential to be able to examine very small areas. But 

 this is impossible, for two reasons : — 



1. The intensity of the beam transmitted would be too small to 

 affect the eye. 



2. The diffracted wavelets from several adjacent structural 

 elements must co-operate for well-defined spectra to be formed. 



But if no sudden changes take place in the character of the 

 markings, it would be possible by obtaining measurements from 

 comparatively large overlapping areas to calculate the effects of 

 individual elements by a process of differentiation. 



