808 SUMMAKY OF CQIIRENT EESEARCHES RELATING TO 



an inverted image of it, in tte erect image of tlie flame, independently 

 of any spectra. 



Now with regard to the spectra. I well remember that the first 

 experiment I performed when the diffraction theory was new was to 

 receive the images on a piece of oiled tissue-paper at the objective back. 

 If ray memory serves me right, you can trace an image of P. angulatum 

 about half an inch from the objective back. The images will necessarily 

 be much out of focus, but, nevertheless, they can be made out. There 

 were black outlines on a light ground in the dioptric beam, and a green 

 image in each of the sis spectra. Eemove the greased paper screen 

 further back from the back lens, and the six spectral images were seen 

 to coalesce with the central dioptric image. The point to be learned 

 from an examination of the back of the objective is the size of the cone, 

 or cones, which form the image at the objective conjugate. 



Thus, the dioptric image of a point in the object is formed by a 

 cone, the base being the bright disc at the objective back. A spectral 

 image is formed by a cone, the spectral disc being its base, and so on. 

 I am of opinion that Prof. Abbe has established experimentally and 

 theoretically that the delineation of this microscopic image of the fine 

 structure of P. angulatum depends on the fusion of these green spectral 

 images with the dioptric beam and with one another. 



(2) The next point is the extinction of the spectra by the dioptric 

 beam, or, more correctly, the effect of the spectra is so feeble in com- 

 parison to that of the dioptric beam, that their power to influence the 

 image is practically nil. 



The answer to this is, that just as much as you increase the diameter 

 of the dioptric beam, so do you increase that of the spectra — a fact which 

 may be experimentally verified in two minutes. Thus, expand the 

 illuminating cone until it nearly touches the expanded spectra, now 

 stop out the dioptric beam, and look at the brightness of the spectral 

 image. Then, without moving the stop, reduce the illuminating cone, 

 and watch the diminution in the brightness of the spectral image. 



Of course, it is impossible to carry on the experiment when the 

 dioptric beam overlaps the spectra, as it is impossible to cut out the 

 dioptric beam without cutting out the spectra as well. But it is for the 

 " dioptricians " to show why the brightness of the spectral image should 

 cease to increase at the point when the dioptric beam overlaps the spectra. 

 The brightness of the spectral image most certainly increases as you 

 increase the dioptric beam as far as you can carry on the experiment, 

 and I can see no possible reason why it should not go on increasing 

 until you reach your maximum cone.* 



(3) The following experiment, although not proving the matter, 

 points very strongly in favour of the diffraction and against the dioptric 

 theory. Examine a P. angulatum with a lens which, when illuminated 

 by a narrow pencil, will not grasp the six first-order sj)ectra, and enlarge 

 the cone until the dioptric beam occuj)ies, say, 3/4 of the back. Now, 

 if a lens of suitable angle has been chosen, the expanded spectra will just 

 cut into the peripheral zone of the objective. If the eye-piece is replaced, 

 delineation will be seen ; but if a stop be placed over that peripheral 

 zone, although the large dioptric beam remains the same, the delineation 

 will have vanished. If the image is a dioptric one, why, in the presence 



* Unless tlie experiment has been tried, one would hardly believe the great 

 brilliance of the spectral image when the dioptric beam has been stopped out. 



