724 Transactions of the Society, 



dissimilar to the object itself). There is not the least rational ground 

 — nor any experimental proof — for the expectation that this mixture 

 should come nearer to a strictly correct projection of the object (be 

 less dissimilar to the latter) than that image which is projected by 

 means of a narrow axial illuminating pencil. This latter image has 

 the most favourable conditious in regard to similarity to the object, 

 because in its production nothing is lost of the diffraction-pencil but 

 the peripheral portions (-which in most cases are of relatively small 

 intensity). All the other images with which it is mixed in the 

 case of a wide-angled cone, are liable to greater dissimilarity, com- 

 pared to a strictly true image, because they depend upon a more 

 incomplete admission of the diffracted light. And it is against all 

 rules of reasoning to assume that a mixture, or superposition, of a 

 variety of images, all of which are more or less dissimilar to a true 

 projection, should be less dissimilar than that constituent which is 

 the least dissimilar one (whatever this one may be). 



This conclusion is in no way in contradiction to the fact, that in 

 many cases a broad illuminating pencil may exhibit indications of 

 structure which remain occult to a narrow axial one, because oblique 

 rays are in that respect more effective than the axial ray. The 

 above discussion turns solely on the approach to perfect similarity of 

 image and object. 



