164 MICEOSCOPIC VISION. 



1st. They usually occur when a small cone of illumination 

 is emploj^ed. 



2ndly. They must be an integral multiple of some real 

 structure ; thus, if the real structure is 12,000, the ghost 

 may be 24,000, 36,000, or 48,000, etc. ; but it can never be 

 18,000 or 30,000. 



3rdly. The ghosts invariably have a focus differing from 

 that of the true structure. Thus the false structures may 

 appear to be above or below the real structure ; the position 

 depending upon the over — or under — correction of the objec- 

 tive. 



A question will naturally arise. What is the resolving 

 power when a f axial cone of illumination is used ? 



The answer is, 70,000 times the N.A. of the objective ; it 

 will then be said that this is only 73% of the amount given 

 as the resolving limit in the tables published in the Journal 

 of the Royal Microscopical Society. In reply, it may be 

 pointed out that the limit given in the tables in the R.M S. 

 Journal is the limit of resolving power when a narrow 

 illuminating beam of extreme obliquity is employed, and 

 that this limit is twice the wave length multiplied by the 

 Numerical Aperture, or 96,000 times the N.A. 



Although this is perfectly true theoretically, yet practic- 

 ally it is not so, for many delicate resolutions which cannot 

 be seen at all with oblique illumination are visible with 

 the f cone ; and further, the narrow oblique beam is a fruitful 

 source of false images. 



It may be next pointed out that the plan of using light of 

 a shorter wave length does not give the advantages that 

 theory would lead one to expect. With a | cone it is found 

 that the resolving limit for light in the region of the F line 

 is 80,000 times the N.A., but that this gain ceases when 

 objectives which have a higher N.A. than 0-8 are employed. 



