110 THE AMERICAN MONTHLY [July 



in ^detecting the spores of a monad, itself only 1-6000 of 

 an inch in diameter and themselves only 1-240000 of an 

 inch ; or more especially (because here was involved real 

 " resolution " of fine detail) the process of division in the 

 nucleus of a cell, itself only 1-20000 of an inch long. In 

 such cases v^rhat will be found and is to be observed is 

 unknown ; accurate periodicity of structure is probably 

 absent ; and mere artificial force of clearness in " reso- 

 lution," even if obtainable (which it seldom is) is worth- 

 less in comparison with known trustworthiness in the 

 image so far as it goes. Taking any such case as this 

 for our test-object, and comparing it with the treatment 

 of the A. pellucida as described, we shall be able to ap- 

 preciate the proportion of both truth and error — for there 

 is truth as well as error — in the " spectrum " theory. 



16. We cannot help, in the first place, seeing much 

 error. While the minuteness of structure to be detected 

 by Dr. Dallinger (in an unknown object) was as great, the 

 method of proceeding described for the A. pellucida is im- 

 practicable, and would be useless even if practicable, real 

 work has to be done by far difi'erent means. The finest 

 lenses, used with a wide and solid aplanatic cone of light 

 could alone do such work ; and moreover, earlier lenses of 

 1*48 N. A. were surpassed in results by apochromatic 

 lenses of 1-40 N. A., better corrected for spherical aberra- 

 tion — the meaning of which we shall see. Supposing the 

 microscopist, however, to know or suppose the measure of 

 minuteness in the divisions of the cell-nucleus, he would^ 

 have to employ (w ith doubtless some modification in 

 detail) arrangements for plane-wave illumination generi- 

 cally similar to those he describes for the diatom. But 

 he would be wrong, and the results would be nil. 



Narrow pencils and annuli have of course been tried, 

 for the contrast they give. The probable reason of 

 failure is want of sufficiently regular peiiodicity m the 

 detail. Only such periodic detail is shown better by 



