THE SCIENCE OF MICROSCOPY. 323 



II. If it be asked in what particulars does this ''science of 

 microscopy " differ from the current practice, an answer may be 

 found in general terms by first considering what that practice is, 

 wherein it fails, and wherein lie the means of improving it. 



For technical purposes the several problems connected with the 

 action of the microscope lenses are settled by direct optical 

 analysis of the course of the pencils of light by which the image 

 is formed. If test objects are employed those are selected whose 

 known character renders it easy to draw the necessary inferences 

 for estimation and correction of the lens action. But in ordinary 

 microscopy the object is an unhioivn quantity which we seek to 

 determine by inference from the images of it formed in the 

 microscope. This reversal of procedure lessens the certainty of 

 our inference. For while it is easy to determine the performance 

 of a lens-combination by ascertaining whether the image formed 

 by it is truly conformable with the hioivn object, it is not easy to 

 prove that this image invariably represents the structural 

 constitution of an unhiown object. Supposing for example it is 

 desired to ascertain the optical efficiency of an objective, a certain 

 means offers itself in the analysis of its image-forming function, 

 which can be effected by tracing the passage and reunion of 

 pencils of light admitted through different parts of the lens in the 

 image formed at its posterior focal plane. But when we have 

 thus ascertained the perfect accuracy of a lens performance, we 

 have no assurance that tlie effects of light and shadow in this 

 optically correct image shall unmistakeably indicate the material 

 conriguration of the object producing it. On the contrary it has 

 been shown that an optically perfect lens may yield images of 

 artificially prepared objects, {e.g.^ lines drawn on an absolute plane 

 surface) which suggest a different structural conformation from 

 that which actually exists. Many natural objects whose " structure" 

 is a matter of inference, are admirably suited by their transparency 

 and strongly defined ''markings" for testing the optical qualities of 

 lenses. Yet there can scarcely be adduced a more striking proof 



