ERRORS DUE TO FIXATION AND DIFFUSION 131 



which has general validity for the detection of diffusion arte- 

 facts is to use superimposed sections, with the section to be stud- 

 ied superimposed upon a similar section which is devoid of the 

 substance under investigation. This technique has been dis- 

 cussed previously in this volume, and in two earlier papers 

 (Danielli, 1946, 1949). So far, however, no studies have been 

 made of diffusion during paraffin infiltration and embedding. 

 Although diffusion in paraffin is unlikely to be significant when 

 such substances as nucleic acid are considered, with many of the 

 substances for which the microincineration technique has been 

 used the errors may be serious. 



It should be feasible to study diffusion in wax, by means of 

 the superimposed-section technique. 



As a result of fixation marked increases usually occur in the 

 refractive-index differences between different parts of cells, and 

 as a result diffraction and light scattering become more pro- 

 nounced. Caspersson suggests that this may be minimised by 

 either of two methods. The first is that commonly practiced in 

 microscopic work — namely, mounting the specimen in a medium 

 of the same refractive index as the part of the specimen to be 

 examined. A second most ingenious device, probably applicable 

 only to frozen-dried material, is to make use of the very slow 

 diffusion of proteins, etc., which occurs in specimens mounted in 

 glycerol. When a frozen-dried specimen is mounted in glycerol 

 the boundaries of the various cell organs are sharp, and the re- 

 sultant abrupt discontinuities in refractive index cause marked 

 losses of light. But after some time in glycerol a small degree 

 of diffusion reduces the sharpness of the boundaries, so that 

 there are continuous rather than discontinuous changes in re- 

 fractive index, and light losses are correspondingly reduced. 



Errors due to Inadequate Chemical Procedures 



The first requirements of a -cytochemical reaction are that the 

 end-products should be known and either non-diffusible or read- 

 ily examined by techniques for detecting diffusion. Attention 

 has already been directed to these problems in preceding parts of 

 this book. For quantitative studies a further chemical problem 

 becomes prominent, for it is necessary to know to what extent 

 the cytochemical reaction is quantitative. The requisite infor- 

 mation is seldom available at the present time. 



