A CRITIQUE OF CYTOCHEMICAL METHODS 231 



which may be measured by visible light (Pollister, 1950, 1952a). It 

 happens that when tissues are so fixed that the cell consists of little more 

 than nucleoprotein, all cellular structures have very nearly the same 

 optical dispersion, and it thus becomes possible to mount the specimen in 

 a medium (an oil) which matches the refractive index at any wave length. 

 Under these conditions unstained structures are invisible even by dark- 

 field or phase contrast, showing that nonspecific light loss is negligible. 

 If such material, colored by a reaction, is measured while mounted in oil at 

 or near the appropriate refractive index, practically all the light loss may 

 be assumed to be due to specific absorption by the chromophore of the test. 

 [One possible source of error is that of anomalous dispersion near the 

 absorption peak of the chromophore as pointed out by Scott (1952) and 

 Ornstein (1952).] Another approach is to measure a cell twice, first before 

 (a blank) then after (a test) development of color, a procedure which has 

 been followed with the Millon reaction for proteins (Pollister and Mirsky, 

 1946; Pollister, 1950). Another method is that used in photometry of the 

 natural absorption of nucleic acids, in which the blank is the second meas- 

 urement made after removal of the nucleic acid by nuclease digestion or 

 chemical extraction (Fig. 6-5 and Table 6-4). This is somewhat less 

 satisfactory than the protein blank because the component of the non- 

 specific light loss due to nucleic acid is also removed and thus becomes 

 added to the apparent specific chromophore absorption. 



When the refractive index of the mounting medium is markedly differ- 

 ent from that of the section (e.g., when an unstained section is in water), 

 the nonspecific light losses become appreciable. The methods of ultra- 

 violet microspectrophotometry have been applied either to living cells or 

 to sections which are mounted in glycerin, after either freeze-drying or 

 fixation (e.g., in acetic alcohol). In the two former materials nonspecific 

 light loss is believed by Caspersson (1950) to be minimized in some cases 

 by the absence of sharp phase boundaries. In nearly all fixed material 

 the nonspecific light loss is always considerable (Fig. 6-2D and 6-5). 

 Apparently no mounting medium for ultraviolet studies closely matches 

 the refractive index of such fixed sections. Hence ultraviolet absorption 

 studies must always grapple with the problem of estimating the scatter 

 and internal reflections. As Caspersson (1950) has said, "the most 

 important of all conditioning factors for quantitative microspectrograph}^ 

 is the elimination of the sources of errors caused by these factors." Cas- 

 persson elected to estimate the nonspecific light losses, where appreciable, 

 in the preparations, not by a measured blank as described, but by the 

 unusual method of computing them from analogy with the light losses 

 which he had previously studied in solutions of colorless salts — in which, 

 of course, all light loss was nonspecific. In spite of the urgency and 

 priority of this problem for any quantitative interpretation of the nucleo- 

 protein ultraviolet absorption curves, there has never been a complete 



