ERRORS DUE TO STATES OF AGGREGATION 129 



obtain reliable results only if the specimen is examined with 



polarised light. 



Although it is probable that serious errors due to orientation 

 of anisotropic molecules are rare, the danger of error is none 

 the less very real. It has been shown by Schmidt (1937), Frey- 

 Wyssling (1948), Caspersson (1950), and others that nucleic acid 

 molecules are markedly anisotropic and quite commonly oriented. 



Errors due to States of Aggregation 



These may be classified into two groups: errors due to the 

 aggregation of individual molecules, and errors due to aggregates 

 of size near the limit of resolution. 



Even when the group which is absorbing light is known, it is 

 not always easy to define its absorption band in a specimen. The 

 position of an absorption band may often be readily altered 

 by the formation of complexes with other substances in the speci- 

 men. Well-known examples of this are the change in colour of 

 astaxanthin from red to blue when it is adsorbed on certain pro- 

 teins, and the wide variations in the spectrum of haem when it is 

 adsorbed upon different proteins. These changes in spectrum 

 are not restricted to the visible spectrum. Caspersson has found 

 a marked difference in the ultraviolet between the positions of the 

 maximum of absorption by pure nucleic acid and the maximum 

 found with nucleic acid in cells. These effects must be expected 

 whether the absorbing molecule is an intrinsic part of the cell 

 or an added dye; they are likely to be affected by the mode of 



fixation. 



The second effect may be most clearly appreciated if we con- 

 sider a chequerboard, as in Fig. 8. The square as a whole is the 

 area whose absorption is under consideration. The hatched 

 areas contain a high concentration of absorbing substance; the 

 clear areas have no absorbing material. Consider the case in 

 which the concentration of absorbing material in the hatched 

 areas is so high that, if a large area had the same composition 

 as the hatched areas, practically no light would be transmitted. 

 Then the actual absorption of light by a chequered area depends 

 upon the relationship between the distance d and the wavelength 



of light. 



If d » a, the chequers will be clearly resolved and the area will 

 transmit 50 percent of the incident light. If A » d, the chequers 



