ERRORS DUE TO SCATTERING 127 



scattering and refraction of light in the object, as well as by the 

 absorption of photons by individual constituents of the object. 

 Since the analysis of the distribution of chemical substances in 

 the specimen depends entirely on the amount of absorption of 

 light, it is necessary to reduce the effects of scattering and refrac- 

 tion to a minimum, and, if these effects are still significant, to 

 make measurements of the errors due to them in order to correct 

 apparent absorption values. Caspersson has set out these prob- 

 lems in some detail. His chief conclusions are as follows: 



1. The object must be clearly resolved. A satisfactory absorption curve 

 cannot be obtained if the diameter of the object is less than three times 

 the wavelength of the light used. 



2. It is not sufficient to measure the absorption of light at one wavelength 

 only, because the absorption bands of the various components of tissues 

 overlap to a considerable degree. 



3. The optical system must fulfill Abbe's sine condition. Otherwise the 

 distribution of light in the image does not correspond to that in the 

 object. 



4. Every point in the object must be illuminated by incoherent light, 

 as, for example, by using Kohler's method of illumination. 



5. A correction must be applied for the amount of light lost by scatter- 

 ing. This is best done by measuring the amount of light actually scattered. 

 A much more common procedure is to measure the loss of light in the 

 specimen at a wavelength at which the specimen is believed to have 

 practically no power of absorbing light. The light loss at this wavelength 

 is then assumed to be caused by scattering. It is then commonly assumed 

 that the scattering is varying inversely as the fourth power of the wave- 

 length, and the light loss by scattering at other wavelengths is calculated 

 on this basis. This method is by no means satisfactory, since, whereas the 

 loss of light varies as 1/X n , where X is the wavelength, the value of n is a 

 function of particle size in the specimen and varies from 2 to 4 for particles 

 between 10 m/A and 10 fi in diameter. Since this is a range of particle 

 sizes to be expected in cells, the correction is obviously difficult to make 

 from observations at one wavelength. From this point of view there is 

 much to be said for working at as large a wavelength as is compatible with 

 adequate resolving power. 



6. To minimise scattering and refraction, the specimen must be mounted 

 in a medium of as nearly as possible the same refractive index as the speci- 

 men. Caspersson suggests that the ratio of the two refractive indexes 

 should not exceed 1.1. 



7. If the points mentioned above are adequately taken into considera- 

 tion, including the limit to the ratio of the refractive indexes of speci- 

 men and mounting medium, then Caspersson calculates that the minimum 

 permissible numerical aperture which can be used is 0.85. 



