4 STUDIES IN GELS I25 



present the slits shown in the image (Fig. 85 b, Muller and Pase- 

 WALDT, 1942). Hence, the diatom wall is not a massive structure, 

 but consists of an outer and an inner lamella, separated by sub- 

 microscopic spaces and connected by pillar-shaped buttresses (Fig. 85b). 



Fig. 82c shows what was meant in Table XIII (p. 120) by "most 

 favourable magnification". A microscopic image or a microphoto- 

 graph can be magnified at will by projection, so that the magnification, 

 or better the image scale, does not provide an unambiguous reference 

 by which to compare different microscopes. Nevertheless there is a 

 limit to the magnification of images, in that the contours become 

 rague when the image scale becomes too large. For this reason there 

 exists a "profitable" magnification which is best maintained in micro- 

 photography and which is designated as "most favourable magnifica- 

 tion". Strong magnifications of the microphotographic negatives ob- 

 tained result in poor definition as shown in Fig. 82c, where the 

 systems of lines are hazy as a result of a magnification of 10,000, which 

 is seven times the "profitable" one of 1500. 



The most successful objects of research for the electron microscope 

 are the submicroscopic particles of suspensoids, such as inorganic 

 coUoids, virus particles, bacteriophages, organic macromolecules 

 which exceed 50 A diameter. Unicellular objects such as diatoms 

 and bacteria are too thick; they furnish black shadow pictures and 

 details are only to be seen if the object is perforated or provided with 

 surface appendages (cilia, flagella). The colloid particles, however, 

 are thin enough to transmit electrons, producing real so-called phase 

 images. 



Fig. 84a shows shadowed macromolecules of haemocyanin from 

 the blood of a snail. This micrograph was the first clear-cut picture 

 of protein macromolecules (Williams and Wyckoff, 1945). Ac- 

 cording to SvEDBERG, the globulat proteins aggregate by 2, 4, 8 etc. 

 to form bigger particles. This rule (see p. 141) found by experiments 

 with the ultracentrifuge, is now substantiated by electron micrographs 

 such as Fig. 84b (Polson and Wyckoff, 1947). 



The agents of virus diseases have been found to be macromolecules 

 of different shapes. The classical tobacco mosaic virus is rod-shaped, 

 as proved by indirect methods (double refraction of flow. X-rays). 

 The electron micrograph (Fig. 84c) shows that the length of the rods 

 is not defined. Their mean length depends on the p^^ of the dispersing 



