544 



LEWIS. 



due to shrinkage, has been deliberately smoothed out, altogether giving 

 the models somewhat flatter surfaces and sharper edges than occurs in 

 the preserved tissue. But this suppression of accidental irregularities 

 has been permitted only when necessary to reveal the true pattern of 

 the cell as indicated by its contacts with others. 



The preliminary count of contacts which the cells have with those 

 about them is partly recorded in Figure 2, by numerals placed within 

 the cell outlines. Of the sixty-three cells there shown, the average 

 number of contacts is 13. 9G. A contact is always a potential facet, 

 and usually, as seen in the figure, is actually such. The contacts 

 of thirty-seven more cells were counted, but, as it happened, the 

 average number per cell remained unchanged — 13.96. In Table I 

 is shown the number of cells, in the one-hundred counted, having the 

 number of contacts indicated at the top of the several columns. The 

 cells range from hexahedra to icosahedra, with no special tendency to 

 form dodecahedra. More than half of them are 13-, 14- or 15-hedra. 

 This is consistent with a typical tetrakaidecahedral form. 



TABLE I. 



Irregularities are due in part to the great difference in the volume 

 of the cells. Calculated roughly from the displacement of the wax 

 models, these cells ordinarily range in size from .0004 to .0014 c. mm., 

 with an average volume very close to one one-thousandth of a cubic 

 millimeter. Since such a cavity if filled with human blood would 

 contain between four and five thousand corpuscles, it is evidently a 

 cell of considerable size. The largest cell modeled measured .00224 c. 

 mm., and the smallest was crowded and flattened to a mere .00008 c. 

 mm. Cells with an excessive number of contacts are usually very 

 large, and of such form as to suggest that an expected division in some 

 particular plane has failed to occur. Thus they may be octagonal 

 above and below, with two tiers of lateral surfaces, so that a vertical 

 division woidd reduce them nearly to normal size and shape. Or 

 they may be hexagonal above and below, with three tiers of lateral 

 surfaces around most of their circumference; these would be reduced 

 by a horizontal division. Certain instances of small cells with few 

 contacts have resulted from an atypical division of cells approximately 



