196 C.-E. A. WINSLOW AND ANNE F. ROGERS 



liquefaction after four weeks is shown in Table 42. Fig. 4 shows 

 graphically the skew curve plotted from these data. 



There is, as would be expected, a gradual falling-away toward 

 the highest amounts of liquefaction, and the abrupt downward 

 falling of the curve toward the non-liquefiers at o is extremely sig- 

 nificant as indicating a sharp difference between the two groups. 

 If it had been practicable to plot the non-liquefiers on this figure, 

 the curve would have gone up at an acute angle more than twice 

 as high as that of the mode of the liquefiers. This angle divides 

 with more than usual definiteness those organisms which liquefy 

 from those which do not liquefy gelatin. 



The correlations of gelatin liquefaction with other properties 

 have been already considered. We have found that a high pepton- 

 izing power is rare among the earth and water cocci and the sarcinae. 

 It is most frequently associated with the smaller individual cells 

 among the non-packet-formers. It is absent or very rare in the cocci 

 which show only faint surface growths. Finally, it appears that 

 the white, colorless forms which produce high acidities, as well 

 as the red chromogens, are non-liquefiers. The yellow cocci which 

 produce little acid are moderately active liquefiers, and the orange 

 forms with a moderate acid production show the highest pepton- 

 izing power. 



IV. CONCLUSIONS FROM THE INVESTIGATION, 

 i. FOUNDATION OF SUBFAMILIES AND GENERA AMONG THE Cocci. 



The extreme variability of the cocci has appeared with great 

 clearness in the present study. Almost every one of the characters 

 measured shows a wide range of fluctuations. In view of the gen- 

 eral laws of variation, the absence of sexual reproduction, and the 

 susceptibility of the bacteria to the direct influence of the environ- 

 ment, this is precisely what should be expected. It makes it, however, 

 clearly impossible to draw sharp and arbitrary lines for any single 

 character by which individual organisms can be naturally classified. 



If, on the other hand, we examine a series of individuals with 

 the idea of discerning central types about which they vary, the prob- 

 lem begins to solve itself, since such types are easily apparent. Cer- 

 tain organisms tend to show the packet grouping some invariably 



