424 JENNINGS— HEREDITY IN PROTOZOA. [ApriU4, 



these 364 halves the correlation between deptii of constriction and 

 length is greater, amounting to .7818; while the increase in length 

 with each 10 microns of increase in depth of the constriction is 

 5.598 microns. 



While the length thus increases, the breadth decreases. This is 

 evident on inspection of Table XII. The correlation between depth 

 of constriction and breadth of body is therefore negative; its coefifi- 

 cicnt, in the case of Table XII., is — -5232. With each increase of 

 10 microns in the depth of constriction the breadth of body decreases 

 2.630 microns. If again we take into consideration only the 364 

 halves in which lengthening has decidedly begun, omitting thus the 

 uppermost row of Table XII., we find that the correlation decreases 

 to — -3316, and the decrease in breadth for an increase of 10 microns 

 in depth of constriction is but 1.252 microns. This appears to indi- 

 cate that a large part of the decrease in breadth occurs in the first 

 stages of constriction. 



If we compare with the means of the 262 halves in which length- 

 ening has not begun, the means of the 364 in which lengthening has 

 begun (Table VIII., rows i and 2), we find that the length has 

 increased from 87.848 to 93.033 microns, while the breadth has 

 decreased from 55.480 to 49.540 microns. If we examine the means 

 at successively older stages, we find, of course, greater dift"erences. 

 Thus, when the constriction has reached a depth of 36 microns, the 

 10 specimens in that stage show the mean length increased to 101.200 

 microns, while the mean breadth is but 46.400 microns. Similar 

 relations are to be observed if we compare the means of the younger 

 and older sets of each lot shown in Table \TII. 



Since, while the length is increasing, tiie breadth is decreasing, 

 the growth tends to decrease the correlation between length and 

 breadth or even to make it negative. Thus, while in the stage before 

 lengthening has begun (row i, Table VHI.) the correlation is .6546, 

 in the 364 specimens of the same lot, after lengthening has begun 

 the correlation has decreased to — .0938 (row 2, Table VIII.). In 

 a second lot, containing 124 halves, when we throw all the halves 

 together the coefficient of correlation between length and breadth 

 becomes —.1136 (row 5, Table VIII.). In the mirclia form, 106 

 halves after lengthening has begun give a positive correlation between 



