Karl Pearson 
405 
than in length. Thus of the six individuals measured thrice, four exhibited no change at all 
in length, three no change at all in breadth during the last 24 hours. It seems therefore that 
individuals either cease growing at a certain interval after division or else have periods of 
rest. The unit oi 5 n does not seem sufficiently small to obtain accurate results for breadth 
measurements. For, anything between 72-5 and 77'5 would be registered as 75. But anything 
just below 72-5 or just above 77-5 would be registered as 70 or 80 respectively. Thus a real 
growth just less than 5 jx might not appear at all, and a real growth just greater than 5 fi 
might appear as a growth of 10 /x. As the standard deviation in breadth for the whole 
series is less than twice 5 /x, this cannot fail to have serious influence in dealing with such 
a short series as 11 pairs. Hence the conclusions reached below are purely tentative, but 
they will suffice to show the importance of allowing for the growth factor in dealing with 
like cases of homotyposis. 
(2) In order that we may safely apply the results deduced for growth from the present 
series to Dr Simpson's earlier series, we must determine whether the former may, as far as its 
statistical constants are concerned, be considered a random samj^le of the latter. Accordingly 
the new measurements were taken as if they had been 44* 'original measurements of length, 
breadth, etc., on separate individuals. 
TABLE V. 
Constants of Tiuo Series. 
Organ 
Mean 
Standard Deviation 
Coefficient of 
Variation 
100 series 
44 series 
100 series 
44 series 
100 series 
44 series 
Length 
Breadth 
Index 
Interval from division in hrs. 
229-050 
68-125 
29-913 
227-954 
74-205 
32-557 
18-023 
19-152 
9-155 
4-029 
18-099 
9-997 
3-470 
11-153 
8-361 
13-439 
7-940 
13-472 
TABLE VI. 
Correlations of Two Series. 
Organs 
100 series 
44 series 
Lengths of 1st and 2nd Members ... 
Breadths of 1st and 2nd Members ... 
Breadth of 1st and Length of 2nd Member-I- 
Indices of 1st and 2nd Members 
Length and Breadth of 1st Membert 
•914+ -Oil 
-782 + -026 
•447 + -054 
•664 + ^038 
•421 + ^055 
•698 + -074 
-574 + -097 
-641 + -060 
•114 + -100 
•594 + ^066 
* In order not to weight individuals and to leave the series as homogeneous in age-variation as 
possible the three pairs measured thrice were only included as far as their first two roeasurements are 
concerned. 
t It is a remarkable result that breadth of second member is more highly correlated than breadth 
of first member with length of first member in both Dr Simpson's series. How far is some element of 
"compensation in division" indicated? 
