542 
Note. 
observed for sizes of genera, girls with no accidents being most frequent, but there is 
not so long a ‘ tail ’ towards high numbers of accidents. 
A. It is assumed that every girl is equally likely to meet with an accident during 
any short interval of time. The form of frequency distribution deduced does not fit 
the facts. 
B. It is assumed that the girls may be regarded as forming two groups, careful 
and careless, a girl of the second class having a higher chance of meeting with 
an accident during any short interval of time than a girl of the first class. The form 
of frequency distribution deduced still cannot be made to give a good fit to the facts. 
C. It is assumed that the girls are of all degrees of ‘ carelessness ’, i. e. of chance 
of accident during a short interval of time, and that the form of frequency distribution 
of carelessness is given by one of Pearson’s curves with origin at zero, so that there 
are or may be girls of zero carelessness. The form of frequency distribution now 
deduced for numbers of girls with o, i, 2, 3, . . . accidents is found to give an excel¬ 
lent fit to the facts, and from the constants of the fitted distribution we can deduce 
the distribution of carelessness amongst the girls. The interpretation thus arrived at 
is confirmed by finding that the number of accidents met with by a girl during one 
interval is correlated with the number she shows during a subsequent interval. 
We may conclude that assumptions A and B are in error, whereas C may 
be right, and since it is reasonable and probable arid otherwise confirmed it probably 
is the right interpretation. 
Similarly, from the view of evolution now discussed, the form of the frequency 
distribution for sizes of genera can be deduced, and it appears to give a very good fit 
to the facts. This is, in so far, confirmatory evidence of the truth of the view. 
G. UDNY YULE. 
