M. Greenwood and J. W. Brown 
479 
for Bohemian cases. The mean body-weight is greater than Pearl's figure and the 
coefficient of variation between that of Pearl and Pearson*. 
Since Pearl and Blakeman did not restrict their inquiry to persons between 25 
and 55 years of age, any comparison of our results with theirs must be of doubtful 
validity. Similarly we cannot control the constants involving age by comparisons 
with the work of other biometricians. In the first study, the correlation between 
age in years and " healthy " heart weight for males was given as "136 + "025 and 
from the same data we find in the case of females r = "155 + "029, in good agree- 
ment with the former value. Our present results cannot be said to differ signifi- 
cantly from the old values and we may quote a passage from the other memoir: 
" When we remember that the healthy heart is on the average much smaller than 
the heart in disease, and that sickness on the average increases continuously with 
age, we shall probably lay less stress on the general a priori idea that the weight 
of the adult heart increases very sensibly with age alone f." We see no reason to 
modify this view ; further the recorded ages are likely to be subject to a con- 
siderable observational error in this class of material. 
The general conclusion therefore both of the present and the first study is that 
age alone is not proved to be sensibly correlated with the weights of the viscera in 
an adult bealthy " General Hospital Population." 
The correlations involving body-weight are the most interesting in our new 
series since, with the exception of Pearl's results, they are so far as we are aware 
the first reductions of this variable in its association with the visceral weights 
which have been published. It may be freely admitted that the weight of a corpse 
must be so appreciably affected by the circumstances immediately preceding 
death that it cannot correspond at all closely to the weight during life. The force 
of this objection is diminished by the large percentage of violent deaths to be 
found in our series, since these cases would not be affected by the rapid wasting so 
often associated with a fatal illness. 
Chiefly for this reason, we are not without hope that the present calculations 
throw some light upon the true correlation between heart weight and body- 
weightj. One inference seems clear enough, viz. that no opinion ought to be 
ventured as to whether the weight of a heart is or is not normal without taking 
into account not only that of the other organs but also the body-weight. This 
leads to the question as to whether a formula of the nature of a regression 
equation might appropriately be employed to connect heart weight with other 
variables. In selecting variables upon which to base a prediction of the heart 
weight, the ideal is that each member of the set shall be highly correlated with 
the heart and that the correlation of the chosen variables one with another shall 
be low. For the purposes of a first trial, we decided to found the prediction 
* K. Pearson, " The Chances of Death, etc.," Vol. I. pp. 293, etc. 
f Greenwood, op. cit. p. 69. 
% It is also to be remembered that there must be an appreciable post mortem loss of weight owing to 
evaporation in the cold chamber. 
61—2 
