August 11, 1899. ] 
The latter were taken from various 
records available in New York City and 
consisted for the most part of misdemeanors 
under the observation of the police force of 
the city, the teachers in the public schools 
and the warden of the penitentiary, although 
the death record kept by the Board of 
Health was also considered. The total num- 
ber of data considered exceeded 400,000, 
made up of cases of assault and battery, 
suicide, and arrests for insanity by the po- 
lice, recorded misdemeanor in the peniten- 
tiary and public schools, record of deaths 
for the city, record of errors made by clerks 
in seyeral of the larger national banks and 
record of strength tests made in the gym- 
nasium of Columbia University. The classes 
of data varied in number from 1,000 to 
more than 100,000, and were for the years 
included within the period for which the 
weather conditions had been tabulated. By 
a somewhat elaborate process of computa- 
tion the exact percentage of each class of 
data occurring under each of the definite 
meteorological groups was determined ; for 
example, the percentage on fair or cloudy 
days, on days when the temperature was 
between 15° and 20°, ets., for all the fifty or 
more conditions studied. 
We have already stated that from the 
meteorological data, the normal prevalence 
of these conditions had been determined. 
It may be readily seen, however, that the 
normal prevalence of a condition equals the 
expected occurrences of each of the classes 
of data for that condition—for instance, if 
30 % of the days for the ten years were 
fair we should expect 30 % of the assaults, 
suicides, etc., to have occurred upon fair 
days if the character of the day had no influence. 
If, however, 35 % did actually occur we 
should infer that the effect of fair days was 
to increase the number of assaults, as, in- 
deed, this study has shown to be the case. 
The conclusions of the paper are based 
entirely upon this comparison of occurrence 
SCIENCE. 
177 
of data under a given meteorological con- 
dition, with the prevalence of that condi- 
tion. Both were reduced to percentages. 
When the occurrence for a given condition 
was found to exceed the expectancy the ex- 
act excess was computed and when below 
the deficiency. More than one hundred 
and fifty curves were constructed showing 
these relations (see ‘Conduct and the 
Weather,’ Monograph Supplement No. 10 
to The Psychological Review), a few of which 
are shown with this paper. 
Moderately high temperatures were found 
to be accompanied by excess in all the mis- 
demeanors considered; low temperatures 
by deficiencies. The temperature group 
80-85° showing an excess of 68 % for as- 
saults by males and 100 % for those by fe- 
males. The next higher group, however, 
shows a drop to 83 % excess for the former 
and a deficiency of 33% for the latter. 
This sudden falling-off for conditions of in- 
tense heat is shown for nearly all classes of 
data, and is undoubtedly due to the fact 
that under such temperature there is little 
energy available for offensive conduct. 
Death, suicide and the recorded error in 
banks alone remain excessive under such 
conditions. 
Figs. 1 and 2 give a comparison of the 
occurrence of assault and death (male) re- 
ferred to the temperature conditions for 
each month of the year. It may be seen 
from them that during the winter months 
the temperature produces but little effect, 
there being but slight excesses or defic- 
iencies for any of the groups. Excesses 
and deficiencies read vertically. The hori- 
zontal lines show differences of 20 %, read 
from the heavier base lines. 
When, however, we come to the spring, 
the higher temperature for the months are 
accompanied by.a very marked increase in 
the number of assaults (April, 70-75°, an 
excess of 64%) and one less marked for 
death. During the heated summer time 
