( 556 j 
TABLE I. Showers at Batavia, 1866—1905. 
| Number Percent. Mean dur. Mean numb. 
RUE of rainy of rainy Nel of showers | of showers 
hours hours in hours per diem 
| | | | 
January. … 28704 LAG 1575 1367 | 3.26 1.14 
February... | 26136 4565 17.5 12ste Ii 365 > | a 
March...... 28680 2026 10.5 939 | 3.92 0.79 
20) | VR | 27792 1704 6.1 664 2.57 OE 
MANNE | 28680 1138 4.0 HOG. ry) O68" POD 
JUNE veele 27768 1222 4.4 ANT 3.02 | 0.34 
Malye. 98798 924 3.9 341 2,4 | 0.98 
Augustus ... 28680 | 516 1.8 212 2.43 0.18 
September .. 27768 897 ote 324 Paar it 0.28 
October ....|| 28704 1219 4.2 480 2.54 0.49 
November .. 97768 2039 | 7.3 709 2.88 0.61 
December .. 28704 3372 ew 1071 | en heal) 0.90 
== neen Ee | En 
WEAR veen, | 338112 | 25078 7.42 8190 3:06 slo OE 
shows some resemblance to that of April. The phenomenon of rain- 
fall, therefore, is subject to an annual variation consisting of a single 
period as shewn by the percentage of the showers and a semi-annual 
period with minima in April and August. Further it appears from 
Table II that showers of 24 consecutive hours rarely occur (14 
cases in 8190), but that, although very rarely, showers of 100, 109 
and even of 147 hours have been observed in February and March. 
2. Besides the numerical representation as given in Table II, also 
an analytical representation is of some importance as, in the constants 
calculated, peculiarities of the curve of distribution often oceur which 
appear neither in the numerical, nor in a graphic representation and, 
at the same time, these constants can be considered as a quantitative 
measure of these peculiarities. 
In treating frequencies of this kind, the number of constants to 
be used must remain restricted to afew, as every constant necessitates 
the computation of an average of higher order and constants based 
on high moments will, in this case, be practically idle. 
It seems, therefore, desirable not to introduce more than two 
