A NEW EVAPORATION FORMULA 33 
for each day from September 1 to October 10 inclusive, 1010. The stations shown 
at the column headings in Table 14 are those of Table 3 in the order given in that 
table. The figures in parentheses at each column heading in Table 14 are the factors 
from Table 3, and should be considered as repeated down the column. Theother figures 
in Table 14, except in the last two columns, represent the daily rainfall at each 
station in inches from midnight to midnight, as observed by the regular U. S. 
Weather Bureau and Canadian Meteorological Service stations, and by co-opera- 
tive observers. The product of the factor of a station by the observed rainfall at 
that station in inches converts the rainfall to units of 0.0001 foot rise of the whole 
surface of the lake. The sum of such products for any day for all of the stations 
available around and near the lake shore gives the total computed rise of the lake 
surface due to rainfall on it that day. These computed rises are shown in the last 
column but one. In the last column are shown the rainfall on the lake from noon 
of the preceding day to noon of the current day in units of 0.001 foot. This was 
computed from the preceding column by assuming that it is the mean between the 
rise, as computed from the rainfall from midnight to midnight of the preceding day, 
and the rise as computed from the rainfall from midnight to midnight of the current 
day. 
The maximum value of the rainfall on Lake Michigan-Huron in one day shown 
in Table 14 is 0.040 foot on September 6, 1910. The maximum value computed in 
the 28 months used in this investigation was 0.060 foot on August 25, 1910. 
For Lake Superior, 7 2 was computed in the same manner as the above, except 
that the rainfall stations and factors shown in Table 4 were used. 
In the computation of h as shown in Table 14, whenever the rainfall at the 
station for a day was missing, the rainfall at the nearest station at which the rainfall 
had been observed that day was substituted for it, and the substitute station so 
used indicated by enclosing it in parentheses. 
Rainfalls of less than 0.01 inch are indicated by a "T" in the Weather Bureau 
records. These T's were not neglected in the computation of h- The product of a 
factor and a T gives a rise of lake surface less than unity (in most cases) in the 
units used. It was found by experience that the T's were properly evaluated by 
counting 3 to 7 T's 1, 8 to 12 T's 2, and so on. 
EXAMPLE OF COMPUTATION OF /,, LAKE MICHIGAN-HURON 
The definition of Ii is given on page 9 and the method of computing it is 
briefly stated in general terms on pages 19 to 24. It is now proposed to give a 
numerical illustration of the computation of 7i for Lake Michigan-Huron. There 
are five steps involved, namely, (a) the computation of the daily barometric cor- 
rection at each of the gages Milwaukee, Harbor-Beach and Mackinaw, (6) the 
computation of the daily wind correction at each gage, (c) the application of the 
corrections computed in steps (a) and (6) to the observed level to get the corrected 
level at each gage, (d) the combination of the corrected levels at all three gages to 
get the weighted mean corrected level of the whole lake surface, and (e) taking the 
differences of the weighted mean corrected level of the whole lake surface from day 
to day. 
EXAMPLE OF COMPUTATION OF DAILY BAROMETRIC CORRECTION 
To illustrate the method of computing the daily barometric correction, 
as summarized on page 22, the computations pertaining to the Harbor-Beach 
gage will be used. The barometric pressures, as read from the weather forecast 
