A NEW EVAPORATION FORMULA 29 
should cover the same time interval. In the computation of e, the mean temperature 
and vapor-pressure on which it depends do cover this interval. But, in the com- 
putation of the vapor-pressure, the time interval covered is from 12:30 p. m. of the 
preceding day to 12:30 p. m. of the current day. The mean of the 8 p. in. value of 
the preceding day and the 8 a. m. value of the current day is approximately the 
mean for the 24-hour period stated in the preceding sentence, which overlaps, ap- 
proximately equally, the last half and the first half, respectively, of the preceding 
and current days. A more exact method of computing the mean vapor-pressure for 
the two days might have been obtained by first computing separately the mean for 
the preceding and the mean for the current day from their 8 a. m. and 8 p. m. values, 
respectively, and then taking the mean of these two means. This would have 
doubled the labor, and it is believed that it would have produced only moderately 
greater accuracy than the method adopted. 
The corresponding values of e for Lake Superior were obtained in the same 
manner, except that different meteorological stations were used as shown in Table 
2 and on Plate 4. 
EXAMPLE OF COMPUTATION OF (j^ - X ). LAKE MICHIGAN-HURON 
The definition of x is given on page 9. It is an unknown to be determined 
indirectly from the least-square solutions by successive trials of various values, the 
final value adopted being the one which makes the sum of the squares of the 
residuals a minimum. 
The final adopted value for x in this investigation is 2.6. The more detailed 
statement of the method of arrival at this value, and the evidence in support of it, 
will be reserved for a later place. This preliminary statement of its value is given in 
order to use that value in the numerical illustration following. 
The definition of w is given on page 9, and the meteorological stations at which 
it was observed are shown in Table 1 and on Plate 1. The computation of 
(77^ — 2.6] for Lake Michigan-Huron for the dates September 1 to October 10 
inclusive, 1910, is shown in Table 11. The tabular values in this table are values 
w 
of jf-f. at the stations shown ; w, in all columns of this table except the last, referring 
to any wind velocity, and is the total travel of the wind in miles each day as recorded 
w 
by self-recording anemometers. The mean value of y^ is shown in the twelfth 
w 
column. In the next column is shown the mean value of ttjtt from noon to noon, 
this being taken as the mean between the two midnight-to-midnight values in the 
column preceding, and assumed to be the mean T^th part of the wind travel for the 
lake for the two days ending at midnight of the current day. In the last column are 
shown values of (^.—2.6), these values being limited to positive ones only, for 
reasons to be explained later in the proper context. These are the values used in 
the observation equation (1). 
The corresponding values for Lake Superior are computed in the same manner. 
The meteorological stations for this lake are shown in Table 2 and on Plate 4. 
