44 A NEW METHOD OF ESTIMATING STREAM-FLOW 
miles per hour from the southeast to lower the mean water surface for a day as 
much as 0.001 foot at Harbor Beach. 
With such a table as No. 19 before one, together with the Weather Bureau 
records of hourly wind velocity and direction, it is a simple matter to evaluate the 
daily wind effects, particularly since they are less than 0.001 foot most of the time 
at the Michigan-Huron gages. For example in the eight months June to September 
inclusive each of 1910 and 1911, there were only 6 days on which the wind effect 
at Harbor Beach exceeded 0.001 foot. The maximum effect in that time interval 
was 0.020 foot on July 24, 1911. 
The preceding two paragraphs serve to illustrate step (b), page 33, in the 
Computation of I u 
ELEVATIONS AT MILWAUKEE. HARBOR BEACH AND MACKINAW FOR JUNE TO SEPTEMBER 1910, 
CORRECTED FOR WIND AND BAROMETRIC EFFECTS 
In a manner similar to the numerical illustration given for August-September 
1910 on pages 33 to 44, the observed elevation of the water surface at each of the 
gages Milwaukee, Harbor Beach and Mackinaw on Lake Michigan-Huron, and 
Marquette on Lake Superior, was corrected for wind and barometric effects for 
each day of the 28 months of observations. The observed elevations so cor- 
rected at Milwaukee, Harbor Beach and Mackinaw for June to September in- 
clusive, 1910, are shown in Tables Nos. 20, 21 and 22, respectively. On days 
on which there was no record of observed elevation the missing value is indicated 
by a dash. The values of the corrected elevations which are in parentheses are 
values detected by the criterion developed from the least-square solutions which 
served to determine the barometric effects. This criterion, for each gage, is 3.5 
times the probable error of a single observation as determined by the last baro- 
metric solution at that gage. For the four gages Milwaukee, Harbor Beach, 
Mackinaw and Marquette, the probable errors of a single observation of the 
corrected rise in the water surface (corrected for wind and barometric effects) as 
determined from the least-square solutions are, respectively, ±0.028 foot, ±0.025 
foot, ±0.023 foot and ±0.020 foot. Hence the criterion by which to detect 
abnormal corrected elevations are, respectively, ±0.10 foot, ±0.09 foot, ±0.08 
foot and ±0.07 foot. Thus, between August 22 and August 23, 1910, the cor- 
rected elevation is shown to have fallen fron 580.26 feet to 580.10 feet at the 
Milwaukee gage (Table 20). This is a change of 0.16 foot, which is greater than 
the criterion for that gage, ±0.10 foot, hence the corrected elevation for August 
23, 1910, at that gage was not used in evaluating the elevation of the mean surface 
of the whole of Lake Michigan-Huron for that day. It so happened that on 
August 23, 1910, the record at the Harbor Beach gage was missing (Table 21). 
On that day also at the Mackinaw gage (Table 22), the indicated rise from the 
preceding day in the corrected elevation is 0.22 foot, which is larger than the 
limit ±0.08 foot for that gage established by the above criterion. Hence for 
August 23, 1910, the evidence shows that the lake surface was affected in such an 
abnormal way, possibly by seiches initiated by abnormal changes in barometric 
pressures, that the corrected elevations are unreliable and therefore that day 
could not be used in the investigation of the laws of evaporation. 
In Tables 20 to 22, the sums of the barometric and wind corrections, added 
to the observed elevations, give the corrected elevations of the water surface. 
This illustrates Step (c), page 33. 
