MICHIGAN ACADEMY OF SCIENCE. 
79 
The epoch chosen is 1880 to 1904. It has been selected ns n compro¬ 
mise between too long and too short a series. A series of rain observa¬ 
tions is too short when the average of a longer period would have a 
value materially different. It is too long when few stations cover its 
full extent and its greater part comes before the beginning of many 
of the series of observations that must be made use of. Sixty-three of 
our 143 stations are of less than fifteen years length of observation, 
most of them for the last ten or fifteen years only, and the observations 
previous to 1880 at the few stations where any were made in the region 
are unsuitable for comparison with these recent data. 
Mr. A. J. Henry* says: “The average variation of a twenty-five year 
period is about 5 per cent,” meaning that the single year’s observa¬ 
tion will, on the average, differ by that amount from the mean of the 25. 
Of the 143 stations from which data have been used, eighty-three are 
in Michigan, and sixty in neighboring territory, furnished by the cour¬ 
tesy of the United States Weather Bureau, the State Weather services, 
especially that of Michigan, and the Meteorological Service of Canada! 
The data are accepted as furnished since an adequate study of the 
original observations could be made only within the services in charge 
of them. The discussion by Prof. Henry at p. 9 of Bulletin D, states 
that the registers of the roof gages of the Weather Bureau stations prob¬ 
ably fall short of the true precipitation by 5 or 10 per cent. It might 
be thought that the eighteen records from such stations included in our 
list would reveal this defect on mapping them beside records from vol¬ 
unteer observers in Michigan and other states. No tendency of the sort, 
however, appears in this region. As to the rain measurements of vol¬ 
unteer observers, it is easy to understand that there are great possibili¬ 
ties for inaccuracy in the conditions of their work and occasional ex¬ 
amples come under one’s eye of what should not be done in the way of 
scientific observation, yet the discussion of their results shows that 
these unpaid workers have deserved the greatest praise for careful and 
unremitting attention to a task that no one who has not carried on 
meteorological observations through fair weather and foul, through 
work days and holidays, can fully appreciate. 
The following little table shows the grouping of the observation sta¬ 
tions by years of record: 
Stations. 
Years of re 
25 
25 
19 
20-25 
2G 
15-19 
25 
10-14 
48 
5-9 
The averages for the complete records (1880-1904) were entered 
directly on a blank chart. The corrections applied to'oilier records are 
based on the fact that there are years and groups of years of heavier, 
others of lighter rainfall than tiie average. Thus at the twenty-one 
complete stations the average rainfall of the five years, 1880-1884,' was 
almost invariably 10 to 15 per cent greater than the average of the 
whole twenty-five years, as is seen in the table below. 
* Weather Bureau Bulletin D, Rainfall of the U. S., p. 320. 
