90 
EIGHTH REPORT. 
son of the fiye-year averages at east-facing Alpena and west-facing Ran¬ 
goon on Lake Huron, illustrates this pretty well. Perhaps the chief 
inference to be drawn is that in selecting reference stations for the 
reduction of imperfect series similarity of situation will be an important 
guide. 
Fig. 6 reproduces a portion of the latest rainfall chart of the United 
States, made by Mr. Henry in 1902, and published in the Monthly 
Weather Review of April of the year. It is the first such chart I think 
that our Weather Bureau has published in which some attempt is made 
to reduce imperfect series to the epoch adopted for study “by a process 
of extrapolation.” It was only done in some cases however, though in 
Europe it is the accepted method of constructing isohyetal maps. The 
period of this chart is 1870 to 1901. It has as good an agreement with 
our chart as could be expected from the employment of twenty Michi¬ 
gan stations where we are using 84. The stations employed are of 
course the best ones. 
Fig. 7 reproduces Section Director Schneider’s map of 1900. I have 
seen no statement of the plan on which it was drawn, but internal 
evidence suggests that short series have not been reduced to agreement 
with their longer term neighbors. The map cannot fairly be compared 
with Fig. 1, which has the advantage of four more years time of obser¬ 
vation, a great advantage indeed in a state where so many of the series 
of observations began in 1897. 
On the pages that follow are reproduced the original data used. The 
number printed at the head of each column refers to the number in the 
list of stations above. 
