at he does come back for more.
We do not refer to the figures of the last two years, which indeed are
higher than those for many preceding years, but to the only averages
that are of much significance in this connection; namely, those for
periods of years going back half a century or more. Five-year averages
back to the Civil War show increasing per capita consumption for
continental United States (see table).

FIVE-YEAR PER CAPITA CONSUMPTION FIGURES

_Five-year Per capita Five-year Per capita
Period Pounds Period Pounds_

1867-71 6.38 1897-1901 10.52
1872-76 7.03 1902-06 11.50
1877-81 7.53 1907-11 10.21
1882-86 9.09 1912-16 10.02
1887-91 8.07 1917-21 11.39
1892-96 8.63

It will be seen that the gain has been a decided one, fairly steady, but
not exactly uniform. In the fifty years, John Doe has not quite come to
the point where he hands up his cup for a second helping and keeps a
meaningful silence. Instead, he stipulates, "Don't fill it quite full;
fill it about five-sixths as full as it was before." That is a
substantial gain, and one that the next fifty years can hardly be
expected to duplicate, in spite of the efforts of our coffee
advertisers, our inventors, and our vigorous importers and roasters.

The most striking feature of this fifty-year growth was the big step
upward in 1897, when the per capita rose two pounds over the year before
and established an average that has been pretty well maintained since.
Something of the sort may have taken place again in 1920, when there was
a three-pound jump over the year before. It will be interesting to see
whether this is merely a jump or a permanent rise; whether our coffee
trade has climbed to a hilltop or a plateau.

In this connection it should be noted that the government's per capita
coffee figures apply only to continental United States, and that in
computing them all the various items of trade of the non-contiguous
possessions (not counting the Philippines, whose statistics are kept
entirely separate from those of the United States proper) are carefully
taken into account.

But for the benefit of students of coffee figures it should be added
that this method does not result in a final figure except for one year
in ten. The reason is that between censuses the population of the
country is determined only by estimates; and these estimates (by the
U.S. Bureau o