ustries of the country keep
satisfactory accounts, and can answer the questions with some
correctness. But manufacturers are likely to suspect the objects of
the census, and to fear that the information given will be open to the
public or betrayed to competitors. Furthermore, the manufacturing
schedule presupposes some uniformity in the method of accounting among
different companies or lines of business, and this is often lacking.
Another source of error in the manufacturing census of the United
States is that the words of the census law are construed as requiring
an enumeration of the various trades and handicrafts, such as
carpentering. The deficiencies in such returns are gross and
notorious, but the census office feels obliged to seek for them and to
report what it finds, however incomplete or incorrect the results may
be. Even on the population returns certain answers, such as the number
of the divorced or the number unable to read and write, may be open to
question.

The wide range of the American census, and the publication of
uncertain figures, find a justification in the fact that the
development of accurate census work requires a long educational
process in the office, and, above all, in the community. Rough
approximations must always precede accurate measurements; and these
returns, while often inaccurate, are better than nothing, and probably
improve with each decade.

Besides, the breadth of its scope, in which the American census stands
unrivalled, the most important American contribution to census work
has been the application of electricity to the tabulation of the
results, as was first done in 1890. The main difficulties which this
method reduced were two. The production of tables for so enormous a
population as that of the United States through the method of tallying
by hand requires a great number of clerks and a long period of time,
and when complete cannot be verified except by a repetition of the
process. The new method abbreviates the time, since an electric
current can tally almost simultaneously the data, the tallying of
which by hand would be separated by appreciable intervals. The method
also renders comparatively easy the verification of the results of
certain selected parts.

Judged by European standards the cost of the American census is very
great. The following table gives the total and the per capita cost of