nterior, and in all
plans for the execution of the law the head of the office was
responsible for success. The law divided the subjects of census
inquiry into two parts--first, those of primary importance, requiring
the aid of the enumerator; and, secondly, those of subsidiary
importance, capable of production without the aid of the enumerator.
The former had to be finished and published by 1st July 1902; the
latter were not to be undertaken until the former were well advanced
towards completion. By this means the attention of the office could be
concentrated on a small number of subjects rather than distributed
over the long list treated in the volumes of the tenth and eleventh
censuses.

Under the federal form of government, with its delegation of all
residuary powers to the several states, the United States have no
system of recording deaths, births and marriages. Hence there is no
such basis as exists in nearly every other civilized state for a
national system of registration, and the country depends upon the
crude method of enumerators' returns for its information on vital
statistics, except in the states and cities which have established a
trustworthy registration system of their own. These are the New
England states and a few others in their vicinity or influenced by
their example. Enumerators' returns in this field are so incomplete
that hardly two-thirds of the deaths which have occurred in any
community during the preceding year are obtained by an enumerator
visiting the families, no satisfactory basis for the computation of
death-rates is afforded, and the returns have comparatively little
scientific value. In the regions where census tables and
interpretations are derived from registration records kept by the
several states or cities they are often made more complete than those
in the state or municipal documents. The census of agriculture is also
liable to a wide margin of error, owing to defects in farm accounts
and the inability of many farmers to state the amount or the value
even of the leading crops. The census figures relate to the calendar
year preceding 1st June 1900, and hurried and careless answers about
the preceding year's crop are almost sure to have been given by many
farmers in the midst of the summer's work.

The difficulties facing the manufacturing census were of a different
character. A large proportion of the ind