manner, as will be seen with regard to the
conditions of economy and of accessibility.
The three conditions are in truth all
connected together, but especially the two last
named.

Even in a paper such as this the question
of classification cannot altogether be
overlooked; but it is one more easy to open than
to close -- one upon which I am not bold
enough to hope for uniformity of opinion and
of practice. I set aside on the one hand the
case of great public libraries, which I leave
to the experts of those establishments. And,
at the other end of the scale, in small private
libraries the matter becomes easy or even
insignificant. In libraries of the medium scale,
not too vast for some amount of personal
survey, some would multiply subdivision, and
some restrain it. An acute friend asks me
under what and how many general headings
subjects should be classified in a library
intended for practical use and reading, and
boldly answers by suggesting five classes
only: (1) science, (2) speculation, (3) art,
(4) history, and (5) miscellaneous and
periodical literature. But this seemingly simple
division at once raises questions both of
practical and of theoretic difficulty. As to the
last, periodical literature is fast attaining to