Abstract
This paper analyzes data on the titles of Japanese books, gathered by means of National Diet Library Search. The books fall into two categories: ordinary books and bestsellers. As for the former, 2082 samples have been taken from eight areas in Nippon Decimal Classification for the years from 2003 through 2012. As for the latter, 400 titles have been obtained for the years from 1973 through 1982 and from 2003 through 2012. The objects and methods of analysis are common to both categories: we have separated main titles and subtitles, and for each of them we have examined the frequencies of several kinds of characters used, the linguistic structures of the titles, and the modalities in the titles. Further, we have considered the effects of the areas on the statistics regarding ordinary books and those of the periods on the statistics regarding bestsellers.
Of statistical significance are the following: as for ordinary books, there are effects of the areas on the length of titles, the frequencies of the kinds of characters used, the linguistic structures of the titles, and the modalities in the titles, both for main titles and subtitles. As for bestsellers, there are effects of the periods on the length of titles and the modalities in the titles for main titles, and those on the frequencies of the kinds of characters used for subtitles.
Lastly, taking wakaru as an example, we argue that different areas may have different reasons for using a particular expression as part of a book title.
https://ror.org/02evnh647 
