In 1914, books accounted for 74% of items circulated from the Cincinnati Public Library. Prints accounted for 13% of circulation, lantern slides, 6%, and music rolls, 3%. The Cincinnati Public Library at this time was one of the largest and most lavish public libraries in the U.S. These statistics indicate the scope of services that a leading public library provided.
Data for U.S. public libraries in 1955 show less format concentration in holdings and greater concentration in circulation. Books comprised an estimated 67% of libraries’ items and 94% of libraries’ circulation. Photos, pictures, and prints, which made up 20% of items, accounted for only 2.2% of circulation. While sound recordings and films accounted for small shares of items and circulation, these formats had relatively rapid turnover in lending (see Table). A film was lent on average 13.3 times per year. A film could be viewed much more quickly (perhaps a half hour for films of this time) than a book could be read, and loan periods for films were probably much shorter than those for books. The ratio of circulation per item suggests considerable interest in viewing films in public libraries’ collections.
|Format||Share of items||Share of circ.||Circ./item|
|Photos, pictures, prints||20%||2.2%||0.2|
|Sound recordings (titles)||1.3%||2.2%||3.4|
and misc. items
|Notes and Sources: see  below|
 Data from Papers and Proceedings of the Berkeley Conference of the American Library Association, July 1915, published in the ALA Bulletin, v. 9. The number of (book) volumes at the end of 1914 was 463,521. The source does not give item counts for the other formats. Total circulation for all formats was 2,164,310. A large number of piano rolls are available digitally here and here.
 Data from U.S. Dept. of Health, Education, and Welfare, Biennial Survey of Education in the United States 1954-56, Chapter 5, Statistic of Public Libraries: 1955-56, Tables 9-13. The number of library systems reporting non-book items was only 26% of the number of systems reporting book volumes. I’ve scaled all reported figures by number of systems reporting. If systems reporting non-book items had larger than average non-book holdings, the non-book figures are over-estimates. Because nearly the same number of systems reported items and circulation, scaling matters little to the circ./item figures. For further analysis, see the underlying data for this table.