Is it really public domain? Checking an old book before you copy it

Someone emails a library a scanned cookbook and asks whether they can sell it as a print-on-demand reprint. The librarian’s first question is never about the recipes. It is: what year is printed on the title page, and where was it published?

For books published in the United States, that one fact settles most cases. The rest of this guide is about the cases it does not settle.

The rule in one line

In the US, every work published before 1931 is now in the public domain. Each January 1 another year joins it: on Public Domain Day 2026, books first published in 1930 came free, among them Faulkner’s As I Lay Dying and Dashiell Hammett’s The Maltese Falcon. Works published from 1927 onward get 95 years from publication, a term set by the Copyright Term Extension Act of 1998, so next January it will be 1931’s turn.

Every book on this shelf passes easily. The oldest, Hints to Purchasers of Horses, is from 1825; the newest, Every Woman’s Cook Book, from 1922.

There is a second, smaller door. A US book published between 1931 and 1963 had to have its copyright renewed after 28 years, and a large share never were. Those books are public domain too, but proving it means searching the renewal records. The Stanford Copyright Renewal Database covers books, and the US Copyright Office has the full catalogs. That is real research, not a glance at a title page, so treat the 1931–1963 window as “maybe” until you have checked.

Four traps that catch careful people

The edition you are holding is newer than the text

A 1902 text reprinted in 1985 is still a 1902 text. But the 1985 introduction, the footnotes, the modernised measurements and any new drawings belong to whoever made them. Our copy of Practical Cooking and Serving is the 1919 printing; its copyright line lists 1902 and 1912, so even a revision from 1912 is long clear. A modern “annotated” edition would not be.

Translations are separate works

A translation carries its own copyright, counted from when the translation was published. Tolstoy has been public domain for a century; a 1990s translation of Tolstoy is not. Check the translator’s date, not the author’s.

Your country may count differently

The US counts from publication for older works. The UK, the European Union and many other places count from the author’s death, usually plus 70 years. Frederick Frye Rockwell published Save It for Winter in 1918 and lived until 1976, so in a life-plus-70 country his text stays protected until the end of 2046. Same book, different answer.

Never published is not the same as old

A handwritten family recipe book from 1890 that was never published falls under different rules: in the US, generally the author’s life plus 70 years. Since most diarists are long dead, many such manuscripts are free too, but the reasoning is different and anonymous ones have their own term.

Where to get the whole book, legally and free

Project Gutenberg has been typing up public-domain books since 1971 and now holds more than 70,000 ebooks. Its texts are proofread by volunteers and checked for US status before release. If you are outside the US, Gutenberg itself tells you to check your local law before downloading.

Standard Ebooks takes public-domain texts, many of them from Gutenberg, and produces carefully formatted, consistently typeset editions. The selection is smaller and leans towards literature; you will not find many cookbooks there.

The Internet Archive holds page-by-page scans of millions of books, many from university libraries. For old practical books it is often the only place a title survives online, and most book pages on this site link to a scan there. Note that the Archive also lends in-copyright books one reader at a time; a 2024 appeals court ruling in Hachette v. Internet Archive went against that lending. The public-domain scans, which you can download outright, are a different category.

HathiTrust is a partnership of research libraries. Its “full view” books can be read and downloaded; some full-view titles are limited to readers in the US, for the death-date reason above.

The Library of Congress puts digitised books, pamphlets and government bulletins in its online collections. For wartime food bulletins and seed catalogs it is hard to beat.

What about the scan itself?

People worry that a library owns the photograph of a public-domain page. In the US, a federal court in Bridgeman v. Corel (1999) held that an exact photographic copy of a public-domain two-dimensional work adds no new originality and gets no copyright. Libraries may still ask you to credit them, and their website terms are a contract question, not a copyright one. Crediting the source costs nothing and helps the next reader find it, so do it anyway.

A five-step check before you copy

1. Find the first publication year, not the reprint year. Title page first, then the copyright notice on the back of it.
2. Before 1931 and published in the US? You are almost certainly clear for the original text.
3. Strip out anything newer: introductions, notes, new illustrations, translations.
4. Outside the US, look up the author’s death year and your country’s term.
5. Work from a scan of an original printing where you can, and credit the library that made it.

None of this is legal advice for a specific project, and a publisher planning a print run should ask a lawyer. For reading, quoting, teaching and cooking from a book like My Best 250 Recipes, the title page is usually all the lawyer you need.