Back in the first half of the 20th century getting a copyright required filing for it with the U.S. Library of Congress. A simple form filed would result in a 28-year copyright term. And, at the end of that term you could file again for a second 28-year term.
But in practice very few books or stories were worth the time filling out the renewal form, especially those short stories published in magazines and various pulp-fiction titles. So, often, neither publishers nor authors did.
The only way to discover this, until very recently, was an even more onerous task of researching in multiple references in person at the Library of Congress to figure out the original registration, and then extension filing, and then the new registration.
And who would pay to digitize this since the only thing it is good for is to prove a text is not copyrighted?
Well, it turns out the New York Public Library would, and did.
And, roughly, 80% of the books published between 1924 and 1964 are now in the Public Domain[en.WP]. Of course the cool bit is that Hathi Trust[O] is already busy scanning these texts. And, for fun, there is a bot equally busily posting about unrenewed texts which have been scanned in over on the Mastodon network. The bot’s name is Secretly Public Domain.
The owner of the bot also has a github repository with python scripts for checking the Library of Congress’s Catalog of Copyright Entries (CCE), imaginatively titled cce-python. Xe also wrote the news brief which inspired this post to increase awareness of the database(s).