About Lectern
“Seek ye out of the best books words of wisdom; seek learning, even by study and also by faith.” Doctrine and Covenants 88:118
Lectern is a library of 352 public-domain books for the study of Latter-day Saint and Christian scripture, history and theology: the Journal of Discourses and the first doctrinal works of the Restoration, the Church Fathers in the Schaff edition, the first Loeb Classical Library volumes, the Septuagint and the Greek New Testament, Josephus and the lexicons for reading all of them. Every one of its 206,526 pages can be searched, read, linked and quoted, and a quotation can be checked against the page it cites.
It is part of Restoration Commons, beside Pulpit, which does the same for every General Conference sermon. A lectern is the stand a book is read from; a pulpit is where the sermon is given.
Where the books come from
The shelf began as a cut of a private research collection of about 8,700 volumes. Lectern holds the ones that are in the public domain in the United States. Most are scans of library copies (many carry the stamp of Princeton, Toronto, Harvard or the Internet Archive on their first pages), read here with their own OCR. Others are digital editions set from the printed books, the Schaff Church Fathers among them, with the printed page numbers in their margins. Nothing on these pages was retyped or rewritten by a model.
How public domain was decided
A book published before 1931 is in the public domain in the United States. The term for 1930 books ran out on 1 January 2026. That rule is simple; the hard part is the copy. A 1914 Loeb volume reprinted in 1989 may carry a revised translation, and a 1902 commentary reissued by a modern publisher carries that publisher's copyright on its new front matter.
So the collection's catalogue gave 461 books dated before 1931, and the first twelve pages of each scan were read for anything later. 293 showed no year after 1930 at all. The other 168 were read by a language model (Claude Haiku) against a written rule: reprint lines (“First printed 1914, reprinted 1951”), library stamps and digitisation notices are fine; a copyright notice, an ISBN, a revision or new introduction dated 1931 or later is not; anything unclear is out. It passed 109 as plain reprints and kept out 56 with new matter and 3 it could not decide. A sample of its calls was checked by hand. 7 files that were not scans and 43 second copies of the same volume were dropped.
Every book's record shows the rule, the year, the words from its title page where they could be read, and what the front-matter check found. If you think a book here is still in copyright, open an issue and it will come down while it is looked at.
How a quotation is checked
The check is code, not a model. The quoted words have to be on the cited page, in order, with no words skipped, starting and ending on whole words. Case, punctuation, hyphens at line breaks and old ligatures are forgiven. Wording is not: a quotation one word away from the page fails, and the result shows what the page says instead.
A citation is verified when its words are on the cited page, page only when the page is real and there was nothing to check, and rejected otherwise, with the reason. The check says whether a page contains the words. Whether the page supports what is claimed about it is still a reader's call.
The text is the scan's own OCR, so a misread letter on a page can make a true quotation fail. The page image is always beside the text, so the page itself can settle it.
For an AI assistant
Lectern is also an MCP server with no key and no sign-in (Connect). An assistant connected to it searches the 9 shelves, reads pages, and checks its own quotations before it gives them to you, and every citation it gives is a link that opens the page. The hope is a small proof that these tools can make old books easier to read closely rather than easier to skip.
Not here yet
Books from 1931 to 1963 whose copyright was never renewed are also in the public domain, and many Latter-day Saint books of those years seem to be among them. Each needs a check of the renewal records first. Books still in copyright that are free to read elsewhere (the Church's own sites, BYU ScholarsArchive, journal archives) could be searched and cited here and read at their source; the engine already supports it.
How it is built
Lectern runs on the engine behind Catchword, a page-cited library of the AI safety record: the same citation scheme, quote check, search and MCP tools. The code is on GitHub.