Maintenance
Part of Misquotes and fact checks: the questions that matter
Misquotes and fact checks by decade: what you can search
Misquotes and fact checks by decade: why pre-1931 print is open, why the 1930s to 1970s are hardest, which corpora to use and how to report a gap.
The difficulty of checking a quotation does not rise smoothly with age. It rises, peaks somewhere in the middle of the last century, and then falls again. A line from two hundred years ago is often easier to trace than one from sixty years ago, and the reason has almost nothing to do with how well the paper survived.
This page explains the shape of that curve and what it means for where you spend your effort.
What to take away
- Checkability tracks digitization and rights status, not age.
- The hardest window is recent enough to be in copyright and old enough to be undigitized.
- A gap in the record is a fact about archives, not evidence about a line.
- Say which decade you could not search. That sentence carries real information.
The curve
Three forces set how findable a decade is. Whether the material was digitized at all. Whether the full text can be shown to you or only a snippet. And how well automated reading handled the physical page.
Findability by decade
Period
- Deep past
- Good editions
- Long print era
- Very good
- Middle decades
- Snippets only
- Late analog
- Catalogs
- Digital decades
- Total, undated
Full text
- Deep past
- Old typefaces
- Long print era
- Uneven titles
- Middle decades
- Rights status
- Late analog
- Rights, microfilm
- Digital decades
- No chronology
Obstacle
- Deep past
- Long print era
- Middle decades
- Late analog
- Digital decades
| Period | Full text availability | Dominant obstacle |
|---|---|---|
| Deep past | Good, in scholarly editions | Recognition of old typefaces, and language |
| The long print era | Very good | Uneven title selection |
| The middle decades | Poor, often snippets only | Rights status |
| The late analog decades | Catalogs more than text | Rights, and undigitized microfilm |
| The digital decades | Total, and undated | No trustworthy chronology |
The decades, in US copyright terms
For material published in the United States, the periods in that table map onto decades through copyright law.
- Before 1931open. All works first published in the United States before January 1, 1931, except sound recordings, lost their copyright protection 95 years later, on January 1, 2026. Corpora can show these pages in full, so a check here is a reading job. Works published in 1931 follow on January 1, 2027, so the open window moves forward one year every January.
- 1931 to 1977the middle decades. Under US copyright law, works published or registered before 1978 get at most 95 years from publication, and only if the copyright was renewed in the 28th year. Works that lapsed for non-renewal did not regain protection, but nothing on the page tells a corpus which ones those are. That uncertainty is why the 1930s through the 1970s show up as snippets or nothing.
- 1978 onwardthe late analog and digital decades. Works created from 1978 are generally protected for the author's life plus 70 years, so this print stays closed; what opens it is the move to the web, where text is everywhere but undated.
Decade by decade: one misquote, one resource
Four documented cases show how the decade of a line decides the check.
The 1830s: a famous line, fully on the record
Goethe died in 1832, and his last words are usually given as "Mehr Licht!", "more light!". The earliest known account, by Karl Wilhelm Müller, gives a longer, ordinary request to open the second shutter so that more light comes in, and Thomas Carlyle's letter of 2 July 1832 records another long version. Because all of this sits in old, out-of-copyright print, the check is a reading job. For this era, search HathiTrust Digital Library, a repository of research-library content administered by the University of Michigan that holds material digitized through Google Books and the Internet Archive as well as by libraries themselves.
The 1930s: a poem from a small magazine
The bereavement poem "Do not stand by my grave and weep" first appeared as "Immortality" by Clare Harner in the December 1934 issue of the poetry magazine The Gypsy. A 1930s little magazine is exactly the kind of middle-decades print that corpora show as snippets or not at all, which is how the poem came to be credited to anonymous sources and to the Hopi and Navajo. For newspapers, Chronicling America, the Library of Congress's searchable database of digitized historic US newspapers, is the open route; much of its content is in the public domain.
The 1980s: a wrong author in the catalog era
The same poem's most notable false claimant, Mary Elizabeth Frye, was first wrongly cited as its author in 1983. To test a claim like that, check the copyright record. The Copyright Office published the Catalog of Copyright Entries in print from 1891 through 1978 and on microfiche from 1979 through 1982; registrations and renewals since 1978 are online at the Copyright Office website, while earlier entries are not in the online catalog. A pre-1978 check therefore means the printed volumes, often held at federal depository libraries.
The 2010s: fact-checking moves online
Quote Investigator, started in 2010 by Gregory F. Sullivan under the pseudonym Garson O'Toole, fact-checks the reported origins of widely circulated quotes and publishes the earliest appearances it found. For any line that surfaced on the web, start there, then date the earliest web copy yourself before it disappears.
The middle rows are where quotation checking goes to die. That period produced an enormous volume of exactly the material that manufactures attributions: mass-market magazines, motivational literature, advertising copy, self-help books, newsletters, corporate house organs. Very little of it is searchable in full text, and much of it was never considered worth preserving.
Why the middle is closed
Not because it decayed. Because of rights. Whether a work can be shown in full turns on its copyright status. The US Copyright Office's own account of how long protection lasts makes clear the answer depends on facts about the individual work, not a single date: for works made for hire and anonymous or pseudonymous works, the term is 95 years from first publication or 120 years from creation, whichever is shorter.
Corpora respond to that uncertainty by defaulting to caution: snippet view, or nothing.
So a phrase can be indexed, present, and unreadable. You may learn that it occurs in a book without being able to see the page or the sentence around it, which is precisely the information a quotation check needs.
The other half of the problem is that much of the twentieth century sits on microfilm and has never been scanned.
Microfilm was a preservation success and a discovery disaster. The content survives, but you can read it only one reel at a time, in person.
Digitization programs prioritize by demand and funding, and popular magazines from the middle decades rank low on both.
What that means for a line with no early history
Suppose the line first appears in the mid-1990s and nothing older turns up. Two explanations fit equally well.
- It was written then, or shortly before.
- It circulated for decades in print that no corpus covers, and surfaced when the web did.
There is no way to choose between those from a search alone, and treating the first as established is the single most common overreach in this work. The correct output names the gap: the search covered these corpora, which are thin or closed for these decades, and therefore an earlier printed appearance would probably not have been found.
Our account of what each kind of archive actually holds is the material for writing that sentence properly, and the grading of confidence rather than verdicts is where the resulting position belongs.
Where the frontier is moving
The situation improves, slowly and unevenly. Institutions digitize by grant cycle, and the technical guidelines the National Archives publishes for digitizing holdings give a sense of what a serious program involves and why it is not quick. Meanwhile, digitization of older, rights-clear material continues to be far cheaper and far more attractive to funders than the middle decades.
The practical consequence for anyone maintaining a file of findings is that null results have a shelf life.
- A negative result from five years ago may be wrong now, in either direction.
Keeping null results alive
- Recheck lines that mattered on a schedule
- Record the date of every search
- Note which decade you could not reach
- Treat old negative results as possibly wrong
Spending effort where it pays
Decade shapes strategy more than most people allow.
For anything that reads as old, start with deep print corpora first and expect a longer, duller ancestor.
For anything that smells like the middle decades, in vocabulary or sentiment, expect the search to fail. Then plan a library visit or an interlibrary request, not more desk searching.
For anything recent, act fast: originals are still findable, and the window to date a page closes quickly.
Our material on how collecting practice changed across periods covers the editorial habits that put lines into circulation, which often traces an origin faster than a phrase search.
Where a line belongs to a technical field rather than general circulation, the material on specialist records points to corpora indexed on entirely different principles.
Common questions
Is there a decade where checking is basically hopeless?
Not hopeless, but expensive. The decades that are both in copyright and largely unscanned require physical access, which means a library visit or a document request rather than an afternoon at a screen.
Does the gap mean older attributions are more reliable?
No. It means they are more checkable. Those are different properties, and a checkable attribution is often checkably wrong.
How should the gap be described to a reader?
Plainly, and specifically. Name the decades your corpora do not cover in full text and say that an earlier appearance there would not have been found. Vague hedging tells a reader nothing.
Is it worth paying for access to closed archives?
For a single line, rarely. Public libraries commonly provide subscription archive access with a card, and interlibrary services can retrieve a specific article far more cheaply than a subscription.
