Quotes by era source check: how to bound a line

You will rarely find the moment a line was first said. You can very often bound it: prove it existed by one date and did not exist in this form before another. A bracket is a real result, it is defensible, and it answers most of the questions people actually have about a notable quotation.

This page is about building that bracket. Quotes by era source check means two things in practice: dating the wording from what a line contains, and dating the attribution from where it appears, each with a named source behind it.

What to take away

Two bounds

The upper bound is the easy one. Any dated appearance of the line proves it existed by then, and the earliest such appearance you locate is your upper bound. Push it earlier by searching harder.

The lower bound sits inside the sentence: something in it could not exist before a certain point. Archaeologists call this a terminus post quem, the point after which something must have been made. The logic transfers.

If a line names a technology, an institution, or a word that came later than a supposed author, the line is later too.

Between those two you have a bracket. It is often decades wide, and a decades-wide bracket is frequently enough to refute a specific attribution outright.

Building the lower bound

Look for anything in the sentence that carries a date.

Any one of these gives a date after which the line must have been written, and the strongest are referents and metaphor sources, because they are hard to argue away. A line that names a telephone, a car, or a computer cannot be older than the object. For a word or a sense, the dated entry in the Oxford English Dictionary is the reference to cite.

Pushing the upper bound down

This is search work, and it is the part that takes time.

Coverage decides how far you can push, so choose the corpus to fit the period. Google Books and HathiTrust search scanned books and let you sort results by date; the Internet Archive adds books, periodicals, and ephemera; Chronicling America, the Library of Congress's digitized newspaper collection, covers the American press. The Library of Congress guide to that collection is a good model of a project stating its own limits.

JSTOR reaches scholarly journals, and Early English Books Online covers English printing before 1700. Coverage sets how far back the upper bound can move: a search can only find appearances the corpus actually holds.

Two brackets, not one

Bound the wording and the attribution separately, because the gap between them is the finding.

When the attribution bracket opens long after the wording bracket, you have the classic pattern: a line circulating anonymously for years, then acquiring a famous name. That is close to decisive evidence against the attribution, and it does not require finding the true author. The cause is familiar: credit drifts toward the best-known name in the field, and our notes on misquotes and fact checks collect case after case.

Writing the result

State both bounds and what each rests on. Something like: this wording is attested by a given year in a named source, and cannot predate a given point because of a specific referent; the attribution first appears decades later.

Three properties make that a good output. It is checkable, because both bounds cite something. It survives being improved, since a later researcher who pushes the upper bound down has refined your result rather than refuted it. And it does not overclaim, because it says nothing about who wrote the line.

The result is a label rather than a verdict: earliest known appearance, with the corpus named. That is the convention our notes on quote context and sources follow. And where the bracket shows a saying older than any name attached to it, the right frame for reporting it is a saying with a history, as in our pages on sayings with long histories.

When the bracket is useless

Sometimes it is very wide and tells you nothing: nothing in the sentence dates it, and the earliest appearance is recent. That combination is common for short general observations, which are exactly the lines that circulate most.

In that case say so plainly. A wide bracket is an honest description of thin evidence, and it still rules out the confident caption you started with. Our account of jokes and sayings that leave almost no printed trace covers why some kinds of material are so thinly attested.

Common questions

Is internal dating not just guesswork?

Not when it rests on a referent. A sentence mentioning a thing that did not exist yet is decisively later than that thing. Vocabulary arguments are weaker and should be treated as suggestive rather than settled.

How wide is a useful bracket?

Any bracket that excludes the claimed author is useful, however wide. A century-wide bracket that begins after someone's death has answered the question.

What if the referent could be a later editor's change?

A fair objection, and the reason to check whether earlier versions of the line lack it. If the dating referent is present in the earliest appearances, the objection fails.

Can the bounds ever meet?

Occasionally, where a work with a known publication date is located and contains the line. Then you are no longer bounding, you have a source, which is the best outcome available.