Quotes by Era: Six Collecting Methods and Their Errors

Sorting quotes by era — by the speaker's century — is the obvious way to organize them, but not interesting. What changes across periods is not the material, since people have always said memorable things, but the technology of collecting.

Each era had a dominant way to gather and reprint quotations. Each method failed in its own characteristic way, and those failures still circulate.

Learn the error signature of each period and you gain something practical: given a suspect line, you can often tell roughly when it went wrong, which tells you where to look.

What to take away

What to take away

Six ways of collecting, and what each one broke

The extract book. The oldest method, and the ancestor of the florilegium: a scholar copies striking passages out of works into a collection organized by topic, for reuse. The extracts are separated from their contexts by design, the whole point is to make them portable. Attribution attaches to a small pool of prestigious names, because those are the authors being read. Anything anonymous in the source tends to acquire the nearest famous name in the collection. Error signature: a maxim ascribed to a major classical figure with no work named, and no ancient text that contains it. A name placed after a sentence makes two claims at once, and this method is where they get fused.

The commonplace book. The personal version, kept by readers for their own use. Entries were copied from memory as often as from the page, headings were the reader's own, and sources were noted inconsistently or not at all. These books were later mined by editors and printed. Error signature: wording shaped by memory rather than by a text, and a topical heading that imposes a meaning the original passage did not carry. What they were mostly full of was sayings with no author attached, which is exactly why they were so easy to mine.

The printed collection of adages. Early print produced large thematic compilations of sayings ascribed to classical authorities, frequently in the compiler's own Latin. When those were later translated into vernacular languages, the compiler's phrasing was translated as though it were the original author's. Error signature: an ancient name attached to phrasing whose real ancestor is a much later compiler.

The mass-market anthology of familiar quotations. The nineteenth-century breakthrough, and the largest single multiplier of error in the whole history of the subject. These volumes were cheap, indexed by keyword, and enormously popular; Bartlett's Familiar Quotations is the best-known volume in this line. Many entries carried a name with no work, edition, or page. Later collections were assembled largely from earlier collections, so an unsourced ascription entered the record once and was then copied forward for generations, gaining authority purely by repetition. Error signature: an attribution that every reference agrees on and no reference cites.

The newspaper short item. Papers filled space with brief paragraphs, epigrams, and jokes, drawn from exchange copy and syndicated matter. Short items travel well and citations do not travel with them, so a line could circulate through hundreds of papers in a season, picking up and shedding names as it went. This is where a great deal of the comic attribution that still circulates entered print. Error signature: many early appearances, all within a short window, with inconsistent or absent attribution, and a famous name that appears only in the later half of the run.

The image card. The current method. A line, a portrait, a typeface, no date, no source, and an engagement mechanism that selects for whichever wording performs best rather than whichever is correct. It also does something none of its predecessors did: it puts a face next to the words, so the attribution is asserted visually and cannot be hedged. Error signature: no source at all, a portrait doing the work of a citation, and wording that differs from the version on the last card you saw.

The one period that pushed back

Between the anthology era and the card era, a different kind of book appeared: the dictionary of quotations with real apparatus. The Oxford Dictionary of Quotations is the standard example, and it includes editions, pages, dates, cross-references, and explicit labels that separate a sourced entry from an ascribed one.

The vocabulary of "attributed to" and "widely misattributed" comes from there. It exists because editors doing the work found much of the inherited record could not be verified.

Full-text search added what no earlier method had: finding a line's earliest appearance, not its most common one. That reversal is the field's most important change.

An ordinary reader can now find that a quotation appears everywhere after a certain year and nowhere before, evidence that settles most disputed attributions.

Search has its own failure mode. A digitized corpus shows what was scanned and correctly recognized. Text recognition on old print is imperfect, so a line can be in a book yet invisible to search.

Confident claims that a phrase "does not appear before" a date are only as good as the corpus behind them. The honest form of that claim names the corpus — Google Books, HathiTrust, or the Internet Archive — and says how it was searched.

Reading the signature

Old is not the same as reliable

The instinct that an older source is a better source is half right. An old appearance is genuinely valuable for dating: it proves the line was circulating by then, and dating is most of what you can establish about a saying.

But age says nothing about the attribution. A nineteenth-century collection carrying a name with no citation is exactly as unevidenced as a caption carrying the same name, and it is more dangerous, because it looks scholarly.

The question to ask of any old source is not how old it is but what it cites. A book from two centuries ago that names an edition and a page is a good source. One that names only a person is a record of what somebody believed.

What this means when you use older collections

Scanned historical material, of the kind described in the Library of Congress newspaper research guide, is what makes this practical for anyone now, and it is worth being explicit about which archive you searched when you report a result.

Common questions

Does era tell me anything about whether a quotation is genuine?

Not directly, but it tells you which failure to test for first. A line ascribed to an ancient figure is most often an extract-collection or compilation artifact; a line ascribed to a Victorian figure is most often an anthology artifact; a line ascribed to someone recent is most often a caption artifact. That ordering saves time.

Why do the same few names keep appearing across all these periods?

Every collecting method concentrates material on names its audience recognizes, and those names accumulate across methods. Each era hands its ascriptions to the next, and the drift runs only toward greater fame.

Fields where a real corpus exists to check against are the ones where the machinery can actually be undone. This page is the account of the machinery that carried the errors.

Should I distrust everything in an old anthology?

No: distrust the unsourced entries in it, which is a different and much smaller instruction. Plenty of entries carry real citations. The habit worth building is checking which kind of entry you are reading before you copy it.