Two words in one language descending from a single root but entering along separate historical paths, so both survive with different forms and often different meanings. English frail and fragile both come from Latin fragilis — frail through gradual sound change via Old French and Middle English, fragile as a learned borrowing taken straight from Latin in the fifteenth century.

How can a word's shape date its arrival?

Because a word entering a language early is exposed to every sound change that happens afterwards, and one entering late is not. Guarantee and warranty descend from one Germanic root borrowed into French twice at different times, so the two English forms preserve two snapshots of the same French word at two stages of its own development.

Why is first attestation not the same as age?

Attestation records the earliest surviving written instance, which depends on what happened to survive and what anyone thought worth writing down. Symbol scholarship keeps these strictly apart: the recorded date of a symbol's first appearance is a fact about the archive, not about the symbol.

Can every widely travelled term be traced to an origin?

No. Wanderwörter — tea, orange, camel — have every intermediate borrowing step independently documented and still cannot be assigned a first coiner rather than a further transmitter. The record is rich rather than absent, and the honest account is the relay itself rather than a terminus.

Frail and Fragile

The diagnostic model

You are seeing

Two near-synonyms coexist with slightly different meanings and no one knows why both exist
A term's first attestation is much later than its evident age
Vocabulary from different periods is treated as one vocabulary
A concept is dated by its earliest written record rather than by its form

The mechanism

A term that arrives more than once leaves a differently-shaped copy on each arrival, so the form of a variant records the route and period it came by, independently of when it was first written down.

The older apparatus

Historical linguistics, which distinguishes the date a word was written down from the date it arrived, and reads arrival from shape. Symbol studies has the same problem and dates by attestation alone.

The false friend

Two genuinely unrelated terms that happen to resemble each other. The comparative method rejects lookalikes that do not follow regular correspondence — resemblance alone establishes nothing.

The discriminating test

Where two variants of one term coexist, check whether their forms are consistent with a single arrival. If they are not, they document separate entry events and can be dated separately.

On your own data

Take a domain vocabulary, group terms into doublet sets, and date each variant by form and route rather than by first citation. The layers should correspond to identifiable adoption waves.

English has both frail and fragile. They mean nearly the same thing and neither is a mistake.

Both descend from Latin fragilis. Frail arrived the slow way, through Old French and Middle English, worn down by every sound change that happened to English along the route. Fragile was picked up directly from Latin in the fifteenth century by people reading it, and arrived almost unworn.

One root, two entry events, centuries apart. The difference in form is a record of the difference in path.

Linguists call this a doublet, and the useful property is that it works backwards. You do not need to know when fragile entered English to know it entered late — the shape says so, because a word present in the language earlier would have been ground down by the same processes that produced frail.

Guarantee and warranty are the same trick with an extra turn. Both come from one Germanic root borrowed into French, twice, at different times: once during the Norman period and again later, after the word had continued changing inside French. The two English forms are two photographs of the same French word at two stages of its own life.

The shape is the timestamp.

Why attestation is a different measurement #

The alternative — dating something by its earliest surviving mention — measures the archive rather than the thing.

Symbol scholarship is careful about this and states it directly: a symbol’s attestation date is not its age. What survived depended on what was durable, what was worth recording, and what happened not to be destroyed, none of which correlate with when the symbol came into use.

And the symbol record shows what you get when attestation is the only instrument available. The ouroboros appears with three attested meanings across three periods — one shape, carrying different content each time, with nothing in the record to say whether the meanings evolved, arrived separately, or coexisted. The labyrinth is a stable design assigned different functions by different cultures. The swastika held near-total stability within its traditions for millennia and then ruptured completely inside a single decade.

Each of those is a real finding. None of them is a date, and none can be turned into one by more attestation, because attestation answers a different question.

What symbol studies lacks is the doublet. There is no equivalent of a worn form against an unworn one, because a shape copied is a shape copied — it does not accumulate the wear that tells you how long it has been in circulation. The instrument does not transfer, and the absence is worth naming rather than papering over.

Vocabulary arrives in waves #

A domain’s terminology is not one vocabulary. It is a stack of adoption events, each leaving terms with the shape of their moment.

Finance has visible strata. Terms from marine insurance and the Italian banking houses. Terms from British legal drafting. Terms from mid-century operations research. Terms from the derivatives build-out. Terms from software. Each wave brought its own morphology, its own preference for compounds or acronyms, its own borrowing conventions — and words from different waves sit side by side in a single sentence, treated as one register.

Doublets are visible in it once you look for them, because the same underlying concept keeps arriving twice under different clothes and both survive with slightly different senses. Someone knows which one to use in which room, and nobody can say why there are two.

Reading a vocabulary as layered rather than flat gets you something specific: an approximate date for each term that does not depend on the first time somebody wrote it down. Which matters most for exactly the terms whose written record is worst, because those are the ones that circulated in speech for years before anyone bothered to define them.

Where it stops #

The method has a hard edge, and etymology states it plainly rather than working around it.

Wanderwörter are words that have travelled through enough languages over enough centuries that no amount of documentation identifies who coined them rather than merely passed them on. Tea, orange, camel. Every individual borrowing step is independently traceable. The chain is fully documented at every link. And the origin is still not recoverable, because tracing a relay tells you the route and not the starting point.

The failure is the opposite of the usual one. It is not that the record is thin. It is that the record is complete and the question was malformed.

For anything that has circulated widely enough, the honest account is the relay itself rather than a terminus — which is not a consolation prize. The relay is the object. It tells you who was in contact with whom, in what order, and when, and a coiner’s name would tell you almost none of that.

Frail and fragile have been in the same language for five hundred years without anyone finding it odd.

They are two dated samples of the same word, sitting in the dictionary, waiting for somebody to read them as measurements.