What speakers mean by it now
English piece is the ordinary word for a part taken from a larger whole: a piece of bread, a piece of cloth, a piece of land. It is also the ordinary word for a complete item that belongs to a class, so that a piece of furniture is not a fragment of a chair but one chair among others. From there the noun stretches into specialized units that still feel bounded: a coin, a chessman, a firearm, a stretch of road, a stretch of time, a poem, a song, a play. The shared idea is a unit you can count or hold apart, whether it was broken off or made as a whole.
The origin of the word piece is a borrowing story, not a native English one, and the older layers of that story are only partly settled.
French into Middle English
The English form is first recorded around 1200, in the thirteenth century, as Middle English pece, with later spellings peece and peice. At that date it meant a fixed amount, a measure, a portion. By about 1300 it also meant a fragment of an object, a bit of a whole, a slice of meat. The source was Old French piece, already in use in the twelfth century for a bit, a portion, an item, and a coin. English took the word through the Anglo-French of the Norman world, which is why dictionaries list Anglo-French or Old French as the immediate parent rather than Latin directly.
Behind the French noun stands Late Latin or Vulgar Latin pettia, also written petia and pecia. In that register it named a broken fragment and, in legal and estate language, a parcel of land. Classical Latin did not furnish this word. The usual account is that Latin borrowed it from Gaulish, reconstructed as *pettyā. So the route is Celtic speech in Gaul, then Latin, then Gallo-Romance, then French, then English.
English already had a native noun for much the same idea. Old English styċċe meant a piece or portion and continued in Middle English as sticche or stucche. Piece did not fill an empty slot; it replaced a home-grown word at the center of the vocabulary.
A Celtic noun, then a disputed deeper root
If *pettyā is correctly recovered, its ancestor is Proto-Celtic *kʷezdis, a feminine i-stem meaning a piece, portion, or quota. That reconstruction is useful because it ties together a family of living Celtic words. Welsh peth means a thing. Breton pezh and Cornish peth or pyth keep related senses. Irish and Scottish Gaelic cuid mean a share or portion. The split between p-forms in Brittonic and Gaulish and c-forms in Goidelic is the expected outcome of Proto-Celtic *kʷ, so the comparison is morphologically tidy even though some of the Gaulish evidence is reconstructed rather than carved on stone.
Romance descendants of the Latin loan are the easy cognates for an English reader: French pièce, Spanish pieza, Portuguese peça, Italian pezzo and pezza. They show how widely the Gaulish word, once inside Latin, spread with the Romance languages.
The next step back is where sources disagree. Some etymologies set up a Proto-Indo-European root *kwezd- meaning a part or piece and attach Russian chast, part, to the same root. Other specialists find no solid cognates outside Celtic. Ranko Matasović, in his etymological dictionary of Proto-Celtic, treats *kʷezdis as of unknown origin and possibly taken from a pre-Indo-European substrate. That is not a minor quibble: it means the deepest layer may not be Indo-European at all. The Gaulish source of the Latin word remains the majority view; an Indo-European pedigree for the Celtic word does not.
Senses that grew after the loan
Once the noun was in English, it kept adding units that could be counted as pieces. Early in the fourteenth century it could mean a period of time. Around 1400 it named a separate article forming part of a class or group, and in the same period it named a coin. The coin sense produced piece of eight, the English name, from about 1600, for the Spanish dollar worth eight reales and stamped with the numeral 8.
In the sixteenth century the word moved from leftover to specimen. From the 1530s it could mean a literary composition. From the 1540s it named an artillery weapon, then by the 1580s a portable firearm, still slang for a gun. From the 1560s it meant a chessman, and also an example or instance of something. Fixed phrases cluster in the same centuries: piece of one's mind from the 1570s, in one piece by the 1580s, of a piece by the 1610s. Piece of work for a remarkable person still carries an echo of Hamlet, though the phrase has had a long life of its own.
The verb is built on the noun. Around 1400, pecen meant to mend clothing by sewing extra bits onto it. By the late fifteenth century the same verb meant to join parts, to put something back together, the sense that survives in piece together an account or piece together a broken object.
A related English noun, patch, is usually derived from a dialectal Anglo-Norman form peche of the same French word. The two English items look unrelated today. Historically they are two outcomes of one Gallo-Romance noun: one path gave the general word for a part, the other the word for a scrap used to cover a hole.
The sibling hiding in a mended cloth
The useful takeaway is that kinship. Patch is not a later metaphor spun out of piece; it is a twin that entered English by a slightly different French door. When a garment is patched, the repair is, in the older vocabulary, another piece, the same Celtic-via-Latin noun coming back as a scrap. That is firmer ground than a reconstructed Indo-European root that may never have existed, and it dates the English family to the centuries after the Norman Conquest, when French kept feeding the language of portions, parcels, and repairs.