"Candy" began as a word for a broken-off lump. The sweetness came with it, but the original idea was a piece.

The trail starts in Sanskrit. Wiktionary gives khaṇḍa as "piece, fragment, candied sugar, dried molasses." It comes from a root meaning "to divide, break into pieces." Crystallized sugar came in chunks, and the word for a chunk became a word for the sugar.

Andy Hollandbeck of the Saturday Evening Post traces the next steps. Khanda passed into Persian as qand, cane sugar. Arabic made it qandi. Merriam-Webster glosses the Arabic qandī as "candied," from qand, "crystallized sugar."

Here the story picks up a partner. Sanskrit had a second sugar word, sharkara, which originally meant "grit, gravel." It became Persian shakar, Arabic sukkar and Medieval Latin succarum. In Old French the two lines met in a single phrase, çucre candi. That meant lumps of rock sugar eaten as a treat.

Hollandbeck says both candi and çucre had reached English by the end of the 1200s. Candi was respelled candy, and çucre became sugar. Merriam-Webster describes Middle English sugre candy as a partial translation of Middle French sucre candi. Its earliest record of the noun "candy" on its own is 1587, still meaning crystallized sugar.

So a bag of candy holds two Sanskrit words that took different roads west. One meant a broken piece. The other meant gravel. Both pictured sugar as something hard and lumpy.

The meaning widened slowly. Dictionary.com dates Middle English candi, "candied sugar," to 1225 to 1275. The verb still carries the old sense: to candy fruit, peel or ginger is to cook it in heavy syrup until it turns transparent. The noun now covers any sugar confection, often with chocolate, fruit or nuts. Wiktionary also notes a possible alternative origin for khaṇḍa in Proto-Dravidian, so the Sanskrit word may itself be a borrowing.