Two Readings for One Character
Most kanji carry at least two on'yomi (Sino-Japanese readings): a go-on (呉音) and a kan-on (漢音). The character 経 is kyō in go-on and kei in kan-on. In practice, お経 (sutra) uses kyō; 経済 (economy) uses kei. The split is systematic, not random.
The Earliest Wave: Korean Monks
Japanese scribes first encountered written Chinese through Baekje and Goguryeo scholars in the 5th–6th centuries. The readings they carried are what we now call go-on — “Wu sounds”, named loosely after the Wu region south of the Yangtze, though that label was only attached centuries later, most likely by champions of kan-on. Go-on is usually thought to reflect the Chinese of the Southern dynasties around Jiankang (modern Nanjing), but no contemporary source confirms it, and how much the Korean route reshaped the readings is unclear. Buddhist sutras, liturgical chants, and temple protocols were encoded in this phonology before Japan had any direct contact with the Chinese mainland.
The Tang Prestige Layer
In the 7th–8th centuries, Japanese embassies (kentōshi) brought back the prestigious Chang'an pronunciation. This kan-on — Han sounds — swept through court vocabulary: 政治 seiji (politics), 文化 bunka (culture), 経済 keizai (economy). But the Buddhist establishment, already centuries old, kept its inherited pronunciation. Monks would not re-learn sutras to match a new secular standard.
The Divide in Practice
The pattern holds across dozens of characters. 業 is gō (karma) in go-on, gyō in kan-on; 極楽 gokuraku (paradise) is all go-on; 道 is the go-on dō in 道路 dōro (road) but the kan-on tō in 神道 Shintō — go-on stuck in plenty of everyday words too. Even temple names follow the bias: 法隆寺 Hōryū-ji, 東大寺 Tōdai-ji — the 寺 is always the go-on ji, never the kan-on shi.