Résumé :
Four manuscripts of Achilles Tatius break off at i 10.3
τοῖς ῥήμασιν
, and a fifth originally did so. All five begin with Longus, where they descend from Vat. gr. 1348. This manuscript too has Achilles Tatius after Longus, but complete, so that they may seem unlikely to descend from it in Achilles Tatius. According to the latest editor, E. Vilborg (Gothenburg 1955), they do not descend from it: they avoid three of its errors (p. lxii, lxvi). Expectation and evidence coincide, then, and that should be that.
Every fragmentary manuscript, however, goes back eventually to a complete one, and whether three errors will bear the weight of a stemma depends on others.