Athens became the imaginary scene of comedies, and the relationships which had become a fixed rule for them were transferred to tragedy also, and soon to other theaters whose setting bore little or no resemblance to that of the theater of Dionysus Eleuthereus. Certainly by the time of New Comedy the convention was firmly established, and except for characters leaving or entering the houses in the background almost every exit or entrance was oriented for the audience with reference to country, harbor, or market place. When Greek comedy was transplanted to Italian soil the convention was taken over, too, and reappears, possibly with some modification, in the plays of Plautus and Terence. Regardless of the convention, however, and the period of its origin there is one blemish which careful stage managers nowadays seek to avoid. When a door closes upon a departing character, it should not be immediately opened again to admit another character, whom the first character must have brushed against in the hall. A slight pause is somehow provided to enable the two characters to avoid meeting and to give the sense of space beyond the room on the stage. Now in Euripides’ _Alcestis_ a violation of this common-sense principle of stage craft seems to occur. At vs. 747 Admetus and the chorus have departed bearing the body of Alcestis to its last resting place. In the ensuing scene Heracles at last learns the identity of the deceased and at vs. 860 rushes out to wrestle with the king of the dead beside the grave. _In the very next verse_ Admetus returns. According to the Hellenistic convention Heracles must have departed and Admetus have re-entered through the parodus at the audience’s left. But which parodus was employed does not in this case greatly matter. The point is that since Heracles was bound for the spot from which Admetus was returning, they must have used _the same parodus_. Nevertheless, later developments show that they did not meet; indeed, certain telling features of the dénouement would have been spoiled if they had. Yet how could they avoid doing so? The play furnishes no reply. So far as I can see the only way in which the difficulty can be obviated is by supposing that vss. 747-860 take place before a slightly different part of the palace from the rest of the play. Scholars, however, do not commonly accept a change of scene in this piece (see pp. 250 f., below).