he produced from his pocket whenever he had a moment to spare—­often on the top of an omnibus.  Only when the first draft was complete did he proceed to set the scenes, as it were, and map out the stage-management.  On the other hand, one has heard of playwrights whose first step in setting to work upon a particular act was to construct a complete model of the scene, and people it with manikins to represent the characters.  As a general practice, this is scarcely to be commended.  It is wiser, one fancies, to have the matter of the scene pretty fully roughed-out before details of furniture, properties, and position are arranged.[14] It may happen, indeed, that some natural phenomenon, some property or piece of furniture, is the very pivot of the scene; in which case it must, of course, be posited from the first.  From the very moment of his conceiving the fourth act of Le Tartufe, Moliere must have had clearly in view the table under which Orgon hides; and Sheridan cannot have got very far with the Screen Scene before he had mentally placed the screen.  But even where a great deal turns on some individual object, the detailed arrangements of the scene may in most cases be taken for granted until a late stage in its working out.

One proviso, however, must be made; where any important effect depends upon a given object, or a particular arrangement of the scene, the playwright cannot too soon assure himself that the object comes well within the physical possibilities of the stage, and that the arrangement is optically[15] possible and effective.  Few things, indeed, are quite impossible to the modern stage; but there are many that had much better not be attempted.  It need scarcely be added that the more serious a play is, or aspires to be, the more carefully should the author avoid any such effects as call for the active collaboration of the stage-carpenter, machinist, or electrician.  Even when a mechanical effect can be produced to perfection, the very fact that the audience cannot but admire the ingenuity displayed, and wonder “how it is done,” implies a failure of that single-minded attention to the essence of the matter in hand which the dramatist would strive to beget and maintain.  A small but instructive example of a difficult effect, such as the prudent playwright will do well to avoid, occurs in the third act of Ibsen’s Little Eyolf.  During the greater part of the act, the flag in Allmers’s garden is hoisted to half-mast in token of mourning; until at the end, when he and Rita attain a serener frame of mind, he runs it up to the truck.  Now, from the poetic and symbolic point of view, this flag is all that can be desired; but from the practical point of view it presents grave difficulties.  Nothing is so pitifully ineffective as a flag in a dead calm, drooping nervelessly against the mast; and though, no doubt, by an ingenious arrangement of electric fans, it might be possible to make this flag flutter in the breeze, the very fact of its doing so would