The most interesting portions of the _Lives of the Novelists_ are those which show us, by the frequent recurrence of the same subjects, what parts of the theory of novel-writing had particularly engaged Scott's attention. For example we find him discussing, most fully in the _Life of Fielding_, the reasons why a successful novelist is likely not to be a successful playwright. The way in which he looks at the matter suggests that he was thinking quite as much of the probability of failure in his own case should he begin to write plays, as of the subject of the memoir; for Fielding wrote his plays before his novels, but the argument assumes a man who writes good novels first and bad plays afterwards. One of his statements seems rather curious and hard to explain,--"Though a good acting play may be made by selecting a plot and characters from a novel, yet scarce any effort of genius could render a play into a narrative romance." Perhaps he expected the "Terryfied" versions of _Guy Mannering_ and _Rob Roy_ to hold the stage longer than fate has permitted them to do. From another point of view also he was interested in the connection of the novel and the drama. He felt that the direction of the drama in the modern period had been largely determined by the influence of successful novels; and he probably overestimated the effect of the "romances of Calprenede and Scuderi" on heroic tragedy.[211] A subject which recurs even oftener than that of the distinction between drama and novel is the question of supernatural machinery in novels. Horace Walpole is commended for giving us ghosts without furnishing explanations. Indeed the _Castle of Otranto_ is highly praised;[212] but so also is Mrs. Radcliffe's work, except on the one point of the attempt to rationalize mysteries. The kind of romance which she "introduced"[213] is compared with the melodrama, and its particular mode of appeal is analyzed in very interesting fashion. In the _Life of Clara Reeve_ the proper treatment of ghosts is discussed at length, for that author had contended that ghosts should be very mild and of "sober demeanour." Scott justifies her practice, but not her theory, on the following grounds: "What are the limits to be placed to the reader's credulity, when those of common-sense and ordinary nature are at once exceeded? The question admits only one answer, namely, that the author himself, being in fact the magician, shall evoke no spirits whom he is not capable of endowing with manners and language corresponding to their supernatural character."