The following is the rule about rhetorical questions in O.O. which is given in most authoritative grammars:

Questions in the indicative in O.R., if they are part of a continuous report of a speech, are put in the infinitive, if they are of the first or third person; in the subjunctive, if they are of the second person, i.e. cur fngio? becomes cur se fugere? cur fugis? becomes cur Me fugeret? and cur fugit? becomes cur ilium fugere?

But this rule is based merely on statistics. It is useful as a rough ruleof-thumb for Latin composition, but as a guide to the interpretation of Latin literature it is worse than useless, because it not only leaves an untidy litter of exceptions to puzzle the learner in many passages of Caesar and Livy, but, like so many grammar-book rules, it obscures the fact that syntactical constructions as well as words and inflexions come into use only as channels for the expression of particular notions in the human mind. The interaction upon one another of habitual methods of expression does cause exceptions to any rule that can be drawn up, and consequently the best accounts, such as are given by Kuhner-Stegmann (ii, pp. 537 ff.) and Riemann, Syntaxe latine (pp. 446–8), are too complicated to be of much use to a schoolboy. Nevertheless, an author's choice of construction is normally determined by the notion which he wishes to express, and it is of the utmost importance that a student should be made to understand that linguistic phenomena are the audible or visible results of a people's effort to think and to express living ideas, and not merely a collection of objective data which can be reduced only to statistical rules.