For the purposes of this research there is, however, a great deficiency of reliable information. In the descriptions of dances and pantomimes which can be gathered from the literature of travel the most important point is generally omitted—whether the performances in question were executed in the presence of the other sex or not. From the general records of savage life one may indeed draw some conclusions bearing upon this question. But these conclusions, even when, taken apart, they seem to be sufficiently reliable, are, when considered collectively, full of contradiction. With regard to a great number of tribes the strict separation between the sexes which prevails in all phases of social life—in work as well as in amusements—provides a negative argument against the theory of female influence on art. This circumstance has of course been adduced by opponents of the selection theory; and it was Mr. Gurney’s chief objection to Wagner’s assertion that the earliest popular dances were erotic pantomimes.[347] But the adherents of Darwin’s theory are no less able to support their assertion by ethnological facts. This is made obvious, for instance, by Wallaschek’s _Primitive Music_, which work, as far as we know, gives the most complete account existent of primitive dances and pantomimes. After having enumerated a great variety of tribes in which the sexes are separated when dancing, the author says, “As a rule, however, both sexes dance together.” And he then corroborates this statement by a still greater number of instances in which dances, often of a decidedly erotic character, are performed either by both sexes together or by one sex before the other.[348] Thus facts confront facts in a most bewildering contradiction which is indeed trying to the believer in ethnological argumentation. No sound statistician would base any conclusions on majorities made up of such confused masses of contradictory instances. The question of sex influence on art would therefore be for ever undecided if there were no expedient for classifying the units and balancing them against each other in greater groups. Such a classification is of course no less possible than it is necessary. But it presupposes a close examination of every single instance quoted. Owing to our insufficient knowledge of some of the most interesting tribes, a satisfactory arrangement is therefore as yet involved in great difficulty. In this work, at any rate, it would not be possible to enter into all these detailed researches. We shall restrict ourselves to a few short indications of the mode in which we think that this much-debated question is to be solved.