Nevertheless it must be remembered that the individual spirit-child, which sometimes has even to follow the married couple on their wanderings, chooses its mother on account of her husband and not in all probability on her own; for it is not of her totem, and it is improbable that the natives assume ties of preference between two beings of different clans, if there are at hand two members of the same clan--the father and the reincarnated child. Now this act of choosing, this special preference of a certain woman on account of her husband, clearly points to a very close tie between father and child. Unfortunately, the writers who report the beliefs in question have not investigated the side we have discussed, and as all hypothetical inferences are dangerous in sociology, we must consider this belief to be highly suggestive but nothing more. Nevertheless, setting one against another the two facts--the social existence of a close tie between father and child on the one hand (as we can affirm it on the ground of the emotional character of this relationship), and the existence of a belief that the reincarnated spirit-child is of the father's totem, and is, so to say, attached to him in his roaming life--it is difficult not to suspect some inner connection between them. Now, if our supposition is right, and if this belief has its social influence in defining fatherhood, it may be said that in the Gnanji and Umbaia tribes the essence of fatherhood is seen in the fact that a given man has determined a given spirit-child to take up its abode in his wife's body, and that the close tie of kinship lies in this mutual affinity or attraction exercised by the man on the spirit-child. This is hypothetical, but we may note another statement of Spencer and Gillen's which appears to bear upon our subject and corroborates our first hypothetical assumption.