The supernatural element in tales shows a peculiar degree of variability. In a study of the varying details it appears a number of times that stories which in one region contain fantastic elements are given a much more matter-of-fact setting in others. I take my examples again from the north Pacific coast. In the tale of Raven’s battle with South Wind we find in most cases an incident of an animal flying into the enemy’s stomach, starting a fire, and thus compelling him to cough. In the Tsimshian version he simply starts a smudge in his house. In most tales of the liberation of the Sun the magical birth of Raven plays an important part, but among the Eskimo he invades the house by force or by ordinary fraud. In the Tsimshian tale of the origin of Raven a dead woman’s child flies up to the sky, while the Tlingit tell the same tale without any supernatural element attached to it. Another case of this kind is presented by the wedge test as recorded among the Lower Thompson Indians. In most versions of this tale a boy who is sent into the open crack of a tree and whom his enemy tries to kill by knocking out the spreading-sticks, escapes miraculously when the tree closes. In the more rationalistic form of the tale he finds a hollow which he keeps open by means of supports given me. The available material gives me the impression that the loss of supernatural elements occurs, on the whole, near the border of the area in which the tales are known, so that it might be a concomitant of the fragmentary character of the tales. That loss of supernatural elements occurs under these conditions, appears clearly from the character of the Masset and Tlingit tales recorded by Swanton. In some of the Tlingit tales the supernatural elements are omitted, or weakened by saying that the person who had an incredible experience was out of his head. In the Masset series there are many cases in which the supernatural element is simply omitted. I am not prepared to say in how far this tendency may be due to conflicts between the tales and Christian teaching or in how far it may be due simply to the break with the past. The fact remains that the stories lost part of their supernatural character when they were told in a new environment.
I think it would be wrong to generalize and to assume that such loss of supernatural elements is throughout the fate of tales, for the distribution of explanatory tales shows very clearly that it is counterbalanced by another tendency of tales to take on new supernatural significance.
An additional word on the general theory of mythology. I presume I shall be accused of an entire lack of imagination and of failure to realize the poetic power of the primitive mind if I insist that the attempt to interpret mythology as a direct reflex of the contemplation of nature is not sustained by the facts.
Students of mythology have been accustomed to inquire into the
