Whilst of some tribes (for example the Indians of Guiana) it is said that there is nothing to indicate that they “know of any spirits, except such as are, or once were, situated in material bodies,”[198] amongst others we are often told that not only ghosts are known but also spirits that are declared never to have been incarnate. An extreme form of the ghost-theory maintains that all spirits were once ghosts whose incarnation has been forgotten; but this is needless, and seems not to be true. It is enough that probably the original inhabitants of the spirit-world were ghosts; that some of those now believed not to have been ghosts were once really so; and that those spirits that were never ghosts are later immigrants, who have obtained domicile by having been imagined in analogy with ghosts. The following list indicates more or less probable reasons why (_A_) ghosts have sometimes come to be regarded as non-human spirits, and (_B_) why certain non-human things have come to be regarded as spirits, or as possessed by spirits, more or less resembling the human. _A._ Spirits that were formerly ghosts, but are now declared not to have been: (_a_) Ghosts whose former life has been forgotten by mere lapse of time. The memory of the dead amongst many tribes does not extend beyond three or four generations. If then the ghost of some unusually impressive personality happens to be remembered, when all his relatives and contemporaries have been forgotten, he seems to be separated from the human race. And if his name was that of some natural object, his ghost, according to Spencer’s hypothesis, may now be regarded as the spirit of that phenomenon. But as to Spencer’s hypothesis,[199] although it gives such a plausible explanation of much nature-worship by real facts as to the working of savage language and thought that it seems to me unreasonable to doubt that it has had some of the effects he traces to it, yet it presses upon me more and more that most cases of nature-worship are to be explained by more particular causes.