So far I have dealt exclusively with Trobriand facts. I have also avoided going beyond what could be immediately documented from our collection of magical data. But obviously there are certain further problems which our theoretical approach raises, and it may be well at least to indicate that I am aware of their existence. The following general considerations on the nature of magical language in its relation to pragmatic speech do not, of course, claim the same degree of ripeness and finality as some of the more limited conclusions established on an exhaustive analysis of our own ethnographic area. But though my argument should not be treated as anything but a suggestive and preliminary statement, I think it better to submit it to prospective field-workers and students of magical facts.