This thesis is concerned with the problem of drawing inferences about populations when little or nothing is known about the form of the underlying distribution. Chapter 1 reviews the general problems, but the thesis as a whole is more concerned with problems of estimation than with tests of hypotheses.