As a subject of study statistics differs from other subjects in being non-existent by itself. A “statistic” is a measure—e.g. mean, variance, standard deviation, median—of a sample, and therefore of a sample of “something”, and that original “something” pertains to some other “subject”—e.g. biology, agriculture, medicine, industry, economics. Or, if we investigate or research in the theory on which our techniques are based, we are at once plunged into mathematics. This somewhat nebulous quality of the subject, together with the peculiarly subtle mixture of down-to-earth algebra and logic on the one hand and subjective intuition and philosophical approach on the other, all wrapped together in a sort of cotton-wool of probability, is often for the beginner a real difficulty and thorn-in-the-flesh, though largely contributing later, it is hoped, to the fascination of the subject. I have found in over ten years of teaching statistics at Sheffield University that the student can be much helped over this initial period by simple experiments in which he collects his own data, in the course of which he builds up an idea for himself of the underlying probability situations and in some of which he learns something of the relevance of design and the practical difficulties associated with the carrying out of such simple-sounding instructions as “randomise.” When it comes to analysing his data he experiences a certain “vested interest” in the results just because they are his own observations.