Some years ago when I. assembled a number of general articles and lectures on probability and statistics, their publication (Essays in Probability and Statistics, Methuen, London, 1962) received a some­ what better reception than I had been led to expect of such a miscellany. I am consequently tempted to risk publishing this second collection, the title I have given it (taken from the first lecture) seeming to me to indicate a coherence in my articles which my publishers might otherwise be inclined to query. As in the first collection, the articles are reprinted chronologically, usually without comment. One exception is the third, not previously published and differing from the original spoken version both slightly where indicated in the text and by the addition of an Appendix. I apologize for the inevitable limitations due to date, and also for any occasional repetition of the discussion (e.g. on Bayesian methods in statistical inference). In particular, readers technically interested in the classification and use of nearest-neighbour models, a topic raised in Appendix II of the fourth article, should also refer to my monograph The Statistical Analysis of Spatial Pattern (Chapman and Hall, London, 1976), where a much more up-to-date account of these models will be found, and, incidentally, a further emphasis, if one is needed, of the common statistical theory of physics and biology. March 1975 M.S.B.

In a contribution (Bartlett, 1971 a) to the Symposium on Statistical Ecology at Yale in 1969, I noted in my introductory remarks that that paper was not intended to be in any way a review of statistical techniques for analysing spatial patterns. My contribution to a conference at Sheffield in 1973 aimed, at least in part, to supply such a review and forms the basis of this monograph; but in these prefatory remarks I must still make clear what I decided to discuss, and what I have omitted. Broadly speaking, the coverage is that included in seminars and lectures I have given on this theme since 1969. We may divide problems of spatial pattern (in contrast with complete random chaos) into (i) detecting departures from randomness, Oi) analysing such departures when detected, for example, in relation to some stochastic model and (iii) special problems which require separate consideration; for example, sophisticated problems of pattern recognition in specific fields, such as the computer reading of handwriting or recognition of chromosomes.

In a contribution (Bartlett, 1971 a) to the Symposium on Statistical Ecology at Yale in 1969, I noted in my introductory remarks that that paper was not intended to be in any way a review of statistical techniques for analysing spatial patterns. My contribution to a conference at Sheffield in 1973 aimed, at least in part, to supply such a review and forms the basis of this monograph; but in these prefatory remarks I must still make clear what I decided to discuss, and what I have omitted. Broadly speaking, the coverage is that included in seminars and lectures I have given on this theme since 1969. We may divide problems of spatial pattern (in contrast with complete random chaos) into (i) detecting departures from randomness, Oi) analysing such departures when detected, for example, in relation to some stochastic model and (iii) special problems which require separate consideration; for example, sophisticated problems of pattern recognition in specific fields, such as the computer reading of handwriting or recognition of chromosomes.

Some years ago when I. assembled a number of general articles and lectures on probability and statistics, their publication (Essays in Probability and Statistics, Methuen, London, 1962) received a some­ what better reception than I had been led to expect of such a miscellany. I am consequently tempted to risk publishing this second collection, the title I have given it (taken from the first lecture) seeming to me to indicate a coherence in my articles which my publishers might otherwise be inclined to query. As in the first collection, the articles are reprinted chronologically, usually without comment. One exception is the third, not previously published and differing from the original spoken version both slightly where indicated in the text and by the addition of an Appendix. I apologize for the inevitable limitations due to date, and also for any occasional repetition of the discussion (e.g. on Bayesian methods in statistical inference). In particular, readers technically interested in the classification and use of nearest-neighbour models, a topic raised in Appendix II of the fourth article, should also refer to my monograph The Statistical Analysis of Spatial Pattern (Chapman and Hall, London, 1976), where a much more up-to-date account of these models will be found, and, incidentally, a further emphasis, if one is needed, of the common statistical theory of physics and biology. March 1975 M.S.B.

In a contribution (Bartlett, 1971 a) to the Symposium on Statistical Ecology at Yale in 1969, I noted in my introductory remarks that that paper was not intended to be in any way a review of statistical techniques for analysing spatial patterns. My contribution to a conference at Sheffield in 1973 aimed, at least in part, to supply such a review and forms the basis of this monograph; but in these prefatory remarks I must still make clear what I decided to discuss, and what I have omitted. Broadly speaking, the coverage is that included in seminars and lectures I have given on this theme since 1969. We may divide problems of spatial pattern (in contrast with complete random chaos) into (i) detecting departures from randomness, Oi) analysing such departures when detected, for example, in relation to some stochastic model and (iii) special problems which require separate consideration; for example, sophisticated problems of pattern recognition in specific fields, such as the computer reading of handwriting or recognition of chromosomes.

In a contribution (Bartlett, 1971 a) to the Symposium on Statistical Ecology at Yale in 1969, I noted in my introductory remarks that that paper was not intended to be in any way a review of statistical techniques for analysing spatial patterns. My contribution to a conference at Sheffield in 1973 aimed, at least in part, to supply such a review and forms the basis of this monograph; but in these prefatory remarks I must still make clear what I decided to discuss, and what I have omitted. Broadly speaking, the coverage is that included in seminars and lectures I have given on this theme since 1969. We may divide problems of spatial pattern (in contrast with complete random chaos) into (i) detecting departures from randomness, Oi) analysing such departures when detected, for example, in relation to some stochastic model and (iii) special problems which require separate consideration; for example, sophisticated problems of pattern recognition in specific fields, such as the computer reading of handwriting or recognition of chromosomes.