Maurice Bartlett

Click to follow
The Independent Online

Maurice Stevenson Bartlett, statistician and probabilist: born Scrooby, Nottinghamshire 18 June 1910; Assistant Lecturer in Statistics, University College London 1933-34, Professor of Statistics 1960-67; statistician, ICI 1934-38; Lecturer in Mathematics, Cambridge University 1938-47; Professor of Mathematical Statistics, Manchester University 1947-60; FRS 1961; President, Royal Statistical Society 1966-67; Professor of Biomathematics, Oxford University 1967-75 (Emeritus); married 1957 Sheila Lockwood (née Chapman, died 1998; one daughter); died Exmouth, Devon 8 January 2002.

The last 70 years have seen the vigorous development of mathematical models in technology, epidemiology, economics and the natural sciences whose dynamics are intrinsically random in some sense; life is such. The study of the behaviour such models might demonstrate is that of "stochastic processes". The converse task, to determine an appropriate model from real-life behaviour, is that of statistical inference.

Maurice Bartlett was a pioneer in these disciplines whose insight and power as a problem-solver brought him an international reputation. He made basic advances in the statistical analysis of time series and in the study of stochastic models of the growth of population, the evolution of epidemics and systems with a spatial dimension (e.g. vegetation, meteorological variables). A much less tangible phenomenon which never lost its fascination for him concerned methods for the identification of "factors" in psychology or performance.

These are just the major heads in a series of investigations in which he simultaneously developed theory, evolved methodology and pursued the particular application to a conclusion.

Success in such endeavours requires physical insight and mathematical power, and Bartlett possessed both qualities in high degree. The mathematical advance which is perhaps most striking was an operator formalism for Markov processes whose full implications became evident only after 40 years. In 1955 he published his pioneer text An Introduction to Stochastic Processes which, by integrating his own work with the then scattered literature of the subject, gave the field of stochastic processes both definition and substance. This was long a guide to those entering the subject.

Bartlett was a clear British type: a man who senses his own quest-point in some terra incognita and who hacks a path to it on his own resources. As with most pioneers, those who subsequently prettified the path received over-much of the credit. Bartlett's arguments were sometimes found wanting in rigour by later authors, but insight and rigour are steps which must alternate, and surely the first is the critical one. Bartlett was no self-publicist; both his written and his spoken exposition verged on the terse. However, the power of his mind and the idealism of his motivation were immediately apparent, and a degree of persistence on the part of the reader or listener was well rewarded.

He was born in 1910 in Nottinghamshire, and educated at Latymer Upper School and Queens' College, Cambridge. He took a post as Assistant Lecturer in Statistics at University College London for the year 1933-34; this was followed by a more substantial stint as statistician at ICI (1934-38). The realities and problems which he encountered there undoubtedly set him on the research course which he then followed, although in fact his publications for those years roamed over genetic, psychometric and agricultural topics as well as methodology.

In 1938 he returned to Cambridge as Lecturer in Mathematics. The university had traditionally allocated the subject of statistics to the Faculty of Agriculture, but the increasingly mathematical nature of the subject brought about the creation of the new lectureship with this emphasis in the Faculty of Mathematics, to which Bartlett was now appointed.

However, the Second World War intervened, and Bartlett spent the years 1940-45 at the Projectile Development Establishment of the Ministry of Supply. The PDE was concerned with the development of rocket batteries against air and surface targets, and Bartlett split his time between London and the testing station at Aberporth, in Cardiganshire. The war revealed a lack of trained statisticians, and with the return of peace efforts were made to mount courses answering this demand. Bartlett, now back in Cambridge, played his role in this with advanced classes which were highly appreciated. The appreciation was indeed such that in 1947 he was appointed as the first Professor of Mathematical Statistics at Manchester University.

In his new post he had colleagues such as Max Newman and James Lighthill in mathematics, Tom Kilburn in computing and Bernard Lovell in radio-astronomy. This flush of returning post-war talent at its most productive brought the university almost to incandescence. With his own Statistical Laboratory, Bartlett was able to develop a range of courses, postgraduate students from all over the world forming a happy group. In the words of one of these, Joe Gani:

We were presided over benignly by Maurice, whose shy thoughtfulness was a byword among us. We were learning a great deal from him; his wide range of interests and his multifarious research activities sustained us in all our endeavours. Every problem we raised inevitably called forth a ready and learned response.

I myself first encountered Bartlett in 1953 when he attended a meeting in Uppsala. My first glimpse of him was on a drizzly railway platform: an unexpectedly bulky, ruddy and cheery figure, almost farmer-like, with a hat whose floppy brim was etched by the several chromatograms of successive seasons.

From Manchester he went on to occupy the Chair of Statistics at University College London (Francis Galton's sometime fiefdom) for the years 1960-67, and then the new Chair of Biomathematics at Oxford University until his retirement in 1975. His research continued steadily during these years, concentrating particularly on biological and spatial models. His Selected Papers were published in three volumes in 1989.

By Peter Whittle