Yates, Frank

(1902-1994), statistical scientist

by David J. Finney

© Oxford University Press 2004 All rights reserved

Yates, Frank (1902-1994), statistical scientist, was born on 12 May 1902 in Didsbury, Manchester, the only son and eldest of the five children of Percy Yates (c.1870-c.1935), seed merchant, and his wife, Edith (d. 1958), daughter of Frank Wright, corn and seed merchant of Ashbourne, Derbyshire. An uncle's gift of a table of five-figure logarithms led to the young, precocious Yates becoming interested in mathematics. He gained a scholarship to Clifton College, which he attended from 1916 to 1920, and from where he went, with a senior mathematics scholarship, to St John's College, Cambridge. He graduated among the wranglers of 1924.

Employment opportunities for young mathematicians were few, but several years as a schoolmaster soon showed Yates's need for more practical stimuli than came from teaching differential calculus to the uninterested. In 1927 he joined the geodetic survey of the Gold Coast (now Ghana) as a research officer. Here began a deep appreciation of Gaussian least squares, and a lifelong love of the slide rule and other aids to efficient, well-organized, and accurate arithmetic. Health problems obliged Yates to return permanently to England. On 2 November 1929 in Lancaster, he married Margaret Forsythe Marsden (b. 1904/5), an analytical chemist and daughter of John William Marsden, a civil servant. Chance brought a meeting with Ronald Aylmer Fisher (1890-1962), whose revolutionary impact on statistical theory and practice was then at its peak. After a very informal interview Fisher engaged Yates (in August 1931) as a mathematician in his department at Rothamsted Experimental Station. Although Yates's family background had given him some insight into agricultural matters, his assimilation into agricultural research was rapid. He showed tremendous aptitude for the combinatorial aspects of experimental design. The Fisher-Yates association became both a deep personal friendship and a powerful research combination. Together they proved (in 1934) the ancient conjecture that no 6 × 6 Graeco-Latin square exists. Yates's own remarkably influential monograph on factorial design (published in 1937) contains the first instance of the subsequently overused practice of indicating levels of statistical significance by asterisks. His name is perhaps better known for the continuity correction (published in 1934) that has become standard practice in the analysis of 2 × 2 contingency tables.

In 1933, even before these seminal publications, Yates succeeded Fisher as head of Rothamsted's statistics department, a position in which he was to remain until 1968. Two important papers in 1936 typify his mastery in producing new systems of experimental design. These introduced several types of incomplete block design, well matched to the needs of comparing new crop varieties and of studying alternative crop rotations; features of these have influenced experimentation throughout biology and industrial technology. So influential was Rothamsted thought on all quantitative experimentation that in 1953 Sir Harold Jeffreys could tell the British Association: 'the standard of presentation of results in agriculture is better than in any of the so-called exact sciences' (Yates and Finney, 219).

One of the finest products of Fisher-Yates collaboration was their volume of statistical tables (1936), a model of compositor's art and a friend to all statistical scientists. A by-product of Yates's work on factorial design was an elegant algorithm that later proved important to the development of fast Fourier transforms. When invited to explain how he devised this procedure, he replied characteristically: 'Well, it's absolutely obvious isn't it?' (private information).

Yates's first marriage ended in divorce in 1933. On 14 July 1939 he married Prascovie (Pauline) Tchitchkine (d. 1976), the divorced wife of Alexis Tchitchkine and daughter of Vladimir Choubersky, a railway engineer. Shortly after the marriage, with the outbreak of the Second World War, the change in Yates's personal life was mirrored by changes in his research. He began with what would later be termed a metanalysis. He compiled all available experimental evidence on the responses of food crops to fertilizers, undertook a critical synthesis, and assessed the benefits to food supplies to be expected from wise use of fertilizer imports; this became a basis for national policy on imports in the face of the submarine menace. Friendship with Sir Solly Zuckerman led Yates to become a special consultant on allied bombing policy and assessments of its effectiveness, a type of study later to become popular in industry under the name of operational research.

After 1945 Yates's energy, determination, and earthiness made Rothamsted a world-renowned centre for research in agriculturally oriented statistical science and especially in the new field of statistical computing. Indeed, his earthiness made him, a man without formal training in agriculture or biology, a very successful deputy director of Rothamsted from 1958 to 1968. His tall, handsome figure was familiar to all as that of a colleague always prepared to help dispose of a difficulty. Though his interest in innovatory experimental design continued, it was now overshadowed by concern for the practice of sample survey. This was initially driven by a wartime need for estimation of national timber supplies, and by his own efforts to learn how farmers use fertilizers. He was one of the earliest members of the United Nations Sub-Commission on Statistical Sampling. His work with the UN led him to publish Sampling Methods for Censuses and Surveys (1949; 4th edn 1981), a work which did much to establish sound principles and technical terminology.

Yates's early experience in surveying led him to extend Fisher's tradition of well-planned statistical arithmetic and, ultimately, to become a pioneer of electronic computing. In 1954 he brought an early Elliott computer to the handling of Rothamsted's growing load of statistical analyses. He had to cope with programming for primitive hardware dependent upon the vagaries of thermionic valves and storage on a magnetic drum, but he saw from the start the need for the generality of structure and applicability which, by the 1990s, had become the mark of major software packages. He played a leading part in the creation and extension of the British Computer Society, of which he was president in 1960-61. He also provided much help to developing countries in respect of statistical and computing needs for agricultural research.

During his long career Yates received many honours. He was elected FRS in 1948, received the Guy medal of the Royal Statistical Society in 1960 (he was president of the society in 1967-8), and was appointed CBE in 1963. In 1966 he was awarded a royal medal of the Royal Society. In 1978 he received an honorary DSc from the University of London. He was the author of more than 150 scientific publications, ranging widely over statistical methodology and its applications. From retirement to the end he remained professionally very active, especially in producing software and as an editor of the Journal of Agricultural Science; for this last he did much to encourage the enlightening use of statistical procedures and clear presentation of their logic. He died on 17 June 1994 in hospital in Harpenden, and was survived by his third wife, Ruth, formerly Hunt (d. 1999), whom he had married in 1981 following the death of his second wife in 1976.


D. J. Finney, Memoirs FRS, 41 (1995), 555-73
F. Yates and D. J. Finney, 'Statistics and computing in agricultural research', Agricultural research, 1931-1981: a history of the Agricultural Research Council and a review of developments in agricultural science during the last fifty years (1981), 219-36
private information (2004)
personal knowledge (2004)
b. cert.
m. cert. [Prascovie Tchitchkine]
m. cert. [Margaret Forsythe Marsden]
d. cert.

CUL, corresp. with Gordon Sutherland  SOUND  BL NSA, performance recording

photograph, repro. in Finney, Memoirs FRS

Wealth at death  
£594,930: probate, 12 Aug 1994, CGPLA Eng. & Wales

Oxford University Press 2004 All rights reserved


GO TO THE OUP ARTICLE (Sign-in required)