Agricultural research now has a long tradition of being served by statisticians, both those who are professionally trained and others who (in no derogatory sense) may be described as amateurs because their primary scientific training is in some other discipline. This paper urges that, as part of good research management, more attention be given to the numbers of statisticians needed and the responsibilities they should undertake.

Questions to be considered include not only ‘Who should analyse specified bodies of data?’ but also ‘Which data require full statistical treatment?’. Poor judgement here can mean that the pattern of statistical activity is governed by the forceful approaches of other scientists more than by the scientific content of a problem; moreover, statisticians may be left with no time for the vital tasks of collaborating with others in the planning of research and displaying initiative in new approaches to quantitative problems. Not all research institutes have yet made their statisticians collaborators and partners in research rather than technicians with sharply delimited duties.

Agricultural statisticians today should be deeply involved in research in cooperation with other disciplines, as innovators in statistical technique, and to some extent in development of statistical theory. They must be encouraged to publish, as partners in research teams and individually, to produce and document computer programs that implement methods required in agricultural research, and to improve statistical understanding among their colleagues in other disciplines. They are scientists whose responsibilities and rights are equivalent to those of other disciplines but need to be interpreted in relation to the research role of their subject. This paper asks for their proper integration into agricultural research, and not for special privilege.