Successive editions of the ICD and DSM classification systems have exercised a major influence over contemporary diagnostic practice and perceptions of the form and structure of disorders. Periodic revision has been based on clinical and epidemiological research, and minimal attention has been given to any possible contribution from historical evidence. To test the potential value of such evidence, the historical manifestations of four eating disorders (anorexia, bulimia, rumination and pica) were considered in relation to the clinical descriptions and diagnostic criteria of DSM–III–R (American Psychiatric Association, 1987) and ICD–10 (World Health Organization, 1992). For each disorder, evidence is presented of continuities and discontinuities with the phenomena recognised currently. Instances of symptom overlap between disorders and the implications of the historical findings for some current classificatory issues are discussed. When spread over several centuries, case numbers remain relatively small, the amount of clinical detail is highly variable, and temporal distribution is uneven. The conclusions that can be drawn, therefore, are necessarily somewhat tentative and subjective.

During several recent international meetings on classification, there have been frequent references to national systems of classification developed and used in Europe, North America and many other countries. The UK has been notably absent from this list. As Professor Kendell, in his brief historical survey of the subject, points out: “British psychiatry does not have, and indeed never has had, any important diagnostic concepts of its own in the way that French, American, and Scandinavian psychiatry still do” (Kendell, 1985).

The National Health Service and Community Care Act 1990 defines needs as “the requirements of individuals to enable them to achieve, maintain or restore an acceptable level of social independence or quality of life” (Department of Health Social Services Inspectorate, 1991). Other definitions of need have also been used by researchers, based on different theoretical perspectives. Psychological theories have used the concept of need as a basis for understanding action. Psychiatry, by contrast, often employs the construct to inform service provision, and plan individual care. These issues are important because implementing the NHS and Community Care Act in a consistent manner will involve agreement on defining and prioritising need.

How can we design relevant psychotherapy research? The answer must be shaped by the objectives and potential consumers of such research. For over 40 years, “does psychotherapy work?” (Eysenck, 1952) held the attention of psychotherapy researchers, and randomised clinical trial methodology seemed the most appropriate empirical option for answering this question. There are now well over 500 studies that attest to the efficacy of psychotherapy (see Smith et al, 1980; Shapiro & Shapiro, 1982; Lipsey & Wilson, 1993, for meta-analytic research summaries); it seems that psychotherapy is one of the best documented medical interventions in history. But the conclusion that psychotherapy “works” is akin to finding that antibiotics “work”. We are left with the daunting task of determining which of the wide variety of treatments (psychotherapies, antibiotics) are appropriate for which variety of illnesses (psychopathologies, infections). Morris Parloff (1982) warned us of this need for specificity in his classic article, “Bambi meets Godzilla”, but our preoccupation with documenting the efficacy of psychotherapy has only recently abated enough to mount empirical studies of such specificity.