Professor Carstairs is best known in India for his study of personality formation in a traditional Hindu society—The Twice ­Born. That study dealt with the social determinants of a 'normal' personality and relied mostly on an imaginative use of the clinical method, most of the evid­ence being essentially anecdotal. This study, co-authored with an Indian psy­chiatrist, Professor R.L. Kapur of the Natio­nal Institute of Mental Health, Banga­lore, is rather different. It deals with the prevalence of mental illness in an Indian village and is based almost entire­ly on a statistical analysis of data collec­ted in a survey with an interview schedule specially designed for Indian conditions. It is an exercise in epidemiology concer­ned not so much with the analysis of personality types but with the incidence of personality disorders and their social correlates. Any such study must rest on some definition of mental disorder. The authors of this book define a symptom of mental illness as ‘an item of behaviour, mood, speech, thinking or sensorium which (a) represents a change from the usual pattern for the individual and (b) is distressful to the subject or to those around him, or both.’ How acceptable is this? The first part of this definition is surely not very meaningful. A man who has been a psychotic for years will be considered healthy and one who has just recovered from a neurosis may be con­sidered ill, unless the term ‘change’ is qualified as ‘change for the worse’, in which case we will beg the question since, now, we have to define what we mean by ‘for the worse’. Nor is the second part of the definition very helpful. The part about something being ‘dis­tressful to the subject’ is surely un­acceptable unless we believe that only a person who never suffers from distress must be considered to be mentally healthy or unless we put in some qualifi­cation about the seriousness of the dis­tress felt. However it is really the part about something being ‘distressful... to those around him’ which brings us to the core of the problem—the fact that mental illness can only be defined in a social context. This becomes even clearer when we come to one possible consequ­ence of mental illness—a deterioration in the subject's performance of his socially assigned tasks. The authors measure such a ...