Elizabeth Bott and the Formation of Modern British Sociology

1. The psy-sciences and the social sciences

There is now a growing interest in the constitutive role of the social sciences in what Ian Hacking (2006) calls ‘making up people’, and through this process defining and constituting the social itself. Rather than seeing the social sciences as somehow outside the social relations which they claim to report, they are thus thoroughly implicated in the social itself. Theoretically, the main inspiration has been from Foucault, and has generated a concern with the technologies, or what Latour calls the ‘inscription devices’ by which social scientists – like other scientists – ‘engineer the social’. The map, the land register, the census, the survey and other routine tools of social research thus become key agents which construct social worlds.

Although this argument is now familiar in general terms, there have been few attempts to explore how it addresses the study of socio-cultural change in modern Britain. Many of the best known studies, from James Scott's, Seeing like a State (1998), through Timothy Mitchell's Rule of Experts (2002) to Nicholas Dirks Castes of Mind (2004), focus on social sciences as a form of colonial knowledge and do not study the metropolitan centres themselves in these terms. This is one reason why Nikolas Rose's Governing the Soul (1999a) remains such an important study even though it was first published nearly twenty years ago. Focusing on the American and (especially) British cases, Rose sees the post-war social sciences as critically involved in the formation of new projects of ‘governmentality’ that sought to ‘rule through freedom’. Rose argues that the psy-disciplines framed a science of the individual which was also a science of society. The project of democracy involved the creation of ‘responsible’ individuals who were capable of self regulation and thus automatically aligned their own selves to the social order of neo-liberal democracies (see more generally, Rose, 1999b). Therefore, rather than endorsing the leftist critique of psychology as repressive, Rose insists on the role of these disciplines in doing exactly what they claim to do, namely in defining the modern, ‘free’ citizen who is also a social being.

Rose points to the way that these new forms of research were dependent on the use of distinctive ‘inscription devices’, means of transforming knowledge into devices that can be stored, manipulated, proliferated and deployed as interventions. Drawing on Latour's actor-network theory, he points to the way that the psy-disciplines succeeded not only through the force of their theoretical arguments, but also in terms of the devices they assembled to produce knowledge: statistical tables, diagrams, narratives, case studies and accounts. Rose brings out how, in the same way that the power of the natural sciences resides in their ability to deploy laboratory devices to assemble the natural world they study, the power of the social sciences rests in their battery of research tools and procedures – whether this be the survey, the interview, observation, the use of official data or whatever.

A number of questions are posed by Rose's challenging account. First, he does not clarify how he sees the relationship between the ‘psy’ disciplines, narrowly defined, and the social sciences more broadly. He focuses very largely on the role of psychology, psychoanalysis and psychiatry, though he does extend his discussions to include other social sciences from time to time, and is certainly attentive to the influence of psychologists on the social studies of work, management and the family. In more recent work with Thomas Osborne, Rose has developed cognate arguments about how the social scientists ‘make up’ the world, focusing for instance on the role of the opinion poll in constructing ‘public opinion’ (Osborne and Rose, 1999), or urban poverty researchers in constructing notions of parcellised and ‘striated’ household space (Osborne and Rose, 2004). These later studies, however, do not obviously focus on the ‘souls’ of the population, and it is not clear how influential the psy-sciences were within the broader framing of the social sciences. This tension is itself linked to an important shift during the post-war years as psychology itself distanced itself from the social sciences and became more closely affiliated with the medical or biological sciences.

Relatedly, there is the question of ‘epochalism’. One of the attractive features of Rose's account is his concern to offer a grounded alternative to the grand social theorizing of epochal shifts.

Nonetheless, despite this claim, we can detect a certain tension between the Deleuzian and Foucauldian aspects of Rose's thinking (see Gane, 2004: 176). His work can be read as a form of a ‘weak’ epochalism, in which the rise of neo liberal self-governing democracy is seen as a fundamental process which defines contemporary social and political forms (see further Rose, 1999b). To be sure, he traces this through a series of ‘local’ innovations and devices rather than through a grand ‘motor’, but nonetheless, the claims about the contemporary cultural dominance of the psy-sciences are sweeping. This is a residue of Foucault's archaeological and genealogical project itself. This relies for its analytical power on showing how otherwise disparate or even apparently opposed views can be seen as embedded in a common discourse formed on a common problematic. ⁴ Most famously, it led Foucault in The Order of Things (1971) to dispute that Marx should be seen as ‘outside’ the bourgeois political economy he thought he was criticising, and instead to place him inside its modern discourse with its central role for the human subject. As Han (2001) relates, this strategy requires Foucault to replace Kant's transcendental humanism with that of an ‘historical a priori’, where the conditions of possibility of knowledge are given solely historically. This has the effect, however, of placing the conditions of knowledge as the product of temporality, hence instantiating a certain ‘epochalism’ which is premised on ruptures and transitions. In this formulation, specific incidents, knowledge and discourses cannot but be read as instances of grand narratives, rather than as more localised incidents which retain their own specificity and effectiveness. And so it is that in Governing the Soul examples of liberal intervention are piled up, but we do not see the use of counter-factuals (eg what were the domains where psychology failed to gain jurisdiction?). Although Rose commonly recognizes the failure of the psy sciences, as in his discussion of humanistic work reform programmes in the 1970s, this is then mobilized into an argument about how it energised further attempts at intervention. ⁵ Yet although it certainly is possible that failures can engender forms of mobilization, in certain cases they marked more fundamental failures.

In pursuing these questions, I want to show how Bott's role involved importing the expertise of the psy-disciplines into anthropology and (especially) sociology, and also involved breaking from the discursive formation of the psy-sciences themselves. It is precisely this double legacy that rendered her contribution so productive. Bott's interests were formed out of the classic period of the Tavistock, when, during the period 1945–1955, the alliance between psychology and anthropology was deep and led to major research innovations: the use of surveys to address questions of national culture, and the ethnographic study of institutions and neighbourhoods. However, this programme was to fall apart in the mid 1950s, a failure which Bott herself helped to bring about. Nonetheless, what Bott also managed to do was to make the in-depth interview, a research method and inscription device, developed by psychoanalysts, the province of sociology.

2. Elizabeth Bott and the Tavistock Institute

After arriving in London having studied anthropology at Toronto and Chicago, Elizabeth Bott was employed from 1951 to 1956 as part of an interdisciplinary research team also composed of Dr A.T.M Wilson, (a medical psycho-analyst), Isabel Menzies,(a non-medical psychoanalyst) and Jim Robb (a social psychologist), in a project on ‘Family relationships’. The project grew out of the Tavistock Institute's concern, evident right from its foundation in 1947, about the nature of family relationships, and especially those between marital partners (eg Wilson, 1948). A further input came from the new Family Discussion Bureau, who advised married couples about how to deal with their marital difficulties. The case workers from these Bureaux thought it might be useful to have ‘knowledge of the background and ways of life of those families who did not feel the need of getting outside assistance in dealing with their problems’ (Robb, 1954b).The idea was to study 20 ‘ordinary’ London families defined as married couples with young children, who had no obvious problems that were seen to require professional intervention, in order to find out more about the inter-personal dynamics which shaped household relationships.

The project was deliberately designed to link anthropology, psychology and social psychology. The respondents were expected to undergo psychological tests, conducted by Wilson and Menzies, whilst Bott and Robb were in charge of the fieldwork. This involved intensive household interviews, with up to 20 visits to each family, in order to work through a detailed semi-structured questionnaire. In Robb's (1954b) account, the questions asked cover the following topics: –

detailed social history covering virtually all the descendants of the grandparents of each partner;

The respondents were also interviewed individually to ‘check’ the interpretations from the psychological tests and give their impressions to Bott and Robb. In her book, Bott's reporting of the tensions within the team as they began to interpret the data runs as follows:

This framing, which sets the anthropologist at odds with the psychologist was clearly a very important feature of Bott's own identity. It was not incidental that she saw Durkheim, with his insistence that social facts could not be reduced to psychological factors, as her main intellectual influence. She identified her reasons for coming to England from North America as her frustration with what she saw at the limits of ‘cultural anthropology’ with its focus on childrearing as the locus of culture. Having been told that British anthropologists were more genuinely interested in ‘social relations’ she set sail, in search of the social (see further, Bott-Spillius, 2005).

Yet this fervent anti-psychologism needs to be treated carefully, and is interesting to consider in light of evidence from the recorded fieldnotes, deposited at the Qualidata Data Archive. ⁶ These notes, drawn up by Robb and Bott, are comprised of detailed case notes on the 20 households who were the subjects of the research. ⁷ What comes out of these notes is that Bott and Robb, far from leaving psychology to the psychologists, wanted to review, contest, and supplement the psychological accounts of their respondents. Their notes are littered with their reflections on the psychological attributes of their respondents, extracted from their interview discussions. Here are three examples.

There is a paradox here. When we read Bott's published book, rather than the fieldnotes such as those above, none of these kinds of psychological discussions appear in its pages: it is the stated social identities the respondents, and their reported social connections that are given pride of place. The overt psychologising has given way to the language of social relationships, roles, networks, and norms, the evidence for which is extracted from people's own in-depth interviews. And, rather than psychology's distrust of stated claims, with its insistence on reading behind what is said for its refusals and absences, and for the role of the ‘unconscious’, the form of what is said is now given much more importance. Something seems to have happened between the fieldwork of this project, in the early 1950s, and its writing up, which transformed the relationship between psychology, narrative, and observation.

In order to unravel this relationship, we need to place Bott's work within the research tradition which the Tavistock Institute led in the post-war social sciences. Drawing closely on American expertise and their experience of wartime research on morale they pioneered new forms of social investigation.

What is striking about the Tavistock's mission was its commitment to empirical research, an emphasis evident in the founding statement of the International

Committee of the Society for the Psychological Study of Social Issues (division of the American Psychological Association), which was proudly reproduced in the Tavistock journal, Human Relations.

We can see this commitment, in Rose's terms, as concerned with policing the mundane: there was no corner of everyday life that was un-interesting to these researchers. This explains why the Tavistock was not interested in the kinds of sample survey methods that were beginning to become popular in other branches of the social sciences but which seemed insufficiently complete or comprehensive for their purposes (notably, Glass, 1954, and in the Government Social Survey, on which see Moss, 1991). The Tavistock's main theoretical frame, Lewin's field theory, had no truck with the idea that one could isolate individuals, measure their attributes using sample surveys, and seek to discriminate between the importance of different causal variables. Rather, field theory insisted on the need to look at social process in its context, in order to explore forms of inter-dependence and inter-relationships. This was deemed to require case studies involving comprehensive programmes of observation. The most important was the study of the ‘Glacier Metal Company’ conducted in the late 1940s (Jacques, 1951). This study saw no less than eight full-time observers ⁸ descend on one London factory where they spent two years conducting intensive observational projects, examining company records, and analyzing all its data on labour issues. Amongst the most interesting features of the project was the concern to study all grades within the factory, from the senior managers to the most junior worker, which was seen as essential to allow the entire social structure of the factory to be mapped out. The framing of the study emphasised the interaction between ‘structure, culture, and personality’, which borrowed ideas from cultural anthropology, psychotherapy and medical practice. Rose sees this study as a form of ‘psychoanalysis of the organisation’, and explores how the project sought out the healthy functioning of the factory by encouraging ‘working through’, where unconscious tensions could be faced up to and constructive forces engaged with.

The Glacier study, alongside several other famous Tavistock projects, such as the ‘longwall’ coal mining study (Trist and Bamforth, 1951b), were based on studies of ‘complete’ institutions where observation could readily be conducted in a bounded setting. However, it was not clear how such methods could be applied the study of families and households which were spatially scattered and dispersed in urban environments. It was especially here that the psychologists turned to the assistance of anthropologists, who were seen to offer expertise which the psychologists otherwise lacked. Kurt Lewin was clear about the potential of this kind of inter-disciplinary alliance based on a common commitment to empirical analysis of case studies. He claimed that the

The most significant anthropologist to be enlisted by the Tavistock was Adam Curle, who worked with Eric Trist on a project exploring how demobilised prisoners of war settled back into domestic civilian life in the 1940s (Curle, 1947; Curle and Trist, 1947). This project was important in registering the difficulty of conducting observational research outside ‘closed’ settings. It was noted that prisoners had presented themselves to psychiatrists with various symptoms of distress, but that these did not appear to be linked to obvious psychological causes. Curle and Trist decided to study the domestic lives of these ex-prisoners to gain greater information about the psychological state of the respondents. Given the difficulty of observing directly in people's own houses, a great play was made about using the ‘in depth interview’, seen as a means of eliciting quasi-observational information. These accounts were then compared with those of friends, neighbours, and employers, so that their declared testimony could be cross-checked, and possibly ‘corrected’.

The analysis is reported in highly normative form, with prisoners distinguished into three types, ‘the settled’, the ‘unsettled’ and ‘the norm’. Curle and Trist concluded starkly that the domestic and family lives of most of those they interviewed was very bleak, characterized by intense conflict and tension between husbands and wives. In the ‘normal’ household, it was usual for there to be major domestic disputes, leading to heated arguments lasting for two hours or more, roughly on a fortnightly basis. However, in a significant minority of ‘unsettled’ households, relations were much worse, with physical abuse of the wife by the husband and a near constant state of animosity and antagonism between the partners. Even in the ‘best’ households, where conflict between husbands and wives was not overt, this was not because the partners saw ‘eye to eye’, but rather that they had learned how to laugh off, or in Tavistock terminology, ‘work through’ their difficulties.

Wilson summarized their findings about the traumatic state of domestic relations Curle and Trist discovered in their post-war studies.

This account of the family was, of course, very different from that being developed in American functionalist sociology by Talcott Parsons and his associates, which saw the ‘nuclear family’ as the appropriate fit for modern industrial society. For Tommy Wilson, Bott's colleague, there was a worry that the relations between husbands and wives lay outside the normal purview of the inspectorial gaze. Whereas social workers, the clergy and doctors can visit individuals within households, and whereas there was a developing infrastructure of intervention around child development, they detected a real difficulty in observing directly the relationships between husbands and wives. Wilson went on:

We see here a characteristic concern to problematise an area of everyday life so that it could be better patrolled. For this purpose, Tavistock researchers developed the idea of ‘desocialisation’ – an idea which has some parallels with current sociological arguments about ‘individualisation’. ⁹

This account was made more pertinent by the sudden rise of divorce levels in the post-war period, and the sense that the conventional family was not able to cope with the stresses placed on it. The Denning Committee on Procedure in Matrimonial Causes urged the need for Marriage Guidance. These kinds of concerns can be seen, as in Rose's terms, as constituting marriage as a terrain for the intervention of the psy-sciences’.

And yet, the interesting thing is that this management was never delivered, and certainly not in a way comparable to intervention around child development. Marriage Guidance Councils remained reliant on voluntary funding. Unlike the compulsory regime of child inspection that was institutionalised at this time, no equivalent compulsory, or indeed, even voluntary, inspection and surveillance monitored the relations between husbands and wives. It is not incidental that Rose's account of the concern by the psy-sciences to construct the normal family is focused nearly exclusively on concerns with child development, and he only addresses the marital relationship in passing, in a brief discussion of the marriage guidance movement and the family therapy movement (Rose, 1999a: 175–6, 205). Psychologists failed to develop the battery of inscription devices and regulation in the area of marital relations compared to that of child development. In fact, as Carol Smart (1984) argues, marital relations remained much more organized through legal and institutional intervention. It is interesting that research on sexual behaviour, which might have been one way of institutionalizing and organizing ‘scientific’ interventions around marital relations, failed to gain institutional respectability, with the pioneering research of Mass-Observation in their ‘little Kinsey’ studies on sexual behaviour being nearly entirely ignored (see Stanley, 1995) and Kinsey's own American work having little direct take-up in the UK.

The reason for this failure can readily be ascertained by pursuing Rose's argument about the concern of psychology to construct ordinary, free and self regulating individuals. Whilst this could permit them to intervene in the project of child rearing, in the name of ensuring formation of ‘mature’ subjects, it could much less certainly be applied to marital relations without violating the very idea that families were ‘self-chosen’ entities that husbands and wives voluntarily entered into. To subject marital relations to the kind of compulsory institutionalized intervention that was meted out to children would, in fact, be to undermine the very frame in which the psy-disciplines operated. Or, to put this another way, the world of patriarchal relations, involving the ‘private’ relations between husbands and wives, was not allowable as a truly legitimate subjects for the interventions of the psy-sciences, even when it was seen as the arena of huge problems and conflicts.

It was in this messy space opened up around the issue of relationships between husbands and wives that Bott was to intervene so powerfully. When the project began, its director, Tommy Wilson, was hopeful that it could be used to generate a psychological-anthropological synthetic analysis of the domestic relation that could mark the further advance of the psy-disciplines into a further area of jurisdiction. The idea was to develop a holistic account of the psycho- and sociological relations between husbands and wives to allow a fuller understanding of how families could be successful. The project, however, ended in failure. As late as 1956 the plan was to write a major monograph, with the psychologist Tommy Wilson drafting 50,000 words on the psychological dynamics of the couples, and Bott writing a series of compressed case notes on each of the households. Bott modestly noted in a letter to her colleague Jim Robb that the Tavistock had also agreed to publish her own work as a ‘technical report’. Even at this stage, Bott noted that ‘the ambitious scheme of integrating the sociological and psycho-analytic analysis has gone by the board’, with Wilson's monograph simply to comprise annotated case studies. But in fact, even this limited monograph was never produced. ¹⁰

We can see what Family and Social Network might have been by considering the example of Bott's collaborator, Jim Robb, whose PhD in social psychology had been conducted in the late 1940s. Robb examined social psychological determinants of anti-semitism, pursuing this through a local study in Bethnal Green, an area with a known tradition of fascist support. Robb pioneered a unique fusion of ethnography and psychology, in which detailed observation about local social relations (deriving from his work as a bartender) was linked to the deployment of psychological tests on 117 local respondents. In the years before anthropologists or sociologists had begin conducting community studies on the British mainland, Robb's book was hailed in a special introduction by the leading sociologist W.J.H. Sprott as a major achievement. What did Robb's work consist of? He anticipated Young and Willmott's famous study of Family and Kinship in East London by noting the unusually close relationships between family members, especially as mediated through their mothers. Unlike Young and Willmott, he saw the nature of parenting as central to this local culture. In crowded, cramped, housing conditions, mothers took their babies into their own beds so that they could be ‘demand fed’ so as not to disturb neighbours. However, after they were weaned this close maternal relationship was abruptly disrupted, and children were quickly neglected by their busy mothers, leading to problems of adjustment and anxiety on behalf of young children which predisposed the boys to gang life, and in general proved conducive to anti-Semitism where Jews were identified as a source of anxiety. Robb's psychological tests showed that anti-Semitic Bethnal Greeners were more traumatised than their more tolerant neighbours, so that he could establish a link between local forms of child rearing, psychological states, and local forms of action.

What is interesting about this study is both that it is one of the first ever local studies in Britain, but also that it is the last of its type. It seems quite plausible that when Robb was enrolled alongside Elizabeth Bott, Tommy Wilson was hopeful that something of the synthesis that Robb seemed to have established in Bethnal Green might emerge. But it did not. Bott's book rests on a bigger failure. She found a way of writing about marital relationships, in terms of networks, relationships, and roles that did not involve a theory of personality or psychological dynamics. In the process, she took a method, the in-depth interview that had been devised by psychoanalysts, and redeployed it within an anthropological and sociological frame.

3. Bott's originality: 1: networks

Bott is often lauded as one of the pioneers of social network analysis, with her idea, that the relations between husbands and wives could be placed in the context of the social networks of the partners, being pivotal to the sociological study of families and to the development of the network method itself (eg Scott, 1990). However, we need to be cautious about this. Network thinking was a strong feature of the Tavistock tradition, and in many respects Bott was simply deploying and refining a mode of analysis that the Tavistock psychologists had already developed.

A central feature of Kurt Lewin's field theory was that social research needed to deal with social pheonomeona in relational terms. Rather than differentiating pathological from ‘normal’ action and focusing attention on the former, it was essential to look at all the elements so that their interrelationships could be fully understood. Bott herself was quite unequivocal in her study that the ‘the basic conceptual model is that of field theory: Behaviour is a function of a person (or a family) in a situation’ (Bott, 1971: 4). This insistence on the centrality of field theory was second nature to the Tavistock, and was upheld by Tommy Wilson as well as in the ‘flagship’ Glacier study.

The general idea of network thinking is strongly embedded in Lewin's field theory (Scott, 1990; Freeman, 2004). Thus, the journal Human Relations was using sociograms as a staple means of identifying social relationships: in 1949 the American psychologist, Festinger, published an account of how matrix algebra could be used to identify cliques, and he also authored a pivotal paper on how the spread of rumours could be linked to chains of communication on a housing estate.

In developing this perspective, the Tavistock research broke from individualistic, even psychologistic, arguments. Indeed, several ideas which were subsequently to be taken up by sociologists can be seen to exist, in embryo form. Probably the most important figure, enthusiastically acknowledged by Bott as a major influence on her thinking (‘the real inspiration’), ¹¹ was Eric Trist. Although a psychologist, Trist's emphasis was not on personality factors or on unconscious drives, but in recognising the social processes which defined social roles and activities. ¹² Bott, indeed, regarded him as strongly influenced by Marxism. Working with a number of colleagues, he developed a series of striking innovations, probably the most remarkable of which, was his analysis of labour turnover in an industrial plant (Rice, Hill and Trist, 1950). Rather than seeing people leaving their employment as a ‘problem’, which could be attributed to the workers' motives, Rice et al. boldly proclaimed that ‘labour turnover as a process is the function of an institution’ (Rice, Hill and Trist, 1950: 350). Amplifying this point, they emphasised that turnover could only be seen in the context of the ‘work contract’ and that ‘an employee begins to leave when he enters, and continues to enter until he leaves’ (Rice, Hill and Trist, 1950: 356). To understand why workers leave their jobs it was more useful to look at the kinds of situations in which they found themselves, rather than their personalities. The implications were boldly stated in their conclusion

We have here an identical formulation to that which Bott reported in FSN. Rather than focus on ‘problem’ workers or families, it was necessary to understand this issue in respect to all aspects of the situation. Trist similarly developed this line of thinking with his work on industrial accidents. Noting that most research examining the reasons for accidents attributed them to the characteristics of the workers (‘accident-proneness’) or dangerous situations, Hill and Trist (1953) developed a third account which prefigured Parsons's arguments about the ‘sick role’: ‘accidents will be considered as a means of withdrawl from the work situation through which the individual may take up the role of absentee in a way that is acceptable to him and the employing organisation’.

It is clear therefore that the Tavistock psychologists had already formulated an approach which was similar to that which Bott championed in FSN: a refusal to attribute actions to motives, and an insistence on the need to link social action to context and the nature of the social relationships in which the individual was embedded. This involved breaking from individualistic and situational analyses. It is this which explains why it was the Tavistock Institute's journal Human Relations which published the first formal account of networks, in the form of the anthropologist John Barnes's (1954) study of Norweigian fishing communities. ¹³ Barnes's use of network metaphor is itself interesting in being derived from electrical circuits, further testifying to the dominance of Lewin's mechanical models of fields, rather than biological metaphors as used by other anthropologists, such as that of the ‘web’ used by Meyer Fortes. Bott claims that the idea of the network came to her ‘out of the blue’, as part of an ‘Archimedean moment’. She came to recognise the possible significance of social networks well after the interviews had been conducted in 1951–2. ¹⁴ Whilst this is no doubt true, we should also see the ideas as completely consistent with the intellectual framing of the Tavistock researchers, and it is not incidental that it met with absolutely no resistance from them.

Where Bott innovated was in her strategic use of the idea of social networks as allowing her to bracket out the personality of family members from her concerns. Social networks, it thus appears, allowed a way of writing up her fieldwork in the name of ‘roles’, relationships and networks without requiring a theory of personality. Conjugal relations are the product of the social networks of husbands and wives. There is no reason to doubt the sincerity of her conversion to the social network approach in order to observe that this had the advantage that it allowed her to write up her part of the project without treading on the toes of the psychologists whose brief it was to write up the psychological sides.

We can see how Bott's key innovation, the elaboration of the ‘network approach’, was so strategically important in offering her an interim means of resolving what seemed like a fundamental impasse. The idea of the social network placed emphasis on the way that people's social contacts conditioned their values and actions. In her study, Bott argued that the social networks of middle-class respondents, which tended to be shared between husbands and wives, encouraged joint conjugal roles, whereas the social networks of working class households were characterised by strong gender differentiation between men and women, and separated gender roles. Network approaches repositioned sociological explanations so that they did not need to rely on an account of personality, which could subsequently be seen as the province of psychology alone. Sociologists did not need to give their own account of personality, different from (whether in conflict with or compatible with) psychologists, but could instead develop explanations that sidestepped a theory of personality. In this respect Bott's work was seminal. She did not invent network analysis, but she used it as a means of differentiating the provinces of psychology and sociology/anthropology. However, in a peculiar way her innovation also testifies to the power of the psy-disciplines in shaping the agenda of these new social sciences.

Bott's originality (2): the use of the in-depth interview

In 1932 the doyens of British social science, Beatrice and Sidney Webb, Fabian founders of the LSE, published probably the first ever comprehensive British guide to social research methods. Their advice for the aspiring social researcher focused closely on the need for observation as the fount of social research. Questionnaires could be useful, but because they were artificial instruments they could confuse those who filled them in and mislead researchers. Statistics needed to be approached with caution. However, noting that it had its origins in psychoanalysis, they did state that the interview

Their caution about how the interview could be used, is, however, striking. There is no sense in their account that the interview can reveal aspects of the respondents' identities, values, or personality. The interview is a means of eliciting information about an institution or situation. It is for this reason that the interview is focused on the ‘key informant’, the person who has some valued knowledge to impart: ‘the person interviewed should be in possession of experience or knowledge unknown to you’ (Webb and Webb, 1975: 142). It is essentially an adjunct to the social researcher as observer.

This tradition of the interview as part of an observational research strategy is familiar. When Booth carried out his path breaking studies of poverty in London, he was not interested in interviewing ‘ordinary’ householders. Instead, he interviewed 400 school attendance officers and got their notes on every family they worked with. He then did further ‘wholesale interviewing’ of police, rate collectors, sanitary inspectors, school teachers, Charity Organisation Society investigators, hospital almoners, trade union officers, and even agents of sewing machine manufacturers, ‘together with individual personal observation of particular streets and even particular households when exceptionally required’ (Webb and Webb, 1975: chapter 9). The idea that there was anything interesting in ‘ordinary’ people's own views about the world was simply absent.

It was in psychoanalysis that the idea of the ‘free’ interview emerged, where it was thought that unstructured conversation could allow respondents to reveal hidden, private features. This was noted by the Webbs as early as 1932, and by the 1950s, Jim Robb noted that ‘many of the freer techniques have been developed by therapists and, although these have been used chiefly by them and by social case workers, social science research workers have also shown considerable interest’ (Robb, 1954a: 66).

Bott and Robb's use of in-depth interviews was psychological in its origins, and sought to elucidate the ‘reference groups’ of their respondents. Following the example of American researchers, Wilson's team was interested in the social identifications of the respondents, the kind of groups that were salient and significant to them. ¹⁵ The interview notes contain full and detailed accounts of the ‘negative’ and ‘positive’ reference groups of the men and women, including full details of when these were shared, and when they were segregated. Here we see the interview being used to elicit information about the respondent's thoughts, in ways that were utterly alien to the older tradition of the interview as adjunct to observation that the Webb's prized.

What did Bott do that was so distinctive? She became interested in how these interviews could be used to elicit respondents' views about their social class identity. The Tavistock had no conceptual vocabulary to deal with class, which was largely lacking from their work, even when studying manual workers. Curle and Trist (1947), when exploring the norms and values of the resettled soldiers emphasised their self perception not as members of social classes, but as ‘aright’, ‘just ordinary’, ‘nothing special’, or ‘quite respectable’. Trist and Bamforth (1951) did not deploy the concept of class in writing their study of longwall coal miners. Whilst, as we have seen, the Tavistock practice was to explore issues relationally, through positioning them within fields, they did not invert or question received normative judgements about what problematic, or desirable, practices were. In this respect, they remained highly anchored within a normative, medical problematic. Curle and Trist thus argued on the basis of their interviews with resettled prisoners of war that there was a coherent normative structure to their views in which

It followed that Curle and Trist identified differences in family forms in terms of different capacities of families to live up to universally shared norms, rather than in terms of conflicting views about what family life might consist of. ¹⁶

What Bott did was to ‘relativise’ more fully this kind of research, so that rather than defining one dominant ‘norm’ to which households aspired with varying degrees of conviction and effectiveness, she recognised that norms themselves were contested and varied. This allowed her to present her families as various types, interpreting such differences in terms of their different kinds of social relations, without seeing one kind as morally superior. By identifying the variability of social norms and values, and the way that the researcher needed to explore individuals’ own takes on them, Bott provided a rationale for the intensive in-depth interview conducted by sociologists, as a fundamental research tool.

What Bott then did was to use the concept of class as a means of summarising these kinds of complex views about their social relations. ‘When an individual talks about class he is trying to say something, in a symbolic form, about his experiences of power and prestige in his actual membership groups and social relationships both past and present’ (Bott, 1971: 163). Through this formulation, class becomes a means of linking together disparate attitudes, values and conceptions of the social. Although there is no reason to doubt her insistence that she was deriving the concept of class inductively, it seems that this involved her using the label of class to summarise more specific social labels and badges. Rather like the concept of network, class was useful in allowing her a way of mediating structure and agency. As she reported,

Bott claims that her interest in class arose from her own experience, arriving as Canadian, and seeking to make sense of British society in terms of the hide-bound conventions that seemed to her to be so powerful. In fact, as her own notes indicate, the respondents themselves did not use the terminology of class to any great extent in discussing their reference groups and identities.

We can note that this intervention is very different to that which David Riesman led in the United States at the same period (Lee, 2008). Riesman also sought to claim the in-depth interview as a valid sociological method, but he did so largely through using it to probe the ‘character’ or ‘personality’ of respondents in a way which did not fundamentally break from a psychological framing. It is this which explains why the in-depth interview never became such a significant method in the US, and why, when it was used, it remained focused on moral concerns (eg Bellah et al., 1985). By contrast, Bott mobilised the in-depth interview in the name of more fully sociological framing in terms of perceptions of class.

This move allowed sociologists to define themselves as key agents in organising often inchoate accounts into coherent form. My detailed examination of the reference groups listed in Bott's archived interviews suggests four important points. First, there are many more negative than positive reference groups, indicating the general point that people's identities depend more on disidentification with salient others rather than on a sense of belonging to a group. This is a point indicated by Bott and Robb in their fieldnotes when they explore their respondents' sense of who were ‘people like us’. They note that rather few households had a strong sense of which people were like them, in contrast to a much more developed sense of who they were not. This was certainly marked in more middle-class households, where several indicated their concern to stand out from the herd. Thus Mr Wraith, a journalist reported that ‘we are unique, unconventional, independent, self assured, doing what we like when we like and caring nothing for other people's opinions’. Mr Bullock, an economist, identified with people who ‘are intelligent, young, socialists, politically active, mildly unconventional’. Mr Appleby, a carpenter identified strongly against ‘blokes at work’ ¹⁷ .This is exactly the way that Bott (1971: 191) puts it in her account that ‘concepts of class are used for general orientation in the society at large, for placing strangers, and for evaluating one's own position and that of others’. Class does not evoke a sense of collective belonging to a group so much as a means of differentiating and positioning.

Secondly, and something especially strongly marked amongst working class respondents, there is little sense of a strong sense of the kind of collective social identity that later came to be seen as a defining feature of the ‘traditional working class’ in the views of Hoggart (1957) and Williams (1958). ¹⁸ Positive identifications are mostly to specific groups within the working class, or more commonly, particular locations. Thus for the Newbolts: ¹⁹

Or, for Mr and Mrs Baldock, upwardly mobile mechanics who owned their own business, their identification was to the East End, but

Or for Mr and Mrs Barkaway,

Thirdly, a remarkable range of terms are used when defining the reference groups which are salient to the respondents, many of which are implicitly classed, but rather few of which use class as a definite term. If we focus on the terms used to talk about social groups, we see that only nine of the 99 terms used to define some kind of social group explicitly name class as such, and most of these are negative reference groups. Four kinds of axes of stratification are evident (a) most powerfully there is a major cultural division with, on one hand, those with practical, often manual skills (Mr Appleby's enthusiasm for ‘people who can do things with their hands’, or Mr Redfern's interest in ‘practical people’). By contrast those without practical skills, such as intellectuals, graduates, professors, drones ‘hangers on of royalty’ and the like are often singled out for criticism. This might be interpreted, in Bourdieu's (1985) terms, as an axis of cultural capital, though we might note that it is the practical which is valued over the intellectual. (b) There is also a cross-cutting moral division between the respectable and unrespectable (in which ‘criminals’, ‘spivs’, ‘fiddlers’, ‘people who borrow’, ‘people with dirty jobs’) are denigrated and (c) a distinction in terms of authority and power (‘bosses’, ‘those in authority’). By contrast, income and wealth are rarely identified as a salient division, except insofar as income can be made morally significant, through reference to ‘people who borrow’ or ‘the indigent poor’.

We can see in many of these references a continued reliance on 19^th century populist idioms (see Joyce, 1990; Lawrence, 1998), differentiating between a large ‘popular’ group and ‘parasites’ (drones, the non-productive, etc), but these have now been spliced into a more modern language of class, with reference to employers, capitalists, and so forth. Bott's achievement is to link all these complex terms to the over-arching importance of class, through her conceptual innovation that class was a means of talking about ‘experiences of power and prestige … and social relationships both past and present’ (Bott, 1971: 163). Today we might prefer to unpack these terms in their complexity rather than read them through the unifying lens of class (see generally Coxon and Jones, 1986 and Bottero, 2005). My fundamental point here concerns her use of the in-depth interview as a means of registering not psychological, but social attitudes and identities, which could then be linked to the political language of class to render issues of sociological interest (see further, Savage, 2008). Now, what people say, and how they say it, becomes of interest to the sociologist.

Bott's use of the in-depth interview has been hailed as being of fundamental importance in addressing people's ‘weltanschung’. But at the time, her use of in-depth interviews was seen as highly problematic, in breaking with the established conventions of how interview data should be used. It was precisely the fact that Bott was not behaving like a social worker that was seized on by some commentaries on the book's deficiencies. The review in The British Journal of Sociology (1958) remarked that

The review concluded that the book

This sense that the ‘in-depth interview’ was a superficial and unsatisfactory means of gaining knowledge was clearly evident from Richard Titmuss's infamous foreword to Young and Willmott's Family and Kinship in East London, published in 1957. Even though this book has endured as one of the classic sociological community studies, Titmuss, the doyen of social policy, was clearly uncomfortable with the reliance on interviews. He noted that the research project ‘raised some difficult questions of presentation. Ultimately they resolved, to put it simply, to favour readability…. this has meant that some chapters now have an impressionistic flavour’ (Young and Willmott, 1957: xi).

What we see here is a fight over jurisdiction. Since the 19^th century the task of household investigation through interview had been the province not of academic social scientists but social workers and visitors. Bott claimed that researchers should conduct their own interviews and that could then produce valid sociological insight in the name of class. It is not surprising that the social workers sought to defend their own ‘empirical’ expertise. And, because the Tavistock study separated out the collection of the psychological data from the social data (since the respondents were supposed to visit the clinic to have tests conducted on them), this meant that the household interviews did not cover moral and psychological aspects which were central to the psychological use of interview methods, so much as the social aspects. By a process of serendipity, Bott was left to write up a wealth of field material around the terrain of an emerging ‘social’. Social networks allowed her to ‘work through’ this material in a remarkable and brilliant way.

Conclusions: Bott's legacy

I have argued that Bott's major contribution is to take a method – the in-depth interview – which was originally generated by psychoanalysis, and make it part of the repertoire of sociology. Rather than social networks, as is usually claimed, what she bequeathed to sociology was its ability to define itself as an empirical discipline, with its own attendant research methods based on intensive, in-depth interviews. This allowed sociology to settle its accounts with psychology and bracket out concerns with personality. Certainly, by the later 1950s, British sociologists and anthropologists were queuing up to distinguish their concerns from those of psychologists. Whereas post-war sociology and anthropology were premised on the need for an account of personality as a necessary part of their province of inquiry, in the hands of new generations of social scientists, this idea of personality seemed unnecessary, indeed ideological. To this extent, they were Foucauldians before Foucault. ²⁰ Here is Michael Banton, writing about race relations in 1960.

The doyen of British sociology, T.H. Marshall (1960), took Barbara Wootton to task for assuming that the job of sociologists was to study ‘pathological’ conditions. David Lockwood (1956), in his influential critique of functionalism insisted on the separation of social integration from system integration, thereby arguing that people's values and identities need not be related to their social roles. In contrast to Young and Willmott's (1957) account of Bethnal Green makes no reference to personality or psychology. Nor does Frankenberg's (1957) study of Glynceriog. The social and the psychological were seen in all these – and other – studies, as doing fundamentally different things, and it was safe for the anthropologist and sociologist to bracket the personality as being of no concern to them.

This new relationship was unthinkable without Bott's intervention. It was she who developed an apparatus for researching social relationships as social networks which did not require an account of the personality of the social actor. She did this, I have argued, in part accidentally, through transforming the research technologies of the Tavistock Institute itself. Her research did two, contradictory, things. It both allowed an extension of the psy-disciplines into new areas of the social sciences, through the deployment of the network approach and the in depth interview, yet it also marked the limits of the province of the psy-sciences.

This dual legacy can best be understood in terms of the dilemmas of liberal governance itself. Liberalism constructs autonomous individuals, able to act as bearers of the democratic project. Liberalism is thereby premised on the supremacy of the individual, but much social life is actually organized through households, often composed of heterosexual male and female partners. Although liberal sciences have powerfully intervened to regulate children, as ‘nascent individuals to be’, it has proved more problematic to regulate the marital relationship itself without violating the very precepts of individual autonomy that are supposed to lie at the heart of the liberal polity. It is for this reason that gender relations within the household pose such an intractable issue for liberal governance and proved to be the stumbling block for the Tavistock's research programme. In the post-war years, the psy-disciplines largely gave up the attempt to regulate marital relations, and averted their gaze from the ‘private’ world of domestic relationship, except insofar as husbands and wives sought out the advice of experts for their difficulties.

It was Bott, as we have seen, who opened up this space for alternative analysis, under the frame of an emerging sociology. By interpreting domestic relationships not in terms of the psychological states of marital partners but in terms of their social relations, the means were developed to provide a different framing of the issues which relied less on a liberal framing, and as the subsequent history of anthropology and sociology demonstrates, could be receptive to feminist and radical interventions.

In this process, Bott was also to make the concept of class available as a means of finding a non-psychological use for the in-depth interview. By making it a matter of scholarly interest to consider what people themselves thought about social class, she provided a new justification for the in-depth interview. Rather than class being a means of experts ‘placing’ people, class now became a subject worthy of study in its own right, hence leading to the elaboration of ‘class talk’ that continues to this day. Bott hence marks both the failure and the success of the psy-sciences: it is this which makes her so emblematic. After providing her ‘sociological’ account of social networks, the very last paragraph of Bott's book suddenly sees her return to the need for psychology.

To understand Bott's work, we should not see her as a nascent sociologist or anthropologist so much as a core Tavistock researcher and loyalist. This helps to explain why, having decided to undergo therapy at the Tavistock in 1956, she later retrained as psychoanalyst. As she has recently recalled (Bott-Spillius, 2005), we should not see this later career as being at odds with her earlier work: it was entirely in keeping with her earlier interests and concerns.

Acknowledgements

I would like to thank Libby Bishop and Louise Corti for helping me access relevant material at ESDS Qualidata, and the Leverhulme Trust for granting a Major Research Fellowship to allow me to work on the archive. I would like to thank the audiences at several seminars for their helpful comments, and Helen Hills, Niamh Moore, Tom Osborne, Nik Rose, Carol Smart, John Scott, Sara Delamont and especially Elizabeth Bott-Spillius for their reflections on earlier versions of this paper.