Chapter 1

GENERAL APPROACHES TO THE ANALYSIS OF COMMUNITIES

1:1 The myth of the community.

The belief that stable and tightly-knit communities have existed in the past and still survive in distant lands is an important myth for industrial and highly mobile societies. It is therefore no coincidence that it was in the turmoil of late nineteenth-century industrialization that the idea of 'community' as opposed to modern 'society' was developed extensively, particularly in the work of Tonnies (1887). It was felt that society was changing, values were being undermined, an older closeness was being lost. This powerful myth both influenced, and seemed to find support in, the work of historians and anthropologists during the first half of the twentieth century. Westerners Visiting remote areas of the world were able to discern those 'communities' which were already just a memory in their own society. Commenting on Indian villages, for example, the anthropologist Srinivas remarked that 'nobody can fail to be impressed by the isolation and stability of these village communities' (Srinivas-1960: 23). The work of social and economic historians often pointed to a community-based society, later destroyed by industrialization and urbanization. As Tonnies described it, the emphasis in 'community' was on blood (bonds of kinship), place (geographical bonds), and mind (the sentiment of belonging to a qroup). All these elements, he argued, had been broken in the transition to modern 'society'. Community in this sense could be defined as 'a territorial group of people with a common mode of living striving for common objectives' (Ruth Glass in Frankenberg 1966: 201)

Yet, despite the assumed existence of 'communities', it has been very difficult to define the term more precisely. A recent summary of the contributions of various sociologists to 'community studies' has concluded that the concept of community has been the concern of sociologists for more than two hundred years, yet a satisfactory definition of it in sociological terms appears as remote as ever' (Bell & Newby 1971: 21). One survey of the very large literature using the concept of 'community' considered ninety-four different definitions, yet was forced to conclude that 'all of the definitions deal with people. Beyond this common basis, there is no agreement' (Bell & Newby 1971: 27 quoting Hillery). Even this minimum definition is not satisfactory since there are, for example, 'community studies' of animals other than man. Another minimum definition that has been suggested is that 'community implies having something in common' (Frankenberg 1966:238). This appears to be the original dictionary meaning of the term. Yet, having something in common does not necessarily imply 'community' in any sociological sense of the word. If it did, then all red-headed persons or all suicidal maniacs would be a 'community' and the term would be practically meaningless.

Two recent attempts to discuss the 'myth of community' and the definitional problems associated with it deal with this topic in a way that is not possible here (Stacey 1969; Bell & Newby 1971: ch. 2). One of their major theoretical steps is to differentiate the geographical and the social aspects of community studies which had been merged by Tonnies and Frankenberg in the discussions alluded to above. Stacey has pointed out that though they may overlap, the social relationships within a defined geographical area are theoretically distinguishable from the sense of 'belonging to a group' which such Physical proximity is said to entail (1969: 135). In fact, a 'community', she argues may be geographically based or it may not. Furthermore, since sociologists are interested in the social relationships rather than the geographical space, it may be mistaken to demarcate the area of interest on the basis of physical space. Stacey argues that 'our concern as sociologists is with social relationships. A consideration of the social attributes of individuals living in a particular geographic area is therefore not sociology, although it may be an essential preliminary to sociological analysis' (1969: 136). Or, as Bell and Newby put it, 'a community study must be concerned with the study of the interrelationships of social institutions in a locality' (1971: 19).

The second fundamental point made by Bell and Newby is the distinction between 'community studies' as a method of study, in other words as a preliminary to further investigation or a way of collecting relevant data, and the community study in which the 'community' is treated as a theoretical concept. In the latter case it is assumed that the concept of 'community' reflects some reality in the observed and external world; that it is a heuristic concept that makes the phenomena more intelligible. Unfortunately, the two uses tend to become confused since it is likely that the method will lead to the collection of data which persuades the investigator that he really is studying a 'community' in the second sense outlined above. Despite the confusion, it is essential to keep in one's mind that ultimately the 'community study' approach can have these two senses. It can either be the selection of a unit of observation, for example 1,000 persons at one point in time, or the analysis of a unit which it is believed has some internal structure which is more than random, in other words a 'system' of some kind.

1:2 The community study as a method.

The present work is designed as a contribution to the use of 'community' in the methodological sense, as a means of collecting and organizing data. It makes no assumptions about the actual existence or absence of 'communities of sentiment' or any other kind of community among the objects of its study. In common with a number of disciplines, it shares an interest in the methodological usefulness of taking small, bounded, collections of items, whether human or non-human, as a convenient focus for analysis. Among the disciplines interested in this approach, the following may be selected at random; social anthropology, sociology, archaeology, history, ethology, ecology, genetics, demography. These are but a few. The main purpose of listing them is to stress that it is the general method, rather than the nature or existence of any supposed 'community' of a specific kind, which is our principal concern. It should also be apparent that any general method of community analysis developed within one discipline will have repercussions on all the others. For example, the development of the 'participant-observation' fieldwork method in British social anthropology in the early part of the twentieth century came to influence both sociology and the study of animal behaviour. Any attempt to devise a general method must therefore be wide enough to deal with data which is intrinsically very different from that for which it is specifically constructed.

A second reason for briefly listing the disciplines interested in this approach is to stress that the following discussion, which is centred on the three disciplines with which we are most familiar, namely social anthropology, sociology and history, only covers one corner of the field. Yet, even to survey such an area means a wider spread than that attempted by whole books devoted to surveying community studies. For example, the work by Frankenberg only considers one discipline, sociology, and, within that discipline, one geographical area, Britain. Bell and Newby widen their scope to include American and European studies, but still remain essentially within the discipline of sociology. Neither of these books attempt to take account of the very considerable literature on non-western 'communities' produced by social anthropologists, or on 'past' communities by historians.

1:3 Some criteria for measuring community studies.

The numerous 'community studies' undertaken in Western Europe and North America have already received considerable, attention in textbooks (Frankenberg 1966; Bell & Newby 1971). To these we might add large numbers of monographs by social anthropologists on nonwestern peoples. These range from the early works on the Trobriands, Tikopia and Nuer, through studies of villages in the Himalayas, India, Mexico, New Guinea and elsewhere, to recent studies of Latin American, African and other 'communities'. There has also been a tradition of community studies by English historians, which has recently been revitalized by American, French and other studies which will be discussed below.

The size of the 'communities' studied by these three disciplines varies considerably. Sociologists tend to take the largest unit. For example, Middletown, studied by the Lynds, grew from 11,000 to 35,000 persons over the period of analysis; Banbury consisted of some 19,000 when it was first studied. Yankee city consisted of over 16,000 persons (Bell & Newby 1971: 85, 181, 106). The figures are misleading, however, for the projects were based on small teams of researchers. Banbury, for example, was studied by a team of three; there were at least 18 persons doing fieldwork on Yankee city, a ratio of less than 1 per 1,000 of the population (Bell & Newby 1971: 101, 106). When only a single investigator is involved, the population is often very small. Gosforth numbered a little over 700 Ashworthy a little over 500, when Williams studied them. About 1,000 to 3,000 per investigator is the norm among social anthropologists. Leach studied about 500 people in depth among the Kachin (Leach 1954: 66); Tikopia contained roughly 1,200 persons when Firth did his early fieldwork (Firth 1939: 41); Pitt-Rivers' Andalusian township contained about 2,000 persons (1954; 4). Such communities are believed to be representative samples within a larger population of, for example, 300,000 Kachin or 200,000 Nuer (Leach 1954: 3; Evans-Pritchard 1940: 110). The size of communities in the past at one point in time is less helpful when considering the works of historians. Obviously, the cross-sectional population needs to be multiplied by the number of years over which the population is studied. Thus, Hoskins studied the Leicestershire parish-of Wigston Magna which had an estimated population of about 400 in 1563; but his study covered a period of over nine centuries (Hoskins 1957: 177).

The three communities studied by Spufford contained nearly 2,000 people, but were only analysed in detail for two centuries (Spufford 1974). The work by Wrigley on the parish of Colyton was based on between 1,500 and 2,000 persons, but only one aspect of their lives in the four centuries under consideration was singled out for analysis, namely the demographic (Wrigley 1966). The work by Greven on an American parish covered a population that averaged between 1,000 and 2,000 over a period of about one and a half centuries (Greven 1970). The general impression one receives is that the ratio of 1 investigator per 2,000 people or less is essential in order to undertake really intensive 'community studies' by conventional methods.

As stated above, the number of persons to be studied has to be interrelated with the temporal dimension of the study. It is well known that the absence of information about the past, as well as a certain theoretical framework, limited early social anthropologists to a study of their 'community' at one point in time. Later in the history of the discipline there was a growing interest in change and time. Thus Firth studied Tikopia on a number of visits over half a century. Land records enabled Kessinger, Leach, Obeyesekere and others to study economic and social change in India and Sri Lanka from the middle of the nineteenth century onwards (Kessinger 1974; Leach 1961; Obeyesekere 1967). Yet it still remains true that most social anthropologists study a society intensively over a period of one or two years, supplementing this with small amounts of data on the past.

A good deal of the early work on western societies was also 'timeless', for example that by Warner and his team, and by Arensberg and Kimball. But a growing number of studies included a considerable time depth. The Lynds studied 'Middletown' in America over the period 1890-1924 and made extensive use of historical records. Williams studied land ownership and social change in Ashworthy from the middle of the nineteenth century. Lison-Tolosano studied Belmonte de los Cabaleros from the sixteenth century onwards. John Davis' recent study of land and family in South Italy covers the period 1861-1961 in some depth (Bell & Newby 1971: 82-91; Williams 1963; Lison-Tolosano 1966; Davis 1973) A hundred years is a long time in sociological or anthropological work, but some historians have studied communities for much longer periods. Perhaps the longest is the work by Hoskins on a Leicestershire parish, which is analysed in some detail from the eleventh to nineteenth centuries (Hoskins 1957). Or again, Spufford considered certain economic aspects of Chippenham in Cambridgeshire from the thirteenth to nineteenth centuries(Spufford 1965).

Six hundred years is not out of the question for a historian; but as yet there has been no true marriage of the two modes, sociological and historical. The very intensive study of daily interactions and everyday thoughts which is the hallmark of social anthropology and micro-social history has not been achieved over long time periods. The explanation lies partly in the very nature of the data, but also in its sheer bulk. It is possible for a single observer to watch one thousand persons interacting over one year; but to study these same thousand persons over two hundred years or more would be impossible. It would be the equivalent of studyinq a town of 200,000 at one point in time, a project which would immediately strike any investigator as impracticable, even if the data were good enough. There are, therefore, considerable technical problems to be overcome before the various social sciences interested in 'communities' in the present and past are able to collaborate effectively.

The degree to which the 'community' is closed or bounded both geographically and socially varies from discipline to discipline. It seemed reasonable to assume, when studying the small island of Tikopia, or the Trobriand islands, that there really was some kind of bounded community. The societies visited by the first social anthropologists were often isolated. Their fieldwork methods were evolved in this context. One of the main theoretical problems for that discipline and others has been the transfer of early techniques to situations where there is no geographical isolation. At first sight the problem might seem least acute when the methods were applied to communities in the past. Medieval villages, for example, might seem to fit the criterion of isolation rather better than modern industrial nations. The concept of an isolated rural community seemed to be based on reality. Now, we know that the idea is largely a myth. The very great degree of short range geographical mobility in England from at least the fifteenth century and the interconnectedness of economy and society from at least the thirteenth century is now well established, for instance by Postan's work (1973). Any particular community in England in the past was probably no more isolated than a Chicago suburb or twentieth century Banbury. For this reason, the theoretical problems afflicting modern sociologists and social anthropologists apply with equal force to historians.

The way in which we locate 'communities', in other words the indices we employ to measure them, has been a topic of more concern to sociologists and social anthropologists than to historians. The latter tend to fall back on some unit of administration such as the parish. Some suggestions concerning the ways in which a community could be measured were made by Redfield in his work on the Little Community. He suggested that they must be distinctive, small in size, homogenous and economically self-sufficient (1957:4). By these criteria, many of the so-called 'community studies' are not studies of communities at all. Elsewhere Redfield suggests the following indices.: frequent interpersonal contacts; wives taken from within the area; group feeling in political emergencies (1957: 120-1). He supplements these by suggesting as indices, a common name, common sentiment, the payment of debts within the 'community (1957: 118). Again the definitions appear to be either too blunt to be helpful, or too tight to enable us to study anything except a completely closed island. A more recent attempt to list some of the criteria to be used has been made in a book which applies some anthropological techniques to a Spanish town. Pitt-Rivers lists some of the reasons which makes him think that his Andalusian pueblo is a 'community'. There is a sentiment of attachment to the pueblo which stems from membership at birth; folklore reflects pueblo divisions; there are minor dialect boundaries between pueblos; there are minor differences in material culture, for example clothing; outsiders are often treated roughly; ritual and religion is localized, for example there are local patron saints; reputations are common property within the pueblo and gossip is strong (1954: ch. 1). Pitt-Rivers, concludes that 'the community is not merely a geographical or political unit, but the unit of society in every context' (1954: 30-1). It would not be difficult to add indices to the list. There is the area within which mystical sanctions such as cursing and witchcraft are effective. There area within which shame and guilt are felt, the 'moral community' as Campbell describes it (1964: 259, 310). Other major indices are as follows: shared conscious models: areas of economic exchange; areas of joint agricultural activity; the marriage arena; the area for recruitment to rituals; dialect; costume; social treatment of the dead; area for support in various crises; administrative area; field for informal social control; range of gossip and scandal; area within which the prestige hierarchy operates.

If one were fortunate enough to find an area where all these criteria overlapped, it would look something like figure 1:1. Such an overlap has been suggested by Skinner for Chinese 'standard market communities' but unfortunately, for most investigators, many of the indices do not normally overlap symmetrically (Skinner 1964). This is fairly self-evident when we are studying industrialized and urbanized societies, but it may come as a surprise that often there is little overlap even in rural non-western and 'traditional' societies. The situation we tend to find in many cases is that illustrated in figure 1:2. A classic

Figure 1:1

illustration of the lack of overlap in a non-urbanized society is that described for rural Bali by Geertz. Taking seven indices (shared obligation to worship at a given temple, common residence, ownership of rice land lying within a sinqle watershed. commonality of ascribed social status or caste, consanguineal and affinal kinship ties, common membership in one or another 'voluntary' organization, common legal subordination to a single government administrative official) Geertz looked at specific Balinese villages'. He found that 'virtually nothing is coordinate with anything else and the crisscrossing of loyalties reaches an almost unbelievable degree of intricacy' (1959: 1001). In other words, those who worshipped together were not the same as those who farmed together; those who farmed together were not the same as those with a common residence;

Figure 1:2

those who lived together were not the same as those with kinship ties and so on. We are therefore left with an incredibly complex set of 'planes' which may or may not overlap. One way of putting the problem on one side is to talk of the 'economic community' or the 'moral community', the 'marriage community' or the 'community of gossip'. one can, in fact, make 'community! into an activity-specific word.

The difficulties are further compounded by the fact that there are different 'levels' of community. Redfield has distinguished three of these; there is the community of what ought to happen (the moral community); the community of what is thought does happen; the behavioural or statistical community of what does, if counted, happen (1957: 44-5). As with the various indices which we have discussed above, these may overlap. People may behave statistically in the way that they are thought to behave and it is conceived that they ought to behave. Yet it is often the case that the three different levels are entirely at variance. This 'anomic' situation, as it is sometimes termed, is likely to cause confusion in the investigator as well as in the people who exhibit the conflicting patterns. It is no longer clear how the analyst is to fit together his informant's ideas of how people ought to behave alongside the statistical situation which emerges from counting cases. The conflict in the observer partly arises from the fact that he is likely to be collecting two entirely different kinds of data. The central methodological characteristic of the 'community study' approach appears to be the total involvement of the researcher in a delimited area, geographical or social. With a ratio of 1 investigator to 1,000 or so, it is possible to get to know most of the members of the population under observation and to indulge in the 1participant observation' characteristic of social anthropology. But two other major kinds of material are collected; statements by informants mainly at the level of what ought to happen, and the observations, including the counting of instances, made by the investigator. It is when these various kinds of information clash, as they often do, that both the most rewarding and the most difficult work begins.

1:4 Some criticisms of the community study.

While 'community studies' proliferate, objections to both the concept of 'community' and the methodology have grown. A number of these criticisms may be stated briefly. It is argued that such community studies are, in practice, non-comparable and non-cumulative. They tend to be more like novels or works of art than the objective products of a supposedly rigorous 'social science'. Each study has to be treated on its own merits and sheds little light, it is held, on other areas. It is further argued that no amount of such micro-studies will help us to piece together the macro-structure of the whole society. As Wolf pointed out some years ago, 'we cannot hope to construct a model of how the larger society operates by simply adding more community studies' (1956: 1066). The same criticism was made by Freedman when he commented on Radcliffe-Brown's influence on Chinese studies. The latter's belief was that 'from this patient induction from studies of small social areas would emerge a picture of the social system of China. Of all the biases to which the anthropological approach has been subject this seems to me to be the most grievous. It is the anthropological fallacy par excellence' (Banton 1966: 124). A third criticism is that the concepts and methods were developed for the study of communities which were assumed to be isolatable and geographically located. Such communities may once have existed, but they are seldom found today. Nor are they easily found in the past records of western societies. It is frequently pointed out that this lack of hard boundaries is more widespread than the work of some early social anthropologists would lead us to believe.

A fourth criticism is that 'communities' tend to lie in the eye and methodology of the beholder. As Bell and Newby have pointed out, the participant observation method and intensive attention to personal interrelations tend to create in the sociologist's mind, if nowhere else, a sense of 'community' and integration. He will find community bonds and community sentiments because he expects to do so. On the contrary, mass observation techniques of census and questionnaire tend to overlook interpersonal bonds and sentiments of 'belongingness'. Consequently, the latter methodology emphasizes the atomistic, individualistic, mobile nature of western urban life. If social anthropologists had only used the questionnaire and census in Tikopia or among the Nuer, while sociologists had lived in urban areas for a period of years and noted interactions, it seems likely that our whole picture of the two situations would have been reversed. A fifth criticism is that, because the early community studies were by-products of early structural-functional anthropological analysis, they were timeless, overintegrated,. 'equilibrium models' which took little account of change or of conflict. Anthropological and sociological workers, in their attempt to escape from ethnocentrism, suffered from what Bierstadt has called 'temporocentrism', which he defines as 'the unexamined and largely unconscious acceptance of one's own lifetime as the centre of sociological significance, as the focus to which all other periods of historical time are related' (Bell & Newby 1971: 63). Although social anthropologists were at first merely making a virtue of necessity in the absence of records of the past, this meant that the tools they developed, which were later applied by sociologists to western societies, were not adequate to deal with societies which do have extensive records of the past. The subjectivity of the whole 'community study' approach was shown dramatically by the famous case where two investigators lived in the same town in Mexico at fifteen-year intervals. They found entirely different 'communities' because of their differing interests (Foster 1960).

1:5 Attempts to improve the concepts.

If the old tools appeared to be too cumbersome to deal with rural Mexico or Bali, it is not surprising that when, in the 1950s, they were used in studying Indian townships, Norwegian fishing communities, or the suburbs of London, such concepts seemed totally inadequate. It was obvious that even if one wished to study a particular geographical area, much more flexible indices and concepts were needed. The difficulties were summed up by Wolf when he wrote that 'Communities which form parts of a complex society can thus be viewed no longer as self-contained and integrated systems in their own right. It is more appropriate to view them as the local termini of a web of group relations' (1956: 1065). Although he introduced the term 'broker' groups, and his use of the word 'web' foreshadows the later term 'network', Wolf was unable to solve the theoretical difficulties. A number of attempts have subsequently been made to create a more flexible and subtle set of analytic categories for the study of highly mobile societies which are still believed to contain 'communities' of some kind.

One of these attempts was published in 1957 in Turner's work 'Schism and Continuity in an African Society'. Earlier discussions of roles and actors, based on an analogy with drama, were taken further with the concept of the 'extended case study' or 'social drama. The latter was defined as 'a limited area of transparency on the otherwise opaque surface of regular, uneventful social life. Through it we are enabled to observe the crucial principles of the social structure in their operation, and their relative dominance at successive points in time' (1957: 93). Turner defined the processes through which a social drama would go as firstly a breach of regular norm-governed relations; crisis; redressive action; reintegration or recognition of schism (1957:91-2). Such an approach had been implicitly employed by a number of anthropologists, but the more explicit discussion seemed to free social anthropologists so that they could study minute processes over time rather than merely take a timeless cross section. Combined with the 'case-study method' , in which anthropologists were exhorted to gather material concerning 'a series of connected events to show how individuals in a particular structure handle the choices with which they are faced' (Van Velsen 1967:14) , this made it respectable to analyse individual actions and motivations. Although there were the dangers of degeneration into a narrative and literary mode of pure description, the concentration on a particular event, rather than on a particular group or larger unit, appeared to make a more subtle analysis of life in small 'communities' possible.

In practice the focus of the 'social drama' approach tended to be on crises of various kinds. It was readily apparent, however, that the method could be generalized to cover a much wider range of 'events', including political processes, both formal and informal. One of the most notable attempts to widen the concept was that made in 1963 in an article on the 'Significance of Quasi-Groups'. Mayer pointed out that for the study of highly mobile and 'complex' systems, it is necessary to move away from the earlier emphasis on enduring 'groups' towards the study of what he termed 'quasi-groups', by which he principally meant the 'action-set'. The action-set 'is not a group ... for the basis for membership is specific to each linkage, and there are no rights or obligations relating all those involved' (in Banton 1966: 109). It is, in fact, the 'set' of people who are mobilized in a certain situation. It is not all of a person's potential contacts, but those people who are called on in a particular faction struggle, crisis, or other event. If a number of such 'action-sets' overlap in membership, the begin to form into a more enduring unit which Mayer terms the 'quasi-group' since it lies half way between the entirely temporary action set and the permanent 'group'. These action-sets were often very small and, argued Mayer, centred on a particular individual who brought them into being on a specific occasion. When combined with the 'social drama' approach, sociologists now had better tools for studying much more fluid and complex structures.

Nine years before Mayer presented his paper, Barnes had introduced the concept of 'network', which Frankenberg describes as '.the first major advance in the language of sociology since role' (1966: 242). Bell and Newby agree that the concept of network is very important indeed for the future of community studies. They state that 'what little empirical data there is directly relating to social networks leads us to believe that it is indeed a powerful analytical tool' (1971: 53). The term and the concept were not entirely new when used by Barnes. Nor was the need for such an analytic device concealed. The year before Barnes' seminal article was published, Redfield gave some lectures on the Little Community in which he commented on Rees' Welsh study in the following words. 'The hearth of the lonely farm is the only centre. There is no community centring upon town or village; there is only a double network of kinship connection and neighbourly connection to hold together, loosely, people who dwell separate front one another...there are no fixed groups' (1957: 6). Despite this earlier use, it was only in Barnes' study that the term was given serious analytic content. During the last twenty years a very considerable literature on 'network analysis' has grown up, and some of the major readings are indicated in the bibliography to this book. A number of definitional and substantive battles have been fought. This is not the place to go over this complex ground, but it is important to sketch in one or two of the landmarks.

The original classic definition by Barnes, when attempting to analyse his Norwegian community, was as follows. He isolated three regions or fields in the social system, the third of which is 'made up of the ties of friendship and acquaintance'. He continues that 'each person has a number of friends, and these friends have their own friends; some of any one person's friends know each other, others do not. I find it convenient to talk of a social field of this kind as a network. The image I have is of a set of points some of which are joined by lines. The points of the image are people, or sometimes groups, and the lines indicate which people interact with each other' (1954: 43). As Frankenberg pointed out, the concept was very similar indeed to that of the 'topological graph' as used by mathematicians and it is perhaps no coincidence that Barnes was a mathematician by training. Later investigators attempted to make further distinctions, principally between the 'general network' of all potential and actual links, and what is sometimes called the 'personal' or 'ego-centred' network on the other. Barnes rejects the latter terms and refers to them as 'stars'. Further elaborations have been made not only to make it possible to differentiate the focus of the network, but also to distinguish degrees of distance from any selected individual in a given 'network'. A growing number of criteria have been suggested as to how interpersonal relations should be measured (interpersonal criteria) and how the overall shape of networks (morphological or structural criteria) should be compared. The 'diversity of linkages', content of linkages, directional flow, frequency of interaction, duration, intensity of interaction, are among the major interactional criteria. Size density, degree of connection, centrality, clustering are some of the measures suggested for analysing the structure of networks.

While most sociologists agree that the network concept is extremely powerful no one has et found a way of utilising it properly. The data-gathering and analysis are so arduous that traditional sociological methodology is inadequate. The point has been made by all those who have attempted to undertake a network analysis. For example, Mitchell has commented that the 'study of personal networks requires meticulous and systematic detailed recording of data on social interaction for a fairly large group of people, a feat which few fieldworkers can accomplish successfully' (1969: 10-11). Or again, he writes that 'the characteristics of the personal network must then be abstracted from the field notes. But the interaction is often so complicated that even the most gifted fieldworker stands to miss a good deal. Some systematization of the categories of information to be recorded would obviously improve the quality of the analysis' (1969: 33). Perhaps the most serious attempt to undertake a full network analysis is that by Boissevain. His findings suggest that the approach is extremely difficult in practice. He points out that 'one of the major unresolved problems in the use of networks (is) size. Social anthropologists as yet lack the methodological sophistication needed to tackle this problem' (1974: 71). He further recounts how he 'began with two informants in 1968 on a pilot study basis, planning to branch out and test findings more systematically on a wider sample. Collecting this data proved to be very difficult and very time-consuming, as did its analysis. Hence, for better or for worse, I have data on only two first-order zones' (1974: 97). For example, one informant had 1,750 persons in his 'star' or 'personal network'. Thus to calculate the degree to which they interacted with each other needed a matrix of 1,750 by 1,750. As Boissevain commented, this alone 'reached the memory limits of all but the largest computers' (1974: 36). To gather and analyse the data for the much more intensive type of work suggested by concepts such as 'social drama', faction-set', 'network' suggests that the organization of sociological and anthropological research will need to be changed. It is perhaps no coincidence that at the very time when the new conceptual tools, some of them derived from mathematics, become widespread, the power to use them has begun to be available in the form of computers.

1:6 Attempts to improve the quality of the data.

The various criticisms made against 'community studies' as well as the growing demands for richer materials in order to test out the conceptual tools outlined in the last section, suggested that data of a new kind was needed. Information on much larger numbers of people over much longer periods of time was required in order to test out hypotheses and to avoid charges of ehistoricity, statistical insignificance and so on. A similar situation occurred within demography, where the need for better data was felt by French demographers in the 1950s. Some of them turned to material from their own nation in the past. In the archives they found data of a quality and duration which they could not obtain from elsewhere. It is therefore not strange if those seeking to improve the quality of anthropological and sociological analysis should turn to past records in order to see if they contain material on interactions of a sufficient quality and quantity. The first section of the book below attempts to see, within a very specific context, to what extent the data does exist. Yet 'data' is not just 'out there' to be harvested, it is not a finite quantity, but rather an organic and infinite growth. Its quantity and quality will depend very considerably on the simple techniques whereby it is collected. It is for this reason that technological developments in data collection, for example in such apparently mechanical things as photography, tape-recording, and computers, will have a very great effect on the very nature of the information that seems to be objectively waiting recovery. One example of perhaps the most ambitious attempt ever made to study a 'community' will show the way in which a certain methodology subtly influenced the type of questions which could be asked of the apparently objective data.

In the 1930s Warner and his associates undertook the study of Yankee city. This highly ambitious project is well described by Bell and Newby (1971: 104-6). It was based above all on the 'Social Personality Card', onto which data collected by observation, interview and from documents and newspapers was collated. We are told that 'There was one card for each adult in the community. On them were recorded, name, residence, age, sex, social status, occupation, maiden name of wife, names of children, membership of cliques and associations, church affiliation, type of house, newspapers and magazines taken, movies attended, doctor and undertaker and summary budget data... Also any information from public welfare and police records relating to each individual was added'. These data were punched on to machine-readable cards and described statistically' in the five ensuing volumes on Yankee city. The general aim was a total study; 'the general objective of our research was to determine the complete set of social relations which constituted Yankee City society' (in Bell & Newby 1971:105). The data, collected over the years 1930-4, was based on 16,785 individuals. We are told that 'Warner appreciates that the key problem about his Social Personality Cards is that they are "data centred on individuals". So whilst they lend "themselves readily to statistical compilation and correlation of attributes and relational characteristics of individuals... difficulties arose, however, when it became necessary to compare one relation between individuals with another"' (Bell & Newby 1971: 106). The methods of data collection and data analysis in this colossal study were obviously crucial. As with the method of gathering fieldwork notes in social anthropology, the method had an enormous influence on the final results.

In social anthropology, there have been various technical developments which have revolutionized the gathering of material since the Second World War. Some of these innovations and their consequences are interestingly set out in a recent work by Chagnon where the use of cameras and computers, as well as more traditional fieldwork methods, are discussed with unusual claritv and in unusual detail (Chagnon 1974). His work, as well as the Yankee city study, emphasize the well recognized point that the collection and subsequent analysis of the data on 'communities' cannot, except at a very theoretical level, be held apart. This is especially so in the case of one of the major tools to play a growing part in improving both the quality of the original data and the final results, namely the computer.

1: 7 Conclusion.

This brief survey of some of the history of community studies has only touched on a few of the more interesting studies. It is necessarily biased towards certain disciplines and certain problems. It does reveal, that while the 'community study' approach can be criticized on many levels, the method of studying small, delimited, sets of people or other objects is of fundamental interest to many different disciplines. Furthermore, it suggests that some advances could be made if we could find large quantities of good data over long periods of time which were suitable for manual and computerized analysis. The ensuing chapters are a preliminary attempt to argue that such data does exist and to explain how it is currently being analysed by hand. Of necessity, this report on a current project cannot serve as a manual, but it may prove of interest to the considerable number of people who are involved with historical data, or the analysis of small groups. It is partly designed to fill the need expressed after a conference on record linkage. Laurence Glasco remarked that 'What was lacking however, in most of the presentations was a sense of detail that would allow one to follow the researcher step-by-step through the solution of a specific problem. It would have been helpful to follow even one such example, observing the exact layout of the basic sources (persons working with census materials have only a vague idea of what parish records look like, and vice versa), a flow-chart which would apply to all possible cases one's sources might present, how the material looked at each intermediate step, and how inconsistencies were resolved'(Glasco 1971). Accounts of how to undertake hand analysis of parish registers have already been provided by Fleury and Henry for France, and for English censuses and parish registers by Wrigley and his collaborators (Fleury & Henry 1956; Wrigley 1966: ch. 4). No satisfactory manual exists for many other kinds of record.