1992blongtermdoc.html

Section 1

[ I did not give any of this at the AAA in Nov 1990. I looked at it and began editing 28/5/91, with Moscow in mind. I found it good. Got into 4.]

Paper for AAA Nov. 1990
DRAFT in progress. NOT for publication; nor quotation without consent
Paul Stirling November 1990
University of Kent at Canterbury

ESRC awards G00232121 1985-7; R000231955 1989-90. I am deeply grateful to the ESRC for funding the research on which this article is based; and to the government of the Republic of Turkey for granting research permission.

Households in Social Change in Turkey:
A Long Term Data Base

Plan

1. An Archive; Anti- theory Theory
2. Turkey's Seven Revolutions.
3. The data in 1986
4. Data into Computer
5. The Data Base
6. Uses?
7. An Example: Household Fission

An Archive: Anti-theory Theory

I have for eight years been working on a project to bequeath to future scholars an archive of my study of two Turkish villages, begun in 1949, and still continuing. I now have an almost complete data base of some complexity. In this paper, I describe this data base, and give a very brief account of how and why I and others - have constructed it. This account requires a minimum of self indulgent but fashionable autobiography.

In 1939, after an Oxford year of undergraduate Latin and Greek, I read Ruth Benedict's Patterns of Culture. After the war, I completed a first degree in Ancient History and Philosophy, with a strong sense that these able people were missing important questions. I decided to become an anthropologist in Oxford. In fifteen months, I was supposed to learn anthropology from scratch, plus Islam, Ottoman and Turkish history, and Turkish language. I was bowled over, after a first degree in conventional ancient history and in linguistic philosophy, by social anthropology; by the idea of studying societies, comparatively, as social systems, as social structures. In April 1949, my wife and I found ourselves in Turkey, to study villages. Oxford trained, I was aware of but also suspicious of 'culture', and of grand evolutionary theories. But I perceived myself somewhat arrogantly as philosophically cynical; an all purpose disbeliever. I had grave doubts about 'structure', 'social system', about the way historians and Islamists wrote about history and Islam.

I did not set off with a set of hypotheses. Rather with the idea of applying the Malinowski model of
field work to villagers in a society with history. Besides the works of Evans-Pritchard, Meyer Fortes, and other obvious classics, I had read carefully three or four examples of village studies. And I had received minimal instructions; do not use an interpreter, and take a census of the village. Meyer Fortes said
'Everything that happens in the village concerns you'. Evans -Pritchard said 'Find what they are most interested in and study that'.
I was looking for 'social structure'; groups, networks, and hierarchy. I had
modest ambitions. I consciously did not hope to understand the way the villagers thought and felt with only an outsider's command of language. I felt quite ill at ease with my superficial knowledge of an unfamiliar world religion, and did not realise then how much a mildly uninformed but detailed ethnography of one example might contfibute to its study.

I and my wife dutifully studied two villages, living in appropriate hardship in one for ten months and in the other for seven. Among other things I collected a census household by household. Much too long afterwards, I wrote a modest, honest, and well received ethnography, which for explicit theoretical reasons, did not argue for any specific theory (Stirling 1965).

In 1971, I went back. The same people were there, in much the same situation, lioving much the same kind of life; with of course their new children, and without those who had died. Yet dramatic changes were clearly happening. Unfortunately, Turkey was under martial law, and after waiting for most of my allocated two months, I was finally given permission only for a week's field work. I did a little more, and by using my existing censuses, my village friends, hearsay, and my very co-operative of ficially appointed Turkish 'protector, I collected once again a census not only of the villages, but also of those who by now had moved their households out to live in towns. Out of this experience, I wrote an article (Stirling 1974), and conceived the ambition to return for another serious study.

I found the changes staggering. But I also found myself totally unequipped to deal with them. Of the two current sets of ideas, 'modernisation theory' and neo-marxism, neither seems to have much bearing on my village data. I felt baffled, because I saw so many interesting processes interlocking in the villages, and yet I could not write coherently about them. I argued that the changes came from outside the villages, through a number of channels, and that these set off a very complex set of effects and causes with multiple feedbacks, both negative and positive. I produced a tentative diagram of these processes; but I knew it was seriously inadequate to the reality. Yet no one seemed to take the slightest notice of the theoretical implications as I saw them.

The plan to return for a serious restudy took a long time to realise. When I finally came, in 1982-3,to draft a research application, I made my theoretical misgivings explicit, and based the application on the need to provide as clearly, fully and untheoretically as possible a record on the history of the two villages for future scholars. I wanted to avoid the nommal practice; that is, to use selections of my empirical data as a sort of parable to illustrate my own necessarily transitory and probably second rate ideas and theories. At least, by using the existing censuses, and conducting a new and much more careful household survey, I hoped that I could provide something of pemmanent value. After all, social historians ere invest huge efforts to recover details of village households from previous centuries, details far less detailed than those already in my note books. So the research plan took a somewhat demographic and factual turn.

But I intended to combine this more formal approach with a Fortes interest in everything; listening, watching, following up opportunities, catching innuendoes, asking intimate questions when the occasion seemed right. This double approach was perhaps unfortunate. Not only did the qualitative, opportunistic field work lack enough specific focus, but in the event time was short, and my team concentrated too much perhaps on details of names, dates, movements, qantities, property, and so on.

In 1983, I went to teach in the Middle East Technical University in Ankara for three years, less just over a semester. I had difficulties with research permission from the Turkish authorities. In the end, I was given permission to start work in the summer of 1985, only to lose it again from October till January. Finally, while still teaching in Ankara, I did the bulk of the fieldwork, with four Turkish field collaborators, from Feb 1986 t0 Aug 1986. Definitely not long enough. Two further six week trips in 1989 and 1990, for checking gaps in the 1986 data, were also all too short.

2. Turkey's Seven Revolutions

Turkey was invented, and brought to territorial, legal and human reality in the years leading up to 1923, by a whole complex of processes and events, over which Mustafa Kemal, later Ataturk, presided with heroic determination and astonishing skill. To call all this a revolution, that is, a turning up side down, is misleadingly simple. From say 1920 to 1986, I detect seven major changes, revolutions if you like; but anyone else could number more or less.

Two of these seven are primary, five follow from them. The first is Ataturk's political and social programme of legislated change. The establishment of a sovereign nation State; the change of laws and the judiciary, the education system, the script, the reckoning of time, the system of official names from Islamic to western and secular. The second is the economic change from a relatively poor agrarian society with largely local markets to an industrialised society; if you like, Turkey's industrial revolution. By the mythical constructions of some, a growth in GNP of twenty times.

The five other changes are part of and follow from these. Each requires long discussion, which it will not get. Three, the change in occupations. To run modern Turkey, millions of people had to learn thousands of new jobs. Turkey now has a huge army of teachers, lawyers, doctors, scientists, technicians, managers, accountants, medical auxiliaries, transport workers, shopkeepers, and of course, 'memur' - civil servants of all kind, besides its factory workers and industrialists small and large. Four, to perform these new jobs people had to leave their villages and move to town, and accordingly the towns have grown at a great speed for seven decades. Five, a massive change in formal education. From a society with 10% literacy, 2% high school graduates, and one university, it has achieved 28 universities and over 95% attendance in primary education. Six, overlapping with formal education, but going far beyond it, there is a huge change in social information, in knowledge, from science, medicine, history, general knowledge down to what I call social knowledge in the villages and the new urban settlements, about all kinds of subjects - repairing diesel engines, coping with child diarrhoea, working in Germany legally or illegally, and a whole variety of ideas and beliefs about society, markets, socialism, nationalism, and about Islam and islamic movements. Seventh, what is absolutely crucial to economic growth and social development, Turkey has learned to use, and to use in vastly greater scale, all kinds of organisations and ways of co-ordinating people, getting things done and keeping a growing population, mobile in all senses, in some kind of order.

I have listed these changes not because I think they are not well known already, but because almost all accounts seem to me to treat them in isolation from each other, and thus to underestimate both the overall scale of change in Turkey, and the immense complexity of interacting factors, jostling each other in a series of accidents, side effects and positive and negative feedbacks. The changes, though well under way in 1950, were still only beginning. So my data covers 35 years of intensely rapid social change, in which with immense and important differences, -Turkey experienced the kind of changes which we call the Industrial Revolution in north west Europe and the USA, and which in those countries was a lot slower. Hence my conviction that it is more important to bequeath the data to future scholars, than it is to record my own impressions and ideas.

3. The Data

The data is of four kinds; field notes, censuses, genealogies, and a cadastral survey of one village.

My field notes are daily notes kept whenever I have been in the villages, or visited migrants. They run from 1949 to 1952, 1955, 1970-71; and visits from 1974 to 1983. Then regular visits in 1984 and 1985; and full field work 1986, 1989 and 1990.. I also have daily notes from Ms. Emine Onaran Incirlioglu, from Mr. Mehmet Arikan, Mr. Vahap Tastan and Ms. Hulya Demirdirek.

I took censuses of all households in both villages in 1950 and 1951-2. These gave the members of the household, their relationship to the head, their birth, their marriages, their land, their animals, their occupations, and their labour migrations .

In three days - of ficially - in each village in 1971, I went through my censuses of 1950-52, and asked about each person and their marriages, children and migrations. My Turkish collaborator, appointed by the authorities, kindly wrote all this down while I talked. I myself wrote down what we could find out about migrants. This census depended on hearsay, but from people I knew and conducted in company in a guestroom full of people. So although hasty in the extreme and unchecked, it contains largely reliable materials. I augmented my data by a brief, highly successful visit to Adana, where I met a large number of people from Sakaltutan; and I also visited some migrants resident in Kayseri from both villages.

[THIS PARA IS REPEAT] Visits from 1974 to 1983 were brief; once or twice in most years, for a night or two at a time. In 1983, teaching at METU in Ankara, I was able to visit more often, but still only for very short visits. The politics and bureaucracy of the Turkish State, and my own misjudgment, held up research permission for a whole year. When I was finally able to begin serious work in the summer of 1985, I was still under pressure to teach in Ankara. Then my permission was withdrawn for four months, and I did not actually start the final periods of intensive field work, until February 1986, supported by my Turkish collaborators.

- 4

I had two objectives; perhaps incompatible. I wanted to use our time with villagers and migrants to find out all we could, to watch and listen, to respond to every opportunity. Information which is not asked for is more reliable, and often more interesting.

But I also wanted countable hard data on a large number of topics. I had therefore tried to devise two sets of cards, one for households, the other for individuals. They were intended to be used both for the dta from the 1950/1 and the 1971 surveys, and the new survey. These were complex, and proved moderately unsatisfactory, partly because of the normal difficulty of definition, exceptional cases, and interpretation by four different workers, partly because of the fact that some questions did not work well, and some important points were overlooked. But most of all the sheer clerical time and labour in filling these cards up after periods with the villagers proved a terrible brake on the research. Half way through, we changed to a simpler system, in which we recorded answers to memorized questions straight on to A4 writing pads. These saved time, but postponed problems of comprehension, interpreation, inconsistency and classification for coding till a later stage.

So at the end of August 1986, I had a mass of data in field notes, on cards, and on handwritten interview sheets. A lot of details were missing, and the records were not always consistent. Moreover, in such a study, problems of definition are crucial, both for theoretical reasons, and for simple counting. But some definitions were fuzzy, and others unsatisfactory. Examples. What is membership of a household? How do we decide who is household head? When does a household cease to exist? How do you deal with life history events in a society where birth dates are very often inaccurate or unknown?

4. Data into Computer.

Before undertaking this research, I had had a limited amount of practical experience of quantitative analysis, but none at all of computers. Yet the ambitious and open-ended plan rested on the assumption that computers would be used. I had attempted to plan for them, knowing mistakes had been made, but assuming computers would somehow make recovering from them easier; in fact, the opposite is the case. On my return to UK in the autumn of 1986, I thus faced a much more formidable task than I relaised, and two decisions.

The first decision was in what form to attempt to enter the data. The choice lay between an immediate recourse to a relational database, or first constructing a set of flat files; typing the data in on a word processor, orgnaising it into unlabelled fields or variables, separated by a marker, in iour case, a spece; and by lines. Once satisfactory, these would be moved to any suitable form for publication and storage.

At this stage, with a large number of known problems and the certainty of the emergence of more, we - that is, Dr. Michael Fischer - decided on the second alternative. This decision was, for a number of reasons, correct; but with hindsight, I could have made things easier and quicker for myself had we done them a little differently. This decision also rested on the assumption that I personally would in time learn elementary programming, to the point where I could comfortably rearrange, correct, and analyse the data myself by writing appropriate programmes. This proved impossible.

The second decision was whether to clean the data first, or go ahead at once. I had a naive belief in the capacity of these unknown computers, and Dr. Fischer underestimated the disorderliness, fuzziness and complexity of the data. In retrospect, the decision immediately to employ computer-skilled typists to enter the data direct from the cards, and then to allow an inexperienced research assistant to do the same with the handwritten records, was a serious error. With a large input of clerical time, I had to set to work to correct errors, inconsistencies and omissions, both of content and of layout.

However, my final report on my first award was received with enthusiasm by my assessors. I received a letter of congratulations from the ESRC. For this interim success, I was deeply indebted to Dr. Michael Fischer, who spent hours of unpaid time. So thus encouraged, and needing more help, I raised my sights and devised a new project to improve the final archive. I proposed a retum to the field to check details and repair omissions, to employ research assistance, and, most important, some contribution to Dr. Fischer's time. The award was granted, but Dr. Fischer's extra time was not allowed by ESRC rules. Unwisely, I acquiesced.

One major problem, or should I say ambition, which we shared, was to incorporate qualitative data from my field notes into the archive for future scholars. The representation on computers of qualitative data

is one of Dr.Fischer's main aims. And in my original layout of the cards, I included spaces for a symbol meaning 'see field notes for more information'. Indeed, the business of reducing my friends and informants to numbers worried me; and I was aware of all kinds of doubts and qualifications, which arose from personal awareness of the context.

With time and vast effort, satisfactory files were laid out and duly corrected for the three household surveys. (see attached example). But our efforts to devise a satisfactory format for the individual life histories were unsuccessful. The problems seemed to be three. First, much detail was missing, and often we had inconsistent information about a person at different dates. Clerical errors and bad handwriting in the field were common. Secondly, the complicated and uneven data on migrations, occupations, incomes and assets was very difficult to classify and code without repetitions and unacceptable simplifications. Eventually, the data typed in 1987 by the typists proved largely unusable.

Because of the hold up on the life history format, one major task which derives from my desire to incorporate qualitative data, remains to be dome. I need to go right through forty years of field notes, to augment details and to makes notes about special and interesting cases. This task will call for a series of consistent judgments by me. The problems still lie ahead.

At last, in November 1989, after much discussion, hesitation, and advice from two experts, I accepted an offer from Dr Nick Ryan of the Computer Laboratory to build a database for my data. using the Ingres DBMS on the University Vax Cluster. This decision cut out one stage of the plan to create a flat file for the individual life histories, before creating a publishable database. Dr. Ryan has been working on and interested in historical household and individual data; and since 1987, the University had acquired Ingres, which was more flexible and complex than previous available relational databases management systems.

This whole undertaking aims to produce a data base of historical relevance which will prove not only useful to historians of Turkish social history and anthropologists, but which will in several ways break new ground in the use of computers for archiving research data.

5. Database

The Database on the Ingres relational DBMS then has, for each village, one section dealing with households, and one with individuals.

Dr. Ryan has developed data entry software, using the associated Applications-By-Forms, a so called 4GL system. He has done most of the development on a DEC VAX Cluster, beginning in December 1989, and working on it whenever he can spare time from his other, full-time, duties. This work continues.

Much of the data is derived from the household censuses collected in 1950-1, in 1971, and in 1986. But in 1986, we also attempted systematically to collect data on individual life histories, to augment a great deal of such data already scattered in the field notes.

The S Village records cover all persons alive in 1950, with all their patrilineal descendants and wives, including migrants from the village. The E Village records cover all persons alive in 1951, but in 1971 I asked about only every third household of the 1951 survey. The 1986 records are derived from 1 in 2 of the 1971 households, with the addition of some close patrilineally related households. In both 1971 and 1986, we asked about and, in many cases, visited emigrant households.

Households
box,center; c Ic s Ic s ^ Ic c Ic c l ln n In n. Date S Village E Village _

Village Migrant Village Migrant_ 1950/1 103 2 212 0
1971 137 65 90 42 1986 153 163 57 58
Individual Life Histories

box,center;clclc^lclcllnln. Sex SVillage EVillage_ No. No. _Male 1104 955

Female 1272 1054 Unknown 49 24 _Total 2425 2033

-6

The detailed information for each household and individual record is grouped and displayed through the use of 'fomms'. A printout of these is attached. The main headings are as follows.

Household Forms

1. Head. Location. Observations (surveys).
2. Household members. 1950/1, 1971, 1986, All 1950-86. Birth, kinship to Head, dates of joining and leaving.

3. Agriculture. Land. Animals. Machinery
4. Home. Location. Type of tenure.
i 5. Other Assets. Other dwellings, urban land, businesses.

Life History Forms

1. Identity. Personal code. Birth. Parents or origin outside the village. Death.
2. Residences. Household Codes. Dates of Moves. Position within household.
3. Marriage. Spouses. Dates. Places. Terminations.
4. Children. Dates. Deaths. Miscarriages.
5. Education. Literacy. Schools. Schooling away from home.
6. Occupations. Dates.
7. Pendular migrations. Movements of members of village household to Turkish and foreign places to earn.

For each form there is a space for notes from the ethnographic record. These are used for any interesting comments; and I intend to add information from field notes before publication.

The construction of this complex data base has raised a range of technical, procedural and conceptual problems. Three examples. First, making it possible to ask the DB a unknown range of questions. Second, coping with materials which are often imprecise and inconsistent to missing. Thirdly, coping with a historical time depth on the basis of single observations at specific times. I append the first draft of an unpublished article which Dr.Ryan is preparing on these issues.

6. Uses?

One of my baser motives for creating an archive for others was precisely to duck this question. And certainly the enthusiasm of virtually everyone who knows about my plans encourages me to think that others may well find uses that I have do not have the energy and knowledge to try out, or that I have not even thought of.

One of many ideas that I imbibed with my anthropological Alma Mater's milk was the altruistic duty of ethnographers to record as much as possible of their experience, so that future scholars might use the 'facts' they collect; an idea embodied in the RAI Notes and Queries (1951) or in Murdock's Outline of Cultural Materials (1961). I left out things which I felt I was not competent to handle, which I perceived as too difficult to collect systematically, and which I thought uncountable. But I did set out to list all the general, countable 'facts' about people and about households. I did not ask myself why in every case, but used my accumulated experience of the kinds of information I myself had used and which others use. I was impressed by praise for those whose ethnography provided evidence against their own theories. And I particularly aimed to make possible the use of my data to illustrate or test ideas which I had not myself had.

It is currently fashionable to emphasise the unreliability of social facts, and to invite ethnographers to search their own assumptions and backgrounds for hidden models which they do not but ought to question. I hope I am acutely aware of and sensitive to the propensity of human groups to construct realities which then take on common sense and pseudo-objective certainty. But to quote John Davis at Coimbra: "things

aw as they are, and were as they were." I would pre;c to say they were. One perennial problem; no individual knower can ever be certain where the line between real reality and constructed reality falls.

To make the point differently, we can all distinguish a convincing, careful, clear, ethnographic A
account, a believable set of data, from confused, implausible, plainly naive data. I have an untested
impression that both those most concerned to attack others as naive believers in facts, and those only
committed to a particular theory, are not themselves the best presenters of their own ethnography.
But even if I say that I set out to record a range of interesting real realities for other people to use, I
still need to answer the question - why did I think these bits of information interesting and not others? The
short answer is plain, if unsatisfying. I make use myself of a large number of ideas and models, not all of
them mutually consistent, to which certain bits of information seem relevant; and I also pick up from others
a whole set of things in which they are interested. What is perhaps more important, is the idea, rooted in but
not equivalent to the idea of a social system, that as Fortes said anything may turn out to be interesting.
Antireductionism, if you like; or intellectual indiscipline.
All the same, I can make a sort of brainstorming list of uses of the DB. But first let me scotch one
hoary but recurrent fallacy. I quote myself. 'The chosen community cannot possibly be 'typical', because
there is no such thing. But the model is bound to throw light on other similar communities, either because
the model fits and enlightens, or because the points at which it does not fit, in so far as they are not
explicable by elementary common sense, suggest new problems.' (Stirling 1965 p.25). Both points are
even more true of this study. The seven - plus or minus - national 'revolutions' which I sketched above,
affect even neighbouring and once similar villages in different ways. So a detailed analysis of changes in
one village is in no way secure evidence for generalising. But the constructed models of processes of
change are likely to have wide applications.
But it does provide solid evidence for two very general but to my mind fundamental truths;
fundamental truths that have recently become more accepted, but are still very widely ignored. First, that
simple reductionist accounts of these changes are implausible. The social processes are intensely complex;
arguments that one single factor is more important than all the others, - common candidates are economics,
world view, class conflict, not to mention older, less plausible ideas about geography, religion and so forth
are simply untenable. What happens constantly is turbulent interactions and cross-cutting of a mass of
'factors' of very different kinds, with all kinds of feedback, positive and negative, escalations and controls.
Anthropologists talk about 'holism'; but the point is rather that in any given social field, almost anything
may be happening; anthropologists are marginally more likely to spot the relevance of the unexpected than
one track interpreters or tunnel visionaries.
Second, the overall direction of change is towards greater variety as well as to greater complexity.
Indeed, spontaneous reactions to external forces throws up more variety. So the standard 'disembodied
ethnography' in which we all indulge is even more dangerous.
Fn: Evolution thrives on variety, and if we equate evolution and progress, then progress thrives on
variety; one powerful argument for 'freedom', including a capitalist free market.
The list of questions, or categories, in the 1986 survey ought to be justified item by item. I could do
so, a little unreliably. Pedantic and boring? They were chosen on hunches; but in retrospect, I can state four
principles. First, information which seemed to me successful and interesting from my 1950/1 and 1971
ethnographies. Second, - another egocentric return to the past - in 1974, I published a diagram of 'observed'
connections between 'variables' in these two villages. I reproduce it here. Every connecting line in the
diagram is a causal 'hypothesis'. I used this piece of thinking also as a source of things to ask. Thirdly,
points remembered from other scholars; household types, for example. Fourthly, ideas of my own. For
example, how do young men get incorporated into the building industry? Fifthly, a negative principle, I
omitted things which I perceived as unquantifiable, beyond accurate recording, impossibly time consuming,
or offensive.
Because of the large number of categories, the DB has the capacity to try out a large number of
models and ideas, from specific and simple to vague and grandiose. Thus it will be possible to ask questions

- 8

about changes in age profiles over different time spans; to document differences between actual and registered birth dates; to document changes in the age of marriage, in the patterns of marriage, in the differences of marriage patterns between migrant and villagers, and between new and longterm migrants. It will also be possible to look at connections demographic data and marriage data on the one hand, and such data as occupations, level of prosperity, land holdings. Adding to this casual list is pointless. Some of this may turn out to be simply vaguely interesting descriptive numbers; but in most cases there are causal connections which may relate to other parts of the processes of change.

7. Example: Households, cycles, fissions, and close kinship

In Turkish Village I argued that, in 1950/51, although only quarter of the households at any point were in fact what I now call multiple, that is households containing two or more married couples, the multiple household was the social norm. The reasons for around three quarters of all households at one point in time containing only one married couple that is simple households, were partly demographic, and partly the result of the social rules governing household fission.

The implications of this argument are two fold, but they have not been accepted, nor even noticed, by most sociologists. The first is that the statistical norm in traditional rural Turkish communities is not the 'extended' or multiple family, but the 'nuclear' or simple family. So there can be no statistical proof of a simple change from extended to nuclear with the change from rural to urban. Second, if the numerical preponderance of the 'nuclear' family in the villages does not prevent the 'extended' being the preferred norm, which most normal people experience for part of their life, then people should be cautious in drawing simple conclusions from a similar but larger preponderance of the 'nuclear' family in the towns and cities.

In 1971, I found a considerable number of men had separated from living fathers without no sense of disapproval. Nevertheless the proportion of multiple households in the villages had risen slightly. I was given a new formal rule: one married son, normally the youngest should remain with the old couple, and there should always be one daughter-in-law present. In fact, practice varies widely, and rules are much less heavily sanctioned by public opinion; but on the whole, almost all young couples still begin married life in the husband's father's household, and most senior couples with grown sons have a son's wife living with them.

By 1986, it was clear that no firm rules were applied. People still said a couple should begin married life in the F house, and this happened normally in the village and sometimes in the towns. I found one large multiple household in Kayseri. In the villages, the proportion of multiple households had not declined at all, in spite of the numbers of young men who had separated from their fathers.

The contrast between the village and urban situations is clear. In 1971, rural households averaged about 6 persons, urban households about 4. In the villages, there was little change by 1986, but the urban households were on average larger, just over 4. In the villages, about 1 in 4 households at any time were multiple (see table for details) but in 1971, only 1 of the urban households was multiple. Interestingly, this rose to 11 by 1986.

The reasons for the sharp contrast between size and type between urban and rural households in this sample are several. Two obvious points. The early migrants from the village were young couples with few children. The move out of the village meant leaving their parents behind, and in many cases was their first separation from the multiple household. They could be neither large households nor multiple households in the early stages. By 1986, some urban dwellers had been there for thirty years, and others for lesser periods. Very roughly, the flow to the towns was growing larger all the time, so in any count of averages, the older households are outweighed by new, young small households.

Thus the difference between the profiles of counts of sizes of households and of types of household is largely a direct result not of changes of norms, or correlations between urban and nuclear, but of the demography of household migration, and of the household cycle.

This claim needs qualifying. How people arrange their domestic units does not rest simply on 'norms', but on practical considerations. A landed village household needs labour, especially female labour. This is in no sense true in urban households, except where a man runs a business in which his household members can contribute laboug women mind shops, for example, if the shop is very close to the home. So one motive for staying together is less strong and less common. In the village, many homes are fairly

- 9

spacious, or can easily be expanded; and now many households move out and build on land they own on the village periphery. In the city, some households are squatters who took over sizable plots, and have room to expand in a similar manner. But many live in flats in apartment blocks, where two married couples find life cramped. This does not prevent young couples moving in after their marriage, but it greatly increases the pressure to separate in the first months or years of the marriage. Thus the urban pattern is not necessarily a sharp break from village practice, but an adaptation of it to a new housing situation. Of course, in time norms in the towns may change, but not in any simple way.

All kinds of events disrupt normal practice; urban events are different, and it is of course true that for all kinds of reasons - jobs, the distance between houses, quarrels, sharp differences of income, urban born wives from different backgrounds, - there is much more variety in town than in the village. Thus the number of simple households is in fact very much higher. But what I wish to stress is that the factors operating are several and not stable over time.

One more qualification. As long as a man has his adult sons in his own household he controls several incomes, and the household is likely to be comfortable, and perhaps accumulate savings. If his sons separate, the situation is not clear. Some maintain firm independence and do not help their natal household. Mostly, there is some degree of financial co-operation, as well as friendly visits and such. But in some cases, the group of father and sons may run a joint systems of finances, especially if business activities and investments are involved. In a few cases, such collaboration survives the father's death. All I wish to illustrate here, is that having a thirty-year period of research, and a complex data base, will enable not only me, but future scholars to find surprising numbers, to test ideas, to work out explanations in terms of their ethnographic knowledge from this and other studies.

Addendum

This draft is incomplete, and needs editing. The most important topic omitted, indeed, a very important one, is the inclusion of qualitative data. The provision in every form of a space for notes enables me to include information from my field note books and interview reports. Whether it will be possible to to expand this further is not yet clear.

I have also failed to give sufficient emphasis to the problems of coping with data which is frequently unrecorded, or obviously mistaken, or inconsistent, or imprecise, or unknowable.