A background note on Gtree

A background note on Gtree. (J. Davis)

BICA Issue No. 3: July 1985

Zuwaya live mostly in Eastern Libya (previously Cyrenaica) in the coastal town of Ajdabiya and in the oases of Ajkharra, Tazarbu, Rabbiana and Kufra. They call themselves a tribe (qabila) and are referred to as one in official documents and private conversation; and they speak of lineages (qabila, aila, haush) even though I think Professor Peters would argue they are not entitled to do so: their sections and lineages do not meet all the criteria he thinks necessary. In practice all Zuwaya can be placed on a patrilineal genealogy of thirteen generations maximum depth (in some of the more isolated oases of the Kufra complex the genealogy has eight or nine generations).

Collecting it was a useful way of beginning fieldwork: it provided a relatively harmless topic of conversation which required a relatively limited vocabulary; it was a way of meeting a large number of people, and of making an initial survey of the Zuwaya. The genealogy itself, which I wrote down in Arabic, became a text of interest to Zuwaya who asked to see it, amended it, identified lacunae, and so on. When started off as a preliminary activity became a permanent theme of research. More than once it got me off a hook: denounced and arrested, taken to a police headquarters, the officer in charge asked to see his genealogy; and after conversation and history, corrections and updating with the latest births, let me off with a caution ('You are very silly to make bad jokes in the presence of outsiders')

The genealogy would have been worth the effort purely in practical terms; but its anthropological value is enhanced if it can be analysed accurately and efficiently. Simple descriptive statistics are difficult to construct by hand: the 160 lineages of the Zuwaya are scattered over 45 different locations, and the men have 82 different occupations. It would have taken weeks to construct a table of lineage and residence for the living Zuwaya, with no guarantee or accuracy. And that was the most basic kind of information. So I coded the information for the computer and got a small grant from the University of Kent to get the coding sheets converted to punched cards (I think this was the last of card-punching machines at UKC: they are now obsolete). The tables of lineage x settlement, produced with SPSS, took 6.4 seconds to create, a few minutes to print. Even that would not justify the five months' labour of coding and checking; and from the earliest stage I provided for further use of the data. I wanted to be able to interrogate it: What categories of lineage provide police officers? What is the kin relationship among schoolteachers in TAzarbu? Are those who died at the battle of Kawz in 1931 drawn from particular lineages? It is possible to specify any relationship of kinship and affinity and to ask 'How many spouses (or waterboard officials, electricity linesmen &c.) stand in this relationship? And then, how many relationships of type x are there between policemen? between spouses? There was a further possibility: the genealogy is patrilineal; but - following a suggestion made by Emrys Peters - I was able to show for one oasis of twenty-seven households that it made sense to suspend belief in the Zuwaya explanation for who lived there ('our ancestors owned it, and we provide the women') and to regard women as the fixtures, the men moving in to marry a core five-generation line of an ancestress and her daughters and granddaughters. This is the 'ideological transformation' Nick Ryan writes of in his paper. To deconstruct and then reconstruct a patrilineal genealogy, even of so few people, took many days of work by hand and head, and would be prohibitively expensive to do for all Zuwaya settlements. I do not think I will ever wish to argue that, although Zuwaya speak patriliny as their principle of residence they are mistaken for the true principle is something other; but it is interesting and may be significant to consider questions such as 'what would be the most economical way of explaining who lives where'. Such playing with data, recasting it in different forms, is extremely time consuming when done by hand and is probably not worthwhile, given human mortality. But the computer, once it is given enough data, will be able to do the job in a few minutes. The programming is now nearing completion. A final task is to provide links between the genealogical information and information from other sources: field notes, other lists and maps; above all an as yet unanalysed set of copies of archives from Kufra and Ajdabiya which contain the marriage and divorce records from 1932-79 and 1943-79 respectively. This archive records details of approximately 20,000 matrimonial events, many of which concern Zuwaya who can be identified on the genealogy.

It may be helpful to describe briefly how the genealogy was coded. Fig. 1 shows one of the drawings from my genealogical notebook. I made these usually during the night or early morning from rough notes made the day before. The coding is fairly straightforward for most of the information: numbers indicate the value of each variable (residence, occupation, sex, civil status and so on). But because I wanted to interrogate the computer about relationships between people, I also encoded their position on the genealogy. The first step was to give a number to each lineage : in the example it is 038. the founding ancestor is then given the identity number 00; and his children are 01, 02, 03 and so on. His eldest children are 0101, 0102, 0103 and so on. This method is convenient in two ways: it allows me to add in further offspring without remembering: some women, for example, are not recorded in genealogies, but do appear in other notes or in documents. The codes are expandible. Secondly, the length of the string of numbers indicates a person's generation; and it is easy to compare strings to identify (patrilineal) relationships:

038 01010102 038 01010203

are siblings; while

038 01010102 038 01010203

are patrilateral parallel cousins. Here is a part of the computer print-out of the coded genealogy, showing part of the lineage in Fig.1:

Table: Part of Usbaq lineage.

038
0204020103
MHMUD
ZAID
MHMUD 038
020402010301
A-RHMAN
MHMUD
ZAID


038
020402010302
ZAID
MHMUD
ZAID


038
020402010303
MHMMD
MHMUD
ZAID


038
0204020104
MFTAH
ZAID
MHMUD


038
020402010401
SLAIMAN
MFTAH
ZAID


038
020402010402
ZAID
MFTAH
ZAID


038
020402010403
MHMMD
MFTAH
ZAID


038
020402010404
HMMD
MFTAH
ZAID


038
0204020105
FAIZA
ZAID
MHMUD


038
0204020106
AZ
ZAID
MHMUD

038
02040202
MHMMD
MHMUD
ZAID


038
02040203
MFTAH
MHMUD
ZAID

Figure 2 is the genealogy drawn by the computer using Nick Ryan's Gtree program. It takes input of this kind and produces a diagram which can be manipulated to produce the most aesthetically pleasing or anthropologically striking effect. When the user points to a particular individual, the program will redraw the diagram with that person as the most junior or most senior person in the line, and will display basic information about residence, status and so on. The program will shortly be able to trace the relationship between individuals.

The program is in fact more flexible than might appear, and will cope with any kind of kinship system. For instance, much of the information about genealogical position as it is encoded here is redundant from the
point of view, since it can construct the genealogy a). if it knows for each eldest child who the father is, who the next sibling may be; and if it knows for each other child who the next siblings (elder and younger) may be; and b). if it knows the spouse of each married person. I could, therefore, have entered the information in a much terser form than I did, with much saving in time and computer memory. But it would be difficult to collect as little information as that in the field and if you did not have a computer with you it would be practically impossible to conduct any preliminary analysis by hand and head. And of course, you would not be able to use the genealogy as a text during fieldwork in the way I found so helpful. At home, and without permanent access to a computer terminal, it is useful sometimes to work from a print-out from the file, and information about lineage position and patronyms is invaluable for human work: more than six hundred Zuwaya are called Muhammad but only three are called Muhammad Muhammad Muhammad; without patronyms it is very difficult for a human to match genealogical data with other records.

The main problem I have in working on this data is remembering that the genealogy is a composite, a variable construct. People give different versions of it in different places and in different situations, and that applies not only to the six-generation band which is 'within living memory' (i.e. an adult says 'My grandfather told me his grandfather said ...'), but also to the top levels purporting to show the main lines of descent from the founding ancestor. It is relatively easy to maintain a sense of the uncertainty and variability of the genealogy when working by mind; less easy to do so when unambiguous and unhesitating data is put into a computer, and such clear and detailed statistics and diagrams come out of it.