free web hosting | free hosting | Business Hosting | Free Website Submission | shopping cart | php hosting


An analysis of contemporary pilgrimage structures

Part 2 (chs. 6-9, references)

Wim van Binsbergen


Part I:
click here in order to access Part I

Part II:




The principal set of people who have a definite relationship with a particular saint are the actual members (i.e. inhabitants) of the territorial segment with which that saint is associated. All these people, male and female, must partake in the routines of the saintly cult, including dedication of meals, at least twice-annual zyara, and observance of the saint’s festival.

            Male members of the segment are not under formal obligations of zyara, although many of them do visit, as individuals, the shrines, and attend the festivals, of the major saints in their own valley and adjacent valleys. Some men are involved in the saintly cult as ritual specialist: as shrine-keepers, and as members of the ecstatic cult in whose songs local saints feature along with international saints, and demons. for most purposes, men rely on the women in their households and compounds to deal with the local saints. Yet men who intend to definitively settle elsewhere, in the realm of a different saint, will find their plans crossed by dreams and omens through which the saints protests against their absconding.

            Women, through their dedication of meals and their zyara, carry the bulk of the saintly cult in Khumiriya.

            This ritual involvement of women is intimately linked to the marriage pattern. Marriage is virilocal: both according to the rule and in c. 95% of actual practice. and since no woman marries into the household in which she was born, every marriage involves a woman’s crossing of segmentary boundaries at least at the lowest level of segmentation (in the rare case she marries within the same compound). Like other Islamic societies, and explicit rule as to the preference of agnatic endogamy exists in Khumiriya. demographic processes, the dynamics of marital alliance, the essentially bilateral kinship system hiding under the patrilineal idiom, and the intergenerational transfer of property, however, are much more complex than that they could be summarized, at the analytical level by the participants’ ideology of patrilateral parallel-cousin marriage. This is not the place to present my very extensive data on this point. Let it suffice to say that roughly 50% of contemporary marriages involve partners belonging to different villages, each with their own distinct set of local shrines and saints. A village-exogamous marriage means that a woman leaves her original set of village-level local shrines behind and adopts a new set, that of her husband’s female consanguineal relatives. it is part of a woman’s extensive incorporation into her husband’s segment[9] that she fully adopts the shrines of that group. Within the compound, hamlet and neighbourhood, elder women coordinate food production, food processing, water hauling and firewood collection. From these female leaders the in-marrying woman will learn about the identity and relative importance of the segment’s shrines and saints. She will soon dedicate some of her household meals to these saints, and join the other women in collective zyara to the shrines. However, she will not as a rule give up her relationship with the shrines in her original segment. Although a woman will not often leave the immediate environment of the village for the purpose of visiting relatives, the hospital, the market, or diviners, she has an unalienable right to visit her original shrines, and thus her segment of origin and her relatives there, twice a year.

            A married woman is involved in two complementary sets of relationship with saints — which mirrors, and in fact sustains, her involvement in both her original segment and that of her husband. The picture is further complicated by the relative nature of segmentation. The greater the segmentary distance a woman crosses for marriage, the more different the two sets of shrines will be. If she marries in a different village within the same valley, the two sets will overlap in that the valley’s main shrine and festival will be part of both sets; in that case marriage will only add a few lesser shrines of her husband’s segment (at the village neighbourhood, hamlet and compound level) to the woman’s pre-existing set. With intra-village local endogamy (c. 50% of all marriages) the differences will be even less significant, and in fact the set of shrines before and after marriage may entirely coincide. The differences are far more conspicuous in the case of a marriage linking people from different valleys or even chiefdoms. But the principle remain the same throughout.

            thus every Khumiri woman has zyara obligations vis-à-vis the local shrines associated with the territorial segment (or better: nested hierarchy of segments at various levels) to which she belongs at a given point in time; for descriptive purposes, this type of zyara will be called local zyara. In addition, all women who have migrated from their segment of birth, i.e. mainly in the context of marriage, retain zyara obligations vis-a-vis the local shrines in that segment; this type of zyara will be called original zyara. For the sake of completeness, we should not overlook the fact that marriage is the main, but not the exclusive occasion for a woman to adopt a new set of zyara obligations: when the household of which she is a dependent member takes up residence elsewhere, a similar situation obtains regardless of her marital status. However, such cases are so rare as compared with the virtual universality of marriage among Khumiri women, that they require no separate treatment.

            Local zyara comes with actual membership of (i.e. residence in) a territorial segment, and unites all adult women of that segment under a female leader. The latter co-ordinates the collective zyara of the segment’s women to the local shrines, as part of her general tasks of female leadership. In fact these collective visits to local shrines present an amazing spectacle of territorial segmentation in action. At the occasion of the festival of a valley’s or village’s main shrine, the various female leaders of segments will have agreed on a time for collective zyara. Compound by compound, hamlet by hamlet, neighbourhood by neighbourhood, one will see small groups of women in their best clothes converge along the village path, and team up on their way to the shrine, only to break up again, segment-wise, on their return, Alternatively, the fact that virtually every woman in a compound, hamlet and neighbourhood derives obligations of original zyara from her own, unique life history, endows her with an individuality in the religious sphere which she will normally be allowed to maintain despite strong social pressures towards incorporation in her husband’s segment. the frequent attribution of misfortune to irate, neglected saints suggests however both the practice of individual shedding of original zyara obligations, and the deep-lying tensions in the marital and inter-generational sphere that would seem to attend the incorporation process.

            Personal zyara to major regional saints in the context of illness or infertility results, finally, in the third type of women’s zyara obliga­tions in Khumiriya. For here again the norm applies that a living human cannot at his or her own initiative terminate a relationship with a saint once entered into. For a variety of reasons (which seem to include female under-nutrition; a very low marital age of women before marital legislation was revised in the 1960s; and a repressive sexual culture instilling profound fears and sexual inhibitions in young people of both sexes) many Khumiri women are recorded to have suffered from impaired fertility in the first years of their marriage. In order to remedy this complaint, women would often resort to pilgrimage to distant shrines of regional saints outside the set of shrines falling under local or original zyara obligations. The personal relationship between a woman and a regional saint invoked for reproductive troubles would ideally last a lifetime; in later years, as a woman would take her daughters and daughters-in-law with her on this personal zyara, the younger generation would automatically inherit this relationship, even though the regional shrine would be too distant to be listed among the territorial segment’s local zyara obligations.

            Numerous are the cases when material misfortune, illness and even death are attributed (via various techniques of divination) to irate saints revenging humans’ lack of respect, breach of promises, failure to dedicate meals and make pious visits, or neglect of duties vis-à-vis one saint while honouring the expectations of another saint. since Khumiri saints are shown to embody, on the one hand, concepts of intra-kin intimacy and inter-generational relations, on the other hand a structure of complementary opposition of segments, it will be obvious — even without a discussion of specific cases — that the social, mental and psycho-somatic dramas enacted in such cases reveal deeply-rooted tensions and contradictions within the Khumiri social process and symbolic order. However, an explanation of misfortune like the Khumiri one would represent a welcome escape clause in any religious system: given a certain degree of recognized non-observance of rules and of opportunism[10] among the living humans involved, the supernatural entities invoked are free to honour or to ignore human requests without succumbing to their professional disease: credibility gap. In fact, not all Khumiri women attend to their original and personal zyara obligations with equal zeal; the factors apparently determining this variation in religious behaviour will be discussed below.

            In modern anthropology, paradigmatic consistency and elegance have become reasons for healthy mistrust. Therefore, the above general­ized description of the saintly cult, and particularly of zyara, in contemporary Khumiriya needs to be substantiated with evidence on actual religious behaviour as stipulated by the models and rules described here. We find ourselves here in the somewhat exceptional situation that such evidence is, in fact, available, and that it corroborates the generalized description with amazing precision.



In the remaining sections of this chapter I shall describe the patterns of local, original and personal zyara as found among the adult women inhabiting the villages of Sidi Mhammad and Mayziya, in the valley of Sidi Mhammad.

            The data were collected in 1968, at a point in my field-work when I had sufficiently mastered the principles of Khumiri popular religion and society to phrase my questions properly; and when my stay in the village of Sidi Mhammad had generated a sufficient amount of trust and rapport to allow me to systematically interview the majority of the adult female population in both villages. In Sidi Mhammad, of the total population of 42 resident adult women, 35 (= 83%) were thus interviewed. The 17% non-response could be shown to form an a-select sample from the total population of 42, with regard to important background variables: relative economic position of their household; number of years of their marriage had lasted; geographical distance across which their marriage had been contracted. (Table 1).

    (a) duration of marriage (years)§)                              
















number of women in response group
















in non-response group

















Mann-Whitney U-test, corrected for ties: z = 1.13; p = .13
§) the analysis is limited to women resident in the village of Sidi Mhammad but born in a different village



(b) distance across which marriage was contracted (km)
























number of women

in response group






















in non-response group






















Mann-Whitney U-test, corrected for ties: z = -1.36; p = .09 [ check bottom row table ! ]

    (c) relative economic position of household*)
    poor medium wealthy total
number of women in response group 21 11 3 35
in non-response group 2 2 2 6

Mann-Whitney U-test, corrected for ties: z = 1.11; p .13
*) one woman was omittted from the analysis since the wealth of her household could not be assessed with certainty

Table 1. Validating the sample of women

My data on Mayziya are less complete: they adequately cover local zyara, but show gaps with regard to original and personal zyara. The analysis of the latter two types (section 8) will exclusively be based on Sidi Mhammad data.

            Zyara is public behaviour and moreover a source of prestige and baraka. It is therefore discussed without reticence, even when the interviewer is a young male foreigner. The interview data were checked against: observational data concerning the various types of zyara; systematically elicited statements about the zyara behaviour of neighbours; and many accidental statements uttered during everyday conservations or open-ended interviews. The correspondence between these data proved to be almost 100%. Moreover the data show great internal consistency, particularly in the extent to which the responses and observational data on local zyara converge for the several women of each segment. This convergence could hardly be a research artifact, because when I collected the data I was not even beginning to realize that Khumiri social organization could be described with a model of territorial segmentation. For all these reasons I consider the data to be of good quality, and amenable to such non-parametric statistical tests as I shall perform upon them.[11]

            The valley of Sidi Mhammad stretches from south to north along the Wad al-Kabir, a river whose tributaries have their sources at the highest peaks of Khumiriya, and which flows into the Mediterranean near the town of Tabarka, c. 15 km north of Sidi Mhammad.


Diagram 2. The wider surroundings of the valley of Sidi Mhammad (click on thumbnail to enlarge)

            Diagram 2 shows the wider surroundings of the valley. This diagram conveys the remarkably small geographical scale of the phenomena at hand. The valley of Sidi Mhammad has an area of about 10 km2, and comprises only six villages: Sidi Mhammad, Mayziya, Tracaya-sud, Tracaya-bidh, Fidh al-Missay and Raml al-cAtrus; together these villages comprise c. 600 inhabitants. Movements between villages id mainly on foot, and here the mountainous terrain imposes severe constraints. Thus from Sidi Mhammad it takes people half a day to reach the major regional shrine of Sidi cAbd bi-Jamal, a distance of barely 10 km as the crow flies. Such a distance forms in fact the effective maximal radius for most purposes of inter-village contacts, including zyara and marriage. While illustrating this point, Table 2 suggests that structures of zyara, and the affinal networks created by marriage, together constitute on relational region, of the sort which Meillassoux has called a marriage field (aire matrimoniale, Meillassoux 1964: 11 andz passim).

  range (km) median (km)
distance across which marriages are contracted .1 — 7.8 . 45
distance across which shrines are visited (all types of zyara combined) .0 — 10.1 . 55

Table 2. A comparison of geographical distances across which women resident in the village of Sidi Mhammad visit shrines and across which the marriages of these women have been contracted.


            Like Sidi cAbd Allah bi-Jamal, Sidi Mhammad is a regional saint. The latter’s twice-annual festival lasts for several days and nights. In addition to the people of the valley itself, who are under obligations of local zyara, the festival attracts, from all over Khumiriya, scores of women who are under obligation of original or personal zyara, and moreover scores of male pilgrims, as well as musicians, showmen, ecstatic dancers, butchers, and peddlers in sweets, candles, incense, haberdashery, etc. while the saint Sidi Mhammad is locally represented by no less than four shrines including two qubbas, he is by no means the only saint of the valley. Diagram 3 shows, in their relative position vis-à-vis the dwelling houses, the location of the eighteen shrines that are found in the immediate environment of the villages of Sidi Mhammad and Mayziya alone. Table 3 summarizes the names and physical characteristics of these shrines.

            A minority of the local shrines are surrounded by cemeteries, and a segment’s right to bury its dead in a particular cemetery, i.e. near a particular shrine, is an important expression of the segmentary structure. However, this aspect is not dealt with in my present argument, which concentrates on zyara. Of the shrines listed in table 3, the numbers 1 and 8 are surrounded by cemeteries that are still in use, whereas abandoned cemeteries are found around the shrines 5 and 7, as well as several hundred meters south of 9 and 13.

Diagram 3. Shrines in the valley of Sidi Mhammad (click on thumbnail to enlarge)

  1 Sidi Mhammad al-Kabir qubba
  2 Sidi Mhammad al-Wilda qubba
  3 Sidi Mhammad (al-Wilda) kurbi
  4 Sidi Mhammad (al-Wilda) kurbi
  5 Sidi Bu-Qasbaya al-Kabir mzara
  6 Sidi Bu-Qasbaya al-Wilda mzara
  7 Sidi Bu-Qasbaya al-Wilda mzara
  8 Sidi Rhuma mzara
  9 Sidi Bu-Naqa mzara
10 A’isha mzara
11 Mzara cAin Raml mzara
12 Hasharat al-Brik mzara
13 Sidi Hammad mzara
14 Sidi Bel-Ahsin mzara
15 Jadda Massauda mzara
16 cAli cAbu ’l-Qassim mzara
17 Sidi Bu-Kharuba mzara
18 Hasharat al-Fras mzara

Table 3. Names and physical characteristic of shrines in the villages of Sidi Mhammad and Mayziya.

            Moreover, many of the saints listed in table 3 have shrines elsewhere, outside the villages of Sidi Mhammad and Mayziya; those distant shrines are not listed here. The local zyara pattern in those two villages is confined to the eighteen shrines of table 3.[12]

            In order to assess whether the pattern of local zyara as found in these two villages is in fact governed by territorial segmentation, we have to go through a number of steps. Firth, the dwelling-houses, representing the lowest level of segmentation, have to be clustered into higher-level segments, according to their location, to the visible boundaries by which they are surrounded, and to the distribution of utilitarian characteristics attributes (threshing-floors, springs, men’s assemblies) over the clusters thus formed. The outcome of this exercise is shown in diagram 4.

Diagram 4. The spatial structure of segmentation in the valley of Sidi Mhammad (click on thumbnail to enlarge)

            The following step is the tracing of the specific pattern of local zyara which obtains in each of the territorial segments thus distinguished. A problem arising at this point is that there are far fewer local shrines than territorial segments. The choice is further limited by the fact that not all shrines are available in the same degree as additional, religious attributes of segments. For two adjacent lower-level territorial segments, which are complementary in that they both form part of a higher segment at the next hierarchical level, it would be impossible to express their segmentary opposition by differential patronage, of some very minor shrine situated at a considerable distance, say at the other end of the village: the catchment area of that shrine would be too small to reach as far as these segments. Similarly, these segments could not distinguish themselves by differential patronage of the village’s or valley’s main shrine, for that shrine would already function as the additional, religious attribute of a higher segment encompassing both lower-level segments.

            Two devices combine so as to solve these dilemmas. First non-patronage, even of a nearby shrine or combination of nearby shrines, can mark a territorial segment just as much as positive local zyara. Secondly, segments can distinguish among themselves not only through the selection or non-selection of local shrines in a particular combination, but also through differences in frequency with which the selected shrines are actually visited. Twice-annual zyara constitutes a minimal frequency for any shrine; four times a year is an average frequency for shrines that are visited with more than minimal zeal. As marking devices, non-patronage and differential frequency dramatically increase the number of possible combinations given a limited number of shrines; yet it must be admitted that differential frequency introduces a non-discrete element that somewhat spoils the neat, digital combinatory logic of the segmentation model.

            These devices are clearly at work in the pattern of local zyara in the village of Sidi Mhammad and Mayziya, as shown in diagram 5.

Diagram 5. The spatial structure of local zyara in the valley of Sidi Mhammad (click on thumbnail to enlarge)

Here for all compounds of both villages the associated patterns of local zyara are shown, on the basis of the interview and observational data discussed above. Combining the information of diagrams 4 and 5 results in diagram 6, which presents the segment’s differential local zyara patterns in the familiar dendrogram format.

Diagram 6. Patterns of local zyara (pilgrimage) in the village of Sidi Mhammad as an expression of segmentation (click on thumbnail to enlarge)

            A number of conclusions can be based on diagram 6. Clearly, territorial segmentation provides the key to existing structures of local zyara. Territorial segments, whose existence is marked by visible boundaries and the distribution of utilitarian characteristic attributes, distinguish themselves in the religious sphere by the veneration of specific combination of local shrines, in specific frequencies. What emanates clearly from diagram 6 is the fact that complementary opposition in segmentation only refers to one level at the same time, irrespective of the distribution of distinctive features at higher or lower levels. Thus segments 1.1.2 and 1.2.2 can afford to be both associated with shrines 1, 2 and 3, which both segments visit frequently. There is no direct complementary opposition between these two segments, since they belong to different higher level segments (1.1 and 1.2 respectively), and the difference between the later is marked by shrines 7 and 11. The complementary segment of 1.1.2 is 1.1.1. (this difference is marked by frequent visiting of shrines 7 and 11, as against shrines 1, 2 and 3); the complementary segment of 1.2.2. is 1.2.1, with differences being marked by frequent visiting of shrines 5 and 6 as against 1, 2 and 3, respectively. The inclusion of complementary segments in higher-level segments renders the combinatory logic of characteristic attributes more complicated, but does not destroy it.

            However, while the model fits empirical reality amazingly well, the fit is, of course, not 100%. Not all complementary segments at all levels are marked by differential local zyara. Thus the sub-neighbourhoods,, and have an identical pattern of local zyara.

            Moreover it turns out that, insofar local zyara is concerned, three and not two segmentary levels are to be distinguished between compound level and village level; this is particularly the case in the village of Sidi Mhammad. Environmental conditions and the ongoing dynamics of territorial segmentation can explain these deviations from the simpler model. Permanent water supplies are scarcer in Sidi Mhammad than in Mayziya: in the former village there are 9 to 17 households to one permanent water source, against only 8 to 9 in Mayziya. Hence the spring-defined neighbourhoods are in fact considerably larger in Sidi Mhammad than in Mayziya, and begin to approach villages. This process of segmentation also manifests itself in the erection of a separate men’s assembly in the southern part of the village of Sidi Mhammad (super-neighbourhood 1.1, called Qaca-Ramal), and in the growing expression of antagonism between people from that part and the rest of the village. The complex historical background, involving competition between rival clans, aspirations of political leadership, the vicissitudes of marriage alliances, the effects of establishment of a colonist’s farm near Qaca-Ramal, and the differential use of cemeteries cannot be elaborated upon here (Van Binsbergen 1971, 1980a, 1980b).

            The ongoing segmentation process also explains the ambiguous position sub-neighbourhood occupies in the dendrogram. But here we encounter not fission (as in the Qaca-Ramal case), but fusion: the segment in question, straddling the boundary between the two villages, historically forms part of Mayziya, but its members have established strong ties of marriage and clientship with their present neighbours, the administrative chief’s family; the latter’s residence in the village of Sidi Mhammad dates back to the 1910s.[13]

            Rather than upsetting the model of territorial segmentation as governing local zyara, these deviations show that model to be dynamic, and capable of responding to the realities of the social and ecological process. Let us now turn to the quantitative data concerning original and personal zyara: forms of religious behaviour that cut across, instead of express, the pattern of territorial segmentation.



Turning now to non-local zyara, we should first assess the relative incidence of the three types of zyara.

            The 35 systematically interviewed women in Sidi Mhammad observed between them 232 zyara obligations vis-à-vis shrines in Khumiriya. Of these, 219 (= 94%) involved local zyara. Each woman observed an average of 6.6. zyara obligations, the total range stretching from 5 to 10. Of this average of 6.6., an average 6.3 involved local zyara (range 5-8, as can be read from diagram 6). The fact that many shrines are associated with the same saint, means that the number of observed zyara obligations vis-à-vis different saints is lower than that vis-à-vis shrines. The women of the sample have an average of 4.1. (range 3-7) observed zyara relations with saints, out of which an average of 3.7 (range 3-5) involve saints associated with the local segments these women belong to at the several hierarchical levels. These data on zyara relationships can be converted into figures on actual pious visits made, by taking differential frequency into account. Per period of six months, the women of the sample make 342 zyaras between them, of which 329 (= 96%) are local zyaras, 10 (= 3%) are original zyaras, and only 4 (= 1%) are personal zyaras. These figures must be considered estimates. Yet they convincingly demonstrate the overwhelming preponderance of local zyara, as stipulated by the structure of territorial segmentation, over the non-local forms that cut across the segmentary structure.

            It is virtually impossible for a woman to resist the strong social pressure and the supernatural sanctions that prompt her participation in the collective local zyara of the segment in which she is resident. Original and personal zyara, however, are a more individual matter, and here observance of existing obligations shows considerable variation.

            The positive data on personal zyara are too limited to allow statistical analysis. The three women concerned are between forty and sixty years old. They have exceptionally high prestige and power because of their age, their very close kinship relations with administrative chiefs, and the wealth of their households. Two are effective female leaders of their neighbourhoods, and as such co-ordinate local zyara. Their maintaining of personal zyara relations with distant shrines, nearly all of which are of regional importance, adds to their local prestige, and renders further independence to their religious and social behaviour as individuals. Moreover, they would hardly be able to fulfil their personal zyara obligations if their social position did not provide them with the financial means to undertake a long journey, and with an extensive regional network of social contacts on which they can rely during that journey and at the distant shrine. The data strongly suggest that many other women in the sample contracted personal zyara obligations at some time in their lives, but had to drop them because of their less exalted social position within their segments of residence.

            Original zyara is a somewhat more common phenomenon. Here we can draw on two sets of data: data on the women resident in Sidi Mhammad; and on the set of women who originate from that village and who (according to the converging evidence of observational data and interviews) either observe, or fail to observe, their obligations of original zyara vis-à-vis the regional shrines of Sidi Mhammad.

            Since about 50% of all marriages are contracted within the same village, and since (cf. diagram 6) not all segments within a village differ as to the set of shrines to which local zyara is directed (although frequencies of zyara tend to differ), not all women in the sample acquired obligations of original zyara at marriage. In fact, only 14 women in the sample did so (= 40%); for the remaining 60%, local zyara and original zyara entirely coincide.

             Of these 14 women, 7 (= 50%) observe their original zyara obligations, while 7 (=50%) do not. Table 4 makes clear that the relative importance of shrines is a crucial factor here. Such importance is measured by the following indicators: the segmentary level at which the shrine functions as an additional, religious attribute; the physical characteristics of the shrine (qubba, kurbi or mzara); and the existence of a twice-annual festival for that shrine.

Importance Range of geographical distance (km) Number of observances Number of non-observances total
high 2.6 — 8.8 8 1 9
middle   .8 — 3.2 1 7 8
low   .1 — 2.1 0 5 5

Table 4. The observance of obligations of original zyara among women resident in the village of Sidi Mhammad, as a function of the importance of the original shrine, and of the geographical distance between that shrine and a woman’s current place of residence.

            Further statistical analysis (Van Binsbergen 1971a: 286f) demonstrates that such conceivable factors as wealth, prestige, and the number of years elapsed since the woman, by marrying and taking up residence in her present segment, acquired obligations of original zyara, do not have a statistically significant impact on the observance of original zyara among the resident women of Sidi Mhammad.

            These data are supplemented by those on women who, originating from Sidi Mhammad, have married outside and therefore are under obligations of original zyara focussing on the valley of Sidi Mhammad. The festival of Sidi Mhammad is the only occasion at which the necessary observational data can be collected; moreover it is by far the most important occasion for women to observe their original zyara obligations. For these reasons I shall concentrate here on zyara to the major, regional shrines of Sidi Mhammad, and ignore zyara to lesser shrines in the same valley. A fortunate implication is that thus importance of shrines as a factor determining observance of original zyara is kept constant, so that other factors may stand out more clearly. It is important to realize that in these cases we are dealing with women who have married not only outside the village, but also outside the valley of Sidi Mhammad — for all villages of that valley would make the pious visit to the shrines of Sidi Mhammad as part of their local zyara obligations.

            On the basis of my village census and genealogies the full set of women involved can be identified. Limiting the analysis to those who currently live within a distance of 20 km (original zyara across wider distances would be practically impossible anyway), the set consists of 22 individuals, 15 of whom (=68%) actually observe original zyara, while 7 (=32%) do not.

            While no data are available as to the wealth and prestige of these out-marrying women in their present, distant places of residence, the data reveal that the number of years elapsed since marriage (i.e. since the departure from the original segment) is significantly associated with observance of original zyara (table 5).

Duration of marriage (years)                      
      3   4   5   7   8 10 12 15 25 total
Number of women Observing   1   1   4   2   2   2   1   1   1 15
Not observing   0   0   0   0   0   1   2   2   2   7

Mann-Whitney U-test, corrected for ties: z= -2.81; p= .003

Table 5. Observance of original zyara obligations among out-marrying women from Sidi Mhammad, as a function of the duration of marriage.

All women who left the village for marriage ten years ago or less, stick to the rule; those who left longer ago, tend to drop observance. Duration of marriage seems to be a surface factor, underneath which a more important one is hidden: the residence, in the segment of origin, of a surviving parent (table 6). Of course, the longer a marriage has lasted, the older a woman is and the less likely she will have surviving parents.

  At least one parent alive, and resident in Sidi Mhammad No parent resident in Sidi Mhammad total
observance 12   3 15
non-observance   0   7   7
total 12 10 22

l =15.30; df=1; p< .001. The l-statistic has the same distribution as X2; cf. Spitz 1961.

Table 6. Observance of original zyara obligations, among out-marrying women from Sidi Mhammad, as a function of parents’ residence in the segment of origin.

This factor points to the social functions of original zyara, as a unique opportunity to visit living kinsmen. Additional statistical analysis (Van Binsbergen 1971a: 288f) however suggests that, besides sociability and psychological kin support in the vicissitudes of marriage and virilocal incorporation, another structural theme is involved here: the inter-generational transfer of property rights (particularly in relation with land, a scarce asset in Khumiriya). The wish to keep in touch with consanguineal relatives around the original shrine is not a sufficient reason for original zyara; for observance of this type of zyara is not significantly associated with the residence, in the segment of origin, of an out-marrying women’s brothers — whatever the wealth of the latter. The continued residence of parents suggests an undivided patrimony. By keeping up visits to her segment of origin, the woman, in accordance with Khumiri views on land tenure, asserts her right to a share equal to that of her male siblings. De facto these rights are waived as, after the father’s death, the surviving sons administer the patrimony on their own behalf: initially under the direction of the eldest sons, until such time when fraternal rivalry necessitates division. At that point a reversal of visiting obligations can be seen: more fully incorporated in her husband’s segment, the woman tends to drop her original zyara obligations, but instead here brothers are under obligation to visit her with presents at the day of the Great Festival (Id al-Kabir, Id al-Adha). The woman’s sons, however, retain a latent right in the land administered by the mother’s brothers, and in exceptional cases these rights are actually exercised, leading to a man’s matrilocal residence.

            Combining the evidence on Sidi Mhammad’s resident women and out-marrying women, the main factors determining observance of original zyara obligations may be summarized as in diagram 7:

Diagram 7. A statistical model of the structure of local zyara in the valley of Sidi Mhammad (click on thumbnail to enlarge)



The ethnography presented here clearly has many interesting openings towards central theoretical concerns in the social science of religion. There is a striking Durkheimian suggestion of one-to-one correspondence in the extent to which the saint and his shrine seem to function, at all levels of social and ritual organization and experience, as a straightforward symbol of the social group with which they are associated. Alternatively, such cutting-across the overall structure of segmentation as can be seen in original and personal zyara, points to the potential of religion to provide alternatives to the structural arrangements that govern the more secular aspects of social life.

            This calls to mind the theories of pilgrimage and regional cults as advanced by the Turners and by Richard Werbner (Turner 1974; Turner & Turner 1978; Werbner 1977). The possible contribution of the Khumiri data to the further development of these theories would at first glance appear to be somewhat negative. Werbner, in an attempt to get away from the classic correspondence paradigm in religious anthropology, has stressed cultic regions’ autonomy vis-à-vis processes of material production, secular social organization, and political structure. The Khumiri case (of predominantly local zyara) however would be an example of extreme correspondence between cult and the secular societal process. Moreover, except in the relatively rare cases of original and personal zyara, which cut across segmentation, the more massive manifestations of the cult (at the village and valley level, and culminating in festivals) do not seem to involve principles different from those operating at the lowest level: the cult of inconspicuous mzaras that are tucked away in some corner of a compound and hamlet, and that have virtually no relevance beyond these small territorial units. In this respect the Khumiri cult of saints, while clearly a regional cult in terms of geographical scope and number of people involved, would not stand out as one when its organizational structure is considered.

            Similarly, it is only in personal zyara to distant saints — the expression of an atomized devotion — that the shrine appears, in Turner’s terms, as the ‘Center Out there’, and that this generalizations apply as to pilgrimage as a distinct social process in its own right. In local zyara, which constitutes the vast majority of pious visits in Khumiriya, the shrine is not a distant place visited at the end of a long and arduous physical and spiritual journey across unknown parts — it has more the nature of a visit to a close and dearly-loved relative, involving a short passage through familiar surroundings, in the company of people one knows well and identifies with. It is for this reason that i have refrained, in this essay, from using the term pilgrimage — except in the title, for signalling purposes only.

            Finally, while both authors would stress the dialectics of inclusiveness/exclusiveness or universalism/particularism as the crux of the cults they describe, the Khumiri data would suggest that this dialectic could hardly be adequately analyzed on the level of popular religion alone. On the one hand the very same dialectic underlies the secular structure of segmentation (where the opposition of complementary segments is resolved at the next level of segmentary inclusion). On the other hand it is on this dialectic that the interplay resolves between formal and popular Islam (which the saints straddle, as epitomes of the former and yet cornerstones of the latter) (Van Binsbergen 1980a).

            Another obvious dimension of the Khumiri data concerns the dialectics between socio-economic structure and the symbolic order (Van Binsbergen 1981). The embeddedness of most of the cult of saints, through patterns of local zyara, in a segmentary organizational structure of localized social units entrusted with material production, biological reproduction, and with the regulation of the social relations upon which these fundamental processes depend, would suggest that in the cult production find expression, and are in themselves being reproduced. Too little could be said here about these contradictions (mainly: those between men and women; and between human patrons and clients) to indicate their relation to the cult of saints. Moreover, the virtual coincidence (table 1) in Khumiriya between the cultic regions as created by various types of zyara, and the area within which the biological reproduction of human population takes place (as indicated by the distances across which marriages are contracted) suggests that the relation between religion and societal reproduction operates at an even more profound level than the sheer underpinning of the structures of segmentary organization, and of authority, that govern the local subsistence economy. The saints’ involvement in women’s reproductive troubles points in the same direction. marital relations, and more in general the tension between male and female, would seem to constitute a dominant axis in the Khumiri cult of saints (cf. Fernea & Fernea 1972; Dwyer 1978; Davis 1979).[14]

            Meanwhile at least one other contradiction would have to be considered, that between the state-supported rural elite (administrative chiefs, officers of the unemployment relief work organization, teachers) and the peasants. The dialectics of their relationships must be understood in the light of the relation between the peasants’ less and less viable subsistence economy (with its manifold links with the cult of saints), and the capitalist economy (with its manifold links with the cult of saints), and the capitalist economy into which Tunisia is increasingly drawn — with as its latest local manifestation the massive labour migration to Libya of Khumiri men in the 1970s. This interplay of competing relations of production seems to offer, finally, a setting for the persistence of Khumiri popular religion, with the cult of saints as its major manifestation, despite the inroad of formal Islam. Popular religion lives on at least to the extent to which the non-capitalist local subsistence economy lives on — albeit that the symbolic order tends to either lag behind, or anticipate, the development of economic relations.

            I shall, however, resist here the temptation of jumping to theoretical conclusions; these will hopefully be drawn elsewhere at greater length and against a fuller background of historical and ethnographic data, and theoretical considerations.



Dallet, J.

1939               ‘Les “mzaras” de Kroumirie’, IBLA (Institut des Belles Lettres Arabes) 1940, 3: 323-42.

Davis, J.

1979               ‘The Sexual Division of Religious Labour in the Mediterranean’, paper read at the Conference on Religion and Religious Movements in the Mediterranean Area, Amsterdam: Municipal University of Amsterdam/Free University.

Demeerseman, A.

1938-39          ‘Le culte des saints en Kroumirie’, IBLA (Institut des Belles Lettres Arabes), 1: 3-28 (1938), 2: 3-27 (1939).

Demeerseman, A.

1939-40          ‘Les croyances relatives aux “Oualis” des mzaras en Kroumirie’, IBLA (Institut des Belles Lettres Arabes), 3: 3-39.

1964               ‘Le culte des walis en Kroumirie’, IBLA (Institut des Belles Lettres Arabes), 27: 119-66.

Dornier, P.

1950               ‘Le recours aux oualis dans les campagnes du Nord de la Tunisie’, IBLA (Institut des Belles Lettres Arabes) 13: 392-396.

Dwyer, D.H.

1978         ‘Women, Sufism and Decision-making in Moroccan Islam’, in: L. Beck & N. Keddie (eds.), Women in the Muslim World, Cambridge (Mass.)/London: Harvard University Press, pp. 595-598.

Evans-Pritchard, E.E.

1949               The Sanusi of Cyrenaica, Oxford: Clarendon Press.

Ferchiou, S.

1972         ‘Survivances mystiques et culte de possession dans le maraboutisme tunisien’, L’Homme, 12, 3: 47-69.

Fernea, R.A. & E.W. Fernea

1972         ‘Variation in Religious Observance among Islamitic? Women’, in: L. Beck & N. Keddie (eds.), Women in the Muslim world, Cambridge (Mass.)/London: Harvard University Press, pp. 385-401.

Gellner, E.

1969         Saints of the Atlas, London: Weidenfeld & Nicholson.

Keddie, N.R. (ed.)

1972         Scholars, Saints and Sufis, Berkely?/Los Angeles/London: University of California Press.

Köbben, A.J.F.

1975         ‘Opportunism in Religious Behaviour’, In: W.E.A. van Beek & J.H. Scherer (eds.), The Relevance of Models for Social Anthropology, London: Tavistock, pp. 87-112. [fout]

Louis, A.

1977         Bibliographie ethno-sociologique de la Tunisie, Tunis: IBLA (Institut des Belles Lettres Arabes).

Meillassoux, C.

1964         Antropologie économique des Gouro, Paris/The Hague: Mouton.

Sallnow, M.

1981         ‘Devotional Pluralism in the Andes’, paper read at the Symposium on Plurality in Religion, IUEAS (International Union of Anthropological and Ethnological Sciences) Intercongress, Amsterdam.

Siegel, S.

n.d.         Nonparametric Statistics, New York etc./Tokyo: McGraw-Hill/Kögakusha.

Spitz, J.C.

1961               ‘De l-toets de de l’-toets’, Nederlands Tijdschrift voor de Psychologie, 16: 68-88.

Turner, V.W.

1974         Dramas, Field and Metaphors, Ithaca/London: Cornell University Press.

Turner, V.W. & E. Turner

1978         Image and Pilgrimage in Christian Culture, Oxford: Blackwell.

Van Binsbergen, W.M.J.

1970         ‘Verwantschap en territorialiteit in de sociale structuur van het bergland van noordwest Tunesiè’, drs. soc. sc. thesis, Amsterdam: Municipal University of Amsterdam.

1971a         ‘Religie en samenleving: Een studie over het bergland van noordwest Tunesiè’, drs. soc. sc. thesis, Amsterdam: Municipal University of Amsterdam.

1971b             ‘ “Saints of the Atlas”: Ernest Gellner’, Cahiers des Arts et Traditions Populaires, 4: 203-211.

1976         ‘Shrine Cult and Society in North and Central Africa’, paper read at the Annual Conference, Association of Social Anthropologists of Great Britain and the Commonwealth, Manchester.

1980a         ‘Popular and Formal Islam, and Supra-local Relations: The Highlands of North-Western Tunisia, 1800-1970’, Middle Eastern Studies, 20: 71-91.

1980b         ‘Interpreting the Myth of Sidi Mhammad’, In: K. Brown & M. Roberts (eds.), Using Oral Sources: Vansina and Beyond, special issue, Social Analysis, 4, Adelaide: University of Adelaide, pp. 51-73. Also in: Proceedings of the International Oral History Conference, 24-26 October 1980, Amsterdam: Historisch Seminarium, Municipal University of Amsterdam, vol. II: 511-547.

1981         Religious Change in Zambia, London/Boston: Kegan Paul International.

in prep.         Shrines and Ecstasy in the Social Structure of Northwestern Tunisia.

Waardenburg, J.

1979         ‘Official and Popular Religion as a Problem in Islamic Studies’, in: P.H. Vrijhof & J. Waardenburg (eds.), Official and Popular Religion, The Hague/Paris: Mouton, pp. 340-386.

Werbner, R.P. (ed.)

1977         Regional Cults, London/New York: Academic Press.

homepage | click here in order to access Part I

[9]              An indication of this incorporation is that very few widows ever move back to their village of origin; for a set of indicators of female incorporation in a context of marriage, cf. Lewis 1965.

[10]             Cf. Köbben 1975, who in fact cites the Khumiri case.

[11]             These tests are not affected by the relatively small number of cases, nor do they imply assumptions as to the scale level (interval, ordinal, nominal) of the variables; cf. Siegel, n.d. The short questionnaire, in colloquial Arabic, used to collect (in addition to observational materials) the quantitative data on zyara and other types of religious performance, will be included in Van Binsbergen, in preparation.

[12]             Of the 18 shrines, the number 8 and 9 are not visited by any inhabitant of either village: 8 is, however, visited by inhabitants of the neighbouring village of Tracaya-bidh.

[13]             A peculiarity of the zyara pattern of the village o Mayziya, and one that is not easily accommodated within our tripartite typology of Khumiri zyara, is that virtually all adult women resident in that villages have an infrequent zyara relationship with the shrines of Sidi Bu-Kharuba and Sidi Bu-Zarura in the adjacent valley of Saydiya, c. 4 km east of Mayziya. Here again segmentary fission provides the explanation: these distant shrines are collectively visited because the majority of the present-day inhabitants of Mayziya are recent immigrants from Saydiya; their migration from that valley has been too recent than that religious and secular ties with that relatively distant place of origin could already have been severed entirely. Khumiri history offers numerous cases of emigrants cutting off such ties after a few decades.

[14]             Davis' juxtaposition of pious men in Islam versus pious women in Mediterranean Christianity seems scarcely to apply to Khumiri popular Islam.

homepage | click here in order to access Part I

page last modified: 2000-05-17 19:59:10