I am indebted to the trustees of the Horniman Scholarship, the Right Worshipful Company of Goldsmiths, the Department of Sociology and and the trustees of the Devon War Memorial Scholar-ship of Exeter University, whose financial assist-ance enabled me to undertake fieldwork in Ahaggar. I am grateful to Professor W. D. Hammond-Tooke, Head of the Department of Social Anthropology at the University of the Witwatersrand, for his generous advice and comments on this paper, of which Part I appears here and Part II in the next issue of Africa.

¹ Murphy (1967), for example, in his study of Tuareg kinship recognizes an incongruity or ‘lack of fit’ between the kinship terminology and the social system, and concludes that ‘imbalance and dissonance may be the very essence of structure’. However, his analysis of the social system makes no reference to its dominant feature, namely its class structure. The current work of certain French anthropologists, such as Andre Bourgeot and Pierre Bonte (much of which is still unpublished), is exempt from these criticisms. See, for example, Bonte forthcoming.

² ‘Vassals’ (Kel Ulli) outnumber ‘nobles’ by about 8:1.

³ The use of the terms ‘nobles’ and ‘vassals’ does not give a sufficiently accurate description of their respective positions and statuses even in traditional times. Although it might be argued that the traditional mode of production was‘feudal’, the feudal analogy is dangerously misleading, and tends to reflect the thinking of many earlier observers of Tuareg society who seem to have perceived most traditional ‘African’ societies as being in some way ‘feudal’.

⁴ Ettebel literally means ‘drum’, which was the symbol of political authority.

⁵ The occasional use of the past tense is necessitated by the fact that Ahaggar society has undergone radical changes in the last few years, following Algerian Independence. The drumgroup, and several other political, social and economic institutions have been destroyed (sometimes coercively) or restructured in accordance with the Algerian ‘social revolution’. The class structure, and much of the social system as described in this paper, is consequently quite different from that now found in Ahaggar. I am not concerned here with these changes, but rather with Kel Ahaggar society as it existed up to about the mid-1960s.

⁶ See note 5.

⁷ A further relationship, known as temekcbit, whereby vassals were obliged to feed their nobility, was closely associated with the temazlayt.

⁸ The ‘wealth’ of the Kel Ulli could be measured effectively in terms of their herds of goats (and more recently camels), and their derivative products.

⁹ It seems that in practice the nobles could camp alongside and demand food, through the instituted temekchit relationship (see note 7), not only from Kel Ulli with whom they had temazlayt relations, but virtually all subordinate people in their drumgroup.

¹⁰ Nobles frequently left their camels in the care of their Kel Ulli, who then had usufruct rights over them (and certain other benefits).

¹¹ The ‘ridicule’ and humiliation suffered by a noble as a result of such action acted as a powerful sanction.

¹² The generation or so covering the latter part of the last century and the early part of this century (from about the 1880s to 1920) was a complex period in the history of the Kel Ahaggar. It should be mentioned, however, that the many changes that took place during this period were as much a result of external forces, in the form of French encroachment on Ahaggar and their subsequent pacification of the Kel Ahaggar, as of forces being generated within Ahaggar itself. These changes go beyond the scope of this paper, but an incident witnessed by Maurice Benhazera, and recounted in his report on the Kel Ahaggar in 1908, gives an interesting picture of the tensions and changes taking place in ‘noble-vassal’ relations. He wrote: ‘I saw an Iouarouaren camp [Kel Ulli of the Taitok drum-group] who were camped near to tents of the Taitok nobles, break camp in order to move further away. They had arrived back from Tidikelt [to the North of Ahaggar] with dates and other produce, and the noble women constantly came to ask for things in the Kel Ulli camp. Consequently, their chief, In Chikadh, gave the order to leave, saying that it was in order to put a stop to the nobles’ pestering’ (Benhazera 1908: 54) (my translation).

¹³ Prior to Algerian Independence there were about 5000 Kel Ahaggar and about 1000–1500 slaves. Slaves were not distributed evenly between descent groups (tawsatiri). For example, in the Kel Rela (noble) and Dag Rali (vassal) descent groups, ‘slaves’ slightly outnumbered ‘Tuareg’. Although the Algerian government has put an end to all forms of ‘slavery’, a few ex-slaves chose to remain (as ‘free-men’) with their former masters, and are still to be found in several camps.

¹⁴ Since 1959 the rainfall in Ahaggar has been particularly low. In only a few years has it risen above the mean annual average, and then often in the form of heavy cloudbursts of short duration with a considerable amount of the rain being lost in run-off. Pasture has remained impoverished throughout most of this period, and I have consequently never seen such ‘section’ camps form.

¹⁵ The Taitok were effectively expelled from Ahaggar after the French pacification, and in the 1949 census numbered only 33 individuals in Ahaggar, while the Tegehe-Mellet became numerically and politically insignificant.

¹⁶ All four are named after the tawsit sub-area, or more specifically the main centre in that sub-area, over which they held land-rights, viz. Kel Tamanrasset, Kel Terhenanet, Kel Hirafok and Kel Tinhart.

¹⁷ e.g. Murdock 1959; Blanguernon 1955; et al.

¹⁸ Their matrilineality ‘makes sense’ in so far as they are not exclusively matrilineal. As we shall see, matriliny is most pronounced amongst the nobility who can hardly be called ‘pastoralists’. See Aberle (1961), Douglas (1969), et al. for a more general discussion of the problems of matrilineal systems.

¹⁹ The ‘stomach’ symbolises the mother's family and, more specifically, matrilineally related women. (The father's family, and the patriline, are symbolised by the ‘back’, arouri.)

²⁰ The Kel Ulli's (and all other subordinate groups’) land-rights were theoretically revokable. In such unlikely instances as non-payment of landrights or revolt these rights might have been revoked, but no such cases have ever arisen.

²¹ The term for the taivsit ‘chief’ among both classes is amrar. In the case of the nobility the tawsit chief was, of course, the drum-chief who held all military and legislative authority over the drumgroup. Among the subordinate classes, the term amrar is translated more appropriately as ‘headman’.

²² These rules are in accordance with Goody's conclusions on the nature of African descent systems, that ‘matrilineal systems of succession and inheritance are intrinsically more lateral, and hence more corporate’ (1970: 628).

²³ The case study which Nicolaisen uses to illustrate these ‘rules’ (1963: 143) is one of the most exceptional amongst the Kel Ahaggar. It concerns actual marriages between the two vassal (Kel Ulli) descent groups of the Dag Rali and Aguh-en-tehle. However, over the last three generations there have been only four marriages between these two descent groups, which involved ‘exchange mafriages’ between two ‘families’. Indeed, with the exception of two other marriages these are the only exogamous marriages contracted by members of the Dag Rali during this period. The example is also unusual in that several of the children are ‘adopted’. Moreover, Nicolaisen does not make clear whether his ‘rules’ apply to both classes, although this is implied. They therefore raise such questions (which cannot be answered without much speculation) as whether residency may have been more matrilocally orientated in earlier times, and whether his ‘rules’ possibly reflect practices which may have been more common amongst the nobility.

²⁴ The apparent opposition between these two modes of transmission of property rights is largely resolved by the specific marital strategies of the two classes (see below).

²⁵ Labour, through the reproduction of the ‘domestic unit’ and the inheritance of slaves, is thus also transmitted predominantly in this line.

²⁶ The term anet ma is also given to the mother's mother's brother's son and the mother's father's brother's son (see Nicolaisen 1963: 449), neither of whom may belong, to the same matriline as ego. This, however, is in accordance with the merging rules applicable to Iroquoian kinship systems. (The core terminology of Kel Ahaggar kinship is Iroquoian.)

²⁷ See Radcliffe-Brown 1952: chs. 1, 4 and 5; Levi-Strauss 1968: ch. 2; Goody 1959: 61–88; de Heusch 1974: 609–19.