UNIVERS ITY OF THE WITWATERSRAND
 A F R I C A N S T U D I E S I N S T I T U T E
 African Studies Seminar Paper
 to be presented in RW
 4.00pm SEPTEMBER 1976
 Title: Classes, The Mode Of Production And The State In Pre-Colonial
 Swaziland.
 by: Philip Bonner
 No. 046
tf.
 Not to be quoted without
 the author's permission
 UNIVERSITY OF THE WITWATERSRAND - AFRICAN STUDIES INSTITUTE
 AFRICAN STUDIES SEMINAR
 Paper for Monday, 13th September Room CM 319 at 4,00 p.m.
 Mr. P.L. Bonner Classes, the Mode of Production and the
 State in pre-colonial Swaziland
 In this paper I seek to do three related things; to identify the
 dominant mode of production in the Swazi social formation and other
 possible subsidiary modes; to establish the existence or otherwise
 of classes In Swazi society; and to elucidate the development of
 the Swazi state. Before I start however I should like to
 make two disclaimers. Firstly I make no pretence at theoretical
 rigour and have merely used the concepts employed In this paper to
 illuminate my material. Secondly the questions raised by such
 concepts are not easily answered by the data I have at my disposal.
 Swazi tradition I do not doubt will help to give answers to those
 questions, but first it will be necessary for those questions to be
 posed. This I did not do during my own fieldwork in Swaziland,
 and for that reason much of my paper will have a provisional ring.
 Nevertheless, despite these limitations I feel justified in venturing
 into this field. Swazi society does not as far as I can see fit
 into recent categorisations of pre-capitalist modes of production
 and its study may help in developing those categories further. At
 the same time the questions raised by this mode of analysis
 undoubtedly turn the spotlight on neglected areas of Swazt society
 and even If only partially or provisionally answered must help our
 understanding of how that society works. For that reason I make no
 apologies at undertaking such an analysis and merely ask that It be
 taken as much as a programme for research as the fruits of a
 completed project.
 The/ ...
- 2 -
 The Tributary Mode of Production
 Hindess and Hirst define a mode of production as "an articulated
 combination of relations and forces of production structured by the
 dominance of the relations of production. The relations of
 production", they go on, "define a specific mode of appropriation
 of surplus labour and the specific form of social distribution of
 the means of productions corresponding to that mode of appropriation
 of surolus labour". The implications of the notions of articulation
 and of dominance are of particular significance for the present
 study. Later Hindess and Hirst make the point that, "a distinct
 structure of relations of production supposes a set of forces of
 production which corresponds to the conditions of the labour process
 it establishes", and go on to reject the concept of an Asiatic mode
 of production on the grounds that, "the conditions of appropriation
 of surplus labour entailed by this mode do not transform the
 labour process or the relation of the labourer to it". As a result
 the relation of the relations of production and the forces of
 production are characterised by, "a certain arbitrariness11! and the
 conditions of proving the existence of the mode of production are
 not met, i.e. forces cannot be deduced from the relations of
 production. (1)
 C
 This is no small matter, as the mode of production characteristic
 of most African social formations seems to correspond more closely
 to the Asiatic mode of production than to any other, and once this
 is rejected the bulk of African and indeed other social formations
 known to history are relegated to some kind of 'modeless1 limbo.
 An alternative must therefore be found, or else in one guise or other
 the Asiatic mode must be revived. What in the first place are the
 possible alternatives? In Hindess and Hirst's classification the
 two most likelv candidates are the feudal mode of production and
 what thev term the second variant of the primitive communist mode.
 The second variant of the primitive communist mode of production
 corresponds/...
- 3 -
 corresponds to what Rey and others have called the lineage mode of
 production. (2) Its identifying characteristics are communal
 appropriation of the social product and its extended or 'complex*
 redistribution among the lineage members. Leaving aside for the
 moment the question of whether the lineage mode of production is
 characterised by classes or is classless, it is clear that it has
 been widely distributed across the continent of Africa. Whether or
 not that means that this has been the dominant mode of production
 is however an entirely different question, for almost wherever one
 looks one finds the lineage mode combined into other aggregates
 like the village or the kingdom which themselves require definition
 ) in terms of modes of production.
 Can these larger aggregates be considered as variants of the feudal
 mode of production? In Hlndess and Hirstsfs terms they cannot, for
 whereas the feudal mode of production presupposes a ruling class
 which exists independently of the state machine, direct producers
 who are politically and legally bound to their exploiters through
 the right of eminent domain, and the absence of communal production,
 most if not all of the larger collectivities of which we have been
 speaking are characterised by the absence of an exploiting class
 independent of the states, the absence of private property in land,
 ) and communal production. (3) The differences implied by all this
 cannot be reduced simply to matters of scale. Amin for example
 suggests that the feudal mode of production is only a decentralised
 and regressive variant of the tributary mode of production (which he
 sees as the dominant mode of production for most of the world's
 historv), in as much as the rights to tribute and to ownership
 previously vested in the state have reverted during periods of weak-
 ness to individual feudal lords. (4) In fact the difference between
 the two is far more fundamental than that. In the feudal mode of
 production the mechanism for the appropriation of surplus (feudal
 rent) and the property relations which go with that, involve
 extensive control over the labour process on the part of the feudal
 lord/...
lord and for this reason this excludes the possibility of communal
 production. (5) In other words the forces of production are
 structured by the relations of production and can be deduced from
 them. In the tributary mode of production on the other.hand (which
 includes the Asiatic mode) no such intensive or continuous structu-
 ration of the labour process takes place. Production can be either
 independent peasant or communal (6) and the exploiter has a purely
 external relation to the process of production intervening to exact
 tribute (or tax/rent in Hlndess and Hirst's terms) only once the
 production process is complete. (7) Consequently the forces of
 production are not in any meaningful way structured by the relations
 of production, thev cannot be deduced from one another, and the
 existence of the tributary or Asiatic mode of production remains
 unproved.
 Does this mean that there are no modes of production corresponding
 to the larger aggregates of kingdom or chiefdom, and that we are
 forced back on some kind of notion of hegemony which is rooted in but
 not comprehended by the lineage mode? I think not. As Taylor
 suggests in his review of Hindess and Hirst, the relations of
 production established in the Asiatic mode of production are only
 compatible with communai forces of production, and I am inclined to r
 think that this is true of all tributary formations, since other
 quasi-feudal mechanisms for structuring and reproducing the labour
 process would be necessary if communal ones were abandoned. At the
 same time Tavlor also suggests that the state in the Asiatic mode
 of production is in fact obliged to intervene in the process of
 production by building canals and the like so as to secure its
 reproduction. (8) This type of large-scale intervention has been
 relatively uncommon in Africa, a fact which has prompted Coquerry-
 Vidrovitch to describe the African mode of production as the Asiatic
 mode without large works. (9) Other kinds of intervention do
 nevertheless occur. Among the Pedi regiments of young men are
 organised/...
- 5 -
 organised by the chief to build walls and agricultural terracing,
 while in the view of Guy, the Zulu state came into being at least
 partly because of the varied nature of Zulu ecology and the need
 to rationalise access to diverse kinds of pastures and soils.(10)
 More generally, the provision of security enables the reproduction
 of production, and prevents the Indiscriminate plundering which is
 often the fate of acephalous groups.(11) In the Swazi state, one
 sees these interventions of several similar kinds: control of
 access to various zones of pasturage, control of the size and
 reproduction of individual homesteads through witchcraft accusations,
 and control over the cvcle of agricultural production through the
 annual incwala ceremony, and through the withdrawal of young men
 from the agricultural cycle into centralised age regiments.(12)
 In sum then there seems to me to be sufficient articulation between
 the forces and relations of production for us to reconsider the
 possibility of tributary mode, although the task of actually
 specifying that in a way that is theoretically coherent I shall
 leave to those who are more theoretically sound.
 Social classes, politics and the state
 "No classes, no politics, no state" Is a passage that recurs
 throughout the pages of pre-Capitalist modes of Production. What are
 its implications for tributary modes of production in general, and
 for that of the Swazi in particular? Following Engels, Hindess and
 Hirst see the existence of the state and of politics as an effect
 of the social division of labour, i.e. of the existence of social
 classes. In social formations such as these "the political level
 exists as the necessary space for the representation of the interests
 of the various classes, and the presence of a state apparatus is a
 necessarv condition of the maintenance and functioning of the
 mechanism of appropriation of surplus labour by the ruling class".(13)
 Otherwise, where social classes are absent, the state and politics
 do not exist and the social formation is only constituted of economic
 and ideological levels. Given this definition it is obviously
 important to impart some precision to our notion of class, and to
 see/...
- 6 -
 see whether classes can then be identified in tributary, and
 particularly the Swazi, social formation. In Marxian theory the
 separation of society into social classes arises from the social
 division of labour between a class of labourers who are separated
 from ownership of the means of production and a class of non-labourers
 who control the means of production, and use this as the means of
 appropriating surplus labour. On the basis of this definition
 Hindess and Hirst reject outright the possibility of social classes
 In the primitive communist mode of production. Since the mode of
 appropriation of surplus labour is communal, they insist, there are
 ipso facto no classes, and hence no possibility of politics and the
 state.(14) Both Rey and Terray take the opposite view, asserting (
 that control by elders over women in the first case and over
 prestige goods in the other enabled the elders "to control the
 surplus product, the partial or total use of which is for the
 reproduction of relations of dependence between the direct producers
 and this group", and out of this process they consider class
 relations to emerge.(15) Without wishing to enter that debate
 there does seem to be one point that should be made, which is that
 if Hindess and Hirst see "the process of the formation of the
 state (ksj identical to the process of transition from primitive
 communism to some other mode of production", and if they consider
 transitions from one mode of production to another to be the product
 of class struggle, some kind of embrvonic class formation must be
 possible in that mode for the transition to ever occur-(16)
 It is that kind of embryonic class formation that seems to me to
 characterise tributary formations in general and that of the Swazi
 in particular. Its origins may perhaps be traced to the process
 outlined by MeiIlassoux.
 Through/-..
- 7 -
 Through historical accidents, usually due to contacts
 with foreign formations, a group takes for all its
 members the quality of 'senior* in relation to
 other groups considered collectively as minor. All
 the economic and social prerogatives of the elder are
 transferred to the dominant class* usually an aristocratic
 lineage. Prestations due to the elder become tributes
 due to the lord who may also gain control over the
 matrimonial policy of the community, and eventually
 over the means of production-land. (17)
 Meillassoux's conception suffers in some measure from the role it
 ascribes to historical accidents, which in themselves presumably
 ought to be explained. Thereafter however it is possible to see
 these embryonic classes taking firmer shape. To take the Swazi
 case as an example, participation in the trade to Delagoa Bay and
 the ruling groups1 monopolisation of its profits helped further
 boost the latter's power, which In turn was consolidated still
 further by their move to the Shiselweni district in the south of
 modern Swaziland, and the opportunities this presented for enforcing
 unequal access to unevenly distributed means of production!. Part
 of that process 1 have dealt with elsewhere, and I should like to
 turn instead to the second phase of Swazi expansion which took place
 at the beginning of the 1820s. (18)
 At this point it is worth noting the comments of Hindess and Hirst
 on the conquest situation. "Conquest11, they point out,
 does not of itself produce state domination. The
 conquering people are not phantoms, they existed prior
 to the conquest and they must have a social organisation
 and a mode of producing the means of subsistence. In the
 first instance, the conqueror's mode of production will be
 represented alongside that of the dominated people. The
 dominant people receive tribute which is redistributed
 according to their social institutions and relations of
 production. No state is formed by thLs relation, dominant
 people/subject people; the means of coercion to obtain
 tribute are provided by the Gentile constitution of the
 dominant people, and the subject people regulates Its own
 affairs by its own institutions. The fact of conquest does
 not/...
- 8 -
 not produce either class society or the state. The
 conditions of transition to class society, of the
 conversion of the conquerors into a non-labouring
 ruling class, are not given in conquest as such. If
 such a transition does take place then it is on the
 basis of class society and irreconcilable class
 antagonism that the state is formed, not on the
 basis of conquest. Conquest only explains certain
 conditions under which the state may be formed, it
 does not explain the mechanisms of the formation of
 the state. (19)
 Conditions In the Svazi conquest area almost exactly mirror those
 outlined above. For several years the Swazi resembled more
 closely an army of occupation camped out in hostile territory,
 than a settled administration. "In those early days11, one oral
 history recalls, "there were no chiefs, only princes and leaders
 of regiments" (20), and the same picture can be derived from the
 evidence of Swazi messengers to Captain Gardener when they visited
 Mgungundlovu in 18.35. The capital of Swaziland, they told him,
 was Elangeni, and a little to the south was another village of
 Lobamba, which between them housed the entire male population of
 the Swazi, numbering no more than a few hundred men.(21) The
 messengers were evidently exaggerating, no doubt for Zulu ears* as
 we know of much more extensive Swazi settlement in this period,
 especiallv in the South.(22) Yet in the area of conquest there /
 was an element of truth In what they said. Few of Sobhuza's
 brothers or sons were assigned chiefdoms in the central areas until
 the closing years of Sobhuza's life, and the type of 'placing* to
 which Kuper refers did not occur on any extensive scale until the
 reign of his successor.(23) The history of Maphalaleni illustrates
 the trend. Maphalaleni was established for LaNdwandwe, one of
 Sobhuza's favourite wives, but so late in Sobhuza's reign that by
 the time she got there Sobhuza was already dead.(24) The same
 pattern repeats Itself throughout central Swaziland. Neither Maloyi
 nor Malunge seem to have taken effective occupation of their chiefdoms
 in the Mbuluzane River area until the reign of Mswati, and on the
 north side of the Komati River none of the Hhohho district was even
 allocated/...
- 9 -
 allocated until the 1840s and 1850s.(25) In the south things
 were somewhat different. At least five of Sobhuza's sons were given
 chiefdoms there, but if Mantintinti Is anything to go by, they only
 took possession comparatively late in Sobhuza's reign.(26) After
 accompanying Sobhuza to Mdlmba, Mantintinti "never set foot alive"
 in the chiefdom he had been given, and it was only "during the time
 of the return of the princes" to neighbouring Velezezweni that his
 successor Mtfonga "was instructed to return".(27) In sum, thenf
 the story told to Gardiner is at least partially confirmed. In the
 area of conquest the Ngwane were, for most of Sobhuza's reign, a
 nation under arms. Little of the conquered territory was settled,
 ' and the bulk of the population clustered for security in military
 towns. Only in the final years of Sobhuza's reign did the situation
 begin to change. Men could now be spared to reinforce the south,
 and an administrative presence was gradually extended in the
 conquered zone. Imperceptibly a shift was taking place to a
 society less overtly parasitic, and less openly reliant on the naked
 use of force.
 The same process of integration accellerated in the reign of
 Sobhuza's successor Mswati (1839-1865). Almost as soon as Mswati
 -\ succeeded his father he was faced with a rebellion by his half-brother
 Fokoti, and once that had been put down Mswati's mother Tandile and
 his paternal uncle Malunge took it as the opportunity to set in
 motion a series of far reaching reforms. On the face of it these
 took mainlv political and ritual forms. The ritual supremacy of
 the king as expressed in annual lncwala (first fruits) ceremonies
 was bolstered by ritual importations from the Ndwandwe, and
 Swaziland's military and administrative structures were systematised
 bv creating nation-wide age regiments as the framework of
 Swaziland's military organisation, and by establishing a far more
 extensive network of roval villages to serve both as rallying points
 for regiments and as centres for monitoring and supervising local
 political activity.(28) The rationale behind these changes was
 not/...
- 10 -
 not however solely political or ritual. The withdrawal of
 young men from the agricultural cycle of their families homesteads
 involved the direct appropriation of surplus labour by the royal
 house since they were then set to work tilling the king's fields,
 while changes in the incwala ceremonies and the establishment of
 royal villages in the provinces were part of a wider process whereby
 members of the royal family were being dispersed Into the regions,
 as a means of securing control over all aspects of their activities
 including the process of production.(29) Why this was needed
 now is less easy to judge, but the most likely explanation is
 Swaziland's declining ability to raid for booty and tribute in the
 regions to the north and the west in the troubled period following
 Sobhuza's death. Under Sobhuza such raiding had been widely
 undertaken. Thonga traditions collected by Nachtigal are full of
 references to Swazi attacks, and when Trichardt passed by the Pedi
 in 1836 he was warned that the entire area to the west of the
 Steenkampsberg was under Swazi control.(30) By the time the
 trekkers arrived in the area, however, in the mid 1840s Swazi
 raiding had all but ceased, and it may well have been the shortage
 of surplus from these 'traditional* sources in a period of stress
 that caused a tightening of the mechanisms of surplus extraction
 in Swaziland itself.(31) r
 The programme of reform evoked immediate opposition and before
 long the regents were forced to back down by a coalition of regional
 interests,(32) Further action in the matter had to await the
 mid lSSO's as Mswati and his regents were now confronted in rapid
 succession bv the rebellions of two of his elder brothers and by
 two massed Zulu attacks. (33) 1832 however marked the end of
 7ulu attacks, and Zulu energies were henceforth consumed by internal
 wranglings over the succession.(34) As a result Mswati was able
 to bring the country more systematically under his control as is
 evidenced bv his attacks on the semi-autonomous Emakhandzambile
 chiefs/*..
- 11 -
 chiefs{meaning 'those found ahead' i.e. those chiefdoms which
 had occupied central and northern Swaziland prior to Sobhuza's
 arrival from the Southf] Of the nineteen Emakhandzambile
 chiefdoms about which I have definite information, fourteen suffered
 in some way or other during this period, and in the cases of the
 others it required the intervention of special factors or unusual
 circumstances to save them from a similar fate. Thus the Mnisi,
 the Thabedze, the Gamedze, the Mngometfulo, the Sifundza, the
 Masilela, and the Mavirobela were all attacked by Mswati's forces
 and it is also reported that the chiefdoms of the Mahlalela and
 Moveni's Magagula would also have experienced similar treatment had
 ' It not been for the intercession of chance on the one hand and a
 trusted royal relative on the other. (35) As for the others
 their autonomies were no less completely circumscribed, with the
 Ngwenya, the Oladla, the Mnclna and the Moyeni's Magagula being
 demoted and placed under trusted officers of Mswati.(36)
 It would of course be absurd to assume that all this was undertaken
 in the immediate interests of surplus appropriation. In the case
 of the Mnisi, e.g., their chiefdom was attacked because of the
 extensive rain-making powers their chief deployed, and ritual factors
 \ may have entered into Mswati's attack on others as well.(37)
 Nevertheless underlying most of these assaults one can see attempts
 to extend politico/economic control. In many instances it is
 difficult to separate these two levels. The Mavimbela, e,g., were
 attacked because they refused a royal wife from Mswati, and on the
 face of it were reacting against the extension of political control.
 Yet for them there was another dimension to the problem, since
 marrving a princess involved the pavment of inflated sums of
 bridewealth and so meant a heavy drain of cattle as well.(38) In
 other cases economic issues are more clearly defined. Thus when
 Mswati tried to extort tribute from the Sifundza and Masilela
 peoples during a period of drought for example, his party was
 intercepted/...
- 12 -
 intercepted as it returned and stripped of all that it had taken.
 Mswati did not react immediately but, according to tradition,
 bided his time until the offenders fears had been lulled. A
 more serious consideration was probably fear of the Zulu, since
 it is likely that these events took place in the drought of 1848.
 Once the Zulu threat had begun to recede however the Masilela
 were made to pay dearly for their crimes. A hunting party was
 arranged to which the Masilela were summoned, and they were then
 surrounded and annihilated by the rest of the assembled host (39).
 The process whereby Mswati extended and rationalised his /
 Dolitical and economic control over subject groups I would see
 as marking the beginning of class society and the emergence of
 the state. In the place of random and indiscriminate plundering
 one now finds the development of more institutionalised mechanisms
 for the appropriation of surplus, whose volume was at the same
 time kept within reasonable limits by the need to retain the
 loyalty and cooperation of subject groups against external enemies
 like the Zulu. Side by side one also discerns the developing
 institutions of the state; the age regiments, which socialised
 the youth of subject peoples into a sense of national identity,
 and whose labour and booty raiding enriched the dominant class; (
 the expanded 11bandla or national council which represented all
 interests in the country and whose participation was required for
 all major political decisions Lthough here Hindess and Hirst's
 comments on the means of representation should also be rememberedj(40);
 the annual incwala_ ceremony in which the king and his people
 are symbolically renewed each year, and so on. Whether at this"
 embryonic stage they should be considered as existing entirely
 independently of the ruling groups is obviously open to question.
 Yet the same question remains open in the case of the feudal
 state. The extent to which or manner in which this operates
 "as/
- 13 -
 " as the necessary space for the representation of the various
 classes " in the feudal mode of production is never properly
 elaborated by Hindess and Hirst, leaving one with the sneaking
 suspicion that even here it only partially fulfills that role.
 The last question that I wish to consider in this section
 is the precise configuration of class interests represented in
 the early Swazi state. Terray in his recent article on "Classes
 and Class Consciousness in the Abron Kingdom of Gyaman",
 ultimately ducks that very question. He begins by defining
 classes in terms of relations to the means of production, but
 then goes on to characterise them as a relation of exploitation (41).
 This will not do, for as Hindess and Hirst among others have
 shown, there may be exploitation without classes ever arising,
 as in the case of "banditry and extortion by political fbosses'" . (42)
 Can we therefore say we have classes in pre-colonial
 Swaziland? I believe we can, if only in embryonic form. Firstly,
 land, the basis of the economic system and the principle means
 of production , is controlled by the rulers and can be redistributed
 by them in a variety of ways. Chiefdoms can have part of their
 territory taken away from them and allocated to other groups;
 they can be removed en masse to other parts of the country or
 thev can have their local rulers replaced by relatives and
 functionaries of the king.(43) Thus, while each individual
 subject has acccfss to the means of production, he can have his
 access restricted to means of greatly inferior worth. Much
 the same goes for cattle and for wives* By a variety of
 mechanisms such as witchcraft accusations, control over the
 appropriate combinations of pasturage, variable brtdewealth pay-
 ments, and preferential marriage patterns, the accumulation of
 cattle/.....
- 14 -
 cattle can be restricted for the commoners and largely confined
 to the ruling class. If one adds to this an appropriation of
 surplus on the basis of these divisions which goes to maintain
 a non-labouring ruling class, then one must concede the emergence
 of class society if only in attenuated form.
 But the question remains who precisely constitute these
 classes ? In his study of the Abron kingdom of Gyaman Terray
 sees a three-class division between a non-labouring group of
 aristocrats and a class of working slaves on the one hand, and
 tribute producing peasants on the other, but it seems questionable (
 to me how clear cut this distinction was(44). Are we to believe
 for example, that the Abron aristocracy was an entirely non-
 labour;ng class and that neither'its women or cadets engaged t
 in productive work? Certainly among the Swazi this was not the
 case, for there a large section of the ruling group took part
 in such pursuits. Instead what ve find there Is a more complicated
 situation, with tribute being extracted from subordinate groups
 in the various forms that we have mentioned, and then being partly
 concentrated in the upper echelons of the ruling class and partly
 filtering down. The important thing to remember here
 however is the barrier that existed to surplus dribbling down V-
 to the groups from which It had been drawn. Essentially this
 was constituted out of Swazi marriage practices* Unlike their
 Nguni counterparts the Swazi'practised a system of preferential
 in-kin marriages, which tended to restrict cattle and other wealth
 to within the ruling class. These restrictions, It is true,
 were neither permanent nor impermeable. The marriage of
 matrilateral cross cousins is much more flexible than its
 patrilateral parallel variant which fixes genealogical relationships
 in a virtually unchanging rhythm , and it was this the Swazi
 practised/
- 15 -
 practised rather than the more inflexible kind (45). The
 Swazi moreover marry their classificatory cross cousins and not
 their actual mother's brother's daughters, and have preferential
 marriages with a variety of other kin (46). Finally a more
 general political expediency could involve an entirely different
 order of preference, and led Mswati to exchange wives with
 both Magagula chiefs (47). from the broader structural point
 of view however the relationships which developed with the
 conquered were decisively different from those of other Nguni
 groups. Although offering a more flexible range of marriage
 ) options than parallel and true cross cousin marriage, the
 various Swazi marriage preferences still concentrated them within
 a restricted group of kin (48). The broad effect of this
 was that while a degree of social and political mobility was
 permitted, differences of political and social status were per-
 petuated, which persist to this day. In Sobhuza's time these
 were at their most intense. Marriages were confined politically
 within the dominant Ngwane, and spatially to their military encamp-
 ments, while wealth tended to circulate in the same restricted
 group. Under Mswati there was a blurring at the edges of these
 , categories, but otherwise they remained essentially intact,
 leaving marriage practices to continue as a crucial determinant
 of class.
 Captives and Booty Raiding in the Tributary Mode of Production
 Jack Goody in his essays on Tradition, Technology and the State
 in Africa sees captives and tributary raiding as an integral part
 of the tributary mode of production(although he does not use that
 term). The low fertility of African soils, the relatively low level
 of technology in most African societies (which make human muscle
 power virtually the only means of power available), and low
 population/...
- 16 -
 population densities, together make it impossible to extract
 surplus from peasant communities on a scale similar to much of
 Asia and Europe, and ensure that it is labour rather than land
 that is the factor of production which is scarce. Out of this
 arises bootv raiding as an important means of acquiring surplus,
 and particularly the raiding of human booty as a source of labour
 power. (49) Terray in his Gyaman study refines these ideas further.
 The low level of surplus appropriation from peasant communities
 he suggests is a consequence of the need of ruling aristocracies
 to enlist the support of peasant communities in the face of threats
 from outside. At the same time the fear of creating a permanently
 dispossessed and disaffected slave population, as well as perhaps
 other factors, ensures the gradual absorption of the captive and
 his family into Gyaman society, and created the need to secure
 further supplies of captives as a means of making good that loss.(50)
 A number of the same points apply to Swazi society.
 The part played by captives in the Swazi economy has been
 considerably larger than is generally imagined.(51) Whether this
 was always the case is difficult to say. When the Swazi was
 centred on Shiselweni in the south of Swaziland between circa 1770 -
 1820 e.g. their highly pastural economy may have meant that their (
 needs for labour power were correspondingly reduced. The situation
 changed once they moved from Shiselweni to Ezulwini. Here the
 Dastorage was less rich and agricultural production more necessary,
 and henceforth captives were raided far and wide.(52) According to
 Ndambi Mkhonta the Ezulwini village once boasted large numbers of
 captives, and the same is likely to be true of all royal capitals.(53)
 Other examples which I have come across without directly questioning
 on the subject are those of the Dube who were attacked and enslaved
 in the reign of Mswati, and of the Thabede who suffered a similar
 fate at. much the same time. (54) Sources of slaves fell in two
 broad/...
- 17 -
 categories, although the distinction was probably blurred in the
 early davs of the state. The first group comprised non-Swazi who
 were raided outside their kingdom's boundaries (the Tltfunjwa).(55)
 In Sobhuza's early conflicts with the chiefdoms of Magoboyi and
 Mkise. for example, captives were taken and their presence in Swazi
 society was used in later years to justify Mswati's right to cede
 the eastern Transvaal* (56) Later it was the Thonga who bore the
 brunt of these attacks and in was they-who in this case were most
 usuallv traded as slaves. The other major source of supply were
 children seized from households within the Swazi kingdom (the
 ) Tigcili). (57) As Tikuba told Stuart in 1898
 It often happened that when a person was killed
 for some crime or other and his cattle and children
 seized, those children were taken by the Swazi and
 sold to the Boers in the Transvaal. (58)
 The Berlin missionary, Merensky9 reported on similar practices after
 his visit to Swaziland in March 1860. "Even now" he wrote in his
 diarv,
 if a man of his (Mswati's) people has many daughters
 or good cattle his soldiers come, surround the Kraal,
 murder the old, and take the goung people and cattle
 as booty. Children are being sold or given to the
 'great of the realm1.(59)
 What were the implications of these developments for the emergence
 of the state and for the development of classes? Terray would
 have us believe that they signal the emergence of slave classes and
 of a slave mode of production, but as Hlndess and Hirst show a slave
 mode of production presupposes private property in land and slaves,
 and a form of commodity exchange corresponding to both*(go) Moreover,
 given the rights and protection made available to captives in
 Swazi society, it seems more appropriate to regard them as a group
 of perpetual minors rather than as a clearly defined class.(61)
 The/...
- 18 -
 The seizure of captives however did have other implications for
 class formation, particularly when considered against the trade
 In captives to the Transvaal. Without going into any details
 this seems to have grown to substantial proportions by the middle
 of the 1850's and then to have boomed to new levels for the first
 half of the next decade.(62) The question that is inevitably
 raised by these developments is to what extent the seizure of
 captives for trade become an object in itself in the attacks |hat
 were made on offending subject groups, and to what extent, in
 Rodney's terms, this.led to the social degeneration of Swazi
 society.(63) In other words to what extent 4o we see a robber ,
 relation rather than a class relation coming to characterise
 Swazi society? I would argue that for the most part the seizure
 of captives did not become the objects of these attacks; that
 they were a by-product of rationalising economic and political
 contro, and that once this rationalisation was completed internal
 captive-taking largely ceased. In its place however one finds a
 shift of emphasis to Mozambique, the devastation and impoverishment
 of that area in the. pursuit of captives, and perhaps some of the
 earliest origins of the underdevelopment of Southern Mozambique,
 P.L. Bonner
 Department of History
 University of the Witwatersrand,
 Johannesburg