Published online by Cambridge University Press: 22 January 2009
The main interlacustrine kingdoms have been presented, on the evidence of their royal genealogies recalling up to thirty reigns, as stretching back to a ‘Chwezi’ period some five centuries ago. This view was promoted especially in the Kitara zone, comprising Bunyoro and regions to its south and, as a close linguistic grouping, extending to Nkore, Karagwe and Buhaya. Rwanda to the south-west and Buganda to the east, though each rather distinct, share some of the same cultural and traditional features. In the central Kitara zone it has been further argued that the ‘Chwezi’ period is represented by various impressive archaeological sites – hilltop shrines, notably at Mubende, with special and archaic objects; complex earthwork enclosures at Bigo and elsewhere; and the concentrated settlement nearby at Ntusi. Certain of these have been claimed as Chwezi royal capitals of ancient Kitara, and specific features have been compared with royal abodes of recent centuries. Such literal interpretation, let alone royalist manipulation, of oral traditions is now considered too simplistic; not only are the Chwezi generally regarded as gods or mythical heroes, but also the role of archaeology is now seen as something more positive than the mere verification of verbal evidence.
Renewed archaeological research indicates that Ntusi was occupied from about the eleventh to the fifteenth centuries a.d. and that the earthworks, including Bigo, and the settlement on Mubende hill fall into the latter half of that span. This cultural grouping thrived on a combination of cattle-keeping and grain cultivation, as is especially clear at Ntusi on fertile ground in the midst of the Bwera grasslands. It may have been the growing strains of a delicately balanced economy as competition increased for cattle and the pastures which led to its eventual breakdown. During the last half-millennium Bwera has been a peripheral and lightly populated district between Bunyoro, Nkore and Buganda. It is difficult to imagine these later kingdoms developing directly out of a supposed ‘Chwezi’ one based at Ntusi and the Bigo constructions.
Two periods of marked change are discernible therefore, one around the middle of this millennium, the other at its beginning. That earlier, mid-Iron Age, revolution witnessed the introduction of cattle on a large scale and the first intensive exploitation of the interlacustrine grasslands. Cattle becoming then an economic asset, it may be inferred that ownership of stock and defence of the pastures became sources of prestige and patronage, with obvious social, political and military implications. This situation opened opportunities for other specializations, including the production of salt for distant distribution. Traditions concerning gods and heroes, and the continuing popular chwezi cults, illustrate the changes and may also echo the cultural and economic importance of iron and its working among agricultural populations from before the pastoral revolution.
4 E.g. Johnston, H. H., The Uganda Protectorate (London, 1902)Google Scholar; MrsFisher, A. B. (Ruth), Twilight Tales of the Black Baganda (London, 1911Google Scholar: mostly about Bunyoro, despite its inappropriate title); Père Gorju, J.-L., Entre le Victoria, l'Albert et l'Edouard (Rennes, 1920)Google Scholar; and the various works of J. Roscoe from 1911.
5 This practice began with Speke who, in the published version of his Journal (ch. 18), records how during his stay in Bunyoro late in 1862 he called one day on King Kamrasi, taking a Bible to explain ‘all I fancied I knew about the origin and present condition of the Wahuma… beginning with Adam…’. Then, ‘proceeding with the Flood’, he pointed out the races of mankind, white, tawny and black, descending from Japhet, Shem and Ham. A few years later Stanley, (Dark Continent, 345)Google Scholar noted that Mutesa, the king of Buganda, was equating Kintu, the supposed founder of that kingdom, with Ham.
6 These written works are listed by Cohen, D. W., ‘A survey of interlacustrine chronology’, J. Afr. Hist., XI (1970), 177–201, here 182–4.CrossRefGoogle Scholar The earliest to be published was that of the prime minister, Sir Kagwa, Apolo, Basekabaka be Buganda [‘The Kings of Buganda’] (c. 1901.Google Scholar See M. S. M. Kiwanuka's translation and introduction, Nairobi, 1971). But ten years earlier a Ganda king-list, which deviates in several details from Kagwa's, had been assembled by another ‘high official’, S. Mugwanya: see Wrigley, C. C., ‘The kinglists of Buganda’, History in Africa, I (1974), 129–39.CrossRefGoogle Scholar See also Henige, D. P., ‘Reflections on early interlacustrine chronology: an essay in source criticism’, J. Afr. Hist., XV (1974), 27–46CrossRefGoogle Scholar, and ‘Ganda and Nyoro kinglists in a newly literate world’, in Miller, J.C. (ed.), The African Past Speaks (Folkestone, 1980), 240–61Google Scholar; also, Twaddle, M., ‘On Ganda historiography’, History in Africa, I (1974), 85–100.CrossRefGoogle Scholar
7 Journal of the Discovery, ch. 9. See also Henige, , ‘Reflections’, 250–1Google Scholar, on Grant's testimony. Speke found it more difficult to obtain a royal genealogy for Bunyoro, recording only the last three kings. Visitors in the 1870s managed to add only two more names: see Henige, , ‘Ganda and Nyoro’, 244–5.Google Scholar
8 In 1875 Stanley was able to gather much fuller details than Speke had on Ganda dynastic history and legend (Dark Continent, 344–80) and to produce a list of 35 kings (p. 380). Henige, ‘Reflections’, makes a much less sanguine assessment of the evidence and argues for substantial changes in and expansion of the ‘official’ king-list between 1862 and 1875. See also Wrigley, ‘The kinglists’. Stanley was certainly aware of the extent to which Mutesa's court was already acquainted with rudiments of Islamic belief and learning (see Dark Continent, ch. 12); and with the help in translation of James Dallington, a UMCA pupil from Zanzibar of Nyasa extraction, and of a scribe called Idi, he produced ‘an abridged Protestant Bible in Kiswahili’. For the impact of this abridged bible, see Mukasa, Ham in Uganda J., II (1934), 67–8.Google Scholar According to Rowe, John (‘Myth, memoir and moral admonition’, Uganda J., XXXIII , 19)Google Scholar, literacy in Arabic script and acquaintance with the Koran had reached Buganda only a few years previously, between the visits of Speke and Stanley. See also Oliver, Roland, The African Experience (London, 1991), 203Google Scholar; and Oded, A., Islam in Uganda (Jerusalem, 1974).Google Scholar
9 Apparently he did not reflect on whether each reign was a true generation. In fact the list does not seem to add up to twenty.
10 For instance Ingham, K., ‘The amagasani of the Abakama of Bunyoro’, Uganda J., XVII (1953), 138–45Google Scholar; Oliver, R., ‘The royal tombs of Buganda’, Uganda J., XXIII (1959), 124–33Google Scholar, discerning institutional changes at certain periods. For descriptive and symbolic aspects, see Ray, B., ‘Royal shrines and ceremonies of Buganda’, Uganda J., XXXVI (1972), 35–48.Google Scholar The essential statement on comparative chronology is Oliver, Roland, ‘The traditional histories of Buganda, Bunyoro and Nkore’, J. Royal Afr. Inst., LVIII (1955), 111–17.Google Scholar
11 Exceptions have been made, notably in Rwanda: Noten, F. L. Van, Les tombes du roi Cyirima Rujugira et de la reine-mère Nyirayuhi Kanjogera (Tervuren, 1972).Google Scholar On shrines, not necessarily burials, materials in them and collections of royal insignia, see Schmidt, P. R., Historical Archaeology: A Structural Approach to an African Culture (Westport, Conn., 1978)Google Scholar; and Sassoon, H., ‘Kings, cattle and blacksmiths: royal insignia and religious symbolism in the interlacustrine states’, Azania, XVIII (1983), 93–106.CrossRefGoogle Scholar
13 Posnansky, M., ‘The excavation of an Ankole capital site at Bweyorere’, Uganda J., XXXII (1968), 165–82Google Scholar; cf. Reid, and Robertshaw, , ‘Ankole capitals’, 86.Google Scholar A broadly comparable capital which was occupied probably during more than one reign over the same general span is that of Ryamurari in Ndorwa. This former ‘Hima’ kingdom lies south of Nkore and west of Karagwe (that is within the north-eastern part of the modern state of Rwanda). The site is of some size and preserves a number of enclosures and other visible features. Excavations undertaken at Ryamurari in 1980–1 are reported summarily by Tshilema Tshihiluka in annexe 10 of Noten, F. Van, Histoire archéologique du Rwanda (Musée royal de l'Afrique Centrale, Annales 112), (Turvuren, 1983), 149–253Google Scholar; and in Africa-Tervuren, xxix (1983), 19–26.Google Scholar I am grateful to M. Tshihiluka for further information about this site and its potential.
14 Journal, ch. 10 et seq. For other accounts and illustrations, see Gutkind, P. C. W., The Royal Capital of Buganda (The Hague, 1963).Google Scholar
16 Baker, , Albert N'yanza, 78.Google Scholar Having come from the coast, Speke employed Swahili prefixes for territories (U-), not those of the interlacustrine Bantu languages (Bu-) normal in later literature.
17 1863, ch. 9. From an allusion in the previous chapter of his Journal, it appears that Speke's information about Kitara, its former greatness and extent, was not simply Nyoro indoctrination. Apparently while in Karagwe, the traditions of that kingdom had been told to him by reference to greater Kitara.
18 But Uzoigwe, G. X. (ed.), Anatomy of an African Kingdom (trans. Muganwa, T.) (New York, 1973)Google Scholar, from original by Nyakatura, J. W., Abakama ba Bunyoro Kitara (Kampala, 1947), 26, 74 and 127Google Scholar, indicates a special connection between the name Kitara and Kyaka district further south, nowadays in eastern Toro and bordering Nkore and Bwera. This agrees roughly with Gorju's discussion (Entre le Victoria, 39–42) of the meaning and location of Kitara.
19 Uzoigwe, Anatomy, 1. See also p. 30 for an even broader sphere extending as far as ‘Abyssinia’. Note also the title of Dunbar's, A. R. book, A History of Bunyoro-Kitara (Nairobi, 1965).Google ScholarBuchanan, Carole, ‘Perceptions of ethnic interaction in the East African interior: the Kitara complex’, Int. J. Afr. Hist. Studies, XI (1978), 410–28CrossRefGoogle Scholar, speaks of the ‘Kitara complex’ in reference to the broad interlacustrine zone and its cultural traditions, in effect as a term of convenience (as we are inclined to do here). Her attempt to justify her usage – ‘after the name of the oldest state system associated with the region’ (p. 413) – may convey the impression, surely not intended, that she is intellectually aligned with Nyakatura and Dunbar.
20 This is an impressionistic conspectus based on local perceptions of linguistic closeness and distance. For classification of the Bantu languages of this zone, see Bryan, M. A., The Bantu Languages of Africa (London, 1951), 104–9.Google Scholar For a more recent discussion and utilization of comparative linguistic evidence, see Schoenbrun, D. L., ‘Early history in Eastern Africa's Great Lakes Region: linguistic, ecological and archaeological approaches, c. 500 BC to AD 1000’ (Ph.D. thesis, University of California, Los Angeles, 1990).Google Scholar
21 Best synthesized by Oliver, : ‘Discernible developments in the interior, c. 1500–1850’, in Oliver, R. and Mathew, G. (eds.), History of East Africa (3 vols.) (Oxford, 1963), IIIGoogle Scholar, and ‘The East African Interior’, in Oliver, Roland [ed.], Cambridge History of Africa, Vol. 3: c. 1050–1600 [Cambridge, 1977], 621–69.CrossRefGoogle Scholar ‘Chwezi’ barely figure in this way in the history of Buganda, which is marginal to the Kitara cultural zone. See Kagwa, Basekabaka, and also Kiwanuka, M. S. M., A History of Buganda (London, 1971).Google Scholar
22 The celebrated statement of this critical line is that of Wrigley, C. C., ‘Some thoughts on the Bacwezi’, Uganda J., XXII (1958), 11–17.Google Scholar
23 See Berger, Iris, Religion and Resistance: East African Kingdoms in the Precolonial Period (Tervuren, 1981)Google Scholar; also Berger, and Buchanan, C. A., ‘The cwezi cults and the history of western Uganda’, in Gallagher, J. T. (ed.), East African Cultural History (Syracuse, New York, 1976), 43–78Google Scholar; Heusch, L. de, Le Rwanda et la civilisation interlacustre (Brussels, 1966)Google Scholar; Cohen, D. W., ‘The cwezi cult’ (review of de Heusch), J. Afr. Hist., IX (1968), 651–7.CrossRefGoogle Scholar
25 This is to simplify the argument in Berger, Religion and Resistance.
26 Lutatenekwa, F., ‘The Bachwezi: ancient kings or living spirits?’, Kale (University of Dar es Salaam), I (1972), 1–6.Google Scholar
27 See Fisher, , Twilight Tales, 98–9Google Scholar; also Johnston, , Uganda Protectorate, 587–97Google Scholar, drawing on several British informants, including Rev. A. B. Fisher and George Wilson. The generally more north-westerly currency of Ndahura's fame, by comparison with that of Wamara, is indicated further by the former's ‘birthplace’ recently shown to Kamu-hangire and Robertshaw north-west of Mubende. Similarly of the ‘middle-rank’ Chwezi, Mulindwa seems to be revered on balance rather to the north of his ‘brother’ Mugenyi, the latter being especially associated with the grasslands of Bwera and their famous herds. See below, n. 34 and at n. 38.
28 Lanning, E. C., ‘Some vessels and beakers from Mubende’, Man, LIII (1953), 181–2CrossRefGoogle Scholar; and ‘Excavations at Mubende Hill’, Uganda J., xxx (1966), 153–63.Google Scholar See also the discussion in Berger and Buchanan, , ‘The cwezi cults’, 53–6Google Scholar, mentioning copper spears and Nyoro royal accession ceremonies here. The allusion is probably to iron spears with copper decoration: on these and other Mubende regalia, see Lanning, , ‘The surviving regalia of the Nakaima, Mubende’, Uganda J., XXXI (1967), 201–11.Google Scholar New archaeological work on the summit of Mubende in 1987 is reported by Robertshaw, Peter in Nyame Akuma, XXX (1988), 37–8Google Scholar, and in the Annual Report of the British Institute in Eastern Africa for 1987–88 (London, 1988).Google Scholar
29 Lanning, , ‘Excavations’, 154Google Scholar, gathered information that ‘not only were dwellings and spirit huts ransacked and burnt, but the tomb huts of past priestesses buried nearby were also damaged. Whilst certain belongings and regalia were saved much was lost, including, it has been said, the drums of the priestess, a ceremonial elephant tusk and a drum called Rusama which had previously been kept at the centre of worship at Masaka Hill, thirty miles south-west of Mubende. Yet the spirit of Ndaula persisted and with the end of hostilities the Nakaima returned to her hill-top abode.’
30 The transfer after 1900 from Bunyoro to Buganda not only of Mubende itself but of the whole of this district (‘the lost counties’), containing many Nyoro traditional sites and royal graves, helped ensure a sentimental attachment.
32 See note 28, especially Robertshaw.
33 See Schmidt, ‘A new look’, on eastern Buhaya, and especially the Rugamora Mahe site. Note also the position of early Iron Age objects in later sites in Rwanda, which include both real and supposed royal graves: Van Noten, Les tombes.
34 This Masaka Hill is not to be confused with the town of Masaka over 100 km to its south-east. The shrine, its relics and features were described some years ago by Lanning, E. C., ‘Masaka Hill, an ancient centre of worship’, Uganda J., XVIII (1954), 24–30.Google Scholar Ephraim Kamuhangire was able to visit Masaka hill again in 1991 and to confirm some of the features. For further discussion see Berger, and Buchanan, , ‘The cwezi cults’, 58f.Google Scholar Note also their account of the shrine of Mulindwa further north in Bunyoro (pp. 56f).
35 In Nyakatura's account (Uzoigwe, Anatomy), Wamara succeeded Ndahura as king of Kitara. When he departed from the land of the living, the kingdom was left to the Bito. This is close to the account of the legend in Ruth Fisher's Twilight Tales.
36 See Fisher, , Twilight Tales, 99Google Scholar, followed by Gorju, , Entre le Victoria, 48Google Scholar; also Johnston, , Uganda Protectorate, 595.Google Scholar Note also Nkore informants recalling Wamara by pointing in this direction, to Mawagola county if not Ntusi specifically: Reid, and Robertshaw, , ‘A new look’, 84–5;Google Scholar cf. Oliver, , ‘Ancient capital sites’, 53.Google Scholar The question of a traditional connection with Ntusi is discussed below. The lake called Wamala lies a little distance to the east across the Katonga. The name recurs elsewhere, usually as ‘Wamala’ in Buganda.
37 D. L. Baines in 1909 (Uganda Official Gazette, used by Wayland, E. J., ‘Notes on the Biggo bya Mugenyi’, Uganda J., II , 21–3)Google Scholar; and Gorju, , Entre le Victoria, 47–8, 53–5.Google Scholar The latter's information does not seem to have been entirely independent, while Baines, from his own account, appears to have relied as much on Baganda informants as on local Hima knowledge (or had the latter channelled to him by the former). For critical discussion, see Schmidt, P. R. in Robertshaw, P. (ed.), A History of African Archaeology (London, 1990), 256f.Google Scholar The interminable debate over the ‘true’ meaning of the name mugenyi, and over which Mugenyi might be meant, only underlines, when considered alongside other attributions in the district, the historical–supernatural association. The evidence does not tell us anything about who ‘built’ Bigo or occupied it. Nor does it help in dating, except to emphasize the antiquity of the place, beyond living memory.
38 For references see note 34; also Williams, F. L., ‘Myth, legend and lore in Uganda’, Uganda J., X (1946), 67.Google Scholar
40 For descriptions of these and of numerous other sites in this region archaeologists remain indebted to the various articles of Eric Lanning in the Uganda Journal, in particular ‘Ancient earthworks in western Uganda’, xvii (1953), 51–62Google Scholar; ‘The Munsa earthworks’, xix (1955), 177–82Google Scholar; ‘The earthworks at Kibengo’, xxiv (1960), 183–96Google Scholar; and for associations with Kateboha, xxiii (1959), 186–8. New investigations at Munsa by Peter Robertshaw in 1988 help in the dating and the sequence of activity, although a follow-up campaign is intended for further elucidation.
41 Schmidt, (‘A new look’, 263–4)Google Scholar argues, largely from the radiocarbon evidence, for a generally late dating and the successive use of the site possibly at separate periods. Despite his valid criticisms of the methodology of previous dating estimates, his conclusions seem unconvincing in the light of the (virtually negative) oral evidence and current archaeological knowledge of the region. That would place Bigo within the last 1000 years but not in the last 300. The radiocarbon readings from samples obtained during the 1960 excavations do not in general contradict this but are too imprecise to help the argument further. The ceramic comparisons with Mubende, Munsa and nearby Ntusi suggest at the present stage of research a dating for Bigo very roughly around five to seven centuries ago. The dating of the Munsa works is similarly unsatisfactory but appears to be broadly similar to that of Bigo, again with evidence of settlement before the construction of the earthworks, or at least of some of them.
42 This relies on Jeremy Meredith's observations of the pottery collections, site by site and layer by layer, analysing by both shapes and decoration, including a variety of roulettes. Most of the radiocarbon measurements obtained so far are set out by Sinclair, P. J. J., ‘Archaeology in Eastern Africa: an overview of current chronological issues’, J. Afr. Hist., XXXII (1991), 179–219.CrossRefGoogle Scholar
43 See above, note 36.
44 See Wayland, , ‘Notes on the Biggo bya Mugenyi: some ancient earthworks in Northern Buddu’, Uganda J., II (1934), 25fGoogle Scholar, mentioning his own and Combe's previous excavations; also Lanning, , ‘Ancient earthworks’, Uganda J., XVII (1953), 56f;Google Scholar and ‘Ntusi: an ancient capital site in Western Uganda’, Azania, v (1970), 39–54.Google Scholar
45 Not only those known as ‘Ntusi male’ and ‘female’ but also bwogero. The latter group of mounds, consisting of a barren clay, is anyway unattractive to cultivators. The others (including Ntusi ‘male’ and ‘female’), consisting of thick organic soil, are suffering encroachment on their sides, the cultivators preserving the tops but not appreciating that they are modifying the shapes of the mounds and continually reducing their sizes.
46 It is speculated from time to time that the ‘male’ and ‘female’ mounds, as they are known, are more than simple rubbish tips but could have served a cultural or ritual purpose (as did perhaps the differently built mounds at Bigo?), either originally or subsequently. The top layers of the ‘male’ mound yielded a few pieces of glass ornament and cowrie-shell beads of Indian Ocean provenance (Reid, Andrew in Nyame Akuma, XXXIII , 27).Google Scholar These may indicate coastal contact about the thirteenth century, which is much earlier than otherwise imagined, or alternatively the use of these mounds in the nineteenth century by Hima herders, either for ritual purposes or simply as suitable places from which to watch their cattle.
47 Nowadays, with the largely Baganda settlement at Ntusi utilizing this thick soil, a large area is given to banana groves. Many of these are for beer; there are cooking bananas too, but in years of less than average rainfall these prove inadequate, and extra supplies are brought by motor transport from the wetter Masaka direction nearer to Lake Victoria. This suggests that in this parkland district an agricultural economy heavily reliant on bananas would not have been sustainable earlier. Sorghum is grown now in moderate amounts, mostly for beer, by combining with bananas. Very occasionally one sees a field of finger millet. Several crops of New World origin are grown – sweet potatoes, maize, cassava, beans and groundnuts – and these reduce the dependence on bananas.
48 Such putative manuring practices may possibly explain the numerous vertical pits dug through the upper and middle layers of the big rubbish mounds. The question whether some of the mounds adjoined cattle-kraals and consisted partly of dung piled with the domestic rubbish has also to be considered; it is hoped that soil and phytolith tests may help elucidate this point, which is of obvious importance for the interpretation of the integrated pastoral-agricultural settlement and economy of Ntusi.
49 A favourite situation for these sites is on the slopes of ridges above damp valleys. Andrew Reid's survey of the district (see Nyame Akuma, xxxiii , 26–9)Google Scholar has increased enormously the number of these sites known. He has excavated on certain examples. One at Kakinga, about 5 km south-east of Ntusi, seems on a single radiocarbon result to belong around the end of the Ntusi sequence, in the fifteenth or early sixteenth century. Alongside this, both Reid and Ephraim Kamuhangire have been investigating the pastoral ecology of the district and current Hima herding strategies. The overall discussion of Ntusi presented here owes much to their enquiries and collaboration. See Reid, D. A. M., ‘The role of cattle in the later Iron Age of southern Uganda’ (Ph.D. thesis, University of Cambridge, 1991).Google Scholar
50 See Sutton, , ‘Ntusi and the dams’, Azania, XX (1985), 172–5.CrossRefGoogle Scholar More thorough inspection of the bwogero mounds in 1987–8, and a test excavation in the loop mound, confirm that they consist of barren soil scraped from the ‘basins’ in the valley bottom. It could still be argued of course that the positions and heights of the mounds, as well as the configuration of the loop feature to make a small enclosure, had some other purposes. But the virtually complete lack of cultural material in this particular part of Ntusi makes sensible speculation very difficult.
51 Notably at two sites, one an iron furnace, the other with typical EIA ceramics, in Rwanda: Grunderbeek, M.-C. Van et al. , Le premier âge du fer au Rwanda et au Burundi (Brussels, 1983), 42;Google ScholarNoten, F. Van, Histoire archéologique du Rwanda (Tervuren, 1983), 32.Google Scholar For background on the Early Iron Age in equatorial and southern Africa, see Phillipson, D. W., The Later Prehistory of Eastern and Southern Africa (London, 1977), esp. 146–7Google Scholar, and African Archaeology (Cambridge, 1985), 171f.Google Scholar In several parts of southern Africa cattle were present late in the first millennium a.d., and it has been claimed that they were kept earlier there by certain of the initial Iron Age groups (Huffman, T. N., ‘Ceramics, settlements and Late Iron Age migrations’, African Archaeological Review, VII , 169–70).Google Scholar While that argument is not necessarily relevant for the interlacustrine zone, it is necessary to gain fuller information on the nature of Early Iron Age settlement in this latter region, especially in view of the generally held assumption of the expansion of iron-using Bantu cultivators from there into southern Africa early in the first millennium a.d. Schoenbrun (‘Early history’, 98) argues that grassland was gaining in the interlacustrine region at the expense of forest during the first millennia b.c. and a.d., largely because of clearance for both agriculture and iron-working, and that around 1000 a.d. the optimal environments for cattle-herding had been attained. The history of the latter is therefore much shorter here than in the highlands east of Lake Victoria, where cattle are attested well before the Iron Age.
52 The pastures of the interlacustrine region are among the lushest in Africa. Here more than anywhere in the continent it has been possible to maintain highly specialized herding in coordination with cultivating communities and to select medium-sized cattle of a not especially drought-resistant breed. Another striking feature of interlacustrine pastoralism past and present, when compared with that of Maasailand and other regions, is the scarcity of goats and even more of sheep.
53 Meredith's studies (see above) indicate that while string roulettes (both twisted and knotted) and the rarer carved-wooden ones were definitely new in this middle Iron Age period, there may be more continuity from the Early Iron Age in the basic shapes of the vessels than has been appreciated. Later types are generally more robust and their texture appears to be more coarse, but this is partly explicable functionally, by the demand at this period for bigger vessels. That in itself is an indication of new cultural and economic factors. On types of rouletted decorations and a working terminology, see Soper, R., ‘Roulette decoration on African pottery: technical considerations, dating and distributions’, African Archaeological Review, III (1985), 29–51.CrossRefGoogle Scholar Until recently, it has been generally thought that string or knotted roulettes were not employed in the interlacustrine and other eastern African regions until early in the second millennium a.d. But the Ntusi sequence now shows that these decorative techniques were normal there from the start of its sequence at the beginning of this millennium. A dating two or three centuries earlier has been claimed, on the evidence of individual radiocarbon measurements, in Rwanda: see Grunderbeek, Van, Le premier âge, 44, 48.Google Scholar Since the latter part of the first millennium a.d. is so poorly recognized throughout most of the East African interior, an earlier dating for this (and indeed other supposedly ‘later’ Iron-Age developments) cannot be ruled out. Further observations on rouletting and a range of historical considerations are offered by Desmedt, Christiane: ‘Poteries décorées à la roulette dans la Région des Grands Lacs’, African Archaeological Review, IX (1991), 161–96.CrossRefGoogle Scholar
54 This section relies heavily on ethnographic observations, which must be projected backwards with caution since detailed archaeological correlations are lacking for most items. Milk containers may also be carved from wood; the remains of a wooden vessel have been recovered from the Ntusi ‘male’ mound, in a level of about the twelfth– thirteenth century.
55 Fine, black graphited wares – see Posnansky, M., ‘Pottery types’, J. Afr. Hist., II (1961), 193Google Scholar – may however be generally late. Smoking pipes of recent centuries are made of this; they are not found in the Ntusi sequence or at Bigo and Mubende (except in the demonstrably late nineteenth- or twentieth-century settlement).
56 Discussed by Sassoon, ‘Kings, cattle and blacksmiths’.
57 See in particular Berger, Religion and Resistance. Iron-working in recent times and the later Iron Age appears to have been an industry quite separate from potting in most areas, whereas there are hints that in the Early Iron Age these two crafts which manipulate in their different ways both minerals and fire may have been closely connected.
58 Grant found Kibiro salt prized in Bunyoro in 1862: A Walk Across Africa, 296. See also Speke, , Journal, ch. 17.Google Scholar Two years later Baker, (Albert N'yanza, 360–2)Google Scholar visited apparently comparable saltworks further south-west along this lake shore at a place he recorded as ‘Vacovia’.
59 Methods of salt-making vary according to the nature and concentration of the sources. Most require some solar evaporation or exploit brines, soils or crusts unspoilt by recent rain. Generally therefore it is an industry of the dry season, that also being of course the time when labour for both the preparation and the transporting of the salt is freed from agriculture. Fuel for boiling is another laborious factor. A description of the existing salt industry at Kibiro is provided by Connah, G., Kamuhangire, E. and Piper, A., ‘Salt production at Kibiro’, Azania, XXV (1990), 27–39.CrossRefGoogle Scholar
60 By Hiernaux, J. and Maquet, Emma in the 1950s: L'âge du fer à Kibiro (Tervuren, 1967)Google Scholar; and by Graham Connah more recently as part of the British Institute's interlacustrine research project: see preliminary reports in Nyame Akuma, xxxi (1989), 46–54, and xxxiv (1990), 38–45Google Scholar, and in Annual Report of the British Institute in Eastern Africa for 1989–90 (London, 1990)Google Scholar; also Connah, , ‘The salt of Bunyoro’, Antiquity, 65 (1991), 479–94.CrossRefGoogle Scholar Further south, at the Uvinza brine-springs on the periphery of the interlacustrine zone, salt-working has been attested back to the Early Iron Age: Sutton, J. E. G. and Roberts, A. D., ‘Uvinza and its salt industry’, Azania, III (1968), 45–86.CrossRefGoogle Scholar The limited radiocarbon tests suggest that it may not stretch back to the first phase of the EIA, but only to the middle of the first millennium a.d.
61 As indicated, not all aspects of change at either ‘revolutionary’ period were necessarily sudden; some facets of the first may perhaps overlap with certain of the second. But it becomes essential to recognize a middle period which was both preceded and followed by radical changes. Until now there has been, for lack of an adequate chronology, a tendency to roll the changes into a single process of ‘northern’ influences or to posit a combination of internal and external factors around the fourteenth or fifteenth century: cf. Steinhart, E. I., ‘Herders and farmers: the tributary mode of production in Western Uganda’, in Crummey, D. and Stewart, C. C. (eds.), Modes of Production in Africa (Beverley Hills, 1981), 114–55, esp. 118.Google Scholar Moreover, the longer history of specialized cattle-herding in this region now attested, together with the suggestion of gradual polarization between the herding and cultivating wings of the communities in the early centuries of this millennium, should reduce the force of the argument based on contrasting lactase levels (see Cook, G. C., ‘Lactase deficiency: a probable ethnological marker in East Africa’, Man, n.s. IV , 265–7)CrossRefGoogle Scholar for imagining a separate geographical and ‘racial’ origin for the herders (Hima and Tutsi). The length of the middle period, or rather the beginning of later Iron Age developments no less than 1000 years ago, is emphasized by Oliver, Roland, ‘The Nilotic contribution to Bantu Africa’, J. Afr. Hist., XXIII (1982), 433–42.CrossRefGoogle Scholar