• 1 The term “vernacular histories” will be discussed in Section 2.

1Deep history models on the peopling of Central Africa generally pay little attention to fishing communities whose activities along the many rivers of Central Africa have played a crucial historical role. The BANTURIVERS project intends to fill this gap. To do so in a meaningful way, we will need, as is usual in sub-Saharan Africa, to combine the strength of several disciplines: anthropology, linguistics, history and archaeology. In this paper, the emphasis is on vernacular histories of past migrations, which we recorded recently, and on those documented in colonial archives.1 We contextualize these histories with the available archaeological information on the wider region and compare them with information on ancestry, itineraries and contact as can be inferred from language data. We will focus here on a stretch of the Congo and Lualaba River where communities with different cultural backgrounds and origins reside next to each other at what appears to be an important cultural crossroads between western and eastern Central Africa.

• 2 A. Moeller, 1936.
• 3 For instance, J. Omasombo Tshonda, 2020.
• 4 J. Vansina, 1990, p. 353, n36.
• 5 M. McMaster, 1988; J. Vansina, 1990.

2This story starts at the Boyoma Falls, near Kisangani, the official end of the Lualaba River, where Wagenya fishermen place large fish traps in the rapids. Heading upstream along the Lualaba—or rather, driving along the Route Nationale 3, since the cataracts make navigation impossible—one crosses Komo villages, while Wagenya and, from Batikalela onwards, Mokpa villages border the river. The last cataracts being located near Ubundu, one can follow the river upstream by boat again and now travel alongside Lengola and Metoko villages. Today, we also find Lokele fishing camps along the Lualaba, but apart from the Lokele, all these communities were present by 1877, when Stanley arrived in the area. When, whence, and how they arrived in precolonial times is known only in the version of vernacular histories that were collected, reinterpreted and reformulated by colonial administrators during the colonial period. Based on these accounts, Governor Moeller presented a grand analysis of Bantu expansions in the Orientale Province.2 It is still the most comprehensive historical work on the region, and the recent monographs on the new provinces rely heavily on it.3 Vansina warns, though, that Moeller’s work must be understood in the context of colonial rule, with its specific goals and ways of thinking.4 Yet, thus far, hardly any alternative or supplementary insights have been offered by sources other than vernacular histories. The exception consists of insights from historical comparative linguistics in the works by McMaster on the Uele region and Vansina, but the information concerning the region of interest here—namely, the lower Lualaba—remains scant, and, considering the advances in Bantu historical comparative linguistics, in need of revision (see sections 3 and 9.4).5

• 6 D. Kopa wa Kopa, B. Ricquier, accepted. The article on classification is currently being written.
• 7 This article does not incorporate information on migration published elsewhere in ethnographic mono (...)

3In the present article, we offer a critical assessment of vernacular histories. We trace original colonial documents later analyzed by Moeller, which are available in the AIMO archives (“Affaires Indigènes et Main d’Oeuvre”), and we recorded new vernacular histories with Mokpa, Metoko, Lengola and Komo informants in 2019 and 2020. We also include Mbole histories from the archives, since, although they are not fishermen, they are frequently cited by the other communities. We focus here on the itineraries proposed in these stories, as well as on claims of temporalities, cultural contacts and identities. Also, new results from historical comparative linguistics place these vernacular histories in a different perspective. The linguistic details will be demonstrated in separate articles.6 Here, we discuss solely the migrations that can be inferred from language classification and language contact as compared to the routes extracted from vernacular histories.7 The aim of the present paper is to untangle the information embedded in these histories in order to outline the main elements of chronological and cultural dynamics in the area and to investigate whether past social interactions described in these complex narratives can be related to the rise of the languages spoken in the area today.

• 8 D.L. Schoenbrun, 2021; A. Ekblom et al., 2017.

4Inspired by Schoenbrun’s use of the term, and encouraged by the article of Ekblom et al., we choose to refer to the histories related by members of the concerned community as “vernacular histories”.8 We feel that this term best covers the nature of the stories. The label “oral traditions” seems inappropriate, considering that they were not recorded in official circumstances or in the form of recitation. They are not solely “oral” either, since some are inspired also by written documents or written down by a member of the community, thus excluding the term “oral histories”. “Vernacular” here is a synonym of “native” or “indigenous” and implies that it concerns histories of a community as narrated by members of that same community. It does not imply that these histories are narrated in the vernacular language, which they were not. Our recent vernacular histories were recorded in Lingala, and we can suspect that also in colonial times the histories were offered to the colonial administrators in a vehicular language.

• 9 J. Thornton, 2004.
• 10 D.L. Schoenbrun, 2017.
• 11 See N. Arazi, G. Senga, 2021.

6A limitation of vernacular histories is that they are set in the present and might be more informative of the political and cultural tendencies of that present than revealing of the past. Indeed, Thornton discusses the example of Kongo origin traditions, documented since 1588 and changing through time in accordance with political evolutions.9 Another issue, highlighted by Schoenbrun, is the influence of written sources.10 A library may not be present in the villages, but it happens that written sources are available locally to members of the community and have become integrated in the newly produced narrative. As an example, during interviews made in Kasongo about the Arabo-Swahili period, an informant gave a very structured and clear narrative about the history of the “Wazula”. Asked about the possibility that some of this history might have been written in the past, he concurred and produced an original colonial document synthesizing the history of the “Wazula”.11 During an interview with a Komo informant, a written source was referred to as well.

• 12 D.L. Schoenbrun, 2017.
• 13 Results from sites further afield, along the Congo River and its tributaries west of Kisangani and (...)

7How are we to deal with vernacular histories then? Schoenbrun offers some practical guidelines.12 Ideally, different versions, both oral and written accounts, are compared and placed in a broader context. The latter may be drawn from other bodies of historical evidence, such as the insights delivered by archaeology, historical comparative linguistics and other disciplines. In the case of the lower Lualaba, we can compare recent interviews with the accounts offered by colonial administrators. Having investigated vernacular histories of different people independently, we can compare the narratives of different groups, with the aim to build what archaeologists would call a regional cultural chronostratigraphy. How are the different histories connected, and do they allow for a chronological sequence of major events? Thanks to a new classification of the relevant languages, we can also reach out to results from historical comparative linguistics. Archaeology should provide valuable additional information for the very ancient or more recent precolonial period.13

• 14 See J.-P. Sanderson, 2010.
• 15 The administrative reforms culminated in 1933. “Territories” and “secteurs” underwent only minor ch (...)

9The archival documents used here are all part of the “Affaires Indigènes et Main d’Oeuvre” (AIMO) archives available for consultation at the Royal Museum for Central-Africa in Tervuren, Belgium. More specifically, all the documents are catalogued in the AIMO “ethnographic” part of the archives. The AIMO was an administrative department of the “Gouvernement Général” of the Belgian Colony, and the documents were written by territorial agents, such as territorial administrators or local governors, and conserved in situ.14 In the case of the documents used to collect the oral histories of the time, they were written by colonial agents with the intention to put people in boxes and link one particular population with one particular territory.15 The migration routes described in the process were not written down for academic purposes, nor in ideal conditions. Indeed, biases were induced, as the people who collected these oral histories were part of a coercive occupation regime—furthermore, the successor to a regime (the Congo Free State under Leopold II of Belgium) well known for the violence of its practices. Nonetheless, even if they do not constitute direct historical data and need to be interpreted with caution, the documents still constitute historical sources by themselves and are filled with information about past oral histories that would otherwise be lost. The documents used in this paper are not homogenous. Indeed, the people who wrote them did it in some cases directly from field interviews, but in others they based their reports on previous documents written by other administrators.

• 16 M. McMaster, 1988.

11The communities concerned in the present paper—namely, the Mokpa, Wagenya, Metoko, Komo, Lengola and Mbole—all speak Bantu languages, as do most of their neighbors. In the wider region—namely, north of Kisangani, and northeast and east of the Lualaba—we also find other languages—namely, Ubangi and Central Sudanic. In her dissertation, McMaster explored these contacts in the Uele region and concerning Boan Bantu speech communities.16 Based on language classification and glottochronology, McMaster argues that the first Boan Bantu speech communities arrived in northeastern Congo around the mid-first millennium BC. At that time, they would already have met with Central Sudanic communities. Ubangi communities would have arrived only in the first centuries AD. Important for the region under study here is McMaster’s conclusion on the Bira branch of Boan Bantu, to which Komo belongs. The ancestor speech community of Komo and its immediate relatives would have moved to the south around the turn of the era.

• 17 J. Vansina, 1990.
• 18 C. Ehret, 1998.

12Also based on historical comparative linguistics, but considering other Bantu branches of relevance here, Vansina offers some later dates.17 “West-Bantu” speech communities, more precisely Boan and “Soan” (the group we call Upper-Congo here), would have arrived in the wider area around Kisangani around 1 AD. Nyanga people, living to the east of the Lualaba and speaking a language from a different Bantu branch, would have arrived from the Great Lakes by 200 AD. Finally, farmers, whose culture would already have been a mixture of West- and East-Bantu traditions, came from the southern savannas from 330 AD onwards. Vansina writes that “people tended to cluster along the major waterways” and that the area showed low population density in contrast to the savannah or mountain areas. Following a different Bantu classification, Ehret distinguishes the Proto-Enya from the Proto-Lega community, the first being “West-Bantu” and coming from the west, the latter East-Bantu and arriving from the southwest—namely, the Kasaï area.18 We will see in both the vernacular histories and the new linguistic evidence that this view requires revision. Ehret suggests the early first millennium BC for their migrations. Ehret also notes that there were “cultural interactions of modest importance” with peoples from the Great Lakes, but that these Proto-Lega communities otherwise had a quite distinct history from “Mashariki Bantu” (i.e. East-Bantu) communities.

• 19 I.e. at the time of the first submission of this paper. In the meantime, archaeological fieldwork h (...)
• 20 A. Livingstone Smith et al., 2017; N. Arazi et al., 2020; L. Champion et al., 2017.
• 21 A. Livingstone Smith et al., 2017; D. Seidensticker, 2021; H.-P. Wotzka, 1995.
• 22 Cf. H.-P. Wotzka, 1995; I. Matonda, K. Bostoen, 2022; P. Coutros et al., 2022; I. Matonda Sakala et (...)
• 23 About this long story, see for example M. Posnansky, 1961; R. Oliver, 1966; M. Eggert, 1981; J. Van (...)
• 24 See for example B. Clist, 1989, p. 80; M. Eggert, 1993, p. 322-323.

13At present, archaeological data are inexistent for the region between Kisangani and Lowa.19 The only points of comparison are research carried out to the northwest, between Kisangani and Bumba, and, to a lesser extent, to the south, in the Kindu and Kasongo areas.20 The chrono-cultural sequence from these sites distinguishes roughly three main chronological phases. The early phase, dated from 300 BC to 0 AD, is characterized by the presence of ovoid vessels with flat bottoms and everted necks, decorated with comb impressions, dragged comb impressions, rocking comb or blade impressions.21 At a continental scale, the early phase shows links with the earliest ceramic traditions identified in Central Africa. Stylistic parallels can be found around Mbandaka, Kinshasa, in the Kwilu and down to the opening of the Congo River and the Atlantic Coast, but also on the coast of Gabon, in the southern tip of the Central African Republic and Cameroon.22 This pottery distribution pattern coincides with the core area of Bantu languages, and thus this early pottery phase has often been linked to migrations from early Bantu speech communities.23 While there is no elegant proof of this yet, it is a widely shared and discussed assumption.24 Keeping this idea as a rough working hypothesis, this would mean that the first Bantu speech communities arrived in the eastern Congo rainforest around 300 BC, a date which is not too distant from the mentioned linguistic estimates.

• 25 A. Livingstone Smith et al., 2017; D. Seidensticker, 2021; H.-P. Wotzka, 1995.

14The middle phase is dated to around 200 to 300 AD in the same area between Kisangani and Bumba. It is characterized by pottery styles clearly related to those of the early phase, with a predilection for flat-bottomed ovoid vessels with everted necks, decorated with stylus and comb or rocking blade impressions. During the middle phase, pottery styles are rather homogenous between Bumba and the area of Isangi on the Congo River, but they display a strong regionalization in shape and decoration at the continental level. However, here as elsewhere in Central Africa, they appear as a continuation of pioneer traditions.25

• 26 L. Champion et al., 2017; A. Livingstone Smith et al., 2017.
• 27 R. Oslisly, 1992, 1994; H.-P. Wotzka, 2006.
• 28 D. Seidensticker et al., 2021; G. de Saulieu et al., 2021; B. Clist et al., 2021.
• 29 See for example P. Nlend, 2014, p. 50.

15The late phase is poorly dated and covers the period from 800 AD to the present. It is characterized by a radical break in pottery traditions, with the appearance of round-bottomed vessels with short everted or horizontal necks, decorated with roulettes and matt impressions—although no detailed determinations have been made yet.26 These traditions also mark the appearance of the pounding on a concave mould technique (“martelage” in French), a pottery roughing-out technique that can still be observed between Bumba and Kisangani among contemporary potters. The late phase also finds echoes at the continental level. For the period between the 8th and the 11th centuries AD, there are very few archaeological data available for much of western Central Africa.27 The causes of this hiatus in radiocarbon dates are debated.28 One will note, however, that this is the period during which the large states of the area—the Kongo, Kuba or Lunda polities, to name but a few—are in the making. While regionalization is still marked, we also see common stylistic features, such as a general shift from flat-bottomed to round-bottomed vessels and the appearance of the “peigne gouge”, a kind of stylus, in pottery decoration.29

• 30 This absence from the archives and colonial maps still has political implications. The Mokpa do not (...)
• 31 J. Yakusu, 1948.

23The Mokpa are nearly absent from the archives.30 Only Yakusu mentions the “Binamokba”.31 He notes that they originated in the “Mobiatoko” valley and then split into four or five groups. One group stayed near the Ubundu Falls (Omanga), a second in Binamonbo, and a third in Wanie-Rukula. The other groups are not further mentioned. This does not mean that the Mokpa are a mysterious and unknown community, simply that they received another, more confusing, ethnonym: “Baleka”. The ethnonym “Wagenya” is not straightforward either.

• 32 A. Moeller, 1922.
• 33 H. Marmitte, 1934. According to one informant, muléka refers to a spear that is used for fishing (s (...)
• 34 A. Moeller, 1932, 1936.

24“Wagenya” and “Mokpa” are present-day ethnolinguistic identities. “Baleka” is the ethnonym most commonly cited in the archives. In its most strict sense, it refers to the riverine people from the Ubundu region, Bamanga rapids, and the Wanie-Rukula region.32 This corresponds to the area inhabited today by people identifying as “Mokpa” but named “Baleka” by others. Marmitte indicates a more general use of the term Baleka, covering people without any affiliation, rather referring to “people of the water”.33 This use of the term is still common today in the region. The term would be of Lengola origin. Indeed, the term “Baleka” is also applied to certain Lengola factions: “Walengola adapted to life on the river”.34

• 35 A. Moeller, 1922; H. Marmitte, 1933.

25The ethnonym “Wagenya” covers a wide range of people from various backgrounds. Colonial administrators point to the Arabo-Swahili for giving all “indigenous adapted to life on the water” the nickname “Wagenya”.35 Moeller gives the example of the Wagenya of Kasongo-Kibombo, Maniema Province, who would include people of all origins—for instance, Wazimba. The Wagenya of the Kisangani region would rather be of “Baleka” origin. People also self-identify as “Wagenya”. Marmitte notes:

• 37 F.H. De Bock, 1922; A. Moeller, 1922.

26We find information on the Baleka and Wagenya in two documents of 1922: notes by administrator De Bock, based on field inquiries but also including notes of other administrators; and Moeller’s annotations to the first document.37 De Bock mentions no more than that the Wagenya of the Falls come from the spring of the Maiko, the Wagenya from Ponthierville/Ubundu from the spring of the Lilo, and those from Lowa from the spring of the Lowa. De Bock adds a temporal reference: the Wagenya would still be able to show the trees planted by their fathers along the Maiko River. Moeller adds more information: the Wagenya would come from the Kambaula River, upstream along the Maiko. They arrived recently at the Falls, from where they chased Lokele people, the only survivors being the group Milambo. Moeller does not discuss the origins of the Mokpa/Baleka, merely that the people in the region around Ubundu, at the Bamanga Falls and in Wanie-Rukula, are identified and/or self-identify as “Baleka”. He notes that the Baleka of Wanie-Rukula are also called “Warega”. Moeller indicates that his observations are preliminary, merely to orient future research.

• 38 H. Marmitte, 1933.
• 39 H. Marmitte, 1934.

27In 1933, the term “Baleka” remained obscure. According to Marmitte, Van Belle considered the Baleka of Ubundu to be affiliated to the Baleka-Mituku.38 Marmitte himself rather thinks that, even though they might have followed the same migration itinerary as the Baleka-Mituku, they are in fact part of the Lengola, adding that they have nothing in common with the Bambole. Marmitte also mentions the Wagenya of Ubundu and Lowa but merely repeats the observations of De Bock. Another relevant group discussed by Marmitte are the Bamanga-Bangiri of Ubundu. Marmitte believes that they come from the Uele, settled on the right bank of the Lualaba near Wanie-Rukula and crossed to the left bank chased by the Komo but under the protection of the “Baleka”. They would have adopted certain traditions of the mentioned Baleka but would speak a different language. Marmitte cites the territory administrator Soors, who classified the Bamanga-Bangiri as part of the Wagenya. In 1934, Marmitte cites Lauwers, who notes that some Bamanga, from Ponthierville (Ubundu), claim to come from Micici, a Lega town on the Elila River, and speak the same language as the Metoko.39 Considering the presence of a village called Bamanga inhabited by Mokpa people today and the fact that the Mokpa and Metoko languages are related, the mentioned Bamanga may have been “Baleka”/Mokpa.

• 40 J. Yakusu, 1948.

28A most interesting document is the two-page account of Jean Yakusu, himself Mugenya.40 In contrast to the mentioned colonial documents, which are interpretations of vernacular history, this account can be considered as “raw data”—namely, vernacular history as narrated by a specific informant. This narrative situates the origin of the Wagenya at the Obiatuku River, from where they would have migrated towards the Lualaba/Congo, together with their “neighbors”—namely, the Binamokba, Banyametoko and Balengola. Upon arrival at the Falls, seven communities were already present: the Isungu/Isumu, Yasanga, Bafaka, Bayembu, Bakango, Batibo and Baliele. The Isungu would have offered a girl as wife to Loa-Moongo, ancestor of the Wagenya. One of their sons would have confronted his maternal uncles, resulting in a conflict. The Wagenya wanted to occupy part of the islands in the rapids. When the Isungu refused, the Wagenya took the islands by force and chased away the Isungu. The next day, the war continued, and also the Bafaka, Bayembu, Batibo and Baliele were chased away. The Wagenya thus became the masters of the region of Stanleyville (Kisangani). In the words of Yakusu: “une petite tribu de rien (dont les descendants de Loa-Moongo) toute puissante battit les plus grandes qu’eux”. According to Yakusu, they were present before the Komo and others now living in the area. The Wagenya only tolerated the Batiakaiya, a community still living along the Tshopo River in 1948. The fugitives then lived with the Bambole of Opala and along the Congo River. Some of them had become Lokele. Yakusu furthermore mentions two Wagenya groups which he designates as “clans” of Lengola origins: the Banakulu and Bakonga or Binalombe.

• 41 On these sociopolitical dynamics within the migrations, see Kopytoff, 1987; J. Vansina, 1990; de Ma (...)
• 42 A. Moeller, 1936.

31The Mokpa stories are mostly informative on cultural affiliations and ancient migration routes (see Map 1 in the Maps section below for an overview). The group in question is presented as being related and carrying the same “culture”: same habits, customs, language, way of life, circumcision, etc. The different ethnonyms designate the common ancestor and his descendants, all founders of family groups, probably clans or fractions of clans.41 The Mokpa would originate from an ancestral Lega or Leka community, the other descendants being the Barega, Bazimba, Mituku, Wagenya, Basongola and Banyanga. According to the informants in Bamanga, the ancestral “Lega” homeland would have been situated in the Semliki mountains, located east of Kisangani, at the border with Uganda. This corresponds to the view put forward by Moeller.42 The Mokpa informants do not mention, however, how the ancestors reached the Lualaba from this mountainous homeland. Instead, they suggest that the Mokpa, Wagenya, Metoko and Songola came from the south, one informant explicitly mentioning a route along the Lualaba, the other implicitly when indicating that the Wazimba stayed there in the forest and that the Songola stopped “halfway”. The itinerary drawn by the informants at Batikamondji is slightly different. They situate the Lega homeland in Shabunda and name the Elila as the way followed to reach the Lualaba. Interestingly, this is in line with Lauwers’ notes that “Bamanga” people would come from Micici, a Lega town at the Elila River.

• 43 A. Moeller, 1922.
• 44 J. Yakusu, 1948.

33The Wagenya account of Yakusu does not include claims on cultural affiliations on a higher node, nor does it trace the migration routes further back than the Obiatuku valley. Interesting, however, is that the Mokpa, Metoko and Lengola also followed the Obiatuku downstream to their current location, as in the recent Batikamondji history. The Wagenya story is rather informative about pre-existing communities. They were apparently many. Yakusu mentions seven communities. Moeller also mentions that the Wagenya chased away other communities, “Lokele” in his words.43 According to the Batikamondji informants, they were “Mbole”. Yakusu and the Batikamondji informants agree that the Bambole of Opala are descendants of the communities chased from the Falls. Yakusu furthermore speaks of intermarriage with the Isungu and Lengola clans who became integrated in the Wagenya community.44 This gives us an insight into previous interethnic dynamics. The Batikamondji informants also mention intermarriage with the Lengola.

• 48 H. Marmitte, 1933; 1934.
• 49 See J. Omasombo Tshonda., 2020.

37We discuss here two archival documents written by Marmitte, “Administrateur Territorial”.48 The first, dated 20 February 1933, “Lowa”, is written upon instructions of Governor Moeller and Libois, “Commissaire de District”. Marmitte indicates that his notes are a combination of information found in the “political” archives, which he considers insufficient, and field data that are scant because of his recent arrival in the territory. This first document covers the origins and migrations of the communities living in the Territory “Walengola-Wasongola-Mituku”.49 The second document is dated 26 January 1934 and was written after Marmitte left for Shabunda, where he became administrator of the Warega Territory following the territorial reorganization of October 1933. This document is a complement to earlier studies by the territory administrator Van Belle and rather concerns the social and political organization of the Metoko. It is based on field trips by Marmitte among the “Mituku Bamoya” and “Mituku Basikate”, already conducted before postage of the first document. Both documents are mostly interpretations, except for one specific story that is paraphrased.

• 50 H. Marmitte, 1934.

38The Metoko are commonly referred to as “Baleka-Mituku” in the archives (cf. supra). The name “Metoko”, also written as “Mituku”, refers to the ripe palm fruit of a species of raffia palm, ketoko, which is abundant in their environment. According to Marmitte, their original ethnonym is “Baleka” or “B[a]rega”, but they were called “Mituku” by the Arabo-Swahili after the abundant palm trees of the region.50 Marmitte translates Baleka as “gens d’eau”, hence “people of water”, as opposed to Bahuse or Babira, who are people of the forest.

• 51 H. Marmitte, 1933.
• 52 Some Gombe families (coming from the west, from the region of the Boyela/Moma) would be allied to t (...)

39Based on the claim that they suffered from Komo invasions and considering the homeland of the Komo proposed by De Calonne, Marmitte deduces that the origins of the “Baleka-Mituku” as well as the Walengola must have been near the sources of the Lindi, Maiko and Lubutu rivers.51 Marmitte adds that this is a “simple hypothesis”. It is, in any event, not proposed by the Baleka-Mituku themselves. According to Marmitte, the last settlement recalled by the “Baleka-Mituku” is on the left bank of the Lualaba, downstream of the Kasuku River, near the source of the Lilo. This land was called Tubila. Marmitte goes on, reciting clan movements that are not always ethnically homogenous:52

• 53 H. Marmitte, 1934.

40In both documents, Marmitte writes that, based on the available information, it cannot be clarified whether the Metoko preceded the Lengola in the region. One story, however, clearly states that at least at the Lilo, the Lengola were the first inhabitants: Upon arrival at the source of the Lilo, the Metoko encountered Lengola people. Following a conflict—namely, Lengola wives having been seduced by two Metoko men, upon which one of the latter was killed by the envious husbands—the Metoko fought and chased away the Lengola.53 Marmitte notes that contacts with the Lengola and the Bakusu resulted in loanwords into the language “Kituku” and even goes as far as stating that linguistic criteria no longer allow for ethnic identification. According to Marmitte, the Metoko claim that the species of raffia palm with which they are identified was already present in the region when they arrived, especially along the Lilo River.

• 54 J. Yakusu, 1948.

42The Wagenya story of Jean Yakusu, cited in Section 4, states that the “Banyametoko” of Lowa rather have their origins in the Obiatuku valley, indicating a joint dispersal with the Mokpa and Wagenya.54

• 55 Lega people also divide their dialects along these administrative borders (BANTURIVERS linguistic f (...)

43The recent interview is broadly in line with the information presented by Marmitte in the 1930s. The ethnonym, for instance, is agreed to be a reference to the abundant raffia palm trees in the region called ketoko, first applied by the Arabo-Swahili (Marmitte) or by “white people” (recent interview). The original ethnonym is said to be Baleka/Balega/Barega. In the archives, Marmitte translates this term as the generic “people of water”, but it cannot be excluded that the name implies instead a link with the Lega people of the Maniema and Kivu provinces. Indeed, the recent Metoko informants situate their homeland in Lega land. They cite Shabunda, Mwenga and Pangi, the three territories where you can find Lega people today (see Map 2 in the Maps section below for an overview of the migration routes).55 In 1933, Lauwers was told by Metoko that they were affiliated to the Lega and came from Micici.

• 56 J. Omasombo Tshonda, 2020, p. 290, 292.
• 57 H. Marmitte, 1934, p. 1.
• 58 H. Marmitte, 1934, p. 7.

45In the interview, the informants mention a divide between the Mituku-Basikate and Mituku-Bamoya. This divide corresponds to two “secteurs” of the Ubundu Territory. The villages founded by the sons of Bamoya and Basikate, as cited by the informants, equal the “groupements” of the respective territories.56 This follows the socio-political dynamics of migratory progression—namely, the settlement of sons (cadets) in villages distinct from that of their father. The listing of villages in Marmitte’s documents does not follow this divide. Still, Marmitte mentions that he visited the Mituku-Bamoya in 1932 and the Basikate in 1933.57 The administrative divide was likely based on ethnic realities of the time. Also interesting in this respect is Marmitte’s observation that two young men were nominated as head of the mentioned secteurs by the colonial administration and that all contemporary clan chiefs were seeking to prove their “droit d’aînesse” to claim this position.58

• 59 H. Marmitte, 1934, p. 1.

46According to Marmitte, it is difficult to ascertain whether the Lengola preceded the Metoko in the region. He knows of only one story in which the Metoko Baniankonge chased away the Walengola Waniaboyombo from the source of the Lilo.59 In the interview, the informants suggest an earlier presence of the Lengola and Komo, indicating that the Metoko were “guests”.

• 63 F.H. de Bock, 1922.

51Rather than establishing a precise route, Hackars tells us about the pressure from the “Arabized” and the reshuffling by the colonial territorial administration. This reshuffling is also indirectly mentioned by De Bock.63 In addition, Hackars points to a limitation of surveys conducted by administrators who are clearly part of a system of domination: because of their position, they can make the interviewees say whatever they want:

• 65 F.H. de Bock, 1922.
• 66 A. Moeller, 1922.
• 67 Fivé, 1932.
• 68 The Lumbuya River is rather an affluent of the Tshopo River. It is unclear how this geographical er (...)

52Nonetheless, most colonial officials seem to agree on placing the origin of the Komo in the northeast. De Bock mentions that they claim to come from Beni.65 De Bock mentions a recent migration some twelve years past of “elephant hunters” that already reached the Lokilo region against the colonial administration’s will. Moeller suggests that the Komo come from the source of the Maiko and might be related to the Bira of Ituri.66 Moeller adds that the “Babela” followed the course of the Maiko, whereas the Komo followed the ridges (lignes de faîte). The same story is told by the Komo of Madula amongst whom Fivé performed field enquiries.67 According to Mabianga, they are named Komo because they live at the ridges. Their original name would be Babia, and the contemporary Babia or Babira, living “downstream” would be their brothers, sharing the same ancestor Mbia. Fivé furthermore writes that the Babula of the Madula, a Komo group living between the Tshopo and the Congo River, come from the Lubuya, which he considers an affluent of the Maiko River.68 The Babula are part of the “Bakumu ya Lubuya”, a group of various clans that followed a similar route from Lubuya, continuing on the Maiko and then downstream along the Lualaba. The cause of the migration would have been a fight between the Bambundji and the Bakilinga, “three good generations ago”. They arrived in the following order and regrouped once in Kisangani as all of them came from the Lubuya: the Babangana, the Babama – the Babula – the Batuanaleka – the Bayangana – the Bakuba – the Batiamondje.

• 69 F. Stradiot, 1931.
• 70 A. Moeller, 1922.

53Stradiot provides a very detailed report on the Komo, discussing their traditions and lineages.69 Concerning their origins, Stradiot writes that the Komo recall the Ituri. If the river is meant, the Komo homeland could have been situated somewhere in a large stretch northwest of the Ruwenzori. Stradiot describes three different migration routes. The Komo from the west followed the (Congo?) river upstream; those from the east followed the Lindi, Mandaie and Loya rivers; those from the center went inland to the south and then crossed the Lindi and Tshopo. The impulse of the migrations was a push from better-armed populations called “Nkutu” and “Banienginiengi”. The migrations were not hampered since the land would have been empty except from the not-so-numerous “pygmy” peoples. Stradiot also mentions that some Komo recall a large salt lake and the presence of white-skinned and long-bearded people. Stradiot links the Komo to the Bali and Bira of Ituri. Finally, Stradiot provides another etymology for the name ‘Komo’. It would refer to the bark of a specific tree, nkumu, which they wear as cloth. This interpretation is also mentioned by Moeller.70

• 71 Fivé 1932; A. Moeller, 1922.

54The archival documents seem to agree that the Komo came from the mountains in the east or northeast (see Map 3 in the Maps section below for an overview). The rivers Lindi and Maiko are frequently cited, sometimes also the Tshopo River, whose bed is in between the two. It is not always clear whether these rivers were followed or crossed, but considering the sequence of citations, we could infer either a northeast-to-southwest route in the case of an Ituri origin, crossing the mentioned rivers. Coming from the Rwenzori mountains, a route from the source of the Lindi towards the source of the Maiko, hence an east-to-west route, could be proposed. The recent vernacular histories cite the same rivers—namely, Lindi, Tshopo and Maiko rivers—but position an origin in the west or even near the Ubangi, as learned in a written source. There appears, however, to be a contradiction in the recent narrative. The name of the ancestor Nkumu Esoko means ‘mountain’. Even though a metaphorical interpretation is offered, this may have had a more literal meaning in the past, thus ultimately pointing to an eastern origin—namely, the mountains of the Great Lakes. The same etymology is also mentioned by Komo informants interviewed by Fivé, who mention that they did not live along the rivers but on the ridges, which confirms Moeller’s notes of 1922.71 If there was at all a movement from the west, it may have been secondary. Then again, the recent vernacular history speaks of a split that occurred at the Maiko, one group heading northwest to the Kisangani area, the other south, to Maniema.

• 72 In informal exchanges during separate missions of fieldwork for the BANTURIVERS Project, also other (...)

55The preference of the Komo not to live along the rivers is also confirmed in other vernacular histories—for instance, by Lengola and Mokpa informants. Both the Komo interview and the archives stress the identity of the Komo as hunters, more specifically “elephant hunters”. In a Mokpa interview, the Komo community is described as having the skills to find food in the forest and resembling hunter-gatherer communities. 72

• 73 Hackars, 1919.
• 74 F.H. de Bock, 1922.

57The first migrations occurred in an unspecified precolonial period. In the recent interview, it is claimed that the Komo were the first occupants of the land around Kisangani, on the right bank. It is not clear whether the Mbole, living on the left bank, preceded them or not. In any event, the Wagenya arrived later and so did Stanley. The archives tell us that more population movements were caused by the arrival of the Arabo-Swahili and also commanded by the colonial administration.73 De Bock indeed mentions recent movements of Komo elephant hunters to Lokilo, dated around 1910.74 The Komo expansion thus seemed to be still on-going in the early colonial period, much to the dislike of the colonial administration.

• 75 A. Van Belle, 1932.
• 76 H. Marmitte, 1934.
• 77 Anonymous, n.d.
• 78 P. Wautier, 1932.

59If the distinction between the mentioned communities appears neat in this recent interview, it is again a complex picture in the archives. The Walengola do not seem to form an ethnically (nor politically) homogenous group. Van Belle reports that the Lengola from the left bank (the Babinde in this case) “claim not to know the Lengola of the right bank”, but they do admit to be “tribal brothers”.75 In 1934, Marmitte writes that the Babira are in fact Lengola of the Ubundu-Kirundu region.76 In the same document, Marmitte mentions an opposition between on the one hand the Babira and Lengola of the Lowa who are Bahuse or “forest people”, and on the other hand the Baleka-Mituku who are “water people”. Furthermore, a geographical split between Lengola groups would have led to a name change—the Lengola from Kilima Pondo to Kisangani being called Balila, as opposed to the Lengola of Ubundu that have kept the ethnonym.77 Wautier, in turn, notes that the Bokuma that are close to the Yamba (Bambole) would also be a portion of the Lengola that separated from the rest of the group.78

• 79 Anonymous, n.d.

60An anonymous report posterior to 1931, states that the Bokuma come from Biondo, near the Ruiki.79

• 80 A. Van Belle, 1932.

61Various origins and migrations routes are depicted by colonial administrators. According to Van Belle, Lengola informants claim to come from the northeast towards Ubundu as a consequence of Komo expansions.80 The Lengola followed the right bank of the river upstream and were gathered between Kilindi and the Lowa River mouth. Not finding iron at the riverbanks, they would have followed the Lowa River upstream towards the east until the mouth of the Lubutu. There they encountered other Komo coming from the east and were forced back to the Lualaba.

• 81 H. Marmitte, 1933.
• 82 Anonymous, n.d.
• 83 H. Marmitte, 1934.
• 84 H. Marmitte, 1933. Marmitte furthermore offers details on clan migrations, which illustrates the co (...)

63As mentioned above, Marmitte deduces from de Calonne’s location of the Komo homeland in the Semiliki, and from the fact that the Baleka-Mituku and Walengola recall Komo invasions, that the homeland of the latter must have been situated at the sources of the Lindi, Maiko and/or Lubutu.81 This is not based on documented vernacular histories but concerns a hypothesis. Marmitte notes that the “natives” do not remember their homeland but merely the last settlements that are geographically close to their current location. For the Lengola, these last settlements are situated at the source of the Obiatuku and on the riverbanks of the Lowa. Contradicting the first hypothesis—namely, that Komo invasions caused Lengola migrations—Marmitte attributes this two-fold split to the presence of Komo people in the area. Some Lengola would have fought the Komo; others sought a way around them. This split is repeated in an anonymous document posterior to 1931 but attributes it to the arrival of the Komo.82 At the source of the Lilo, on the left bank of the Lualaba, the Lengola would have been chased away by the Metoko.83 Concerning encounters, Marmitte also writes that some Komo, coming from the Maiko, became allied to the Lengola upon colonization.84

• 85 See for example I. Kopytoff, 1987, concerning the crucial ideological dimension of the “first occup (...)

65The geographical information enclosed in the recent vernacular history locates the homeland in the Bafwasende Territory, more specifically Uma, Bafwasende and Opienge (see Map 4 in the Maps section below for an overview). From there, the Lengola would have followed the Obiatuku River until the Congo River. Next, they continued upstream as far as the rapids and then followed the Ruiki. This is on the left bank of the Congo River. After that, they continued in the forest. The Lengola would have been the first occupants of that part of the forest as well as of the stretch of river upstream of Ubundu, a recurrent theme in this type of vernacular history.85 Also, in the archives, the homeland is located in the northeast. Several origins are mentioned, but all are situated northeast and not too far away.

• 86 A. Van Belle, 1932.
• 87 H. Marmitte, 1933.

66The archival documents seem to disagree on this status of first occupants. According to Marmitte, the Lengola preceded the Komo as well as the Baleka-Metoko who chased them from the Lilo River on the left bank of the Lualaba. Van Belle, on the other hand, mentions the presence of the Metoko preceding, or at least at the same time as, the Lengola and suggests further joint dispersals.86 Other sources, such as Yakusu (cf. supra) and the narrative produced by the informants in Batikamondji, also consider that the Lengola migrated together with the Metoko, Wagenya and Mokpa. Marmitte furthermore presents confusing claims on the presence of the Komo pre- or postdating the Lengola on the right bank of the Lualaba.87

• 88 Anonymous, n.d.; H. Marmitte, 1931.
• 89 A. Moeller, 1922; J. Yakusu, 1948. Today, the word “lokele” refers to any spoon. The freshwater mus (...)

69The colonial administrators again had difficulties identifying the “Bambole”. The archives mention both the Bambole and the Ba M’bole, which would be different groups sharing a very close name.88 Various hypotheses are put forward on the etymology of the word: M’Bole could come from Bantu na m’bole, “les gens du bas”, or it could be a nickname given by the “arabisés” due to the fact that these populations consumed a lot of onions called m’bole. In 1915, Van de Capelle notes that the Topoke and Lokele declare that together they form the M’Bole. The Lokele are the people from the river and the Topoke the people from inland. Around the same period, Moeller writes that “Lokele” would be a nickname based on the local name of a species of freshwater mussels and is also used to designate people from various origins—for instance, the Yakusu and Yatumbu who, according to Yakusu, were driven out of Kisangani by the Wagenya and would have been of “Mbole” origin (cf. supra).89 This is a fine example of the general picture of this region, where identities are more shaped by clan (shifting) alliances than by strict ethnic criteria. The term Mbole thus may refer to a larger group, including Lokele and Topoke peoples, as well as to a more restricted group, currently located in the Opala area and between the Lomami and Congo River.

• 90 P. Wautier, 1932; H. Marmitte, 1931; Van de Capelle, 1915.
• 91 F.H. de Bock, 1922.

70The various reports present in the AIMO archives situate the origins of the Mbole in a zone roughly around Basoko and the Aruwimi (see also Map 5 in the Maps section below for an overview). Both Wautier and Marmitte situate it in the north of Basoko—in a region between the lower Aruwimi and Yahila region according to Marmitte, and near Bokombo in the north of the Lese-Itimbiri according to Wautier; Van de Capelle in a region to the east of Basoko, on the Aruwimi, in a zone between Banalia, Mongandjo and Yangambi; and Wautier, speaking about the Tooli clan, to the south of Basoko near Lukutu.90 Only De Bock locates the origins of the Mbole further to the northwest on the right banks of the Congo River, “downstream of the Itimbiri, from the region of Budja, between the Molua River and Mokua River”.91 De Bock, nonetheless, also mentions another possible origin—namely, that the Bambole might come from “the Aruwimi woods”. Despite these differences, the general idea is that they come from the north/northwest of the area they occupied at the time.

• 92 Van de Capelle, 1915.
• 93 H. Marmitte, 1931.
• 94 H. Marmitte, 1931; F.H. de Bock, 1922; Van de Capelle, 1915.
• 95 Van de Capelle, 1915.
• 96 H. Marmitte, 1931; Anonymous, n.d.; P. Wautier, 1932.

71The Mbole “ethnic group” is composed of numerous clans that followed various migration routes. In a number of cases, they migrated with clans belonging to other ethnic groups such as the Lokele and settled in zones unoccupied by other competing ethnic groups, mainly in the Lomami region.92 In the words of Marmitte (own translation): “The Bambole declare that when they arrived on their present land, they encountered no other peoples than the very few BOTCHWA (“pygmies”) from the Tshuapa valley.”93 The Mbole migrations took place in a context where alliances with other groups were unstable and conflicts frequent, including among themselves.94 Those conflicts where at the root of their migrations—for example, with the Azande, Basoo and Turumbu, according to Van de Capelle, and the “arabisés”, according to De Bock, in the Basoko region. This took place in later stages of their migrations as well—for example, with the Lokele on the Loha River,95 and with the Lengola in the Lobaye and the Ubundu region.96

• 97 H. Marmitte, 1931.

72Based on a genealogical reconstruction, Marmitte situates the start of the Mbole migration in the 18th century, and a second phase at the beginning of the 19th century according in this case to elders (Likaka and Afakuniuki) that were still alive at the time.97

• 98 H. Marmitte, 1931.

83Little is mentioned on pre-existing communities, except for the region around the Falls, for which we get much detail in Yakusu and in documents from colonial administrators. These communities are identified as Mbole or Lokele, both said to be part of a larger “M’Bole” group. Many were chased away and became integrated in other groups, such as the same Mbole and Lokele, which complicates tracing their past. For all we know, these communities may have had different origins, and the claim of a Lokele or Mbole identity may be due to their later integration into the mentioned communities. Apart from the Falls area, the sources claim that the region was “empty”, perhaps with the exception of hunter-gatherer communities. This is mentioned for the Komo, for instance, but considering the large area inhabited by the Komo this may in fact apply to regions far east of the Lualaba. The Mbole are said to have encountered hunter-gatherer communities, but according to Marmitte these would have come from the south, from the Tshuapa River.98 In summary, if the Falls had already been popular for a longer period, the riverbanks of the Lualaba are claimed to have been scarcely or even not inhabited before the arrival of the present-day communities in the late precolonial period. That of course is a perspective based on vernacular histories, and we can imagine a number of reasons for why people and events prior to the arrival of the M’bole are not remembered. To dive deeper into history, we will need to apply other methodological tools.

• 99 A. Livingstone Smith et al., 2017. This article was written before the start of archaeological fiel (...)
• 100 L. Champion et al., 2017.
• 101 See A. Haour, 2018, for a successful connection between memory and archaeology.

84To obtain an absolute dating of this comparative chronology and follow the track of history beyond memory, the BANTURIVERS team needs the input of archaeological data, especially for the period before 1875. As mentioned in Section 3, for now we can make comparisons only to archaeological data from the area between Kisangani and Bumba. The relevant chronocultural sequence distinguishes three main chronological phases.99 The early phase is dated to 300 BC – 0 AD, probably concerning the earliest migrations into the wider region. The middle phase is dated around 200 – 300 AD and provides a cultural continuation of the previous period. The late phase covers the last 1200 years and displays significant cultural changes, with evidence of strong northern influences. Based on the vernacular histories, we would be inclined to conclude that the migrations cited in the present article occurred during this late archaeological phase, but only new archaeological excavations can confirm this hypothesis. The late phase is characterized by abrupt changes in pottery styles and technologies—namely, the appearance of the pounding on a concave mold roughing-out technique and the appearance of roulette decorations. We know from surveys undertaken south of our study area that this change also occurred in the southern stretches of the Lualaba.100 However, at this stage, we do not know how many distinct pottery styles were in use during the late period, so we lack the subtle distinction or the spatial and temporal analysis of co-existent pottery styles that would allow a correlation with vernacular histories. As is often the case, the correlation of data from vernacular histories with archaeology is a difficult endeavor. Archaeologists in the BANTURIVERS team will make this connection by working closely with village elders in determining the location of excavations.101

• 102 The topics presented in 9.2 and 9.3 merit further research. The present paper being focussed on mig (...)

89Another claim we can read in the vernacular histories is one linked to the use of resources. Every community self-identifies with specific subsistence strategies and/or the dichotomy water vs. forest. This identification is a co-construction, neighboring communities confirming the subsistence-related identity of the other. The Mokpa and Wagenya, for instance, identify as “people of the water”, having developed special fishing techniques and/or being skilled in navigation and even combat on water. This is confirmed by the Lengola, who admit having needed the help of the “Balega” to cross the water and having learned their river-related skills when integrating in their society. The Komo self-identify as hunters, and this is confirmed when a Mokpa informant compares their knowledge of the forest to that of hunter-gatherers. The hunting identity of the Komo is also linked to a self-identification as warriors, which may again question the claim of first occupancy. The Metoko and Lengola informants also self-identify as people of the forest, describing the difficulties in crossing the Lualaba River. However, unlike the Komo, they do not claim to be skillful hunters. In the case of the Metoko, for instance, the exploitation of raffia palms is generously mentioned, not least because of the link with the ethnonym.102

• 103 A. Moeller, 1932, 1936.
• 104 Linguistics is a frequently consulted tool for historical research concerning the Bantuphone commun (...)

90Moeller’s outline of the region’s early history also incorporates insights from language comparison.103 If Moeller’s conclusions were largely based on intuitions of similarities between the concerned languages, we now have evidence based on statistical methods—namely, phylogenetics.104 The BANTURIVERS project is currently working on a new classification of the Bantu languages of the northeastern DRC and on contact between Mokpa and the other languages spoken on the lower Lualaba. The resulting insights into linguistic affiliations, dispersal routes and contact scenarios provide interesting comparative material with the information retrieved from vernacular histories (compare Map 6 in the Maps section below with the other maps).

• 106 According to the classification in R. Grollemund et al., 2015.

91A first observation is that the migration routes cited in the vernacular histories correlate particularly well with the dispersal routes inferred from the classification of the concerned languages. The new classification confirms the affiliation between Enya, Mokpa and Metoko, with the latter being slightly more distant. Enya and Mokpa are closely related, confirming a recent split. This indicates that they probably reached the region between Kisangani and Lowa jointly and differentiated upon arrival. The trio forms a subgroup of the Lega group, the other branches including Songola-Zimba, the Lega varieties and Nyanga. The closest relatives of the Lega group are Zambian languages such as Bemba, spoken in the southeast, and Great Lakes languages spoken east and northeast of this area, and on a higher node also Luban languages spoken south of the rainforest.106 Taking the current location of the daughter languages into account, it can be assumed that coming from the savannas south of the rainforest, where we still find the Luban languages, the remaining three groups diverged near the western shores of Lake Tanganyika. This is where the three groups meet today, with Taabwa as Zambian language, Bembe and Buyu as Lega, and the Great Lakes languages just north and across the lake. The communities speaking Lega languages then entered the rainforest and settled in the area around Shabunda. In the most recent classification, the subsequent divergence separates the northern from the southern Lega languages, with Mokpa, Enya and Metoko rather related to Nyanga and Kwame. Considering the geographic location of the latter, the Lowa River forms a plausible route westward, as mentioned in several vernacular histories. This would also still imply travelling through Lengola land, at least if they occupied roughly the same area at that time as they do today. A northern route, meaning that the Lega communities would come from the Semliki mountains near the Ugandan border, is not corroborated by linguistic evidence.

• 107 Y. Bastin et al., 1999.

93Lengola proves to be a special case. As already suggested by Bastin and colleagues and confirmed in the new classification, Lengola is most closely linked to Bodo and Nyali, two languages spoken in the northeast, in fact just north of the languages affiliated to Komo.107 The claim as recorded recently that the Lengola would originate from the region Bafwasende – Opienge is thus plausible from a linguistic perspective. The new linguistic results position Lengola, Bodo and Nyali as a separate branch of Central-Western, not as part of Boan or Upper-Congo (see below). From this classification, we can infer that the claim that Komo and Lengola are “brothers” would rather point to specific clanship bonds than shared ancestors.

• 108 D. Kopa wa Kopa, B. Ricquier, accepted.

95A second correlation can be found in the history of contacts. Linguistic evidence suggests the presence of pre-existing communities. The three languages Enya, Mokpa and Metoko have labial-velars as an inherent part of their phoneme inventory.108 These sounds, [k͡p] and [g͡b], are common in northern Bantu languages but absent in Lega and other East-Bantu languages. The omnipresence of these phonemes throughout their lexicon, even in basic words such as ‘to die’, ‘to hear’ and ‘dog’, indicates intense contacts, perhaps even a language substrate. All other Bantu languages of the area contain labial-velars—hence all could be candidates. However, only Lokele and an Olombo variety called “Mbole” have an equally wide use of these phonemes. Interestingly, the latter variety is spoken near one of the locations where the Yasanga fugitives would have settled, who would have been speakers of the language “Mbole”. Considering that Metoko also contains labial-velars, contacts with these “Mbole” people must have occurred before the divergence of the three languages. Consequently, based on linguistic evidence, the presence of a community speaking an Upper-Congo language such as Olombo or Mbole along the Congo River between Kisangani and Lowa can be assumed, and it preceded the arrival of the Wagenya-Mokpa-Metoko.

• 109 A. Moeller, 1936.

97Moeller presented an important historical work on the migrations of the Bantu peoples in his “Province Orientale”.109 His interpretations are based on oral histories collected by colonial administrators. However, these histories were rarely presented literally. Usually, they are echoed only in the conclusions of the administrators concerned. Moeller combined this information with similarities or discrepancies in traditions, political structures, languages and ethnonyms, and he accumulated the information to arrive at “grandes lignes”, broad strokes of migrations. He thus links the migrations of the Wagenya, “Baleka” and “Baleka-Mituku” to that of the Warega and Bembe. The itineraries presented on his map in fact resemble the routes drawn by linguistic evidence. However, the northeastern origin, in the Semliki mountains, preceding a homeland on the Ulindi is not confirmed by historical comparative linguistics. One of the reasons to propose this homeland near the Ugandan border is the presence of people called “Balegga”. The claim of a northeastern origin cannot be found in the archival documents consulted or in the vernacular histories recorded with Mokpa informants at Batikamondji and Metoko informants in Ubundu. It is striking, however, that a Semliki homeland is proposed by recent informants in Bamanga. This may simply indicate that they have access to theories such as Moeller’s. Combining the information in the original archival documents, the recent vernacular histories (except one) and the linguistic evidence, we are rather inclined to reject this hypothesis.

• 110 This has been well demonstrated, for example by W. de Mahieu, 1975, for the Komo, and by C. Gray, 2 (...)

101Considering that clanship and ethnic affiliations were very dynamic, as we can read in the archives and as mentioned in the recent interviews, whose history is referred to in vernacular and language histories? How should we interpret the link between these two types of evidence? A straightforward answer is that both types of evidence are linked to the most recent cultural layers identifying today as ethnolinguistic communities, meaning that they currently correlate language to cultural affiliation. Both the linguistic and historical evidence allow us to uncover one previous layer. If, in the archives, the memory of these predecessors is still alive and included in detail in the accounts, the memory became opaque in the recently recorded interviews, nearly a hundred years later. Today, generic names are mentioned, such as “Mbole”. The linguistic evidence may indeed corroborate a link with a community speaking Mbole or an affiliated Upper-Congo language. However, the predecessor communities mentioned in the archives were not necessarily ethnolinguistic communities. The names may refer to clans or other types of segmentary affiliation, for whom a linguistic identity was secondary, easily exchanging one language for the other. In the Wagenya account, it is also mentioned that one Wagenya clan is of Lengola origins, and a Lengola informant mentions that the Lengola at one point adopted the “Lega” language and acquired their customs. In the past, elements such as language or certain rituals may have been more fluid, and there may have been a greater importance of clans. Consequently, history should probably be told in terms of clan history.110 Unfortunately, this type of detail is missing from the vernacular histories we recorded, and it cannot be retrieved from the linguistic evidence.