Rock-cut stratigraphy: sequencing the Lalibela churches.

Fauvelle-Aymar, Francois-Xavier ; Bruxelles, Laurent ; Mensan, Romain 等

[ILLUSTRATION OMITTED]

Introduction

The site of Lalibela on the northern plateau of Ethiopia has gained worldwide fame as a historical riddle and a tourist attraction. This complex of a dozen rock-hewn churches (Figures 1 & 2) was occasionally mentioned (under the names of Warwar or Dabra goha) in Ethiopian sources written in geez (Bosc-Tiesse 2009; Derat 2009). But these sources point to legendary origins and provide little information useful to the historian or archaeologist.

[FIGURE 1 OMITTED]

[FIGURE 2 OMITTED]

The site is named for King Lalibela (d. after 1225) of the Zagwe dynasty (c. eleventh to thirteenth centuries), to whom later written sources (from the fifteenth century) ascribe the foundation often churches (Derat 2006). Little is known about King Lalibela, although one of the rare relevant contemporary documents suggests that one church--Medhane Alem--was cut from the rock during his reign (EMML 6907, fol. 208v). As for the rest of the site, there is no compelling evidence confirming that it was really built in the thirteenth century.

Francisco Alvares (a member of a Portuguese embassy to Ethiopian King Lebna Dengel) visited Lalibela in the early sixteenth century (Alvares 1540 [1943]), providing an early record of the organisation, topography and state of preservation of the site. But it is not until the Italian occupation, and again from the 1960s onwards, that the site, now more easily reachable from the capital Addis Ababa, attracted the attention of amateurs, scholars in architecture (Monti della Corte 1940; Bianchi Barriviera 1963, 1966) and art history (Lepage 1997; Gervers 2003a), liturgists (Fritsch 2008) and archaeologists (Phillipson 2009). A programme of preservation and tourism management was also implemented under the aegis of UNESCO, the World Monuments Fund, and the Ethiopian authorities (the site of Lalibela was placed on the World Heritage list in 1978).

Interest has intensified in the last decade or so, mainly devoted to the iconography on the walls of the churches (Lepage 1999, 2006), the typology of the monuments' architecture (e.g. Lepage 1997; Gervers 2003a; Phillipson 2006) and its liturgical functions (Fritsch & Gervers 2007; Fritsch 2008). But comprehensive archaeological investigation of Lalibela has been inhibited by its obvious complexity, the reluctance of the clergy to allow scholars to generate a corpus of knowledge independent from the religious narrative, and the general persuasion that the nature of the site precludes the recovery of any stratigraphic information. Hence it seemed that the exploration of Lalibela had come to an impasse.

The work reported here, which formed part of a multi-disciplinary project, was aimed at providing a sequence of the site using archaeological methods independent of the art historical evidence, as advocated by Gervers (2003a & b) and Phillipson (2009: 123-81). Since the creation of one church-builder was removed by the chisel of the next (Gervers 2003a: 28), the stratigraphy is hard to read, but it is nonetheless present in the order of cutting, in the dumping of the rock pieces and in subsequent sedimentation. Our investigation did not require entry to the churches (the main concern of the ecclesiastic authorities), but focused on the geological formation of the area, and the way it had been quarried.

The Lalibela complex is traditionally divided into three groups of churches (Figure 2). The first group, located in the northern part of the site, includes five monuments (listed here in the order in which they are usually visited): Medhane Alem, Maryam, Denagel, Masqal and the complex of Debre Sina/Golgota/Sellasse (which comprises three churches). This Northern Group is separated from the Eastern Group by a seasonal stream, the Jordan (Yordanos), which runs in a deep gully that collects water from the entire massif. This gully shows evidence that it is partly man-made. The Eastern Group has five monuments: Gabriel-Rufael, the Betalehem, Marqorewos, Amanuel and Libanos. Another group, to the west, only comprises the church of Giyorgis.

Method

The method comprised a systematic survey and record of all the morphological, functional and stylistic anomalies of the church buildings. We also observed and evaluated the potential of the stratified dumps of spoil from rock-cutting and the processes of sedimentation that had occurred on the site, especially where cutting new spaces in the massif had rendered them liable to new forms of natural impact (see below).

Construction sequence

We use the term 'anomaly' for all kinds of unexpected, irregular or apparently abnormal and useless features found in the architecture of the site. Drawing on the method of 'superimpositions' widely used in rock art studies (see e.g. Russell [2000] for an elaborate case study), our interest in the morphological, functional and stylistic anomalies of Lalibela relies on the idea that each single anomaly preserves a memory of two successive actions made at the same place (though they may be very distant in time). Many examples can be encountered. In the Northern Group, doorways overlooked a void, indicating that a pathway outside the door had been superseded and dug away (Figure 1). In the north wall of Medhane Alem, an opening high up and asymmetric to the wall gives access to earlier chambers (Figure 3). The north-west facade at Gabriel-Rufael (Figure 4) shows a hanging staircase leading to a now removed courtyard, another replacement courtyard served by new openings, and a now 9m-deep courtyard cutting away most of the previous features, with a cistern dug in its base. These are some of the types of small local sequences that can be deduced.

[FIGURE 3 OMITTED]

Spoil heaps

A study based on satellite imagery, checked in the field, indicated that the dumps of debris resulting from the making of the monuments are found almost everywhere on the site, mainly at each of the four sides of every church (Figure 2). Very few previous investigators have evaluated the potential of these spoil heaps (however, see Finneran 2009). Yet the disposal of the rock fragments must have posed a serious problem for the constructors, and from an archaeological point of view, it is very likely that these refuse dumps contain a complete stratification of the site from its beginnings, thus mirroring the phasing of its successive transformations, with broken tools, ceramic sherds and other artefacts preserved in the heaps of stratified stone chippings (Figure 5). Subsequent overbuilding of these archaeological deposits, for example by tombs, may also provide a convenient terminus ante quem for the dating of the monuments.

In at least one case we found evidence that the refuse was not rejected in the vicinity of the monuments under excavation but rather collected and gathered at another disposal site that was used as a 'grey zone' either because of the existence of ruined and useless structures or in order to conceal potentially undesirable monuments. This example lies to the immediate north of the Eastern Group, where we detected a huge dump clearly overlapping ancient ground levels and portions of a large-block masonry wall found in situ (Figure 2). Here we can only suggest that massive, built structures are to be uncovered below this artificial hill, which again points to a very important archaeological potential for the site.

Sedimentation

Lalibela was always prone to the generation of sediments, which would collect naturally in the open-air cuttings and courtyards. Furthermore, cutting ever-deeper monolithic walls naturally increased the process of weathering, sometimes weakening them to the point of collapse. There are several cases on the site where one can observe remnants of now-disappeared rock-cut buildings. This is the case in the courtyard of present day Libanos, which was cleared of its heaps of sediment after about 1970 (compare the picture in Phillipson 2009: 143) revealing a number of features belonging to former buildings (Figure 6).

The widening of the open-air courtyards for the purpose of cutting out new monuments created virtual water tanks liable to collect all the water of the massif. This situation would sometimes lead to unexpected and possibly catastrophic events. This can be seen in the front courtyard of Gabriel-Rufael (Figure 4), where the original ground level was lowered by 9m, and an outflow was bored through the monolithic wall bordering the courtyard. It is likely that the constructions at Lalibela continually modified the geomorphological and hydrological systems, which could both accelerate the process of sedimentation and provoke its removal. One can suggest that these two processes did not always function at the same time. The clearance and redeposition of sediment implies a strong centralised power, able to maintain the monuments against the elements, while the accumulation of sediment suggests times when such power was on the wane.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

The purpose of this article is not to publish the complete catalogue of observations, measurements and anomalies (which will be done elsewhere) but to draw attention to the methodological arguments of our proposed phasing. It should be noted that no attempt will be made here to convert this phasing into a chronology.

A new phasing

The phasing deduced from the stratigraphic examination of the different parts of the site is summarised in Figure 7. The rocky massif in which the Lalibela complex is set consists of basaltic scoriae embedded between two dark flows of basaltic lava, part of the Aiba formation (Asfawossen Asrat et al. 2008a & b). This substratum in which the churches were hewn cannot be described as volcanic tuff (contra most authors, e.g. Phillipson 2009: 124). The church-builders aimed for the scoriae, which they could cut, but avoided or left unhewn the much harder basalt (Figure 8). It is remarkable that all the constructions at Lalibela were dug in the same deposit of basaltic scoriae, to the exclusion of basalt proper. This testifies to an intentional choice probably informed by a good empirical knowledge of the properties of the bedrock.

[FIGURE 6 OMITTED]

This empirical knowledge seems to have been active over a long period of occupation. We are able to trace an early phase of settlement, which may have begun long before the site was transformed into a Christian religious complex. We call this phase 'Troglodytic' (Figure 7). This term is not intended to have any chronological or ethnological connotation. It is characterised by small entrances to narrow tunnels running a few metres under, and generally parallel to, the surface of the rock, leading to little dome-shaped chambers. Small flights of stairs also belong to this phase. Though most of these features have been erased by subsequent transformations of the site, some openings can still be observed hanging above the modern level of circulation (e.g. Figure 3).

In a second phase which we have called 'Hypogean', the inhabitants of the site continued to occupy, or reoccupied, the galleries of the previous phase but transformed some of them into a network of cubic chambers, heightening ceilings, and ornamenting entrances with Aksumite-like pillars and doorways. At the time, these doorways (today hanging over the void) connected with open-air courtyards, newly-excavated passages deeper in the massif, and with peripheral corridors or trenches that were attached to the interior spaces. Associated with this phase are monuments such as Masqal and Denagel, respectively north and south of Maryam, and parts of others, such as the central chamber of Gabriel-Rufael. The east part of the courtyard of Maryam, where an ancient circulation level can still be observed, also belongs to this phase (Figure 9). It may be useful to some readers to add that the designation 'Hypogean' is used here in a completely different sense than Phillipson (2009), who uses it to designate all non-built architectural feature whatever its morphology and function.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

The question remains as to whether these monuments could, at this stage, have already been Christian churches. In the case of Gabriel-Rufael, for instance, the northern room, generally regarded as the sanctuary of the church, was in fact a chapel (Fritsch 2008: 88); it follows that the monument as a whole was never meant to be a church any more than a castle equipped with a chapel is a church. In addition, we agree with Phillipson (2009: 146-8) that since the opening of its windows truncated previous features, this chapel is a rearrangement of a space having previously existed under the very different form of our second (Hypogean) phase of occupation. We thus believe that most Hypogean monuments were not begun as churches, but transformed into churches at a later stage.

The next phase in our sequence is called 'Monumental 1'. It includes the first properly monolithic monuments excavated into the heart of the massif, which were cut out and cored so as to take a form nearer to that of conventional buildings. This sometimes entailed widening and deepening previously-existing open-air courtyards, as in the case of Maryam. In the course of these transformations, some pre-existing Hypogean monuments were preserved as protruding relics because the extension of the excavation would otherwise have broken into subterranean spaces. Other structures had a facade cut and designed on the exterior while being internally transformed into churches. The churches of Maryam and Medhane Alem clearly belong to this Monumental 1 phase, as is also the case with a now-disappeared monumental access (above the 'Tomb of Adam') to this set of churches (Figure 1). It is possible, though not certain, that most of the monuments of the Eastern Group, which are probably part of a fortress complex, belong to this phase.

We call the next phase 'Monumental 2'. It is mainly characterised by a considerable lowering of the outside levels. The aim was apparently to create facades, accesses and windows to newly excavated churches below the ground level of the previous ones (Figure 1). As a result, the monumental access to Maryam found itself truncated and unusable as such. During this phase, not only were new churches probably created (Golgota-Selassie), but also other non-religious monuments were transformed into churches (e.g. Marqorewos), the whole site becoming a Christian religious complex that lost some of its civilian or defensive features. For this reason, it seems that the Monumental 2 phase is marked by an ideological split from the previous phase.

The description of the site given by Alvares at the beginning of the sixteenth century indicates that the site may have been in its Monumental 2 phase at that time. From this point on is what we call the 'Filling-in' phase. This was a time when there was no longer any centralised (religious or political) power able or willing to maintain the site as a ceremonial centre. It is not known what exactly took place in Lalibela between the sixteenth and the mid twentieth centuries, and we cannot rule out the possibility that the site went through one or several periods during which it exerted some sort of strong spiritual attraction, thus reinforcing the sanctity of its churches. But whatever the case may be, the site underwent several centuries of change: sediment accumulated and filled up the trenches, some monuments collapsed and their debris was not cleared, and new settlements were established on the top of the debris dumps and filled-up trenches. This was the situation encountered by Sandro Angelini in the 1960s (Angelini 1967) when he started to clear the trenches and the courtyards.

Discussion

Recent research on Lalibela has followed two different lines of interpretation. The first one, represented by Claude Lepage and Emmanuel Fritsch, considers the whole site (Lepage 2002), or at least the major part of it (Fritsch 2008), to be the result of an architectural program that was undertaken and completed during the reign of King Lalibela, or at the latest during the late Zagwe dynasty. This approach is based on a sequence in architectural development (Fritsch & Gervers 2007; Fritsch 2008), drawn from reference to liturgical sources and the few documents at our disposal such as the History of the patriarchs of Alexandria, and inscriptions carved on a number of wooden altars still kept in the churches. To these written sources Lepage (1999, 2006) adds the interior wall paintings and the outside relief-carving of Maryam, which he confidently ascribes to the same period, while Fritsch (2008) considers the presence of certain architectural features as markers for a mid twelfth- to thirteenth-century attribution.

By contrast, Phillipson (2009:123-82, esp. tables on p. 178 and 180) is inclined to ascribe most Lalibela monuments, which would not all necessarily have been initially conceived as churches, to a long period of time between the seventh/eighth and the early twelfth century. With proper caution, he derives his early chronology from architectural features (mainly monolithic doorway frames imitating wooden lintels and protruding beams) that are reminiscent of typical 'Aksumite' architecture (Buxton & Matthews 1974). He ascribes the latest monument on the site--that of Golgota, the last in Phillipson's sequence--to the time of King Lalibela because local oral tradition locates Lalibela's grave at precisely this place. It is worth noting here (contra Phillipson 2006, 2009) that the presence of an Aksumite style need not imply an early date since Aksumite features could have been incorporated long after the fall of Aksum. The church of Yemrehanna Krestos located a few dozen kilometres from Lalibela is a good example of this, since it is generally ascribed to the twelfth century, in spite of its typical 'Aksumite style' architecture. Gervers (2003a), who also favours a long sequence, places it almost entirely after King Lalibela's time and extends it as far as the fifteenth century, considering that Golgota church is a testimony to the rehabilitation of the Zagwe dynasty under the reign of King Zara Yaeqob in the mid fifteenth century.

The sequence deduced from stratigraphic observation aligns best with the models of Phillipson or Gervers in that it points to several phases of occupation and transformation of the site. However, our sequencing suggests a much longer chronological development culminating in two important monumental phases. Whether the last of these should be assigned to the thirteenth or the fifteenth centuries remains in question.

Conclusion

Most previous researchers attempting to understand Lalibela have followed a typological methodology, classifying types of churches, architectural features or decoration styles. Such an approach has contributed significantly to the field. But it must also be admitted that this approach can be partly circular in the sense that each specialist may unwittingly produce as many idiosyncratic types as required to reflect his/her own field of competences and to fit the expected historical sequence he/she has in mind.

We have shown here that, in spite of the absence of conventional deposits, there is considerable potential for stratigraphic analysis of the site, even without archaeological excavation. This is based on the local sequencing of features, the distribution and layering of soil heaps and the study of sedimentation, including sediments now removed.

While it does not resolve detailed questions of date, our initial reading of the phasing of the site offers a framework of development throughout the life of the site that can be enhanced by future research.

Acknowledgements

This article is based on the results of a multidisciplinary mission carried out in May 2009 under the aegis of the Ethiopian Authority for Research and Conservation of Cultural Heritage (ARCCH, Addis Ababa) and the French Centre for Ethiopian Studies (CFEE, Addis Ababa), with the cooperation of the French Institute for Research in Preventive Archaeology (INRAP, France), the support of the Cultural and Tourism Office and the ecclesiastic authorities of Lalibela. The team was composed of historians (M.-L. Derat, E-X. Fauvelle-Aymar), an art historian (C. Bosc-Tiesse), a liturgist (E. Fritsch), archaeologists (R. Mensan, C. Menard), a geomorphologist (L. Bruxelles, INRAP), a topographer (O. Onezime, INRAP), a cartographer (Loic Di Maria), and experts in heritage (Abebe Mengistu, Fasil Ayew). Brian Clark, Iyassu Demissie, Niall Finneran and Alemu Hayle participated in some aspects of the field mission. This mission was funded by the CFEE, the Department of Social Sciences and Humanities of the Centre National de la Recherche Scientifique (CNRS, France), and by generous support from Ethiopian Airlines. We express our thanks to all these individuals and institutions, as well as to David Phillipson and another anonymous referee, whose comments helped to improve the text, and to Leila Qashu for editing the English.

Received: 4 March 2010; Accepted: 18 May 2010; Revised: 3 June 2010

References

ALVARES, A. 1540 [1943]. Verdadeira informacao das terras do Preste Joao das Indias: pelo Padre Francisco Alvares; prefaciada, anotada e actualizada na grafia por August Reis Machado. Lisbon: Agencia geral do Ultramar.

ANGELINI, A. 1967. Le chiese monolitiche di Lalibela. Atti dell'Ateneo di scienze, lettere ed arti in Bergamo 33: 3-24.

ASFAWOSSEN A., g. AYALLEW & M. DEMISSIE. 2008a.

Geological and geotechnical properties of the medieval rock-hewn churches of Lalibela, northern Ethiopia: implications on their deterioration and conservation. Paper read at the International Conference on Paleoanthropology, Paleontology and Archaeology in Ethiopia, Addis Ababa, ARCCH, 14 January 2008.

ASFAWOSSEN A., M. DEMISSIE & A. MOGESSIE. 2008b. Geotourism in Ethiopia. Archaeological and ancient cities, religious and cultural centres: Yeha, Axum, Wukro and Lalibela. Addis Ababa: Shama Books.

BIANCHI BARRIVIERA, L. 1963. Le chiese in roccia di Lalibela e di altri luoghi del Lasta. Rome: Istituto per l'Oriente.

--1966. Restauri alle chiese di Lalibela. Rassegna di Studi Etiopici 22: 135-46.

BOSC-TIESSE, C. 2009. Gouverner et definir un territoire. Geopolitique, art et production manuscrite au Lasta entre 1667 et 1768. Annales d'Ethiopie 24: 87-148.

BUXTON, D.R. & D. MATTHEWS. 1974. The reconstruction of vanished Aksumite buildings. Rassegna di Studi Etiopici 25: 53-77.

DERAT, M.-L. 2006. The Acts of King Lalibala: structure, literary models and dating elements, in S. Uhlig (ed.) Proceedings of the 15th International Conference of Ethiopian Studies: 561-68. Wiesbaden: Harrassowitz Verlag.

--2009. Du Bugna au Lasta: centre et peripherie dans le royaume d'Ethiopie du XIIIe au XVIe siecle. Annales d'Ethiopie 24: 65-86.

EMML 6907. Ethiopian manuscripts microfilmed for the Ethiopian Manuscript Microfilm Library, Addis Ababa, and for the Hill Manuscript Library, Collegeville, a yet uncatalogued Project Number, National Library, Addis Ababa.

FINNERAN, N. 2009. Settlement archaeology and oral history in Lasta, Ethiopia: some preliminary observations from a landscape study of Lalibela. Azania 44(3): 282-91.

FRITSCH, E. 2008. The churches of Lalibala (Ethiopia) witnesses of liturgical changes. Bollettino della Badia Greca di Grottaferrata 5: 69-112.

FRITSCH, E. & M. GERVERS. 2007. Pastophoria and altars: interaction in Ethiopian liturgy and church architecture. Aethiopica 10:7-51.

GERVERS, M. 2003a. The rehabilitation of the Zague kings and the building of the Dabra Sina--Golgotha--Sellassie complex in Lalibala. Africana Bulletin 51: 23-49.

--2003b. The Dabra Sina--Golgotha--Sellassie complex in Lalibala, in B. Teffera & R. Pankhurst (ed.) Proceedings of the 6th International Conference on the History of Ethiopian Are 388-414. Addis Ababa: Addis Ababa University.

LEPAGE, C. 1997. Une origine possible des eglises monolithiques de l'Ethiopie ancienne. Comptes Rendus des Seances de l'Academie des Inscriptions et Belles-Lettres 141(1): 199-212.

--1999. Les peintures murales de l'eglise Beta Maryam a Lalibala, Ethiopie. Comptes Rendus des Seances de l'Academie des Inscriptions et Belles-Lettres 143(3): 901-966.

--2002. Un metropolite egyptien batisseur a Lalibala (Ethiopie) entre 1205 et 1210. Comptes Rendus des Seances de l'Academie des Inscriptions et Belles-Lettres 146(1): 141-74.

--2006. Un bas-relief royal a Lalibala (Ethiopie) vers 1200. Comptes Rendus de l'Academie des Inscriptions et Belles-Lettres fasc. 1 (January-March): 153-73.

MONTI DELLA CORTE, A.A. 1940. Lalibela. Le chiese ipogee e monolitiche e gli altri monumenti medievali dei Lasta. Rome: Societa italiana arti grafiche.

PHILLIPSON, D.W. 2006. A new sequence, chronology and evaluation of the Lalibela rock-cut churches. Orbis Aethiopicus 10: 138-48.

--2009. Ancient churches of Ethiopia, fourth-fourteenth centuries. New Haven (CT) & London: Yale University Press.

RUSSELL, T. 2000. The application of the Harris matrix to san rock art at Main Cave North, Kwazulu-Natal. The South African Archaeological Bulletin 55 (171): 60-70.

Francois-Xavier Fauvelle-Aymar (1,2), Laurent Bruxelles (1,2), Romain Mensan (1), Claire Bosc-Tiesse (3), Marie-Laure Derat (3) & Emmanuel Fritsch (3)

(1) TRACES-UMR 5608 (UTM-CNRS-INRAP), Maison de la recherche, 5 allee Antonio Machado, 31058 Toulouse, France (Email: Fx.fauvelle@yahoo.fr; Laurent. bruxelles@inrap.fr; mensrom@gmail.com)

(2) School of Geography, Archaeology and Environmental Studies, University of the Witwatersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa French Centre for Ethiopian Studies (CNRS-MAEE), P.O. Box 5554, Addis Ababa, Ethiopia (Email: Cnrs-cfee@ethionet.et; derat@univ-paris1.fr; emmanuelcssp@wanadoo.fr)