«The engineering of the prehistoric megalithic temples in Malta A. Torpiano Faculty for the Built Environment, University of Malta, Malta ABSTRACT: ...»
The engineering of the prehistoric megalithic temples in Malta
Faculty for the Built Environment, University of Malta, Malta
ABSTRACT: The prehistoric megalithic structures of Malta and Gozo date back to a civilization of 4500 to 5500 years ago. Although now in ruins, their longevity is remarkable, and must
be due to the inherent durability of limestone, properly selected, as well as to the underlying engineering principles and construction. Prehistoric civilizations are often, erroneously, perceived as technologically primitive, however these prehistoric structures are technologically remarkable. This paper proposes engineering principles underlying its longevity. Hitherto, many of the features of the extant structure have been explained as having “decorative” functions. The author suggests that these features should be assessed in the light of the possible engineering and constructional processes adopted; this approach is based on the belief that, particularly for civilizations in which energy resources were stretched, it would not make “resource” sense, if these features were not there for a specific purpose – it would not be, using modern terminology, sustainable.
1. INTRODUCTION The megalithic temples of Malta have been dated back to the 4th millenium BC. The oldest of the major temple sites, Ggantija in Gozo, has been dated, on the basis of recalibrated radiocarbon dating, to 3600 to 3000 BC, the “youngest”, Tarxien, to 3000 to 2500 BC. This is therefore looking a civilization that lasted at least 1000 years. It is still not clear where it came from and where it disappeared to, and why.
The temple sites understandably underwent a number of changes over the 3500 – 2500 BC millenium. The Lower Temple in Mnajdra, for example, belongs to the older Ggantija phase, whilst the Middle Temple belongs to the Tarxien phase. It is therefore not surprising to find a significant difference in the characteristics of the stonework, in the quality of the workmanship, and in the decorative details. On the contrary, it is indeed surprising to observe that the structural principles, on which the temples are based, apparently remain so consistent over this period.
These principles also appear to be surprisingly sophisticated.
To our knowledge the Maltese prehistoric megalithic temples are the oldest expressions of free-standing stone architecture. This extraordinary statement is made even more amazing not only by the fact that more than thirty prehistoric sites have been uncovered in the barely 320 sq.km. that make up the Maltese Archipelago, but also that the structures with similar characteristics, found in other neighbouring Mediterranean countries, such as the Balearics, the Iberian Peninsula, or Sardegna, generally have a much younger pedigree, and, in any case, do not have the whole range of characteristics of the Maltese megalithic structures.
Malta’s prehistoric temples are a series of megalithic structures that were “discovered” in the 19th century, although some of the major sites were recorded by travellers at least as far back as the 17th century. The 19th century and early 20th century saw the first systematic archaeological campaigns to uncover more of these structures. The early campaigns were probably not as scientific as they should have been, and included some degree of reconstruction, and also the removal of some earth mounds that, in hind sight, might have been central to understanding the way these temples were built. In order to understand the structural behaviour, and especially the construction methods, adopted by the prehistoric builders, it is obviously necessary to, first of all, translate, in our minds, the extant ruins into hypothetical complete structures. It is also necessary to avoid being misled by what may have been mistaken reconstruction carried out since the 19th century, but also, perhaps, before this “modern” period of interest in the monuments.
Plate 1. Plan view of Model of Ggantija Temples, showing coalescing of two complexes In his letter of the 19th November 1840, J.
G.Vance13 describes the excavations carried out, at Hagar Qim, at the request of the Governor General Sir Henry Bouverie, and records how “the remains of ancient architecture” were uncovered, and how “a number of well-proportioned blocks…., scattered in different directions…., some lying in heaps, others singly” were found at the site. Although these descriptions are quite detailed, it is not easy to relate the megaliths illustrated in this letter to the way the configuration of the massive stones in this temple complex would be read today.
In subsequent years, the scattered blocks were lifted back in place, or what was presumed to be their original place. A degree of reconstruction and re-interpretation, often without detailed records, took place in many of the important temple sites, and these may lead to mistaken hypotheses about the original structures. In Mnajdra, for example, it is known that the front section of the Middle Temple was extensively re-modelled in the 1930’s, as was the rear perimeter wall.
Some photographs of the excavation, of both the Hagar Qim and Mnajdra sites, record that the land surrounding the temple structures was much higher than is now the case. The photographs also record the removal of an earth mound at the back of Mnajdra – how much of this mound was modern, and how much prehistoric is nowhere recorded.
2. SHAPE AND MODELS The temple structures have a number of consistent characteristics. The megalithic structures are assembled from large stone blocks, so-called megaliths, apparently without the use of any bedding mortar. The stability of these stone structures therefore depends on the structural form, as well as on the mutual interlocking of the rough surfaces of adjacent blocks. The constructional features include the use of (i) upright free-standing block assemblies, that form “trilithon” portals along the main axis of the temples, (ii) other upright blocks that are used to define the semi-circular apses, which are arranged sequentially along the axis, and (iii) horizontal megaliths, laid in “courses”, forming horizontal arches that corbel out, one course above the other, to form vault-, or dome-, like structures. These inner walls are surrounded by external walls with specific constructional features, such as the alternation of tangential and transverse orientation of the megaliths to form an interlocking outer ring. The space between the inner walls and the outer wall is, generally, filled in.
The basic typology for a temple unit consists of a longitudinal axis, normally also an axis of symmetry, terminating in an apse, and along which is a varying sequence of pairs of lateral apses, or lobes. The external shape of one temple unit is thus broadly ogival, merging into a concave façade. The temple complexes, particularly the four major ones, however, consist of more than one temple unit, and the basic configuration is modified as one temple unit follows another in time, and coalesces with the previous one to form one single complex.
The temple plan typologies range from the three lobes of Mgarr, or Mnajdra (Upper Temple), through the five lobes of Ggantija or Mnajdra (Lower and Middle Temple), to the seven lobes of Tarxien. The temples have obviously been modified over the centuries, sometimes extensively.
The odd apse, in these configurations, is the one at the end of the axes of the temples – which is more or less developed depending on the site. It is reasonable to presume that the sequence of pairs of apses, along a linear axis, owes at least as much to constructional and structural requirements, as to the requirements of ritual.
One of the most fascinating aspects of the study of the prehistoric megalithic temples is that amongst the archaeological material excavated are found what must be considered as amongst the oldest architectural models, now preserved in the National Museum of Archaeology. The larger of these models refers to what appears to be a typical pair of apses, which pair forms the basic unit in the construction of the temples. It also illustrates what must have been the external appearance of the structures, and perhaps gives a hint of the original roofing system. In the Mnajdra Middle Temple, there is what must then be considered as one of the earliest architectural drawings, or more properly, engravings – the elevation of a typical temple, having similar characteristics as exhibited on the small models. It is likely that both the models and the wall engraving are post-construction representations of the temple structures, possibly with some votive meaning; however, it would be attractive, albeit implausible, to consider these as instructions to the prehistoric builders!
Plate 2. From left.
(i) Engraving of façade, (ii) Votive model, (iii) Façade reconstructed from fragments
3. STRUCTURE It is known that the structural stability, and, no less important, the constructional feasibility, of the stone roof of a single, circular, cell, such as the girna or the nuraghe, is based on the stability of the complete horizontal compression ring of stone. A simple description of the corbelled stone chamber could be a series of stone rings, one on top of the other, each ring having a diameter smaller than the previous one. Every complete ring is stable by virtue of its resistance to the compression, (that is, the reduction in the length of the circumference), which would be necessary for the ring to fall through a space that has a smaller diameter. The individual components of the incomplete ring, however, must either be independently stable, or else they have to be supported by some “falsework” until the ring is completed. The individual components of the ring could be independently stable, if, for example, each stone is corbelled off a lower stone, with the projecting part not being too large compared to the part resting on the stone.
Each complete ring of stone, resting on a previous, and larger, ring of stone is inherently stable, because in order for it to collapse under uniform loading, which can only happen by falling inwards, it requires a reduction of its circumference. In other words, if the individual components of the ring are in contact with each other, the stone ring will develop a horizontal compression force to resist such collapse. The stability of the ring therefore depends on the contact between the vertical faces of these “voussoirs”, and hence on the shaping of these voussoirs to a wedge shape, so as to achieve full contact. If such a contact were absent, the stability would then depend only on the individual stability of each corbel. A vertical slice through a corbelled structure would be stable only if the structure were closed at the top – that is, only if the slice formed a complete vertical arch.
However, the dome has two mechanisms by which loads can be carried. The first mechanism is that of the horizontal ring, as discussed earlier, and the other mechanism is that of the vertical arch that exists in any vertical section through the dome. The two mechanisms co-exist, and the load is distributed between these two mechanisms in accordance with the relative stiffness of the two – and depending, obviously on the geometrical configuration of the dome. In the so-called “false” dome, the stone rings have horizontal interfaces, and therefore the frictional resistance, vital for vertical arching action, may be diminished (although it is not absent). In the “true” dome, with the voussoir joints cut normal to the curved surface, the vertical arching action may be more significant, although, in many structures this is also impaired by, say, uneven settlement under the dome perimeter. Arching action, in fact, also requires rigid abutments. Furthermore, when the dome has an occulus, that is, an opening at its crown, the vertical arch is incomplete, and therefore the important load-carrying mechanism is the horizontal, or circumferential, ring, at least in the regions around the occulus, whether the dome is termed “true” or “false”.
The structural mechanics of the single cell funerary chamber have been extensively studied.
Cavanagh and Laxton3 have published an analysis of the Mycenean Tholos Tomb, which it must be remembered, is at least a thousand years younger than the Maltese prehistoric temples. Cavanagh and Laxton discuss what they identify as the three mechanisms that are available for the stability of the tholos tomb. In addition to the vertical arch action, and to the horizontal ring action, they propose that the system of corbelling is, by itself, also possible, provided the corbelling is taken high enough to bridge the span of the structure. This is only feasible in the context of an overburden that balances the over-turning moment that results as the corbelled wall goes higher and higher. Although, in their paper, Cavanagh and Laxton quote the observations of the original students of the tholos tombs, Cockerell and Donaldson, as saying that “in its horizontal position at least, the arch was clearly understood by the architect who designed these chambers”, they discount the importance of horizontal ring action in the Mycenean tholos tomb, primarily because of the major interruption represented by the entrance shaft that pierces the circular shape.
Although this approach ignores the mutual interaction of the different mechanisms, the interruption to the horizontal stone ring, presented by the “entrance shaft”, is a very relevant issue to the discussion of the structural system of the prehistoric temples in Malta. The basic structural unit seems to be that formed by a pair of apses. The horizontal circular compression ring is clearly interrupted by the temple axis, an axis that presumably was made necessary by ritual requirements of entry into the spaces created by the stone structures. The particular system of joining pairs of apses, adopted in the local megalithic structures, consists of the “trilithon” portals that not only mark the axis of the temples, but connect one half of a horizontal circular ring, in one apse, to the other half of the horizontal circular ring, in the opposite apse. In other words, the “trilithon” portals form the structural continuity necessary for a modified form of “dome action”, which includes both corbelling mechanisms, as well as horizontal compression actions. It is also very likely that any roof structure, now missing, would have contributed a further mechanism of load transfer through a modified vertical arch action.