say which came first. Another  source of inspiration lies in pre-temple domestic architecture, such as the oval huts found at Skorba. The deposit in one of these contained terracotta figurines and goat skulls, suggesting that it may have served as a shrine of sorts. However, no intermediary stage in the development has yet been found.

Temples are most often found in pairs and occasionally in groups of three or even four (see the plan of Mnajdra, below). These smaller complexes, in their turn, tend to be found in clusters as the map (right) shows. Colin Renfrew believes that they reflect the territory of particular groups, seeking a sense of self-identity through shared beliefs and rituals— perhaps tribes or clans.

Satellite view of Malta & Gozo showing the location of Neoliathic temples

Tarxien. Model Temple

Mnajdra. Sketch of a Temple Facade

The entrance to the temple was framed by a pair of orthostats capped by a massive stone lintel, to form what is known as a trilithon. A good example survives at the site of Ta Hagrat (right). Inside there was often a small courtyard area, laid with paving stones, which may have been open to the elements. The court was flanked by a pair of D-shaped chambers or apses containing altars and other cult installations. As time went on the number of apses was increased from a single pair to two or even three pairs. The terminal chamber was originally the main focus of ritual activity but finally the builders decided to do away with it and replace it with an elaborate niche. Its function was effectively taken over by the inner pair of apses.

The interior wall faces were either smooth Globigerina or Coralline orthostats with a coat of plaster. They were often painted with red ochre, traces of which have been found at a number of sites. Here too, the upper part of

Ta Hagrat. Entrance

Ta Hagrat. Model Temple

torba—a layer of crushed Globigerina on a rubble base, which was then repeatedly wetted down and pounded until it was hard enough to polish.

The problem of how the temples were roofed is still unresolved. A model found at Ta’ Hagrat would seem to indicate that long stone slabs were laid across the building. The scale of the model suggests they were of massive size and the evidence from sites like Mnajdra and Hagar Qim shows that the builders used a technique known as corbelling (arranging the blocks in the upper part of the wall so that each course slightly overlaps the one below) to reduce the amount of space that needed to be covered. Unfortunately, no such slabs or even broken pieces of them have survived at any of the excavated sites. Of course they may well have been removed by later farmers and broken up for building material but it is odd that the destruction would have been so

Most temples were provided with interior furnishings, generally interpreted as altars, which were usually placed at the rear of the apse. Many of these are simple blocks of stone, often decorated with pitting or reliefs in vegetal or animal motifs. There are also trilithon altars similar to the doorways but (normally) on a smaller scale. Some are as tall as 2.5 metres, however, which is an awkward height for offerings and sacrifices. Double-decker altars have also been found, a trilithon below and an upper slab supported by a central pillar. Finally, there are carved pedestals (some elaborately so) standing about a metre high that support shallow basins. At Hagar Qim, one of the apses contains a round-topped cylindrical pillar about a metre and a half high. In later times, these were known as baetyls and were found all over the Mediterranean world in Classical times. A square sectioned version stands in its own little shrine, built into the outer wall of the temple. Baetyls were used to represent the deity when a true depiction would have been wrong for whatever reason, and that certainly looks like the case here.

Hagar Qim. Altar

The stone blocks, some of which weighed several tonnes, were probably moved by means of sledges and the so-called ‘cart tracks’ found inthe south of the island (left) may well have been for their rails. It has been suggested that the Globigerina spheres found at Tarxien were used to move blocks of stone, but they would have been far too soft to support their weight over any distance. The blocks were apparently manoeuvred into their final position using levers—many of the uprights have notches at their base where the pole would have been inserted. Ramps of earth and rubble were most likely used to raise the lintels.

The trefoil temple is essentially a more formally designed version of the prototype. Subsequently, it was made more elaborate by the addition of an outer pair of apses— the 5-apse temple. The simplest explanation for the change is that more interior space was needed to deal with changes in the cult activity that went on there. In a number of instances, there is evidence for a doorway or screen to separate the outer apses from the interior, dividing the temple into a relatively public area and one that was probably restricted to priests or the like. In time the innermost chamber was reduced to a vestigial niche and altar, reducing the number of apses to four. Finally, a unique building with three pairs of apses was built at Tarxien.

The archaeological evidence supports Evans’ broad sequence. At Ggantija, the four apse North Temple is demonstrably later than the five apse South Temple since the outer wall of the latter was rebuilt to accommodate it. Similarly, at Tarxien the six apse temple was clearly inserted between the two earlier four apse temples, which were drastically modified to accommodate it. It is now seems clear that, rather than a steady and gradual development, there were a couple of fairly brief episodes of temple building—the Ggantija phase (around 3500 BC) and the Tarxien period some 500 years later. At each site the evidence indicates clearly that the temples were built successively, but that the construction of one did not mean the abandonment or destruction of its predecessor—although the nature of its function may have changed.