89However, the site at Clickhimin in Shetland has an example of this early fort architecture, and also of the new material culture -particularly of the Everted Rim pottery jars with plausible links to Brittany -that appears with it there. The unexcavated island promontory semibroch Dun Baravat in Lewis preserves unique architectural evidence which probably shows why the high hollow wall was originally devised, and the similar cliff-top site Rudh an Dunain in Skye -also unexcavated -is reasonably dated to the Early Iron Age by pottery in a nearby cave. The new dating of the Old Scatness broch, also in Shetland, to the third or fourth century BC appears to confirm that a period of dynamic change occurred in Shetland in the Late Bronze and Early Iron Ages, and that it could have involved foreign contact of some kind. To explain the extraordinary phenomenon of scores if not hundreds of broch towers appearing all over Atlantic Scotland in the Middle Iron Age, from about 200 BC onwards, all the architectural evidence, as well as that of the contemporary material cultures, needs to be considered. expansion and first flowering: middle iron age 2 (C.200 bc to ad 200) Broch architecture (Fig. 1)
Basic featuresThere is no dispute about what a classical broch tower is, only about how many there were, and how standardized. It is a round dry-stone tower averaging about 20 m in overall diameter with a wall about 5 m thick. The original great height depended on the lightness of the masonry, the upper wall being a double one with a gap of up to 1 m wide bridged by a horizontal Figure 1 Reconstruction by the author of Dun Carloway in its original state, soon after being built, based on a drawing by Alan Braby with modifications. The intra-mural cell originally shown in the foreground has been omitted as there is no evidence for one. The main difference from the original drawing, however, is the design of the raised wooden floor in the interior. This is assumed to have been ring-shaped, thus leaving the central area of the ground floor -with the main fireplace -clear. The design of the main roof is speculative but the arrangement shown here -with the base of the thatch protected by a stone parapet (the uppermost part of the outer skin of the double wall), and resting on the top level of lintels which brace the two skins apart -seems plausible. However, there would have to have been some kind of ring beam at the bottom of the roof frame to prevent the roof from spreading and collapsing the delicate structure of the top of the hollow wall.