The ground has been back-filled, the opening closed, and the grass grows undisturbed. Yet for several generations, local people knew that something lay underneath: a small area of ground that would intermittently subside and collapse, as if the earth were slowly remembering a hollow it had tried to forget.

What lay below was a souterrain, an early medieval underground structure, typically stone- or earth-cut, thought to have served as a place of refuge or cold storage, and often associated with nearby settlement. In 1993 the collapsed area widened enough to allow entry, and archaeologists Lee Snodgrass and Paddy O'Leary surveyed what they found. The opening, roughly three metres across and two metres deep, gave access to four passages radiating outward. The first, running northwest for around eight metres and mostly earth-cut, connected three dome-shaped chambers linked by creepways, the low narrow tunnels one must crawl through to move between sections. Chamber 1 showed a possible construction shaft, a vertical cut used during building and then left in place. Chamber 3, the furthest along that passage, narrowed at its northeast end and sloped downward toward a spill of stones. Two further passages extended northeast and southeast but were inaccessible during the survey. The fourth passage was more formally constructed: its entrance was framed by three lintels, large flat capstones, set on upright orthostats, and on two of those lintels the surveyors noted cupmarks, shallow circular depressions pecked into the stone surface whose age and purpose remain open questions. This passage led toward two further chambers, both blocked and unreachable. The reuse of older marked stones in a souterrain's fabric is not unusual, but it lends the structure an additional layer of ambiguity, the possibility that the builders were incorporating material with an earlier significance they may or may not have understood.

The site has since been back-filled, and there are no surface remains. The record of what was there exists only in the archaeologists' survey and in subsequent published accounts, including O'Mahony's 2007 study. The slope above Roaringwater Bay looks like ordinary farmland because, now, it is.