At first sight, the second trench seemed to be the most promising of the three. We had decided to dig here because the weathering pattern on the cliff face suggested that there might once have been an overhang at this point, while the bank in front of it dipped back down, as if a mound had formed over the collapsed roof. This proved to be correct. The overhang had indeed collapsed, probably during the Mesolithic, as deposits similar to those that characterised the Mesolithic occupation and later periods over the rest of the site overlay it. We therefore hoped that it might have held back some of the deposits that had formed inside the former shelter. This would have helped our interpretation of the main shelter, beside Trench I, where the deposits inside had long been cleared away.
The first season’s work appeared promising. A number of Mesolithic style lithics were found and there were hints of a number of discrete soil formation horizons. However, work in 1997 showed that over the millennia, water dripping off the cliff had thoroughly altered all the deposits, washing material downhill and moving it some distance from where it was dropped. Iron pan had formed towards the base of the lowest deposit and small solution gullies had formed in the underlying sandstone bedrock.
One thing that did appear curious was discovered towards the end of the 1997 season. The former roof of the shelter had fallen forwards, wedging itself into the ground at an angle of around 45°. It had been trimmed flat, with tool marks resembling those in the main shelter, although this lay at least 0.15 m below the ground surface. Why this took place could not be determined, as it did not appear to be part of work either John Harris’s or part of the creation of the Park c 1800.