feet below the scarp-top, the point being lower by several feet than the base of the low, irregular scarp found in the gallery from N to V, at a, b, and c. That there could be no rise in the orginal rock between M and a, b, and c was easily seen from the sharp slope down at the surface. The true state of the case immediately flashed upon me, as I stood, candle in hand, in the gallery at M and peered down the deep shaft at the man who held up his lamp to light this beautifully-worked scarp which towered for 21 feet, top and base being seen. It was a moment of relief, for the eccentricities of the rock cuttings beyond N had given me several bad quarter-hours. It was clear that the apparent turning at N was only superficial, that the line of scarp at its base must continue past N to somewhere near the point O, and that at that point we must expect a turning towards the north-west, as a, b, and c were so much higher than the base of the scarp. If this theory were correct, our winding gallery from N to V had been following along the top of the rock, inside the face of the scarp, probably along the top of the rock-base, of a great tower or bastion, the outer face of which we were yet to find. That at the point V we had again reached the true scarp (having crossed the width of the bastion) seemed possible, and the levels admitted of this.
All this, however, remained to be proved. I first decided to follow along the scarp-base from M by a gallery some 15 feet immediately below the gallery already opened. However, as we had previously opened a shaft from above at the point N, it seemed more economical, considering the earth-question, to deepen this shaft. Here we were again troubled with large stones. When these occurred within the limits of the dimensions of our frames, the task of break-up was easy, but it was an anxious moment when, after a frame had been fitted in, a head of a stone would be seen projecting 2 feet into the shaft just below it and extending into the earth-wall, how far no one knew. The fear was that the removal of the stone would widen the shaft so as to make insecure all above, but happily we managed to keep our shaft safe until we reached the rock. As I had foreseen, the base of the scarp continued past the point N and on to the point O, where it takes the expected turn to the north-west to form the bastion. From O to P there is a rise at the base of the scarp of 3 feet. At O' the top of the scarp appears in the gallery, having here a height of only 4 feet. However, there are evident signs (small and large chippings) that the top had been quarried away. At P the scarp is only 2 feet high, but as we turn the corner there is a sudden drop of 6 feet at the base, so that the scarp is 8 feet high. The rock is also scarped from P to P' (which was as far as we followed it), the scarp facing south-west, the line P—O, of course, facing north-east. As I have said above, when I discovered that we had been working across the top of the rock instead of around its scarped edge in our cross tunnel from the false corner at N to V, I thought that probably at V we had again reached the main line of scarp. Accordingly I set a gang of men to work from V towards U to meet the gang working from O towards P. The earth from the line O—P was at first carried to the surface up the shaft
