They had followed down the glacier for a day, and the ice became so very rough they were obliged to leave their sledges behind them and proceed with the greatest care. As they continued to descend, the glacier gradually dwindled and then suddenly ended in a low wall of ice. In the valley beyond were some frozen lakes, beyond which the valley changed into a series of deep, narrow gorges, filled with long lines and confused heaps of stones and other débris, shed from the glacier as it melted. Perhaps in this valley, more than anywhere else, lay the evidences of what was happening and what had been. There lay the glacier, inert and dead, while the summer sun was gradually wasting its huge mass. On either side its shrinking tributaries had already severed their connection with it, and receded up the mountain sides. Everything pointed to receding glaciation. Not only were the gorges filled with the stones and débris of the moraines, but thousands of feet up the hill-sides they lay in clear-cut lines, showing how vast the proportions of the glacier had once been. Other glaciers observed gave the same indications. One of these, named the Ferrar Glacier, after the discoverer, was described as containing probably as much ice as any hitherto known in the world. Two others, the Barnes and the Shackleton Glaciers, each contained a great deal more; but all were greatly shrunk from the tremendous proportions they must once have had, and which could be traced by the lines of moraines. It was from observing these glaciers and speculating upon the immeasurable quantity of ice that once must have slowly flowed along them to the sea, that a theory was formed as to the real significance of the great ice barrier lying to the south of the Antarctic Circle. The opinion was generally entertained that the ice in this barrier is afloat. It had receded from the time when Ross first discovered it. Even while the expedition was in the neighbourhood it receded. In September 1902, a depôt was established on the ice at a line drawn between a volcanic peak at the extreme end of Minna Bluff and the top of Mount Discovery. In November 1903, this spot was found to have moved 608 yards out of the alignment, moving east of north. On the sides of Mount Terror, 800 feet above the existing surface of the barrier, moraines marked its original height. From these facts it was reasoned that at one time the ice forming the cap over the Antarctic continent was enormously deeper than it is to-day, and that it then extended out into the Antarctic Ocean for probably hundreds of miles further than it now reaches. The slow but steady flow of the great ice stream had absolutely filled the sea off the coast of Victoria Land, and the barrier, now existing, was the remains of this once mighty mass.