To clear up the changes which had happened at the subglacial catchment of glacier d’Argentière, an extensive study, with 31 borings and a coring down to the bottom (240 m) was performed in 1979/80, just upstream from the catchment, in an overdeepened area. The behaviour of the water level during boring with a hot water jet, and just after, was different from one bore hole to another, mainly because transient leaks appeared in the walls of bore holes. Next, the water level fluctuated slowly, in the same way in most of the deep bore holes, showing that glacier ice below about 100 m deep is slightly pervious. What is so measured is the pore pressure of water in deep ice. The piezometric gradient between bore holes, and the time lag between fluctuations of water level, which increases with distance from the right bank, shows that there is no waterway at the bottom of the overdeepened area, save at its up-stream end. Most of the melt water must flow between ice and rock along the right bank, its free surface rising by about 150 m during the increased discharge in June. No clear-cut correlation between the bottom pore pressure and the air temperature or the discharge at the subglacial catchment down-stream was found.