secondary source rods add a further contribution to the discharges, which can be slightly reduced, but the figures for this are not available. Approximately 98% of discharges from PWR plants occur in liquid form. Given the relative radiological impact, which is on average 100 times higher for gaseous emissions than liquid discharges at sea-shore sites, it may be useful to look at further increasing the ratio of liquid emissions. However, this solution assumes appropriate outlet channels, which have already been taken into account in the impact studies. Given the low radiological impacts and the volume of effluent to be treated, detritiation is not realistic. EDF has a fuel development programme whose aim is to increase the electricity generated by each tonne of uranium and to reduce the quantity of waste. However these developments have tended to lead to an increase in tritium production. The ARCO association highlighted the fact that at Flamanville, tritium discharges have doubled due to an increase in electricity generation of just 4.5%. The La Hague fuel reprocessing plant is the main source of tritium releases, mostly in the form of liquid discharges, given the relative impact which is 1,000 times lower than for atmospheric releases. Given the potentially large volumes to be treated (of the order of 40,000 m 3 per year) and the very low activity concentration at the time of discharge, detritiation is not achievable with currently available technologies. Reducing waste volume further up the process would lead to occupational exposure levels that would be incompatible with process optimisation. Various areas for discussion, R&D and process improvement have been identified, in particular: the need to assess which process changes or fuel reprocessing options have a genuine industrial future (voloxidation, pyroprocessing, etc.); the necessary discussions regarding design of a reprocessing plant with fewer site-specific advantages than La Hague in terms of radiological capacity, which would not therefore be granted the same effluent release permits. PHENIX, SUPERPHENIX and fast breeder reactor designs were not extensively discussed, despite the fact that the forthcoming GEN IV programme will soon be making the headlines. Tritium production is higher than in PWR plants and 95% of the tritium formed in the fuel passes into the molten sodium in the reactor coolant system. The borated control rods add a significant contribution and tritiated methane is formed. Tritium releases, standardised for one GWe, are overall twice as high as in the PWR design. The tritium could be recovered and reused from the molten sodium in the SUPERPHENIX reactor. The radiological impact of this operation needs to be more closely analysed. A small fraction of the tritium produced in nuclear reactors is recycled by small producers in order to create synthetic tritium-labelled molecules for industrial uses (luminescence) and for research. Although these sources only represent a small volume of waste, they can cause significant inadvertent environmental marking. The radiological impact of this is difficult to identify and depends largely in speciation. An appropriate outlet channel should be provided in order to manage these sources. Tritiated waste is stored in various centres run by the French National Radioactive Waste Management Agency (ANDRA). Some tritium-marking in ground water has been traced to the “Centre Manche” (CSM), in places reaching values of several hundred Bq/L, leading to various assessments of potential developments over time. Work to characterise the source and the transfers to outlet routes needs to be continued and the predictions need to be reinforced by taking appropriate measurements. The Soulaines and Morvilliers centres have been subject to prudent management of the tritium radiological capacity and to tough rules as to the intake of waste packages. This situation has meant that environmental marking has remained low. However, ANDRA does not have actual authorisations for tritiated waste, which leaves many sources of tritiated waste without a permanent home. The solution described in the CEA report drafted under the 2006 Act is 50 years interim storage, until a new disposal site is opened, with the appropriate authorisations. Some defence sites have nuclear purification, recycling and storage facilities, for manufacturing weapons subassemblies and for reprocessing or neutralising waste. This is the case at the CEA centres at Valduc, Marcoule and Bruyères-le-Châtel. Monitoring of these facilities has shown a hundred-fold decrease in releases since the 1970s and 1980s, with gaseous releases of 95 TBq (0.3 g) at Valduc in 2007, for instance. This reduction in releases has been achieved by detritiation of highly tritiated waste by using heat treated to separate tritiated water from a solid substrate. The tritiated water is then trapped on zeolite. The resulting volumes are low and the waste is stored at the Valduc centre. The practice is optimised and research is continuing into the recovery and use of the tritiated water. The working group noted a lack of data about certain facilities that are the exclusive responsibility of the French Ministry of Defence (reactors on sea-going vessels).