In all tokamaks the power threshold for H-mode access, Pth, is observed to increase approximately linearly with density [1]. However, a minimum Pth as a function of ne has been demonstrated on several tokamaks including COMPASS-D [2, 3], ALCATOR CMod [4] and DIII-D [5]. The Pth for the ITER ELMy H-mode may exceed the available heating power except at densities appreciably lower than the desired operating condition of ~1020 m-3. An operating scenario under consideration for ITER starts with the transition into H-mode at low ne followed by an increase in ne to the required operating condition. Hysteresis between the L-H and the H-L transitions is expected to maintain the H-mode during the ne rise. Therefore, it is important to characterise H-mode access in present day tokamaks under conditions of very low density. The departure of Pth on JET from the ITER ne scaling at very low densities was first reported by Horton et al in [6], Righi et al. in [7] and Sartori et al. in [8]. Low density data from COMPASS-D, JT-60U and JET have also been compared with Alfven Drift Instability theory by Igitkhanov et al. in [9]. This earlier JET work is extended here with experimental results from ne, Bt and q95 scans under conditions of low plasma density both with the Septum and the Septum Replacement Plate (SRP) in the MkII Gas Box (MkIIGB) divertor configurations.