But it is supposed that the Nile levels at that time were about 2 metres lower than they are now, and it is necessary to consider in what way such a change of conditions would modify the views of what Lake Mœris was and did, as given in the foregoing arguments and calculations. The supposition, or certainty, that the Nile in the time of Herodotus was about 2 metres lower in level than it is now, is based on the following observations, which Mr. Petrie has given me. He estimates that the rate of rise has been about 4 inches a century. This, he states, is shown by a Roman wall at Tanis and by the town-level of Naukratis, both old towns in Lower Egypt. The old tombs at Memphis are now under water. At Edfu the High Nile rises shoulder high on the walls, which shows a rise of 4 inches or more per century. At Aswân (Assouan) the records of High Niles on the Roman Nilometer show that they were lower than now by an amount calculated at a rate of 4 inches per century. There is also, Mr. Petrie adds, other evidence of the same sort, but less definite, giving the same general result. If now we suppose the Fayûm (Lake Mœris) filled to R.L. 20·50 and emptied to R.L. 17·50, there is nothing to be changed in the calculations, except the maximum and minimum surface levels of the lake. Thus there would be a rather, but not much, larger area reclaimed and the Edwah-Biahmu bank would have been formed along the edge of the lake at lowest water, instead of in two metres of water. This modified view of its formation would seem to be more probable than that which supposed it to have been formed in water. If, however, we suppose the lake still filled to R.L. 22·50 as a maximum, while its lowest level reached R.L. 17·50, the discharges found to have been necessary to fill the lake (under the conditions previously assumed excepting as regards minimum level), must be increased by 50 per cent., and the figures representing the return-flow be doubled. In all probability the maximum level of the lake was somewhere between R.L. 22·50 and 20·50, and may be taken as varying from R.L. 22·00 to 21·00.