The origin of these shallow freshwater lakes, which form a characteristic feature in the scenery of East Anglia, has been much debated; but with the knowledge obtained from a study of the submerged forests the explanation is perfectly simple. During this period of slow submergence each of the shallow valleys in which the broads now lie was turned into a wide and deep navigable estuary, which extended inland for many miles. When the subsidence stopped the sea and tides soon formed bars and sand-banks at the mouths of the estuaries, and lateral tributaries pushed their deltas across. The Norfolk rivers, being small and sluggish, were driven to one side, and could neither cut away the sand-banks nor fill up with sediment such wide expanses. These estuaries therefore were silted up with tidal mud and turned into irregular chains of lakes, separated by irregular bars and sand-banks. The lakes, instead of becoming rapidly obliterated and filled up by deltas which crept gradually seaward, remained as freshwater broads; for as soon as a bank became high enough for the growth of reeds and sedges the river mud was strained out and only nearly clean water reached the lagoon behind. Thus a depression once left, provided it was out of the direct course of the river, tended to remain as a freshwater lake until vegetable growth could fill it, and the river mud was spread out over the salt-marshes or went to raise the sand-banks till they became alluvial flats, and thus still more thoroughly isolated the broad. A few centuries will see the disappearance of the last of the broads, which have silted up to an enormous extent within historic times; but the fact that so many of these broads still exist may be taken as clear evidence of the recent date of the depression which led to their formation. When we look at ancient records, and notice the rapidity with which the broads and navigable estuaries are becoming obliterated, we cannot help wondering whether the measure of this silting up may not give us the date of the last change of sea-level. It should do so if we could obtain accurate measurements of the amount of sediment deposited annually, of the rate at which the sea is now washing it in, and of the rate at which the rivers are bringing it down. All these factors, however, are uncertain, and it is particularly difficult to ascertain the part played by the muddy tidal stream which flows in after storms and spreads far and wide over the marsh. Though all the factors are so uncertain, we can form some idea of the date of the submergence. Many years ago I made a series of calculations, founded on the silting up of our east coast estuaries, the growth of the shingle-spits, and the accumulation of sand-dunes. The results were only roughly concordant, but they seemed to show that the subsidence stopped about 2500 years ago and was probably still in progress at a date 500 years earlier. This question of dates will be again referred to in a later chapter.