As to the source of rivers, we need add nothing to what has already been said about the “congregation of waters” (_congregatio aquae_) and about springs and fountains. It was commonly thought in medieval and classical times that two or more rivers may rise from one source and flow off in diverse directions. The most striking example of this was furnished by the Scriptural four rivers of Paradise, which, though rising from one stream, were believed to find their way to at least three different seas. In commenting on the rivers of Paradise, Gervase of Tilbury expressly asserts[928] that not only is it possible for more than one stream to rise from the same headwaters but that the same rivers may again mingle and again separate their waters. Giraldus Cambrensis describes[929] how the Shannon of Ireland rises in a lake between Connaught and Munster and thence divides into two branches flowing in opposite ways, one southward to the “Sea of Brandan,” the other northward into the Northern Ocean. It is true that in regions of imperfect drainage development, like Ireland, northern North America, and parts of the Amazon Valley, two small streams occasionally do spring from a single source. On the other hand, it is entirely contrary to the laws of hydrography that two or more full-grown rivers should either leave a lake and depart across country in different directions or, except in the case of deltas, owe their origin to the separation of the waters of a single large stream. Classical and medieval geographers, however, were not acquainted with this law, and the words of the Bible justified the writers of geographical books, even down to as late a date as the eighteenth century, in making broad rivers divide into separate branches and wander at random over the country. Giraldus Cambrensis noted several other peculiarities of rivers.[930] For example, he remarked that the stream at Wicklow which flows across the harbor (we may presume in a channel through mud flats) is brackish at ebb tide; a similar river at Arklow is fresh.[931] Tide water, he said, does not mingle with the River Conway in North Wales.[932] Elsewhere he observed that the term _aber_ in Welsh is applied to all those places where one stream flows into another.[933] The River Dee is not affected by rains, but the winds make it rise.[934] It changes its bed every year, and, as its course forms the boundary between England and Wales, these changes are interpreted as omens foretelling whether the English or the Welsh are going to be the more successful in their combats with each other during the succeeding year.[935] THE NILE FLOOD