The curves established for the years 1908 to 1910 by the Argentine hydrographical service enable us to analyse the mechanism of the flood with a good deal of confidence. The beginning of the flood at Bajada Grande in October corresponds to the first flood of the upper Parana. During this first phase the curve of the Bajada is parallel (thirteen days later) to that of Posadas. There is the same parallelism in November, December and January. If the summer rains are light on the upper Parana, the flood is late on the lower Parana, and the water is still low there in December (0.20 below the low-water mark on December 31, 1910). At the beginning of March, before the maximum of the flood, the curve of Bajada Grande differs from the curve of Posadas. It is the time when the flood of the lower river is caused by the rise of the Paraguay. The secondary floods of June and July again have their origin in the upper Parana, but, as they are added to the flood of the Paraguay on the lower river, they reach a higher level there than at Posadas; the difference gradually disappears as the flood of the Paraguay subsides. It is the addition of the late floods of the upper Parana to the flood of the Paraguay that causes on the lower river the abnormal floods that occur there at irregular intervals (in 1825, 1833, 1858, 1878, 1905 and 1917). Below the Bajada the height of the floods progressively declines. On the estuary they are no longer perceptible; variations of level are due entirely to the tides. In the channels of the delta of the Parana the tide does not reverse the current as it does in the estuary of the Uruguay, but it causes a slight rise of the water; and this has been observed sometimes, at very low water, as far as Rosario. It is near Corpus, about forty miles above Posadas, that the upper Parana escapes from the restraint of the Brazilian tableland, which imprisons its valley, from the falls of the Guayra, in a deep fissure between lofty basalt cliffs. Below Posadas the river leaves the region of hills and red earth. Below Corrientes it flows everywhere over its own alluvia. Even above Corrientes its form has surprising characteristics of youth. The precise survey done on its banks has brought to light a very distinct break of its fall above Villa Urquiza, about 400 miles from Buenos Aires. The fall, which from Corrientes onward remains between sixty and forty millimetres per kilometre, sinks suddenly to thirteen over a stretch of twenty-five miles, and then rises again to thirty to forty-five millimetres.[124] Below Rosario the mean descent is twelve millimetres to the kilometre, below San Pedro only six.