Below Parana, as far as the estuary, the careful observations that have been made since 1903 on the movement of the river have enabled us to learn some of its laws.[128] We can distinguish four sections of unequal length. From Parana to Diamante the river remains in touch with the cliffs of the left bank. It is not straight; it describes a series of linked crescents of equal radius, which seem to be traces of so many meanders. Only one in two of the windings of the cliff is followed by the channel. The wandering of the river is confined within limits as in a fixed mould. The Paracao shallow, which for a long time prevented ships from reaching Santa Fe (gradually deepened by dredging from eight to nineteen feet between 1907 and 1911) is at the angle where two of these curves meet. On the right bank the secondary arms continue to follow the river (Parana viejo, Riacho de Coronda).[129] [128] In studying the variations of the bed of the Parana it is necessary to avoid comparing maps drawn at dates separated by long intervals. The differences of such maps are such that they do not enable us to follow the processes by which the actual forms have been derived from earlier forms. The analogies which they show are sometimes due, not to the permanence of the topography, but to the return of a complete cycle of changes, or of conditions analogous to the earlier conditions. [129] The secondary arms of the right bank, north of Santa Fe, were not explored until 1870. Sullivan's map (1847) only mentions the Riacho de San Jeronimo, which is visible for a short distance below 20 deg. S. lat. The right bank was the domain of the Indians, and the Correntinos would not venture near it. In 1870 ships began to use the San Javier arm, on which many colonies arose. Further north the Parana Mini has been used since 1890 for exporting _quebracho_ timber.