For this purpose any probable value of S will suffice. 5. =Discharge Curves and Tables.=--Ordinarily it will be possible, by plotting the observed discharges as ordinates, the gauge readings being the abscissæ, to draw a discharge curve and from it construct a discharge table. Unless the channel is of firm material and not liable to change, there are likely to be discrepancies among the observed discharges, so that a regular discharge curve will not pass through all the plotted points. If the discrepancies are not serious, they can be disregarded and the curve drawn so as to pass as near as possible to all the points, but otherwise trouble and uncertainty may arise. The soundings should be compared in order to see whether changes have occurred in the channel. If such changes do not account for the discrepancies, the cause must be sought for in some of the recorded velocities. If no sources of error in these can be found, such as wind, it is possible that the velocity has been affected by a change in the surface slope owing to some change in the channel downstream of the length. Failing this explanation, the discrepancies must be set down to unknown causes. With an unstable channel and where accuracy is required, the sectional areas and velocities should be regularly tabulated or plotted so that changes may be watched and investigated. To do this it may be necessary to take surface slope observations, or to set up extra gauges which will show any changes in the slope. If, downstream of the discharge site, there is any place where affluents come in and bring varying volumes of water, or where gates or sluices are manipulated, and if the influence of this extends up to the discharge site, the water-level there no longer depends only on the discharge, and a discharge table must be one with several columns whose headings indicate various conditions at the place where the disturbances occur. In order to show how the gauge readings and discharges vary from day to day throughout the year, a diagram should be prepared showing the gauge readings and discharges as ordinates, the abscissæ being the times in days starting from any convenient date as zero. Such a diagram, showing only gauge readings, is given in fig. 56, CHAP. XII.