In the case of a large stream which shifts its course, the reading of a gauge does not give a proper indication of the water-level. In other words, the distance of the gauge from the two ends of the reach is subject to alteration. The case is the same as if the stream was stable and the gauge was shifted about. In such a stream there ought, if accuracy is required, to be a group of two or more gauges for each point where there would be only one if the stream was not a shifting one. Also, owing to erosion of the bank or the formation of a sandbank, it may often be necessary to shift the gauge. When possible it should be kept in a fixed line laid down at right angles to the general direction of the stream. When shifted, its zero level should be altered in such a way that the reading at the new site at the time of shifting is the same as it was before shifting. When the gauge is moved back to the original site its zero should be placed at its original level, though this may give rise to a sudden jump in the reading for the reason given in the first sentence of this paragraph. 3. =Plan and Sections.=--Making a survey and plan, and laying down on it the lines for longitudinal and cross-sections, and taking levels for the sections, are ordinary operations of surveying. If any land is liable to be flooded, its boundaries should be shown on the plan and on some of the cross-sections. Unless the water is shallow, it is necessary to obtain the bed levels from the water-level by soundings, the level of a peg at the water-level having been obtained by levelling. All the sections should show the water-level as it was at some particular time, but the water-level will probably have altered while the survey was in progress, and allowance must be made for this. The pegs at all the cross-sections and on both banks of the stream--for the water-levels at opposite banks may not be exactly the same--may, for instance, be driven down to the water-level when it is steady, and thereafter any changes in it noted and the soundings corrected accordingly. In order to ascertain what changes are occurring in the channel it may be necessary to repeat the soundings at intervals and, if there is much erosion of the bank, to make fresh plans.