It is important that the temporary works for control of the stream during the early period of construction should be so designed and constructed as to afford a maximum of safety to the work and to those engaged on it, with a minimum of pumping and interruption. It must be borne in mind that the problem almost invariably involves more than one handling, or one diversion, or one channel for the water. In other words, when the masonry has progressed to a certain point, the manner or channel of diversion must be changed to permit further progress. In fact several changes either laterally, or in elevation, or both, may be required. It must here be stated most emphatically that a sound plan for stream diversion must clearly contemplate and amply provide not only for every step that must be taken but also for the order of procedure.

This does not mean that the works should necessarily be planned of a capacity to exceed the largest known flood, nor that chances as to capacity or time available should never be taken; but it does mean that the engineer should not allow himself to be drawn into any situation where he does not know what to do next. Even if a time comes when he can do nothing but sit on the bank and watch the flood roll by, he must have clearly in mind the thing to be done when the river has subsided to a working stage. To be caught with a plan that is in the slightest degree indefinite or incoherent is almost certain to be expensive as well as foolish. This does not mean that the plan may not be modified, or even changed entirely, if found necessary as the work progresses. The balancing of capacity of channel against probable stream flow, and the cost of temporary works against the cost which would result from a flooding of the work, is the gist of the entire problem. With increasing depth of pit it becomes increasingly desirable that it be not flooded, particularly if the loose material is such that flooding the pit is liable to result in its being filled by the loose material. With increasing flow it rapidly becomes expensive and then impossible to handle the flow without flooding the pit.

One alternative involves cost of temporary works up to a capacity which is so rarely exceeded that the chance may be taken. The other involves an estimate of the number of times the work may be flooded, the cost of emptying the pit of water and loose material, cost of repairs to temporary works and the value of the lost time. The net remaining working periods should be such that substantial progress may be made between floods.

In fact, it is the satisfactory combination and balancing of all the conditions which determine not only the design of the temporary works, but indeed whether the project itself is feasible or possible.

Every project of magnitude must differ more or less from every other project, and each must be considered as a problem by itself. Nevertheless, it may be profitable to discuss certain types of actual diversion works to see what general principles may be laid down.