The same principle can be adopted for any distributary whose off-take is in the upper part of a branch, _i.e._, where the branch is many times larger than the distributary, and where it is not possible or not desirable to regulate the water level of the branch. For a distributary further down the branch, the turns of branch and distributary can be arranged as explained above for a canal bifurcation. The orders given to the gauge reader are, as before, to give the channel whose turn it is, full supply and to send the balance down the other channel. When the turn of distributary is over it becomes the turn of the branch. The distributary would not be closed if this would cause the full supply in the branch to be exceeded. Care must be taken that every distributary receives full supply during part of the time when the branch is receiving full supply. If its turn came only when the branch was receiving a residuum supply, or rather when the residuum supply was reaching the distributary off-take--for in the case of a distributary whose off-take is far down a long branch the two things are not the same--it might, in the event of the supply in the main canal falling exceptionally low, receive no water at all. The time taken by a rise in travelling down a canal is very much the same as that taken by a fall and each takes effect more or less gradually. When a branch receives, at any point, a temporary increase in its supply, owing to the closure of a distributary for, say, three days, there will be a rise lasting for three days at a point further down. The rise will take some time to come to its height, and some time to die away. There will be about three days from the commencement of the rise to the commencement of the fall, or from the end of the rise to the end of the fall. If, either in the main canal or in a branch, there is any distributary into which full supply cannot be got, its turn can be increased accordingly. Owing to the shortness of the turns, and to allowance having to be made for the time occupied by rises and falls in travelling down the branch, the fixing of the turns for distributaries near the tail of the branch requires a good deal of consideration. Matters are facilitated by making a sketch (Fig. 25) in which the widths of the channels, as drawn, are roughly in proportion to the full supply discharges. If 14 copies of the sketch are made the arrangements for each day can be shown on them, full supply being shown black and residuum hatched. Distributaries would be shown as well as the main channels. [Illustration: FIG. 25.] The irrigation registers of course show how the irrigation of the different channels is going on from year to year and if changes in the turns become necessary they can be effected.