[Illustration: FIG. 70.] Converging breakwaters also tend to stop littoral drift, and the space inside them acts as a harbour of refuge in storms and as a sheltered place where dredgers can work (_Rivers and Canals_, CHAP. XI.). They have to be heavily built and are very expensive, and they are generally adopted only when there is an important seaport, and when they can be put to all the uses above indicated. APPENDIX A =Fallacies in the Hydraulics of Streams= (CHAP. I., _Art. 4_, and CHAP. VI., _Art 2_).--In an inundation canal in India the supply during floods was excessive. Orders were given that a flume be made at the head, as shown in fig. 71. The sides were to be revetted, as shown in fig. 19 (CHAP. VI., _Art. 3_); the length, excluding the splayed parts, was to be 200 feet, and the floor was to be a mattress well staked or pegged down. The order stated that “by this means we cannot get into the canal much more than its true capacity.” With 9 feet of water, a surface fall of 4 inches in 300 feet would give a velocity of some 6·5 feet per second, and a further fall of about 8 inches would be required at the head of the flume to impress this velocity on the water. The flume would reduce the depth of water in the canal by 1 foot, _i.e._ from 9 feet to 8 feet. This would not be in anything like the proportion desired. Moreover the flume, unless the bed was extremely well protected, would be destroyed. The above is a case of exaggerating the effect of an “obstruction.” [Illustration: FIG. 71.] Again, on a branch canal it was observed that “wherever cattle crossings exist there is a deep silt deposit which practically blocks the branch.” The deposit exists because the sides of the channel are worn down. A wide place always tends to shoal (CHAP. IV., _Art. 9_). If the deposit obstructed the flow of water there would be a rush of water past it, and it could not exist. The Gagera branch of the Lower Chenab Canal--the left-hand branch in fig. 72--was found to silt. It was proposed to make a divide wall (fig. 72) extending up to full supply level. The idea is unintelligible. The silt does not travel by itself but is carried or rolled by the water. As long as water entered the Gagera branch, silt would go with it. The authorities, who had apparently accepted the proposal, altered the estimate when they received it, and ordered the wall to be made as shown dotted and of only half the height. This was done. The idea seems to have been that the wall would act as a sill and stop rolling silt. This is intelligible, but see CHAP. IV., _Art. 2_, last paragraph. Moreover, there was a large gap, A B, in the wall. The work is said to have proved useless, and proposals have been made to continue the wall from A to B. In this form it is conceivable that it may be of use. [Illustration: FIG. 72.]