Matters connected with the hydraulics of open streams seem to lend themselves in a peculiar way to loosely expressed remarks and fallacious opinions. The set of a stream towards a bank is sometimes supposed to profoundly affect the discharge of a diversion or branch. Its effect is simply that of “velocity of approach,” which, as is well known, is quite small with ordinary velocities, and is merely equivalent to a very small increase of head. Narrow bridges or other works are sometimes said to seriously “obstruct” a stream without any observations being made of the fall in the water surface through the bridge. This fall is the only measure of the real obstruction.[7] CHAPTER II RAINFALL 1. =Rainfall Statistics.=--The mean annual rainfall varies very greatly according to the locality. In England it varies from about 20 inches at Hunstanton in Cambridgeshire, to about 200 inches at Seathwaite in Cumberland; in India, from 2 or 3 inches in parts of Scinde, to 450 inches or more at Cherrapunji in the Eastern Himalayas. Rain is brought by winds which blow across the sea. Hence the rainfall in any country is generally greatest in those localities where the prevailing winds blow from seaward, provided they have travelled a great distance over the sea. Rainfall is greater among hills than elsewhere, because the temperature at great elevations is lower. Currents of moist air striking the hills are deflected upwards, become cooled, and the water vapour becomes rain. This process, if the hills are not lofty, may not produce its full effect till the air currents have passed over the hills, and thus the rainfall on the leeward slopes may be greater than elsewhere, but on the inner and more lofty ranges the rainfall is generally greatest on the windward side. Thus the rainfall may vary greatly at places not far apart. An extreme instance of this occurs in the Bombay hills, where the mean annual falls at two stations only ten miles apart are respectively 300 inches and 50 inches. In temperate climates the rainfall is generally distributed over all the months of the year; in the tropics the great bulk of the rain often falls in a few months. The fall at any one place varies greatly from year to year. To obtain really reliable figures concerning any place, observations at that place should extend over a period of thirty to thirty-five years. The figures of the mean annual fall will then probably be correct to within 2 per cent. The degree of accuracy to be expected in results deduced from observations extending over shorter periods is as follows:-- No. of years 25 20 15 10 5 Error per cent. 3 3¼ 5 8 15