Rankine gives the ratio of the available rainfall to the whole fall as 1·0 on steep rocks, ·8 to ·6 on moorland and hilly pasture, ·5 to ·4 on flat, cultivated country, and nil on chalk. These figures are only rough. The figures for rocks and pastures are too high. The loss from evaporation and absorption is not proportional to the rainfall. It is far more correct to consider the loss as a fairly constant quantity in any given locality but increasing somewhat when the rainfall is great. The available rainfall in Great Britain has generally been overestimated. Sometimes it has been taken as being ·60 of the whole fall. More commonly the loss is taken to be 13 to 15 inches. This is correct for the western mountain districts, where the rainfall is about 80 inches and the soil consists chiefly of rocks partly covered with moorland or pasture. In other parts of the country, especially where flat, the loss is often 17 to 20 inches. All the above figures are, however, general averages. The proper estimation of the available rainfall at any place in any country depends a great deal on experience and judgment, and on the extent to which figures for actual cases of similar character are available. Regarding the “run-off” from saturated land during short periods, see CHAP. XII., _Arts. 1_ and _2_. 3. =Measurement of Rainfall.=--A rain-gauge should be in open ground and not sheltered by objects of any kind. The ordinary rain-gauge is a short cylinder. This is often connected by a tapering piece to a longer cylinder of smaller diameter. In this the rain is stored safely and is measured by a graduated rod. The measurement can be made more accurately than if the diameter was throughout the same as at the top. In other cases the water is poured out of the cylinder into a measuring vessel. If the rain-gauge was sunk so that the top was level with the ground, rain falling outside the gauge would splash into it and vitiate the readings unless the gauge was surrounded by a trench. Ordinarily the top of the gauge is from 1 to 3 feet above the ground. When it is 1 metre above the ground the rain registered is said to be on the average about 6 per cent. less than it should be, owing to the fact that wind causes eddies and currents and carries away drops which should have fallen into the gauge. The velocity of the wind increases with the height above the ground, and so does the error of the rain-gauge. Devices for getting rid of the eddies have been invented by Boernstein and Nipher (_Ency. Brit._, Tenth Edition, vol. xviii.), but they have not yet come into general use. The Boernstein device is being used experimentally at Eskdalemuir. It would appear that much splashing cannot take place when the ground is covered with grass, and that in such a case the top of the gauge could be 1 foot above the ground, thus making the error very small.