It must be remembered that when rain continues for long periods, ground which in the ordinary way would generally be considered permeable becomes soaked and eventually becomes more or less impermeable. Mr. D. E. Lloyd-Davies, M.Inst.C.E., gives two very interesting diagrams in the paper previously referred to, which show the average percentage of effective impermeable area according to the population per acre. This information, which is applicable more to large towns, has been embodied in Fig. 16, from which it will be seen that, for storms of short duration, the proportion of impervious areas equals 5 per cent. with a population of 4.9 per acre, which is a very close approximation to the 4.5 per cent. obtained in the example just described.

Where the houses are scattered at long intervals along a road the better way to arrive at an estimate of the quantity of storm water which may be expected is to ascertain the average impervious area of, or appertaining to, each house, and divide it by five, so as to get the area per head. Then the flow off from any section of road is directly obtained from the sum of the impervious area due to the length of the road, and that due to the population distributed along it.

[Illustration: FIG. 16.—VARIATION IN AVERAGE PERCENTAGE OF
EFFECTIVE IMPERMEABLE AREA ACCORDING TO DENSITY OF POPULATION.]

In addition to being undesirable from a sanitary point of view, it is rarely economical to construct special storm water drains, but in all cases where they exist, allowance must be made for any rain that may be intercepted by them. Short branch sewers constructed for the conveyance of foul water alone are usually 9in or 12 in in diameter, not because those sizes are necessary to convey the quantity of liquid which may be expected, but because it is frequently undesirable to provide smaller public sewers, and there is generally sufficient room for the storm water without increasing the size of the sewer. If this storm water were conveyed in separate sewers the cost would be double, as two sewers would be required in the place of one. In the main sewers the difference is not so great, but generally one large sewer will be more economical than two smaller ones. Where duplicate sewers are provided and arranged, so that the storm water sewer takes the rain-water from the roads, front roofs and gardens of the houses, and the foul water sewer takes the rain-water from the back roofs and paved yards,

it was found in the case previously worked out in detail that in built-up roads a width of 36 ft + 2 (8 ft 7 in) = 53 ft 2 in, or, say, 160 sq. ft per lineal yard of road would drain to the storm water sewer, and a width of 2 (6 ft 10 in) = 13 ft 8 in, or, say, 41 sq. ft per lineal yard of road to the foul water sewer. This shows that even if the whole of the rain which falls on the impervious areas flows off, only just under 80 per cent. of it would be intercepted by the special storm water sewers. Taking an average annual rainfall of 30 in, of which 75 per cent. flows off, the quantity reaching the storm water sewer in the course of a year from each lineal

30 75
yard of road would be —- x 160 x —- = 300 cubic
12 100
feet = 1,875 gallons.

[Illustration: FIG. 17.—SECTION OF "LEAP WEIR" OVERFLOW]

The cost of constructing a separate surface water system will vary, but may be taken at an average of, approximately, l5s. 0d. per lineal yard of road. To repay this amount in thirty years at 4 per cent, would require a sum of 10.42d., say 10-1/2d. per annum; that is to say, the cost of taking the surface water into special

10-1/2 d. x 1000
sewers is ———————— = 5.6, say 6d. per 1,000
1875
gallons.

If the sewage has to be pumped, the extra cost of pumping by reason of the increased quantity of surface water can be looked at from two different points of view:—