As mentioned before, the only ventilation in this case is such as will permit the issuing sewer gas to find its way into the house. It is by no means unusual to find no provision at all for ventilation, or to find the ventilating pipes so small that they are totally useless. In more cases than one, Burton found the soil pipe carried up as a rain-water pipe into the attics, where it received rain-water from two gutters, one from each side of the roof, and discharged all the sewer gas which escaped by it. Generally, the drinking-water cisterns are situated in such attics. It may be noted, in the other drawing (Fig. 21), that a trap is fixed on the main drain, which will keep back almost all sewer gas, and that ventilating pipes are so arranged that a constant circulation of fresh air exists through the whole drainage system, and will carry away with it any little sewer gas which passes through the trapping water. The most perfect water-supply arrangement does not necessitate the existence of cisterns in the house at all. This is beside the mark, for the reason that in London, to which Burton confines his remarks, the supply of water to the greater portion of the town is intermittent, so that cisterns are a necessity. Water, even in London, is almost always delivered in a sufficiently pure state to be drunk, but it is a very common thing for it to be contaminated in the cisterns. Even if there be no actual disease germs carried into the water, there is liability of deterioration from the mere fact of a large quantity of water being stored for a long time before use. If the cisterns are of so great size as to hold as much water as is used in, say, three or four days, it follows that all water drawn has remained in these cisterns for an average time of several days. This is by no means likely to improve its quality, but, on the contrary, if it does nothing else, it renders it flat. There are far more dangerous causes of contamination than this, however. The commonest of these is to be found in direct communication between the drains and the cisterns through the overflow pipes of the latter. This is shown in Fig. 20. It will be seen that there is a trap on the pipe by way of protection against the sewer gas. This is a by no means uncommon arrangement; but, as will be readily understood, such a trap is absolutely of no good. An overflow pipe to a cistern is merely an appliance to be put in use in case of an emergency; that is, in case of derangement of the ball valve through which the water enters. As a matter of fact, an overflow may not occur from year’s end to year’s end--probably does not--and, as a consequence, the trap soon becomes dry, and the temporary security afforded by it is lost. In 37 per cent. of houses inspected, Burton found direct communication between the drain or soil pipes and the drinking-water cisterns.