As regards the power consumed, the engines of each pumping station worked up to about 236 indicated horse-power, and their regulated duty for each train, including the anticipatory pumping, was equivalent to 5·5 minutes of work for every mile of the length of tube they had to exhaust. As the running speed averaged 40 miles per hour, or a mile in 1·5 minutes, the 236 horse-power during the 5·5 minutes of pumping must be regarded as equivalent to 865 horse-power during the actual passage of the train. Now, making full allowance for piston friction and extra friction on curves, for the power expended in getting up speed, for the excess of air-pump resistance due to the changes of temperature experienced by the air under exhaustion, and even for the very large actual amount of friction in the engines employed, the work done should have been represented by an expenditure of 240 horse-power during the passage of the train. If to this is added an allowance for leakage, such as the experiments at Dalkey indicated would be amply sufficient with the longitudinal valve in good condition, it may be said that Mr. Brunel had a right to expect that the duty would be performed with an expenditure of 300 horse-power; whereas it actually required 865 horse-power, or nearly three times the amount. The explanation of this waste is simple. Serious and unexpected causes of failure developed themselves in the longitudinal valve, and led to an excessive amount of leakage. A great part of the normal duty of the engines was, as has been stated, to exhaust the tube previous to the entry of the train; and when, owing to leakage, the amount of air to be so pumped out was greatly increased, it became necessary that the operation should be commenced much earlier. There was thus a longer time during which the leakage could take place, and a still greater amount of air to be pumped out. It therefore followed that a large increase of leakage involved waste of power in an enormously increased proportion.