Let the flagship throw over a flat board from abreast the foremast and make a preconcerted signal.
Simultaneously, let another ship, say the fourth ship, measure the distance from the foremast of the flagship with a range finder; there is no reason why the error in measuring this distance, about 1 200 yards, should be greater than ½ of 1 per cent.
Simultaneously, let each ship in the fleet start a stop watch and begin to count the revolutions of the engines.
When the board gets abreast the range finder say of the fourth ship, let this make a preconcerted signal, let all ships stop the stop watches, and note how many revolutions were made during the interval, and what was the exact length of that interval.
If the fleet be going 10 knots, the interval in time will be about 3.6 minutes, that is 216 seconds. There seems no reason why a mistake should be made greater than I per cent in measuring the time.
Therefore, it would seem that this method would enable the admirals and captains to ascertain the speed at any time, and the corresponding revolutions, with an error not much greater than 1 per cent.
It may be objected that there is no need of ascertaining the fleet speed with such extreme accuracy, and that the patent log and speed curves supplied the ships give means of ascertaining the speed closely enough for practical purposes.
To this it may be answered that everybody knows that the patent log is far from accurate, and that the speed corresponding to any given number of revolutions varies with the degree of foulness of the bottom, and the force and direction of the wind and sea. Furthermore, that, if we are to regulate the fleet speed by means of the revolutions of the engines, we may as well do it by an accurate method as by an inaccurate method; and to carry out the method herein suggested would cost practically nothing, would not take up the time of anybody not on duty, and could be done at any time.