[Illustration: Fig. 3.--Diagram of variations in tractive stresses and tonnage taken as a function of the speed.] All those masses of water which receive a horizontal acceleration from the keel run counter, on the contrary, to the propulsive stress, and it becomes of interest, therefore, to bring them to a minimum. The vertical stress is limited by the weight of the boat, and, theoretically, with an infinite degree of speed, the boat would graze the water without being able to enter it. The annexed diagram (Fig. 1) shows the form that calculation has led Mr. Pictet to. The sides of the boat are two planes parallel with its axis, and perfectly vertical. The keel (properly so called) is formed by the joining of the two vertical planes. The surface thus formed is a parabola whose apex is in front, the maximum ordinate behind, and the concavity directed toward the bottom of the water. The stern is a vertical plane intersecting at right angles the two lateral faces and the parabolic curve, which thus terminates in a sharp edge. The prow of the boat is connected with the apex of the parabola by a curve whose concavity is directed upward. [Illustration: Fig. 4.--Diagram of the variations in the power as a function of the speed.] When we trace the curve of the tractive stresses in a boat thus constructed, by putting the speeds in abscisses and the tractive stresses in ordinates, we obtain a curve (Fig. 2) which shows that the same tractive stress applied to a boat may give it three different speeds, M, M', and M'', only two of which, M and M'', are stable.