Meusnier, experimenting in various ways, experimented with regard to the resistance offered by various shapes to the air, and found that an elliptical shape was best; he proposed to make the car boat--shaped, in order further to decrease the resistance, and he advocated an entirely rigid connection between the car and the body of the balloon, as indispensable to a dirigible.[*] He suggested using three propellers, which were to be driven by hand by means of pulleys, and calculated that a crew of eighty would be required to furnish sufficient motive power. Horizontal fins were to be used to assure stability, and Meusnier thoroughly investigated the pressures exerted by gases, in order to ascertain the stresses to which the envelope would be subjected. More important still, he went into detail with regard to the use of air bags, in order to retain the shape of the balloon under varying pressures of gas due to expansion and consequent losses; he proposed two separate envelopes, the inner one containing gas, and the space between it and the outer one being filled with air. Further, by compressing the air inside the air bag, the rate of ascent or descent could be regulated. Lebaudy, acting on this principle, found it possible to pump air at the rate of 35 cubic feet per second, thus making good loss of ballast which had to be thrown overboard.