A hurricane initiation mechanism, believed to be new, is proposed for the eastern tropical North Atlantic Ocean. It starts with an outbreak of warm dry air from the Sahara Desert moving out over a fairly large region of ocean just west of the big bulge of Africa. Critical to the hypothesis is the experimental fact that heat diffuses significantly slower in air than water vapor does. In summer and early fall the desert air of the outbreak is warmer than the ocean surface it first encounters. Thus this air layer is cooled from below, which is initially stabilizing. However, water vapor diffuses up into the dry air faster than the air’s heat diffuses down to the sea surface, all over the generating region simultaneously. Consequently, a horizontally large layer of air somewhat above the sea surface becomes buoyant (less dense) and rises up as a unit, and the pressure of this layer decreases by the perfect gas law. Then the water vapor in the ascending air condenses around dust particles brought in from the desert, releasing heat and producing an additional upward acceleration of the already ascending air. Atmospheric pressure lowers further in accordance with Bernoulli’s law: where the (vertical) speed is greatest, the pressure is least. Measurements are suggested to validate the hypothesis if they do not already exist.