It is shown that introducing gravity in the energy minimization of drops on surfaces results in different expressions when minimized with respect to volume or with respect to contact angle. This phenomenon correlates with the probability of drops to bounce on smooth surfaces on which they otherwise form a very small contact angle or wet them completely. Theoretically, none of the two minima is stable: the drop should oscillate from one minimum to the other as long as no other force or friction will dissipate the energy. Experimentally, smooth surfaces indeed show drops that bounce on them. In some cases, they bounce after touching the solid surface, and in some cases they bounce from a nanometric air, or vacuum film. The bouncing energy can be stored in the interfaces: liquid-air, liquid-solid, and solid-air. The lack of a single energy minimum prevents a simple convergence of the drop's shape on the solid surface, and supports its bouncing back to the air. Therefore, the lack of a simple minimum described here supports drop bouncing on hydrophilic surfaces such as that reported by Kolinski et al. Our calculation shows that the smaller the surface tension, the bigger the difference between the contact angles calculated based on the two minima. This agrees with experimental finding where reducing the surface tension, for example, by adding surfactants, increases the probability for bouncing of the drops on smooth surfaces.