Just a couple of points I thought I'd throw in. First off, the pertinent law is the conservation of energy, not mass. As gravitational potential energy is converted into kinetic energy, and the longer any given water molecule has been accelerated the faster it must be falling, then on average a molecule nearer the sink than the tap will be traveling faster than one nearer the tap than the sink. If you take any given area, say a cubic centimetre, the molecules at the top of that area also must be moving slower than those at the bottom, so the water is kind of being stretched out vertically. This in itself does not necessitate that the cross-sectional area must decrease. If this were the only concern, the cross-sectional areas may be the same at the top as the bottom, but the density would be greater at the top. What makes the cross-sectional area decrease, rather than the density, is (on a fundamental level) the intermolecular interactions between the water molecules. For the density to decrease, the average distance between molecules has to to increase, meaning an increase in potential energy. This requires an increase in temperature, but no heat is being transferred to the water (bar the negligible friction with this air, and not even that in a vacuum). Therefore the average distance between molecules has to be constant, and so it is the cross-sectional area that has to decrease. This is explained macroscopically by the laws of fluid mechanics, involving field lines and viscosity and other things I can't remember much about.