Spin-coupled theory is used to investigate the bonding in several hypercoordinate and "normal octet" compounds of main group elements. It is found that d basis functions play much the same qualitative role in hypercoordinate and normal molecules, acting as polarization functions. There are no obvious demarcations in the energy penalty per bond of excluding such functions. No evidence is found to support the traditional notions of sp(n)d(m) hybridization. The spin-coupled approach, also known as the full-GVB model, provides a very clear and simple picture of the bonding in all of the molecules studied. In SF6, for example, the sulfur atom contributes six equivalent, nonorthogonal sp(x)-like hybrids, which delocalize onto the fluorine atoms. Each of these two-center orbitals overlaps with a distorted F(2p) function, with the perfect-pairing spin function dominating. The spin-coupled description of PFS is entirely analogous, with remarkably little differentiation between axial and equatorial bonds. A key consideration for all of the hypercoordinate species studied is the polarity of the various bonds. It is suggested that less emphasis than hitherto be placed on the "octet rule" and that the so-called democracy principle be asserted : any valence electron can participate in chemical bonding if provided with sufficient energetic incentive. This idea is pursued for phosphorus and sulphur halides, for XeF2, and for the CHS5-, SiH5-, and SiF5- ions. It is argued that there are no significant qualitative differences between the hypercoordinate nature of first-row, second-row, and noble gas atoms in appropriate chemical environments.