In the course of work to develop a new and consistent model of hydrates and hydrate inhibitors we have made some unexpected predictions about the structure of hydrates that appear to be supported by the currently available experimental data. Based on the data and on the accepted theory of thermodynamic stability, we conjecture, that the hydrate formed by certain mixtures is structure II - in a certain composition range - whereas the associated simple hydrates are structure I. This is true for a number of ternary systems such as methane-ethane-water. The structure of nitrogen hydrate has been the subject of some debate. We found that it is impossible to derive a consistent set of model parameters for the nitrogen simple hydrate, assuming nitrogen forms structure II at ambient temperatures. However, structure II will form from a mixture of nitrogen with very small amounts of ethane or propane added. The phase transitions are a natural consequence of the physical factors involved : empty lattice stability; number of cages in each structure; affinity of hydrate former for each cage. The conclusion about which is the stable hydrate structure can be drawn from a fairly simple analysis of the equations and holds equally well for several models published by other authors. Our discovery was, however, a consequence of the application of rigorous and consistent multiphase-split algorithms. Experimental confirmation would give extra support to the current theory and parameters. Precise measurement of the conditions at which the transitions occur would be a very useful contribution to the determination of the correct hydrate model parameters.