The suggestion that a substrate whose architecture matches that of one of the faces of ice will make an effective nucleating agent goes back over 50 years. Over the years a number of experimental tests of this suggestion have been offered together with a variety of theoretical explanations to rationalize the disappointing nucleation abilities of most substances. In our study we have selected what might appear to be an ideal nucleating substrate, namely, BaF2, whose (111) face lattice constants are a near match to the basal face Of I-h ice. Yet thin-film ice on BaF2(111) proves to be unstable. We report a novel infrared spectroscopic probe of thin ice films and present a theoretical analysis to account for the instability of ice on BaF2(111) and similar substrates. We find that the details of the water-surface interaction override the advantage of a close surface-ice lattice match. We summarize the conditions that need to be met for an effective nucleating substrate.