Naturally occurring nanoscale features are all around us. Before we can manipulate them, we first have to develop a basic knowledge of how they are produced in nature, and why they are stable. We have been studying natural minerals in the hematite-ilmeni te (Fe2O3-FeTiO3) system, because they exhibit unexpectedly strong and stable remanent magnetization. Through TEM studies we have shown minerals in this system contain exsolution lamellae down to ~single unit cell thickness ~ 1-2nm (McEnroe et al., 2001 , 2002: Robinson et al., 2002, 2004a). With detailed analyses and mapping chemical elements, we have shown stranded metastable diffusion profiles are common features at moderate temperatures and that these dictate the sizes and patterns of subsequently e xsolved nanoscale lamellae. A surprising result has been the apparent magnetic stability of the nanoscale lamellae. These samples, from Proterozoic rocks, have retained a magnetic memory, for far longer than any hard drive that will ever be needed, near ly 1 billion years. Magnetic measurements show that the medium destructive fields needed are 75mTeslas minimum and commonly >100mTeslas, above that of standard demagnetizing equipment, and far above the Earth's magnetic field of 0.05mT. This extraordina rily high but variable magnetic stability is partially a function of which phase is host and which is lamellae, and what is the state of internal stress. Thermal stability is ~ 600oC or greater, depending on composition. With these properties, materials made out of these compositions could have extremely hard memories, and be resistant to high temperatures, making such material very attractive for magnetic storage of data for space and security applications. We propose to combine magnetic-property measu rements, nanoscale chemistry, atomic simulations, phase relations and thermodynamics of the mineral system, to try to understand why these nanophases are so stable and about the processes that formed them.