In the present chapter, we describe results of experimental investigations and theoretical analysis of phase selection and nucleation of pores in small samples of undercooled diopside liquid when it is enclosed by a solid crystalline surface layer. The formation of the surface crystalline layer starts with nucleation and growth of highly dense diopside crystals. At the moment of impingement of these crystals on the sample surface, the crystallization pathway switches from diopside to a wollastonite-like (WL) phase. The origin of such switch can be explained by the fact that the formation of the WL-crystal produces less elastic stress energy than the same amount of diopside. This difference is due to the lower (as compared to diopside) density of the WL-crystal phase, which is closer to the liquid density. The relative content of the two crystalline phases can be changed by varying the sample size. Due to the density misfit the growth of the WL-crystalline layer leads to uniform stretching of the encapsulated liquid. This negative pressure leads finally to the formation of one small pore, which rapidly grows up to a size that almost eliminates the elastic stress and, therefore, dramatically reduces the driving force for further pore nucleation. The nucleation process of the pore is experimentally found to occur in a very narrow range of the relative widths of the surface layer (compared to the sample size) and, consequently, of negative pressures. We consider this fact as an indication that pore nucleation proceeds via homogeneous nucleation. The above-given qualitative explanation of the observed phenomena is corroborated by detailed theoretical calculations of elastic stress fields and their impact on phase selection and pore nucleation. Good qualitative and partly even quantitative agreement between experiment and theory is found. An overview on other systems with similar or related properties is included as well. The findings of this research are quite general because the densities of most glasses significantly differ from those of their iso-chemical crystals. By this reason, the studied phenomena are of high technological significance for the development of different types of glass-ceramic materials and the understanding and control of sinter-crystallization processes. The latter problem is also considered in the present chapter.