The morphogenetic properties causing germ-layer spreading and stratification in amphibian gastrulation were called "tissue affinities" by Holtfreter. The differential adhesion hypothesis (DAH) attributes such liquid-like tissue rearrangements to forces generated by intercellular adhesions within and between the migrating cell populations. This theory predicts that, among the primary germ layers, the cohesiveness of deep ectoderm should be the greatest, that of deep mesoderm should be intermediate, and that of deep endoderm should be the least. Also, the cohesiveness of differentiating neural ectoderm should increase after induction, causing it to internalize and segregate from epidermis. The DAH also explains why the cohesiveness of "liquid" tissues, whose cells are free to rearrange, should be measurable as tissue surface tensions. Using a specially designed tissue surface tensiometer, we demonstrate that (i) aggregates of Rana pipiens deep germ layers do possess liquid-like surface tensions, (ii) their surface tension values lie in precisely the sequence necessary to account for germ-layer stratification in vitro and in vivo, and (iii) the surface tension of deep ectoderm just underlain by the archenteron roof is twice that of not-yet-underlain deep ectoderm. These measurements provide direct, quantitative evidence that the "tissue affinities" governing germ-layer flow during early stages of vertebrate morphogenesis are reflected in tissue surface tensions.