1. Fragments of epiblast, or of epiblast together with endoderm, were placed in holes cut in the primitive streaks of young chick embryos, and the embryos then cultivated in vitro for about 24 hours. The grafts were sometimes extruded, but in many cases they healed more or less well. In those cases in which partial healing occurred the grafted presumptive ectoderm has usually developed into mesodermal structures, such as somites, though in a few cases it has been induced to form neural tissue. In some cases healing was so complete that no trace of the graft can be found, and in these embryos also the greater part of the presumptive ectoderm must have formed mesoderm. 2. The conversion of presumptive ectoderm into mesoderm has only been accomplished by the method of the above experiments, which involves the performance by the graft of the normal gastrulation movements. In the Amphibia the performance of such movements is unnecessary, and it is suggested that in the chick the importance of the movements is not that they are a part of the gastrulation process but merely that they involve the breakdown of the epithelial structure of the grafted ectoderm. 3. The relevance of these results to the germ-layer theory is discussed, and it is pointed out that, although it is clear that there is no fundamental physiological distinction between the ectoderm and mesoderm before invagination, it is still possible that such a distinction may exist between the endoderm and the ectoderm-mesoderm system.