The diffusibility of the vegetalizing factor was examined by a transfilter culture using an ethanol-fixed swimbladder of the crucian carp (Carassius auratus) as the inductor and presumptive ectoderm from gastrulae of Cynops pyrrhogaster as the responding tissue. Nucleopore filters, about 12-14 pm thick, with nominal pore sizes of 0.05, 0.1, 0.6, 0.8, 3.0 and 8.0 pm were interposed between the interacting tissues. The responding pieces of ectoderm were removed from the assemblies after contact for 0.5, 1, 3, or 24 hr and cultured in Holtfreter's solution for 10 days at 20°C.The inductions observed were almost entirely mesodermal, although masses of endoderm-like yolky cells were seen in explants and neural tissues in a few cases. Filter membranes with pores of 0.05 to 8.0 pm did not interfere with the vegetalizing effect.Under an electron microscope, small cytoplasmic cones of the responding cells of the presumptive ectoderm were observed in the pores of the interposed filter after 3 hr's contact. The cones grew longer as the cultivation time increased, but even after 24 hr there was no contact between the interacting tissues. Since 3 hr's contact between the interacting tissues was sufficient to cause full vegetalization on the transfilter culture with the swimbladder, the formation of the cytoplasmic outgrowths had no significance in the induction.The transfilter method, which was first developed by GROBSTEIN (5, 6 ) in pioneer work on the mechanisms of development of the salivary gland, has been employed in experiments on various organs (1, 4, 19, 24, 25). The method was devised to test whether cell-to-cell contact between interacting tissues in induction systems is necessary for induction. Experiments have shown that direct contact between interacting tissues is not usually essential, although induction of the kidney tubule by the spinal cord may require close cellular interaction (22).Just before development of the neural plate in amphibian embryos, the notochordal anlage and the presumptive neuro-ectoderm come into contact, suggesting that cell contact may be significant in primary induction (23). This possibility has been tested using several methods which prevented direct contact between competent ectoderm and inductive materials (2, 3, 7, 9). NIU and TWITTY (13) also examined the effects of treating small pieces of presumptive ectoderm with a medium conditioned by previous cultivation of axial mesoderm. In some of these earlier experiments, no inductions occurred (3, 7) and in others, the results were either questionable or only neural cells without organized tissue were induced (2, 13). However, applying a modification of GROBSTEIN'S technique in studies on primary induction in amphibian embryos (14, 17, 20), induction of the brain was observed using a dorsal blastopore lip as the inductor. TOIVONEN et al. (20) employed a new type of filter membrane, the Nucleopore
