We investigated the inductive signals originating from the vegetal blastomeres of embryos of the sand dollar Peronella japonica, which is the only direct developing echinoid species that forms micromeres. To investigate the inductive signals, three different kinds of experimental embryos were produced: micromere-less embryos, in which all micromeres were removed at the 16-cell stage; chimeric embryos produced by an animal cap (eight mesomeres) recombined with a micromere quartet isolated from a 16-cell stage embryo; and chimeric embryos produced by an animal cap recombined with a macromere-derived layer, the veg1 or veg2 layer, isolated from a 64-cell stage embryo. Novel findings obtained from this study of the development of these embryos are as follows. Micromeres lack signals for endomesoderm specification, but are the origin of a signal establishing the oral-aboral (O-Ab) axis. Some non-micromere blastomeres, as well as micromeres, have the potential to form larval skeletons. Macromere descendants have endomesoderm-inducing potential. Based on these results, we propose the following scenario for the first step in the evolution of direct development in echinoids: micromeres lost the ability to send a signal endomesoderm induction so that the archenteron was formed autonomously by macromere descendants. The micromeres retained the ability to form larval spicules and to establish the O-Ab axis.