In laboratory experiments, spherical 1 mm-wide glass and basalt particles are heated, and the hot particles collide at about 1 m/s with a flat glass target that is at room temperature. When the particles are heated below 900 K, the collisions are essentially elastic with coefficients of restitution of about 0.9, but above 900 K collisions become increasingly inelastic and the coefficient of restitution decreases with increasing temperature. At 1100 K the glass particles approach sticking but, simultaneously, at the same temperature the particles melt on timescales of minutes. The basalt particles approach sticking at 1200 K. Only above 1400 K do basalt grains in contact with each other fuse together, forming compounds on timescales of hours, and at 1500 K basalt grains completely fuse together. Therefore, cooling basalt grains only have a 100 K window for compound formation, and velocities very likely have to be below 1 m/s for sticking in the first place. We predict that this puts constraints on compound chondrule formation and particle densities in the solar nebula.