We revisit the commonly accepted notion that the difference between interaction and reaction cross sections is negligible at relativistic energies, and show that, especially in small mass number region, it is large enough to help probe nuclear structure. For analyses of the difference, we construct "pseudo data" for the reaction cross sections using a phenomenological black-sphere model of nuclei since empirical data are very limited at high energies. The comparison with the empirical interaction cross sections suggests a significant difference between the reaction and interaction cross sections for stable projectiles on a carbon target, which is of the order of 0-100 mb.

5 pages, 5 figures