Abstract
The electron spin dipole—dipole interaction in O2 is calculated nonempirically by the use of several approximate wavefunctions of the HL and LCAO—MO types. It is pointed out that the experimental value determined by Tinkham and Strandberg can hardly be explained by simple LCAO‐type approximations; the calculated value would agree with the experiment only if the wavefunctions of this molecule were almost completely ionic.