On a two-point boundary-value problem in geophysics