Self-consistency and a generalized principal subspace theorem

J. Multivariate Anal. 133:27-37. Generalizes principal-subspace results using self-consistency, including location-mixture and skew-normal settings. Forward-citation near miss for DS17's beyond-elliptical axis: it broadens the distribution classes supporting subspace restrictions, but still optimizes squared approximation error and contains no profiled Fisher block, endogenous tilt equation, or margin-certified empirical obstruction. Abstract and theorem context screened; primary theorem text remains library-access pending.