No label defined (Q4354292)

A necessary and sufficient criterion for linear rectifiability of measures is given in terms of the average density introduced by Bedford and Fisher. More precisely, it is shown that a measure in a Euclidean space is linearly rectifiable if (and only if) the lower one-density is positive and finite and agrees with the lower average one-density almost everywhere. The analogous statement for upper densities fails to be true. The proof uses tangent measures as introduced by Preiss and comes in two steps. In the first step the hypothesis on the densities is used to construct, by means of probability distributions on the space of tangent measures (tangent measure distributions), at almost every point a flat tangent measure with the property that the mass of the unit ball equals the lower one-density. In the second step it is shown that such a tangent measure can only exist if the measure is linearly rectifiable.