If we stop infinite-system DMRG at some superblock size L, we can interpret the final wave function in two ways. We can take it as an approximation to the exact state for the superblock of size L andevaluate expectation values. The accuracy is limited not only by the truncations, but also by the factthat the first truncations were carried out for extremely small superblocks: the choice of relevantshort block states is likely to be a not too good approximation to those one would have chosen forthese short blocks embedded in the final system of length L.