Report 1 considered across-the-board ratio adjustment of area census counts using a known population total T; Report 2 considered the same adjustment using an estimator, T, of T. In this report suppose that, once again, we have areas (typically, states) indexed i=l, ..., I. Suppose, also, that we categorize each person in a second manner, e.g., according to J age-race-sex cells as in our empirical results below. We number the cells from j=l to J. Let Tj be the true total number of persons, summed over the I areas, that fall into cell j. In the manner of Report 1 suppose that we know Tj, from a PES or similar source, for each of our cells j.