Some years ago, Bolton, Peterson, Wills and Wills (1976, BPWW) reported a statistically-significant excess of QSOs within 2' arc of flat spectrum radio-selected QSOs (specifically, they found 5 such objects while only 1 was expected by chance). None of the QSO pairs had the same redshifts so the result could be regarded as evidence for non-cosmological redshifts, non-uniform QSO distributions, or a statistical fluctuation. Because of the potential importance of BPWW's result, we later used the same technique to examine an independent sample of about 150 QSOs with flat radio spectra and found results that are consistent with the known surface density of QSOs. If we assume a surface density of 3.3 QSOs per square degree brighter than B = 19 (e.g. Marshall et al. 1983), the 150 fields each of 2' arc radius should contain 1.7 random QSOs above this limit and we found 2 (e.g. Wills 1978). A third object is very close to B = 19, and another field contains an object that is extremely blue, with B < 19, on the Palomar Sky Survey, but below our plate limit (i.e. B > 21); we have so far been unable to obtain a spectrum for it. One of the 3 confirmed QSOs is 119″ arc from the radio-emitting one, so rather small changes in the magnitude limit, search radius and choice of epoch at which the magnitudes are measured can result in there being anywhere between 1 and 4 secondary QSOs (or 1–3 if the variable object is not a QSO), compared with 2 expected by chance, and perhaps 10 that would be predicted by BPWW's results.