In this paper independent evidence in favour of the hypothesis proposed by Thakur and Sapre (1979) that a QSO consists of a bright central object embedded in an extended nebulosity has been presented. αU−B and αB−V, the spectral indices in (U-B) and (B-V) colours, have been calculated for a sample of 80 QSOs with redshiftz≧0.76. In Figure 1 Δα = (αB−V − αU−V) has been plotted against 〈α〉 = (αB−V + αU−B)/2. In this figure the QSOs in which detectability of neubulosities has been predicted by Thakur and Sapre (1979) occupy a separate but adjacent part of the diagram as compared to those for which such a prediction has not been made. In Figure 2 (U-B)0 has been plotted against (B-V)0 for the same sample of 80 QSOs, where (U-B)0 and (B-V)0 are the (U-B) and (B-V) colours corrected for galactic extinction. In this plot also, a similar separation of the two classes of objects is discernible. Another empirical criterion—namely, (B-V)0≧0.32 for the detectability of nebulosities around QSOs—has been suggested.