Abstract
The CMB anisotropy at ℓ ≲ 60 seems to have some special features which include (i) a dipole modulation and (ii) a decrease in power. It is known that both of these effects can be generated if a curvaton-type field has a super-horizon perturbation. It is also known that this will generate non-gaussianity fNL in the same range of ℓ, whose magnitude has a lower bound coming from the magnitude of the observed CMB quadrupole. I revisit that bound in the present paper, and point out that it may or may not be compatible with current data which should therefore be re-analysed.