We compare a numerical dynamo model with a model of the palaeosecular varia
tion and find that, although there is good agreement in the meridional dist
ribution of the secular variation, the amplitude of secular variation in th
e dynamo model is too small. Increasing the Rayleigh number does little to
increase the amplitude of the secular variation, while it increases the axi
al dipole component of the field to geophysically unreasonably large values
. However, by introducing lateral variations in heat flux across the core m
antle boundary into the dynamo model, we are able to match the amplitude of
the secular variation. In addition, we find that the dynamo model then exh
ibits persistently anomalous behaviour beneath the Pacific region.