Does the Sun shrink with increasing magnetic activity?

We have analyzed the full set of Solar and Heliospheric Observatory (SOHO) Michelson Doppler Imager (MDI) f- and p-mode oscillation frequencies from 1 996 to date in a search for evidence of solar radius evolution during the r ising phase of the current activity cycle. Just as Antia et al. in 2000, we find that a significant fraction of the f-mode frequency changes scale wit h frequency and that if these are interpreted in terms of a radius change, it implies a shrinking Sun. Our inferred rate of shrinkage is about 1.5 km yr(-1), which is somewhat smaller than found by Antia et al. We argue that this rate does not refer to the surface but, rather, to a layer extending r oughly from 4 to 8 Mm beneath the visible surface. The rate of shrinking ma y be accounted for by an increasing radial component of the rms random magn etic field at a rate that depends on its radial distribution. If it were un iform, the required field would be similar to7 kG. However, if it were inwa rdly increasing, then a 1 kG field at 8 Mm would suffice. To assess contrib ution to the solar radius change arising above 4 Mm, we analyzed the p-mode data. The evolution of the p-mode frequencies may be explained by a magnet ic field growing with activity. Our finding here is very similar to that of Goldreich et al. (1991). If the change were isotropic, then a 0.2 kG incre ase, from activity minimum to maximum, is required at the photosphere, whic h would grow to about 1 kG at 1 Mm. If only the radial component of the fie ld were to increase, then the requirement for the photospheric field increa se is reduced to a modest 60-90 G. A relative decrease in temperature of th e order of 10(-3) in the subphotospheric layers, or an equivalent decrease in the turbulent energy, would have a similar effect to the required inward growth of magnetic field change. The implications of the near-surface magn etic field changes depend on the anisotropy of the random magnetic field. I f the field change is predominantly radial, then we infer an additional shr inking at a rate between 1.1 and 1.3 km yr(-1) at the photosphere. If, on t he other hand, the increase is isotropic, we find a competing expansion at a rate of 2.3 km yr(-1). In any case, variations in the Sun's radius in the activity cycle are at the level of 10(-5) or less and, hence, have a negli gible contribution to the irradiance variations.