Quenched Chiral Logarithms

I develop a diagrammatic method for calculating chiral logarithms in the quenched approximation. While not rigorous, the method is based on physically reasonable assumptions, which can be tested by numerical simulations. The main results are that, at leading order in the chiral expansion, (a) there are no chiral logarithms in quenched $f_\pi$, for $m_u=m_d$; (b) the chiral logarithms in $B_K$ and related kaon B-parameters are, for $m_d=m_s$, the same in the quenched approximation as in the full theory; (c) for $m_\pi$ and the condensate, there are extra chiral logarithms due to loops containing the $\eta'$, which lead to a peculiar non-analytic dependence of these quantities on the bare quark mass. Following the work of Gasser and Leutwyler, I discuss how there is a predictable finite volume dependence associated with each chiral logarithm. I compare the resulting predictions with numerical results: for most quantities the expected volume dependence is smaller than the errors, but for $B_V$ and $B_A$ there is an observed dependence which is consistent with the predictions.