Looking for a varying $\alpha$ in the Cosmic Microwave Background

We perform a likelihood analysis of the recently released BOOMERanG and MAXIMA data, allowing for the possibility of a time-varying fine-structure constant. We find that in general this data prefers a value of $\alpha$ that was smaller in the past (which is in agreement with measurements of $\alpha$ from quasar observations). However, there are some interesting degeneracies in the problem which imply that strong statements about $\alpha$ can not be made using this method until independent accurate determinations of $\Omega_b h^2$ and $H_0$ are available. We also show that a preferred lower value of $\alpha$ comes mainly from the data points around the first Doppler peak, whereas the main effect of the high-$\ell$ data points is to increase the preferred value for $\Omega_b h^2$ (while also tightening the constraints on $\Omega_0$ and $H_0$). We comment on some implications of our results.