Expansion history and f(R) modified gravity

We attempt to fit cosmological data using $f(R)$ modified Lagrangians containing inverse powers of the Ricci scalar varied with respect to the metric. While we can fit the supernova data well, we confirm the $a\propto t^{1/2}$ behaviour at medium to high redshifts reported elsewhere and argue that the easiest way to show that this class of models are inconsistent with the data is by considering the thickness of the last scattering surface. For the best fit parameters to the supernova data, the simplest 1/R model gives rise to a last scattering surface of thickness $\Delta z\sim 530$, inconsistent with observations.