On the size of the genus of a division algebra