Estimating variation in bony characters and a comment on the Kluge-Kerfoot effect