I can't comment on it being a bug or not, but obviously something changed between 10.6 and 10.7. As a work around, can you run your calculations and then go back and select those lines where Bearing is Null and calculate them to 0?
That should just about do it....