Of course, what Trevor said about the radius is accurate, but I'm going to try to embellish it, a bit. If you were to assume that the corner between the journal and the web was absolutely square, then there is an infinitely (theoretically speaking) small area over which bending forces can be distributed. With a radius, small though it may be, you add an infinite number of points (again, theoretically speaking) over which those same forces can be distributed. You've, of course, seen or used the technique of drilling a small hole at the end of a crack in plexiglass to stop the crack from growing. This is similar.
Real mechanical engineers here will correct me if I'm wrong, I'm sure.
Richard L.