I think that the MAC in the comment above was being confused with lift distribution (dependent on aero and geom. twist, etc.). The MAC and its position are purely geometrical, as Howard's diagram indicates, but we have to keep in mind that in locating a.c.'s and thus cg's through MAC analysis, we are making the great simplification of assuming that the lift is the same at each infinitessimal point of the wing. This makes the computation of the aerodynamic center of the wing the same as finding the center of mass of a plate of constant thickness and density and the same size and shape as the wing, a significant over-simplification. There are aspects of the 3-D wing and its flow that alter this, but some apparently balance others. So we just have to get used to the variances and get on with the design job, being aware of trends. %-thickness of wings also marginally affects the approximately quarter-chord position of the a.c. The best reason to use MAC's, IMO, is to be able to predict the a.c. movement with varied sweep and taper. It seems to work pretty well, but I'd guess that the MAC (Edit 3: Oops, I meant a.c. or old center of lift) of each panel is further in toward the wing root than computed.
I don't think use of the diagram is always safe, because small inaccuracies in line positioning on paper can lead to large errors in where the lines cross. Also the a.c. of an elliptical wing of any sweep or aspect ratio is easier to compute than that of a straight tapered wing, but very difficult to approximate with diagrams. It's actually inside that of a straight-tapered wing of the same aspect ratio. This MAC's spanwise position is also listed wrong on several internet sites.
Edit: I did not address Howard's last computation, but finding the spanwise lift center is a good way of positioning the wing laterally in the fuselage, as long as the thrust line then passes through the center of mass.
Edit 2: ...or, better, of positioning the fuselage along the span, of course.
FWIW.
SK