Srednicki's QFT book uses scalar [tex]\phi^3 [/tex] as an example of a QFT. To make a calculation to all orders, Srednicki claims (chpt 19) you calculate all 1P1 graphs for 2 external lines, giving the self-energy. He then says you calculate all the 1p1 graphs for 3 external lines, giving the 3-point vertex function. Then you calculate n>3 vertex functions by drawing 1p1 graphs, but using the 3-point vertex for vertices and exact propagators for propagators. Then any amplitude can be calculated by tree level graphs using these n-point vertices and exact propagators. I think this is mostly correct, but doesn't Srednicki need to say that you have to calculate the 3-point vertex function using the exact propagator? If not, you'd be missing a lot of diagrams that can contribute to a scattering process.