The procedure developed in Smith (1974) to model the radiation source for type 11I bursts is modified to include scattering of radiation in the source itself. Since the inhomogeneities in the source must have the same statistical properties as the inhomogeneities used in tracing radiation from the source to the observer, these two parts of the type II1 problem are no longer uncoupled. Thus we use inhomogeneities consistent with the scattering inhomogeneities of Steinberg et aL (1971) and Riddle (1974) and apply the procedure to an archetype 'fundamental-harmonic' pair observed at Culgoora on 28 September, 1973 at 0319 UT. We find that it is impossible to model this burst with a source which is homogeneous in the sense that every part of the source has the same energy density in plasma waves. The density inhomogeneities in the source severely hamper amplification of the supposed fundamental. Possible ways out of this dilemma are discussed, including second harmonic pairs and a source with an inhomogeneous distribution of plasma waves. It is concluded that none of the possibilities are completely satisfactory to explain present observations and suggested that critical observations are missing.