Experimental polymer melting as observed in differential scanning calorimetry (DSC) is generally complicated by the influence of lamellar thickness distributions and/or reorganizational effects. Idealized polymer melting in the absence of such effects is critical both for a basic understanding of the DSC results and for decoupling instrumental from sample size broadening effects. A model system which does not have the mentioned complicating effects was developed, consisting of a polyethylene-indium powder mixture. The melting behavior of this model system was experimentally investigated and simulated by a "shell model" that treats melting in a DSC sample as the melting of a series of shells. All of the parameters in the model are obtained experimentally; none is adjustable. The simulation was generally quite satisfying, although a somewhat greater deviation was observed for extreme heating rates (1 degrees C min(-1) and 40 degrees C min(-1)). In addition to the "shell model" simulation, a single Gray type melting peak was also found useful in approximating the experimental curve.