ON GRAPHITE AND THE VARIATIONS IN THE ULTRAVIOLET EXTINCTION CURVE

A silicate-graphite model of interstellar extinction has been compared with recent observational information about the ultraviolet part of the extinction curve. The model has large (> 0.01-mu-m) and small (0.005-mu-m) grains of each composition, and the number densities of the four components are assumed to vary independently and randomly within certain limits. A total of 50 extinction curves have been generated and subjected to analysis similar to that for the ultraviolet observations. By fitting a Lorentzian profile to the 4.6-mu-m-1 graphite hump, we show that the model produces no correlation between the central wavenumber and the profile width. This is contrary to often stated expectations for any graphite model, but in agreement with observations. Several observed correlations in the ultraviolet extinction curves can also be understood within the context of this model. However, the model predicts a correlation between the linear background extinction and the far-ultraviolet extinction which appears inconsistent with the observations. It is suggested that the presence of a nongraphitic mantle on the larger silicate grains may improve the model in this respect.