A precise knowledge of the wavelength dependence of interstellar extinction (i.e., the absorption and scattering of light by interstellar dust grains) and of any spatial variability in this dependence is important for two distinct reasons. First, the extinction depends on the optical properties of the dust grains along a line-of-sight and potentially can reveal information about the composition and size distribution of the grains. Further, changes in the extinction from place to place may reveal the degree and nature of dust grain processing occurring in the ISM. Second, the wavelength dependence of extinction is required to remove the effects of dust obscuration from observed energy distributions, since most astronomical objects are viewed through at least some small amount of interstellar dust. Spatial variations in the extinction potentially limit the accuracy to which energy distributions can be ``dereddened.'' Such uncertainties might be acceptably small for very lightly reddened objects, but quickly can become debilitating along modestly reddened sightlines.
This paper follows from a talk given at the meeting ``Ultraviolet Astrophysics Beyond the IUE Final Archive'' (Fitzpatrick 1998) and addresses the second point raised above, i.e., the correction of energy distributions for the effects of interstellar extinction. The goals are to provide a summary of what is known about the wavelength dependence and spatial variability of interstellar extinction and to present strategies for the best removal of the effects of extinction from astronomical data. In Section 2, I briefly review the chief features of IR-through-UV extinction and the nature of the known spatial variations. These variations are illustrated with a set of optical/UV extinction curves which span the (currently) known extreme limits of extinction variations. Constraints on the wavelength dependence of extinction and some general correlations between extinction curve shape and interstellar environment are also noted in Section 2. Strategies for dereddening are discussed in Section 3 along with estimates of the appropriate uncertainties which ought to be incorporated into an error analysis. Some final comments regarding extinction along extragalactic sightlines and at wavelengths shortward of 1150 Å are given in Section 4.
The Appendix of this paper describes the construction of a new estimate of the shape of a mean and an environment-dependent IR-through-UV extinction curve. These curves, which may be obtained from the author, are suitable for dereddening multiwavelength spectrophotometric data, such as has become available with the Hubble Telescope's Faint Object Spectrograph, and for deriving extinction relationships for any photometric system.