Testing the empirical relation between ultraviolet color and attenuation of galaxies

We test the empirical relation between ultraviolet color and attenuation as derived for starburst galaxies with a wide assortment of galaxy types detected by the Galaxy Evolution Explorer and find that it systematically overestimates the far-ultraviolet attenuation of our sample by similar to 0.5 mag. Our efforts to find an additional parameter that could improve the starburst reddening relation were unsuccessful. In particular, UV - K-s colors (in non-matching apertures) show no correlation with the offset from the starburst reddening relation, suggesting either that UV - K-s colors are a poor tracer of present to past average star formation history (the b parameter) or that the intrinsic dust distribution/geometry may be responsible for moving galaxies off the correlation. It is possible to reduce the systematic overestimate of A(FUV) by using the linear correlation derived from our sample, which simply lowers the starburst predicted values of A(FUV) by 0.58 mag. The scatter, however, remains large at 0.89 mag.