THE diffuse interstellar bands, which have been known for about 36 years1, are still one of the mysteries of astronomy. One hope has been that with definite identification of these bands will come not only further information about interstellar matter and its condition but, more specifically, more information about the nature of interstellar grains. Even without identification, however, there have been properties of the lines which could conceivably carry information about their origin. When van de Hulst2,3 considered the question of the origin of the diffuse lines from the point of view of the optics of small particles containing absorptive impurities, he noted that the shape of the extinction curve in the neighbourhood of the absorption lines could be one of various possibilities, depending on the size of the particles. Although this result was based on an approximate low index of refraction theory it turned out to be quite correct—at least qualitatively—as confirmed by detailed Mie theory computations4–7. As a consequence of these extensive calculations it was conclusively shown that if the 4430 band (the strongest of the diffuse bands) is symmetric there can be no simple accounting for these bands as originating from absorbing atoms or molecules dispersed through the interior of the grain, because the effective grain sizes needed to account for the shape of the extinction curve are such as to produce highly asymmetric additional extinction about the average. But new observational data on extinction have been reinterpreted8,9 and it now seems that the 4430 band is probably highly asymmetric and indeed may actually consist of a lower extinction at the shorter wavelengths followed by an additional extinction at the longer wavelengths, the whole being similar to a dispersion curve as predicted for the small dielectric or dirty ice particles4 which give rise to the interstellar extinction. We have followed up this idea and have extended the calculations to polarization10 as well as extinction in the 4430 region produced by several of the proposed interstellar grain models including silicates and graphite. In so doing we have found that the polarization may be the more useful and definitive way of getting information about the grains11.