The amount of dust estimated from infrared to sub-millimetre (submm) observations strongly depends on assumptions of different grain sizes, compositions and optical properties. Here we use a simple model of thermal emission from cold silicate/carbon
dust at a range of dust grain temperatures and fit the spectral energy distribution (SED) of the Crab Nebula as a test. This can lower the derived dust mass for the Crab by ~50% and 30-40% for astronomical silicates and amorphous carbon grains compared to recently published values (0.25M_sun -> 0.12M_sun and 0.12M_sun -> 0.072M_sun, respectively), but the implied dust mass can also increase by as much as almost a factor of six (0.25M_sun -> 1.14M_sun and 0.12M_sun -> 0.71M_sun) depending on assumptions regarding the sizes/temperatures of the coldest grains. The latter values are clearly unrealistic due to the expected metal budget, though. Furthermore, we show by a simple numerical experiment that if a cold-dust component does have a grain-temperature distribution, it is almost unavoidable that a two-temperature fit will yield an incorrect dust mass estimate. But we conclude that grain temperatures is not a greater uncertainty than the often poorly constrained emissivities (i.e., material properties) of cosmic dust, although there is clearly a need for improved dust emission models. The greatest complication associated with deriving dust masses still arises in the uncertainty in the dust composition.
We present the properties of the first 250 $mu$m blind sample of nearby galaxies (15 < D < 46 Mpc) containing 42 objects from the Herschel Astrophysical Terahertz Large Area Survey (H-ATLAS). Herschels sensitivity probes the faint end of the dust lum
inosity function for the first time, spanning a range of stellar mass (7.4 < log$_{10}$ M$_{star}$ < 11.3 M$_{odot}$), star formation activity (-11.8 < log$_{10}$ SSFR < -8.9 yr$^{-1}$), gas fraction (3-96 per cent), and colour (0.6 < FUV-Ks < 7.0 mag). The median cold dust temperature is 14.6 K, colder than in the Herschel Reference Survey (18.5 K) and Planck Early Release Compact Source Catalogue 17.7 K. The mean dust-to-stellar mass ratio in our sample is higher than these surveys by factors of 3.7 and 1.8, with a dust mass volume density of (3.7 $pm$ 0.7) x 10$^{5}$ M$_{odot}$ Mpc$^{-3}$. Counter-intuitively, we find that the more dust rich a galaxy, the lower its UV attenuation. Over half of our dust-selected sample are very blue in FUV-Ks colour, with irregular and/or highly flocculent morphology, these galaxies account for only 6 per cent of the samples stellar mass but contain over 35 per cent of the dust mass. They are the most actively star forming galaxies in the sample, with the highest gas fractions and lowest UV attenuation. They also appear to be in an early stage of converting their gas into stars, providing valuable insights into the chemical evolution of young galaxies.
