In the context of a cold big bang (CBB) cosmological model, I estimate the temperature, calculate the evolution, and discuss the anisotropies of a homogeneous radiation background emitted at high redshift by Population III objects and thermalized by a mixture of carbon/silicate dust and iron or carbon whiskers. Assuming that Population III objects supply dark matter remnants and produce the universal helium abundance of ~24%, the resulting radiation, if thermal, should have temperature 0.5 K<~T0<~9 K. To calculate its thermalization I limit standard dust density using constraints from maximum allowed metallicity, quasar reddening, Type Ia supernova observations, and high-redshift object visibility and assume a small mass ratio of whiskers to standard dust. For high redshift of generation (zi>~100) and highly conducting, high aspect ratio (length/diameter >~1000) iron whiskers, the final spectrum meets the FIRAS spectral distortion limits and does not exceed limits on IR-background light. Energetic considerations probably require zi<~40Ωb if the observed cosmic microwave background (CMB) is star generated. Such a model is marginally viable with baryon density Ωb=1, but probably ruled out if Ωb<~0.5. I comment briefly on possible other thermalization mechanisms. In whisker-thermalized models, CMB anisotropies are imprinted at lower redshift and through somewhat different physical processes than in the hot big bang (HBB). Nonlocal thermal equilibrium of the dust causes the Sachs-Wolfe effect to disappear on small angular scales. Radiation pressure probably ejects whiskers from luminous regions such as galaxies. If significant intergalactic magnetic fields are absent, CMB radiation drag probably fixes whiskers in the CMB frame, hence large-scale Doppler anisotropies (other than the dipole) do not exist. ``Intrinsic'' temperature fluctuations and the integrated Sachs-Wolfe effect operate essentially as in the HBB. Numerical results for the horizon, diffusion, and other length scales relevant to CMB anisotropies for fiducial thermalization models are provided. Finally, I note that CMB polarization may be higher in the whisker model than in the HBB and that acoustic oscillations should not appear in the CBB power spectrum, and I comment on the amplitude and possible frequency dependence of the anisotropies.