Puzzlement about thermal redshift

Discussed is the classical theoretical description of the experimentally established thermal redshift of spectral lines. Straightforward calculation of the observable spectrum from a canonical ensamble of monochromatic radiators yileds overall blueshift rather than redshift. It is concluded that the customary explanation of the thermal redshift as a second order Doppler effect does not bear closer examination, and that in fact, the phenomenon ''thermal redshift'' is not yet fully uderstood in classical terms.