We investigate the practice of assigning high spin temperatures to damped Lyman α absorption systems (DLAs) not detected in H i 21-cm absorption. In particular, Kanekar & Chengalur have attributed the mix of 21-cm detections and non-detections in low-redshift (zabs≤ 2.04) DLAs to a mix of spin temperatures, while the non-detections at high redshift were attributed to high spin temperatures. Below zabs= 0.9 , where some of the DLA host galaxy morphologies are known, we find that 21-cm absorption is normally detected towards large radio sources when the absorber is known to be associated with a large intermediate (spiral) galaxy. Furthermore, at these redshifts, only one of the six 21-cm non-detections has an optical identification and these DLAs tend to lie along the sight-lines to the largest background radio continuum sources. For these and many of the high-redshift DLAs occulting large radio continua, we therefore expect covering factors of less than the assumed/estimated value of unity. This would have the effect of introducing a range of spin temperatures considerably narrower than the current range of ΔTs≳ 9000 K , while still supporting the hypothesis that the high-redshift DLA sample comprises a larger proportion of compact galaxies than the low-redshift sample.
Monthly Notices of the Royal Astronomical Society, Vol. 356, no. 4 (Feb 2005), pp. 1509-1518