Unlike the neutral gas density, which remains largely constant over redshifts of 0 < ∼ z < ∼ 5, the star formation density, ψ * , exhibits a strong redshift dependence, increasing from the present day before peaking at a redshift of z ≈ 2.5. Thus, there is a stark contrast between the star formation rate and the abundance of raw material available to fuel it. However, using the ratio of the strength of the H i 21-cm absorption to the total neutral gas column density to quantify the spin temperature, T spin , of the gas, it has recently been shown that 1/T spin may trace ψ * . This would be expected on the grounds that the cloud of gas must be sufficiently cool to collapse under its own gravity. This, however, relies on very limited data and so here we explore the potential of applying the above method to absorbers for which individual column densities are not available (primarily Mg ii absorption systems). By using the mean value as a proxy to the column density of the gas at a given redshift, we do, again, find that 1/T spin (degenerate with the absorber-emitter size ratio) traces ψ * . If confirmed by higher redshift data, this could offer a powerful tool for future surveys for cool gas throughout the Universe with the Square Kilometre Array.