All equilibrium model simulations show a warming in the mid-troposphere and cooling in the stratosphere. It has been suggested that this contrast in trends between the troposphere and stratosphere might prove a useful detection fingerprint (Karoly, 1987, 1989). In addition, high signal-to-noise ratios have been obtained (Barnett & Schlesinger, 1987) for free tropospheric temperatures. However, stratospheric cooling may not solely be attributed to greenhouse forcing, and can arise due to volcanic pollution injections and ozone depletion. Model simulations are also inconsistent with recent observations (Angell, 1988, and section 220.127.116.11), in particular at the level at which warming reverses to cooling. Perhaps most significantly of all, though, is the lack of reliable instrumental data. The instrumental record for upper atmosphere temperatures extends back only to the late 1950s with the use of radiosonde, whilst satellite-based data is even younger.