Andrews concludes: The bottom lines that emerge from all this are: 1. The only published temperature time series that are not distorted by adjustments and which can therefore be considered reasonably reliable are the UAH TLT series and the SST series after about 1950. (The various radiosonde series, which go back to 1958, also appear to be generally reliable, and more use could be made of them.) 2. Climate models show too much air temperature warming at the surface and in the troposphere. They show probably about the right amount of sea surface warming after 1950 but probably too little...