Question: how do you normalize between measuring temperatures with non-calibrated mercury thermometers (as they were taken in previous centuries) and digital thermometers accurate to within a tenth of a degree? Even with a calibrated thermometer, the way you take the temperature changes the reading—for instance, wet bulb vs. dry bulb methods. For historical comparisons, the data set just isn’t there because methodologies have changed.
When the facts don’t fit your story change the facts.