This is a topic we have covered here a lot – downward revisions to temperatures decades ago that increase the apparent 20th century warming. Here is a great example of this from the GISS for Reykjavik, Iceland. The GISS has revised downwards early 20th century temperatures by as much as 2C, despite Iceland’s Met office crying foul. It is unclear exactly what justification is being used to adjust the raw data. Valid reasons include adjustments for changes in the time-of-day of the reading, changes to the instrument’s location or type, and urbanization effects. It is virtually impossible to imagine changes in the first two categories that would be on the order of magnitude of 2C, and urbanization adjustments would have the opposite sign (e.g. make older readings warmer to match current urban-warming-biased readings).
Arctic stations like these are particularly important to the global metrics because the GISS extrapolates the temperature of the entire Arctic from just a few thermometers. Changes to one reading at a station like Reykjavik could change the GISS extrapolated temperatures for hundreds of thousands of square miles.