The data is being "corrected" to "take away errors", this is clearly visible as well as explained by Heller in his many videos on the subject. Another issue is that the measuring network has notably shrunk in size in the last decades with the measurements from locations which have been taken out replaced by interpolated data. Heller claims that "uncorrected" historical data shows that the average temperature in the 30's was higher than it is today and that this is the reason for the "corrections" - simply to make the data fit the narrative. Looking at actual historical data seems to corroborate his claim of the records being "corrected", US state and territory temperature extremes [1] also show a tendency to have the highs somewhere in the 30's (23 states) with 1936 as the hottest year in 13 states. These are "uncorrected" temperature records.
[1] https://en.wikipedia.org/wiki/U.S._state_and_territory_tempe...