[xyzzy123]

I don't think the author adequately considers changes to the local environment of the thermometer, changes in the equipment itself (eg replacement, movement) possible changes in recording methodology and so on.

There was no real effort to understand the possible sources of noise or error in the measurements.

[misslibby]

But local changes of environment would have been expected to increase temperature readings? More asphalt, more buildings...

[jacquesm]

Not necessarily, it could just as easily go the other way (thermal sink dropping the average).

[tnvaught]

Huh? https://en.wikipedia.org/wiki/Urban_heat_island

[bobthepanda]

Also, 1920 is a very arbitrary starting year. Why then?

In 1920 the Industrial Revolution has been going on for over 100 years, the US petroleum industry was already big enough to have been affected by antitrust action, and Henry Ford's assembly line has been enabling mass motorization for the better part of a decade.

[BoumTAC]

1920 is 100 years ago, so I don't think it's an arbitrary starting year. Also the older you start, the worst the data quality will be.