All world temperature series may be illegitimate.

According to HadCet, the data has been adjusted to account for urban heat island affects

-- Long Term Temperature Records Contradict GISS Temperature Record, by Mark Fife

Two points:

1. Urban Heat Effect. That does not necessarily mean they adjusted it correctly. It seems most UHE adjustments have been too low. The correct average adjustment for an UHE-station seems to be about 2C. But only at night. I'd like to know by how much they adjust east station.
2. Homogenization. All these temperature data series seem to be homogenized today. Which, I gather, means deleting outliers (from my reading of the Wikipedia entry). In 2018 Lansner & Pedersen published as study "Temperature trends with reduced impact of ocean air temperature". The took all temperature data for the past 110 years. Split stations into 2 groups.
   • OAA : Ocean Air Affected
   • OAS : Ocean Air Sheltered

   OAS stations showed no recent warming. OAS stations in USA showed no warming since the mid-1930s! OAA stations show recent warming. So it looks like all warming is due to ocean air warming the land. This is not really compatible with what we're told about the greenhouse gas effect.

   The killer here is: it looks like stations are currently homogenized based on proximity. homogenization is only theoretically legitimate when one finds a set of similarly located stations which all change for the same reason. In this case it's done by dropping outliers. One should not homogenize OAA and OAS stations. Before accepting any homogenized data one must check that no OAS & OAA homogenizations affected the other type of station.

   Although Lansner & Pedersen don't make this point about homogenization, their results clearly show that homogenizations will be invalid unless differences between OAS and OAA are taken into account (which I doubt they are - otherwise the Wikipedia entry would tell us).