4/4 The year 1914 is characterised by a strong El-Nino (the uptick is seen consistently in land, SST and NMAT temperatures), the start of WWI and a sudden change in data sources (figure 2 below). Whether any of those events is responsible for the NMAT changes during this time, remains to be seen.

2/4 The result is puzzling (figure 1 below, ClassNMAT seen in panel a and b): GMST based on NMATs is indeed *very* cold in the 1900s, following the SST-based GMST time series up until 1913. Yet, from 1914 onwards, the NMAT-based reconstruction is close(r) to the LSAT-based reconstruction...