On Christmas Day 2008, I received an e-mail from the German researcher Ernst-Georg Beck. He asked me to analyze an atmospheric CO2 time-series from 1820 to 1960. A CO2 time-series from 1820 was something special. Next day I went to work, analyzed the time-series, wrote a short note and e-mailed back a comment. It was all done in an hour. I wrote that there was nothing special about this time-series. The CO2 variations coincided with North Atlantic water surface temperature from 1900 to 1960 (Yndestad et al. 2008) .
In August 2009 he came from Germany and visited me in Ålesund, Norway. He wanted to
know more about the CO2 time-series signature. In a golden moment, the wavelet spectrum
analysis revealed a coincidence between the CO2 signature, the Atlantic Water temperature, and the lunar nodal tide spectrum (Yndestad et al. 2008) . Beck found what he was looking for. After the CO2 signature was estimated, I started to ask him about possible errors in the timeseries. Slowly I began to understand, the magnitude of the job he had done. This was a life’s work, which should have been presented as a PhD thesis.
Summary of the work
The estimated Marin Boundary Layer (MBL) CO2 time-series was derived from around 90,000 measurements associated with meteorological data selected from a basis of around 200,000 recorded measurements. The data had been presented in 979 technical papers. Among the authors were two Nobel Laureates. The standard analytical procedures of the time had been employed, including intercalibration between some laboratories. The uncertainty was quite large the first decades but narrowed over time to a level within 3 % (Beck 2007) . For measurements over time in the same laboratory, the relative uncertainty would be reduced. That should give adequate comparisons with modern data including a fairly good overlap with the Mauna Loa data from 1958 the last 5 years. Thus, the data set should be considered at least partly adequate which would justify further detailed studies by the international scientific community. After all, the resolution of these in situ data is far beyond the ice core data being widely accepted.