Didn't see this in the article - the paper is entitled "The projected timing of climate departure from recent variability".<p>Direct link:
<a href="http://www.nature.com/nature/journal/v502/n7470/pdf/nature12540.pdf" rel="nofollow">http://www.nature.com/nature/journal/v502/n7470/pdf/nature12...</a><p>From the methods summary section:<p>They used the projections of seven climate variables: near-surface air temperature, sea surface temperature, precipitation, evaporation, transpiration, surface sensible heat flux, and ocean surface pH.<p>"For each model and variable, we used the period 1860-2005 from the historical experiment (the longest time span common to all models), to establish the historical bounds of climate variability. The projections...[were used] to simulate the period 2006-2100...to identify the year at which mean annual values of a given variable would exceed historical bounds."<p>"In total, for all variables and experiments, we processed 89,712 years of data comprising 1,076,544 monthly global maps, interpolated to an equal-area grid with a resolution of 100 km."<p>I think the 9 in 1960-2005 must be a typo. Everything is 1860 in the study.