New paper shows maximum daily temperatures were higher in the 1930's than in 2010

A paper under discussion for Climate of the Past plots global warmest daily maximum temperatures from the HADCRU global temperature dataset and shows maximum daily temperatures were higher in the low-CO2 1930's than at the end of the record in 2010. This is contrary to AGW theory, which predicts more extreme daily maximum temperatures to occur after 1950. 

TXx is defined in Table 1 above as the warmest daily maximum temperatures in the HADCRU dataset, and plotted below.

Clim. Past Discuss., 10, 2105-2161, 2014 www.clim-past-discuss.net/10/2105/2014/ doi:10.5194/cpd-10-2105-2014 Investigating uncertainties in global gridded datasets of climate extremes

R. J. H. Dunn¹, M. G. Donat², and L. V. Alexander^2
1Met Office Hadley Centre, FitzRoy Road, Exeter, EX1 3PB, UK
²ARC Centre of Excellence for Climate System Science and Climate Change Research Centre, UNSW, Sydney, NSW 2052, Australia
Abstract. We assess the effects of different methodological choices made during the construction of gridded datasets of climate extremes, focusing primarily on HadEX2. Using global timeseries of the indices and their coverage, as well as uncertainty maps, we show that the choices which have the greatest effect are those relating to the station network used or which drastically change the values for individual grid boxes. The latter are most affected by the number of stations required in or around a grid box and the gridding method used. Most parametric changes have a small impact, on global and on grid box scales, whereas structural changes to the methods or input station networks may have large effects. On grid box scales, trends in temperature indices are very robust to most choices, especially in areas which have high station density (e.g. North America, Europe and Asia). Precipitation trends, being less spatially coherent, can be more susceptible to methodological changes, but are still clear in regions of high station density. Regional trends from all indices derived from areas with few stations should be treated with care. On a global scale, the linear trends over 1951–2010 from almost all choices fall within the statistical range of trends from HadEX2. This demonstrates the robust nature of HadEX2 and related datasets to choices in the creation method.