Statistics were never my strong point and I cannot debate this issue with the depth of argument put forward by Richard Probert-Jones (October 2012: p. 278) and others. Nevertheless I was not convinced by the undoubtedly carefully-worded comment in the ‘Weather news’ section in the same issue concerning the newly-published Met Office 30-year averages for 1981–2010. Comparing these with the 1961–1990 averages, the writer states that (the figures) also suggest that the annual mean temperature is now around 0.5 degC higher than during the 1961–1990 period (and) the rainfall and sunshine amounts have both increased slightly compared to [sic] the previous averages’. It seems to me that it is unsound, if not actually wrong, to compare two 30-year periods which overlap by ten years. This means that 10 years of the data (1981–1990) are used in both sets of averages. Surely one can only sensibly compare two 30-year periods which each use entirely separate sets of data e.g. 1951–1980 compared with 1981–2010 or, indeed, any sequential or separated 30-year period such as 1921–1950 compared with 1971–2000 etc. The writer goes on to state that these figures are consistent with a global warming trend. It is my view that comparing just two sets of any figures cannot demonstrate a ‘trend’. Surely one requires at least three sets of data to even begin to hint at a ‘trend’.

(Editor's note: the comparisons were made by the Met Office – see–2010-averages)