This is all getting rather interesting. Some very numerate people have been looking at Phil Jones' claim about statistical significance in the temperature records and the consensus seems to be that Jones has got it wrong.
First out of the blocks was Doug Keenan, who noted in the comments here that using the methodology described in Jones' IPCC chapter and data to the end of 2010, the confidence interval for the temperature trend still covered zero.
Meanwhile, Jeff Id wondered whether monthly data might produce a different result and Lucia looked at a slightly different version of the data - HADCRU3V. Their conclusion was that if you used the alternative dataset and used monthly data, the warming was indeed signficant and there was much tut-tutting over the failure to note these changes.
And lastly, with something of the feel of a coup-de-grace, Doug Keenan noted that Jones has previously inveighed against the use of HADCRUT3V for these sorts of calculations.
It will be interesting to see if we get some sort of a correction.