The Met Office has released the source code for analysis of the land surface temperature data.
It consists of two Perl scripts: station_gridder.perl and make_global_average_ts_ascii.perl.
The first one produces the gridded output for the temperature anomalies across the globe by year since 1850. The second takes the output of the first and produces a text file that can shows the global temperature anomaly since 1850.
I've run them both on the same data as my program (and produced an annual average version of the output of my script). Here's a graph that shows their trend (in red) and my trend (in green). Nice, I'm getting essentially the same results as they are.
Note. The output of station_gridder.perl is similar to the current CRUTEM3 file, but not the same. There are small differences.
Here's a graph that compares CRUTEM3 (in green) and the output of station_gridder.perl (in red) on the released data.
And finally a plot of all three data munches: CRUTEM3 (in green), station_gridder.perl (in red) and me (in blue):