Now this is how to carry out an argument. Are the CRU data “suspect”? An objective assessment. Take the publicly available raw data set, chose two arbitrary subsets and see how the CRU data compares with it. Answer: the CRU adjustments (for entry errors, changes in recording station practice, outliers) fit the raw data subsets, showing that the CRU adjustments have created a representative sample of the raw data.

To be fair, the CRU probably did this during its work, but the fact that anyone who understands the mathematics can do it as well is critical.