In Table 3, some minus signs are missing.
All correlation coefficients for the cases A and B are actually negative (–0.108, –0.128, –0.120; –0.019, –0.013, –0.012).
The summary statistics for Anscombe's quartet in Table 4 unfortunately contain stray "1"s and are mostly incorrect.
Means in x are 9.0, variances in x are 11.0 (reported correctly in the table), means in y are 7.5, variances in y are 4.12*, and Pearson CCs are 0.81*.
Thanks to Sampsa Hautaniemi for pointing this out!