You clearly did not read the full paper by Capers Jones and understand the context of his study. Capers Jones is the world’s leading expert on software metrics and is well-respected in the IT industry.
Table 16 in Appendix B represents the synthesis of data sources from numerous client studies collected over the years and “normalized” to give you a balanced view of the state of the art. The statistical methods applied to the data produced the clustering effect that you see.
You may question the methodology, but the fact is, Capers Jones has collected the largest and most reliable dataset available for this kind of research. If you can find a better dataset or a better study, I’d like to see it.