What I was told by Belden (and it makes perfect sense) is that different color wires must have insulation applied by different extruders (where liquid PVC coats bare wire pulled through a small die). The tolerances on the die, and therefore, insulation thickness can vary by .001". Using 8451 as an example, this makes capacitance change by about 5%. However, by grounding a (balanced interface) cable shield only at the source end, most of the CMRR-degrading effects of the capacitance imbalance disappear. My generic seminar (see http://www.jensen-transformers.com/an/generic%20seminar.pdf) explains in detail.
I fail to see how the circuit in Figure 8.19 is "balanced". When one of the signal lines is grounded, the interface is unbalanced by definition. Also, anyone who believes that two conductors in a shielded cable have equal capacitance to the shield is a fool! If the wires are different colors (no joke), their capacitances can easily differ by 5% or more. I've measured this myself on many commercial cables (try Belden 8451, for example). Bill Whitlock, president & chief engineer of Jensen Transformers www.jensen-transformers.com and Fellow of the Audio Engineering Society
It is very good and comprehensive. But it seems to me that we need have a different approach for electrostatic or magnetic problem. Sometimes they are even against each other. The point is what is the scenario of which, electrostatic or magnetic play more dominant role.