To Tha and davenn, consider this. The direction the signal current is flowing does not affect the direction of the noise current. Induced noise is the same direction in both conductors; that's why it is called common mode noise. No cancelling occurs in the wires. The cancellation happens at the receiver. For a balanced signal pair, the signal phases are + and - (or 0 degrees and 180 degrees) while the noise is + and +. It is critical that the noise arrive with equal amplitude and phase on both conductors, and controlled twisting achieves this. When the receiver circuit does an invert-and-add, the signal goes to 2x and the noise goes to zero.
For an unbalanced signal, where the 2nd conductor in the pair is a return, the signal again arrives at the receiver as a difference, this time between the receiver input and its signal reference (gnd). The noise again arrives in phase on both conductors, so when the input goes up due to noise, the input reference goes up also. The receiver sees signal at 1x and noise at zero. This works only if there is no other connection between the transmitter GND and the receiver GND. If there is, the receiver input moves due to noise but the receiver GND input does not. This is why ground loops "cause" noise. They don't. They allow the existing noise *not* to be cancelled. Close enough.
"
So this wire to wire capacitance is only higher due to increased surface area, not because of the spiralling geometry.
"
Turn it around. The spiralling geometry causes increased length, which causes increased resistance, inductance, and surface area, and the increased surface area causes the increased capacitance.
ak