Consider a bundle of wires carrying binary data and the worst case (or almost-worst case) scenario of them switching all at the same time. The wires with their partial mutual inductances would be aiding one another.
It can be found that if the currents are the same, then the magnetic fields are the same and if the currents have the same direction, these are aiding inductors.
The loops are very independent, each wire forming its own LC circuit with some conceptual capacitance on the other end of the wire. Simulating this with 5 mutual inductances (expressed with coupling coefficients) in LTspice or PSpice yields a ringing peak of about 1.4 V instead of about 1.07 V with no mutual inductance.
From the simulations, no more than 1.4 V of peaking can be obtained maybe because they weren't intended for such a large number of mutual inductances. PSpice at least supports a maximum of 6 mutual terms.
However, it can also be found that, considering the mutual inductances, there is a very large matrix of the form:
Then, wouldn't all the mutual inductances add up to the self inductance to yield a very large ringing peak for every wire in the bundle?