No, because you need to know the source impedance to calculate the bandwidth. Therefore the cable itself cannot have a bandwidth (within reason, say into the GHz range).
The part I've bolded is one source of my confusion I think. Doesn't that mean the cable itself is ultimately limited to a GHz range bandwidth that is so wide that it isn't significant for the application ("within reason")? I'm attempting to drill down to basic principles here, rather than practical audio implementations.
That one needs to know the source impedance to determine bandwidth... OK, that is determining the characteristics of the filters described by the interaction between the cable and the devices before and after it. I think I get that part.
[partly off-topic aside-]
Can a device have absolutely zero measurable source impedance? Although I may be mistaken, I think I’ve read about amp designers playing with designs which achieve negative impedance, although not necessarily useful. If so, wouldn’t that create runaway oscillation? I don’t think that would imply infinite bandwidth however.
However, when we get to digital/RF signal transmission there is usually a defined impedance, so for those cables (which also have to be designed with matching characteristic impedance) there is a given signal loss per length of cable. Still not quite a bandwidth because it's dependent upon cable length.
I understand how signal loss is not the same as bandwidth. Signal measured at two locations can have the same bandwidth but different levels. Conversely they could have the same level but different bandwidths (dependant on how level is measured, it would have to be a narrow bandwidth measure, well within the corner frequency limits of both). What I don't get is why transmittable frequency range limits that are dependant on cable length (or not) would not be described as ‘bandwidth’.
Thanks for your help in understanding this.