The secrets of spill: Crosstalk - magnetic spill
Magnetic crosstalk brings us to a higher dimension of problems.
As I said earlier, it should never be a problem in digital equipment as any crosstalk can be placed safely out of the no-man’s-land between 0 and 1, which is the only situation where it could cause errors. Crosstalk performance in analogue recording is however of major significance. Where electricity is tractable and follows set paths, for the most part, magnetism is feral by nature and goes where it likes, recognizing few rules of civilized behaviour.
So on an analogue multitrack you expect crosstalk from one track to the next, particularly on the 24-track format where the guard bands are narrow. Oddly enough, magnetic crosstalk is worst at low frequencies (which is probably a good thing). Old hands might remember the Alesis HR-16 drum machine from the 1980s which had one particular kick drum sample that was notorious for spreading over up to six tracks, such was its low frequency content.
Once again it is timecode that gives us the worst problem. Timecode from track 24 can crosstalk into track 23, and many engineers would prefer to leave track 23 free as a guard band. Crosstalk works the other way round too, and high level signals on track 23 can bleed into the timecode track, making the synchronizer throw a wobbly.
Putting a bass instrument on track 23 is certainly not a good idea, although thankfully there is no lasting damage to the timecode track and the problem is fixable. Another crosstalk problem that occurs in analogue recording is where you want to bounce from one track onto an adjacent track. Here, the crosstalk occurs actually within the record head. Signal from the head element in record mode leaks directly through the spacer into the adjacent track in play, then it is sent through the mixing console to the record track again, and round and round in the inevitable feedback loop.
This is often interpreted as a fault condition, but it is simply part of the nature of analogue recording.