With respect H.M, your descriptions are all on the H.T. side of the coil... the primary side would not fail and 'come back' which is what I'm saying.
It doesn't exactly fail and come back: what happens internally is that the insulation on the wires (enamel paint, basically) in the windings just begins to crack after long years. Then, when the coil gets warm inside and presses the (square) iron core against these windings, they short out a few turns here and there from those pressures: the coils made before about 1978 had just waxed paper wrapped around those iron cores. Thus, they act up when hot, not so much when cold. Often, this scenario is coupled with slight arcing (visible in the [real] dark), at the mounting screws of the coils where they attach to the little cross-brackets that hold them to the frame. This symptom is most noticeable on the later bikes, where the frame ground is right at the coils, and on those bikes a "bad" coil will show tiny arc-weld spots between the iron core and the mount bracket if you take it apart to look-see. When this arcing occurs, the bikes sometimes can bite your leg if riding in the wet and the whole front of the bike (and your legs) get soaked enough. It usually seems to happen at idle the most, when your leg is resting against the sides of the frame and holding up the bike while you're standing in the water...
This has happened to me on my 750 (in 1986) and my brother's CB500 (1976), and on my old 1971 CL350 (which was how I inherited that particular bike) in about 1972, so it does not take a long time for this failure to show up. In all cases, an aged coil was the culprit (not wires or plug caps) and a new one solved the biting bug.
Less often, I have seen the spark wire burn in two. This makes a "booster gap" if the wire is still securely held in the coil body, and the spark plug on the bad wire will appear to have a brighter, if yellow, spark (out in the atmosphere) than its sibling on the other wire. This is only true at low pressures, though: in the chamber, it flames out easily and most of the coil's spark energy then transfers to the other plug, which shows a bluer (less bright) spark when outside. This whole scenario led me to my [in]famous "pull on the wires" test, which I do to bikes whenever they come to me for a tuneup. I've found 6 (that I remember) since 1997 with at least one broken wire. (Not all were SOHC4 bikes.)
