Here's the poop behind the ohms rating for these coils and caps:
Intended coil discharge time: 1.1ms-1.6mS range (shorter on 500/550/750F and late 750K engines, due to hemi chamber, longer on swirl-charge engines like early 750K).
Coil wattage: about 24 watts at 1500 RPM, 15 watts at 8500 RPM.
Discharge voltage: 6.5k to 7.5k volts, stock.
Picking all numbers right in the middle, as a typical designer might do:
19.5 watts / 7000 volts = 2.79 mA maximum spark current available, able to "pump" the spark into up to 2.5 megohms. (This is important for starting the spark across the gap, or "setting up the plasma field" as it was once called, long ago...)
If spark discharge time is supposed to be 1.35mS, then the resistance has to slow the discharge of the coil's inductance while it tries to pump this current to the sparkplug.
So, we take the watts (19.5) and divide it by discharge time (.00135) and get the ideal discharge watt-second load of 14,444.4 ohms.
Now...this ohms must be split between the sparkplug and something else (in this case, the resistor cap or wire), because the plug starts at a very high resistance (like 1.5 megohms) until it builds up the plasma field that will guide the spark across the gap, then it drops to about 2000 ohms during the discharge itself (i.e., the actual spark jump event) because the plasma field acts like a conductor of sorts. So, it's not a strict, linear resistance arrangement: you're trying to hit a moving resistance target while the coil is discharging. This changing resistance, when drawn in a graph, looks like your bike's speed when hitting the brakes hard: it begins to stop at one rate, then stops harder as the tires bite and heat and the brakes get a more solid grip.
So, Honda's engineers elected to split the ohms in the middle, too, at about 14.444k/2 = 7.222k, and their OEM plug caps fell into the range of about 6-8K with their production tolerances, circa 1970 or so.
This translates to about 7.5mm of spark gap in the open air, which is precisely what their Honda Electrical Testers of the era used to have for us to check coils with: the spark had to jump at least 6.5mm, and typically would reach about 8mm.
In actual use, a 2k resistance (like 12" of Dyna suppression wire) will still produce most of a 50/50 spark transfer to the plug gap, although the spark will be shorter duration (like .9 to 1.1 ms), while heating up the wire a little bit and slowly causing it to burn out. Not to worry: Dyna coils have replaceable wires.
A 10K resistance is actually Honda's upper spec limit, requiring cap replacement. It will cause a delayed spark start, which shortens the actual jumping time for the arc, and in wet conditions can cause the spark to exit to other places than just into the plug.
A 5k resistance is pretty close to Honda's original plan. Coils that came out on other Japanese bikes in the 1980s had fewer turns ratios (on bikes with bigger alternators), which are cheaper to build. The electronic ignitions were used to support the higher currents of these coils with no troubles, because points would burn up faster (witness points life with Dyna 3-ohm coils as an example).
So, the NGK 5k caps are a good choice, given our current parts supply situation.
