Yes IIRC even 2 would not work.
My thoughts:
Resistance goes up as the wire heats.
Initially, the shorter wire heats the most, but has the least resistance.
As its resistance increases, more current flows to the next wire(s) in sequence.
During this time, the experimenter is turning up the power to increase current (fixed voltage? Resistance in sequence to adjust power?)
When the longest strand begins heating and resistance increasing, the load begins rebalancing so the increased power (amperage?) now is moreso going through the shortest wire as it was at a lower load before the wires began heating.
But now, we have enough power to heat all wires, much of it now going through the shortest one, as the other wires also have a higher resistance than when they first started heating up.
Resistance can go up by 20% it seems, but the difference short to long is 14" to 16" so the shortest wire is overloaded.
Of course, as others point out, when the first wire melts, the same total current is going through fewer wires, overloading each in sequence, more spectacularly as the shorter ones fail.
(When I say power, I assume the power supply keeps the voltage as constant as possible and the added power is more current…)
My thought would be that you want to turn up the power very slowly so the whole stays as close to equilibrium steady state as possible; the problem is the evolving changes in heat/resistance causing varying imbalances and the power is turned up too high before the system is stabilized.
Maybe a little due to the draft. The wire is very soft at that point. You can see that it droops into a perfect catenary when it’s close to breaking, and only ever breaks close to the end where the tension is highest (but not too close, since the copper supply wire acts as a heatsink). So if the falling wire causes a small disruption to the one next to it, it wouldn’t surprise me.
BTW, your selected gauge and lengths were right in the ballpark of where they needed to be. They got hot in the several tens of volts range, which given a 120v source meant decent freedom to dial up or down. A much lower gauge would have been prohibitively long for an acceptable resistance.
Yes I remember the voltages were in that range, I don’t think they got far north of 50V and I was at the lower end of the scale but still with a good amount of adjustability.
I think we are on the same page. I proposed the same explanation in post 21.
That rebalancing would only happen on a current-limited power supply, though. Or one where there was significant resistance in the common portions of the circuit.
A variac is a voltage-limited supply. So would be a triac-based dimmer switch. So they aren’t affected when one wire is cut.
If the power source is a voltage source, then the current in each wire is independent of current in the other wires. At least ideally.
I have worked with variacs before. They are current limited - just like a regular transformer. The current through the primary/secondary winding is limited by the gauge of the wire.
The plan was to have voltage adjustable, and the voltage was measured on a meter as I adjusted the variable voltage controller (of some type I don’t recall). But I do suspect that there was a current issue as well that was inherent in the voltage supply/regulation.
One thing I sort of vaguely remember that I had to turn the knob further than I expected when the full array was connected to get the expected voltage. This makes me suspect: 1: the power supply had trouble providing enough 2: as resistance increased due to increased temperature, voltage increased, but I don’t think that amount of extra rotation of the knob would have popped the longest wire once the others popped.
Too bad that part of the experiment was a side issue to sidestepped as it is pretty darn interesting.
That’s not what anyone means by current-limited, though to be fair I probably should have said “current source” or “constant current.” Anyway, the fuse would blow first.
The load here is well below the capacity of the variac and hence it acts as a constant voltage source.
Incidentally, the reason I have so much nichrome laying around is because we used it to cut the solar panel stays on our cubesat.
We used a constant-current high-power LED (buck converter) driver to power them. Initially we used power resistors to get the ~8 V from the battery supply down to the amount required, but that was an extreme amount of power due to the very low voltage across the short nichrome loops. The LED driver was efficient, putting around 2.5 A through the nichrome while not just dumping the excess voltage into heat.
Unfortunately, our best guess is that it failed somehow, not releasing the panels. We communicated with the cubesat a couple of times, indicating that it was running but with extremely low signal power, which probably meant that the antennas (which were behind the panels) didn’t extend.
Most likely, it just needed to be engaged for longer. We probably should have had an actual deployment sensor, increasing the duration of the cut until it happened. We gave it very generous margins over what worked on the ground, but it seems it wasn’t enough in space. Hard to know for sure, though.
It’s super cool that you did this, thanks.
Glad to! Can’t have all these supplies sitting around going to waste. I probably have a few thousand feet of the 36 AWG still…
So finding new ways to get the wires to explode would have been very helpful 
Sorry it didn’t work.
We considered that! Instead of using the nichrome to cut (Spectra) fishing line, just tie them down with the nichrome itself and vaporize it. The idea had some downsides, though…