On TN31's, I measured as low as 5.5A to the LED with bridged resistors and maybe 22 AWG, but I measured 6.2A to the LED with 20 AWG I believe, and I think 18 AWG is used to get the 6.5A to the LED.
Thanks Tom. That’s great info right there. Something to keep in mind for sure.
I replaced the resistors with solder on my TN31 and used 18awg wires. Works perfectly.
Ohhh - ok, those results were on different lights, so not a controlled comparison. Also O-L mentioned this and I believe it - the Teflon coated wire with less strands performs better, so believe he said 20 AWG Teflon is as good or better than 18 AWG silicon wire, something like that. I am still very confident that the wire thickness is the difference of why my fully modded TN31's in the past were under-achieving, compared to others (vinh), which I later found out was using heavier gauge wires. 6.5A on a TN31/K40 is about the max. The K50 sounds like it max's out higher for sure. After working on Fenix lights, I'm a total believer in heaviest possible wire - their LED wires are the largest I've ever seen - believe it's 16 AWG (smaller than 14, larger than 18), and for them, weight does matter, so they didn't mind adding weight in the wires.
The strand count does matter. I have been using Bell wire on lights that I need to keep the wires really small for profile reasons. Lights where putting in heavy wire would set the reflector too high. It’s a single strand of heavy wire with very thin plastic coating. I have even used the wire that is used for installing cable TV. It’s a single strand of very heavy wire. I used it on multi-emitter lights. I pulled it out of the cable wire and then coated it myself with non-conductive glue.
I was taught in my electronics classes that electrons flowed around the outside of the wire/ wires. So in theory a cable/wire with 30 strands would conduct/flow better than a wire with one strand of the same diameter because the one with more strands has more surface area compared to single strand. Of course this is only theory.
Doesn’t the skin effect only apply to very high frequency AC current?
The buck driver does switch on and off very fast but the replacement wires to the LED only carries DC.
High current DC wiring tends towards more smaller strands. I don't buy that junk about single conductor wire being better for DC. I think people 'discovering' some benefit with single conductor wire over stranded wire are missing some other signifcant factor and then misidentifying the cause.
My classes dealt only with automotive electronic systems so it was DC voltage but I was not the teacher nor did I question them about it. But nowadays I work with engineers so I can ask one at work and see what they say about it ( of course they might not be correct either but they have that degree and will NOT hesitate to tell you so)
I haven't done direct tests to compare wires (strand counts) - sounds like it would possibly be easy to do. I know O-L mentioned it and I thought there was other agreement/confirmation. Hope 18sixfifty can report if he noticed differences - I assume he meant less strands were better? Not sure...
Another factor in the wire debate is the quality of the wire itself. There’s a lot of cheap wire out there. The Teflon wire is generally a very high quality silver coated oxygen free copper wire. So is it the strand count, or just the overall quality of the wire? I think it’s the higher quality of the wire that’s making the difference.
+1 ouchy, similar to what comfy said, could be other factors -- well all I do know is 20 or 18 AWG seems better than 22 AWG, least it's been my experience 
.
Would measuring the resistance with a DMM and crocodile clips help with the process of figuring this out? Just throwing a thought out there if anyone is going to test them all.
It’s going to be very difficult because we’re talking miliohms and most meters can’t accurately measure many levels between 1.0Ohm and 0.0Ohm.
I like what djozz is doing, pushing a high current across the switch and then measuring the voltage drop. Much more accurate than the DMM method. switchtesting , (june13th2015:additional test by MRsDNF in post#44)
Guys have measured resistance of wires, etc., but I had no clue how... Yes, regular DMM's don't read low enough, but think specialized meters can do it. Forgot who, but someone measured the resistance of the traces on the SinkPAD's and Noctigons early on - thread about the stars I think.
Update: Here it is, post #5: https://budgetlightforum.com/t/-/19753
Thinking about it, I do have access to precise bench DMM’s that can measure up to 6 digits. (Agilent 34401A)
I’ll have to compensate for the leads’ resistance, but should yield pretty solid results.
It's easy to measure voltage drop (loss) through two given pieces of wire. What's not so easy is making sure they are the actual advertised wire gauge and of comparable cross section. I've noticed that it's not uncommon to find that you can for instance use the 24AWG hole in wire strippers on some 22AWG wire and not nick any of the strands, and on different brand/source 22AWG wire you can't. To be able to say for sure which is 'better' between solid vs. stranded or fine conductors vs. coarse you have to be comparing apples to apples. Just going by what's printed on the insulation isn't good enough and the kind of thing I mean when I say 'misidentifying the cause'.
You would have to test the wires with a digital low resistance ohmmeter or a meggar
Whichever is easiest should suffice I think - measuring the resistance or voltage drop. Of course more testing increases accuracy though so I welcome it if you have the time 
I’ve found that to strip 24AWG stranded teflon wire, I have to use the 26AWG stranded hole to strip it. You have to be careful though. I think it’s because of the tighter, more compressed teflon coating that you need the extra bite.