Help me solve this enigma

Long post, sorry, but I want to explain everything right, because this is really puzzling me.

I got these two flashlights from DX.
When they came I first tested them with 3xNiMH AAA, and they had pretty much the same brightness. I was also surprised by the high output.
I tossed the 3xAAA carrier in one of them, replaced it with a 18500 LiIon cell with some padding and used it for about two weeks. I left the other one unused in a drawer along with its carrier and 3 NiMH cells.

Today, in the process of reviewing these flashlights for my website (along with a few others from DX), I looked at the output of the light I've been using and thought "hmm, wasn't this brighter"?
I fished out the NiMH-powered one from the drawer, turned it on, and sure enough it was significantly brighter.
At first I figured the LiIon cell was at fault; I hadn't used the flashlight much, but the cell is quite old (like, five years), and we all know how LiIon cells don't like aging.
But when I swapped the 18500 and the 3AAA carrier around, the situation didn't change much.
I measured the 18500's voltage and it reads 3.9, so it's far from being discharged.
I then started thinking maybe the LEDs were being overdriven so badly they had already degraded, but it seemed unlikely. The light was bright, but not THAT bright.

On a whim, I then swapped the tailcaps around. I don't even know why I did that, because they can obviously have no influence on the output... or so I thought.
With the other tailcap, the LiIon flashlight is suddenly bright again, and the other one is dimmer.
I've checked a few times to see if I'm not just high on Mentos (I eat a lot of them), and it is perfectly repeatable: one of the two tailcaps causes the light to be dimmer.

Puzzled, I then took the multimeter and measured resistance.
The "bright" tailcap has a resistance of about 0.3/0.4 ohm (which, if I understand correctly, is probably caused by imperfect contacts and is functionally equivalent to no resistance at all). The "dim" tailcap's resistance is inconsistent. On one measurement it's about 4 ohm, on another it varies wildly between 3 and 25 ohm, on another it's 8 ohm, and on one it got as high as 100.
It's gone insane.

Question 1: WTF??!
How can a simple switch develop a wildly variable resistance? Do I have tiny goblins in there playing with micro-sized resistors?
I should mention the variable resistance seems to have an effect only on the multimeter's readings. When the cap is on the flashlight the output is dimmer, but it doesn't change.

Question 2: am I right in assuming the "bright" tailcap has no resistance?
I can handle simple cells-motor/light circuits, but when we start talking about resistance and ohms my brain starts to hurt, so I don't know if that 0.4 ohm resistance changes under load or what.

I then measured resistance between the head contacts and the contacts of the LEDs (the back side of the board is exposed), and it reads 0.2 ohm. Measuring the resistance between the screw threads in the head gives the same result, so I'm pretty sure the 0.2 ohms are added by the multimeter and there are no resistors whatsoever in the head.

If that is the case, and if there's no resistance in the "bright" tailcap...
Question 3: how can it be that running on low-internal-resistance rechargeable cells, and from my variable power suppy (set to allow anything to draw as much current as it wants up to the supply's maximum of five amps), the LEDs didn't go into thermal runaway and die a horrible death?
These are the new 20.000mcd LEDs... could they be self-resistored or something?

Please help. I feel like I've broken some fundamental law of physics and disproven years and years of studies.

Try cleaning the contacts on the "dim" tailcap. I'm betting you have a little corrosion (oxidation) on the contact/spring. Rub it with some sandpaper or steel wool or similar and I bet it improves. Contact cleaner will also work if you have any handy.

Roger

Yeah, It sure seems to be oxidation built up on the tailcap contacts. I would use deoxit or rubbing alcohol. One other thought make sure the spring is making good contact with the cap and if there is a retaining ring, make sure it is also tightened down.

Don't use anything abrasive it'll just make things worse in the future, I used a spray contact cleaner and cotton buds.
Norm

It happens on any point of the spring I touch with the probe. It also happens if I touch the contact in the switch (you can touch it from the center of the spring).
Regardless, I'll try cleaning the contacts with alcohol & cotton buds and report the results.
Thanks.

What about questions 2 and 3?

I have a similar pair of torches, but with the dimmer LEDs. (I haven't used them much as I'm waiting for some more Crees...) Anyway, another torch I bought locally, anodised blue with 14 LEDs, has such a bad tail clicky that it would flicker and vary markedly in its brightness while just sitting still on a table! However, each time I took the switch completely apart and cleaned it (without harsh scratching) it would get better for a while.

It eventually got too annoying so I swapped it with the tail clickie from one of the DX torches (exact same thread) and it has worked mostly fine since. It looks a little funny though - a black tail clicky on a blue torch. And of course one of the black DX torches has a blue tail clickie.

I say mostly fine since I recently thought it was rather dim, even though it should have only been part way through its set of NiMHs. I measured the voltage on the cells and they were fine, so I put them back in again and the torch was suddenly back up to full brightness again. Hmmm. Bad connection somewhere.

My conclusion - when these little torches work they're not too bad for size, cost, etc. Pity they don't work well all the time, and are less than fully reliable.

Oh yeah, one more thing. I measured 450mA on one of the DX torches with fresh NiMHs, which means an average of 50mA per LED. Yeah, that's overdriving them.

Oops, that took a while to type, and it didn't help going off to look at other threads.

Any contact will have a small resistance, but your good tail cap's resistance will be pretty low.

As for thermal runaway, driving LEDs that hard constantly probably won't do much for ensuring they have a long life. But I figure they should still last long enough for even more efficient LEDs to make it into the budget range. Just think - Cree XR-E P4s can now be bought for under US$5 each.