JT the only way someone would die in your scenario is if both links failed at the same time(on a 2 bolt anchor. The odds can't be figured the same.

I don't know if you play poker but the odds of getting pocket aces are not the same as the odds of two aces showing up somewhere on the table during the game, and that is what you are doing, treating those scenarios the same. I thought stats was your strength?
Scott

I agree with you in principle, however in this case, aren't we saying that there is a 2 in one million chance that the anchor will only withstand somewhere between 5.5-11 kN, a value that is still sufficient for rappel. Realistically, the only way to achieve forces in that range is if someone was foolish enough to fall directly on the anchor attached by a spectra sling, with no mitigation from energy absorbed in the body or friction as the person scrapes down the rock, and that assumes they attached themselves to the quicklink directly and not the bolt hanger, or the sling.

The real danger in this scenario is the static fall on the anchor, because if the person falls just a little farther, we're talking forces that that could break slings, biners, not to mention backs.

I'm beginning to regret that I made any calculations in the thread. It might have been better had I said that the all we know from the testing is that there is too much variability in the failure rates, and left it at that.

Jay

Insults aren't answers. Let's see if you can get it right Jay. If the odds of a link failing are 1 in 500,000, what are the odds that two links will fail at the same time?

Oh, and I a teacher so show your work.

You didn't answer my question...

The probability of a link being weaker than the mean minus 3 sigma is .00135, so the probability of two of them being weaker than the mean minus 3 sigma is .00135^2, or .00000182, roughly odds of 500,000:1.

Edit, Hypothetical question: Say the odds of an anchor failing are 500,000:1. Would these odds be good or not? Would you rap off an anchor that had 2 chances in a million of failing? Probably you would. I know I would. Does that mean that it would be ok to knowingly install anchors that had 2 chances in a million of failing? In my opinion, no. Here's why: Imagine that everybody installed anchors that had 2 chances in a million of failing. How many anchors are there in the world? I have no idea, but let's say that there are 1 million. Then, chances are that 2 of those anchors will fail, and someone will die. Thus the only justification for installing anchors that have 2 chances in a million of failing is that most everybody else is installing better anchors. Kant would not approve.

In other words, we have to keep in mind that the probability of failure is 2 chances in a million per anchor. The probability of an unsafe anchor being installed somewhere increases each time such an anchor is installed.

Jay

Personally, I'm glad you provided to calcs. It is the kind of interesting input that provides value to threads like these. I think you have been very clear that, with the limited sample size, any analysis of the data is of limited value, this should also be relatively obvious. I think the testing of the links provides a context for the discussion, gives some actual values to play with, and helps explain how the process works, and helps to explain why a substantial data set is important to statistical analysis. Also, provides a context to explain the significance of distribution in statistical analysis. All interesting things.

If you have a moment, I wounldn't mind a brief description of "bimodal failure (Edit: oops I mean distribution)".

Insults aren't answers. Let's see if you can get it right Jay. If the odds of a link failing are 1 in 500,000, what are the odds that two links will fail at the same time?
Oh, and I a teacher so show your work.
S

The 1 in 500,000 value was for the probability of two links failing in the one anchor. The probability Jay calculated for one link failing was around 1 in 740.

I have red all of dis tred and decidered I wood rather repel on too of dose Chineez links den one expenseive one.

Your right, I'm no math teacher, and my english teacher was always displeased with my lack of proofreading. I'm going to make some wild assumptions and use my limited memory of basic stats to see if I can calculate my chances of dying on two china made quicklinks with poor quality control.
My assumptions:
A. 1 in 100 of the quick links is significantly below specs and would fail in a low range of force.
B.1 in 1000 climbers would do something stupid or unfortunate enough to generate forces in the range necessary to cause failure of both links.
If we square 1/100 (for two such links at the same time) then multiply by 1/1000, that is a wildly speculative but I think conservative estimate of 1 in 10 million chance a climber will experience such an event.

We all kind of expect that the longer we climb, the higher our chances of getting hurt.

If you've done any climbing at all then you've rapped on things substantially less strong than this links, you just didn't know how much weaker it was.

I have no opinion on quick links in general. I would, however, not use these particular Chinese quick links, based on the results of this testing.

Jay

I do find it rather interesting that Sample 3 held the most force, as that was the sample that got additional testing due to the attempts to use the webbing. It is certainly possible that these attempts influenced the results for this piece, but given the variation in peak loads its also possible that it was legitimate.
.