The most recent (2013) edition of Climbing Anchors, by Long and Gaines, makes some interesting statements in the preface about catastrophic failures of anchors. Long first says that: when belay anchors fail, it's usually in cases where the gear was placed in a horizontal crack, and the anchor was subjected to a shock load
sideways to the intended direction of pull. He attributes this to
anecdotal reports. He also says the following, which I find very surprising.
Because statistically, the majority of climbing accidents occur from leader falls, we'd expect catastrophic anchor failure to result from a leader taking the dreaded Factor 2 ripper right onto the anchor. Not so. Most of these failures involved a leader belaying a second up on a toprope.
So from this information, it sounds like these failures often happen the first time the anchor is ever loaded, i.e., it was such a poor anchor that it wouldn't hold even a top-rope fall.
Is it possible to find systematic statistics anywhere on the types of anchor failures (top-roping, leading, ...) and how survivable they were? In the case of the leader belaying the follower, it seems like the failure might actually be survivable in many cases, since there might still be a lot of gear down below that hadn't yet been cleaned. (But the belayer had better maintain a serious death-grip on the rope!) Even if, as Long says, few failures are from factor-2 falls, those are probably the ones that would not be survivable at all, since there is no other gear present.
I'm also having a hard time reconciling the statements that:
- "anchor failures are usually from a load "sideways to the intended direction of pull,"
- they usually happen while the leader is belaying the follower.
When the leader is building the anchor, his rope is trailing right back toward his last piece of pro, so it should be dead obvious what the direction of pull is going to be when he catches a fall by the follower. The exception would be, I guess, when there is a traverse from the last piece to the belay station, and the follower takes a fall after cleaning that final piece (and in this case, the anchor failure is probably not going to be survivable).
I would also be interested in statistics about how many accidents are caused by anchor failures while climbing, as opposed to rappelling. The latter seems more common, which probably isn't so surprising, since people usually try to make their belay anchors really bombproof.