The belay at that point was visually complex. This should not have been a problem. We were all experienced. The light was difficult. I took Pete's PAS without realising he keeps it stored in a different way than most climbers: rather than keeping the twist lock on the end (possibly with a keeper), he keeps it attached to the middle. This and the fact that I was using someone else's kit to secure someone other than me should have made me run a double check, especially as I was clipping him to the back of the powerpoint and hence operating by feel not sight. I opened the gate and the gate snapped back shut. Except in that half a second the PAS jumped out of the gate. I checked the twist lock was done up. I checked the twist lock was in the powerpoint. I didn't check the PAS was still in the twist lock, nor perform an alternative inspection by giving the PAS a yank. Because the powerpoint was being squashed against the rock, the PAS was held in place by being trapped between the powerpoint and the chockstone, so this issue was not spotted.

I would not normally do such a second test, but then I would not normally be operating out of sight, with three people on top of each other, in a cave with a blinding light in my face meaning my vision was poor. Nor using someone else's kit to secure someone else. I should have triple checked.

Interestingly, Pete failed to secure himself to a route in Norway with the same PAS. So this keeps happening to him. In that instance, the PAS got entwined with a big sling, and it was the locker on the sling that he clipped to the anchor, rather than one on the PAS - he had used that one at some other point on the climb for something else. Two data points don't make a conclusion, but I am left wondering if one should consider always keeping the locker on the end of the daisy, shortening it by other means, and consider using a retainer on said locker - many do just that.