The error rates seem REALLY tight. I was a bit surprised. If you look at the speleothems that are used in IntCal they are nothing like that tight and they are only in the 25 -50k range.
Would love to hear thoughts on this from those more knowledgeable.