Passage
Some lake floors preserve paired layers: a pale layer of mineral grains washed in during the melt season, and above it a dark layer of organic matter settling once inflow slows. Where the pattern is undisturbed it can be counted downward from the surface and each pair taken as one year, giving the pollen, ash and charcoal a chronology of a precision no other method for that span approaches. For decades the counts were reported as though the pairs were being read off a calendar.
Two findings complicated that reading. In basins fed by steep catchments, a single heavy rainfall in late summer can wash in a pulse of pale grains that resembles, by any criterion a core makes available, a second annual pair. And where inflow is weak, a year can pass without contributing enough of either material to be distinguished, so that two years of organic matter merge into a single dark layer. The first yields counts that are too high; the second, counts too low.
It is tempting to conclude that the two offset, so that a long count is about right. They do not offset, for a reason worth stating carefully. They are not random noise but products of particular conditions, and which condition prevails in a basin turns on the same variables — melt vigour, intense summer rainfall — whose history the chronology is built to recover. A basin’s errors therefore run predominantly one way for long stretches, and which way they run is not independent of the signal under study.
Practice has adapted, though not by rescuing the original claim. A count is now anchored, where possible, to a layer dated from elsewhere: an ash bed traced to a known eruption. When the count disagrees with the anchor by thirty pairs in four thousand, the anchor settles the total but says nothing about where in the interval the surplus or missing pairs lie. What results is an interval estimate at every depth except the anchors themselves. Questions the older counts had appeared to answer — whether a pollen shift at one lake led one at another by a decade — are returned to the undecided. Candour of that kind is a gain; it is not a gain in resolution.
