7,500 dated shells put a price on the blur in the marine fossil record

Dating shells is the expensive part. The data needed to test sedimentation rate as a control on age mixing has long been wanted and rarely gathered, because assembling it is time-consuming and very expensive ^[10]. That is what makes this result useful rather than merely tidy. If burial rate dominates, and burial rate can be deduced from the environmental context ^[9], then a quantity that used to require a laboratory budget has an estimate available before the sample bag is opened.

The price of establishing that shortcut once: more than 7,500 fossils dated by radiocarbon and other methods, from shallow coastal seafloor out to the edges of continental shelves ^[6], through a project twenty years in the making ^[7], pooled from research groups based in eight countries ^[8][17]. That is roughly 375 dated specimens a year ^[16]. No individual study is going to pay that bill again, which is the whole argument for a proxy.

The mixing mechanism is not exotic. A square metre of productive seafloor can be perforated by hundreds to thousands of isolated and interconnected tunnels ^[1], and the animals working them stir the sediment together with whatever shells and skeletal fragments are lying in it ^[2]. The result, called time averaging ^[3], is that well-preserved specimens found next to each other may have lived hundreds or thousands of years apart, according to Rafal Nawrot of the University of Vienna ^[4].

Two things follow for anyone reading a layer. First, temporal resolution is a local variable rather than a constant: deltas accumulate sediment fast while other areas receive only a fine dusting over long periods, and rates in one place change through time ^[11]. A single core can therefore carry different resolutions at different depths, and two assemblages of the same age can encode intervals of very different length.

Second, the richest sample is not automatically the best-resolved one. Fossil yield at a spot tracks how many organisms were there to be fossilised ^[12], and productive settings are exactly the ones described as riddled with burrows ^[1], so the sites that reward digging are also doing a lot of the smearing ^[18]. Sedimentation rate is the primary control the consortium identifies ^[15], but productivity pulls sampling toward the noisier end.

Daniele Scarponi of the University of Bologna puts the ordering plainly: before interpreting an assemblage, you need to know the interval it represents, because time averaging dictates which questions can be asked ^[13]. His comparison is between Pompeii, buried fast enough to freeze a community, and a graveyard used continuously across generations ^[14]. Both yield data; they do not yield the same data.

The operative test for a paleoecological claim, then, is whether the interval it asserts is shorter than the interval its sediment permits. Where burial was slow, a "community" reconstructed from one layer is a composite of generations, and it is a composite already filtered by decay and scavenging, since even shells and bones disintegrate if they are not buried quickly ^[19]. The study does not clean the record. It says how dirty a given piece of it is likely to be, cheaply, and in advance of the coring.