1310 THE STRUCTURE OF EVOLUTIONARY THEORY
be consistent with truly catastrophic removal—a "counterintuitive," but actually
rather obvious, option that paleontologists had never conceptualized because the
literal signal matched their expectations, and they therefore never questioned the
meaning. Signor and Lipps argued that if all taxa under consideration had truly died
at once, their recorded disappearance in the stratigraphic record would still be
sequential, as based on probability of fossilization. After all, a top, large-bodied
carnivore like Tyrannosaurus, with a local geographic range and a relatively small N,
might only yield a fossil once every several meters during the stratigraphic range of
its actual existence; whereas a foraminifer that lived as billions of individuals at every
moment of a continuous oceanic core, should provide abundant specimens in every
mm of its stratigraphic existence. Thus, even if the dinosaur and the foram species
died at the same instant in a worldwide catastrophe, the last dinosaur fossil might still
appear several meters below the extinction boundary, while foram fossils should
persist to the last stratum.
This artifactual pattern, now appropriately named the Signor-Lipps effect, can
be distinguished from a genuine petering out that truly gradual extinction would
produce—most obviously by testing for correlations between time of disappearance
before the boundary and expectations of preservability in the fossil record (as
measured by "waiting time" between fossils in normal strata between extinction
events, not by working with abundances near the boundary itself, where circular
reasoning may so easily intervene). Regardless of the outcome in any particular
application—and some studies have yielded consistency with Signor-Lipps, whereas
others seem to show genuine petering out— paleontologists had never formulated, or
even conceptualized, this important methodology until the catastrophic impact
scenario forced their attention to such questions.
As a different approach to the same basic situation, one might decide instead to
take a group long interpreted as showing a clear literal signal of petering out—and
then go to the most promising stratigraphic record in the world, pulling apart every
single bedding plane to see if surpassingly rare specimens might occur nearer, if not
right up to, the extinction boundary (for you only need one to disprove complete
extinction). This "needle (or dinosaur) in the haystack" strategy represents the "flip-
side" of the Signor-Lipps approach to the general problem of sampling in science—
either use a global and statistical method to extract a clear signal from broad data, or
sample with such intensity in a more limited area that you can effectively survey the
available "universe," and no longer even require the art of statistical inference. But
why would one even think of sampling with such intensity, absent the prod of the
impact hypothesis and its prediction of putative success. After all, if your world-view
enjoins petering out, and your data (read literally) clearly display just such a pattern,
why would you petition the National Science Foundation for cash, and then spend
several summers sweating in a desert, pulling apart every bedding plane in a single
place virtually guaranteed to yield nothing. Such behavior could only point to an
unsound mind—