Re: Frosted Popper-Tarts (was Re: Underlying basis...)

["Thomas R. Holtz, Jr." <th81@umail.umd.edu>]

At 11:53 PM 7/19/99 EDT, Philidor11@aol.com wrote:
>A past event such as the K/T bolide impact cannot be tested experimentally 
>because prediction is impossible.  A possible set of relationships among 
>animals consistent with all known data cannot lead to predictions which can 
>be tested immediately in a laboratory. 

Ah, you seem to have conflated "test/experiment" with "laboratory
experiment".  Sure, in physics, chemistry, and some parts of geology and
biology the tests are done in a laboratory; in astronomy, other parts of
biology and geology (such as paleontology), etc. these tests sometimes have
to be done outside the lab.  Nevertheless, they are still tests.

For example, discovery of the iridium spike at Gubbio, Italy led to the
hypothesis "an impact of a bollide occured at the K-T boundary".  In order
to test this hypothesis, additional data was looked for.  This additional
data (other correlates of impact, such as iridium spikes at other
localities, tektites, impact quartz, a crater, etc.) represent tests of the
original hypothesis: had no additional data been found, the intitial
hypothesis would be considered wanting.

In the example of the hypothesis "paired fins evolved prior to jaws in
vertebrate evolution" I gave in the previous posting: we could test this by
looking at fossils and find which of the following occur among early
vertebrates: jawed fish without paired fins, pair finned fish without jaws,
fish without fins or jaws, and fish with both fins and jaws.  (Okay, the
last two examples are non-informative cases).  If we find that jawed fish
without paired fins in the fossil record, but not pair finned fish without
jaws, we can say that the known data reject the hypothesis; if (as we do)
find pair finned fish without jaws, but no jawed fish without paired fins,
we support this hypothesis.

>However, a hypothesis in the historical sciences generally can lead to 
>predictions of future discoveries.  For example, discovery of material 
>associated with an impact from 65 mya might be predicted.  A hypothesis about 
>evolutionary relationships among animals can lead to predictions that animals 
>with certain characters will be found dating from certain times and in 
>certain places,

Indeed: however, these future discoveries themselves represent tests of the
original hypothesis.  They will often produce new hypotheses, but the same
applies to experiments involving mixing two fluids in a beaker.

(Just an aside about laboratory experiments: because of the concern over
experimental error, there is always the need for multiple runs of the same
experiment, as well as controls, and so forth.  Even lab experimental
results are tentative: future discoveries could potentially overturn their
results).

>In any case, though the event itself might not be directly testable, the 
>search for predicted evidence might itself be considered a testing process.

Yes.

>Once new data becomes available, an apparent black swan (to use the analogy 
>previously discussed) might not lead to immediate rejection of the hypothesis 
>because of difficulty with the quality of the data, so I guess even 
>refutations are tentative.  By the way, would 'all known swans are white' be 
>a hypothesis or an observation?  Doesn't a hypothesis involve a conclusion 
>drawn from the data (or sometimes the lack of data, as with that thing in 
>Scotland)?

"All known swans are white" would seem to be more of an observation,
supporting the hypothesis "all swans are white".

                        Thomas R. Holtz, Jr.
                        Vertebrate Paleontologist
Deptartment of Geology                  Director, Earth, Life & Time Program
University of Maryland                  College Park Scholars
College Park, MD  20742       
Webpage: http://www.geol.umd.edu        Phone:301-405-4084
Email:tholtz@geol.umd.edu               Fax:  301-314-9661