Current plate motions can be accounted for by a balance of active forc
es, slab pull, ridge push, and, for continental plates, trench suction
, with drag beneath the plate as a resistive force. If we assume that
the same forces have acted through time, we can reconstruct plate moti
ons from the geometry of past plate boundaries. Paleozoic reconstructi
ons are made with paleomagnetic, tectonic, climatic, and biogeographic
data, as no ocean floor remains. PALEOMAP reconstructions are used to
estimate past plate speeds and to test simple dynamical models in ord
er to determine which ranges of forces best accounts for the observati
ons. Over the last 600 m.y., plate speeds averaged over 40- to 100-m.y
. intervals show considerable variation; Gondwana's speed oscillates f
rom 20 to 60 km/m.y. over a long timescale (200-400 m.y.) with conside
rable noise superposed. Over the Paleozoic Era motions for large conti
nental regions average 28 km/m.y.; force balance models based on prese
nt-day observations suggest that continental regions without a large a
ttached slab would move 30 mm/yr. The opening and closing of the ocean
between Laurentia and Gondwana 560-400 Ma is used to test dynamical m
odels and the parameter values assumed. In the late Precambrian, Laure
ntia rifted away from Gondwana. In the earliest Cambrian it was near 4
0 degrees S; by Late Cambrian and Ordovician it had moved to the equat
or. During the Silurian and Devonian, Laurentia reversed direction and
later collided with Gondwana at 40 degrees S. In a model of the force
s acting on the plates, slab pull, ridge push, and trench suction are
assumed to balance plate drag. Only certain ranges of ridge-push and t
rench parameters can model both the opening and subsequent closing of
the ocean. The dynamic models, with parameter values inferred from pre
sent rates, bracket the rates required by the reconstructions.