The purpose of this investigation is to encourage a fresh look at Plei
stocene Beringia. Heretofore, flooding of Bering Strait has been cited
as the only barrier to migration, with marine sea transgressions bein
g a ''sea gate'' that closed off migration during glacial interstadial
s and interglaciations. However, the possibility exists that glacial a
dvances were also barriers, with marine ice transgressions being an ''
ice gate'' that closed off migration during glacial stadials and glaci
al maxima. This possibility proceeds from the Marine Ice Transgression
Hypothesis (MITH), which states that marine ice sheets form on the br
oad Arctic continental shelf of Northern Hemisphere continents when se
a ice thickens, grounds and domes in shallow water, and then transgres
ses landward as continental ice sheets and seaward as floating ice she
lves (Hughes, 1987). Landward transgression is onto coastal lowlands.
During Pleistocene glaciations, a marine ice sheet extending from Spit
sbergen to Greenland may have transgressed the circumpolar continental
landmass at its lowest and narrowest gap, central Beringia, and calve
d into the Pacific Ocean. Four models of Beringian glaciation are pres
ented, based on the distinction between marine glaciation and highland
glaciation. Central Beringia was glaciated only in highlands in the t
raditional model (Hopkins et al., 1982), was also glaciated by a self-
sustaining ice shelf floating over the deep ocean basins of the Bering
Sea in the model by Grosswald and Vozovik (1984), was glaciated by a
marine ice sheet that covered highlands, the continental shelf, and su
pplied the ice shelf in a model for maximum Pleistocene glaciation, an
d was glaciated by a marine ice sheet in the Chukchi Sea that merged w
ith highland glaciers, transgressed the continental shelf of the weste
rn Bering Sea, and calved into the southern Bering Sea along the edge
of the continental shelf in a model for the last glaciation. Field tes
ts are suggested to assess the viability of these four models. The fir
st model is already established for highland glaciation in Alaska, but
less established in Siberia. The last model should be the easiest to
evaluate for marine glaciation. The last model limits human migration
across the Beringian land bridge to brief intervals between stadials a
nd interstadials of the last glaciation cycle, when bath the ice gate
and the sea gate were opened to human migration. This model can influe
nce the sea change now underway among Quaternary scientists studying p
eopling of the Americas, based on the archaeological, linguistic and e
thnic diversity among native American populations.