The little airplane banked to the right. From my seat on the port side I could see its shadow crossing the ice. The skis made it look rather like a duck coming in to land on water, webbed feet outstretched. As the pilot leveled the aircraft, a huge cliff came into view, the dark brown of its rocks contrasting sharply with the pristine whiteness of ice and snow that faded into the horizon.

The steeply inclined layers of this Precambrian sandstone were distorted by concertinalike folds. I took several photographs. As we rounded the cliff, another came into view. Resting on top of the sandstone was a thin capping of rock almost as white as the background: Cambrian limestone. "Fascinating," I thought as I raised my camera again. "The basic geology here is very similar to that of western North America."

My colleagues and I had come to the Pensacola Mountains of Antarctica to study how the two geologic subdivisions--East and West--of the icy continent relate to each other. East Antarctica is an old Precambrian shield lying to the south of Australia, India and Africa; West Antarctica is part of the geologically young and active volcanic "ring of fire" that surrounds the Pacific Ocean. The uplifted rim of the East Antarctic shield meets West Antarctica along the Transantarctic Mountains, of which the Pensacolas form a northern extension.

It had been a long trip down: 14 hours from Los Angeles to New Zealand in a commercial jet, 10 hours from New Zealand to McMurdo Station in Antarctica in a ski-equipped Hercules transport and, finally, five hours across the continent to the Pensacola Mountains, bypassing the South Pole en route. Now, after setting up our base camp, we were finally at the mountains near the southern margin of the same ocean that laps the beaches of Los Angeles.

We still had to get to the rocks, however. In Antarctica such excursions take time. Having selected a possible crevasse-free landing site, our pilot brought the Twin Otter down for a "ski drag." That is, he put some weight on the landing gear but maintained enough airspeed to take off again. We circled and carefully examined these tracks. Crevasses can be hidden under snow, but here there were no telltale signs of blue cracks. Coming around again, we touched down and stopped quickly so as to reduce the chance of hitting rough ice beneath the snow. It was a bumpy landing, nonetheless, although the aircraft appeared to have suffered only superficial damage. We roped ourselves together for safety and started to walk across the windblown snow to the base of the cliff, leaving our anxious pilot to examine the plane.

Fossil Clues
THE BOUNDARYbetween the two rock types exposed in the Pensacola Mountains is one of the most fundamental in Earths history. After the birth of the planet 4.5 billion years ago came the four-billion-year-long interval of time known as the Precambrian. Toward the end of this era--about 750 million years ago, while the first soft-bodied, multicellular creatures were developing--the brown sandstones of the underlying Patuxent Formation we had just sighted were deposited. The strata were laid down in a rift valley that opened within the continental shield. As the rift deepened, rivers poured in, dropping their eroded soils onto the valley floor.

About 540 million years ago, an explosion of multicellular animal life ushered in the Cambrian period. Myriad cone-shaped skeletons of the creature Archaeocyatha collected in shallow seas that had advanced over the sandstone. These formed a reef along the rim of East Antarctica that was eventually transformed into limestone. (The cap on the Patuxent Formation is called the Nelson Limestone.) Because Archaeocyatha was a warm-water animal, what is now the western margin of the East Antarctic shield must have been situated in tropical latitudes during the Cambrian period.

The rifting event that led to the Patuxent sandstones' being deposited reflects the separation of East Antarctica from some other continental landmass. The divergence opened the Pacific Ocean basin about 750 million years ago. (Subsequently, igneous rocks from island volcanoes and material scraped off the subducting ocean floor accreted onto East Antarctica, forming West Antarctica.) This rifting occurred long before the supercontinent Pangaea--from which the present continents broke off--was formed. Pangaea was assembled only at the end of the Paleozoic era, approximately 250 million years ago. It started to fragment during the Jurassic period of the Mesozoic era, about 170 million years ago, creating the Atlantic and other young ocean basins.

Making our way up a ridge toward the top of the cliff, we saw that the lowest layers of the Cambrian strata--which lie below the limestone--were made of pink conglomerate and coarse sandstones. As the sea advanced over the deepening rift and the subsiding margin, it had ground the Precambrian rocks into boulders, pebbles and sand grains. The deposits became more fine-grained as we climbed, and the quartz sandstones immediately underneath the Nelson Limestone had the appearance of old friends. They were full of vertical worm burrows known as Skolithus.

These tubes are the only traces of ancient filter feeders, which extracted nutrients from sediments and left a clayey residue around their burrows. "Just like western North America," I noted out loud, "but then just like the Durness rocks of northwestern Scotland, too." Indeed, strata deposited by the seawater that advanced to cover most of the continents 540 million years ago--as shown by the presence of Cambrian seashores in such places as Wisconsin--are remarkably similar on all continents.

Matching Mountains...
THERE IS NOTHING like personal experience with rocks, however, to set a geologist thinking. My first impressions of the Transantarctic Mountains in 1987 raised a question that stayed near the forefront of my mind: Could the continent from which Antarctica rifted apart at the end of the Precambrian possibly have been western North America? Or were their margins at that distant time merely in similar environments on either side of an even more ancient Pacific Ocean basin?

The answer has far-reaching implications. The global paleogeography of the time ("paleo" is a prefix that geologists use to indicate "historical") is currently a mystery. To know how the continents were distributed could provide clues to the vast environmental alterations that preceded the Cambrian period. Late in Precambrian times there were several ice ages, and the oceanic and, presumably, atmospheric chemistry changed greatly. Multicellular animals evolved, heralding a biological profusion that included the far-distant ancestors of vertebrates and, hence, of human beings [see "End of the Proterozoic Eon," by Andrew H. Knoll; Scientific American, October 1991].

It is clearly difficult to map out with much certainty the geography of an ancient time on a dynamic planet with continents that move. Alfred Wegener and other pioneers of the theory of continental drift had noted that several North and South American mountain ranges truncated at the Atlantic margins match up neatly across the ocean with mountain ranges in Europe and Africa. Nowadays magnetic data and satellite images of the ocean floor showing fractures--appearing rather like railway tracks, along which the continents slid apart--allow us to reconstruct the supercontinent Pangaea very accurately.

A number of lines of evidence indicate that Pangaea was not the original configuration of the continents. When iron-bearing rocks solidify from lava, they become magnetized in the direction of Earth's magnetic field. The magnetization of rocks that congealed from pre-Mesozoic lava is quite different in North America and Africa, suggesting that in an earlier era these continents moved separately. Volcanic rocks that were fragments of ancient ocean floor have also been found in mountain ranges of Pangaea such as the Famatinian belt (Argentina), the Mozambique belt (Africa) and the older Appalachians. These early Paleozoic and Precambrian ophiolites--as the rocks are called--demonstrate that former ocean basins closed when the supercontinent amalgamated. Struck in the 1960s by the presence of early Paleozoic ophiolites in the Appalachian Mountains of the Maritime provinces in Canada, the imaginative Canadian geophysicist J. Tuzo Wilson asked: "Did the Atlantic Ocean open, close and then reopen?"

In reconstructing continental configurations prior to Pangaea, we get no help from the ocean floors. Although the Pacific Ocean basin already existed, ocean floor of such antiquity has long been thrust under the continents bordering the basin. Geologists therefore have no oceanic "railway map" for continental drift before Pangaea. We have to fall back on evidence from the continents themselves, just as Wegener did when attempting to reconstruct Pangaea before modern oceanography and satellites.

...and Margins
WITHIN PANGAEA there are some ancient continental margins that have no obvious counterparts. The Pacific margins of North and South America, Antarctica and Australia were all formed near the end of the Precambrian, between 750 million and 550 million years ago. The Appalachian margin of Laurentia--the ancestral shield of North America--also rifted away from another continent at that time. Since Wilson asked his famous question, the counterpart to this margin has usually been assumed to have been western Europe and northwestern Africa. But there is no firm evidence for such a juxtaposition.

In 1989 I led another field trip to Antarctica, as part of the International Geological Congress hosted by the U.S. The object of the expedition was to help bring Antarctic geology--long the private domain of a very small group of especially hardy souls (even among geologists)--into the mainstream of global earth science. Various experts on the Himalayas, the European Alps, the Appalachians, the Rockies and many other regions participated.

Soon after, one of these scientists, Eldridge M. Moores, was browsing in the library of the University of California at Davis when he came across a short article by Richard T. Bell and Charles W. Jefferson of the Geological Survey of Canada. They pointed out similarities between Precambrian strata in western Canada and eastern Australia and concluded that the Pacific margins of Canada and Australia might have been juxtaposed. Sensitized by his recent trip, Moores realized this would imply that the Pacific margins of the U.S. and Antarctica had been juxtaposed, a thought similar to my own. After some quick library research, he sent me a map highlighting the structural parallels in the interiors of the Laurentian and East Antarctic shields. "Is this crazy?" he asked.

Similarities in the internal structures of displaced continents can be powerful evidence of former juxtaposition. Moores drew particular attention to a report citing that along the Transantarctic Mountains--in a place called the Shackleton Range (after the famous British explorer Sir Ernest Shackleton)--lie rocks similar in age and character to those underneath much of New Mexico and Arizona. He also pointed out that roughly billion-year-old rocks like those characterizing the Grenville province--an aged band of rocks running along the eastern and southern margin of North America, from Labrador to Texas--had been found near one Antarctic shore. He called his hypothesis--the idea that the continents had been juxtaposed--SWEAT, for Southwest U.S.East Antarctica.

Fired up by the possibility that my question might finally have an answer, I reproduced Mooress reconstruction using the PLATES software at our institute at the University of Texas at Austin. The program allows us to group together pieces of continents and move them over the globe with geometric precision. A short time later my colleague Lisa M. Gahagan and I had removed any uncertainties about matching the boundaries: the scale and general shape of the two old rifted margins were indeed compatible. Moreover, the boundary between the Grenville rocks of Texas and the older rocks of Arizona and New Mexico projected into Antarctica--just where I knew there was a similar boundary under the ice, between the Shackleton Range and some tiny rock outcrops along the frozen shores of the Weddell Sea. It seemed as if the rocks right under my feet, those that form the Llano uplift in Texas and from which the Texas State Capitol was built, were reappearing electronically in Antarctica!

If the western edge of North America was joined to East Antarctica and Australia, then some other continent must have rifted off the Appalachian margin. Paul F. Hoffman, now at Harvard University, and I have suggested that the eastern side of North America's Laurentian shield was wedged against the Precambrian shields of South America, known as Amazonia and Rio de la Plata. In manipulating the three shields on the computer screen, it occurred to me that the Labrador-Greenland prominence of Laurentia might have originated within the recess in the South American margin between Chile and southern Peru, often referred to as the Arica embayment. Both the promontory and the embayment are believed to date from late Precambrian times. But while they are of the same size and general shape, they were extensively modified when the Appalachian and Andean mountain chains rose. So a precise geometric fit is not to be expected.

Crystalline enigma
MY SUGGESTION provides a possible explanation for a long-standing enigma of Andean geology. Along the otherwise youthful and active Peruvian margin are found 1.9-billion-year-old crystalline rocks. Hardolph A. Wasteneys, then at the Royal Ontario Museum, dated zircon crystals from the Arequipa massif, along the coast of southern Peru. He demonstrated that these rocks were highly metamorphosed when North Americas Grenville Mountains were formed, 1.3 billion to 0.9 billion years ago. They may therefore represent a continuation of the Grenville province of eastern and southern North America into South America.

The hypothesis of a South American connection for the eastern margin of Laurentia unexpectedly brought my career full circle. I grew up in Scotland and cut my geologic teeth on its rocks. Northwestern Scotland and the submerged Rockall Plateau--off the western margin of the British Isles--remained part of North America until the North Atlantic Ocean basin had almost finished opening. Scotland was at the apex of the Labrador-Greenland promontory. When nestled (electronically) in the Arica embayment, the rocks of the Scottish Highlands that I studied for my doctoral degree in the 1960s appear to continue into equally old rocks of Peru and Bolivia. Given how well studied the Scottish Highlands are, they may provide critical tests for a former North America-South America connection.

Assuming the SWEAT hypothesis and the Pan-American connection, we can try to reconstruct the global distribution of continents and oceans in the late Precambrian. Most geologists believe that the relative areas occupied by continents and ocean basins have not changed since the late Precambrian. If, therefore, Antarctica, Australia, North America and fragments of South America were fused into a pre-Pangaean supercontinent, now named Rodinia, then there had to have been vast oceans elsewhere. Ophiolitic relics caught up within the continents indicate that these oceans lay between India and today's East Africa (the Mozambique Ocean) and within Africa and South America (the Pan-African and Braziliano oceans, respectively).

Between roughly 750 million and 550 million years ago these ocean basins were destroyed, and all the Precambrian nuclei of Africa, Australia, Antarctica, South America and India amalgamated into the supercontinent of Gondwana. It was during this time interval that the Pacific Ocean basin opened between Laurentia and the East Antarctic-Australian landmass. Isotopic dating of volcanic rocks in Newfoundland shows that the ocean basin between Laurentia and South America did not open until the beginning of the Cambrian. North America may therefore have separated out in a two-stage process.

Reconstructing the travels of North America requires an essential piece of information: the magnetization of ancient rocks. Such data allow geologists to figure out the latitude and orientation of the rocks when they formed. But because Earth's magnetic field is axially symmetrical, paleomagnetic measurements cannot tell us about the original longitude of the rocks. Present-day lava from Iceland and Hawaii, for example, could reveal to a geologist 100 million years from now the latitudes and the orientation of these islands but not their vast difference in longitude. It would not be apparent that the islands are in different oceans.

Traditional reconstructions of Laurentia always place its Appalachian margin opposite northwestern Africa during the Paleozoic era. I decided to plot the relation of North America to Gondwana differently, taking advantage of the fact that the longitude of the continent is not constrained by paleomagnetic data. It turned out that North America could have made what one of my graduate students referred to as an "end run" around South America during the Paleozoic, starting from next to Antarctica.

When Luis H. Dalla Salda, Carlos A. Cingolani and Ricardo Varela of the University of La Plata in Argentina saw the sketch of the end run, they became excited. They had recently proposed that a Paleozoic mountain belt, whose roots are exposed in the Andes of northern Argentina, could have formed when another continent collided with Gondwana. Moreover, the western margin of this Famatinian belt includes Cambrian and Lower Ordovician limestones (between 545 million and 490 million years old) containing trilobites characteristic of North America. Perhaps, they reasoned, this is a "geologic calling card" left behind when North America collided with South America during the Ordovician period, 450 million years ago.

It appears that after rifting from South America at the end of the Precambrian, North America moved quite far away. During the Cambrian period, when Gondwana was undergoing glaciation, North America was equatorial. Ocean floor was then subducted beneath the South American craton, and North and South America collided again during the Ordovician. We think that the older part of the Appalachian Mountains, which terminates abruptly in Georgia, was once continuous with Argentina's Famatinian belt. This construction places Washington, D.C., close to Lima, Peru, during mid-Ordovician times.

End of the Run
AFTER THE ORDOVICIAN collision, the continents separated again, apparently leaving North American limestone with its characteristic trilobites in northwestern Argentina. My Argentine colleagues and I have suggested that these rocks tore off the ancestral Gulf of Mexico, known as the Ouachita embayment. Blocks carried up by Andean volcanoes from below the limestones have recently been dated at around one billion years old, just like those of the Grenville province that probably occupied the embayment.

It is possible that the North and South American continents interacted again before North America finally collided with northwestern Africa to complete Pangaea. French geologists studying the Paleozoic sedimentary rocks of the Peruvian Andes have found that they are made of debris that must have eroded from a neighboring landmass. They assumed this continent, occupying the area now covered by the Pacific Ocean, to have been an extension of the Arequipa massif in Peru.

It may, however, have been North America. As Heinrich Bahlburg of the University of Heidelberg in Germany has pointed out, ancient warm-water North American fauna mingle with cold-water fauna of southern Africa and the Falkland (Malvinas) Islands in the 400-million-year-old (Devonian) strata of northwestern South America. Together with a deformation along the eastern seaboard of North America known as the Acadian orogeny, and the truncation of mountain structures along the South American margin, the evidence points to Laurentias sideswiping northwestern South America during the Devonian. There are even Ordovician limestones with South American trilobites--another calling card--at Oaxaca in Mexico. Only after North America finally moved away from the proto-Andean margin did the Andean Cordillera of the present day begin to develop.

Some 150 million years later North America returned to collide with northern Europe, Asia and Gondwana. Pangaea--with the Urals, the Armorican Mountains in Belgium and northern France, the Ouachitas and the youngest Appalachians as sutures--arose from the collisions of these continents. After a 500-million-year odyssey, North America had finally found a resting place. But not for long. In another 75 million years it separated from Africa as Pangaea broke up, to move toward its current position.

During the southern summer of 19931994--six years after my first glimpse of the Pensacola Mountains and glimmerings of North America''s odyssey--I returned to Antarctica. This time, with my colleague Mark A. Helper, two graduate students and two mountaineers, I explored the Shackleton Range and Coats Land near the Weddell Sea. According to my computer simulations, this is where North Americas Grenville rocks had projected 750 million years ago. Antarctic geologists have long regarded these areas as anomalous.

At the end of our visit to Coats Land, we roped together, picked up our ice axes and climbed back to another small aircraft. Weighing down our packs--and the aircraft, which groaned into the air--were the rock samples we had gleaned that day. In the laboratories of my colleagues Wulf A. Gose and James N. Connelly, we sat down to analyze those rocks.

Persuasive evidence
OUR IDEAS about how Earth looked before Pangaea, first described in this magazine in 1995, have stimulated a great deal of activity within the geologic community. They offered the first testable hypothesis regarding global geography in late Precambrian and early Paleozoic times--the critical era when single-celled organisms evolved into soft-bodied multicellular creatures, then invertebrates with hard shells and, ultimately, primitive vertebrates.

Over the past decade, interest in the Rodinian supercontinent that preceded Pangaea has spawned research centers and international programs to study this supercontinents assembly, geography and fragmentation. One related result of this scientific ferment is the "snowball Earth" hypothesis, which proposes that Earth was covered with ice at sea level all the way to the equator 600 million to 700 million years ago, at the time of Rodinias fragmentation and the formation of the Pacific Ocean basin.

The snowball Earth hypothesis posits an extreme global environment that challenges our understanding of climate past, present and future. If confirmed, it would mean that a dramatically chilly period directly preceded the explosion of multicellular life that occurred approximately 545 million years ago. Because forecasters rely on the distribution of continental landmasses in designing computer climate models, our rather esoteric study of ancient supercontinents has clearly taken on added significance in recent years.

With the absence of ocean floor predating Pangaea and the fragmentary nature of evidence from the continents, opinions regarding this period of Earth history inevitably differ. Some experts even doubt the very existence of the late Precambrian Rodinian supercontinent described in this article--doubts difficult to reconcile with the thousands of kilometers of preserved late Precambrian rifted continental margins.

Other researchers have used the same data we have relied on to reach radically different notions of the way this pre-Pangaean supercontinent may have looked. Instead of a connection between the southwestern U.S. and East Antarctica, for example, some experts propose that the U.S. Southwest and Mexico were connected to southeastern Australia. And an older idea has been revived, to the effect that Siberia was rifted from the proto-Pacific margin of North America. Nevertheless, two lines of evidence persuade me that our concept of the way Earth looked before Pangaea is the correct one.

First are the fruits of our 19931994 trip to Antarctica: the rock specimens we obtained from Coats Land. The paleomagnetic data obtained from those rocks do indeed show that this part of Antarctica could have been adjacent to the core of present-day North America when the rocks formed as volcanic deposits some 1.1 billion years ago. Extensive lava flows of this age lie exposed near Lake Superior and extend in the subsurface through Kansas to Trans-Pecos Texas, the Keeweenawan province. Although identical deposits exist throughout southern Africas Umkondo province, my colleagues Jim Connelly here at Austin and Staci Loewy of the University of North Carolina at Chapel Hill have demonstrated that our Coats Land rocks contain lead isotopes that match those of North America's Keeweenawan province--but are quite distinct from the isotopic composition of the Umkondo lavas of Africa.

Second, evidence increasingly suggests that lower Paleozoic limestones of the Precordillera of northwestern Argentina originated in North America--yet another geologic calling card revealing North America's former presence off the Pacific margin of South America. Workers from both continents who have analyzed the rocks of Argentinas Precordillera have shown unequivocally that they originated in North America.

It remains unclear whether this ancient North American limestone arrived in South America as a Madagascar-like microcontinent or through transfer resulting from a continent-continent collision--as Italy was much later transferred from Africa to Europe when those two continents collided. Yet however they were transferred to South America, these limestone rocks offer the strongest possible evidence that North America did indeed make an end run around the Pacific margin of South America [see illustration on page 17] during Paleozoic times and that ancestral North America probably originated somewhere between the present Antarctic-Australian and South AmericanAfrican parts of a pre-Pangaean supercontinent.

IAN W. D. DALZIEL has been studying the geology of Antarctica, the Andes, the Caledonides and the Canadian Shield since earning his Ph.D. at the University of Edinburgh in 1963. Currently he is research professor and associate director at the Institute for Geophysics of the Jackson School of Geosciences at the University of Texas at Austin. In 1992 Dalziel received the Geological Society of Londons Murchison Medal. In addition to his extensive geologic travels, he loves to visit wild places, preferably with his family. When in Austin, he sculls on Town Lake.