PLATE TECTONICS

Encyclopædia Britannica Article

By Tjeerd H. van Andel, Stanford University

Theory dealing with the dynamics of the Earth's outer shell, the lithosphere. According to the theory, the lithosphere consists of about a dozen large plates and several small ones. These plates move relative to each other and interact at their boundaries, where they diverge, converge, or slip relatively harmlessly past one another. Such interactions are thought to be responsible for most of the seismic and volcanic activity of the Earth, although earthquakes and volcanoes are not wholly absent in plate interiors. While moving about, the plates cause mountains to rise where they push together and continents to fracture and oceans to form where they pull apart. The continents, sitting passively on the backs of plates, drift with them and thereby bring about continual changes in the Earth's geography.

The theory of plate tectonics, formulated during the late 1960s, rests on a broad synthesis of geologic and geophysical data. It is now almost universally accepted and has had a major impact on the development of the Earth sciences. Its adoption represents a true scientific revolution, analogous in its consequences to the Rutherford and Bohr atomic models in physics or the discovery of the genetic code in biology. Incorporating the much older idea of continental drift, the theory of plate tectonics has made the study of the Earth more difficult by doing away with the notion of fixed continents, but it has at the same time provided the means of reconstructing the past geography of continents and oceans. While its impact has, to a considerable degree, run its course in marine geology and shows signs of reaching the limits of usefulness in the study of mountain-building processes, its influence on the scientific understanding of the Earth's history, of ancient oceans and climates, and of the evolution of life is only beginning to be felt.

For details on the specific effects of plate tectonics, see earthquake and volcano. A detailed treatment of the various land and submarine relief features associated with plate motion is provided in tectonic landform and oceans.

Principles of plate tectonics

The plate tectonics theory has a long and tortuous history. Yet, the theory itself is elegantly simple.

The surface layer of the Earth, from 50 to 100 kilometres (31 to 62 miles) thick, is assumed to be composed of a set of large and small plates, which together constitute the rigid lithosphere. The lithosphere rests on and slides over an underlying, weaker layer of partially molten rock known as the asthenosphere. The constituent lithospheric plates move across the Earth's surface, driven by forces as yet not fully agreed upon, and interact along their boundaries, diverging, converging, or slipping past each other (Figure 1). While the interiors of the plates are presumed to remain essentially undeformed, their boundaries are the sites of many of the principal processes that shape the terrestrial surface, including earthquakes, volcanism, and orogeny (i.e., the deformation that builds mountain ranges).

The most conspicuous feature of the Earth's surface is its division into continents and ocean basins, a division that owes its existence to differences in thickness and composition between the continental and the oceanic crust. The continents have a crust of granitic composition and hence are somewhat lighter than the basaltic ocean floor. Also, they are 30 to 40 kilometres thick as compared to the oceanic crust, which measures only 6 to 7 kilometres in thickness. Their greater buoyancy causes them to float much higher in the mantle than does the oceanic crust, thus accounting for the difference between the two principal levels of the Earth's surface. The boundary between the continental or oceanic crust and the underlying mantle, the Mohorovicic discontinuity, has been clearly defined by seismic studies.

Plate boundaries

As conceived by the theory of plate tectonics, the lithospheric plates are much thicker than the oceanic or the continental crust; their boundaries do not usually coincide with those between oceans and continents; and their behaviour is only partly influenced by whether they carry oceans, continents, or both. The Pacific Plate, for example, is purely oceanic, but most of the others contain continents.

At a divergent plate boundary, magma wells up from below as the release of pressure produces partial melting of the underlying mantle and generates new crust. Because the partial melt is basaltic in composition, the new crust is oceanic. Consequently, diverging plate boundaries, even if they originate within continents, eventually come to lie in ocean basins of their own making. In fact, most divergent plate boundaries seem to have formed within continents rather than in oceans, probably because a hot, weak layer, sandwiched at a depth of about 15 kilometres between two stronger ones, renders the continental crust more vulnerable to fragmentation than its oceanic counterpart. The creation of the new crust is accompanied by much volcanic activity and by many shallow tension earthquakes as the crust repeatedly rifts, heals, and rifts again.

The continuous formation of new crust produces an excess that must be disposed of elsewhere. This is accomplished at convergent plate boundaries where one plate descends—i.e., is subducted—beneath the other. At depths between 300 and 700 kilometres, the subducted plate melts and is recycled into the mantle. Because the plates form an integrated system that completely covers the surface of the Earth, it is not necessary that new crust formed at any given divergent boundary be completely compensated at the nearest subduction zone, as long as the total amount of crust generated equals that destroyed.

It is in subduction zones that the difference between plates carrying oceanic and continental crust can be most clearly seen. If both plates have oceanic edges, either one may dive beneath the other; but, if one carries a continent, the greater buoyancy prevents this edge from sinking. Thus, it is invariably the oceanic plate that is subducted. Continents are permanently preserved in this manner, while the ocean floor continuously renews itself. If both plates possess a continental edge, neither can be subducted and a complex sequence of events from crumpling to under- and overthrusting raises lofty mountain ranges. Much later, after these ranges have been largely leveled by erosion, their remains continue as a reminder that this is the “suture” where continents were once fused.

The subduction process, which involves the descent into the mantle of a slab of cold rock about 100 kilometres thick, is marked by numerous earthquakes along a plane inclined 30°–60° into the mantle—the Benioff zone. Most earthquakes in this planar dipping zone result from compression, and the seismic activity extends 300–700 kilometres below the surface. At a depth of 100 kilometres or more the subducted oceanic sediments, together with part of the upper basaltic crust, melt to an andesitic magma, which rises to the surface and gives birth to a line of volcanoes a few hundred kilometres behind the subducting boundary. This boundary is usually marked by an oceanic deep, or trench, where the overriding plate scrapes off the upper crust of the lower plate to create a zone of highly deformed, largely sedimentary rock. If both plates are oceanic, the deformed sediments and volcanoes form two island arcs parallel to the trench. If one plate is continental, the sediments are usually accreted against the continental margin and the volcanoes form inland, as they do in Mexico or western South America.

Along the third type of plate boundary, two plates move laterally and pass each other without creating or destroying crust. Large earthquakes are common along such strike-slip, or transform, boundaries. Also known as fracture zones, these plate boundaries are perhaps best exemplified by the San Andreas fault in California and the North Anatolian fault system in Turkey.

Most of the seismic and volcanic activity on Earth is therefore concentrated along plate boundaries where mid-ocean ridges, trenches with island arcs, and mountain ranges are generated. Some seismic and volcanic activity also occurs within plates. Interesting examples of this interplate activity are linear volcanic chains in ocean basins, such as the Hawaiian Islands and their westward continuation as a string of reefs and submerged seamounts. An active volcano usually exists at one end of an island chain of this type, with progressively older extinct volcanoes occurring along the rest of the chain. Such topographic features have been explained by J. Tuzo Wilson of Canada and W. Jason Morgan of the United States as the product of “hot spots,” magma-generating centres of controversial origin located deep in the mantle far below the lithosphere. A volcano builds at the surface of a plate positioned above a hot spot. As the plate moves on, the volcano dies, is eroded, and eventually sinks below the surface of the sea, while a new one forms above the hot spot. Hot spot volcanism is not restricted to the ocean basins; other manifestations occur within continents, as in the case of Yellowstone National Park in western North America.

Plate motion

The movement of a plate across the surface of the Earth can be described as a rotation around a pole, and it may be rigorously described with the theorem of spherical geometry formulated by the Swiss mathematician Leonhard Euler during the 18th century. Similarly, the motions of two plates with respect to each other may be described as rotations around a common pole, provided that the plates retain their shape (Figure 2). The requirement that plates are not internally deformed has become one of the postulates of plate tectonics. It is not totally supported by evidence, but it appears to be a reasonable approximation of what actually happens in most cases. It is needed to permit the mathematical reconstruction of past plate configurations.

The joint pole of rotation of two plates can be determined from their transform boundaries, which are by definition parallel to the direction of motion and so form small circles around the pole. The rate of motion can be computed from the increase in age of the crust away from the divergent plate boundary usually by means of magnetic anomalies (see below Hess's seafloor-spreading model). Because all plates form a closed system, all movements can be defined by dealing with them two at a time.

It is, of course, conceivable that the entire lithosphere might slide around over the asthenosphere like a loose skin, altering the positions of all plates with respect to the spin axis of the Earth and the equator. To determine the true geographic positions of the plates in the past, which is so important in paleoclimatology and paleoceanography, investigators have to define their motions, relative not to each other but rather to this independent frame of reference. The hot spot island chains serve this purpose, their trends providing the direction of motion of a plate; the speed of the plate can be inferred from the increase in age of the volcanoes along the chain. It is assumed, of course, that the hot spots themselves remain fixed with respect to the Earth, an assumption that appears to be reasonably accurate for at least some hot spots.

Quite another method of determining absolute plate movements relies on the fact that the equatorial waters of the ocean are, and always have been, very fertile. The high biological productivity yields an enormous quantity of calcareous microfossils, which, like a gigantic natural chalk line, marks a narrow equatorial zone. The displacement of the equatorial deposits over time, traced by means of deep-sea drill cores, enables investigators to determine the direction and rate of plate movement.

Because the plates all interlock, any change in motion anywhere must reverberate throughout the entire system. If two continents collide, their edges will crumple and shorten, but eventually all motion must stop at this boundary and large parts of the system elsewhere will have to adjust. Earth scientists are thus able to reconstruct the positions and movements of plates in the past so long as they have the ancient oceanic crust to provide them with plate speeds and directions. Since old oceanic crust is continuously consumed to make room for new crust, this kind of evidence is eventually exhausted. Little oceanic crust of Early Cretaceous age (about 144 to 97.5 million years old) remains, and none is older than the Jurassic—the geologic period that began approximately 208 million years ago. Consequently, this method fails for the history of drifting continents during earlier geologic periods (i.e., the Paleozoic and Precambrian eras), making it necessary for investigators to turn to another, less effective technique (see below Paleomagnetism, polar wandering, and continental drift).

Historical overview

Precursors

Any major new idea in science appears to lead instantly to a search of the past for those who might once have proposed similar concepts and with whom the current proponents should therefore share the credit. In the case of plate tectonics, the primary candidate is obvious: Alfred Wegener of Germany, who explicitly presented the concept of continental drift for the first time at the outset of the 20th century. Though plate tectonics is by no means synonymous with continental drift, it encompasses this idea and derives much of its impact from it.

There might have been predecessors even to Wegener. The outlines of the continents bordering the Atlantic Ocean are so similar that many probably noticed the correspondence, and some might have drawn the conclusion that the lands on both sides were once joined together. The earliest reference to this peculiar geographic feature was made by the English philosopher Francis Bacon. In his Novum Organum (1620), Bacon pointed out the correspondence but did not go beyond that. Such was also the extent of the contribution of the great French naturalist Georges-Louis Leclerc, Count du Buffon, a century later. Neither can François Paget qualify as a forerunner of continental drift theorists: even though he stated in 1666 that an undivided continent existed before Noah's flood, he explained the creation of the Atlantic Ocean by having part of that continent sink into the sea.

The first credible proponent of continental drift was Antonio Snider-Pellegrini, a belated advocate of catastrophism (the view that geologic history consists of a sequence of numerous violent catastrophic events), who in 1858 ascribed the biblical flood to the former existence of a single continent that was torn apart to restore the balance of a lopsided Earth. More recent and much more sophisticated was the work of the American geologist Frank B. Taylor, who, disdaining the then-prevailing contraction model of mountain building, postulated in 1908 that the arcuate mountain belts of Asia and Europe resulted from the equatorward creep of the continents. His analysis of tectonic features foreshadowed in many ways modern thought regarding plate collisions, and he anticipated Wegener's publications by only a few years. Curiously, however, his work instantly sank into oblivion.

Alfred Wegener and the concept of continental drift

Wegener was by training and profession a meteorologist (he was highly respected for his work in climatology and paleoclimatology), but he is best remembered for the foray into geology that led to his formulation of the concept of continental drift. In 1910, again because of the geography of the Atlantic coastlines, Wegener came to consider the existence of a single supercontinent during the late Paleozoic era (about 350 to 225 million years ago) and named it Pangaea. He searched the geologic and paleontological literature for evidence attesting to the continuity of geologic features across the Indian and Atlantic oceans, which he assumed had formed during the Mesozoic era (about 245 to 66.4 million years ago). His efforts proved rewarding, and he presented the idea of continental drift and some of the supporting evidence in a lecture in 1912. This was followed in 1915 by his major work, Die Entstehung der Kontinente und Ozeane (The Origin of Continents and Oceans).

The idea of large ancient continents composed of several of the present-day smaller ones had been put forth in the late 19th century by the Austrian geologist Eduard Suess. Suess, however, was not thinking of continental drift. In the spirit of his day, he assumed that portions of a single enormous southern continent—designated Gondwana, or Gondwanaland—foundered to become the Atlantic and Indian oceans. Such sunken lands, along with vanished land bridges, were frequently invoked in the late 1800s to explain sediment sources apparently present in the ocean and to account for floral and faunal connections between continents. They remained popular until the 1950s, stimulated believers in ancient Atlantis, and even made their way into literary works.

Yet, it was already known that the concept of isostasy rendered large sunken continental blocks geophysically impossible, and Wegener characteristically introduced his continental drift proposal by pointing this out. Only then did he proceed to conclude that, if the continents had been once joined together, drift of their fragments rather than their foundering would have been the consequence. The assumption of a former single continent could be tested geologically, and Wegener next displayed a large array of data. Even today his evidence, ranging from the continuity of fold belts across oceans and similarities of sequences of strata on their opposite sides to paleobiogeographic and paleoclimatological arguments, would be judged worthy of serious consideration. He further argued that, if continents could move up and down in the mantle as a result of buoyancy changes produced by, say, erosion or deposition, they should be able to move horizontally as well. The driving forces he considered, however, were unconvincing: both pole fleeing and the westward tidal force appeared to most to be entirely inadequate.

Wegener's proposition was attentively received by many European geologists, and in England Arthur Holmes pointed out that the lack of a driving force was hardly sufficient grounds to scuttle the entire concept. As early as 1929, Holmes proposed an alternative mechanism—namely, convection of the mantle, which remains today a serious candidate for the force driving the plates. Wegener's ideas also were appreciated by geologists in the Southern Hemisphere. One of them, the South African Alexander Du Toit, remained a lifelong believer. After Wegener's death, Du Toit continued to amass further evidence in support of continental drift.

Evidence supporting the hypothesis

Much was thus to be said for the idea that the continents were joined together in the Paleozoic, and supporting evidence has continued to accumulate to this day. The opposing Atlantic shores match well, especially at the 1,000-metre (3,300-foot) depth contour, which is a better approximation of the edge of the continental block than the present shoreline, as Sir Edward Bullard cogently demonstrated in 1964 with the aid of computer analysis. Similarly, the structures and stratigraphic sequences of Paleozoic mountain ranges in eastern North America and northwestern Europe can be matched in detail. This fact was already known to Wegener and has been strengthened substantially in subsequent years.

Often cited as evidence have been the strikingly similar Paleozoic sequences on all southern continents and also in India. This Gondwana sequence—so called after one of Suess's large continents—consists of glacial tillites, followed by sandstones and finally coal measures. Placed on a reconstruction of Gondwana, the tillites mark two ice ages that occurred during the long march of this continent across the South Pole from its initial position north of Libya about 500 million years ago until its final departure from southern Australia 250 million years later (Figure 3). The first of these ice ages left its glacial deposits in the southern Sahara during the Silurian period (which extended from about 438 to 408 million years ago), and the second did the same in southern South America, South Africa, India, and Australia from 380 to 250 million years ago. At each location the tillites were subsequently covered by desert sands of the subtropics, and these in turn by coal measures, indicating that the region had arrived near the equator.

During the 1950s and '60s, patient work in isotopic dating showed that the massifs of Precambrian time (from about 3.8 billion to 570 million years ago) found on opposite sides of the South Atlantic did indeed closely correspond in age and composition, as Wegener had surmised. It is now evident that they originated as a single assemblage of Precambrian continental nuclei later torn apart by the breakdown of Pangaea.

Disbelief and opposition

More common than interest or approval, however, was a disbelief so strong that it often bordered on indignation. One of the strongest opponents was the British geophysicist Sir Harold Jeffreys, who spent years attempting to demonstrate that continental drift is impossible because the strength of the mantle should be far greater than any conceivable driving force. He refused to abandon this viewpoint in spite of the massive evidence in favour of plate tectonics. It was in North America, however, that opposition to Wegener's ideas was vigorous to the point of excess and very nearly unanimous. This is illustrated by reports from a symposium on continental drift organized in 1928 by the American Association of Petroleum Geologists. The mood that prevailed at the gathering was expressed by an unnamed attendant quoted with sympathy by the great American geologist Thomas C. Chamberlin: “If we are to believe Wegener's hypothesis, we must forget everything which has been learned in the last 70 years and start all over again.” The same reluctance to start anew was again displayed some 40 years later by the same organization when its publications provided the principal forum for the opposition to plate tectonics.

Wegener was attacked from virtually every possible vantage point, his paleontological evidence attributed to land bridges, the similarity of strata on both sides of the Atlantic called into question, the fit of Atlantic shores declared inaccurate, and his very competence doubted. It also did not escape attention that he did not possess proper credentials as a geologist.

What might have been the cause of this overwhelmingly negative response in the light of such a substantial amount of supporting evidence? The unsatisfactory quality of Wegener's driving mechanism has commonly been cited as the reason, but that seems too simple, especially since the absence of a mechanism did not delay the acceptance of plate tectonics. The roots of the resistance most certainly reached far deeper. It would be unusual for the practitioners of any science to flock to a new concept—particularly a revolutionary one of such profound consequences—before the need for a thorough overhaul of the existing conceptual edifice had become compelling and obvious to most, its supporting evidence daily crumbling, and its explanatory power reduced below any acceptable level.

Whatever the cause, continental drift, having been rejected by the vast majority of geologists the world over, retreated into obscurity and remained there for roughly three decades. Ironically, though, Du Toit so successfully kept the fires burning in the Southern Hemisphere that it remained quite respectable there to profess oneself an adherent of continental drift during the very years that such a confession north of the equator would have exposed one to ridicule and disbelief.

Renewed interest in continental drift

Paleomagnetism, polar wandering, and continental drift

The fact that some rocks are strongly magnetized has been known for centuries, and geologists recognized more than 100 years ago that many rocks preserve the imprint of the Earth's magnetic field as it was at the time of their formation. Volcanic rocks such as basalt are especially good recorders of paleomagnetism, but some sediments also align their magnetic particles with the Earth's field at the time of deposition. Investigators therefore have at their disposal fossil compasses that indicate, like any magnet suspended in the Earth's field, the direction to the magnetic pole and that yield the latitude of their origin.

During the 1950s, paleomagnetic studies, notably those of Stanley K. Runcorn and his coworkers in England, showed that in the late Paleozoic the north magnetic pole—as seen from Europe—seems to have wandered from a Precambrian position near Hawaii to its present location by way of Japan. This, of course, might mean that the magnetic pole itself had migrated or that Europe had moved relative to a fixed pole. Either continental drift or polar wandering was therefore a reasonable explanation. Paleomagnetic data from other continents, however, soon yielded apparent polar wandering paths different from the European one. Separate wanderings of many magnetic poles are not acceptable, but the paths could be brought to coincide by joining the continents in the manner and at the time suggested by Wegener.

Impressed by this result, Runcorn became the first of a new generation of geologists and geophysicists to accept continental drift as a serious proposition worthy of careful testing. Yet, the band of the converted remained small, and most geologists found sufficient reason to doubt the paleomagnetic results, which were often conflicting due to the primitive nature of the early techniques. Since then, more sophisticated methods capable of removing the overprint of later magnetizations have made paleomagnetic data strong supporting evidence for continental drift and a major tool for reconstructing the geography of the past. In the meantime, however, much progress had been made on an entirely different and quite independent front. It was from evidence extracted from the oceanic crust that plate tectonics would be born.

Gestation and birth of plate tectonics

When Wegener developed his ideas, and for many years thereafter, relatively little was known regarding the nature of the ocean floor. After World War II, however, rapid advances were made in the study of the relief, geology, and geophysics of the ocean basins. Due in large part to the efforts of Bruce C. Heezen and Henry W. Menard of the United States, these features, which constitute more than two-thirds of the Earth's surface, became well enough known to permit serious geologic analysis.

Several major topographic and tectonic features distinguish the ocean basins from the continents. The first of these is the mid-ocean ridge system. Mid-ocean ridges are broad, elongated elevations of the ocean floor rising to about 2.5 or 3 kilometres below sea level, with widths ranging from a few hundred to more than 1,000 kilometres. Their crests tend to be rugged and are often endowed with a longitudinal rift valley where fresh lava flows, high heat flow, and shallow earthquakes of the extensional type are found. Mid-ocean ridges nearly girdle the globe.

Trenches constitute another type of seafloor feature. In contrast to mid-ocean ridges, they are long, narrow depressions containing the greatest depths of the ocean basins. Trenches virtually ring the Pacific; a few also occur in the northeastern part of the Indian, and some small ones are found in the Atlantic. Elsewhere they are absent. Trenches have low heat flow, are often filled with thick sediments, and lie at the upper edge of the Benioff zone of compressive earthquakes. Trenches border continents, as in the case of western Central and South America; but they also may occur in mid-ocean, as, for example, in the southwestern Pacific.

Mid-ocean ridges and, more rarely, trenches are offset by fracture zones—transverse features consisting of linear ridges and troughs approximately perpendicular to the ridge crest that they offset by a few to several hundred kilometres. Fracture zones often extend over long distances in the ocean basins but generally end abruptly against continental margins. They are not volcanic, and their seismic activity is restricted to the area between offset ridge crests where earthquakes indicating horizontal slip are common.

Hess's seafloor-spreading model

The existence of these three types of striking, large seafloor features, which had gradually become evident during the late 1940s and '50s, clearly demanded a global rather than local tectonic explanation. The first comprehensive attempt at such an explanation was made by Harry H. Hess of the United States in a widely circulated manuscript that he had written in 1960 but not formally published for several years. In this paper Hess, drawing on Holmes's model of convective flow in the mantle, suggested that the mid-ocean ridges were the surface expressions of rising and diverging convective flow while trenches and Benioff zones with their associated island arcs marked descending limbs. At the ridge crests new oceanic crust would be generated and then carried away laterally to cool, subside, and finally be destroyed in the nearest trenches. Consequently, the age of the oceanic crust should increase with distance away from the ridge crests, and, because recycling was its ultimate fate, very old oceanic crust would not be preserved anywhere. This, incidentally, took care of an old and troubling paradox: only rocks younger than Mesozoic had ever been encountered in the oceans, whereas the continents bear ample evidence of the presence of oceans for more than three billion years.

Hess's model, later dubbed seafloor spreading by the American oceanographer Robert S. Dietz, appeared to account for most observations and was received with interest by many marine geologists. Confirmation of the production of oceanic crust at ridge crests and its subsequent lateral transfer was not long in coming. Fracture zones had thus far been widely regarded as transcurrent faults that gradually displaced one crestal block to the right or left relative to the other. Given this interpretation, the abrupt termination of many fracture zones against continental margins raised intractable problems. The aforementioned Canadian geologist J. Tuzo Wilson solved these problems in 1965 with a single ingenious stroke. Suppose, he argued, that the offset between two ridge crest segments is present at the outset. Each segment generates new crust, which moves laterally away. Along that part of the fracture zone lying between crests, the crustal slabs move in opposite directions, even though the axes or rift valleys themselves remain stationary. Beyond the crests, adjacent portions of crust move in parallel and are eventually absorbed in a trench. Wilson called this a transform fault and noted that on such a fault the seismicity should be confined to the part between ridge crests, as is indeed the case. Shortly afterward Lynn R. Sykes, an American seismologist, showed that the motions deduced from earthquakes on transform faults conform to the directions of motion postulated by Wilson and are opposite those observed on a transform fault.

The seafloor-spreading model also required that the oceanic crust should increase in age with distance from the ridge axis, and Wilson had already pointed out that volcanic islands in the Atlantic indeed show this pattern. Such islands are few, however, and it is in the nature of these piles of lava and ash that the moment of their birth is difficult to ascertain. Additional evidence was needed, and it soon came from magnetic surveys of the oceanic crust.

A magnetic survey of the eastern Pacific floor off the coast of Oregon and California had been published in 1961 by two geophysicists, Arthur D. Raff and Ronald G. Mason. The results were puzzling and gave rise to many farfetched interpretations. Unlike on the continents, where magnetic anomaly patterns tend to be confused and seemingly random except on a fine scale, the seafloor possesses a remarkably regular set of magnetic bands alternately higher and lower than the average Earth field. These positive and negative anomalies are strikingly linear and parallel with the mid-ocean ridge axis, show distinct offsets along fracture zones, and, when displayed in black and white, generally resemble the pattern of a zebra skin. The axial anomaly tends to be higher and wider than the adjacent ones, and in most cases the sequence on one side is the approximate mirror image of that on the other.

In his convection/seafloor-spreading model, Hess had attributed the formation of the oceanic crust mainly to the hydration of a peridotitic mantle, a process not judged likely to produce such regular magnetic anomalies. Alternatively, it seemed possible that partial melting of the mantle might yield a basaltic magma, which, after congealing, would be a much better medium for retaining a strong imprint of the Earth's magnetic field. This second hypothesis has since been amply confirmed by deep-sea dredging and drilling.

It had further been known since early in the century that the polarity of the Earth's magnetic field reverses from time to time. Studies of the remanent magnetism of stacks of basalt lavas extruded in rapid succession on land had, since the late 1950s, begun to establish a sequence of reversals dated by isotopic methods.

The Vine-Matthews hypothesis

Assuming that the oceanic crust is indeed made of basalt intruded in an episodically reversing geomagnetic field, Drummond H. Matthews of Cambridge University and a research student, Frederick J. Vine, postulated in 1963 that the new crust would assume a magnetization aligned with the field at the time of its formation. If the field were normal, as it is today, the magnetization of the crust would be added to that of the Earth and produce a positive anomaly. If intrusion had taken place during a period of reverse magnetic polarity, it would subtract from the present field and appear as a negative anomaly. Subsequent to intrusion, each new block would split and the halves, in moving aside, would generate the observed bilateral magnetic symmetry. Given a constant rate of crustal generation, the widths of individual anomalies should correspond to the intervals between magnetic reversals. Correlation of magnetic traverses from different mid-ocean ridges demonstrated in 1966 an excellent correspondence with the magnetic polarity-reversal time scale just then published by the American geologists Allan Cox, Richard Doell, and Brent Dalrymple in a series of timely papers. This reversal time scale went back some three million years, but since then further extrapolation based on marine magnetic anomalies (confirmed by deep-sea drilling) has extended the magnetic anomaly time scale far into the Cretaceous period, which spanned from about 144 to 66.4 million years ago.

As an aside, it is of interest to note that in Canada Laurence W. Morley, simultaneously with Vine and Matthews but entirely independently, had come to the same explanation for the marine magnetic anomalies, but publication of his paper, delayed by unsympathetic referees and technical problems, occurred long after Vine's and Matthews' work had already firmly taken root.

These confirmations persuaded a large number of marine geologists that seafloor spreading was a reality. Continental drift, however, was not much in their minds, as they focused mainly on the explanations that the concept provided for a host of oceanic features. Land geologists were disinterested, viewing the affair as primarily an issue for their marine colleagues.

Two concerns, however, remained. The spreading seafloor was generally seen as a thin skin most likely having its base at the Mohorovicic discontinuity—i.e., boundary between the crust and mantle considered of such major importance in the early 1960s that plans were undertaken to sample it by deep drilling in the oceans. If only oceanic crust were involved, as seemed to be the case in the Pacific Ocean, the thinness of the slab was not disturbing, even though the ever-increasing number of known fracture zones with their close spacing implied oddly narrow, long convection cells. More troubling was the fact that the Atlantic Ocean, though it had a well-developed mid-ocean ridge, lacked trenches adequate to dispose of the excess oceanic crust. There, the adjacent continents needed to travel with the spreading seafloor, a process that, given the thin but clearly undeformed slab, strained credulity.

Toward a unifying theory

Working independently but along very similar lines, Dan P. McKenzie and Robert L. Parker of Britain and W. Jason Morgan of the United States resolved these issues. McKenzie and Parker showed with a geometric analysis that, if the moving slabs of crust were thick enough to be regarded as rigid and thus to remain undeformed, their motions on a sphere would lead precisely to those divergent, convergent, and transform boundaries that are indeed observed. Morgan demonstrated that the directions and rates of movement had been faithfully recorded by magnetic anomaly patterns and transform faults. He also proposed that the plates extended approximately 100 kilometres to the base of a rigid lithosphere, which had long been known to be underlain by a weaker asthenosphere marked by strong attenuation of earthquake waves. In 1968 the French geophysicist Xavier Le Pichon refined these propositions with a computer analysis of all plate data and proved that they did indeed form an integrated system where the sum of all crust generated at mid-ocean ridges is balanced by the cumulative amount destroyed in all subduction zones (Figure 4 ). That same year, the American geophysicists Bryan Isacks, Jack Oliver, and Lynn R. Sykes showed that the theory, which they enthusiastically labeled the “new global tectonics,” was capable of accounting for the larger part of the Earth's seismic activity. Almost immediately others began to consider seriously the ability of the theory to explain mountain building and sea-level changes.

Only a few years later, details of the processes of plate movement and of boundary interactions, along with much of the plate history of the Cenozoic era (the past 66.4 million years), had been worked out. Yet, the driving forces—notwithstanding a brief flurry of discussion around 1970—remained mysterious and continue as such. The vast accumulation of data bearing on plate history and plate processes has yielded surprisingly little information about what happens beneath them. Pull by the subducting slab, push at the spreading ridge, convection in the asthenosphere, and even tidal forces have been considered, but in every case the evidence has to be admitted as inconclusive. Many favour convection, but, if this indeed is the driving force, the flow pattern at depth is clearly not reflected in the surface movements of the plates, constrained as they are by each other.

Plate tectonics as an explanation for earth processes

Since its inception in 1967–68, plate tectonics has had a pervasive impact on the Earth sciences. If it has not fully lived up to the proud name “new global tectonics,” it has nevertheless exerted enormous influence by clarifying areas of obscurity, reconciling seemingly conflicting evidence, unifying to a remarkable degree events occurring in distant parts of the globe, establishing new pathways toward knowledge, and opening the door to new subjects of investigation, all the while raising a myriad of new and important issues. A few examples may suffice to illustrate the dimensions of the revolution it has wrought.

Pangaea and its breakup

Magnetic anomalies, transform faults, and hot spot trails permit rigorous geometric reconstructions of past plate shapes, configurations, and movements. As previously noted, reasonable agreement on such reconstructions can be achieved as far back as the late Cretaceous, but beyond that time little oceanic crust remains. Much effort in reconstructing the geography of the past 100 million years or so has shown that, though some reconstructions are straightforward, others imply that the strict application of the assumption that plates are not internally deformed is not justified. Such is the case, for example, for the Pacific basin as opposed to the simple history of the Atlantic.

When dealing with geologic time beyond the Cretaceous, other means have to be applied in reconstructing ancient geography. Geologic data, for instance, can be used in the time-honoured manner to determine the proper fit of continents to make Pangaea. Only minor controversy continues here, such as in the vexing case of the position of Madagascar or of the many fragments that now constitute the Mediterranean, southeastern Europe, and the Middle East. Old collision boundaries are marked by sutures—i.e., zones of deformation that betray points of contact between formerly separate continents even after much erosion. Good examples include the Urals between the old central European and Siberian continental blocks, and the Indus suture north of the Himalayas between India and Asia. Some sutures, on the other hand, are well concealed, as, for example, the one in northern Florida and southern Georgia that marks a collision between North America and northwestern Africa during the Paleozoic.

Having reconstructed Pangaea and identified the pieces that came together to form it through largely traditional means, investigators are able to trace continental migrations using paleomagnetic data to obtain their paleolatitudes and orientations for appropriate instants in time. What cannot be determined are their paleolongitudes, so that knowledge about the widths of intervening oceans remains limited. Paleolongitudes can be estimated only by relatively crude means such as faunal and floral affinities. Accordingly, Earth scientists will never know as much about the geography of the Precambrian, Paleozoic, or early Mesozoic as they do about the last 100 million years of Earth history. Yet, enough information has become available to permit some undoubtedly flawed but fascinating and useful reconstructions.

A widely used Paleozoic reconstruction by a group of geologists at the University of Chicago shows that throughout the Paleozoic, and probably for some time before then, Gondwana existed, consisting of South America, Africa, and Australia. It was surrounded by an ever-shifting array of landmasses, some large—about the size of North America, China, or Siberia—and others small, though numerous and destined eventually to become parts of Europe and Asia.

Paleoclimates and ancient oceans

For Gondwana itself, geologic data suffice to recreate in broad outline form not only its coasts, plains, and mountain ranges but also its deserts, polar ice caps, and tropical jungles. These reflect broad climatic zones, which were themselves controlled mainly by insolation, seasonality, and the rotating of the Earth. Large landmasses may sometimes have modified climate to produce monsoons. Given the positions and sizes of continents, paleoclimatic zones may be inferred and then checked against climatic conditions deduced from the fossil record of sedimentary rocks. The results show a satisfactory correspondence between expectations and record, much better than the one obtained by plotting paleoclimatic data on the present-day configuration of continents. They leave little doubt that it is better to accept that the continents have drifted about than to assume that they have remained stationary.

The supercontinent Pangaea was completely surrounded by a world ocean extending from pole to pole and spanning 80 percent of the circumference of the Earth at the equator. The equatorial current system, driven by the trade winds, resided in warm latitudes much longer than it does today, and its waters were therefore warmer. The gyres that occupy most of the Southern and Northern hemispheres were also warmer, and consequently the temperature gradient from the equator to the poles was much reduced in comparison to the present.

Early in the Mesozoic, the equatorial current became circumglobal when the Tethys seaway split Gondwana from its northern counterpart, Laurasia. The equatorial surface waters, now able to circumnavigate the world one or more times, became even warmer; but whether the return flows continued to keep the high latitudes warm or diminished in strength and therefore somewhat increased the latitudinal temperature gradients is not certain. Nevertheless, in the middle and later Mesozoic the Arctic and Antarctic surface water temperatures were at or above 10° C (50° F), and the polar regions were warm enough to support forests. As the dispersal of continents following the breakup of Pangaea continued, however, the surface circulation of the oceans became much more complex. During the middle Cenozoic, the northward drift of Australia and South America created a new circumglobal seaway around Antarctica that remained centred on the South Pole. A vigorous circum-Antarctic current developed, isolating the southern continent from the warmer waters to the north. At the same time, the equatorial current system became blocked, first in the Indo-Pacific region, next in the Middle East and eastern Mediterranean, then at Gibraltar, and finally, about 5 million years ago, by the emergence of the Isthmus of Panama. As a result, the equatorial waters were heated less and the mid-latitude ocean gyres were not as effective in keeping the high latitudes warm. Because of this, an ice cap began to form on Antarctica some 20 million years ago and grew to roughly its present size about 5 million years later. This ice cap cooled the waters of the adjacent ocean to such a low temperature that the waters sank and initiated the north-directed abyssal flow that marks the present deep circulation. The Quaternary Ice Age arrived in full when the first ice caps appeared in the Northern Hemisphere about 2 million years ago.

There is no certainty that the changing configuration of continents and oceans can be held solely responsible for the onset of the Quaternary Ice Age, even if such factors as the drift of continents across the latitudes (with the associated changes in vegetation) and reflectivity for solar heat are included. There can be little doubt, however, that it was a major contributing factor and that recognition of its role has profoundly altered all concepts of paleoclimatology.

Plate tectonics and mountain building

Subduction and continental collision raise mountain ranges. Consequently, the implications of plate tectonics for the processes of mountain building have attracted much attention. One of the earliest to apply the new theory was the Cambridge geologist John Dewey, who analyzed the Appalachian and Alpine orogenies. Many other researchers have subsequently undertaken similar work in the Mediterranean system and the American Cordilleran ranges, as well as in the Appalachians.

The collisions that gradually joined North America to Pangaea during the Paleozoic exemplify the role of plate convergence in mountain building. Some 500 million years ago, a subduction zone existed along what is now the eastern seaboard of North America. During the course of the next 67 million years or so—the Ordovician period—this region underwent a major phase of mountain building. When it was over, the direction of sediment transport had reversed, indicating that a large source area had appeared to the east of the subduction zone. The thick sequence of deposits from this source in eastern North America bears witness to its size, which is incommensurable with the island arcs that normally accompany a subduction zone. Subduction, sedimentation, and volcanism then continued until a small continent comprising what is today northwestern Europe collided with the northeastern tip of North America. This collision raised a major chain of folded mountains, the remains of which are now found from New England through eastern Canada and Scotland into Norway. This Caledonian range shed sediments to both sides, forming the Catskill delta in eastern North America and the Old Red Sandstone in northwestern Europe. Subsequently, the southern edge of North America collided with South America, and a little later Africa arrived, producing the southern Appalachians. Miscellaneous fragments, today part of central Europe and the western Mediterranean, filled in the gap and completed this portion of Pangaea.

This account fits the plate model fairly well except for two things: (1) the occurrence of paroxysms of mountain building when no continental collision can be assumed and (2) the need for sizable landmasses, presumably small continents, floating offshore to provide a necessary sediment source. Both problems appear quite often wherever plate movements are used to explain the details of the formation of major mountain ranges. Yet, the model itself is simple. It allows for three possible cases: ocean–ocean collisions, ocean–continent collisions, and continent–continent collisions. In the first two situations, oceanic crust is subducted either under other oceanic crust or under a continent—a steady process presumably accompanied by an equally steady scraping off and deforming of sediments and basaltic crust at the plate edge and by volcanism farther away. The New Hebrides trench and island arcs in the southwestern Pacific exemplify an ocean–ocean collision, while the Andes mountain ranges of South America represent a long-lasting collision of oceanic crust and continental crust at the leading edges of plates.

Even in these cases the process appears to be more complicated than the theory suggests, demanding modifications that, though eminently plausible and often proven, are ad hoc and thus diminish its value as a unifying principle. The South American Andes, for example, have had an episodic history not readily matched to the evidently uneventful spreading of the seafloor in the adjacent Pacific. The volcanism of the segments of the Cascade Range in Oregon and Washington in the northwestern United States is only a few million years old in its present form, whereas subduction along the Pacific coast dates back much farther than that. All along the west coast of North America, mountain building has occurred in distinct major phases, ultimately caused in all probability by changes in the rate and direction of plate movements that complicate the basically simple subduction process.

One of the complicating factors is much in evidence in the coast ranges that extend from California to southern Alaska. Large sections of these mountains fit poorly with the geology of the surrounding terrain and are thought to have originated somewhere in the ancient Pacific. Perhaps they once were oceanic plateaus, island arcs, or other thick pieces of oceanic crust that would not subduct but instead became lodged against the continental edge, and so disturbed the processes of subduction and mountain building. Some of these “exotic terranes” have been shown by their paleomagnetic properties to have come a long way from their points of origin, perhaps as far as the Southern Hemisphere.

The Indus-Himalayan-Tibetan region in northern India and China represents a continent–continent collision that began 40 million years ago when all the oceanic crust between northward-drifting India and south-central Asia had been consumed and the continent-bearing edges of the two plates met. The suture of this collision, the site of the former subduction zone, lies inconspicuously in the high plateau region north of the Himalayas. This lofty mountain range itself formed long after the collision as India—its momentum not fully spent even today—drove another 2,000 kilometres into the underbelly of Asia. As a result, Asian lithosphere was displaced laterally along enormous transcurrent faults, which are the sites of the numerous devastating earthquakes in Iran, Afghanistan, and western China. The Himalayas themselves are the product of the northward, low-angle underthrusting of the upper slice of the Indian-Australian Plate under the Eurasian Plate, making it clear that a continent–continent collision involves a good deal more than the crumpling of opposing front edges.

Plate tectonics and life

Inevitably the continuous rearrangement over time of the size and shape of ocean basins and continents, followed by changes in ocean circulation and climate, have had a major impact on the development of life on Earth. Active interest in these aspects of the Earth science revolution has lagged behind that in other areas, even though as early as 1970 the American geologists James W. Valentine and Eldridge M. Moores attempted to show that the diversity of life increased as continents fragmented and dispersed and diminished when they were joined together.

Subsequently, however, the study of plate activity as a force in the evolution of life has leaped forward. A few simple examples of such research must suffice to illustrate the impact of the plate theory. Toward the end of the Paleozoic, during the Permian period (about 286 to 245 million years ago), there was a drastic drop in the variety of animal forms inhabiting the shallow seas around Pangaea. Well over half of the total number of known families became extinct. This drop can be attributed in large part simply to the decrease in biogeographic variety that marks a world consisting of a single continent rather than one comprising many widely dispersed landmasses. Other factors, such as a sharp decrease in the area of shallow-water habitats or a change in ocean fertility due to upwelling, have also been invoked. Moreover, the extinction had a complex history. High latitudes were affected first as a result of the waning of the Permian ice age when the South Pole slipped beyond the southern edge of Pangaea. The equatorial and subtropical zones appear to have been affected somewhat later by a global cooling. On the other hand, the extinctions were not felt so strongly on the continent itself. Instead, the vast semiarid and arid lands that emerged on so large a continent, the shortening of its moist coasts, and the many mountain ranges remaining from the collisions that led to the formation of the supercontinent provided strong incentives for evolutionary adaption to dry or high-altitude environments.

The impact of plate movements and interactions on life is perhaps most clearly demonstrated by what happens when continents diverge or collide. When the Atlantic Ocean began to open during the middle Mesozoic, the similarity between the faunas of opposite shores gradually decreased in almost linear fashion—the greater the distance, the smaller the number of families in common. The difference increased more rapidly in the South Atlantic than in the North Atlantic, where a land connection between Europe and North America persisted until well after the middle Cenozoic. The inverse, the effect of a collision between two hitherto separate landmasses, is illustrated by the consequences of the Pliocene emergence of the Isthmus of Panama. In South America a highly specialized fauna had evolved, rich in marsupials but with few predators. After the emergence of the isthmus had made it possible for land animals to cross, numerous herbivores migrated from north to south. They adapted well to the new environment and were more successful than the local fauna in competing for food. The invasion of highly adaptable carnivores from the north contributed to the extinction of no fewer than four orders of South American land mammals. Only a few species, notably the armadillo and the opossum, managed to migrate in the opposite direction. Ironically, many of the invading northerners, such as the llama and tapir, subsequently became extinct in their country of origin and found their last refuge in the south.

Early plate activity

Whatever the forces may be that drive the plates, they consume energy. By far the largest part of this energy is derived from the decay of radioactive isotopes within the Earth, and the energy flow has therefore declined through the 4.5 billion years of the Earth's history—rapidly at first and then at a slowly diminishing though not negligible rate. Accordingly, it is quite likely that the behaviour of the lithospheric plates on the early, more energetic Earth was different from what it is today, and what prevails at present will certainly differ from what will prevail in the future. A thickening lithosphere, a decreasing heat flow, a temperature gradient that decreases with depth within the Earth, and enlarging—though perhaps less vigorous—convection cells in the mantle have all been postulated as unidirectional changes that affected the behaviour of the lithosphere. It is possible, for example, that the initial plates were too small, too hot, and hence too light to be subducted. In this case, the first subduction would mark the coming of age of classical plate tectonics, and, indeed, clear evidence is lacking for subduction until rather late in the Precambrian.

The evidence that bears on the existence, nature, and movements of possible lithospheric plates during the first several billion years of Earth history is very limited. The continental nuclei of the early and middle Precambrian seem to have been small and might be regarded as small plates on a more vigorously convecting mantle, though admittedly other explanations are equally possible. These nuclei are thought to have been embedded in strongly deformed complexes of sediments and basic igneous rocks, the greenstone belts reminiscent of the sutures that mark the closure of ancient oceans. In most cases, however, paleomagnetic data do not leave room for the existence of sizable oceanic areas between such nuclei. Investigators are thus forced to contemplate the possibility that, in the Precambrian, intensive deformation took place within plates, perhaps commensurate with the postulated thinner lithosphere and higher flow of heat toward the surface. On the other hand, current knowledge of this long but obscure portion of Earth history is so deficient that some geologists have emphatically denied that there might have been a remote past to which classical plate tectonics could not be applied.

Dissenting opinions and unanswered questions

The dissenters

Scientific revolutions as far-reaching in their consequences as the plate tectonics revolution cannot be expected to be easily, or ever completely, accepted. Nevertheless, once the theory had fully emerged, acceptance was quick and widespread and by the late 1960s its influence in the West was pervasive. Such was, however, not the case in the Soviet Union, a country located largely in the continental interior far from present-day plate boundaries. Soviet scientists viewed as central to any issue of global tectonics the vertical movements of continental interiors, phenomena not satisfactorily dealt with by the plate tectonics theory. A leading spokesman for the Soviet position, the academician Vladimir Vladimirovich Belousov, strongly defended a model of the Earth that postulated stationary continents affected almost exclusively by vertical motions. The model, however, only vaguely defined the forces supposedly responsible for the motions. In recent years, a younger generation of Soviet geologists has very gradually come to regard plate tectonics as an attractive theory and a viable alternative to the concepts of Belousov and his followers.

Opposition to plate tectonics was by no means limited to the Soviet Union. Critics were heard—in fact, some quite loudly—elsewhere as well. Sir Harold Jeffreys continued his lifelong rejection of continental drift on grounds that his estimates of the properties of the mantle indicated the impossibility of plate movements. He did not, in general, deign to take note of the mounting geophysical and geologic arguments that were in favour of a mobile outer shell of the Earth.

Others proffered different explanations of the accumulating evidence, most of which were rather farfetched, as, for instance, the suggestion that new crust was formed at trenches and destroyed on mid-ocean ridges. Still others, notably the American geologists A.A. Meyerhoff and Howard A. Meyerhoff, attempted to assemble data that contradicted the theory and thereby show that the supporting evidence was wrong, insufficient, or simply misconstrued. Demonstrating a remarkable command of often quite obscure literature, they issued a series of negative commentaries in the early 1970s; but they failed to convince the majority of their colleagues, partly because they did not offer alternative explanations for the evidence.

The only serious alternative had been proposed in 1958 by the Australian geologist S. Warren Carey in the form of a new version of an old idea—namely, the expanding Earth model. Carey accepted the existence and early Mesozoic breakup of Pangaea and the subsequent dispersal of its fragments and formation of new ocean basins, but he attributed it all to the expansion of the Earth, the planet presumably having had a much smaller diameter in the late Paleozoic. In his view, the continents represented the pre-expansion crust, and the enlarged surface was to be entirely accommodated within the oceans. This model accounted for a spreading ocean floor and for the young age of the oceanic crust; however, it failed to deal adequately with the evidence for subduction and compression. Carey's model also did not explain why the process should not have started until some four billion years after the Earth was formed, and it lacked a reasonable mechanism for so large an expansion. Finally, it disregarded the evidence for continental drift before the existence of Pangaea.

Unanswered questions

As the philosopher Thomas J. Kuhn has pointed out, science does not always advance in the gradual and stately fashion commonly attributed to it. Major breakthroughs often come from a leap forward that is at least in part intuitive and may fly in the face of conventional wisdom and widely accepted evidence while strict requirements for verification and proof are temporarily relaxed. Revolutions thus often become widely accepted before the verdict from rigorous analysis of evidence is completely in. Such was certainly the case with the geologic revolution, which also confirms Kuhn's view that a new paradigm is unlikely to supersede an existing one until there is little choice but to acknowledge that the conventional theory has failed. Thus, while Wegener did not manage to persuade the world, the successor theory was readily embraced 40 years later, even though it remained open to much of the same criticism that had caused the downfall of continental drift. What is the state of the new paradigm? Is it likely to suffer sooner rather than later the same fate that inevitably awaits all scientific theories?

In 1974, almost alone among the doubters who tried to discredit the new theory with contrary evidence, the American geologist John C. Maxwell, in a closely reasoned paper, enumerated all the points on which he believed plate tectonics had failed to offer an explanation. Many of these points have since been resolved, but more than a few remain to suggest that the theory, though in essence valid, may be incomplete.

The greatest successes of plate tectonics have been achieved in the ocean basins where additional decades of effort have confirmed its postulates and enabled investigators to construct a credible history of past plate movements. Inevitably in less rigorous form, the reconstruction of early Mesozoic and Paleozoic continental configurations has provided a powerful tool with which to resolve many important questions. On the other hand, the new paradigm has proved less useful in deciphering mountain-building processes or in offering explanations for the complex history of sea-level fluctuations. The American geologist L.L. Sloss has devoted a great deal of effort to demonstrating that continents do indeed rise and fall in unison, but the possible mechanisms for such a process remain elusive.

Where plate boundaries adjoin continents, matters often become very complex and have demanded an ever denser thicket of ad hoc modifications and amendments to the theory and practice of plate tectonics in the form of microplates, obscure plate boundaries, and exotic terranes. A good example is the Mediterranean, where the collisions between Africa and a swarm of microcontinents have produced a tectonic nightmare that is far from resolved. More disturbingly, some of the present plate boundaries, especially in the eastern Mediterranean, appear to be so diffuse and so anomalous that they cannot be compared to the three types of plate boundaries of the basic theory.

There is further evidence, as held by the American geophysicist Thomas H. Jordan, that the base of the plates extends far deeper into the asthenosphere below the continents than below the oceans. How much of an impediment this might be for the free movement of plates and how it might affect their boundary interactions remain open questions. Others have postulated that the lower layer of the lithosphere peels off and sinks late in any collision sequence, producing high heat flow, volcanism, and an upper lithospheric zone vulnerable to contraction by thrusting.

It is understandable that any simple global tectonic model would work better in the oceans, which, being young, retain a record of only a brief and relatively uneventful history. On the continents, almost four billion years of growth and deformation, erosion, sedimentation, and igneous intrusion have produced a complex imprint that, with its intricate zones of varying strength, must directly affect the application of plate forces. Seismic reflection studies of the deep structure of the continents have demonstrated just how complex the events that form the continents and their margins may have been, and their findings sometimes are difficult to reconcile with the accretionary structures one would expect to see as a result of subduction and collision.

Notwithstanding these cautions and the continuing lack of an agreed-upon driving mechanism for the plates, one cannot help but conclude that the plate tectonics revolution has been fruitful and has immensely advanced scientific understanding of the Earth. Like all paradigms in science, it will most likely one day be replaced by a better one; yet there can be little doubt that, whatever the new theory may state, continental drift will be part of it.