Shocks and Rocks Seismology In

Shocks and Rocks

Seismologyin the Plate Tectonics Revolution

Thestory of earthquakesand the great earth science revolutionof the1960s

by Jack Oliver

Irving PorterChurch Professor Emeritus Departmentof GeologicalSciences and Institutefor the Studyof the Continents(INSTOC) CornellUniversity Ithaca, NY

AMERICAN GEOPHYSICAL UNION WASHINGTON, Historyof GeophysicsVolume 6 Publishedunder the aegis of the AGU History Committee

Library of CongressCataloging-in-Publication Data

Oliver,J. E. (JackErtle), 1923- Shocksand Rocks: seismology in the plate teetonics revolution: thestory of earthquakesand the great earth science revolution of the 1960s / Jack Oliver. p. cm.--(Historyof geophysics:v. 6) Includesbibliographical references (p.-) andindex. ISBN 0-87590-280-4 1. Platetectonics. 2. Seismology.I. Title. QE511.4.045 1996 551.1 '36'09-de20 96-2723 CIP

ISSN: 8755-1217 ISBN: 0-87590-280-4

Printed in the United States of America

AmericanGeophysical Union 2000 Florida Avenue, N.W. Washington,D.C. CONTENTS

Preface

Acknowledgements xiii

The Pre-1960s 1 Observingthe Shaking

Learningabout the Earth and about Earthquakes 13

3 Artificial Sources 23

The 1960s 4 ContinentalDrift and SeaFloor Spreading, The Forerunners of Plate Tectonics 31

TheOrigin and Early Days of theLamont Geological Observatoryand Its Programin EarthquakeSeismology 41

Scienceby HypothesisTesting 53

Scienceby Serendipity 61

Scienceby Synthesis 75

After the 1960s 9 Full-FledgedPlate Tectonics 91

10 Strategy,Philosophy, and Things Like That 105

11 Epilogue 125

Bibliography 127

Index 131

Appendix:Seismology and the New GlobalTectonics PREFACE

Duringthe decade of the1960s, science of thesolid earth underwent an astonishingand awesomeupheaval. In justa few years,geoscientists con- structeda newway of describingand understanding the dynamics of ever- changingearth, past and present, and so found a routeto explanationfor how most,if not all, of the greatfeatures of the earth'ssurface that have harboredand plaguedand enchantedhumans throughout their existence cameto be. Continents,ocean basins, mountain ranges, deep sea trenches, earthquakes,and volcanoes suddenly became explicable as consequences of earthmovements that, on a globalscale, have a remarkablysimple and readilyunderstandable pattern. The long-sought key to the ponderous and agonizinglyslow movements of earththat, over millennia, have deftly shapedour surroundings was found during that decade, or somost scien- tiststhink today, more than a quarterof a centurylater. For thosescientists studying the earth,and a few skepticsaside, the 1960swere a time of astonishmentand discovery,of delightand euphoria, ofpride, and of stimulation reinforced by successafter success. Word of the achievementsspread quickly throughoutscience and beyond. The upheavalwould be called "the plate tectonics revolution," "plate tectonics" beingthe term invented to describethe essence of thenew dynamics. It has beenclaimed by somethat the discoveryof platetectonics is the greatest achievementever of modernearth science, although those who hold brief in thisregard for oneor two otheroutstanding revelations such as recognitionof thegreat age of theearth or thepostulating of thetheory of evolution might challengethat statement. In anycase, the achievements of the 1960s were outstanding and their storyseems to merit telling, not once from the singular perspective of a par- ticularscience historian, but multiplyfrom the diverseperspectives of scientistsand otherswho witnessedand/or participatedin that revolution. Thisbook is onesuch attempt. It focuseson themany critical contributions of thefield of seismologyto thenew dynamicsduring the early stages of the revolutionand it is writtennot by an historianbut by a participant.As such,it shouldcomplement the now numerousother histories of the discoveryof platetectonics, most of whichreport, but none of whichis solely devoted to, the role of seismology. In an effortto keepthe text flowing smoothly, and to holdits lengthin check,I citeonly key referencesand discussnon-seismological topics in rathergeneralized fashion. For those who would like additional reading on thesubject as seen from a perspectivedifferent from mine or with emphasis differentfrom mine, or who would like a more completeset of ences, I include at the end of this section a list of some of the available literature.It is a measureof the breadthof interestin the subjectto note that the following categoriesare representedin the list of authors:science writer, historianof science,philosopher of science,sociologist of science, earth scientist,non-earth scientist, participant in the revolution,and non- participantin the revolution.Their views of the revolutionand their pur- posesare sometimescommon, sometimes diverse, but they all sharerecog- nition and appreciationof a major happeningin science.Some 25 years after the fact, it should be clear and obviousto the reader that the prime purposeof this book is not to add glory to the revolutionand its participants, but rather to describejust how a scienceworks when things are goingwell, in the hopeof acceleratingthe pacetoward greatdiscoveries of the future. As is the casein most, if not all, scientificrevolutions, or paradigm shifts to use modern terminology,the essenceof plate tectonicsand the characteristicsthat distinguishit from previoustheories on the nature of earthdynamics are remarkablysimple. In contrastto the staticearth models that dominatedearlier geologicthinking, plate tectonicscalls for horizontal movementsof rock massesat or near the earth's surfacethrough hugedistances (perhaps thousands or tensof thousandsof kilometers)over long intervalsof time (perhapstens or hundredsof millionsof years).The correspondingspeeds of a few cm/yr for a land masssuch as a continent, althoughextremely slow in normalday to day context,are sufficientlyhigh geologicallyspeaking to startlethose accustomed to thinking of rocksor land massesas immobileand fixed in globalperspective. The comingof plate tectonics,then, corresponded to a shiftin emphasis of geologicthought from what has been termed the "fixist" schoolin which rocksare thought to remainthroughout their existencenear the place where they were formed, to the "mobilist" schoolin which rocks are thoughtto movelaterally and geologicallyrapidly through significant frac- tionsof earth'scircumference. In this sense,plate tectonicsis like its earlier but neverwidely acceptedforerunner, the conceptof continentaldrift. Plate tectonicsis built upon,and incorporatessome of the basicelements of, continentaldrift, but differsfundamentally from early or orthodoxcontinental drift theoryin that the prime entitiesthat moveas units are not definedby theborders of the continentsbut arethe mostlylarger and usuallyless obvious segmentsof the earth'ssurface that have come to be known as the "plates." Takentogether, fewer than a dozenlarge plates form a mosaicthat cov- ersvirtually the entiresurface of the earth.Obvious earth featuressuch as continents,oceans, islands and mountainsare, in the very simplestper- spective,not the prime elementsin the dynamicmodel; it is the lessreadi- ly apparentplates that arethe basicunits, and it is their movementsthat can be describedin simple geometricalfashion. The lesserunits suffer more complex and intricate secondaryeffects that make the story of earth dynamicsmore comprehensive,more varied, more interesting,and often more confounding,but thoseeffects now fall into placeand can be under- stoodas consequencesof the simpleprimary motionsof the basicplates. The effectsof the ponderousmotions are awesome.Driven by heat escapingfrom the earth'sdeep interior,the elementsof the mosaic,i.e. the plates,are continuallyin motion and continuallymodifying their shapes throughaddition of new surfacematerial at somesegments of plateboundaries and destructionof old at others.Huge land massesriding atop the platesmay travel from onehemisphere to another,or from the tropicsto the arctic,and may be split apart, or sheared,or deformedand piled atop one anotherin collision,thereby providing a settingfor a variety of lessergeo- logicprocesses that build the featuresof rocks,terrains, and landscapesthat humansfind familiar and beautifuland essentialto supportof civilization. That so simple a basicmodel of earth was found to accountfor surroundings so complexand so perplexingas thosewe find in the real earth is a strikingexample of successin sciencethrough synthesis, and also of the beautyof explanationthrough simplicity that characterizesgreat advances in scienceand that merits attentionwell beyond the inner circlesof earth science. In fact,and as is the casein mostrevolutions in thought,the beautiful simplicityof the plate tectonicsconcept is a sourcenot only of pride and elationfor the earth sciencecommunity, but also of mild embarrassment and consequenthumility. Why, indeed, didn't we, and couldn't we, earth scientistsarrive at that simple solutionearlier? How had we misdirected and immobilizedour thinkingso that somethingthat in retrospectseems so obviousremained hidden for so long?What might we have doneto make this revolutionoccur at an earlier time and in lessabrupt and more pre- dictablefashion? Such questions suggest that someintrospection on just how the new order cameabout is appropriateand may well be instructive. Well, how, indeed,did we manageto achievethis new level of understandingof earth?Was such discovery the productof genius,a consequence of a flashof inspirationin an exceptionalmind secludedand isolatedfrom the mainstreamof geologicthought which, it would turn out, was in need of reorientation?Well, not exactly.The plate tectonicsrevolution had its shareof cleverand inspiredideas by innovativeand creativeindividuals of course, and I do not wish to detract from deserved credit to the scientists who had thoseideas. But somethingas grandiose and asfar-reaching as the plate tectonicsrevolution is rootedin backgroundsof knowledgeand pro- longedseries of major and minor eventsthat go well beyondthe brilliance of a few individuals and the isolated incidents that come to be known in history as "the discoveries".For such a broadly-basedand far-reaching advanceto be achieved,there first had to comea long, difficult and ago- nizing period of sweat and toil, of cleverstrategic and tacticalthinking, of accumulationof observation,and of failed attemptsat synthesisthat would set the stagefor the climacticand successfuldevelopments to follow.This book thereforeis not solelyabout the role of seismologyduring the plate tectonics revolution of the 1960s, for it also touches on and summarizes eventsbefore, and to a lesserdegree after, that decade. The writing of the book was stimulatedby a requestin 1993from the Committeeon History of the American GeophysicalUnion for a mono- graphon this subject.I presumethat I was chosenfor this taskat leastpartly becauseI am an author,together with BryanIsacks and Lynn Sykes,of a paperentitled "Seismology and the New GlobalTectonics" published in the AGU's Journalof GeophysicalResearch, 15 September1968. That paper,which I shall call the NGT paper,was timely,widely-read and hencevery well- known at that time. Perhapsmore than any other,it brought the seismological componentinto, and in the processmade fundamentalcontribu- tionsto, the nascentsubject of plate tectonics.It attempteda comprehensive review of all available evidencefrom the field of earthquakeseismology that bore on the then new and rapidly developingtopic of what is now known as "platetectonics," but that term was not in useat the time, and we referredto the topic as "the new globaltectonics." The NGT paper contributedsome important fundamentaland new ideas to what would becomethe plate tectonicsconcept, cited abundant supportingobservations, and communicatedthe informationreasonably well through figures and a text that minimized seismologicaljargon. The paper thus became influential, partly among those who were already caught up in the early phasesof the plate tectonicseuphoria from their studiesof geomagnetismor other phenomenaand who henceappreciated and welcomedconfirming and supplementaryevidence from seismology, and partly amongthose in otherspecialties of earthscience who were in the processof becomingacquainted with the conceptof plate tectonicsfor the first time. The NGT paper was describedby someas a milestone,bringing seis- mologyfirmly into the subjectof plate tectonics,and plate tectonicsfirmly into seismologyfor what, it seemedthen and still seemsat present,was for once and for all. Among other things, that popular paper left its three authorslabeled, perhaps over-generously, as authoritiesin the field. This monographis thereforepartly aboutthe eventsthat led up to and resulted in the NGT paper,including earlier publications on relatedsubjects by the three authors. The NGT paper in originalform is includedas an appendix.References to it in the text are in the form: (NGT, Fig. n). The history of seismologyand plate tectonics,however, is hardly the story of one particular paper, no matter how timely or well-received. Scienceis more than just a few isolatedevents that happen fortuitouslyat the right time and place.Other seismologicalpapers of various types and about the same or earlier vintage also contributedinformation that would be important to the development of the concept of plate tectonics. Furthermore, and of tremendousyet often overlooked importance,for many decadesprior to the 1960sseismologists engaged in activitiesthat provideda backgroundof informationon the earthand its earthquakesthat was critical to the developmentof models of the dynamic earth and its moving surficialplates. I hope this book is as much a tribute to them as it is to thoseof us who were fortunateenough to be in the right place at the right time when seismologyoffered the opportunityfor contributionsto the revolution.And, furthermore,in the period sincethe revelationof the '60s, seismologistshave continuedto contributeto the refinementof thoseearly modelsas they labor now in a seismologicalworld with plate tectonicsas the ruling paradigm. In this monograph,therefore, I try to sketchand tracecertain streams of seismologicalefforts, and other activitiesof relatednature, through an interval of historymuch longerthan that decadeof the '60s.Sometimes the storybegins well back into the 19th century,sometimes even earlier.Those early activities,it would turn out, led to a level of understandingof the earth'sinterior and its processesthat provided a fertile foundationfor the excitingnew ideas aboutthe dynamicsof the earth that arosein the 1960s. I also discuss the events of the '60s in some detail of course, and summarize and commentbriefly upon post1960s research on this subject.With the advantageand clairvoyanceof the historicalperspective, I draw attention not only to the triumphs of seismologybut also to oversightsand missedopportunities along the way. My purposein doing so is not to den- igrate in any way the effortsof seismologistsof the past,for I admire and respectall of them.Rather I citewhat now seemobvious fumblings or oversightsof the past in order to stimulaterising young scientistsof today to think and reasonin ways that will help them avoid suchmissteps in the future.Any history,it seems,should have enlightenment,encouragement, and assistancefor succeedinggenerations as a prime goal, and this book tries to do its share on this score. As the readerwill quicklyrecognize, this book is basedon my personal perspectiveof the eventsand activitiesof the time. It is not a scholarly tome basedon endlesslibrary researchand including innumerablecita- tions.It has factual material,opinion and selectedanecdotes. In placesit seemsautobiographical. A historywritten by, and basedon the personal perspectiveof, a participantor observerwill, of course,nearly always result in a subjectiveand somewhatbiased view of the historicalevents. At least, however,in this type of historyit is obviousto the readerthat suchbias is likely. The other extremeof perspective,that of the historianwho comes along later to produce a scholarly,carefully-researched, and annotated record of what happened may seem more objective at first glance. However,my experiencewith suchhistories when they involve eventsthat I know from personalassociation suggests that they alwaysend up with a view of what happenedthat is at leastas distortedas the variousviews of the participants.Try as we might, in other words, and regardlessof our backgrounds,we never managethe perfectand completehistory of anything. For whatever it is worth then, what follows is my currentview, based on memory,long past and recentdiscussion with many colleagues,and someliterature search, on what happenedas seismologyjoined with other earthscience disciplines to producethe conceptof platetectonics, a concept that somehave describedas the greatestadvance ever in human understandingof the earth. For thosewho experiencedthe revolutionand who may feel that I have not alwaysreported things as they rememberthem, I can only respondthat (a) I have tried to give as fair and accuratea depic- tion as I can, subjectto limitationsof spaceand memory,and (b) I encour- agethem to make differentviews known as appropriate. My purposein writing this book is, of course,to tell the storyof seismology'srole in the comingof platetectonics, at leastas I sawit, for all who careto read it. I hopethat the audiencewill includenot only seismologists and scientistsin relateddisciplines, but alsoreaders, scientist or non-scientist, with a generalinterest in how one branchof scienceoperated to produce one of those anomalousperiods when the ups overwhelmedthe downsfor a time and thingswere goingvery well indeed.To facilitateread- ing by those with limited familiarity with seismology,I have tried to includeenough introductory material to make the book readableby non- specialists. The organizationis straightforwardand largely,though not strictly, chronological.The earlypages are aboutthe pre-1960swhen techniquesfor quantitativeobservation of earth's vibrationswere being invented and developed,and when stylesfor handlingand analyzingthe observations waysthat would turn out to be importantduring the discoveryof plate tec- tonicswere being formulatedand established.The relevantresults of these pre-1960 studies of earthquakesare summarized. Also included is a descriptionof the evolution,and the results,of seismicstudies using artificial sourcesthat range in size from the pops of compressedair bubblesto the wallopsof hugenuclear explosions. Such studies contributed important informationon the earth that bore,directly or indirectlyalthough perhaps not so charismaticallyas earthquakestudies, on the developmentof plate tectonics. For the criticaland excitingyears of the 1960sI focuson the earthquake studiesthat played a strongrole in the plate tectonicsstory but alsotry to draw attentionand give credit to lessobvious and lesserknown seismo- logicalresults of other types.For the post-1960speriod, I discusssome of the seismologicalstudies that followed immediatelyafter the formulation of the conceptof plate tectonicsbut make only a cursoryeffort to summarize currentactivity in earthquakeseismology. This is a historybook, not a review of a branch of current science. Near the end, however,and with the dreamscharacteristic of an aging scientist,I could not resistthe temptationto passon a few philosophical remarksthat might inspiresome hard-driving young scientistsomewhere to biggerand better discovery.I point out a few areaswhich, in my opinion, hold exceptionalpotential for seismologyof the future, and some strategiesand tacticsfor realizingthat potential. As the book closes,I hopeit will be no lessthan blatantlyobvious that I am optimisticabout the future of earth science,and that I encouragethose who would study the earth to push aheadat full steamand with the conviction that if they do so they can readily eclipsethe accomplishmentsof earlier eras,even of the very specialand excitingera of discoveryof plate tectonicsdescribed here, and now somequarter of a centurypast.

Eachlimerick's a bit of tomfoolery And hardlya triumphscholaroolery But limerickspulled from theair And appliedhere and there Are a kindof syntacticaljewelry. SOME HISTORIES OF THE PLATE TECTONICS REVOLUTION

Allegre, C., 1988, The Behaviorof the Earth,Cambridge, Massachusetts: Harvard University Press. Frankel,H., 1980a,"Hess's Development of His SeafloorSpreading Hypothesis," in T. Nickles(ed.), ScientificDiscovery: Case Studies, Boston: D. Reidel,pp. 345-366. Glen, W., 1975, ContinentalDrift and Plate Tectonics,Columbus, Ohio: CharlesE. Merrill. Glen,W., 1982,The Road to Jaramillo, Stanford: Stanford University Press. Hallam, A., 1973,A Revolutionin the Earth Sciences,Oxford, England:Clarendon Press. Heirtzler,J., 1968,"Sea-Floor Spreading," Scientific American, 219, pp. 60-70. LeGrand, H., 1986a, "Specialties,Problems, and Localism:The Receptionof ContinentalDrift in Australia:1920-1940," Earth Science History, 5, pp. 84-95. Marvin, U., 1973, ContinentalDrift: The Evolutionof a Concept,Washington, D.C.: Smithsonian Institution Press. Menard, H., 1986, The Oceanof Truth,Princeton, New Jersey:Princeton University Press. Stewart,J., 1990,Drifting Continents and Colliding Paradigms, Indianapolis, Indiana, Indiana University Press. Sullivan, W., 1974, Continents in Motion, New York: McrGaw-Hill. Takeuchi,H., S. Uyeda, and H. Kanamori, 1967, Debateabout the Earth, San Francisco:Freeman, Cooper, and Co. Tarling,D., and M. Tarling, 1971,Continental Drift: A Studyof the Earth'sMoving Surface,London: G. Bell and Sons. Uyeda, S., 1978,The New Viewof theEarth, San Francisco: W. H. Freeman. ACKNOWLEDGMENTS

As noted in the Preface,this book was written at the request of the Committee on History of the American Geophysical Union. Orson Andersonwas the Committee'schairman at the time and BenjaminHowell Jr. the direct contact. BenjaminHowell Jr.,Charles Drake, BryanIsacks, Lynn Sykes,Muawia Barazangi,and GeorgeHade read an early draft and made helpful com- mentsand suggestions,most of which resultedin revisions.However, I, not they,bear full responsibilityfor the final content.Cornell University provided logisticalsupport during the writing; Sue Petersontyped the manu- scriptand innumerablerevisions. Writing a book on the historyof a facetof science,perhaps more sothan normal day to day activity in science,inevitably leaves one awed by real- ization of the scaleof the collectiveefforts of large numbersof thosescientists who contributedalmost endlesslyand tirelesslyand often almost anonymouslyto progressin that branchof science.No historianor history could possiblydo full-justiceto thosecontributions. I can only acknowl- edgein a generalway the collectiveefforts and effectsof suchscientists and regretthat they cannot all be cited individually. Various institutions, in addition to Columbia and others mentioned in the text, supportedthe effortsof scientistsdescribed here as did various governmentand private funding agencies.Particularly prominent were National ScienceFoundation, Air ForceCambridge Research Laboratories, Air Force Office of Scientific Research, Office of Naval Research, and Advanced ResearchProjects Agency, all, except for NSF, parts of the Departmentof Defensethat supportedbasic research in earth science. Finally, during that period of the 1960's that brought the scientific excitement,upheaval and euphoriaon which this book is based,I met and married my wife, Gay,and sheproduced and caredfor our two daughters, Nelly and Amy. In appreciationand gratitude,I dedicatethis book to her and to them. 1

Observingthe Shaking

A magnitudeeight every time MakesP, S, andRayleigh sublime Then those undulations Formearth's free oscillations Fora grandseismological chime.

ll modernscience isbasically empirical, inessence merely the organizationand comprehension of reliable observations. No matter how sophisticatedthe theory, how intricate the reasoning,how ingeniousthe schemesfor organization,or how contortedthe philosophical analysisof the scientificprocess, the ultimate test of scienceis in the observations.Any branch of science,seismology of courseincluded, is thereforerooted in, controlledby, and oftendriven in new directionsby, its observations. What can be, and is, observed and then communicated appropriatelyultimately determinesthe courseof science.To understand the historyand the evolutionof seismologythen, we must have a general understandingof the developmentand growth of its observationalcapaci- ty, and of the observationsthemselves. Unlike somenatural phenomena,such as cosmicrays or radioactivity, which requirethe useof sophisticatedinstrumentation before the phenomenon can even be recognized,earthquakes have probably never been unknown, ignored or overlooked,at least sincehumans began living in active seismicareas. The largestearthquakes are readily felt by all those near the center of the disturbance,and accountsof earthquake-induced destructionand casualtiesand fear appearearly in recordedhistory. Earth scientists,ancient and recent, have sought to relate non-instrumentally determined"macroscopic" evidence, such as shaking,faulting, elevation changeand geographicallocation, to otherkinds of geologicalobservations in an effortto understandthe earthquakephenomena. The early Greek philosophersnoted and commentedupon earthquakes;so did the Chineseand Arabs;so have Western scholars. As just 2 Observingthe Shaking example,Darwin, during his travelsaboard the Beagle,studied shaking, shorelinechanges, and volcanic eruptions correlated with the greatChilean earthquakeof 1835,and, somewhat incredibly, at that earlydate proposed a mechanismto explainAndean earthquakes that bearssome similarity to, althoughit falls short of, modern ideas about plate tectonics.Neither Darwin'sperceptive thoughts nor thoseof othersthat were alsobased on macroseismiceffects played a role of greatsignificance in the development of plate tectonicsduring the 1960s,however. Seismology's contributions were to arise insteadnot so much from the macroscopicdata as from the microscopicevidence that would comefrom the operationof globalor regionalnetworks of sensitiveseismographs capable of detectingextreme- ly minutemotions of the earth'ssurface at largedistances from the center of the disturbances.No human, unaidedby instruments,can detectsuch motions. The developmentof globalnetworks and the sensitiveindividual seis- mographsthat are the components of thosenetworks is an inspiringtale of dedicatedindividuals, clever and imaginativeideas, perseveranceand hard work, and technologicalevolution both within and outsidethe discipline of seismology(See Howell, 1990for a comprehensivereview). What follows is a brief sketchof the principlesand certain of the problems involved,particularly as they relateto the plate tectonicsstory. The goal of recording,or measuring,or just detectingthrough some device,the groundmotion associated with an earthquakehas been around for manycenturies. Some early attempts to build seismographswere based on suchphenomena as steelballs falling from the mouthsof sculptured dragons(!), rippleson a pool of mercurydisturbed by earthmotion, and fallen rectangularor cylindricalblocks once stood on end. Suchattempts seemprimitive or incongruouswhen contrastedwith the bestof modern instruments.Nevertheless they emphasizean importantand reasonably obviouspoint, namely the influenceof the stateof contemporarytechnolo- gy on the activityin a particularbranch of science.As we shallsee, and as in most sciences,the developmentof seismologicalinstrumentation follows,with onlya modestlag in time,development of new technologiesthat canbe incorporatedinto seismologicalinstrumentation. For the visionary, that dependencyraises the issueof what modernand near-futuretechnol- ogyholds in storefor near-futureseismology; certainly the electronicsrevolution is in the processof making currentinstrumentation seem as cum- bersomeand antiquatedas some of theearly seismographs we areabout to discuss.For the historian,it raisesthe questionof just when seismology was capableof producingthe observationsthat would enableit to con- tributeimportantly to the developmentof plate Observingthe Shaking 3

The earliest crude seismic instruments, such as those mentioned above, were designedto detectthat an earthquakehad happenedand perhapspro- vide a crudemeasure of the intensityand directionof the shaking.That was not enough.Soon seismologists recognized that greatvalue lay in recording in somedetail the complexmotion of the earthas it changedwith time. Thus beginningin the late 19th centurythe goal was to build an instrumentthat couldobtain a recordof the groundmotion at onepoint somedistance from the centerof an earthquakeas a functionof time. Once that goal was to a degree attained, instrumentscapable of doing so became common and widely distributedat locationsaround the earth,and seismologybecame the prime meansfor learningabout the earth'sdeep interior. At first it seemsthat detectingand recordingthe groundmotion of an earthquakethat shakesdown buildings and causesnausea in nearby observersshould be a simplematter, and indeed that particular task is not difficult. But earthquakesof interest to scientistsvary greatly in size, througha rangeof at leastten ordersof magnitude,and the amplitudesof their wavesalso vary with distance.Detecting an earthquakewhose source is beneathone's feet is far differentfrom detectingthat sameearthquake at a point halfway around the world. Thus the amplitudesof seismictran- sientsthat a seismologistmight wish to detect,record and studymay range from many feet to a fractionof the minute distancebetween two adjacent atomsin a solid!And the frequenciesof interestvary from as high as 100 cyclesper secondor so to as low as a singlecycle of oscillationin nearly an hour! The challengeof developinginstruments to monitorthis wide range of seismicmotions is thusa substantialone that, eventoday after more than a centuryof effort, continuesto evolve. When a seismicwave passesa particular location,everything in the vicinity- buildings,rocks, seismologists and instrumentsm movesmore or lessin unison.Except in the caseof "shaky" structuresor catastrophic failures, there is little relative motion to be measured.Detecting earthquake-generatedground motion would be easierif one had a fixed platform whose positionwas unaffectedby the earthquake.Then the seismic motionof a movingpoint on the groundcould be comparedto a fixed point on the platform. But there is, of course,no such fixed platform. In this respect,the seismologistexperiences a dilemma somewhatlike that of Archimedeswho claimedthat given a lever long enoughand a place to stand, he could move the earth. Seismologistsresolve the matterprincipally through the useof an inertial massthat is not firmly fastenedbut, rather,loosely coupled to the earth. To illustratehow thistechnique works in principle,consider the simplecase of a masssuspended from the lower end of a coiled springwhose 4 Observingthe Shaking end is fixed to the earth. If constrainedto move freely only in the vertical direction, the mass will have a simple natural period of oscillation. Furthermoreits oscillationscan be dampedso that it doesnot vibratewild- ly. Now if, as a consequenceof an earthquake(or any other sourceof groundmotion), the earthat the point of attachmentof the springoscillates with a period lessthan that of the free mass-springsystem, the masswill tend to remain fixed in space.With the earth moving and the massmore nearly stationary,one need only measureand recordthe distancebetween the two in order to obtaina registrationof earth motion,in shorta seismo- gram. In fact, even for periodsof oscillationmuch longerthan that of the mass-springsystem, a relative motion correspondingnot to displacement but to accelerationof the earth canbe measured.This principleis the basis of nearly all seismographsand they are termed inertial seismographs. Another type of seismographbased on measurementof the distance betweentwo well-separatedpoints in the earth is called a strain seismo- graphbut suchdevices are far lesscommon. The simpleinertial seismographdescribed above detects only the vertical componentof ground motion. But to specifyground motion completely,two otherorthogonal components must be recordedsimultaneous- ly. In order to measurea horizontal component,a modified sensoris required.Such a sensorcan be readily constructedby suspendingthe inertial masson what amountsto a sortof swinginggate, a boomsupported by hingesrotating about a near-verticalaxis, rather than a spring.The free periodof sucha sensor,or horizontalpendulum, can be adjustedsimply by tilting slightlythat axisof rotationfor the hinges.The free period of a vertical sensorcan be adjustedby varyingthe springconstant or the geometry of the suspension.In practice,it is not overlydifficult to build horizontal and verticallyoperating sensors with free periodsas great as 15 to 30 seconds. Such sensorscan detectmuch, though not all, of the spectrumof groundmotion generated by a large earthquake. Once the sensors are in hand, the next task is to convert the minute relative motion between earth and mass into a form that allows it to be recorded as a functionof time. In the early daysof seismologicalinstrumentation, the late nineteenthand early twentiethcentury, the relativemotion of the inertialmass was commonlyamplified by a systemof mechanicallevers. In systemswith considerableamplification, significant friction had to be overcome,so sometimeshuge masses,some as large as a small room, were employed.The massdrove a leversystem that culminatedin a needlethat scratcheda track on smokedpaper coveringa continuouslyrotating and translatingdrum. Earth motionsappeared as deflectionsof what Observingthe Shaking 5 otherwisehave been a straightline on a sheetof paperthat, onceremoved from the drum, was flat and rectilinear,easy to store,and easy to read. Dippingthe smokedpaper in shellacpreserved the recordindefinitely. In a later advance,in order to cut down frictionor improve gain, small mirrorswere attachedto the mass,or somewherein the mechanicalsystem, so that an opticallever couldsupplement or replacethe mechanicalone. A spot of light then wrote a photographicrecord similar in format to the smokedpaper record described above. Still later,sensitive electromagnetic methodsfor detectingrelative motion between the inertialmass and earth were devised.Motion betweena coil mounted on the massand a magnet on the earthgenerated current in a galvanometerconnected to the coil,and so moveda light spotthat couldbe recordedphotographically. This tech- niquehad many advantages and so became more or lessstandard for a con- siderableinterval of time.Almost all of the data on earthquakesthat helped to producethe platetectonics revolution were obtainedfrom seismographs usingthis technique, or a slightvariation of it basedon variablereluctance. Almostall modernseismographs continue to rely on the inertialdetec- tor. The electronicage, however, has broughtamplifiers, filters and other devices,and particularlydigital recordingtechniques that are clearlythe wave of the presentand future. Digital methodswere beginningto take root in seismologyduring the 1960sbecause of the easeand versatilityof digitaldata processing and becauseof the enhanceddynamic range of the recordingsystem, but they were not an importantfactor in seismology's contributionto the plate tectonicsrevolution during that decade. To emphasizethe point, referredto earlier,that seismologicalinstru- mentationfollows technologicaldevelopment in general,one might note herethe parallelsbetween seismograph development and the evolutionof the phonograph.Early phonographs, of course,were also largely mechanical,transmitting irregularities of a groovescratched in wax andthen sensed by a needlethrough a mechanicallever systemto a mechanicalspeaker. Later, mechanical to electrical transducers were added as were electronic amplifiersand electromagneticspeakers and suchsystems prevailed for a time. During the 1960s,seismologists celebrating the comingof plate tectonicsat a party listenedto musicfrom a deviceanalogous to the seismographsof the day.Now phonographs,like seismographs,are dominated by digital techniquesand digital recordingsystems. The precedingparagraphs describe briefly and in very generalized fashionthe evolutionof the seismograph.No suchevolution ever actually occurssimply or unidirectionallyin practiceof course,at leastin its earlier stages,because the humansinvolved have differentperspectives, different goals,different backgrounds and differentideas on how to proceed. 6 Observingthe Shaking

, unify such an effort requiresa commonpurpose or a central funding source.Early seismologistsdiffered somewhat on the formerand certainly did not have the latter,for the field of seismologyas a whole had very limited funding during the early 20th century and before.The sparsefunds that were availablecame from diversesources, perhaps the organization that employedthe scientist,perhaps the scientisthimself or, rarely, an organization concernedwith the local or regionalearthquake hazard. Thus, in thoseearly days,each seismologist operated more or lessindependently and with strongfinancial constraint, and the diversityand decidedlack of standardizationof instrumentsthat the scienceproduced showed it. For example,during roughly the first half of the 20th century,some seismographstations measured only one componentof ground motion, perhapsthe vertical. Others measured only two orthogonalhorizontal components.If three componentsof groundmotion were measured,the frequencyresponse of the verticalrarely matched that of the horizontalsand usuallywas so differentas to inhibit inter-comparisonalmost entirely. Designersof sensitiveseismographs also had to contendwith noise.In the spectrumof seismicbackground noise in the earththere is a sharpand persistentpeak in the period rangeof aboutfour to nine secondsdue to microseismsgenerated primarily by gravity wavesof twice that periodon largebodies of water. The earliestarriving, compressionalseismic waves from a distantearthquake often have abundantenergy in periodsshorter than that of the noisepeak. Furthermore,the energyof later arrivingseis- micbody waves, and mostseismic surface waves, is predominantlyin peri- odsgreater than that of the peak.Instrument designers, therefore, tried to avoid the peak and built both "short-period"and "long-period"seismographs.Often, and unfortunately,the verticaland horizontalcomponents differed in frequencyresponse as a consequence. Other factors contributed to the diversity of seismographs. Seismologistsliving far from the earthquake-proneareas tried to build instrumentsof sufficientsensitivity to detectbackground noise in the earth and hence many teleseismicevents. Seismologists living in active areas built less sensitive instruments that would not be thrown off scale or made inoperativeby a nearby shock.Seismologists such as Milne, Wiechert, Galitzin,Benioff, Wood, Anderson, Press, and Ewing focuseda part of their effortson instrumentdesign and producedseismographs with specialchar- acteristicsthat becamewidely usedand that bear their names.As was true for many physicistsin the early days of that subject,a seismologistwas oftena machinist,engineer and technologicalinnovator as well as a scientist.Hugo Benioff,for example,a versatileand talentedearth scientist, was alsoa consultantto a maker of musicalinstruments, an occupation Observingthe Shaking 7 certain parallels with seismographicinstrumentation may be found by thosewho understandthe basicphysics of both kinds of devices. Adding to the difficultiesthat came from lack of standardizationof seismographswas limited capacityfor calibration.Most instrumentsof that era were not regularly calibrated,but should have been for the perfor- mances of some instruments were sensitive to external environmental effectsand seismographsof the era did not normally operatein controlled environmentsas they do today.The locationof many early instrumentswas oftenin the basementof a sciencebuilding of the scientist'suniversity, and hencein whateversurroundings the campusof that institutionmight be -- metropolitan, suburban or rural; coastal,island, or continental interior. Arrays of seismographswere sometimesespecially installed in activeseis- mic areasto measurelocal earthquakesbut usually at placesselected for convenienceof operationrather than favorableenvironment. One shortstory might illustratethe modusoperandi of early seismologists. From one scientific meeting I can recall a paper whose title announceda newly-developedseismograph with an abrupt right-angled bend in the boom that supportedthe mass.I attendedthe talk, eagerly expectingto hear of someunexpected quirk of instrumentaldynamics that producedan ingeniousnew way to observeseismic waves, only to learn that the bend was only thereso that the instrumentwould fit into the available,but alsovery confined,space open to useby that investigator.As most seismometersoccupy an area of lessthan 3 or 4 squarefeet, it was obvious that spacewas very tight indeedin the quartersof that sciencedepartment! Anotherfactor affecting the locationand operationof seismographstations was the need for accurate timing of the seismic recordings. Seismologistsquickly recognizedthat, in order to compareoptimally the seismic data from one station with that of another elsewhere in the world, accuratetiming, preferably to a small fraction of a secondwas needed. During the first half of the century at least, radio time signalswere not availablefor much of the world as they are now. Therefore,some seismographswere locatedat astronomicalobservatories where their pendulum clockscould be frequentlycalibrated against the stars.Other seismographs were locatedat remoteplaces that were the sitesof otherkinds of scientific observatories,or religious missions,or military outposts,or whatever. Sometimes recorded time at such sites was accurate, but often it was not, and timing dependedon a pendulum clock checkedonly sporadically againstfortuitous short wave radio signals.After WWII, radio stations capableof broadcastingstandard time signalsthroughout much of the world beganto operate,and seismologistsbenefited accordingly, but prior to that developmenttiming was sometimesso bad that earthquake 8 Observingthe Shaking tionsin certainparts of the earthmight be in errorby hundredsof kilometers,perhaps two orders of magnitudeworse than the standard of today.As we shallsee, improved precision and accuracyof locationof earthquakes throughoutthe world duringthe 1950sand 1960swas a factorin the developmentof plate tectonics. The end productof the diverseefforts and effectsdescribed above, as carriedout or experiencedby a heterogeneousgroup of scientistsworking with little funding,scattered throughout the world, and havinglittle in common except for an intenseinterest in how the earth shakes,was, of course,correspondingly diverse. What evolvedwas a globalarray of variouskinds of instrumentsoperating only on landat locationsthat otherwise seemedselected almost at random.There was a varietyof recordingfor- mats;there was initially no centerfor dataarchiving. Readings were com- municatedby mail, often monthsafter the event.Observations in the form of seismogramswere circulatedfrom seismologistto seismologistrarely andthrough the courtesy of individuals.From the viewpoint of oneindoc- trinatedin orderand organization,the situationin earlyseismology must haveseemed chaotic. But, in spiteof growingpains it wasnot all bad. Much of it wasgood, and the early efforts were the basis for the improved system that would follow. As early seismographicobservations on teleseismsaccumulated, and it quicklybecame apparent that, for sufficientlylarge earthquakes, stations throughoutthe world recorded data that were complementary and had to be studiedtogether, seismologists felt theneed to amassand analyze data at a centralplace. They began to setup organizationsand communication modesto accomplishthis task. Interestingly, one of thevery first groups to recognizethe value of a globallycoordinated seismograph network was madeup of membersof the religiousorder of the Societyof Jesus.The Jesuitshad sometalented scientists, a strong interest in learningjust how the earthworked, and a uniquenetwork of globally-distributedmissions. They establishedseismographs at as many as possibleof thosemissions, sentthe data to a centralpoint, and from therepublished locations and otherinformation on earthquakesthroughout the world. Theywere an importantfactor in theearly development of thescience of seismology. Two internationalcenters sprang up in Europe, one the Bureau Centrale Internationalede Seismologiein France and the other, the InternationalSeismological Summary, or Center,in England.Both pub- lishedlocations of eventsand the ISC in particularpublished extensive readings,mostly arrival times of seismicphases, for all stationsreporting on a particularearthquake. The ISC becamea kind of final archivefor most suchdata. Other organizations, usually the centers for localor regional Observingthe Shaking 9 national networks also sprang up. Some examples are the Dominion Observatoryin Canada(where seismology co-existed with astronomy),the JapaneseMeteorological Agency (likewise with meteorology),and the U.S. Coastand GeodeticSurvey (seismology with geodesyand surveying).The USC and GS performeda valuable serviceby providing relatively quick preliminary locationsof earthquakeepicenters and distributingthem by postcard."Quick" in the 1950smeant within a few weekswhereas the ISC "final" locationsoften took many monthsor yearsas the centerwaited for communicationof data that sometimesarrived only in the form of station bulletinsprepared and distributedannually. In contrast,using data from today'sglobal and electronically-connectednetwork, hypocentral locations are available within minutes or hours of the time of the shock. Not all of the early networks were government operated. The SeismologicalLaboratory of CalTechand the University of California at Berkeley,for example,both operatedlocal networks primarily for study of earthquakesin southernand northernCalifornia, respectively. For all of theseservices, local, regional or global,emphasis was on the time of arrival of seismicphases. As we shall see in the next chapterthe arrival times were a very important sourceof informationon the earth's interior. However, there is much additional information in the form of amplitudes,frequencies, wave character,etc. The seismologistwho wished to study such featureshad to requestcopies of the original seismograms separatelyfrom eachof the individual stations.For a globalstudy that was a chore,often an exercisein languageand communication.And therewas often a long delay while seismogramswere beingcopied (no handy Xerox copier then), and errors or gaps in informationtransmittal often crept in alongthe way. Somefurther standardizationand organizationwas clearly needed.It would becomepossible as new funds for supportof the science of seismologybecame available because of governmentinterest in a nuclear testban treaty. With the startof the atomicage in the mid 1940s,a new era of the subject of seismologybegan, although it was sometime beforeseismologists, or anyone,recognized how largethe impactwould be. The very first atom- ic explosionin New Mexico during WWI! was recordedby seismographs. In the post WWI! years as the Cold War flourished,larger and larger nuclear explosionswere tested,some so large that they could readily be detectedby seismographsthroughout the world. Networks of seismographs designedand operated to monitor earthquakesrecorded these events,as did clandestinenational networks of seismographsdesigned to monitor the test explosionsof other nuclear powers. Eventually nuclear testingand the Cold War itself causedsuch high levels of concernthat 10 Observingthe Shaking meansfor curtailingnuclear testing was soughtby the politicalpowers. In 1958, seismologistssuddenly and unexpectedlyfound themselvesmem- bers of political delegationsto internationalnegotiations on a nucleartest ban treaty.They were the technicalexperts on undergroundtest detection. For most,it was a strangeleap from the musty basementof the laboratory where earthquakeinstruments were maintainedto the statelyand proto- colized halls of Genevawhere seismologybecame both hopelesslyand hopefullyintermingled with internationalpolitics, from that dateto at least the present. The scientific-politicalhistory of the role of seismologyin nucleartest ban negotiationsis well beyondthe scopeof thisbook but a consequenceof that political initiative is not. As a result of the attentionfocused on seismology as the principalmeans for detectingdistant underground nuclear explosions,it becameevident to governments,as it had long beento seismologists,that seismologycould, and becauseof hope for a treaty should, be improved,that given additionaleffort, there was potentialfor major advancein seismology,and that the field of seismology,compared to man•; scienceswas underfunded.In 1959,a panelof seismologists(of whichI was one)and otherscientists under the chairmanshipof Lloyd Berknermade a numberof recommendationsfor strengtheningthe U.S. capabilityin seismology.As a consequencenew funding flowed into the field and activity accelerated.Bright new studentsand scientistsand engineersfrom related areaswere attractedinto the subjectand seismologybegan to flourishin the early 1960sas it neverhad before.Coincidentally it was alsothe time when the forerunnersto what would becomeplate tectonicsbegan to appear. One recommendationof the BerknerPanel, authored by David Griggs and Frank Press,deserves special mention here becauseit had an effect especially relevant to the plate tectonics revolution. During the InternationalGeophysical Year 1957-1958,Press and Maurice Ewing had establisheda globalnetwork of matchedthree-component long period seis- mographsat ten widely-dispersedstations. It now seemsa modestattempt to build a standardizedlong-period global network, but it was a majorstep at the time. All the seismogramswere archived at one place (Lamont GeologicalObservatory). It was an effort designedto facilitate research and, in a modestway, it was successfulin doingso. The BerknerPanel built on this experienceand the Griggs-Pressrecommendation called for 100 such stations,each with three matched short-periodand three matched long-period components,distributed around the world. All data, in the form of analogrecords on photographicpaper, were to be sentto onecen- ter where they would be microfilmed for distribution to researchersas Observingthe Shaking 11

The US Coastand GeodeticSurvey undertook the taskof installingthe networkand operatingit and the data facility.Most of the instrumentswere installedat existingstations where they had the careof interestedseismol- ogists.It was a huge success.For the first time there was a network of numerousstandardized seismographs throughout the world (it was called the World Wide StandardizedSeismograph Network) and it transformed muchof the activityin seismologyto a higherlevel, particularly, as we shall seein a later chapter,the effortrelated to plate tectonics. At this point, and on that note, it would seem reasonableto end this chapteron observationsand move on to the next on the analysisof same. However,I am reluctantto do soabruptly and therebyleave the impression that seismologywould necessarilyhave been better off had early seismolo- gistsgotten together and standardizedtheir observationsearlier, foregoing in the processthe less-coordinated,instrument-developing activities. Certainlystandardization was a greatboon. Furthermore the WWSSN has beenfollowed by still greaterand more elegantstandardization based on digitalrecording and moderncomputing techniques, all muchto the bene- fit of the scienceof seismology.I support this trend and dreadto think of how backwardseismology might be without it. However,I alsosupport a certainlevel of lessregimented, less fettered activity in a science. The free-wheeling,unorganized activities of early seismologywere accompaniedby a spirit of wonder and explorationand innovation.I am now concernedthat, with the highly-organizedsystems of today,seismologistsof the presentand futuremay losethe freshnessand the versatility and the stimulusof new ideasthat pervade a field becausemany of its par- ticipantsdo thingsthat are differentin every way from thoseof their colleagues.It is easyto fall into the trapsset by routineand conventionand it requiresspecial effort on the part of the individual to break free of those traps. A sciencethrives on freshness,innovation and vision. I hope this point will becomeeven more clear as we proceedto following 2

LearningAbout the Earthand About Earthquakes

Theearly seismological players Thoughfound wanting at timesas soothsayers Jobperformed quite superiorly As theyprobed earth interiorly Toreveal multi-concentrical layers.

manylocations around the earthas describedin the lastchapter, and 'hethedevelopment earth's continuing ofreliableseismic seismographsactivity thatand typically operation produces ofthemannu- at ally more than a hundredshocks large enoughto be recordedworldwide, eventuallygenerated a huge quantityof informationon seismicmotion of many pointson the earth'ssurface following earthquakes at a wide variety of hypocentrallocations. The data were somewhatheterogeneous to be sure, sometimeslacking in standardizationand timing accuracyand calibration,and collectedonly from the land-coveredportion of the earth'ssurface,but they would turn out to be a treasurehouseof informationon the natureof earthquakesand on the structureand compositionof the earth's interior. Determiningjust how to analyze thosecomplex observations so that the informationthey containedcould be best organized,comprehended, and madeuseful to othersas a part of sciencewas no simpletask, however. But early seismologistswith inspiration,insight, perseverance,dili- gence,and good fortune led the way and eventuallythere evolved an understandingof both earthquakesand the interiorthat becamea foundation for the conceptof plate tectonicsand, indeed,for mostof our under- standingabout the earth'sinterior today. Let us sketchbriefly the manner in which that knowledgeevolved. Virtuallyall of the datawere recorded in analogform, so-called wiggly- line recordsin which the tracedescribed one componentof groundmotion as a functionof time. Early in the game, one point was obvious.Near the source,most of the earthmotion sensed by observers,or seismographs,was 14 LearningAbout the Earth and Earthquakes of short duration, perhapsonly a few secondsat most.At distant seismograph stations,the motion,though lower in amplitude,lasted much longer, perhapstens of minutesto an hour or more.The messagewas unavoidable. As is the casewhen a flashof lightning generatesa long, low acousticrum- ble of thunder, complexpropagation of seismicwaves as they travelled through the earth was turning a near-impulsiveseismic disturbance into a prolonged wave-train at distance.In fact, the seismogramfor a distant shock,or teleseism,typically began with a seriesof distinctnear-impulsive events.The serieslasted many minutesand the eventswere superimposed on a lower level of unrestalso generated by the earthquake.The impulsive eventswere followed by a long-drawnout, oscillatorywave train of large amplitudesand sometimesa half hour or more in duration. Recognizing that therewere messagesabout the deep interior hidden in thoseimpulses and wave trains,seismologists set out to make senseof the complexity.One focusof attentionwas the very first wave to arrive. The seismicrecord often began abruptly,a consequence,it turned out, of the compressional,or P, wave, the fastestwave traveling through the interior.Partly because,as the first arrival, it could be so easilyidentified and its arrival time measuredso precisely,the P wave becamea prime sourceof very useful information.For one thing, and particularlyafter its velocitybecame known, data on P wavesfrom a numberof stationscould be usedto locatethe initial earthquakedisturbance accurately in both space and time. Successiveapproximations and some bootstrapping were required, but eventually this technique became highly refined and it remainsthe best method for locatingnatural seismicdisturbances today. Findingthe velocityof the P wave, or to statethe mattermore precisely, the travel-timeof P to any distantlocation, took sometime and effort,but once in hand, P-wave travel times also became an invaluable source of information on the earth's interior. Early in the gameseismologists discovered what somemust have antic- ipatedbut to otherswas a completeand astonishingsurprise. To the level of precisionprevailing at that time, the travel-timeof P wavesbetween two points on the earth'ssurface was the sameas that betweenany two other pointsanywhere on earth,so long as eachwas the samedistance from the earthquakeas its counterpartin the other pair! This simplebut important observationmeant that the earth at depth was sphericallysymmetric, that there is no large lateral variation at any given depth level. The velocity varieswith depth but it doesnot vary much laterally.And the precisionof the resultwas very high for that time, muchbetter than 1%. Thisresult seemed natural enough to onegroup of scientists,those who commonlydealt with simplephysical or mathematicalmodels of earth LearningAbout the Earth and Earthquakes 15 the planets.To them, mostlyphysicists and geophysicists,this observation was yet anotherstep in their game of making nature simple, and readily amenableto study throughmathematical modeling. They liked what was effectivelya layeredsphere that couldbe easilyrepresented mathematically. To others,particularly those steeped in observationsof the geologyand geographyof the earth'ssurface, simple spherical symmetry was a surprise. Exceptfor the fact that the surfaceis nearlya perfectsphere geometrically (neglectingellipticity for a moment),the featuresof the surfacesuch as continents,oceans, mountains, islands, etc. show no signof sphericalsymmetry whatsoever!The simpleseismic P-wave travel timestherefore seemed a little incongruousto those,mostly geologically-oriented, observers. Thesetwo differentperspectives, one a view of earth by the modellers as relatively simple, the other a view of earth by the field observersas rathercomplex, prevail today and this schismwill receivemore attention as we progressin this storyof plate tectonics.Among studentsof the interior, the view of the earth as sphericallysymmetrical held the day for many decades.But as we shallsee, one of seismoiogy'searly importantcontribu- tions to plate tectonicsis based on the observationof departuresfrom sphericalsymmetry in certainparts of the interior.One might say that this studybroke a psychologicalbarrier that had beenimposed by long accep- tanceand relianceupon the sphericallysymmetric, simple model. And this trend is continuingat presentas modern seismologiststry to map tiny velocityanomalies, i.e. departuresfrom sphericalsymmetry throughout the earth.Nevertheless, spherical symmetry remains a fine first approximation and early work on this topicwas well done and importantand shouldnot be denigrated,even though the resultmay have createdthat psychological barrier that had to be overcome in the 1960s. The techniqueof studyingthe travel time of distinctimpulsive events on the seismogramwas soonextended from the first-arrivingP wave to includeall other suchevents including shear, or S, wavesand variouscom- binationsof P and S wavesthat had left the sourceas one type and reflected, refractedor convertedalong the way. Eventually these "travel-time studies"provided a completeset of travel-timecurves for a wide variety of phasespropagating in the earth.And thosetravel-time curves became the basisfor deducingmost of what is knownof the structureof the deepinte- rior. The outer core,for example,was found and measuredon the basisof travel times and grossamplitude anomalies.The inner core, the various regionsof the mantle,even the mantleitself, were found largely through study of seismicwave travel times. Many seismologistsparticipated in suchstudies; two groupseventual- ly led theway. Gutenberg and Richterat Cal Techproduced a setof 16 LearningAbout the Earth and Earthquakes that documentedthe traveltimes of mostseismic body wave phases.Their work was basedheavily on thoroughobservations and analysisof those observationsmade by them through tirelesseffort and perseveranceand was a magnificentcontribution to sciencethat has clearlystood the testof time. More or lessconcurrently, Jeffreys and Bullenat Cambridgeindepen- dently developedcomplete seismic body wave travel timesin similarbut slightlydifferent style, relying heavily on assembledreadings by othersas collectedat internationalcenters and on soundand carefulstatistical analysis.At the completionof the individual efforts,the differencesbetween the resultsof the two groupswere minor, and earth scientistshad reliabletrav- el time curvesthat couldbe usedwith confidencefor locatingearthquakes and deducingthe structureof the earth'sinterior (for the sphericallysym- metricalearth!). The determinationof near-completetravel time tablesfor all principalseismic body wavesis surelyone of the mostimportant scientific studies of earth ever made. Someinsight into how scienceworks, insight that may be instructivefor youngscientists, can be gleanedfrom the sequelto thisstory. Gutenberg and Richterpublished their resultsin the normalform of scientificpublication, that is as a seriesof articlesin scientificjournals. Different segments of their work appearedat differenttimes and all werenot readilyavailable as a com- pact,well-organized unit. Jeffreysand Bullenpublished in journalsbut also publisheda completeset of traveltimes in convenientform in a handybook- let that couldbe readily obtained.Consequently, the Jeffreys-Bullentables (or JB tablesas seismologistscall them) becomewidely-known and were used as the world standardfor many years.The lessonhere is that good communicationof results,not just routinepublication of them,is oftenwhat make a pieceof scientificwork influential. The intensivestudy of body wavesthat led to the travel time tablesalso broughtsome evidence of otherkinds to light. For the core,or mostif it, P waves were observedbut not S waves, strongevidence, indeed, that the outer core is liquid. For the mantle however, the vast portion of earth betweencore and crust,the travel timesof P and S wavestraveling along the same path seemedcompletely compatible and reasonable,i.e. the P wave traveled consistentlywith velocityslightly lessthan twice that of S, about the ratio expectedfor body waves in an elasticsolid suchas rock. Thus it seemedfrom seismologythat the mantle,with the possibleexcep- tion of isolatedmagma bodies here and therenear volcanoes,was solid. Meanwhile studieswere being made of glacial rebound,the recovery over thousandsof yearsof the continentalcrust following removalof the ice load of the Pleistocene.Particularly for Scandinaviawhere the observationswere exceptionallygood, such studies indicated that at leasta part LearningAbout the Earth and Earthquakes 17 the upper mantlewas flowing, behavinglike a liquid of high viscosityand not the purely elasticsolid seismologistshad deduced.Eventually the con- flictingseismic and reboundresults were reconciledby assigningthe mantle propertiessomewhat like thoseof "silly putty," which respondson the long term like a viscousliquid but on the shortterm like a solid,propagating both compressionaland shearwaves. The mantlewas indeedsolid, but at leastsome of it could alsoflow very slowly. Then Gutenberg,in a study of seismicwave travel times and ampli- tudesat moderatedistances from the source,found in the upper mantle a "low velocitylayer," i.e. a layer in which the seismicvelocity, at leastfor S, decreaseswith increasingdepth rather than increasingwith depthas it normally doesfor the restof the mantle and then remainslower than the velocities of the overlying layer for a certain range of depths.These results, which among other things were subtle and demonstratedGutenberg's remarkableintuitive senseof seismicphenomena and their causes,were corroboratedby others.Although not interpretedas suchat the time, the conceptof a "low velocity layer" in the upper rnantlebecame irnportant during the plate tectonicsrevolution as a possiblesign of weaknessand ability to flow and hencean indicationof "asthenosphere,"a weak layer beneaththe "lithospheric"(or non-flowing)plates above. Gutenberg's dis- coveryis a prime exampleof a soundly-basedpre-plate tectonics observational resultthat couldbe reinterpretedto fit and supportthe plate tectonicsmodel as it was beingconceived. In spiteof the powerful evidencefrom the travel time curvesthat the earthwas, to a goodfirst approximation,spherically symmetric and hence generallylacking in lateral variation, the obviousobservations of surface relief and near-surfacegeology that fail to fit this simplepattern, could, of course,not be ignored. Consequentlyseismologists sought lateral differences,especially in shallowlayers of the earth,the crustand the bounding uppermostmantle. Body wave studiesbased on near-earthquakesdata provided someinformation in a few placesnear localearthquakes. Teleseismic body wavesreflected from the surfacealso provided some information on the crustnear the point of reflection,but earlyon suchevidence was neither reliable nor definitive. Eventuallyattention turned to the later-arrivingtrain of waves (sometimescalled the "coda"or tail) that followedthe earlier-arrivingand pulse- like bodyphases. They were surface waves, energy trapped near the surface becauseof the freesurface and shallownear-surface layers of slowvelocity. They were of two types,named after their discoverers,Rayleigh, and Love, respectively.The surfacewaves had large amplitudes,typically the largest on the seismogramof a particular shock,and were drawn out into 18 LearningAbout the Earth and Earthquakes oscillatory,or dispersed,wave trains.The waveswere dispersed,it turned out, in accordancewith the complexrock velocity structure of the near-surface.Oceanic paths, for example,produced wave trains of completelydifferent characterthan did continentalpaths, largely becauseof the water layer but alsobecause of lateraldifferences in the underlyingcrustal rocks. Eventuallysuch data, in conjunctionwith otherinformation discussed in the nextchapter, showed, unequivocally and contraryto somepending geolog- icalhypotheses, that the crustalrocks of the oceanswere differentfrom those of the continents.Hence the oceanswere not simplysubmerged portions of old continentsas somehad suggested.This was an importantadvance during the early stagesof the studiesof the oceanfloors that would lead eventually to the conceptof seafloor spreading, the forerunnerof platetectonics. Still longer surfacewaves provided informationon the upper mantle that corroboratedto a degreeGutenberg's evidence for a low velocitylayer there.Eventually even longer waves, and the so-called"free oscillations" of earth they producedafter multiple circumlocutions,would provideinfor- mation on the deep interior complementaryto that derived from body- wave travel times,but suchinformation, though important otherwise, was not a key factorin the plate tectonicsrevolution. At any rate, the studiesoutlined so briefly in the foregoingcollectively producedan understandingof deepearth structure that would be especially useful during the plate tectonicsrevolution, particularly to thosewho soughtto understandjust how the earth could produceand maintain the moving systemof platesthat was the essenceof the theory. The attention of early seismologistswas not exclusivelydirected to studiesof the deep interior,of course.A significantfraction of their collec- tive effortwas devotedto attemptsto understandearthquakes. One important questionconcerned the basicnature of the earthquakesource. What happensto makethe earthshake? Were earthquakes natural underground explosions?Or collapses?A shiftingor rupturing or faulting of rockmass- esunder stress?Was there a singlecommon cause of earthquakesor a variety of causes?The greatSan Franciscoearthquake of 1906,among others, providedpart of the answeras geologicalobservations of surfacedisplace- ments coupledwith seismologicalstudies left little doubt that the earthquakehad occurredbecause the SanAndreas Fault had brokenor ruptured, generatingseismic waves in the process.The idea that the rupturing along a fault plane suchas that in Californiawas a consequenceof tectonicstress was soongeneralized and carriedfurther by seismologists,and applied to almostall tectonicearthquakes. In Japan,and in an importantpioneering study, Honda showedthat, if the directionsof first motionsof P wavesat a numberof seismograph LearningAbout the Earth and Earthquakes 19 tionswere projectedback to the focusof the earthquake,there was a complex but neverthelessconsistent spatial pattern of wavesradiated from the shock.Initial compressionswent in somedirections, initial rarefactionsin others,and the radiation pattern was just that expectedfrom a rupture along a segmentof a plane with appropriate orientation in space and appropriatedisplacement of rockson opposingsides of the fault. Eureka!! From this information,seismologists could not only deducethat the earth had rupturedbut alsodetermine the orientationof the fault plane in space and somethingabout stresses deep in the earth.Honda's brilliant lead was followedby others.Byerly and his studentsat Berkeley,for example,developedthe methodfurther and appliedit to selectedlarge earthquakes, mostly thosethat had happenedrecently. Initially, however,only a small number of earthquakeswas studiedin this manner. In the 1950s,John Hodgson of the DominionObservatory in Ottawa and the sonof ErnestHodgson, a pioneerin seismologyin Canada,developed a projectthat was visionaryand of a style that would turn out to be very important in the plate tectonicsstory• But "-'•"•u,L,,•, tunate,y, • it' had one serious problem;it was before its time. Hodgson set out to apply Honda's and Byerly'smethod on a globalscale using data on many earthquakesdistrib- uted throughoutthe world. The objectivewas to seekconsistencies in the globalpattern of focalmechanisms that might revealsomething fundamental aboutearth tectonics.The projecthad somelimited success,but basical- ly it ran agroundand failedto producethe globalresults hoped for, simply becausethe raw data were not yet reliableenough. Lack of standardization in the seismographicnetwork, plus the poor communicationof data at that time, led to so many inconsistenciesin the interpretedreadings of the first motion of P that focal mechanisms were often unreliable and hence misleading.Inconsistencies were socommon, in fact,that it was thoughtat first by somethat the basicassumptions of Honda'smethod were incorrect,and some lost faith in it. When the World Wide StandardizedSeismograph Network began to producedata however,it becameclear that the method did work well much of the time, and it also became clear that the inconsis- tencieswere artifactsand not reliableobservations of earth.Thus Hodgson was vindicatedand he deservescredit for developinga well-conceivedand importantnew kind of project,but it was otherswho madea timelyentrance to the field just as the WWSSN becameproductive who were able to make the criticalstudies that addedto and supportedplate tectonicsin its early stages.Focal mechanism studies of the type developedby thesepioneers continueto play a major role in seismotectonics. Seismologistssought other kinds of information on earthquakesin additionto focalmechanisms, of course.As the capabilityfor precise 20 LearningAbout the Earth and Earthquakes tion of earthquakesin spaceand time and for crude measureof their size developed,it becamepossible to attackmany questionsabout earthquakes. Where do they occur?When do they occur?How largeare they?How often and in what sequencedo they occur?Why do they occur?Most of these questionsrefer to a branchof seismologycalled seismicity, essentially the geographyof earthquakes.It would turn out, particularly,that just where earthquakesoccur was a mostimportant piece of informationduring developmentof the plate tectonictheory. Mapping and understandingof the pattern of the earth's seismicity developedgradually. Early in the game,of course,it becameevident that earthquakesdo not occur randomly over the earth. Rather they occur repeatedlyin certainnarrow and elongatezones, zones that oftenare, more or less,coincident with other readily observedtectonic features, such as volcanoes,mountains, island arcs,and trenches.Initially thosezones were not recognizedas interconnectedon a global scale,but once a sufficient numberof earthquakeshad beenlocated, a globalpattern emerged. It was recognizedthat earthquakestend to occurin narrow belts that encirclethe earth. Later it would be recognizedthat such a coherentglobal pattern implied a driving mechanismof globalscale. In 1949Gutenberg and Richterpublished an authoritativeand compre- hensivebook that quickly becameknown informally as "the bible of seismicity." Through exceptionaleffort they had located or relocatedmost earthquakesthroughout the earth that were sufficientlylarge to be well- recorded,and in that book they presentedthe informationin text and on globaland regionalmaps. By far the besteffort of its kind at the time, their work producedmuch informationon seismicitythat is now considered basic.However, Gutenberg and Richter'swork was basedon data from the heterogeneousseismograph network of the first half of the centuryand on painstaking,but limited, calculationsby hand calculator.Hence locations were not nearlyso accuratenor the data nearlyso completeon a particular shockas they becamein the 1960swith the adventof the WWSSN and digital electroniccomputers (NGT, Fig. 15). Nevertheless,the gross pattern of the major global seismicbelts, thoughnot all of the detailsthat would be importantto the developmentof plate tectonics,was recognizable.It was clear that earthquakestend to occur in narrow belts that encircle the earth, that sometimes intersect, and that rarely end by taperingoff to negligibleactivity. All seismologists,and many other earth scientists,were generallyfamiliar with that global pattern. Strangely,so far asI know,and thereis an importantlesson here, none of us thoughtto askwhy certainfeatures of that particularpattern were as they were. For example, when two seismicbelts intersectedthey LearningAbout the Earth and Earthquakes 21 crossed.There were alwaysthree arms to the intersection,not four. As we shall see,the globalpattern of the beltsis consistentwith the story of the plates.It might have led us closerto that storyearlier had we emphasized study of the seismicpattern but somehowthat point escapedus for sometime. We had all learnedin high schoolscience classes that we had to "ask the right questions,"but in this casewe ignoredthat advice. As Gutenbergand Richterdocumented in their book, mostearthquake foci are shallow,but in certaindiscrete and limited zonesevents as deep as 700 km occur.Deep earthquakeswere known prior to publicationof their text, mostlyfrom studiesin Japan,where Wadati had shownin a pioneer- ing and initially controversialstudy that deepearthquakes do indeedoccur, and that they occurredin a thin zone that dipped neatly beneath the Japaneseisland arc. It alsobecame known early that nearly all deep earthquakeswere associatedwith islandarcs. But the globalpattern of the deep earthquakezones, and the arcs,would not be explaineduntil the 1960s. Many attempts to understandthe pattern of the arcs and the deep SrlOCKSwere made prior to that time of course,t.!d .... U t till• •.•,•1tcll•l. •hA_,•...•.;D kJILkkJll rifteD. _• For example,Benioff associated the deepearthquake zone of Wadati with a major fault somewhatlike that, but without the huge displacements,of moderntectonic theory. As a consequenceof the early work, suchzones are now termedBenioff, or Wadati-Benioff,zones. Although I havethe highest regard for Benioff and his many contributionsto earth scienceand am pleasedto honor him, I prefer the name, Wadati-Benioffzone, becauseit recognizesWadati's extraordinaryefforts as well. Studiesof deep shocks during that era were hampered,because, with the exceptionof the Japanese arc,deep earthquake zones were generallynot well-instrumentedwith seis- mographslocated in the areaabove the earthquakeloci. Thusthe depthsof most shocksin those early days were determinedfrom waves reflected from the surfaceto distantlocations. Although this methodis sound,and still used today,its applicationsuffered then becauseof deficienciesin the quality of data and data interpretationat that time. Studiesof seismicity,particularly the work of Gutenbergand Richter, includedmuch informationon sizes(or magnitudes)of earthquakes,frequency of occurrence,aftershocks and foreshocks,etc., but this kind of information,though doubtlessly important otherwise, played only a minor role in the developmentof plate tectonicsand then largely in a qualitative sense. In addition •o •he s•udiescited aboveand basedon body wave •rave] times,surface wave dispersion,and grossseismic wave amplitudes,during the pre-1960sinnovative seismologists produced some less conventional studiesbased on other characteristicsof the seismographicdata. As 22 LearningAbout the Earth and Earthquakes group,studies of this type were not importantto the plate tectonicsrevolution. A few that were will be cited in later chaptersas their relevancedic- tates. To summarizethis chapterbriefly, it might be said that seismological studies of natural earthquakesduring the first half of the 20th century made unprecedentedsense of the information recordedby seismographs and establishedseismology as the principal sourceof informationon the earth'sdeep interior.But it was obviousto seismologiststhat major problems and major opportunitiesfor further advanceremained. Let us now turn our attentionfrom natural earthquakesto artificial sourcesof seismic wavesin orderto seewhat additionalinformation they providedto aid and abet,and in somecases to hamper,the plate tectonics 3

Artificial Sources

Deepsubterranean inspection By themethod of seismicreflection Meansknowing much clearer Eachlithological mirror, A stepto geologic perfection ?

romthe foregoing it should be more than clear that, though sometimes destructiveand terrifying, natural earthquakes, through the seismic wavesthey generate,are an invaluablesource of informationon the earth'sinterior. And, convenientlyfor the scientistswho study them, if not the public, natural earthquakesare abundant hence frequent, powerful, widely-distributedover many regionsof the earthand throughdepths ranging to 700km, and, assources of seismicwaves, inexpensive, even free. Thus the substantialearly effort to study natural earthquakesthat we have been discussingwas mounted,and thus earthquakesbecame a criticalsource of informationas the conceptof plate tectonicswas beingdeveloped. Thereare somedrawbacks to the use of earthquakesas sourcesof seismic waves for scientificpurposes, however. Earthquakes cannot be predicted precisely,so specialinstrumentation cannot be laid out before the shockoccurs. A fortuitouselement is always presentin study of natural earthquakes.Location and origin time are always determinedfrom analysis of the seismicwaves the quake generatesand henceare never so well- known as they might be if directgeodetic and chronologicalmeasurements could be made. And, of course,earthquakes do not occurat many of the placeswhere a seismologistmight like to have the sourceof seismicwaves for scientificpurposes. Early seisrnologistsquickly recognized these shortcomings and alsothe advantagesthat might be gainedfrom use of artificial sourcesthat couldbe preciselylocated and timed, and monitoredby arrays of specialinstru- mentsthat need only be deployedtemporarily. As a consequence,and one way or another,artificial sources have been used in seismologyfor many 24 ArtificialSources

, decades.A representativeexample is the study of seismicwaves generated by the detonationof a huge quantity of surplus military explosivesat Helgolandfollowing World War I. Thoughoutdated now, at the time that studyprovided unprecedented information on the crustof WesternEurope. Artificial sourcesof seismicwaves need not be largeto be useful,however.Weight drops and hammer blows generatedetectable seismic waves that penetratethe eartha few tensof feetand soprovide information on the topmostlayers of soil and bedrock.At the other extremeof scale,and as noted earlier,waves from large nuclearexplosions can be detectedafter travelingthrough the very centerof the earth.Studies of intermediatescale use as sourceschemical explosions of varioussizes, rapidly expandingair bubblesin water, and huge truck-mountedvibrators whose oscillatory signals canbe made to simulateimpulsive sources with appropriatecomput- er processing.In this chapter,and elsewherein later chapters,we shalldis- cuss studies of seismicwaves generatedby various of these artificial sources,concentrating, of course,on thosethat had an obviousinfluence on the developmentof plate tectonics.The variety in suchstudies is so great that we must beginwith somecategorization. Almost all studies that use seismic waves generated by artificial sourcesof small to moderatesize fall into one of two classes,typically des- ignatedby the terms "reflection"or"refraction." In seismicreflection studies,a sourceat or near the surfacegenerates compressional waves that are partially reflectedat buried interfacesmore or lessbelow the sourceand returnedto the surfacewhere they arerecorded by an array of seismicwave detectorsdeployed near the sourcelocation. The entire,and relativelycom- pact, operationis moved along the surfaceand the sourcerepeated fre- quentlyso that eventuallya seismicprofile is obtained.A seismicprofile is analogousto the echogram,or water bottom profile, commonlyobtained by the depth sounderon a boat or ship, likewisethrough the use of sonic sourcesrepeated at adjoiningsurface locations. In seismic refraction studies, the source is also near the surface but the sensorsare deployedalong a line extendingto horizontaldistances of, typically,about ten timesthe depth of the deepestinterface being probed. The key. information is obtained from waves that travel obliquely from the source to a near-horizontal interface below, are refracted so as to travel along the interfaceuntil again refractedup to surfacedetectors. For this methodto work, the wave velocitymust, in general,increase with depth, which it normally doesin the earth. The information obtainedby these two methods is complementary. Refractiontends to providebetter information on rockvelocity, and reflection better resolutionof rock structure,although these generalizations ArtificialSources 25 oversimplificationsand do not always hold. With this basicbackground, however,we can discusssome more specificapplications and summarize the results. The mostcommon application of the refractiontechnique is in study of the thicknessand velocity structureof the continentalcrust. Initially such investigationsused near-earthquakes as the source,and with somesuccess. Such work led to the discoveryby A. Mohorovicic of the crust-mantle boundarythat bearshis name,or an abbreviationof it. But then,for the rea- sonsnoted above,artificial sources,such as quarry blaststhat were set off for anotherpurpose but that couldbe timed and locatedprecisely, gradu- ally replacedearthquake sources. Later, chemical explosives of the order of a ton or more of TNT were fired in speciallydrilled holes,or in bodiesof water,with the primary purposeof generatingseismic waves; these procedures are still used. Seismicrefraction studies of the crusthave been carried out in many countries,including the USA and most notably the SovietUnion where a high level of activityprevailed for decades.The particularresult of all this work that bore mostheavily on the developmentof plate tectonicswas the clear demonstration of the existence of the Moho, or crust-mantle boundary,at a depth of about40 km mosteverywhere beneath the continents.The depthvaries somewhat from placeto placebut is alwaysof that order.Thus when comparablestudies were alsomade of the oceanbasins there was evidence to demonstrate the contrast in thickness and hence the clear distinc- tion betweencontinental and oceaniccrust, a point of considerableimpor- tancein the globaltectonic story. As a result of the success of refraction work on land, the earliest attemptsat seismicstudies of the sea floor were based on the refraction method.After somepreliminaries in which explosivesources were deto- natedand instrumentswere operated on the oceanfloor, a simplertwo-ship techniquewas devised.One shipfired explosivecharges at shallowdepths; the other listenedwith hydrophones(pressure sensitive devices) also at shallow depths. Successiveshots were fired as the ships moved apart. Wavesrefracted through the oceanicsediments, the oceaniccrust and the underlyingmantle were recorded. Thiswork was pioneeredon a smallscale by MauriceEwing beginning in the 1930s,but only beganto flourishin the postWW II years,still under the leadershipof Ewing who was by then building the LamontGeological Observatory,now Lamont-DohertyEarth Observatory.Raitt at Scripps Institute of Oceanography,Hersey at Woods Hole Oceanographic Institution and Hill at CambridgeUniversity also developedsuch pro- gramsduring this era. The earliestefforts at obtaininga representative 26 ArtificialSources ple of the seismicproperties of deep oceanbasin crust were plaguedby problemssuch as a) working too closeto shore,b) usingexplosive charges that were too small and c) failure to observetruly reversedprofiles. Eventually,and in spiteof thehigh costof operatingships and thehuge size of the oceanbasins, enough data were collectedso that the typicaldeep sea crustalcolumn could be described(Officer et al, 1952)and somevalid gen- eralizationsabout the nature of the rocksof the deep seafloor everywhere could be made. The principal,and surprising,result was that the oceaniccrust is everywhere muchthinner than the continentalcrust. It is only about5 km thick in contrastto 40 km for the continentalcrust. Hence the Moho, or top of the mantle, is only at a depth of 10 km or so beneaththe oceansurface. This observationmeant that oceanicareas could not simplybe placeswhere part of an old continenthad subsided,as many earth scientists,some perhaps influencedby Plato'sstory of the lost continentof Atlantis,thought at the time. The oceancrust was somethingdifferent, and musthave been formed in a differentway, a key point later as the conceptof sea-floorspreading, an early stepin the developmentof plate tectonics,was proposed. Seismicrefraction work at sea provided yet anotherobservation that was an important,if not critical,clue to the understandingof globaltectonics.The oceaniccrust, it turnedout, is overlainby a layer of sedimentsthat is, in general,remarkably thin, perhapsless than a kilometeror so in total thickness.Had the oceanbasins existed as they now are for a largefraction of earthhistory, the sedimentsshould have been much, much thicker, given the ratesof sedimentdeposition that are known. Thus thosewho thought the oceanbasins had existedas they are now throughoutmuch of earthhistory,the followersof the so-calledpermanence of oceanbasins hypothesis, facedwhat becamea well-knownenigma when the observationsof sediment thicknessclearly demonstratedhow thin thosesediments are. Earth scientistseventually were drivenby the enigmato a new modelfor creation of oceanbasins, namely sea-floorspreading. Sea floor spreadingresults in geologicallyyouthful oceanbasins, and hencethin sediments,and it was a major step toward plate tectonics.Still more informationon the history of the oceanbasins was availablefrom seismicexploration of the sediments but, it would turn out, it was the reflectiontechnique, not refraction,that would producethe bulk of it. With the exceptionof samplingby drilling which at best can only be carriedout on a limited scale,seismic reflection studies in generalprovide the best and most detailed information on the uppermostlayers of the earth. By far the most extensive,and intensive,use of the seismicreflection techniquehas been, and continuesto be, in the petroleum ArtificialSources 27 where it is the basisfor a multi-billion dollar per year explorationeffort. Initially,and shortlyafter WW I when it was recognizedthat seismicexploration might be rewarding in the searchfor hydrocarbons,the petroleum industryattempted to utilize the refractionmethod for its purposes,but the superiorresolution and more convenientoperation of the reflectiontechnique quickly made it preferable.Thereafter, the reflectiontechnique has been used almost to the exclusion of others in industrial work. Until the 1950s,almost all of the effort was on dry land where it was, of course,applied to explorationof sedimentarybasins, the habitat of most petroleum.Plate tectonicswould eventuallyprove useful in the searchfor understandingof thosebasins, but the seismicobservations of dry land basinsthat were availablein quantity and quality by the 1960swere not a major factorduring the early stagesof the developmentof plate tectonics. Instead it was seismic reflection studies of sediments on the sea floor, includingparticularly the deep seafloor, that had an importantimpact. Seismicreflection studies of the deep seafloor, it might be said, began in earnestabout 1950.As a young graduatestudent of Ewing at that time, I was privileged to observesome early testingof a primitive versionof the reflectiontechnique in an areaof the Atlanticnear Bermuda.Small charges, perhaps 10 pounds of TNT, were fired just beneath the sea surfaceand recordedby shallowhydrophones suspended from the ship.What seemed to be reflectionsfrom within the sedimentarylayer were observedin that early study,but they couldnot be correlatedfrom the siteof one shotto the next,because the shotsin that preliminaryattempt were too widely-spaced. Later Maurice Ewing, his brotherJohn Ewing, and co-workersdeveloped a method for detonatingexplosives thrown from the moving ship much morefrequently, and they beganto trackreflecting horizons and get revealing seismicprofiles of the deep sea sediments,often including reflections from what we now know to be the baseof the sedimentsor the top of the oceaniccrust. The entire sedimentarysection was being probed. This developmentwas a bonanzafor thosestudying marine geologyof the deepsea. Other academicresearch institutions joined the effort.And, as interestin offshorepetroleum flourished, the industryentered the field and developedthe techniquefurther and beganto apply it to offshorebasins, mostly in areas of shallow water. One major advance,adopted by both camps,was the replacementof explosivesby compressedair "guns" as sourcesthat were safe, reliable and highly repetitive. A new era began. Seismicexploration of the sea floor proliferated.With industry exploring the shallowwater basinsin great detail, and the researchinstitutions probing the enormousdeep sea floor and other shallowwater areaswherever they could, a huge new supply of unprecedentedand highly 28 ArtificialSources informationon the sea floor sediments accumulated rapidly. It wasa time of historicadvance in human capacityto observean importantpart of the earth's interior. Of course,and as is nearly alwaysthe casewhen a part of the earth is observedfor the first time in a novel way, thosewho saw the new observationswere surprised,fascinated, and challengedto interpretthem so as to advanceunderstanding of the earth.They soughtknowledge of the history of the oceanbasins and, of course,of tectonicsof major as well as lesser scale.For one thing, the reflectiondata verified, thoughnot without some agonyon the part of the interpretersdue to the inconsistentreflective char- acterof the oceaniccrust and sediments,the surprisingthinness of oceanic sedimentsthat had been revealedby refractionstudies. Data from both kinds of studiesalso showedthat, though thin, the sedimentstended to increasein thicknesswith distancefrom mid-oceanridges. It was another factof considerableimportance that providedincentive and supportfor the seafloor spreadinghypothesis as it wasbeing proposed. And the reflection data, when combinedwith informationon the age of a particularreflector that could be sampledat outcropby coring,permitted somequantitative measurements of rates of sediment accumulation. On one point the reflectiondata unfortunatelyproved misleading,at least during the preliminary stagesof the work; they gave false overall impressionsto scientistsstudying them. The problemarose because both the techniqueand the baseof knowledgewere evolving.The techniqueof reflection profiling, as describedearlier, developedgradually, becoming more sophisticatedand more revealingwith time. In the early stagesthe method workedwell wherethe sedimentswere layeredand flat-lying,but oftenit did not provideuseful data where the sedimentswere deformed.Now thereare manyparts of the seafloor where the sediments,and especiallythe youngest ones,are indeedflat, that is,never deformed. Thus much of the earlymarine reflectiondata showedflat-lying sediments and little else,and certainlylittle evidenceof deformation.Consequently, those who saw thosedata in broad perspectiveand enteredthe debateover permanenceor lack of permanence of the oceanbasins tended to favor permanence.Maurice Ewing, surely one of the greatestof earth scientistsand one who, more than anyoneI know, foresawan upcomingrevolution in geology,was caughtin this trap. He initially objectedto the spreadinghypothesis partly because of lackof evidence on deformationof the seafloor and abundanceof evidenceapparently favor- ing non-deformation,and partly becauseof the appeal of modelsfavoring permanenceand the statureof the proposersof suchmodels. Later as better evidencebecame available he recognizedthe difficulty,reconciled the dis- agreement,and adopteda moreflexible ArtificialSources 29

The difficulty is best illustratedby the seismicreflection data for the trencheswhere they were particularly tricky. The observationsinitially showedonly the obviousflat-lying sedimentsin the trenches,suggesting that there had been little deformation there. Later, as the trenches and the processthat forms them becamebetter understood,it turned out that the flat-lyingsediments are all very young and in the areaof the trenchnot yet subjectto deformation. Elsewherein the trench area however, there is a great deal of severecompressional deformation of sediments,including young material, caughtin the landward wall of the trench,the so-called accretionarywedge. That deformationcould not be resolvedby the technique in its early stages.Hence it initially appeared,ironically, that the trenches,now known to be locatedat the sitesof principaldeformation in plate tectonics,were stable.The seawardwall of the trench,meanwhile and on the other hand, revealedextensional features in the form of grabens.It would turn out eventuallythat the grabenswere evidencein favor of the subductionprocess for formation of the trenches,but initially they were interpretedby someas indicationof extensionalorigin for the entire trench feature.Subduction, which will be discussedin detail in later chapters,is a key elementof the plate tectonicshypothesis, and eventuallyand as quality and resolutionimproved, all seismicreflection data in the trenchareas were reconciled with the subduction model. At presentthe seismicreflection technique continues as the predominant seismictool for exploringthe oceanfloors and sedimentarybasins at sea,and on land, and the resultscomplement those obtained by sampling the sedimentsand crustunder the Deep SeaDrilling Project.The seismic reflectiontechnique is now alsobeing used routinely to study the entire thicknessof the continentalcrust and the uppermostmantle. In a later chapter ! describe the initiation of the COCORP project which was designedspecifically for this latter purpose.COCORP is modeledafter, and is a consequenceof, the successfulapplication of seismicreflection profiling for explorationof variousother parts of the earth during pre-platetec- tonic days. COCORP uses explosive sourcesand also "Vibroseis,"the truck-mountedvibrators referred to earlier.It is providing a new type of fundamentalinformation on the deepcrust of the continentsand hencehas becomea continualsource of new discovery. One other artificial source of seismic waves, nuclear explosions, deservesfurther comment.Except in the SovietUnion where they have been used as sources of seismic waves in refraction studies of the crust and upper mantle, and in the U.S. where testsrelated to seismicdetection of underground nuclear shots have been conductedwithin the context of nucleartest ban treaty negotiations,nuclear explosions have not been 30 ArtificialSources onated specificallyfor the primary purposeof generatingseismic waves. However,most nuclear explosions detonated for otherpurposes are detected by the standardseismograph network and henceprovide seismicdata, usually data with unique characteristics.One story about such data is worth telling, partly becauseof the scientificvalue of the study and partly for the wry humor. When large nuclearexplosions were being testedclandestinely during early phasesof the coldwar, preciselocations and timesof detonation,even the firing itself,were kept secret.Bullen, uninvolved officially with the testing, neverthelesscollected seismographic data on the eventsbecause of his interestin checkingthe Jeffreys-Bullentravel time tablesdiscussed earlier. He found that, using standardseismological methods, he could determine the locationand the origin time of the clandestineexplosions reasonably well from the seismicdata alone. In fact, the times consistentlyturned out to be sonear to the evenminute that Bullensuggested publicly that the det- onations were timed to fire then. This deduction, which was never verified publiclybut which was almostcertainly correct, was somewhatto the chagrin of the testerswho obviouslyshould have chosena randomand hence lessobvious firing time if they wished to keep it secret.In any case,in the remainderof his study Bullenverified the J-Btables and he did soby using artificialsources of seismicwaves in a novel and productiveway. As an aside,I note that nuclear explosionsfired undergroundand in the atmosphere,the stratosphere,and the deep oceanall producedseismic waves of unusualcharacter and henceof specialinterest to seismologists, but studiesof them revealednothing criticalto the developmentof plate tectonicsand hencethey will not be discussedfurther here. Next, with the precedingbackground information on seismologyas a base,we turn to the streamof ideasand eventsthat would providethe setting and the opportunityfor seismologyto contributedirectly to the plate tectonics 4

ContinentalDrift and SeaFloor Spreading, The Forertmners of Plate Tectonics

Theplates in dynamicmosaic Throughhistory both fresh and archaic Likebold engineers Forsome two billion years Havekept earth from becomingprosaic.

earthe end of the16th century, 1596 A.D. to be exact,Abraham Orteliusin Antwerppublished the third edition of hisThesaurus Geographicus.For that editionhe addeda shortpassage to a section from earliereditions that discussedmyths suchas Plato'stale of Atlantis. That new passagecontains a simplebut truly greatidea, the notionthat the continents of North America, South America, Eurasia and Africa were once joined togetherand have sincedrifted apart, creatingthe modernAtlantic oceanin the process.The passagewas apparentlyoverlooked or disregard- ed by scientists,historians of science,and everyoneelse until it was pointed out in 1994, nearly four hundredyears later, in an articleby JamesRomm (Romm,1994). So far aswe know at thiswriting, that publicationby Ortelius is the first recordof that profoundidea in all of history.Even though others would have the sameidea independentlylater, it seemsquite reasonableto assumethat Orteliuswas indeedthe very first to have it, becausehe was in an especiallyfavorable position to be amongthe first to seethe criticalgeo- graphicalinformation on which the idea is founded. AbrahamOrtelius was oneof the leadingcartographers of the 16thcen- tury, the age of great geographicaldiscovery. As schoolchildren know, it was a time when, inspired by the voyagesof Columbus and Vascode Gama,by religiouszeal, and by economicgain, adventuresomeseafaring men sailed to remote and unknown corners of the earth. Somewhat like modern scientistsprobing the atom, or the universe,on the interior of the earth,those observers of previouslyunexplored geographical regions, like present-dayscientists publishing in journals,brought their observationsto a centralcollecting point for organizationand distribution.There a cartog- 32 ContinentalDrift andSea Floor Spreading rapher collatedthem to producea map. The role of the cartographerwas, in a sense,analogous to that of the synthesizeror theoreticianof modern sciencewho assemblesdisparate observations, organizes them into a uni- fied and self-consistentstory, or theory,and publishesit. Orteliuswas just sucha synthesizer.He was not so much a designeror drawer of beautiful maps or a deviserof map projectionsor styles,as he was an assembler of data of others into a consistent whole. He is best known for producing the first modern world atlas, Theatrum Orbis Terrarium,a very successfulcommercial and intellectualendeavor that was translatedinto severallanguages and effectivelysuperceded the chartsof Ptolemywhich had prevailedas the standardworld map for somethirteen or fourteencenturies! Ptolemy's maps, though a magnificentcontribution to knowledgeat the time they were drawn, neverthelessincorporated certain featuresbased on myth or imaginationfor areasthen lackingin observations,and hencewere sometimesmisleading and certainlyin need of major revisionby the time of Ortelius. As an assemblerof mapsof localitiesand regionsby variousother car- tographersinto a patternthat was moreor lessconsistent and realisticon a globalscale, Ortelius was thereforein a positionto be the first, or at least one of the very first, to see the initial, reasonablyaccurate, map of the earth's surface,or at least of the parts of the surfacesurrounding the AtlanticOcean and criticalto the greatidea. Like many greatscientific dis- coverersof the past and present,when he saw and comprehendedthe key data in appropriateperspective, the greatidea cameto him. There is an important and encouraginglesson for discovery-minded young scientistshere. Ortelius was clearly a wise and learned man, but thereis no suggestionin any of his work that he meritsthe label of genius. His discoverywas not a product of genius;it was a product of inspiration broughtabout by associationwith freshnew observationsof an important object,in this casea largeportion of the surfaceof the earth.Many, probably most,big discoveriesof scienceare made in analogousfashion. A truly exceptionalmind is not essential.Becoming associated early, or first, with importantnew observationsis commonlythe key to discovery.Those who seekto discovermay enhancetheir chancesfor discoveryby maneuvering into a positionwhich providessuch association. There is a secondlesson to be learned as well. Ortelius' magnificent idea, though publishedwhere it was availableto all, apparentlyhad no effecton muchof anything,certainly not the courseof science.The ideawas either quickly forgotten, or unnoticed,or ignored. It lay hidden until Romm'spaper was publishedhundreds of years later and well after the idea had been had independentlyby othersand eventuallyexploited. ContinentalDrift andSea Floor Spreading 33 lesson?To have an important impact on science,an idea, no matter how good or how correct,must also be timely and must be communicated appropriately and forcefully,not merely buried on a library shelf somewhere. Even though scientistsregularly strive, as they should,to be first with a great idea, it is neverthelessoften the casethat what comesto be widely-knownas a greatand originalidea by someonewas had earlierand in lesstimely fashionby others,and then overlooked.A greatidea mustbe communicatedto, and driven to the attentionof, many othersif it is to have an impact.This observationprovides some justification, and might provoke somesympathy, for thosescientists who at timesmight appearoverzealous aboutmaking their own controversialideas widely-known. Ortelius may well have been the first to proposethe conceptof continentaldrift, but his achievementis otherwisenot unique.Later, others, on seeinga reasonablyaccurate map of the Atlantic and the continentsbor- deringit, conceivedand recordedthe sameidea or relatedideas (see Romm for further discussion).Still others surely had similar thoughts without recordingthem. None prevailed, however, or had any lasting effect until 1912 when Alfred Wegener, the great German meteorologist and astronomer,not only had and publishedthe sameidea onceagain, but also assembleda wide variety of data bearingon this spectacularand basicconcept that would be called "continentaldrift." Wegener also, and wisely, took the time and trouble to communicatehis resultswidely and enthusi- astically.As a consequenceof the combinationof his idea and his efforts,he would eventually becomeknown as the "father of continentaldrift," but not until after yearsof controversy. Earlier, and certainly during the time of Ortelius, geologicalobservations relevant to the conceptof continentaldrift were so scarcethat they lent neither supportnor contradictionto the idea. But by Wegener'stime, many of the geologicalobservations that are now widely recognizedas supportfor the conceptof continentaldrift were known, and Wegenerwas ableto make a casewhich includedmany valid and importantpoints based on thoseobservations. For an idea to becomea part of science,it is of course essentialthat appropriateand sufficientobservations for testingand lend- ing supportto the idea be available.Wegener's initiative was timely in this regard;Ortelius' premature. Wegenertook painsto make his caseknown throughlectures and pub- licationsand his efforts did not fall upon deaf ears, but neither did they convinceall who heardhim. His ideaswere acceptedby some,few in number and definitelythe minority,and rejectedor relegatedto a statusof little importanceby others,definitely the majority and including many prominent leadersof earth science.From the time of first publicationin 1912 34 ContinentalDrift andSea Floor Spreading the plate tectonicsferment of the 1960s,Wegener's concept of continental drift was a subjectof controversyand debate,but duringthat interval,and particularlyamong North Americanearth scientists, it usuallyseemed to be losingstature because of certainflaws, and it had had little enoughcre- dencewhen first presented.Its basicsimplicity and rationality,however, gave it stayingpower so that it was never lost.An anecdotebased on my personalexperience might illustratethe stateof affairsat that time. In 1947, when I was a graduate student in physics at Columbia Universityand in need of financialsupport, I took a job as researchassis- tant with Maurice Ewing, a professorin Columbia's Department of Geology.Ewing was a geophysicistand the projectI worked on concerned soundwaves in air, so,strictly speaking, I had little contactwith, and little needfor any knowledgeof, geologyin thatjob. That was fortunate,indeed, for at that time I had never taken a coursein geologyand knew almost nothing of the subject.However, my colleaguesin Ewing's group were mostlyworking on otherprojects that were of a geologicalnature, so I soon developedan interestin geologyand a desireto learn somethingabout it. As a first step I bought and read a book that I had chancedupon in the bookstore.It was written by GeorgeGamow and entitled"The Biography of the Earth."Gamow, a physicist,tried in that book to tell the historyof earth in a self-consistentmanner and in a stylethat would appealto non- scientists.I think it fair to say that it was his attempt at a plausibleand interestingtale of how earthhad evolvedrather than a strictreview of the state of earth science at the time. Among other things,Gamow's book incorporatedWegener's story of the drifting continentsand it left a neophytelike me with the impression that continentaldrift was an establishedpart of earth science.Later, as I learnedthat my colleagueswere workingon projectsdesigned to determine the answersto such questionsas whether the oceanbasins were young or old, or why trenchesand mountainbelts existed,I was secretly perplexed. It seemed that they were unaware that continental drift explainedall thesethings. In fact,they never even mentionedthe concept. However,as the greenestof new recruitsto the subject,I was understand- ably,and I hopedtactfully, reluctant to draw this point to their attention.So I heldmy tongueand waited for my understandingof geologyto grow.Still later,as I beganto takeformal courses in geology,I learnedthat thehypoth- esisof continentaldrift was indeedwell known to geologists,but in those same coursesI was also heavily indoctrinatedin prevailing arguments againstthe concept.Thus as a consequenceand at the time,I relegatedcon- tinental drift and Gamow's book to positionsof lessprominence in my thinking.Like mostNorth Americanearth scientists of that era,I joined ContinentalDrift andSea Floor Spreading 35

school of data collectors who felt that far better observations of earth were in order before anyonejumped onto the bandwagon of some particular hypothesis. Collectingnew observationsturned out to be the right thing to do, of course,as it almostinvariably does when controversyor uncertaintyarises in science.At the time of Wegener'sproposal of the conceptof continental drift, and even more so in the decades that followed, there was considerable observationalinformation available on the geologyof the continents. But the geologyof the seafloor, which coversmore than two-thirdsof the earth's surface,was very poorly known, almost unknown at that time. Consequently,because of scarcityof observationsfor almost all marine areas,some felt free to claim that the continentsdrifted through the sea floor;others felt equallyfree to claim that the seafloor was merely sunken continentalcrust and that it made no senseto claim that one pieceof continental crustcould drift throughmore of the same.Controversy there was, and it would be resolvedby moreand better data. Then World War II came along,stirred up the world and in the processindirectly stimulated efforts to observethe seafloor thoroughlyand comprehensively.Eventually those effortsproduced the new informationon bottom topography,on seafloor structure and nature, and, especially,on magnetic anomaliesover the oceansthat would be criticalto the developmentof the conceptof seafloor spreading,a key stepin the discoveryof plate tectonics. In the postWW II yearsas observation of the seafloor began to flourish, certaintypes of observationson land were importantas well. The first sign that I sensedof revival and strengtheningof the hypothesisof continental drift camein the 1950'sas paleomagneticstudies of continentalrocks by British(notably Irving, Runcornand Creer) and variousAmerican paleo- magneticists(among them Graham, Cox, Doell and Dalrymple)showed that directionsof remnantmagnetism varied with ageof rocksat a singlelocality and alsovaried among rocks at multiplelocations in justsuch a way ascould be explainedby drift of the continentswithin a magneticfield that held its positionrelative to the rotationalpole. These early data were at the very least suggestiveand they drew somerenewed attention to Wegener'sconcept of continentaldrift, but as I saw it at the time, mostscientists were skeptical.It was relativelyeasy for them to be that way becausethe observationswere sketchy,sometimes challenged, and sometimesinterpreted in other ways. Most scientistsof the time were not sufficientlystimulated to reorienttheir work toward studyof continentaldrift as a consequenceof the new devel- opments.Another personal anecdote may be illustrativeof thosetimes. In 1954,just after completionof graduatestudy, I made a trip to South Africa for the purposeof installinga specialtype of seismograph. 36 ContinentalDrift and SeaFloor Spreading that visit, a local, distinguished,and considerablyolder, professor of geology,Lester King, kindly invited me to his home for breakfastone day and then graciouslytook me on a tour of the local geology.On that day I was much impressedby King'sknowledge of not only localgeology but alsoof the geologyof all of Africa and even SouthAmerica. And I shouldhave been impressedfor I was to learn later that King enjoyedan international reputationin geologicalcircles; senior professors at my home institution, ColumbiaUniversity, knew him well. Later,and on a day near the end of my visit to Africa, King took me to his researchlaboratory where he introduced me to his co-worker who, under his direction,was busy slidingplastic scale models of the continents around the surfaceof a globe and trying to fit them togetherin various ways. King was doing researchon continentaldrift, a matter of someaston- ishmentto me as I had alreadylabeled him an astuteand learnedgeologist and as I had been taught in the graduate school of a distinguished Americanuniversity that continentaldrift was impossible.Of course,it was an error of judgmenton my part that I now regret,and that I would apologize for if King were still alive, but at the time my indoctrinationin anti- drift was so strongthat my mistakenresponse was to forgiveKing tacitly for an idiosyncrasyrather than learn from him! That same error would be repeatedby almost every one of my colleaguesat Columbia,for, on invitation,King visitedus a few yearslater and carriedon a formal publicdebate over continentaldrift with WalterBucher, a seniordistinguished professor at Columbia.King clearlywon the debate, and in the process,although he had not convincedthem, he stimulated membersof the audience,which consistedmostly of studentsand faculty, to considerand to arguethe matter of continentaldrift in the monthsthat followed.However, I think it is fair to saythat all, or almostall, of us eventually returned to our anti-drift leaningsas thosediscussions waned in a half year or so.We would all change,of course,a few yearslater as sea-floor spreadingand plate tectonicsarrived. During the period of the debateand throughoutmost of the 1950sas bestI canrecall, when the subjectof drift was raisedor debated,earthquakes werenot directlya part of the argument,pro or con.Seismic studies on land, and particularlyat sea,based on artificialsources were growingin quantity and quality during this period, however,and their importancein the drift debategrew, but mostly,as indicated previously, the crudeearly seismic evidencewas initially interpretedas supportfor the anti-drifters. In the late 1950s,earthquakes became a part of the chain of eventsthat would lead to the conceptof seafloor spreadingand henceultimately to plate tectonics.At that time marine geologistsBruce Heezen and his ContinentalDrift andSea Floor Spreading 37 worker,Marie Tharp, were engagedin a lengthyproject to preparea spec- tacularnew physiographicmap of the oceanfloors (Heezen et al, 1959).It was probablythe first suchattempt anywhere to make sucha map for the entire sea floor, and the projecthad only becomefeasible as the Lamont archiveof oceansoundings grew. Furthermore, the physiographicmap that showedbottom featuresonly schematicallyand pictorially avoided the restrictionthrough military classificationof useof mostsounding data for the deep seathat prevailedthen. At that stage,the data were limited and spotty,of course,and oftena certaindegree of interpretationwas required. Tharp, who did most of the actualplotting and drawing (and who hence, and perhapslike Ortelius,was often the first to seenew portrayalsof the collectedobservational data) noticed that segmentsof mid-oceanridges commonlyexhibited a centralvalley, or rift. Furthermore,she noted that the earthquakesthat occurredbeneath the seafloor oftenwere locateddirectly beneathsuch a rift. That, it would turn out, was a very astuteand important observation.It becamethe basisfor speculationby Ewing and Heezen and Tharpwho promptlyused the known locationof seismicbelts beneath the seafloor to postulatethat the scatteredsegments of mid-oceanridges observedby that time were actuallypart of a single,huge, globe-encircling ridge. It was almosttwice the earth'scircumference in length and it was soonlikened, in overallappearance, to the stitchesof a baseball.This result was an importantstep, for the existenceof sucha featureof globalscale, tectonicallyactive as implied by the seismicactivity, meant that the process that was deformingthe earth, or at leastthe seafloor, was also of global scale.Earth's tectonicfeatures were not just a consequenceof randomly located,local events;they were related through some mechanismthat encompassedmuch of the earth. The globe-encirclingrift systempostulated by Ewing, Heezen and Tharp receivedconsiderable attention from many earth scientists,among them Harry Hess,a Princetonprofessor with a penchantfor off-beatand innovativehypotheses designed to explaingeologic observations ignored or overlookedby others.Hess had a long-standinginterest in marinegeology.During WWII, he had surveyedparts of the Pacificand found and namedguyots, the flat toppedseamounts now known to be therein abun- dance.He had helped to measuregravity at sea in the Caribbeanand he had proposedwhat becamea very well-known,although now considered incorrect,hypothesis to explainthe 5 km thick oceancrust that was being found by marine geophysicistsalmost everywhere beneath the deep sea. Hessproposed, plausibly but incorrectly,that it was a surficiallayer of ser- pentinizedperidotite derived from the peridotiticmantle below with the addition of 38 ContinentalDrift andSea Floor Spreading

Then in the early 1960s,Hess (1960) hit the jackpot.He proposedwhat wouldbecome known as the seafloor spreading hypothesis, the ideathat the seafloor was spreadingapart at the oceanridges, specifically at the riftswith- in thoseridges, and that materialwas welling up from the mantlebelow to createnew seafloor at the cracksthat markedthe riftsand hencethe spreading centers.He proposedthat the spreadingcontinued until entire ocean basinswere formed.This conceptwould becomea majorbuilding block as the platetectonics model was developed.Although, as is oftenthe casewhen a greatadvance in scienceis made,others, particularly Arthur Holmes(1944), had had similar or relatedthoughts earlier, those ideas were mostlyover- lookedor treatedas curiositiesand speculationwhen they appeared.Hess' proposalwas attractiveand it caughton. It had substance,and it was timely becausenew observationsof the seafloor that would supportthe ideawere justappearing and becauseother scientists were independentlybeginning to thinkalong related lines. Dietz (1961),for example,coined the term "sea-floor spreading"and publishedan influentialand widely-readpaper on thetopic. Dietzbased the paper partly on work by Drakeet al (1958)which pointed out the parallelbetween sedimentary troughs of the moderncontinental margin and the pairedgeosynclines of the Appalachians. The criticalevidence in supportof Hess'model came from the map patternsof magneticanomalies at sea,and from measurementson layeredvol- canicand sedimentaryrocks that revealedthat the earth'smagnetic field has reverseditself in the past and just when thosereversals occurred. Vine and Matthews(1963), in a simpleyet eleganthypothesis, used Hess' concept of spreadingto showhow thesetwo kinds of magneticinformation, the field reversalsand the spatial anomaly patterns,could be related. The Vine- Matthewshypothesis became a great successand the sea-floorspreading hypothesisreceived a substantialboost. It beganto get increasingattention from geoscientists,especially and primarily thoseworking on geomagnetism.Initially, and exceptfor the seismicitypattern that suggestedglobal continuity for the mid-oceanridge system,there was no directcontribution to the sea-floorspreading hypothesis from earthquake seismology, but thatwas soonto changeas a consequenceof a suggestionmade by the prominent Canadiangeophysicist, Tuzo Wilson. Wilsonwas a big thinkerwho seemedalways eager and compelledto seekgrander meaning in the observationsof the earththan did many of his fellow scientists.He had longbeen fascinated by problemsof largescale tectonicsand in the 1950shad proposedsome provocative and well-known hypothesesrelating to orogenicbelts and islandarcs, but thosehypotheses were eventuallyrejected, at leastin the form proposedat the time.As soonas it becameknown, Wilson took up the sea-floorspreading idea, focusing ContinentalDrift andSea Floor Spreading 39 tially on one aspectof it in particular.He championedthe conceptof the transformfault, an explanationfor the peculiarrectilinear offsets commonly found along the mid-oceanridges. Wilson showedhow such offsetsand relatedobservations might be explainedas a consequenceof spreadingat the riffs.If the transformfault hypothesiswas correct,the rectilinearpattern provided strongsupport for the seafloor spreading hypothesis. Then, in a paper on the subject(Wilson, 1965), he suggesteda crucialseismological test of the transformfault hypothesis. At that point in time, althoughearthquakes had beenused to definethe world-encirclingmid-ocean rift systemand otherseismic data had provided relevantinformation on marine and continentalgeology, earthquake seismologistshad not beenmuch involved with, or concernedwith, theseevents thatwould turn out to be forerunnersof the platetectonics revolution. In fact, mostearth scientistsin all specialties,except perhaps geomagnetism, were paying at bestonly casualattention to what was goingon. As an earthquakeseismologist at LamontGeological Observatory, where therewere many colleaguesin marinegeoscience, I was aware of the paleo- magneticresults at sea,the geomagneticfield reversals,Hess' proposalof sea-floorspreading, the Vine-Matthewshypothesis, and Wilson'sideas all shortlyafter they appeared,and I was casuallyinterested, but like almost every other earth scientistI was not sufficientlymoved or excitedby the .L•w• •u r,Lar I reorientedmy daily effortsto move into this subject. Wilson'sperceptive suggestion for a seismologicaltest of the transform fault hypotheseswas madewidely availableto seismologiststhrough publication,but it generallyfell upon deaf ears and was passedover. However, one young Lamont seismologist,Lyrm Sykes,recognized the opportunity and capitalizedupon it. In sodoing he not only providedcritical support for the transformfault and seafloor spreadinghypotheses, but he alsostimulat- ed his Lamontcolleagues in seismology,including me. Consequently,and alsobecause of its facilitiesand archives,the Lamont programin earthquakeseismology became a major factoras the sea floor spreadingstory evolved into the plate tectonicsrevolution. Three later chap- tersconcern primarily papers by seismologistsLyrm Sykes, Bryan Isacks and JackOliver and area centralpart of thisbook. The studieswere all carriedout at what was then the Lamont GeologicalObservatory, now the Lamont- Doherty Earth Observatory,of ColumbiaUniversity. As the environmentat Lamont,and particularlyinteraction with other programsthere, especially that in geomagnetism,was criticalto the natureand the successof thoseseis- mologicalstudies, the next chaptergives a brief descriptionand historyof thatinstitution, selectively emphasizing those aspects of thathistory and that settingthat are especiallyrelevant to our 5 The Origin and Early Days of the LamontGeological Observatory and Its Programin EarthquakeSeismology

DocEwing, the head of LGO Whenasked by a visitingCEO Wherehe stored his ships Betweenannual trips Said"I keepthem working at sea-oh!"

heseeds of Lamontwere sown near the end of WorldWar II when MauriceEwing joined the faculty of theDepartment of Geologyat Columbia University.The Department faculty had properly recognized thegrowing importance of geophysicsin earthscience then, and chose Ewing to provide strengthand leadershipin that subject.That decision turnedout to be an exceptionallywise one,but at the time probablyno one- faculty member,administrator, or even Ewing himself-foresawthe magnitude and scope of what it would lead to. When Ewing retired from Columbiaabout a quarterof a centurylater, he left behinda largeand pres- tigious institutionwith an internationalreputation and some400 employ- ees,almost half scientistsor studentsof science,facilities that includeddeep searesearch vessels and laboratoriesoutfitted to studya wide variety of topics in earth science,and scientificprojects operating at locationsscattered throughoutthe world. Ewing was an extraordinaryman of exceptionalvision and bold and daringdreams. He was a superbscientist and a hard-workingtireless leader who led by "perspirationas well as inspiration."He was a forceto be reck- oned with in earth science,and he was widely respected,and sometimes envied,as such.When he cameto Columbia,at aboutthe age of 40, he had alreadyrecognized, correctly, that the sea floor was the greatfrontier of earth scienceof that era.Aided by increasesin both the statureand the fundingof sciencefollowing WW II, Ewing began to build, at the main campusof Columbiaon MorningsideHeights in New York City, a researchorganiza- tion with lofty goals,a distinctivestyle, a spirit of innovation,and a level of inner confidenceborn of soundstrategic thinking and continualsuccess. 42 Originof theLamont Observatory

A criticalpart of Ewing'sphilosophy was the recognitionof the great importanceof new observationsof the unknown.He sawfresh observations as the key to discovery.He soughtto observethe earth everywhereand in every way that he could, and he gave specialattention to the relatively unknownsea floor. If he tooka shipto seafor the principalpurpose of doing seismicwork, the ship was alsooutfitted with other kinds of equipmentas well, and the scientificstaff was calledupon to coreand dredgeand photo- graphthe oceanbottom, sample and measureparameters of the water column, take biologicalsamples, record the geomagneticfield, and measure precisewater depthseverywhere along the ship'strack. To him, time spent at seanot making all possiblekinds of observationswas time and money wasted. Although nominally a marine scientist,Ewing never felt confinedto studyonly the sea.When he found that he couldapply someof his earlier experiencewith acousticwave propagationin ice and in shallowwater to makeearthquake observations provide information on the oceanicand con- tinentalcrusts, he begana majorprogram to operateseismographs on land, and later on the seafloor, in orderto acquireappropriate data. He was also the first,or amongthe first,to advocatethe landingof a seismographon the moon. Ewing's inspirationalstyle and leadership,his lofty goals,his adven- turesome,challenging, and scientificallyfruitful activities,and his ability to tap fundingsources to supportthem, soon brought young scientists and studentsof scienceto his fold. Most o• thosewho ioined him were fortunate indeed,for they were trainedand inspiredto careersthat would makethem leadersof sciencein •heir own right. Among that early group for various periodsof time, and in moreor lessthe orderin whichthey joined, were Joe Worzel,Frank Press,Nelson Steenland, Ivan Tolstoy,Milton Dobrin, Paul Wuenschel, Gordon Hamilton, Dick Edwards, Sam Katz, Bruce Heezen, lack Oliver, Chuck Drake, Bill Donn, JohnEwing, lack Heacock,Bernie Luskin and Chuck Officer. During the late 1940s,Ewing's fledglinggroup at Columbiaquickly grew •o strain the facilitiesand quartersassigned to him on the main Columbiacampus. Room for expansionand new kinds of equipmentwas needed,and •he seismographsrequired a siteless noisy than the Manhattan location.Columbia recognized •he value and the potentialof the Ewing-led activity,and the competitionfor his servicesfrom otheruniversities, and met his need by providing the use of a 100 acre estate,the former home of ThomasLamont, some fifteen miles up the Hudson river from the main campus.In 1949,Ewing's group, which then numbered about a dozengrad- uates•udents and severaltechnicians, moved from the Morningside Originof the Lamont Observatory 43 campusin New York City to the village of Palisadesin RocklandCounty, and the Lament Geological Observatory was formed with Ewing as Director. Also moving to and joining the Observatoryat about that time was a smallgroup in geochemistryled by ProfessorLawrence Kulp. It too would grow to produceimportant earth science and many distinguishedearth scientists,and it became an important componentof early, and modern, Lament but its activitiesand its achievementsare beyondthe scopeof this book on seismelegy. Ewing encouraged,and was supportiveof, expansionof Lament activ- itiesinto almostany area of earth science,so long as the effort promisedto be of high quality.Programs blossomed at Lament in areassuch as micropa- leontologyand oceanographyas well as variousbranches of marine geology,geophysics and geochemistry.Ewing's emphasis on observationresulted in largeand unusualarchives of suchdiverse and information-ladenthings as oceansediment cores, seismic and acousticsoundings, magnetograms and bottomphotographs. Seismograms from the Palisadesseismograph station were archived,along with data from outlyingstations and data marked for discardby nen-Lamentstations and donatedto Lament instead.A complete collectionof the WWSSN microfilmedseismic data referredto earlier was obtained.This huge data collectionwould turn out to be a critical resourcethat gaveLament scientists an advantageousposition when the sea floor spreadinghypothesis arose and as the plate tectonicsconcept was beingdeveloped. The foregoingemphasizes Ewing's insatiable drive to collectnew kinds of observationsabout little known parts of the earth, but that emphasis shouldnot be interpretedto meanthat he or his organizationfailed to stress the analysisand the interpretationof thosedata. Ewing, more than any other scientistI ever encountered,foresaw that somehowa great revolution in earth sciencewas on the horizon,and he strovemightily to causethat revolution. It is sadlyironic that, for the reasonscited earlier, he was not the one to have the key ideasthat led to seafloor spreadingand thenplate tectonics, and that when thoseideas did appearthe early and incompleteobservations of the time happenedto suggestthat thoseideas were wrong. Nevertheless,the contributionsto earth scienceby Ewing were monu- mental and widely recognizedas suchby other scientists.In his career,he mademajor advances in, and progressedthrough, a seriesof topicsranging from globalacoustic wave transmissionin the deep seasound channel, to turbidity currents,to fundamentaldifferences between the continentaland oceaniccrusts, to the globe-encirclingmid-ocean ridge, to earthquakesur- face waves in continents, the ocean basins and the earth's mantle, and to 44 Originof theLamont Observatory effectof atmosphericdisturbances on water waves,to namejust some of his contributions.He was truly a giant of his time. Ewing'searly oppositionto seafloor spreadingand then nascentplate tectonicshas somehowled someto infer that he tried to suppresssupport for theseconcepts by controllingthe work or thepublications of scientistsat Lament during the period when thoseideas were being testedand developed.Of course,I cannotclaim to havemonitored all of Ewing'sactions, but I would like to go on recordhere as noting that I find thesestatements or inferencessurprising and contraryto what I observed.During that critical period,I was a seniorstaff member at Lamentas well as a facultycolleague of Ewing. As head of the earthquakeseismelegy program and one of the foundinggroup of Lament,I was clearlya part of the innercircle of leaders. Furthermore,as laterchapters of thisbook document, my work, and that of my colleaguesand students,was certainlysome of the mostpro-plate tectonics science to come from Lament. At no time, however, did that work or the correspondingpublications encounter any interference,or barriers,or even negative comment from the Director of the Lament Geological Observatory.Ewing was a strongand sometimespartisan leader who did not hesitateto demonstratehis opinionsand the depthof his conviction,but he was too muchthe solidscientist driven by the basictruth of observation in scienceto opposethose truths onceattention was drawn to them and the case made. In fact,during my 24 yearsof associationwith Ewing,first as a student and then as a colleagueunder his direction,I oftenhad occasionto consult with him and to seekhis approval.Almost invariably his reactionto a sug- gestionor a proposalwas positive.Nearly alwayshe not only approvedbut he alsoencouraged an expansionof what I had proposed.I canrecall only a singletime in thoseyears when I encounteredopposition from him and that waswhen I clearlyoverstepped my authorityand tried mistakenlyto spend somemoney that had beenespecially granted to him for anotherpurpose! And even then he did not get nearly so upsetas I would have had I beenin his position.I have oftenthought that his encouragementand his habit of a positive-but-do-moreresponse to an underling'sinitiative was a key to Ewing's successas a leader and administrator. Furthermore,for thosewho would picture Ewing as long a staunch opponentof plate tectonics,I draw attentionhere to a rarely cited paper, publishedin 1967by Johnand Maurice Ewing, which suggesteda slight modificationof the seafloor spreadinghypothesis shortly after it appeared. The modificationhad to do with a possiblechange in the rate of plate motion about ten million yearsago and clearlyimplied an opennessand willingnessto acceptand testthe hypothesisat that very early date.I Originof theLamont Observatory 45 find myselfin completeopposition to, and often somewhatastonished over, some statementsthat have been made and sometimespublished about Ewing's supposednegative reaction and obstructionto certain developmentsin the plate tectonicsrevolution. Of course,it might be that Ewing treatedsome of his scientistsin oneway, othersin another.However, I, and someothers, think it is morelikely that the conflictingviews of Ewing are a consequenceof differentinterpretations of his motivesas he called upon Lamont scientiststo justify thoroughly and in the best scientificfashion what they proposedto publish.Some saw this practiceas unjustifiedand prejudicialopposition to a pet theory,others saw it as merely tough,hard, but neverthelessreasonable and good,science. As in any organization,communication among the various groupsat Lamontwas, of course,for purely practicalreasons, never ideal or complete, and it becameless efficient as the Observatorygrew. In the earliestyears, the numbersof scientistsand studentswere small,and regardlessof interestor discipline,almost all worked in one building, the former mansionof the estatenow known as Lamont Hall. However by the early 1960s,as the fer- ment over sea floor spreadingand plate tectonicsbegan, numbers had grown and additionalbuildings had becomeavailable, so the organization had subdividedinto variousgroups, mostly along disciplinarylines, and they were spatiallyseparated. Thus the groupsin geomagnetism,paleomagnetism, marine seismology, and earthquakeseismology were no longerquartered in the samebuild- ing, and communicationamong the groups,though fosteredon a personal level by friendships,openness and commoninterest, was inhibitedby the less than ideal spatial proximity. Often it took some time for news and excitementin onegroup to spreadto others.Thus when the fervorover continentaldrift and seafloor spreadinghit Lamont,it beganwith thoseworking in someaspect of magnetism,and then spreadto seismologyand elsewhere. More precisely,it was the high level of enthusiasmfor thoseconcepts that spreadin that manner.The ideas, and the principlesof the concepts were known aboutearlier by many at Lamont.We had all learnedformally aboutWegener's theory of continentaldrift, and somehad evenheard Hess proposethe conceptof sea floor spreadingwhen he first presentedit. However,we had somehow,I think, mentally assignedthose concepts to a categorythat we might have labeled"interesting but supportingdata less than compelling."Also in that categoryin our mindswere otherlarge scale tectonichypotheses we had learnedabout from text bookslike Umbgrove's "Pulseof The Earth."Lake's idea, for example,of shearplanes deep in the interior and intersectingthe surfaceat the great arcs was one 46 Originof theLamont Observatory

VeningMeinsz's concept of plasticflow in the mantleto producethe trenches and their associatedgravity anomalies was another.Generally we found suchspeculative hypotheses to be provocativeand interesting,but they lay fallow somehowfor they did not seemimmediately relevant to our dataand henceour work. The luke-warm nature of our interestin global tectonic hypotheseswas soonto change. Probablythe first pro-drift scientistat Lamontwas Neil Opdyke who worked in paleomagnetism.Unlike many of us who had doneour graduate work at Columbia and then moved onto the scientificstaff, Opdyke had arrived at Lamontby a circuitousroute. Ted Irving, a distinguishedpaleo- magneticistand an early advocateof drift, and I becameacquainted on a bus during a geologicalfield trip in California. On learning from me that Lamontwas seekinga bright youngpaleomagneticist, Irving immediately recommendedOpdyke, and I passed that information to Ewing who promptly hired him. At first on learningof him, I didn't know whether Opdyke was from Africa where he was working,or from Australiawhere Irving wasworking, or from Englandwhere Opdyke had receivedhis Ph.D. under Keith Runcorn,another leading paleomagneticistand drifter, or somewhereelse. Consequently,I had some concernover how Opdyke would fit into the ColumbiaUniversity-New York City area environment. That concernvanished when it turned out to my surprisethat Opdykehad attendedColumbia as an undergraduateand, in fact, was captainof the footballteam there! The latterpoint was particularlyto my chagrinas I had playedon that samefootball team just a few yearsbefore Opdyke had done so.At Lamont,in addition to his solid observationalstudies on paleomag- netic stratigraphy,Opdyke was an early and influentialvoice in favor of drift, not a popular view at the time. In additionto the groupin paleomagnetism,Lamont had otherswork- ing in geomagnetism.They observedthe earth'sfield at sea along ship's tracks, and analyzedthose data. Jim Heirtzler was the leaderof the group that also included Walter Pitman, Xavier LePichon, Ellen Herron, Maurice Davidsonand John Foster.Initially, and for years,the data on magnetic anomaliesat seawere confounding,but a greatmoment of enlightenment arrived when it was discoveredthat one track showedthat the magnetic anomalieswere almostperfectly symmetric about a ridge. It was strongevidence in support of Hess' and Vine and Matthews' ideas of symmetric spreadingat a ridge, and the Lamontgeomagnetic group caughtfire and hurried to interpretall of its uniqueset of marinemagnetic data in that light and to advancethe subjectof marine tectonicsin the process.They were spectacularlysuccessful, as hasbeen documentedelsewhere (Glen, 1982). Wordof the remarkabledevelopments in geomagnetismsoon spread Originof theLamont Observatory 47 the earthquakeseismology group. John Foster, then a graduatestudent, vis- ited me in my officeone day to encourageattention from seismologists,and Heirtzler and his cohortsgave a privatepresentation of their resultsto several seismologists,including me, on anotheroccasion. As we shall see,that presentationwas a key factoras Lynn Sykesmade the decisionto pursue TuzoWilson's suggestion, or perhapsit was Wilson'schallenge, on the testing of the transformfault hypothesisbased on the data of earthquakeseismology. With that brief and cursoryintroduction to what was going on elsewhere at Lamont as background,we turn our attentionnow to a more detaileddiscussion of the earthquakeseismology program there, where the studiesthat are a principalfocus of this book were to be carriedout. One of the out-buildingsof the Lamont estatewhen it was given to ColumbiaUniversity was a spaciousroot-cellar, well buriedbeneath the soil for temperaturestability. Shortly after the LamontObservatory was initiated, the floorof the root-cellarwas excavatedto bedrockand a largeconcrete pier pouredin contactwith, and on top of, the Palisadesdiabase sill, a thick igneousrock formation familiar to all inhabitantsof the New York city area becauseit forms the spectacularcliffs along the west shoreof the lower Hudson River. The root-cellar thus was converted into the Lamont seismograph vault, and in a few yearsit becamethe site of one of the finestseis- mographstations in the world. The Palisadesstation, as it was formallydes- ignated,had a variety of instrumentsthat collectivelymeasured three componentsof groundmotion in severalfrequency ranges, and it was especially notedfor the Press-Ewinginstruments operating in the low frequency,or long period, segmentof the seismicspectrum. In thoseearly days,Ewing was activein, amongother things,research on earthquakesand, especially,earthquake-generated seismic surface waves.He providedinspiration, guidance, insight and, asusual, contagious enthusiasm.His prime partnerin this endeavorwas FrankPress who func- tionedas head of the Lamontearthquake seismology program, first while a graduatestudent and then as a youngprofessor. Ewing, the intuitiveveter- an scientistwith a capacityand a compulsionto wring the mostfrom a given setof observationaldata, and Press,trained initially in mathematicalphysics but rapidly learninggeophysics and geology,formed a powerfulteam that quickly made its mark in the world of seismology.Their specialty,which wasbuilt initially on Ewing'sexperience with dispersiveelastic waves in ice and in shallowwater, and on Press'facility with wave guidetheory, was the dispersionof seismicsurface waves. Such waves are guidedby the earth's surfaceand the layersof generallylow velocityassociated with it, suchas the oceanicwater layer and the continentalcrust. Whereas others 48 Originof theLamont Observatory the surfacewave "coda"had focusedon the arrivaltime of the coda'sbegin- ning, Pressand Ewing soughtto understandthe entireoscillatory train as a consequenceof dispersion,the dependenceof phase and group velocity upon period or wavelength.They were remarkably successfuland soon becameworld leadersin this subject.They set the tone of Lamontresearch in earthquakeseismology for muchof the nextdecade or so,i.e. well into the 1950s. As the graduatestudent of earthquakeseismology next in line following Press,for a Ph.D. thesis,finished in 1953and publishedin 1955(Oliver et al, 1955),I followed the Ewing-Presslead and comparedseismic surface wavescrossing various parts of the Pacificwith the Ewing-Presstheoretical modelsof dispersionthere. That studywas solidenough, showing that large segmentsof formercontinental crust were not lying beneaththe deepsea, as somewho had proposedcontinental subsidence to accountfor the ocean basinswould have had it, but by then the primary resultswere not particularly surprisingor unexpected,so the paper had limited impact,and justly so. It might be instructiveto note, however,that one secondarypiece of information that was publishedbut that I, and everyone else, promptly ignored,could have been a clue to the spatialvariation in age of seafloor acrossthe Pacific,and hencea stimulusfor the idea of seafloor spreading well beforeHess' suggestionof that concept.However, the evidencewas eithertoo subtle,or we too naive, to interpretit properly.The data showed that the bedrocksurface was deeperin the westernPacific than in the eastern Pacific,a configurationnow attributedto the greaterage, and hence coolertemperatures and greaterdensities, of rocksmore distantfrom the spreadingcenter of the easternPacific. In otherwords, if we had beensuffi- ciently insightful,we could have looked at thoseearthquake seismograms and found a cluethat, with sufficientimagination, might haveled us to generationof the greatidea of seafloor spreading,a key componentof platetectonics.I makethis point at somelength to draw to the attentionof youngsci- entiststhat reliableobservations, no matterhow subtle,may carrya message of unanticipatedimportance and shouldnot be dismissedor ignoredwith- out thoroughconsideration. During the mid and late 1950s,the Lamont earthquakeprogram that was foundedon the Ewing-Pressera of earlieryears grew in sizeand scope. In 1955,Press left to take a positionat Cal Techwhere he becameDirector of the SeismologicalLaboratory. It was an early step in what becamean outstanding career.He eventually moved from there to MIT and then to Washingtonwhere he was ScienceAdvisor to PresidentCarter and then Presidentof the National Academyof Sciences.When Pressdeparted, I was namedto his vacatedfaculty position at Columbiaand alsobecame head Originof the Lamont Observatory 49 the earthquakeprogram at Lamont,both positionsthat I held throughmid- 1971,i.e. throughthe time of the plate tectonicsrevolution. The late 1950sand early 1960swere a time of greatchange in the nature of the earthquakeprogram at Lamont.An importantfactor in that change was a new building that was constructedon the Lamont groundsfor the earthquakeseismology program in about1960. In additionto more and better officesand laboratories,it providedspacious storage facilities for archiv- ing seismograms.The new building was somewhatdistant from Lamont Hall, the former home of the earthquakeprogram and also the site of the director'soffice. Consequently, although it was certainlyneither designed nor plannedfor by anyone,a net effectof the move of the earthquakegroup, plus concurrentincreases in other activitiesof the Director,was that Ewing relinquishedthe day-to-dayassociation with seismologythat he had had when the quarterswere adjoining,and new directionfor the programbegan to originatealmost entirely from othermembers of the earthquakegroup. Though still rooted to a degreein surfacewave studies,the program diversifiedmarkedly in the 50s and 60s as it moved into subjectssuch as model seismology,microseisms, nuclear test detection,microearthquakes, instrumentationdevelopment, and deploymentof seismographsin such unusualconfigurations and locationsas a network of 10 stationsspanning the globe,a deep mine in New Jersey,the deep sea floor of the eastern Pacific,and the moon. The diversificationwas greatly aided, of course,by the new government funding that became available to seismologists throughthe spaceprogram, NSF expansion,and the needfor basicresearch relatedto nucleartest detection. During that intervalof expansion,the char- acterof the earthquakeseismology program evolvedfrom a small,narrowly orientedoperation that focusedon a very limited number of topicsto a much larger one coveringmany topicsof greatvariety. Scientists, students and techniciansin greaternumbers and with greaterdiversity of talentsand interestsjoined the program. Lamont's reputation for innovative researchin seismologyattracted numerousgraduate students of high quality.They cameto learn, and they broughtwith them the vitality,energy and enthusiasmthat fueled the program further.Each stimulated and influencedthe othersin numerousposi- tive yet often intangibleways. Collectivelythey contributedto the setting that would make the Lamontcontributions to plate tectonicspossible. Many of thosewho carriedout and completedtheir graduatework at Lamontthen joined the staff as researchscientists in newly added projects.One often hearsit stateddogmatically that studentswho completetheir undergradu- ate or graduatework in scienceat one institutionshould go elsewherefor graduatestudy or for postdoctoralresearch. That principle was 50 Originod ethe Lamont Observatory regularlyat Lamont,particularly in the caseof graduatestudents, with little sign of negativeeffects and abtmdantindications of positiveones. In fact, the extraordinaryespirit de corpsof Lamont was at leastpartly a consequenceof that practice. Studentswho begantheir graduatestudy in one aspectof earthscience oftenshifted emphasis to anotheraspect, or redirectedtheir effortsthrough a spectrumof interestsduring their time at Lamont as studentand then researchscientist. George Sutton, for example,did seismicwork at seabased on explosivesources, studied earthquake focal mechanisms and earthquake wave propagation,studied local earthquakesin equatorialAfrica, and designedand participatedin installationof tinconventionalseismographs at Palisades,on the moon,and on the floorof the deepsea of the easternPacific Ocean. Paul Pomeroy installed seismographsaround the world, built an unusuallysuccessful instrument for detectingwaves of long period, and studiedwaves generatedby nuclearexplosions in variousenvironments. JimDorman, trained as a conventionalgeologist, learned geophysics, investigated aspectsof seismicwave propagationand, when digital computers first appeared,became Lamont's expert in computersystems and applied that new capability to various aspectsof seismology.Jim Brune studied wave propagation,microseisms, focal mechanisms, and microearthquakes, and developedan ingeniousmethod for visualizingand analyzingthe dis- persionphenomenon. These former students,and many others,eventually left Lamontto takefaculty positions at otheruniversities but theyleft behind a legacybased on their accomplishmentsand spirit. Also contributingheavily to the seismologicalprogram of that era were JackNafe, a Columbia-trainednuclear physicist who joinedthe geologyfac- ulty to pursuehis interestsin geophysicsand who broughtelegance, techniques,rigor, and depth of understandingof modem physicsto the program, and a number of visiting scientistsincluding Hans Berckhemer, StephanMueller, Inge Lehmann,Yasuo Sato, Bruce Bolt, George Thompson and others.John Kuo, initially a short-termvisitor to Lamont, stayed to build an outstandingcareer at Columbia. Orson Anderson,a solid state physicist,used ultrasonic waves to investigatein an innovativefashion fun- damentalproperties of certainmaterials of the earth.George Hade left a job in a localauto repair shopto becomea superlativeseismological engineer- technicianwho designed,tested and then installedseismographs at remote locationsthroughout the world, in the processhandling difficult technical and diplomaticproblems with equal aplomb and boostingthe morale of thosehe encounteredeverywhere along the way. Although the earthquakeseismology program was administratively distinct,and physicallyseparated from the quartersof, the marine Originof the Lamont Observatory 51 ogy program,there was neverthelessample interactionand communication betweenthe two groups.The marine group carried out seismicreflection and refractionstudies of the seafloor. It was led by JohnEwing, Bill Ludwig, and RobertHoutz during the 1960s,but other Lamont scientists,some of whom held major administrativeposts such as Joe Worzel, Lamont's associ- ate director, and Chuck Drake and Jack Nafe, each for a time chairman of Columbia'sgeology department, found time to participatein the seismic work at sea. The individuals named, plus many unnamed others,contributed not only expertiseand energydirectly or indirectlyto the earthquakeseismology program,but alsothey all addedintangibly to the spiritand camaraderie of the Lamontgroup. There was sharingof interest,enthusiasm and vitality. Opennessand congenialityabout science, and almosteverything else, was pervasive.There was someof the intra personalrivalry that characterizes scienceand on which sciencethrives, of course.However, any frictionwith- in the organizationthat resultedwas typicallyresolved on the basisof what was bestfor the science.Few choseto misseven one of the Lamontparties which were numerous,open to all and alwayswarm, cordial,and joyful. Lamontpersonnel worked hard and played hard. They took pride in doing the dirty and difficult kinds of jobs that involved ships and waves and trucksand sledsand dirt and ice and travel and midnight oil becausethe work was done for the higherpurpose that was science.It was in this happily conducivebut neverthelessdemanding kind of settingthat the Lamont seismologicalcontributions to the plate tectonicsrevolution arose. To organizethose contributions to the plate tectonicsrevolution the fol- lowingpages are dividedinto threechapters, each chapter focusing on a certain scientificpaper but alsoreporting in somedetail related papers and relevant otherevents of the time. The next chapteris centeredon Sykes'seismologicaltest (Sykes,1967) based on Wilson'schallenge in his paper on transformfaulting. Using data on seismicityand on earthquakefocal mechanisms,Sykes provided strongobservational support for Wilson'smodel. The paperis a clearexample of scienceby hypothesistesting and the chapter bearsa correspondingtitle. The followingchapter is centeredon a paper by Oliver and Isacks(1967) which reportsthe discoveryof the phenomenon of the down-goinglithospheric slab in a convergentarc structure,a phe- nomenonthat is the essenceof the subductionprocess. This caseis a clear exampleof discoveryby serendipityand the chapteris soentitled. The next chapterto follow is essentiallythe climaxof the book.It concernsthe paper by Isacks,Oliver and Sykes(1968) entitled "Seismology and the New Global Tectonics."This paperis a clearexample of scienceby synthesisand the title of the chapterreflects that view. It is a point worthy of somenote that 52 Originof theLamont Observatory three widely differentapproaches to scienceall producedresults that are mutually consistentand were important,even critical,to the development of the plate tectonicsrevolution. The messagehere for young scientistsis, of course,that no one styleof doingscience is obviouslysuperior or shouldbe exclusive,and furthermore that sciencewould be lesseffective if forcedinto any one suchmode. I hope this point is made sufficientlyclearly so that it will be noted by all peer reviewers!I shudder to think of how backward sciencemight be if all researchof the past had been confined,as somepeer reviewershave erro- neouslyrecommended, to only projectsfor which the hypothesisis "clearly and explicitly"statedor the problem"sharply 6

Scienceby HypothesisTesting

If it offsetsa ridge, it's a transform A geodynamicalkind of a danceform Twixttwo plates that are mobile In a patternthat's global It can'tbe a tectonicchance form.

ostof the students who began graduate study in seismology at Lamontduring the 1950s and early 1960s held bachelors' degrees in eitherphysics or geology,or perhapsa branchof engineering. Few universitiesat that time gavedegrees in geophysicsor otherwise attemptedto blendgeology and physics at theundergraduate level. Thus mostnew Lamontstudents of seismologyspent a part of theirtime reme- dyingacademic deficiencies in one or the other of these topics. In addition, mostnew students were inexperienced in research and had yet to make the transitionfrom the learning mode of theclassroom to theapprenticeship styleof graduatestudy that called for full-fledgedparticipation in all aspectsof scientificresearch. WhenLynn Sykes arrived at Lamont as a newgraduate student in seismologyin the fall of 1959 he had no such deficiencies. Hehad B.S. and M.S. degreesfrom MIT withemphasis on geophysics, sohe was already indoc- trinatedinto the worlds of both geology and physics, and he also had sup- plementarytraining in mathematicsand electricalengineering. Furthermorehe had alreadyspent a summerof researchat WoodsHole OceanographicInstitution. He arrived,in everyrespect, ready to gointo meaningfulresearch. In classes, hequickly demonstrated thestrength ofhis trainingand his outstanding ability and competence. He also demonstrat- edstrong personal attributes aswell. He had those intangible qualities that quicklymade him outstanding asa researcher.Inparticular, hehad a decid- edknack for finding and exploiting opportunities in science overlooked by others.Such a knack,cultivated or inherited,is a criticalasset of outstanding scientists. 54 Scienceby HypothesisTesting

Lynnhad chosenLamont for graduatestudy after careful consideration of all of his opportunities.As an applicantwith fine credentials,he was acceptedfor admissionby graduateschools nationwide. He read journals and attendedscientific meetings and found that papersby Lamontscien- tists were prominent and often predominantin the field of his interest. Thenhe visitedLamont and, I learnedmuch later, was favorablyimpressed partly becauseI spenttwo hoursof a beautifulSaturday in the springtalk- ing with him aboutscience and graduatestudy at Lamont.I tell this story not for self-adulationbut to make the point by exampleand in the context of thisbook that successin scienceby an organizationmay dependas much on matterssuch as attentionto recruitingas on more obviousfactors. Lynn beganhis researchat Lamont,competently but ratherstraightfor- wardly,by followingthe style of seismologyset earlier by Ewingand Press, as he carriedout somevery solid studieson seismicsurface waves propa- gatingover oceanicpaths. During the courseof this work, he noted,to no one's great surprise,that the locationsof the earthquakesin the South Pacificthat he was usingas sourcesof surfacewaves were oftengreatly in error.Then, in trying to overcomethis obstacleto bettersurface wave studies,he found that thoselocation errors could be reducedby as much as an order of magnitudeby usingthe reliablenew data that had becomeavail- ablefrom the WWSSN and othersources in conjunctionwith newly developed computationaltechniques for hypocentrallocation. In so doing,he not only resolveda problemwith the studythat he wasmaking, but he also recognizedthat he had found a techniquethat he couldprofitably extend to other areas. To take advantage of the new opportunity, he began to relocate hypocentersin variousparts of the world, particularlyalong the segments of the mid-oceanridge systemand then later alongisland arcs. The reloca- tions employeda new computerprogram developedby BruceBolt who had foreseenthe importanceof newly developeddigital computersin that aspectof seismology.It was yet anotherexample of an importantadvance in technologyhaving a near immediateimpact on a branchof science. Sykessoon found that, with the increasedprecision, he couldfind in variousparts of the world new examplesof the peculiarrectilinear patterns of ridgesand fracturezones that were alreadyknown in selectedlocations elsewhere(but that had not yet been associatedwith the phenomenonof transformfaulting). He wrote a seriesof papersthat helpeddefine such fea- turesin variousremote regions of the earth rangingfrom the SouthPacific to the Arctic,and in so doingadded to knowledgeof the overallconfigura- tion of the world rift system.Lynn (Sykes,1966) also made a studyof seismicity associatedwith the Tonga-Kermadecarc, and, amongother SciencebyHypothesis Testing 55 showedthat the deepshocks there are confinedto a very thin zone,less than a few tensof kilometersthick. This observationwas importantfor the Tonga-Fijistudy described in thenext chapter and will be referredto there. In theprocess of suchstudies, of course,Lynn developed a familiaritywith the data sourcesand data archives,and the computingand processingpro- cedures,that would servehim in goodstead later asthe platetectonics revolution began. As Sykesproduced these early studies of ridges,and ridgeoffsets or fracturezones, he not onlydiscovered and delineated those features but he alsofound something very peculiar,and, it would turn out, very significant,about the spatialpattern of seismicactivity and its associationwith differentparts of a fracturezone. To understand this important point, visu- alize an elementof an idealizedriff-fracture zone system(NGT, Fig. 4). Thinkof a slightlydistorted cross, or mathematicalplus sign, in whichthe lowerhalf of the verticalline of the plus signremains vertical but is offset somewhat,at the horizontalline, from theupper half. Now think of the two verticalsegments as designating a north-south trending ridge that is offset at the unbrokenhorizontal line, which in turn correspondsto an east-west fracturezone. The east-westfracture zone is much longerthan the offset betweenthe two ridgesegments and it continuesto thewest and to theeast beyondthe pointsof intersectionwith the segmentsof the north-south trendingridge. Sykesfound that the seismicactivity associated with sucha feature occurredalong both of thenorth-south ridge segments, but onlyalong the intermediatesection of the fracturezone that lay betweenthe ridge intersections.The extensionsof the fracturezone beyond the ridge intersections (tothe east and west in theexample,) were almost completely inactive. That wouldturn out to be a key observation.The entire lengths of thehuge fracture zonesof the seafloor were, in otherwords, not seismicallyactive; only thesegments of thefracture zone between the ridge intersections were. The full significanceof thatobservation was soon to be revealed. In 1965,as a resultof his interestin the seismicityof the area and his Lamont connectionwith the Tonja-Fijideep earthquakeprogram (dis- cussedmuch more fully in thenext chapter), Sykes was working in Fiji for a few months.While Lynn was there, Jim Dorman at Lamontread a paper writtenby TuzoWilson (Wilson, 1965) and publishedin a recentissue of Nature.The paper concerned the new classof faultsthat Tuzowas cham- pioning;they were called transform faults. In essenceWilson's hypothesis predictedthat, for the fracturezone configuration just described, seismicity shouldfollow just the ridge-intermediatesegment-ridge pattern that Sykeshad alreadyfound. There should be little or no activityalong 56 Scienceby HypothesisTesting extremities of the fracture zone. On the other hand, if the more conventional, transcurrentfaulting, hypothesis was usedto explainthe offset,seismic activity shouldhave occurredalong the entirelength of the fracturezone. That point was important and interestingin itself but there was an additional very important point. The hypothesispredicted the focalmechanisms of shocksat eachloca- tion along the ridge-fracturezone featureand a pattern for them different from that to be expectedif the fracturezones were simplytranscurrent faults ratherthan transformfaults. Thus, if the ridgeshad beenoffset by transcurrent faulting along the fracture zone, the earthquakefocal mechanisms shouldhave the samesense of movementas the offsetridges suggested. On the other hand, if the offsetridge segmentshad never been togetherand were the loci of seafloor spreading,then the earthquakefocal mechanisms shouldhave the oppositesense of motion. This opportunityto use focal mechanismsin a criticaltest of an importantnew tectonichypothesis came alongjust as the techniquefor determiningfocal mechanisms was evolving to the point whereit couldbe very useful.Stauder at St.Louis had advanced the techniquenotably usingWWSSN data and begun to producereliable and informativefocal mechanisms,although his resultswere not initially castin the plate tectonicscontext. Brune at Lamonthad usedthe technique in yet anothercontext, and Isacksat Lamont,as describedlater, was focusing attentionon focalmechanisms of, particularly,deep thrustearthquakes in Tonga-Fiji.Sykes' attention was directedprimarily toward the mid-ocean shocks that turned out to be extensional or transcurrent in nature. Dorman wrote a letter to Sykescalling Wilson's paper to his attention and Lynn quickly recognizedthe opportunity.He saw how to testWilson's transformfault model rigorously.If, with the new WWSSN data,he could improve the reliability of the focal mechanismsof earthquakesas he had improved the locationsof hypocenters,he could test Wilson'shypothesis not only with the seismicitypattern but also with the focal mechanisms and their spatialpatterns. The idea was not only good;it was timely.At the time, as was the case for the sea floor spreadingconcept in general, Wilson'shypothesis of transformfaulting was, I think it is fair to say,being taken somewhatlightly by the bulk of the scientificcommunity. Some cyn- ical scientistseven expresseddisdain for anyonewho treatedit seriously enoughto work hard at testingit. Lynn must have had somemisgivings and an interval of indecision over whether he should launch a full scale effort to test Wilson's model exhaustivelyin the faceof someguaranteed criticism. Furthermore, he was alreadyengaged then in a study of seismicityand deep structureof island arcs(Sykes, 1966). At that criticalpoint however,and as notedearlier, Scienceby Hypothesis Testing 57 and I and one or two other seismologistswere invited to the quartersof the geomagnetismgroup at Lamont by its leader, Jim Heirtzler. Jim and his cohortsgave us a very enthusiasticand excitingbriefing on the work his group was doing on the spectacularsymmetric magnetic anomaly data along the oceanridges. The casefor spreadingas postulatedby Hess,and associatedmagnetic anomaly formation as postulated by Vine and Matthews, was being supported sensationallyby the geomagneticians. That meetingconvinced Lynn that seafloor spreadinghad to be taken seriously and that the seismologicaltest of Wilson'smodel was an important thing to do and worthy of an all-out effort. To make a short story of a long and challengingtask, Lynn made that effortand Wilson'stransform fault modelwas supportedin timely fashion. Sykes'data-laden paper was just what Wilson'sidea neededat that time when it was being dismissedor downplayedby someas arm waving speculation. The paper was presentedorally by Lynn at the NASA Goddard Symposiumin New York City in 1966and publishedin JGR (Sykes,1967). Sykes'paper, and figuresfrom it, becamewidely known. It was publicized heavily as supportfor his ideasby Wilsonamong others of course,and the seafloor spreadinghypothesis of Hessreceived a much neededboost. At the NASA-Goddard meeting, Sykes met Bullard, Vine, and Mckenzie for the first time, and the Lamont magneticsgroup and Mark Langsethof Lamont gave talksbearing on seafloor spreading.Lynn wisely publisheda versionof his talk in JGR(Sykes, 1967), in additionto another versionin the symposiumvolume which was much delayedby late sub- missionof other papers.Had he not done so, his work might well have beenpreceded in print by the work of other seismologistswho were beginning to recognizethe opportunityfor them in the early stagesof plate tectonics. In correspondencewith me, Lynn hasnoted meetingsor associationsof somesort with Griggs,Urey and King from outsideLamont and Ludwig, Opdyke, and the magneticsand seismologygroups within Lamont that aided or encouragedhim in his plate tectonics-relatedstudies. To summarize then, Sykes' seismologicalcontributions at this early stageof developmentof the global tectonicstory in essenceincluded the following: a) recognitionwith othersthat precisionof locationof hypocentersand of focalmechanisms could be much improved throughuse of modern WWSSN data and newly developedcomputing procedures. b) demonstrationthat the spatialdistribution of seismicitysupported Wilson's transform fault model and not the more conventional transcurrent fault 58 Scienceby HypothesisTesting

c) demonstrationthat the pattern of focal mechanismsand individual focalmechanisms supported Wilson's transform fault model as well d) establishmentof seismologicalmethods as a valid and important meansfor testingand contributingto the developmentof new tectonic modelsbeing proposedat the time. In demonstratingthe much improved reliability of focal mechanisms and their utility in the searchfor improved understandingof tectonics when WWSSN data were used, Sykes validated Hodgson'sconcept of studyingglobally distributed focal mechanisms as a way to investigatetec- tonicson a large scale,and made clear that great carehad to be taken to insure that the raw data were reliable. Other seismologists,particularly Stauderand Bollingerat St. Louis University,had by this time beenwork- ing on a similar tack and were also obtainingreliable focal mechanisms. However, they were not so much in the early mainstreamof communication on seafloor spreadingand globaltectonics as were the Lamontpeople and so their recognitionof the significanceof their resultsin this context laggedsomewhat, a situationalso experiencedelsewhere by others.There were significantdelays from place to place in the transmittalthroughout the world of scienceof the excitingnew informationthat would eventually grow togetherto becomeplate tectonics,and any particularindividual's contributiondepended to a degree on just when new informationhappened to reachthe individual. Beingin the right placeand being there at the right time were indeedimportant. This point will be stressedfurther in a later chapter. One other point needsemphasis here. In retrospect,and to a newcom- er to the science,what Sykesaccomplished seems rather straightforward and the obviousthing to have done.Wilson proposed a hypothesis;Sykes testedthe hypothesis.What happenedseems just the nextlogical step in the progressof the scienceat that time. And in fact it was. But the next logical step of the scienceis not nearly so easyto seewhen the actualevents are taking place as it is when thoseevents are viewed in retrospect.Wilson publishedhis idea in Nature, a journal widely distributedinternationally and seenroutinely by a large fractionof the scientificcommunity. Yet no other seismologistrecognized and pursuedthe opportunityas Sykesdid. Lynn had the goodfortune to (a) be working with the data and techniques that were necessaryto solvethe problemand (b) to be locatedwhere up-to- date communicationon new developmentsin the sciencewas maintained. Youngscientists seeking to make importantdiscoveries of their own might notethese factors as they designtheir own careers.But they might alsonote that a certaindegree of visionand daringwas requiredof Sykesas he broke with conventionto make a noteworthy Scienceby Hypothesis Testing 59

By the sametoken, we mustnote that not all importantadvances in the scienceare made in like manner.We now turn from our exampleof science by hypothesistesting to an exampleof scienceby serendipity.As we shall seeonce again in the nextchapter, sound strategic and tacticalthinking are importantbut so too are fate and good 7

Scienceby Serendipity

Thereasoning style was induction Thedata, a deepquake production With wavesmuch less stronga At Fiji thanTonga Eureka!A thingcalled subduction!

Aware ofthe exciting advances in the sea floor spreading story being madeby Lamo,nt geomagneticists an,dothers outside Lamont, and with an insiders view of Lynn Sykes positiveseismological test of Wilson'stransform fault hypothesisas that test developed,other Lamont seismologistsbegan to pay increasingattention to the evolving tectonic story.But, I think it is fair to saythat, althoughit is clearin retrospectthat it was a time of specialopportunity, beyond some extensionsof Lynn's early work no one at first saw quite how to deviseand embarkon a major and radically differentnew seismologicalproject designed specifically to take advantageof thoseexciting new developmentsand opportunitiesin largescale tectonics. We managedto do just that, but it was partly because fate and fortuneintervened and gave us a helpinghand. For it would turn out that the projectthat would producean important new advancein the evolvingtectonic story had alreadybeen initiated for another,and what seemedat the time a separate,purpose. It was an effort that I shall refer to here as the Tonga-Fijideep earthquakeproject. The Tonga-Fijiproject was begun in late 1964strictly for the purposeof observ- ing and understandingdeep earthquakes. It was conceivedon the basisof soundreasoning about the stateof knowledgeof the earth, and on a strategy rootedin observationof the unknown,but it was not begunwith the specificintent of testing, or collectinginformation relevant to, the new ideasof oceanfloor dynamics.Happily however,and in timely fashion,the Tonga-Fijiproject produced striking and unanticipatednew evidenceon the structureand dynamicsof island arcs and henceon the processthat would cometo be recognizedand known as "subduction."The Tonga-Fiji 62 Scienceby Serendipity projectalso brought the conceptof mobilelithospheric plates into the story of earthdynamics. These new ideason what happenedat the greatisland arcsand other major arcuatefeatures, the sitesof most deep earthquakes, would thenbe blendedwith the conceptsof seafloor spreading,transform faults, the dynamic mosaicof lithosphericplates, and related ideas to becomethe conceptof plate tectonics. This chapterhence focuses on the Tonga-Fijideep earthquakeproject, on the eventsand people that gave birth to theproject and subsequentstud- iesbased on it, on the strategyand reasoningthat led to the project,and on theserendipitous consequences of the observationalprogram that made the projecta part of the plate tectonicsstory. In anotherbook (The Incomplete Guide to the Art of Discovery)written for anotherpurpose, I choseto use the Tonga-Fijiproject as the prime exampleof a casehistory of discovery. Readerswith a specialinterest in how discoveriesoccur, or in anotherview of the Tonga-Fijiproject, may wish to refer to that book (Oliver, 1993)as well asthe following.Here I retellthe storywith a somewhatdifferent perspective,trying particularlyto set it appropriatelywithin the flow of seismologicalcontributions to the plate tectonicsrevolution. As noted earlier,during the 1950sand 1960sthe Lamont earthquake seismologyprogram that was rootedin the early Ewing and Pressstudies of low frequencysurface waves expanded aggressively into new subject areas.Seismic waves of ultra low frequency,and ultra high frequencywere studied.Large earthquakes, small earthquakes,weather disturbances,and sometimesartificial sourcesgenerated the waves. New kinds of instruments were developedand deployed at various locationson earth and eventuallyon the moon. Young seisrnologystudents entered the field, learnedhow the science operated,and then caughtthe spirit of seekingopportunities in seismology that had previouslybeen hidden, or ignored,or overlooked.To thoseof us who were caughtup in the bustleand swirl of activity,it was a marvelous time, and it was made sobecause we somehowhad acquiredan innerconfi- dencethat we wereworking in a fertilesubject in whichdedication and hard work would revealimportant and hithertounknown information about the earth,the domicileof humans.We had all beeninfected by thebug of earth science,probably by Ewing,and we reveledin the diseaseit gaveus. That period of the 50s and 60s was also a particularlygood time for innovativescience in general.Science was full of vitality,funds for support of imaginativeand oftenrisky projects were available, and thebureaucracy of fundingsources was neitherentrenched, nor stereotyped,nor self-pro- tective.A soundand imaginativeproject that explorednew scientifichori- zonshad a good chanceof being funded and, oncefunded, such Scienceby Serendipity 63 were carried out with a level of dedicationand pride and unselfishdrive and persistencethat stoodout. Some projectsthat were organized under, and nominally a part of, Lamont's earthquakeprogram spilled out into topics beyond the strict realm of earthquakeseismology. During the InternationalGeophysical Year (1957-58)which broughtinto existencethe Lamontworldwide network of Press-Ewingseismographs referred to earlier,Lamont also supplied some personnelfor the remote scientificfield stationon T-3, the ice island float- ing in the Arctic Ocean.As I had had someprevious experience in Arctic science,I was given the task of finding and hiring appropriatescientific personnelfor that station.The work and the living conditionsat that isolated postwere certainlynot routine.There was the difficult environment, all snow and ice. There was limited companionshipand dangerof cabin fever.And the runway for planesthat deliveredpersonnel and supplies was built on snow that made it unsuitablefor use during the summer. Personnelwere flown to the stationduring late spring and had to remain until early fall. Obviouslynot everyonewould or could work effectively under suchconditions. Individuals with someunusual personal traits as well as appropriateskills had to be sought. For the summer of 1957,I had hired a scientistfor that field position well in advance of the departure date, but as that date approachedhe reneged,and I had to find a last minute replacement.With no other appli- cantsfor the Arcticjob, I chosethe best,as yet unhired,candidate for summer work elsewhereat Lamont. He was a junior in Columbia College, majoringin geologyand physics.He had spentthree summersdoing seismic reflectionwork in the steamyswamps of the Gulf Coastof Louisiana, but had no previousArctic experience and so was marginallyqualified on that score.Nevertheless, a professorwho had had him in classspoke highly of him, so I calledhim at homein Louisianawhere he had gonefollow- ing the end of springsemester. It was a Thursdayafternoon. In a few words, I describedthe projectand the job, the trying conditionsand the remote location,and the enforcedisolation, and then askedif he wanted the job. Therewas a pauseof a very few secondswhile he ponderedthis situation that was completelynew to him, and then,decisively he responded"Yes!" "All right,"I said,"Be here on Monday morning,ready to go." He arrived on schedule,adjusted quickly to his new lot, departedshortly thereafter for T3, and, fortunatelyfor both of us, spenta successfuland rewardingfield season on the ice island. I tell this storyat somelength because that individual was BryanIsacks who, after completingthat summerin the Arctic and his senioryear at Columbia, became a graduate student and then a researchscientist 64 Scienceby Serendipity

Lamontand eventuallyplayed a key role in the Tonga-Fijiproject and most of the otherLamont seismological contributions to platetectonics. His name will appearfrequently in the followingand, in fact,that phonecall began a personalassociation between us that hasprevailed to this day and through employmentat two institutions,Columbia, and Cornellwhere Bryanis currently Chairman of the Department of GeologicalSciences and I am a Professor Emeritus. When I see him now, I often think of how different the lives of both of us and many otherswould have been if he had said "no" insteadof "yes"to my outlandishlast minute offer of summerconfinement and adventureon the frozenArctic Ocean. Fate, once again, was kind. Both Bryan and I followed somewhattortuous paths from that first phone call to eventualcollaboration on the Tonga-Fijiproject. Let us now turn our attentionto how that projectbegan. During the mid 1950s,my personalresearch efforts were partly devoted to studyof surfacewaves in the continents,particularly the propagation of highermodes of guidedwaves. The resultswere, I think, somesolid and lastingcontributions to the scienceof seismology,but as suchthey got the attentionof seismologistsand few others,certainly not the generalpublic or the inhabitantsof the politicalworld. Unexpectedlythat degreeof isola- tion of seismologistsfrom other elementsof societywas soonto change. In the late 1950s,while studyingseismograms from the Palisadessta- tion, I found an indicationof somesurface waves that had been generated by an undergroundnuclear explosion in Nevada. It was the first time such waveshad been seenat sucha large distancefrom a rather small,buried, nuclear source.I soon found myself catapultedinto the political world whereinterest in a nucleartest ban treatyhad beenaroused and intensified. As a consequenceof that interest,those few seismologistswith experience relatedto monitoringclandestine nuclear explosions were thrustinto positions as technicalexperts at the treaty negotiations.Geneva, the meeting place for internationaldiplomacy, became the new centerof activity for a smallnumber of seismologistsfrom countriesof the Eastand the West. It was heady businessfor awhile as one sometimesgot the impression that the future of the world dependedon whetherseismologists could dis- tinguishan earthquakefrom an explosion,but after a few yearsof experiencewith intensepolitical bickering and little to showfor it, I yearnedto return to spendinga larger part of my time on fundamentalscientific research.Therefore I began to cast about in searchof a new activity to which I coulddevote a part of my time. Fortunately,the situationwas such that therewas no need for hastein making a decisionabout just when to begin or just what aspectof scienceto pursue.There was, in otherwords, an unusuallygood opportunityfor developinga projectbased on a Scienceby Serendipity 65 thoughtout, soundstrategy. I gave the matter very carefuland prolonged thought and eventuallyhit upon the subjectof deep earthquakesas one that lookedpromising. There were two reasonsfor this decision,one strategic, one practical. Strategically,deep earthquakes seemed a prime and timely target.That deepshocks occurred and that they occurredin zonesdipping beneath arcs to depthsof about700 km was known, mostlyfrom work by Japaneseseis- mologistsled by Wadati.Except for somespeculative hypotheses however, almostnothing was known about the nature of deep shocks,or why deep shocksoccurred where they did. Furthermore,deep shockswere so prominent and so frequentthat they had to be importantcomponents of the seismologicaland tectonicstories of earth dynamics.In addition,past observationsof deepshocks were relativelysparse and spottyso that it seemedthe subjectwas ripe for an observationalprogram. In typical Lamont style,we could set out to observethem thoroughly,even though there was no particular hypothesisto be tested. On the more pragmatic side, there was good reasonto think that a soundproject to study deep earthquakescould be funded. Solid earth scientistshad realizedthe advantagesof internationalcooperation during the IGY (1957-58)and they soughtto maintain and extend thoseadvantages througha sequelto IGY calledthe Upper Mantle Program.It was designed to stimulateearth scientiststo focustheir attentionon that poorly known part of the earth, the upper 1000km or so.The existenceof the internation- ally fosteredproject caused the National ScienceFoundation to look favor- ably on proposalsfor studiesof phenomenaof the upper mantle, and, of course,that was just what and where the deep earthquakeswere. In fact, they were most of the modern action in that part of the earth and hence made an appealingtarget. At Lamont,we bandiedthe idea of a projecton deepearthquakes about and then held a graduateseminar on the subjectto exploreit further. The styleof the seminarwas intensive;each student was requiredto read all of the assignedpapers and cometo classprepared to lead the discussionand critiqueof them. They (and the professor!)thus developedtheir ability to make a hard and tough evaluationof the literature and to formulate and select ideas for future research. I cannot recall the names of all of the handful of attendees,but both Lynn Sykesand BryanIsacks were amongthem. Lynn was stimulatedby what he heardin the seminarto extendsome of his studiesof hypocentrallocations to includemore deep events.He was also an enthusiasticsupporter of an observationaleffort designedto observe deepshocks more thoroughly,but at that point he was not in a positionto commita large portion of his time to a field 66 Scienceby Serendipity

BryanIsacks was alsoenthusiastic about the projectand, indeed,could devotehimself to it and, it would turn out, did. After completingthe Arctic field work and then finishingcollege, Bryan had enteredgraduate school and begun work in seismology.Among other things he participated, togetherwith GeorgeHade, Lamont'sall-purpose seismological engineer, in the installationand operationof instrumentsat a Lamont stationin a deep mine of the New JerseyZinc Company in Ogdensburg,New Jersey. The mine observatorywas designedinitially to accommodatestrain meters and seismographsfor the purpose of studying earth tides and seismic waves of ultra long periods.Bryan's efforts there took anothertack as he observedand studiedwaves at the opposite,high frequency,or shortperi- od, end of the spectrum.As a result,his Ph.D. thesisfocused on small,local earthquakesand on the peculiar,high-frequency, compressional and shear waves traveling through the earth's upper mantle between Caribbean earthquakesand the mine in New Jersey.Such waves, which had beendis- coveredand studiedby Ewing and othersa few yearsearlier, would turn out to be of specialsignificance during analysisof the Tonga-Fijidata. Fate, it seemed,or at leastcoincidence, was at work onceagain. As theplans for it crystallized,the Tonga-Fiji project of coursecalled for tacticaldecisions as well as strategicones. A site had to be chosenfor a detailedobservational program. After considerabledeliberation, we picked the Tonga-Fijiarea, principally because of the high level of deepearthquake activity there, but also becausethe political and logisticalsituation was favorable,and becausethere had been little if any previousseismological observationaleffort there.It seemeda placewhere a large amountof data could be collected in a reasonable time. We decided to run a small network of seismographstations in that area,with eachstation recording three componentsof ground motion in a high frequencyrange that was selected becausethe shockswould all be relativelyclose to, saywithin 1000km of, the recordingstations. The projectwas funded by the National Science Foundationas part of the US Upper Mantle Program.The taskof installation of the network on the islandsof Tongaand Fiji followed. Installingunusual and unfamiliarscientific instruments at remoteloca- tionsin foreigncountries where such activities are non-routine is no simple matterof course,and variouspractical and logisticalproblems arose. They were eventuallyresolved, principally by Bryan Isackswho, with the same verve and determinationthat had earlier taken him to the Arctic on the spur of the moment, moved his wife and three small children to Fiji for more than a year.From that base,and with intermittenthelp from GeorgeHade, Lynn Sykesand others,he arrangedfor and installeda reliableseismo- graphnetwork that spannedthe countriesof Tongaand Scienceby Serendipity 67

Two stations,principally because of their locations,would turn out to be critical.One was locatedon an island of the outer arc of Tonga,not far from the deep Tongatrench. The principal seismicactivity of Tonga,like that of most island arcs,occurs within a zone that outcropsnear the inner wall of the trenchand extendsto depthwhile dipping at a fairly steepangle (about 45ø) beneath the island arc and in a direction normal to the trench i.e. to the west in Tonga.Earthquakes occur throughout the zone which is thin, perhapsonly a few tens of kilometersthick as shown by Lynn Sykes, (Sykes,1966) and which extendsfrom the surfaceto depths of about 700 km. Seismicwaves from earthquakesin that zone,including the very deepestshocks, therefore traveled obliquely up to the Tongastation, traversing, or passingnear to, the seismiczone for almostthe entirepath length. The other criticalstation location was in Fiji where the distancefrom the hypocenterof a very deep earthquakewas about the sameas it was from that samehypocenter to the stationin Tonga.The Fiji station,however, recordedwaves that traveledobliquely through an aseismicportion of the earth'smantle. The geometryand lengthsof the paths to the two stations, in other words, were nearly identical.The differencewas that one path traversedthrough or near the seismiczone; the other path traversed an aseismicand hencemore normal part of the mantle. As the projectbegan and beforethe data beganto accumulate,the dis- tinctionbetween the two kinds of paths based on presenceor absenceof seismicactivity did not get much attention.Seismologists of that era were accustomedto thinkingof the earthas sphericallysymmetrical so that vari- ationsof velocityand otherparameters might occurwith a changein depth, but not with a changein lateralposition. It was the time of what somehave referredto as the "layeredonion" structure of the earth,with eachlayer of the onion different from the layer aboveor below it but lackingin lateral differenceswithin the layer. There was good reasonto hold this view, at leastas a first approximation,because, as noted earlier,seismic wave travel timesare almostentirely a functionof distanceand they are almostinde- pendent of location.Hence the sphericallysymmetric view of earth prevailed, even though it was perfectlyobvious that there was great lateral variation in one particular deep earth phenomenon,earthquake activity. Most parts of the earth had no seismicactivity at depth, a few placeshad continuallyrecurring shocks through a rangeof depths. To reconcilesuch an obviousobservation with the sphericallysymmetric model, the differencesin seismicactivity were, often implicitly, attributed to variationsin dynamic responseto stress,and not to variationsin compositionor temperatureor elasticproperties or some other factor.At somepoint during the evolutionof the Tonga-Fijiproject, I had the 68 Scienceby Serendipity that the deepearthquakes were occurringwhere they did becausethe material of that part of the earthwas somehowanomalous and unlike the mantle at comparabledepths elsewhere, but the projectwas not begun with suchspeculation as the basis. Furthermore, in keepingwith the styleof seismologythat prevailedat the time and in which seismologistsfocused attention on spatialvariations of velocityand little elsewithin the earth,when we consideredthe possibility of an anomaly, we guessed,or tacitly assumed,that any seismicanomaly that manifesteditself in the Tonga-Fiji regionwould be an anomalyin velocity. That guesswas off the mark. A velocityanomaly was indeedeventual- ly discovered,but the revelationabout earth structurethat followed was basednot on subtle spatialdifferences in velocity but rather on blatantly obviousspatial variations of anotherseismological parameter, attenuation. Attenuationof seismicwaves in the earth is typically difficult to measure precisely,so seismologistsof that era had often bypassedthe subjector given it little attentionat best. In Tonga-Fiji,gross spatial differences in attenuationwere huge and robustand unexpectedlyeasy to measure;those differences in attenuation became the critical clue that led to the understandingof a major componentof the plate tectonicsstory. Bryan Isackswas the first to recognizethe unusualnature of the new observations.Like Abraham Ortelius and Marie Tharp, Bryan, living and working at the centralcollecting point for data from the entire Tonga-Fiji network, was the first to seethe new data. Seismogramsof the frequent deep shocksin Tonga-Fijiquickly revealedto him the startling,unantici- pated,and criticallyimportant piece of information.High frequencyshear wavesfrom deepshocks traveling up the seismiczone to Tongawere often as much as three ordersof magnitudelarger in amplitude than the same waves traveling up the path to Fiji that was comparablein distancebut aseismic!High frequencyshear waves were being attenuatedseverely along aseismicpaths but propagatedefficiently through, or near to, the seismiczone. It was a strikingcontrast in observationfor a sciencein which a differenceof only a few percentin someparameter was oftensufficient to merit intensestudy. We were delightedwith that new pieceof what had to be importantinformation on the earth,but it was many monthsbefore we cameto recognizethe full meaningof it, and just how importantit was. Meanwhile, new observations and coordination of our data with other kinds of observationsimproved our knowledgeof the phenomenon.The zone of low attenuation included the seismic zone and a hundred, or at least some tens of kilometers,beneath it. The zone not only contrasted stronglywith its surroundingsin termsof attenuation,but it was alsoof slightly higher velocity than its surroundings.Background Scienceby Serendipity 69 from stationsthroughout the world told us that high seismicattenuation of high frequencyshear waves in the mantle was the casemost everywhere and hencethat it was the seismiczone of Tongathat was anomalous.It was the path from the deep shocksto Fiji that was typical. We puzzledfor monthsover the meaningof our new datauntil at lastthe momentof enlightenmentarrived abruptly; the so-called"Eureka phenomenon" that has beenreported by other scientistsin the caseof surprisingdis- coveries,struck us. It happenedas Bryan and I sat in my officeat Lamont staringat a sketchon the blackboardthat showedthe seismiczone of anom- alouslylow attenuationdipping beneath Tonga-Fiji. Why was it there?And what was it? Then,searching for a clue,I said"The attenuationproperties of this zone are somewhat like those of the shallow mantle between the Caribbeanand Palisades"(as Ewing and later Isackshad observedit). "Why don'twe assumethat the mantlebeneath the Pacificeast of Tongais the same as it is in that part of the Atlantic.Then we could draw it like this." And I drew thenow familiarfigure showing a horizontallayer of oceaniccrust and mantlethat bendsin the vicinityof the trenchand then dipsbeneath the arc and followsthe seismiczone to greatdepths. Almost before I completedthe new sketch,Bryan said "Of course.It's underthrust!"We sensedimmediate- ly that we had the answerand that it was a big one.The rocksthat underlay the normaldeep seafloor to a depthof about100 km or sowere descending as a layer into the interior in the vicinity of the trench-islandarc system. Furthermore,they couldbe tracedto, and hencemust have moved to, at least a depthof 700 km. It was underthrustingon a huge scale!Suddenly, as this model becameclear, all sortsof thingsbegan to fall into place as we envi- sioneda new kind of dynamicsfor island arcs.We were filled with excitementand euphoriaas we beganto recognizethat we had the key to a remark- ablerange of importantproblems of earthscience. For example,at that time a majorenigma had arisenas a consequenceof the growingsuccess of the seafloor spreadingconcept. If, as the spreaders claimed,the seafloor was partingto makeroom for materialfrom below and new surfacearea was being createdat the ridges,how was the entireearth respondingso as to accommodatethe new surfacearea? Some claimed that the earth was expandingat just the proper rate to provide the needed increasein surfacearea. To accountfor the expansion,they challengedthe constancywith time of the gravitationalconstant. Others claimed the earth was not expandingbut that surfacematerial was beingdestroyed elsewhere at just the rate that it was beingcreated at the spreadingcenters. But amongthis latter group therewas disagreementover just where the surfacematerial eventuallydescended into the interior.Some said it sank beneaththe continentsthrough some poorly understoodmechanism and 70 Scienceby Serendipity

, , , a spatialpattern that producedthe observeddeformation of surfacerocks there.Others, perceptively and moreor lesscorrectly, thought the deepsea trencheswere the placeswhere surfacematerial was descendinginto the interior.Some, like Holmes,had earliervisualized great convection cells in the mantlethat broughtmaterial up at the ridgesand carriedit down at the trenches.But none,so far as I know, at that time visualizedthe prominent role of the strong,near surfacelayer of lithospherein the process,nor sensedclearly the full setof relationsbetween that down-goinglithosphere and other phenomenaof island arcs-thetrenches, volcanoes, earthquakes, etc. Our model,with its down-goingunderthrust slab of lithosphereasso- ciatedwith the arcsquickly began to serveas the basisfor explanationof thesephenomena. It subduedthe earthexpanders and the non-arcsinkers, and provideda specificmechanism and model for what was happeningat the arcsthat becamethe foundationfor almost all subsequentstudies of arcsand a criticalelement of globaltectonics. For us, it was a majorconjectural step to go from evidenceon the spatial variationof attenuationand elasticproperties in the earth to a mechanical model with huge slabsof stronglithosphere moving about on a weak, viscousasthenosphere i.e. a model basedlargely on rheologicalproperties. Therewas someintuition and speculationinvolved of course,but the rea- soningalways had a basein studiesby othersof otherkinds of evidence,as well as its prime basisin our seismologicalobservations. We made the key assumption,as we later learnedAnderson had doneearlier in anothercontext, that efficient,high frequency,faster seismic wave propagationcorre- latedwith high strength,and attenuationand slightlylower velocitieswith low strength.Thus Gutenberg's"low velocity layer" describedearlier becameevidence for asthenospherebeneath lithosphere. Studiesof gravity and glacialrebound had producedrheological modelswith a stronglayer over a weak one (Daly, 1940)that couldnow be reconciledwith seismicobservations. While in schoolstudying earth science, it had alwaysseemed strange to me that earth scientistsused two distinct- ly differentnomenclature systems to describethe upper part of the earth's interior, one the crust-mantlesystem, the other the lithosphere-asthenosphere system,yet little effort was made to reconcilethe two. Thus I was relievedand pleasedwhen we seemedto have found a way to do so. With the assumptionthat efficientpropagation indicated strength, the correlationbetween degree of attenuationand the spatialposition of ray pathscould be used to determinethe thicknessof the lithosphere.In our study,that figure turned out to be about100 km, a numberthat is still con- sidereda goodfirst approximationfor a layer that almostcertainly varies considerablyin thicknessfrom place to place and for which the Scienceby Serendipity 71 ment of the depth to the lower boundaryis tricky.Even the definition of that boundary is subjective.Other aspectsof the model seemedto make goodsense intuitively, although it was sometime beforequantitative studies were made, oftenby others. For example,it seemedthat becauseof the very low heat conductivity of the earth,the down-goingslab had to be significantlycolder than its surroundings.The earthquakeactivity seemed concentrated in or near down- goingoceanic crust at the top of the 100 km thick slaband not distributed through the entire thicknessof the lithosphericslab which, of course, includedthe entire oceaniccrust and part of the upper mantle. And the deformedshape of the slab-flatbeneath the oceanfloor, curved abruptly near the trench,and then near planar againat depth-fit our intuitive sense of what shouldhappen to it as a resultof hydrodynamicaleffects, although only afterwe did somesimple experiments to observethe effect,but on this particularpoint we neverproceeded much beyond the intuitive stage. As the story developed,and so long as we confinedour attentionto observationsand physicalmodels that fell within the realm of geophysics, we felt reasonablycomfortable about the conclusionswe were reaching.A modelthat callsfor underthrustingof hundredsor thousandsof kilometers and correspondingdeformation on a large scalemust have geologicalcon- sequencesthat go well beyondthat realm, however.As it seemedthat the underthrustingphenomenon must alsobe relatedto, and was probablythe causeof, volcanicactivity of the arc,we had to determinewhether the vol- canologicalor petrologicalevidence supported or disprovedthe model. We searchedthe literaturebut found no evidencethat we could recognize as definitive on this point until we found an unusualpaper by Coats (1962)in which he proposedunderthrusting of the seafloor to depthsof a hundred or so kilometersto accountfor petrologicalobservations in the Aleutians. His results seemedfully compatiblewith our suggestionof underthrustingto still greaterdepths, and henceto reconcileAleutian geology with our model. We were relieved and encouragedto find this supportingevidence, and Coats'paper, which had not receivedgreat attention initially,quickly and deservedlybecame well-known as an importantand innovative contribution to earth science. Needlessto say,the successwe were having with the Tonga-Fijidata and the down-going slab model made Bryan and me into enthusiastic adherentsto what was a small but growing group of advocatesof a new earthdynamics. Lynn Sykeswas alreadyon that tack and he was stimulated further by the Tonga-Fijiresults. We all beganto expand our thinking from the islandarc scaleto the globalscale, and, althoughwe did not work out the particularconfiguration of the platesand the geometricalscheme 72 Scienceby Serendipity their motions(that was being done by JasonMorgan at about the same time), we neverthelesshad the generalconcept of a global set of moving elementsof lithospherein mind. In factwe describedour ideasin print as the "mobilelithosphere" concept. A paragraphof our paper,(Oliver and Isacks,1967) which was submitted to JGRprior to the time of the famousAGU meetingof 1967when the paper was presentedorally, stated: "Although the conceptpresented here of a lithospherethat is discon- tinuous,underthrust at island arcs,and spread apart or flexed in other areasis basedpartly on assumptionand speculation,the seismicdata sup- portingthe conceptare sufficientin substanceand the implicationsto geo- tectonics are so broad that it merits serious consideration." We were on the right trackwith that paragraphall right, but more tentative than we would have been just a few months later. In fact, in September of that year (1967) at the International Symposium on ContinentalMargins and Island Arcs held in Zurich, we gave a similar paper,later publishedin the CanadianJournal of EarthSciences (0liver and Isacks,1968), but with considerablygreater emphasis on the mobilelithos- phereconcept and its potentialfor resolvingcomplex problems of tectonics and surfacegeology. As we analyzedthe Tonga-Fijidata and beganto think aboutthe down-goingslab of lithosphereand the movingplates on a more global scale,we were, of course,bringing in and integratinginformation from othersources such as geomagnetism and marinegeophysics to fill out the story,but we felt that our independentrecognition, based on seismic data, of the important role played by the lithospherewas original and unique. But, of course,as nearly always happensin science,that wasn't quite so. A few weeksbefore the AGU meetingof 1967,Walter Elsasser, a well- known physicsprofessor from Princetonand Maryland gave a talk at Lamontduring which he presentedhis theoretical model of earthtectonics in which a rigid outer layer that he calledthe "tectosphere"played an important role. In someways his tectospherewas clearlyequivalent to the lithosphere as we used that term, and so we had a good discussionabout our mutualinterests following his talk and,I think,reinforced each other's views. Elsasser'sideas were a remarkableexample of soundconjecture by a theo- reticianwho had a solid senseof appreciationfor the observationsof earth science,but Elsasserpublished his ideasin a rather obscureplace (Elsasser, 1967)and they havenever received the attentionand creditthey merited. Our paper was definitelynot the only study of relevanceto plate tec- tonicsthat was presentedat that remarkable1967 meeting of AGU. Several excitingpapers on geomagnetismand sea floor spreadingwere given Scienceby Serendipity 73 that meetingand sowas JasonMorgan's paper on the globalconfiguration of the plates and the simple schemefor describingtheir motions.For the first time, I think, the idea that somethingof enormousimportance was happeningto earth sciencebegan to spreadwidely beyond the handful of early investigatorsand through the entire earth sciencecommunity. The atmosphereat that meetingwas electrifying.People were literally dashing throughthe halls from one sessionto another,stopping briefly to cry "Did you hear that one?!"when they encountereda friend. I have never attend- ed anotherscientific meeting where the level of excitementwas comparable to, or even approached,that of the 1967AGU meeting. Many came to discussour paper with us after it was given, and we receivedenthusiastic support from people like Tuzo Wilson,who was trying at that time to integrateeverything that was relevantinto the tectonic story and so was delightedwith our results,and Dave Griggs,who was pleasedand enthusedto find somehard evidencethat indicated that his early ideasand his model experimentson mantle convectionin the laboratory (Griggs,1939) were on the right trackand receivingsome support from observationsof the earth.And of coursethere were someskeptics, but their numbers decreasedremarkably over the next year or so as the story was strengthenedfurther. Before our paper was published,we learned that Katsumata(1967), Utsu (1967),and even Honda (1956)earlier, had found in Japan,as we had for Tonga-Fiji,that the seismiczone there correspondedto a zone of low attenuation,so we referencedthem as support for our observationsby adding a note in proof to that effect.The Japanesestudies reinforced our observationsbut the early word aboutsea floor spreadingand relatedmat- ters had not reachedJapan and their resultswere not interpretedby them to indicatelithospheric underthrusting as we had done with the Tonga-Fiji data. Our paper appearedin JGRin August of 1967,and it was widely cited by other authorsand often referredto by speakersfor awhile. Two figures from the paper have often beenreproduced. One showedthe cross-section through the down-going slab with zones of high and low attenuation labeledas such.The other,identical geometrically, showed the samezones labeledas lithosphereand asthenosphere.The secondfigure was commonly usedfor a while by othersto illustratethe essenceof the down-goingslab model of arc dynamics.(See NGT, Fig. i for a three-dimensionalversion of this figure.) The text of the paper suffereda bit becauseof a fluke in the editing process.At the time, JGR was in the processof abolishingfootnotes, so when our manuscriptarrived with an early footnoteto a line in the 74 Scienceby Serendipity duction, the editor promptly moved it into the text, thereby disturbing what I, at least, saw as the smooth flow of the text of the introduction. In any case,the time of that paperis long pastand it is now largelyforgotten; it was certainlynever so well known as the paperthat is the focusof the next chapter.Nevertheless, for me the thrill of discoverywas greaterin the Tonga-Fijistudy than in any scientificeffort I haveparticipated in beforeor since. Those who have read this chapter carefully will recognize that serendipitywas a majorfactor in the discoverythat arose.There was some soundstrategic and tacticalthinking as the projectwas conceivedand as it developed,but it was only becauseof a seriesof fortunateand fatefulcoin- cidencesthat the essenceand the full scopeof the discoveryarose and evolved. Discovery-boundscientists can strive to position themselves favorablyso that an importantdiscovery is morelikely to happento them, but whether it does is still somewhat a matter of fortune, still a matter of being,through chance, in the right placeand at the right time. We turn now to yet anotherstudy, this one of decidedlydifferent style than eitherLynn Sykes'paper of 1967or that by Oliver and Isacks(1967), but onethat also producedimportant and widely-knowncontributions to the development of the conceptof plate tectonics,and that broughtthe interplaybetween seismologyand platetectonics to the attentionof almostall earth 8

Scienceby Synthesis

First theycalled it Tectonics,New Global Todescribe the lithosphere, mobile Plate Tectonics it became But what's in a name? It's thescience that's ignoble or noble.

ith thesparkling 1967 meeting of AGU a stimulusas well asa highlight,interest and activity in researchrelating to the developing story of globaldynamics accelerated in the period imme- diatelythereafter. It was still too early to describewhat was happeningas a consequenceof a "bandwagoneffect," but individualsand smallgroups at scatteredlocations were sensingthe winds of changeand developingthe fervorthat would eventuallycreate that unmistakablebandwagon. The early successof Lynn Sykes' seismologicalpaper on transform faultsand the widespreadattention given to the paperby BryanIsacks and me on the down-goingslab, plus the excitementgenerated by the 1967 AGU meeting and the news of assortedadvances elsewhere, stimulated Lyrm, Bryan and me to join togetherfor intensivefurther action.We were alreadypart of the small fractionof membersof the earth sciencecommunity, mostly geomagneticiansand a few geotectonicists,who had sensed that big thingswere happeningin earth scienceand that biggerthings still were likely to follow.It was easyto seethat seismologywas almostcertain to be a major contributorto thosedevelopments. We alsorecognized that we had been fortunateenough to get a temporary,and almostcertainly short-lived,lead on other seismologists.Therefore we felt that, at Lamont with its archivesof data, supportingfacilities, and colleaguesin related fields,we were in a strongand probablyleading position to make a truly major advancein both global dynamicsand seismologyby relatingthose two subjectsthoroughly and comprehensively.In fact, we convincedour- selvesthat we had a once-in-a-lifetimeopportunity and we set out with great enthusiasmand urgencyto make the most of it. I make thosestate- 76 Scienceby Synthesis mentsrealizing, of course,that not all scientistsmay have seenour status and their statusat that time in the light that we did, but I describehere our attitude and our perspective,as best I can recall it, as a frank report and evaluationof how someof us were thinking at that time of changeand unrest in earth science. In any event,Lynn, Bryan and I made an ambitiousplan with the ulti- mategoal of writing a grandpaper that would reporta comprehensivestudy of everyaspect of the field of earthquakeseismology that boreon the developing storyof globaltectonics. We intendedto testthe new ideason tectonics againstall relevantseismological observations, to developand enhance the storyof globaltectonics with new ideasderived from the seismological perspective,and to call attentionto problemsand opportunitiesin the field of seismologyrevealed through the perspectiveof the new tectonics. To make a long storyshort, we did all of thosethings as bestwe could and we published a lengthy paper in JGR of September1968 entitled "Seismologyand the New GlobalTectonics." Had we written it a few years later the title of that paper would have been "Seismologyand Plate Tectonics."At that time, the term "platetectonics" was not in useand sofar as I know had not beendevised. We usedthe term "new globaltectonics" or sometimes,"mobile lithosphere concept" for the nascentsubject that would eventuallybecome commonly known as "platetectonics." We referredto the entitiesnow known as the platesas "blocks,"or "thin blocks,"or "tabular blocks,"or "lithosphericblocks ." Therewas considerableapprehension on our part as we publishedthe paperbecause it was a ratherabrupt and radicaldeparture from convention and we fearedthat someoneor somethingunforeseen might ariseto prove that it was a major misstep.However, and fortunately,the paper was a major success.It was widely read and widely cited, partly I suppose becauseof the innovative content, partly becausethe comprehensive reviewof somuch data gave it an air of authority,and partly becauseit was readableby non-specialists.And it was timely, bringing the news and excitementof the developingplate tectonicsstory to the large and varied internationalaudience that JGRreaches, just as that audiencewas primed for that news.For the authors,I am happy to say,the paper turned out to be the satisfying and once-in-a-lifetimeaccomplishment that we had dreamedit might be. In the following, I discussthe paper and the effortsthat went into it, attemptingnot merely to describethe methodsand resultsin the scientific fashionthat the paperrelies upon almostexclusively, but alsoto portraythe nature of some of the activity,the spirit, and particularly the interactions with other scientistsand their work. Thosewho want completedetails Scienceby Synthesis 77 the sciencemay, of course,consult the originalpaper as found in the appendix. From here on in this chapter,as in earlier chapters,that paper, "Seismologyand the New Global Tectonics,"will be referred as the "NGT paper." As notedpreviously, the early work of the geomagneticistson the relation betweenmagnetic anomalies and sea floor spreadingwas a strong stimulusfor Lamontseismologists as, during the mid 1960s,we first moved into the stream of activity that would produceplate tectonicswith the papersdescribed in the two precedingchapters. By the springof 1967how- ever,concepts and ideasarising from elsewherewere providingadditional stimuli, particularlyfor phenomenaof global scale.Elsasser's theoretical model of tectonics mentioned earlier became known to us. Then Jason Morgan of Princeton(Morgan, 1968) presentedhis geometricalmodel basedon Euler'stheorem that describedthe motionof the platesso simply and elegantly.In the processhe made and reportedthe first representation of the configurationof the six major platesof the earth.Xavier LePichon, who had comefrom Franceto Lamontto study oceanographyaand who was oncean outspokenopponent of continentaldrift, correctlysensed the changethat was in the air and, combiningMorgan's model with information on spreadingrates from the magneticanomalies, calculated and plotted the relative rates and directionsof motion at plate boundariesover much of the world. LePichon'smaps and calculations(LePichon, 1968) becamean importantbase for relatingseismological parameters to plate motionsin the NGT paper,and we were fortunateonce again to have the authorof a paperso relevant to our work asa colleagueat Lamont.A sign of recognitionby othersof the importanceof seismologicalevidence to the new tectonicsarose as McKenzieand Parker (1967),referencing our paper on Tonga-Fiji,used the plate conceptto explainfocal mechanisms of earthquakes,volcanism and otherfeatures of the northernPacific. Finding that otherswere on a similartack was, of course,a spurto us aswe worked on the NGT paper. As we beganwork on the NGT paper,Lynn, Bryanand I recognized that a large intensiveeffort was ahead of us and that it would have to be done expeditiouslyif we were to stay at the forefrontof seismological researchon this topic.In organizingthe work, we consideredthe possibility thatone of uswould takeprime responsibility for thepaper with theoth- ers contributingsecondarily, but we rejectedthat plan becauseit would haveput toomuch of theburden on oneperson. Instead we agreedthat all threeof us would work as hard as we couldand with a senseof urgency until thepaper was in final form.So as to fosterthis arrangement we agreed that, after the paper was completed,we would determinethe order 78 Scienceby Synthesis authorsby lot. We did soand the paperhas a footnoteon the firstpage indi- catingthat authorshipwas orderedin that manner. The footnote provoked some to suggestthat we must have had a major disputeover authorshipthat causedus to turn to resolutionin that way, but that suspicionwas incorrect.Relations among Lynn, Bryan,and me were amicable at the time and have remained so ever since. Furthermore,our namesall becamefairly well-known as a resultof that paper so order of authorshipwas not nearly so importantas somemight have imagined.One wag, I think it was Jim Gilluly, a prominentsenior geologist,said he thoughtthe paper must have been written by Charles Dickens because the names of the three authors were similar to those of characters in Dickens' novel "Oliver Twist!" The organizationof the NGT paperand of the studieson whichit was basedwas the result of rather carefulconsideration. We choseto organize the effortsand the text accordingto the principal effectsof the new global tectonics,and not accordingto the classicaldivisions of seismology.Thus majorsubdivisions of the paperhave titleslike "ridges,""island arcs" and "movementson a global scale,"rather than "travel time curves.... focal mechanisms"or "seismicity,"although these latter topicsare discussedin minor subdivisions.This style of organizationseemed to work out well, and given the successof plate tectonicsas a unifying themefor all of earth sciences,it is surprisingthat it is not morecommon. All moderntextbooks of beginninggeology that I know,for example,organize the subjectof solid earth sciencealong classicallines and then includea chapteron plate tec- tonicssomewhere later in the book. It seemsthat alternativelyone might presentplate tectonicsat the start as the underlyingscheme of how earth works, and then discussvarious kinds of observationsand evidenceas they fit into, and are a consequenceof, that plate tectonicsmodel. We hoped that the NGT paper would appeal to a large and diverse audienceso we tried to limit jargon, to make the introductionand most other sectionsreadily readableby non-seismologists,and to includesome figuresthat were cartoon-likeand designedto illustrateconcepts rather than to presentdata. One figure,the blockdiagram (NGT Fig. 1) that illus- tratesthe basicelements of what is now plate tectonics,became very popular and hasbeen reproduced frequently and in a wide variety of places.It appearsin many textbooks,and it is consideredone of a set of what some (seeLe Grand, 1986)have calledthe "icons"of plate tectonics.The "icons" in thissense are a few figuresthat, takentogether, communicate graphical- ly and almostat a glancethe essenceof plate tectonicsto the uninitiated. Yearslater, recognitionin retrospectof the specialand crucialimpor- tanceof thoseicons in the spreadingof the conceptof plate tectonics SciencebySynthesis 79 within and outsidethe scientificcommunity has left me firmly convinced that scientistsshould make much greater effort to conveytheir primary sci- entificresults in pictorialfashion, with simplefigures to accompanycon- ventionaltext, be it jargon-filledor jargon-free.Few readersof a scientific journalstruggle through every word of text;most skim the abstract and leaf throughthe figures.The essenceof the paperwill reacha far largeraudi- enceif conveyedby the figures. As, Lynn,Bryan and I werepreparing the NGT paper,other seismolo- gistsat Lamontwere catchingthe platetectonics fever. Jim Dorman had been the first Lamont seismologistto capitalizeon the advent of digital computing.It was the time of the IBM 650,a majortechnological marvel thenbut no matchfor a desktopof today.Nevertheless, that early computer openedmany new horizonsin geophysics.Having long been familiar with the relativelycrude maps of seismicityfound in Gutenbergand Richter'sbook, and thennoticed how J.P.Roth• in Francehad improvedthe detailand informationcontent of suchmaps by usingmore modern data, I suggestedto Jim,and he agreed,that he mightuse the computerto go one stepfurther and plot the new hypocentral data of theWWSSN era. Perhaps, we thought,a still more revealingset of maps would result.Jim and Muawia Barazangitook on thatjob, and with considerableeffort produced a setof globalmaps of seismicitythat alsobecame icons of the plate tectonics story. The mapsshowed far moreepicenters and moreprecisely and accu- ratelylocated epicenters than previous maps, and some maps showed epi- centersfor shocksin only certainranges of hypocentraldepths. All of the mapswere published in a separatepaper by Barazangiand Dorman (1969), and the oneshowing shocks at all depthswas alsoreproduced in the NGT paper(NGT, Fig. 15).That map was a dramaticdemonstration of how the seismicbelts define the plate boundaries and in turnhow mostearthquakes occurat thoseboundaries. Everyone who sawit graspedthat fundamental relationshipat firstglance and thepaper became widely-known. Barazangiat thetime was a graduatestudent relatively new to Lamont, havingarrived there from his nativeSyria by way of the Universityof Minnesotawhere he had obtained a mastersdegree in geophysics.He would follow his first and very notablepaper in seismologywith many other contributionson seismologyand tectonics,some mentionedelse- wherein thisbook. When the seismicitymaps were published,Jim Dorman was a former Lamontgraduate student turned researchscientist. He was broadlybased in geologyand geophysics.He not only foresawthe potential of digitalcomputers in earthscience but he wasinstrumental in bringing Lamontits first centralcomputing 80 Scienceby Synthesis

In a relatedeffort at aboutthe sametime, Paul Pomeroyat Lamontwas using the newly-found computingcapability to make films that showed frameby framethe evolutionof globalseismicity with time. However,with rare exceptions,at that early stagethe temporalpatterns were so complex that we could not characterizethe activity in a simplemanner. That subject continuesas a prime researchtopic today as improved methods of investigation are being developed.Pomeroy's work and expertiseprovided support for Barazangiand Dormanas they made their now-famousmaps, however. The widespreadattention given the Barazangi-Dormanmap stimulated othersto try to make still more informative seismicitymaps. Some did soby usingnew computergraphic techniques capable of producingfigures with severalcolors. Somewhat ironically these maps, though well-done and full of information,never receivedthe attentionor the widespreaddistrib- ution of the Barazangi-Dormanmaps, simply becauseat that time it was so much easier to reproduce,publish and otherwisemanipulate black and white figuresthan coloredones. Furthermore the black and white maps producedan immediatevisual impact on the viewer becauseof the stark black-white contrast, whereas the more gently-toned colored maps requiredprolonged concentration and resolutionby the viewer beforea comparableimpact was felt. I draw attentionto thesematters to emphasize againthe specialimportance of designingfigures for scientificcommunica- tion that do not merely hold contentbut that alsoconvey that contentread- ily and at first glance. The Barazangi-Dormanmap of seismicityfor the entireearth (with sup- plementsfor the polar regionsthat were missingfrom the mercatorprojec- tion) was the basisfor many deductionsand conclusionsof the NGT paper. The narrow globe-encirclingseismic belts formed a pattern that enclosed certainlarge areas and sooutlined the plates.Almost all seismicityoccurred in thosebelts and at thoseplate boundaries,but a few shockswere enig- maticallyscattered within platesand throughportions of the interiorsof continents.The patternhad otherdistinctive characteristics. At intersections of seismicbelts, three segmentscame together, not four or more. Belts,in other words and as noted earlier,did not crossone another.It was just the pattern to be expectedfor the boundariesof a mosaicof irregularplates. Zonesof platedivergence, the spreadingcenters with theirridge-transformfault configuration,displayed only moderateseismic activity and, as Sykes had found,that activityoccurred at the ridgesand at the segmentsof fracture zonesbetween ridges, not alongentire fracture zones. Earthquakes at the divergentmargins were all of shallowfocus, and seismicactivity in general was moderate,not only in frequencyof occurrencebut alsoin size.No very largeearthquakes occurred at the divergent Scienceby Synthesis 81

Most of the earth's seismicactivity, including the very largestearthquakes,and all of the shocksat depth includingthe very deepestshocks, occurredat the generallyarcuate convergent margins. In sum, the global pattern of seismicityturned out to fit in astonishingdetail the geometrical patternof the plates.Furthermore, it turned out that, with few if any exceptions, earthquakesalways occurwithin lithosphericplates or at points of interactionbetween two lithosphericplates. No earthquakesoccur in the asthenosphereor elsewherein the earth. Thus global seismicityprovided spectacularsupport for the plate tectonictheory; perhaps at the time it madethe mostconvincing argument yet for thosejust being indoctrinated into the subject.The claimsof circularrea- soningbased on the use of seismicitydata to outline the platesand then in turn the use of the outline of the platesto explainthe seismicitywere dis- pelled as other kinds of evidencesuch as the patternsof trenches,ridges, arcsand volcanoesfell into place in the new scheme.Still anomalousand unexplainedwere the shallowseismicity scattered through parts of the continentsand a few deep shocks,particularly beneath Spain, but thosewere minor componentsof the globalpattern. It is hard to expresshere the joy and satisfactionfelt by seismologistsas, almostovernight, we went from knowing only that earthquakesoccurred in a distinctivepattern to under- standingjust why it happenedthat way. The successfulmerger of the plate tectonicsmodel and the global seismicity patternswas on the onehand satisfyingand on the otherhand hum- bling. Our egos were alternatelyinflated and deflated as we found that details and featuresof the seismicityrevealed important informationon tectonics but at the same time called attention to the fact that some of those observationshad been availableto us earlier and ignored or overlooked. The three-armednature of the intersectionof seismicbelts, for example, couldhave been pointed out much earlierbased on crudemaps of seismicity but, to the best of my knowledge,never was. Beforethe mid 1960swe seismologistscould and should have been more observant and more inquisitiveabout many suchmatters than we were. We focusedon topics that were in vogue then and ignoredor failed to resolvesome that would be cruciallater on. I hopethat callingattention to thoseembarassing short- comingsand oversightsof that era will encouragemodern scientists to step backoccasionally and searchfor thingsof comparableimportance that they might be overlookingtoday. In addition to the globalgeographical patterns of earthquakefoci and maximum earthquakemagnitudes, another kind of global information becamethe basisfor an importanttest of the plate tectonicstheory in the NGT paper.It concernedthe focalmechanisms of earthquakesand 82 Scienceby Synthesis larly the spatialpattern of the orientationsof thosefocal mechanisms.The efforts of Honda, Byerly, Hodgson, and others were coming of age. Whereasthe work of thosepioneers had sufferedbecause of unreliabledata and primitive computing technology,by the time of the NGT paper Stauder, Sykes, Isacks, Brune and others had recognized the much improvedquality of the new data from the WWSSN. Reliablemechanisms were being determined then by Stauder and colleaguesat St. Louis University and by a few others.With new computingfacilities seismologistsbegan to producereliable focal mechanisms for shocksthroughout the world. Sykes had begun with examinationof the ridge-transformfault areas,and he joined other Lamont seismologistsalready working on the zonesof convergence,to includethe entireworld. Isacks,in particular,had beenfocusing on Tonga-Fijiearthquakes and, oncethe plate conceptarose, came up with the notion of slip vectorsas an indicatorof relativeplate motions.He demonstratedthe agreementof slip vectorswith convergence in Tonga,with the hinge fault at the northernend of the arc, and with the transformfault leading to Fiji, and presentedthe resultsat the 1967IUGG meetingin Zurich. Althoughthis work was carriedout and presentedorally beforeMcKenzie and Parker'spaper with a relatedtheme appeared, the paperdid not appearin print until 1969(Isacks et al, 1969)partly as a result of delay in preparationand partly becausea reviewerrequired six months to completethe review!That was a very long delay indeedduring thisperi- od when the sciencewas advancingso rapidly. By the time of the NGT paper,more than eighty slip vectorsbased on focalmechanism studies for variousregions of the world couldbe plotted on a globalmap and comparedwith the theoreticaldirections of slip calculated by LePichonfor the six-platemodel (NGT, Figs.2&3). The overallfit of focalmechanism data to the LePichonmodel was remarkablygood and, althoughthere were someminor discrepanciesthat could be attributedto the contortionsand complexitiesof the intraplate zone, the focal mechanism data provided yet anotherkind of strongsupport for the plate tectonic theory.The dynamicsof the plates,in otherwords, seemed to be the principal causeof most of the modern world's earthquakes,and the type of motionat a particularplace reflected the entire globaldynamic story. Althoughbut oneof the manyrevelation during the plate tectonicsrevolution,it was a pleasantsurprise to recognizesuddenly that therewas a simple,readily understood,relation among major shocksthroughout the world. A few yearslater I accompaniedClarence Allen of Cal Techon a trip to the Philippineswhere a major earthquakehad occuredalong the great PhilippineFault. After somedifficult travel to a remotevillage near the epicenter,we found ample geologicalevidence for the fault motion and, Scienceby Synthesis 83 enough,the fault, which had moved about three meters,had done so in a left-lateralsense, just as the globalplate theorypredicted! It was euphoric to realizeand confirmthat, with just a map, the plate model, and a few cal- culationsmade at an officedesk, we couldnow predictaccurately the sense of movementof the earth during a major earthquakeat a remote site in anotherhemisphere. The remarkablefit of worldwideseismic data to a very simpletheory of globalscale seemed awesome enough, but just as remarkablythe moving platemodel also seemed capable of producingexplanations for a wide range of seismologicaland otherphenomena at regionaland localscales as well. For the convergentzones, or "sinks"as they were sometimescalled in contrastto the "sources"of surfacematerial at theridges, the NGT paperreported a variety of seismologicalevidence that was relatable to the platemodel. There was, of course,as described in the precedingchapter, the zoneof high attenuation and high velocityassociated with the deep seismiczone of island arcsas shownby Oliver and Isacks(1967) for Tonga-Fiji.There was evidencefor a similar zone beneathJapan from Katsumata,Wadati and Utsu. And there were signsof a similarvelocity anomaly in theAleutians from studiesof travel timesby Cleary (1966),Herrin and Taggart(1966), and Carderet al (1967)of waves from Longshot,a nuclearexplosion buried there. Finding velocity anomalies was easier when the source was artificial because the time and location of the originof the seismicwaves were precisely known. Seismicdata from smallerartificial sources provided other kinds of useful information,particularly in the vicinityof the trenches.Refraction studies, penetratingonly to very shallowdepths, showed that the top of the mantle wasat greaterdepths beneath the trench than it wasbeneath the normal deep seafloor. Based on this information,and on the gravity data that revealed large negativeanomalies associated with the trenches,some suggested the trencheswere zonesof extension.Others thought that the mantlewas merely warped downward at the trenches.Some thought the crustwas thinner or down-droppedand the mantlepulled apart there. However it turnedout that the refractiondata couldbe interpretedstraightforwardly if the deeperposi- tion of thetop of the mantlewas explained as a consequenceof the descentof the lithosphericslab, of whichthe mantle,of course,was the majorpart. As noted earlier, the extensional model of trench formation received a temporaryboost when it was shown by reflectionstudies that sediment- filled grabensoriented so as to indicate extensionnormal to the trench occurbeneath the outerwall of sometrenches. However it was quicklyrec- ognizedthat the bendingof the lithospherein the down-goingslab model would alsoproduce such extension, a kind of skin effectas the slabbent as it went throughthe zone of maximumcurvature. It was alsosuggested 84 Scienceby Synthesis the NGT paper that the grabens,which of coursewere filled with marine sediments,might incorporatesome of thosesediments into the regimeof the down-goingslab and so carry them to depthsof a hundredkilometers or sowhere, following Coats, they would be a significantfactor in the gen- erationand formationof volcanicrocks of the arc. The grabensthus provided supportand not contradictionfor the down-goingslab of the plate tectonicsmodel (NGT, Fig. 8). Stauder(1968) made a beautifulstudy of focal mechanismsof shallow shocksin the Aleutian arc, demonstratingthat the slip vectorsof the large compressionalearthquakes beneath the inner wall of the trenchmaintained an orientationconsistent with the directionof regionalplate motionand not the localorientation of the arc.On the otherhand, the outerwall earthquakes associatedwith the zoneof grabenswere extensionalin nature,with the axis of extensioneverywhere normal to the arc.These results were precisely what the plate modelpredicted, and soprovided strong support for it. Someattacked the model of the downgoingslab on the groundsthat seismicreflection studies often showed undisturbedflat-lying sediments on the floors of the trenches and hence indicated that the trenches could not be the sitesof greatcompressional deformation. As notedearlier, this objec- tion was counteredby pointingout that the flat-lying sedimentswere probably very young and that they were undisturbedbecause they were lying on the outer part of the trenchfloor that had not yet reachedthe deformation zone.Indeed, against the inner wall of the trenchthere was a largelow densityzone that could be attributed to the part of the oncesimilar flat- lying trenchfloor sedimentsthat had sincebeen deformedand piled up againstthat wall as they were scrapedfrom the down-goingslab. Some of those sedimentshad been scrapedfrom the slab while others,perhaps mostlythose in grabens,were carriedto depth with the slab. Later,and after the NGT paper,as the resolutionof the seismicreflection techniquewas improvedand samplesof sedimentswere collectedby coringand drilling, it becameclear that indeed all the evidenceon both the flat-lyingand the deformedsediments was consistentwith the down-going slabmodel, or, as it is now known, the subductionmodel of trenchdynam- ics. In fact, the global view in plate tectonicscontext showed that young active trencheswere in general relatively free of large volumes of sediments, whereas older and less active trenches had more sediments. Of coursethis generalizationwas oversimplifiedbecause the natureof nearby sourcesof sedimentswas an important factoras well. In any case,the dif- ferencesamong trenches with regardto sedimentvolume and distribution that had oncebeen an enigmato marine geologistsseemed to be explain- able under the new model of subductionthat was basedinitally on Scienceby Synthesis 85 data. Other kinds of geologicdata were beginningto fall into placeas well and would continueto do solater and well after the time of the NGT paper. One interestingpoint of the NGT paper concerneda possiblerelation between the plate model and the generationof tsunamis.Most major tsunamishave occurredin associationwith shallowearthquakes in convergent zonesat locationswhere the plate model predictsthrust mechanisms. This associationmakes sensebecause a thrust earthquakeproduces the large vertical displacementof the sea floor that would generatea large tsunami,but a strike-slipearthquake does not. It was yet anothercase of the globaltheory explaininga local effect.Whether other phenomena,such as submarinelandslides, slumps, turbidity currents,or earthquakesassociated with normal faulting and which alsoproduce a large componentof vertical displacementof the surface,are factorsin the generationof some tsunamisremains an open questionhowever. In the convergentzones, in additionto the low level of shallowseismic activity of extensionalnature associatedwith the outer wall grabens,and the very high level of shallowthrusting activity associatedwith the inner wall, there was also some minor, also compressional,shallow activity, behind the arc to accountfor. Although this activity is more scatteredand lesswell delineatedthan the activity associatedwith the grabensor the zone of thrustingat the inner wall, compressionin the lithospherebehind the arc seemedreasonable enough and compatiblewith the model. Later, however, it would turn out that extension could also occur behind the arc at certaintimes and placesbut suchextension also could be explainedby the model.This phenomenonis discussedfurther in the next chapter. And then, of course,there were the deep shocksthat occurredin the convergentzones only. They were commonthrough a wide rangeof depths in somearcs, but sparseand not sowidely distributedin others.At the time the NGT paper was beingprepared, Isacks and Sykeswere in the process of makinga specialstudy of focalmechanisms of deepshocks, and prior to publicationof that study (Isackset al., 1969) they reporteda summary of the resultsin the NGT paper.The resultswere somewhatunexpected but, as usual, seemedcompatible with the down-goingslab model. The deep earthquakes,it turned out, were not shearingmotions corresponding to the movementof the slab through or past the asthenosphereas some had expected.Instead they correspondedto stressesacting within the lithospheric slab.Either the axis of maximum compressivestress or the axis of leastcompressive stress was orienteddown the dip of the seismiczone and hencedown the dip of the slab(NGT, Fig.11). Furthermore,the spatialpattern of focalmechanisms, and consequently stressorientation, varied from arc to arc. In somearcs, where the 86 Scienceby Synthesis zone penetratedto greatdepths, i.e. sevenhundred kilometers,down-dip compressivestresses seemed to dominateat all depths.For otherarcs, intermediate depth shocksseemed to correspondto down-dip extensional stresses.Later, and after the time of the NGT paper, Isacksand Molnar (1969) would show that the focal mechanismsof deep and intermediate depth shocksvaried within arcsand from arc to arc in sucha manneras to supportthe idea that lithosphericslabs were sinkingfreely until the leading edge of the slabeventually sank to a depth where it encounteredresis- tanceto further sinking.Thus stresseswithin slabsdiffered depending on whetherthe particularslab was sinkingfreely or being impededby resis- tanceat depth.This subjectwould becomemore complex as it becameclear that contortionsof the slab and other factorsaffect the seismicactivity as well, but no data that representa seriouschallenge to the slabmodel have so far been reported. The NGT paper presenteda figure (NGT Fig. 14) to illustrateschemat- ically four ideasabout the stateof a downgoingslab including (a) a sinking or underthrustingslab freely penetratingthe asthenosphere,(b) a slab encounteringresistance after it penetratedto some great depth in the asthenosphere(c) a slabbeing assimulated at depth into the asthenosphere and (d) a sinkingslab broken by extensionso that a detachedpiece preced- ed the main slab. This figure was intended to illustrate hypothesesthat might provide the basisfor further research.So far as I know,25 yearslater all four are still viablehypotheses and all are being investigatedcurrently alongwith someother variationsconceived more recently. For the mid-oceanridges, the sourcesof new surfacematerial, the NGT paper of coursereviewed Sykes' by then well-knownresults described in a previouschapter and basedon patternsof seismicityand focalmechanisms at the ridgesand fracturezones. It also drew attentionto the tendencyfor swarmsof earthquakes,perhaps related to volcanicactivity, to occuralong the ridges. Modest attention was devoted to detailed study of certain regionssuch as the Gulf of Aden and the Gulf of Californiawhere the seismic data provided valuableinformation bearing on the tectonicsand the genesisof thosepull-apart features(NGT, Figs.5 & 6). Elsewherein the paperthere are numerousreferences to one or another kind of seismicevidence bearing on particular geologicfeatures or regionssuch as the Alps, or the SanAndreas Fault, or SouthernChile, or the Gulf of California.In a sense,these references to specificselected geological featuresof regionalscale were but a harbingerof thingsto come,for in the light of the new plate tectonicsmodel all featuresof the earthof such scalewould soon come under re-examinationby investigatorsscattered throughoutthe world. In anothersense, the referenceswere a sign, Scienceby Synthesis 87 as they did from seismologistswho had limited experienceand expertisein geologicalstudies of thoseregions, that a new framework for understanding and integratinggeology on a regionalas well as a globalscale had suddenly becomeavailable. It would turn out that not only seismologistsbut almostevery earth scientistwould experiencean expansionin breadth of interestand understandingwell beyondthat of the individual'sparticular specialtyas a consequenceof the power of unificationof the plate tectonics concept. Not all of the attentionin the NGT paper was directedtoward the influenceof seismologicalevidence on problemsof globaldynamics or regional tectonics.The NGT paper also included a speculativesection on possible effectsof the new globaltectonics on the field of seismology.Probably most importantof all in this regardwas the new and dramaticallydifferent per- spectiveon earthdynamics that plate tectonicsengendered. Also important was the early demonstrationof how seismicactivity relatedin surprising detail to the new plate model. Another importantchange, both technicallyand psychologically,was the departurefrom the spherically-symmetric,layered earth model of deeper portionsof earth. Sphericalsymmetry had never been claimedby anyone for the thin and heterogenoussurficial crustal layer, of course.But the entire earth from directlybeneath that crustallayer to the very centerhad previouslybeen visualizedfrom the classicallylayered-onion perspective of sphericalsymmetry. The new plate tectonicsmodel extendedthe asym- metry of that near-surfacecrustal layer to the depthsof the deepestearthquakes,almost seven hundred kilometers, and suggestedthe possibilityof still deeperasymmetry as a consequenceof slabpenetration. And finally it was forecastthat the new earth dynamicswould integrate seismology much more closelywith many of the disciplinesof geologythan had been the case. Sciencehas a tendencyto fragment into specialtiesduring intervals betweentimes of majoradvance, the intervalsof what Kuhn calledproblem solving,and then pull itself togetherwhen a new framework for integra- tion, or what Kuhn called a new paradigm, is discovered.Plate tectonics was clearlya majornew paradigmin that sense,and the whole of earthsci- encebecame ripe for integrationwhen it becameviable. It is, of course, hard to measurequantitatively just how plate tectonicsaffected the overall activitiesof any one individual or any particulargroup of earth scientists, but thereis an abundanceof ancedotalevidence to supportthe claim that the thoughtsand activitiesof nearly every earth scientist,seismologists of courseincluded, were affectedin somevery substantiveway by the coming of plate 88 Scienceby Synthesis

Finally,I would like to describethe circumstancesand goings-onbehind a few selectedportions of the NGT paperin an effortto conveynot somuch the scienceof the momentbut ratherthe magicand excitementthat seemsto pervadethe air when scientistscome across a new way to explainvast quantities of diverseobservations, many of a kind that had receivedlittle or no prior attentionor noticefrom thosescientists, or perhapsany scientists. For example,with the underthrustingof mobilelithosphere in arcsas a conceptin hand, and with LePichon'smodel of relativeplate motionsas a spur,for the NGT paperwe plotteddown-dip lengthsof deep earthquake zonesagainst LePichon's theoretical convergence rates for variousarcs. The resultinggraph (NGT, Fig. 16) showeda clearlinear correlation for the data of most arcs,with the slope correspondingto a time of about ten million years.That time couldhave been the durationof the mostrecent interval of seafloorspreading if spreadingwere intermittent. Intriguingly, the time wasjust that suggestedby Ewing and Ewing in their studyof the variation of thicknessof sea floor sedimentswith distancefrom a spreadingcenter. Or the interval could have been the time for assimilation of the slab into the mantleto the point where it could no longersustain earthquakes. The importantpoint for this discussion,though, is that at that particular time it was not only possiblebut alsoincredibly easy to make a simple little plot by hand on a singlesheet of graph paper and so discoversome- thing fundamentalabout how the entire earth had organizeditself with regardto deep earthquakeactivity. Just a shorttime earlierno one would have dared to dream that sucha simple schemeof organizationprevailed for that phenomenon.Just a shorttime, perhapsa few years,later almostall of the simplerelations like that had alreadybeen revealed. For thosewho would discover,being there at the right time was indeedimportant. In a similar example,Bryan and I took my desktopglobe, shaded the geographicallocations of deep seismicactivity, measured the shadedareas with a planimeter,corrected for dip, divided the total by the lengthof the world rift systemand the ten million year time constant,and so got a tentative figure for the averageannual rate of sea floor spreadingalong the entire rift system.The half-velocity was 1.3 cm/yr. What that number meantwas subjectto the assumptionswe had made of course,but the very fact that we could casuallypick up a nearby globe,do a few very unso- phisticatedthings, and come out in an hour or two with an answerthat seem reasonableand consistentwith other data such as spreadingrates basedon magneticanomalies, left us with the euphoricfeeling that we really were on the right track to understandingthe earth. Searchingfor yet anotherquick and hitherto unanticipatedconfirma- tion of the plate model, I recruiteda new Lamont graduatestudent, Scienceby Synthesis 89

Molnar, to make a global study of P and especiallyS waves propagating through the uppermostmantle, the shallow part of the earth that is normally a part of the lithosphere.The studywasn't quite soquick and easyas the two just described,for some1500 seismograms had to be examined,but Pete, who would becomean outstandingearth scientist,found that, sure enough,where the modelcalled for a single,unbroken plate, high frequency seismicwaves propagatedwell (Molnar and Oliver, 1969). Where the model calledfor a plate boundary,high frequencywaves were attenuated. The resultswere consistentwith the plate model everywherethat a test couldbe made. The plate modelhad producedan explanationfor why literally thousandsof earthquakeseismograms, filling the shelvesof observa- toriesaround the world, for somepaths had one appearance,for somedif- ferentpaths another. Testing the predictionsof a theory is a commonway to testthe theory.Plate tectonics successfully passed this particular test, and many others,with little difficulty.In so doing,it invariablyleft thosefortu- nate enoughto be at the right placeat the right time brimmingwith the joy of discovery. From the foregoing,it shouldbe clearthat the NGT paper and related studieswere transforminga part of earth scienceand makingthe subjectof seismotectonicsinto a fertile field for further study.When Lynn, Bryan and I wrote the NGT paper we hoped, of course,that the paper would have suchan effectand that it would becomeone of the basicpapers of the science,but we had to weathera period of someanxiety while we awaitedthe responseof the scientificcommunity to the paper. Like most scientists,I have never been one to put much stock in omens or mystical signs. However, when, during this period, the Lamont publications office assignedto the NGT paper the contributionnumber 1234,I almostbecame a numerologist! Now let us turn from the NGT paper and contemporaryactivities and eventsand summarizewhat happenedduring the next few yearsas others joined and contributedto what was becomingthe plate tectonicsbandwagon, and plate tectonicsbecame a well-rounded,widely-known, com- prehensivetheory of earth 9

Full FledgedPlate Tectonics

In therealm of thenew plate tectonics Seismologyfunctions as phonics Whilepaleomagnetical data And geologicalstrata Are simplytwo forms of mnemonics.

national meeting each year at that time) probably best marks the hemomentnational when meeting the newsofAGU and in the the spiritspring of theof 1967 upcoming (therewasearth only scienceone revolutionbegan to reacha substantialfraction of the earth sciencecommunity. Partly as a consequenceof that meeting,the next few years, running throughthe remainderof the 1960sand the early 1970s,would seea spateof contributionsfrom scientistsin a variety of disciplinesthat had not previouslybeen much heard from on the subject.The combinedforce of evidence from that diversityof sourcesin supportof the new earth dynamicsgrew rapidlyand soonbecame irresistible to all but the mostdie-hard skeptics. By the end of that period,the conceptof plate tectonicswas firmly established and knownby that nameto mostearth scientists, and to much of the restof the world as well. Scientistsin other fields and interestedlaymen quickly learnedabout and were fascinatedby the geologicalrevolution. The subjectwas popularbecause it was easyto graspthe principlesand the simplebeauty of plate tectonics,and becausecertain of the effectssuch as earthquakes,volcanoes and mountain-buildingwere topics of long standinghuman concernand fascination.Intensive research stimulated by the new paradigmcontinued beyond the burst of innovativecontributions of the late 60s and early 70s and continuestoday. But now plate tectonics reignsas the establishedand almostunchallenged paradigm of solid earth dynamicsas it hasfor some25 years.Of course,it is possiblethat plate tectonicswill be supersededby a still bettermodel of globaldynamics at some time in the future, and of courseimprovements in the early model have been made in that quarter of a century,but in my judgment,beyond sug- 92 Full FledgedPlate Tectonics gestionsfor additionalmodest improvement of the basicmodel, there is no signon the horizonat presentof a still betterand basicallydifferent model. As work relatedto platetectonics spread through many disciplines,and publicationson the topicproliferated, it becamea near-impossibilityfor one individual to monitor or to summarizewhat took placein every specialty, and I make no attemptto do so thoroughlyhere. However, to give the fla- vor of what was happeningand to set a backgroundfor somecomments aboutthe developmentsin seismologythat followedthe early seismological contributionsdiscussed in precedingchapters, I try hereto summarizesome advancesin other disciplinesand to cite a few contributionsthat might be thoughtof as milestones.I do this with sometrepidation, knowing that this brief summarywill surelybe incompleteand that somefriends and col- leagueswill be rightfullydisappointed because I havefailed to includetheir favoritecontributions. To them, I apologize.The uncomfortablenature of thissituation for the authoris perhapsone good argument for the writing of suchhistories by historianslaboring only well after the participantshave disappearedfrom the scene.Those historians might not get it completely right either,but objectionswill not comefrom eyewitnesses. Let us beginthe summariesby focusingon synthesesof globalscale. In some specialties,the opportunityfor global synthesiswas unparalleled. The geomagneticianswere ableto surveyand then correlateand datemag- neticanomalies over largefractions of the oceanicareas of the earthand, as the historyof reversalsof the magneticfield was extendedback in time and refined, to determine the age of the oceaniccrust almost everywhere beneaththe deep sea.The task requiredsome modest effort of course,but it was incredibly simple comparedto the prodigiouseffort that innumer- able geologistshad contributedcollectively over more than a century as they mapped the agesof the continentalrocks that were lessthan half the area of the ocean basins. The oceaniccrust was all young, of course,less than about 200 million yearsold, and the mapsof oceaniccrust showed how the agesof the rocks of the seafloor increased with distancefrom the divergingplate boundaries and how it appearedthat crustwas disappearingat the convergentplate boundaries.The geomagneticianswere challengedfor a time by skeptics who questionedthe correlationsof the anomaliesfrom one ship'strack to an adjoiningbut distantneighbor, but, as the data gapswere filled in, they, and almostall remainingskeptics of the plate tectonicsmodel itself,had to succumbeventually and join what by then had becomethe almost irre- sistableplate tectonicsbandwagon. When drilling of the deep seafloor commencedunder the JOIDESpro- gram,some of the very earliestresults from drilling showedthat the age Full FledgedPlate Tectonics 93 the seafloor, as determinedfrom paleontologicaldating of the sediments immediately overlying the crustalrocks, was just that predictedby the magnetic anomaly-basedstudies. Once drilling data were obtained at a numberof placesand for a rangeof differentages, and shownto be consistent with the magneticanalyses, those combined data becameperhaps the strongestconfirmatory evidence yet for the validity of the plate tectonics model,or at leastthe sea-floorspreading component of it. They convinced all but the most recalicitrantskeptics. For me, having oncewatched fellow graduatestudents in the early 1950slabor endlesslyand fruitlesslyover the early and perplexingscanty informationon what seemedalmost randomly scatteredanomalies of the magneticfield at sea,it was astonishingthat thosemagnetic data had suddenly fallen into placeso beautifullyand had revealedthe agesof the sea floor everywhere.But it was alsoencouraging because that revelationsug- gestedthat other confoundingpuzzles of earth sciencemight ultimately be resolvedin similar fashionand with similar enlighteningconsequences at some time in the future. Also on a globalscale, and oncethe basicpattern of the platesand their movementwas postulated,the spatial patternsof a variety of other phe- nomenabegan to make sense.The configurationof earthquakebelts, the spatialdistribution of shallow and deep shocksand of the maximum size of shocksin a particulararea, as noted earlier,were just a start. Physiographicfeatures fell into place.The oceanicridges and fracture zoneswere characteristicof the divergingplate marginsat sea,and the rift valleyscharacteristic of divergingplate boundarieson land. Island arcs,or just arcs,and deep sea trenchesoccurred at the convergingmargins. The global distributionof volcanoesmade sense,those with explosiveacidic volcanismwere located at the convergentmargins where exotic surface material was carried down into the mantle, thoseproducing more basic rockswere at the divergentmargins. That left certainvolcanic island and sea mount chains unexplained temporarily, but Wilson, among others, demonstratedthe regularprogression of agesin suchchains and attributed it to passageof a lithosphericplate over a "hotspot"in the asthenosphere, each island being formed through volcanism just as it was above the hotspot. What caused a "hotspot" was a mystery for awhile, but Morgan accountedfor the hotspot phenomenonas a consequenceof plumes of mantle material rising from deep in the earth to a region near the surface beneatheach hotspot. Other geophysicistsbuilt upon this idea and tried to use hotspotsas referencepoints to determine motion of the plates relative to an axisfixed in the interior and not simply relativeto another,and 94 Full FledgedPlate Tectonics moving,plate. The detailednature of plumesis still controversialbut the conceptseems a reasonablebase for a plausiblemodel. Somescientists showed how variouskinds of geophysicalobservations thatwere not a part of the early storyfit the modeland providednew informationon earth dynamics.Early measurementsof heat flow throughthe seafloor pioneeredby Bullard at Cambridgeand Revelleand Maxwell at Scrppswere shownto be more or lesscompatible with seafloor spreading, althoughsome doubt was castlater on the interpretationof someof these observationsbecause of suspectedconvective circulation of fluids within the sediments and crustal rocks. Sclater and Francheteau (1970) and others showedhow oceanicrocks, formed hot at the spreadingcenters, cooled and contractedas they moved away from the center.The earthmaintained iso- staticequilibrium so, as the rockscooled, the averagedepth of the ocean increasedin a regularand predictableway. The measurementsfit the theory and thisstudy was widely recognizedas one kind of confirmationof the plate tectonicsmodel. Also, to my chagrin,the calculationsexplained the obscureresults I had obtainedon depthsof the oceaniccrust from the surface wave studies in my thesis describedearlier, and then blissfully ignored.Fortunately, no one noticed. In the 1920s,an ingeniousDutch geophysicistnamed VeningMeinesz beganto developwhat would becomea reliablemeans for measuringgrav- ity at seato a precisionof a few partsper million. It was basedon swinging pendulumsmounted in submarinesthat submergedto depthduring observation so that accelerations due to water waves were avoided. After World War II, observationsof gravity proliferated throughout the oceansas opportunity for scientificobservations on submarinesexpanded. Lamont, led by Worzel, Talwani,Drake and othersoperated such a program.The most strikingresult of observationof gravity at seawas the revealingof largenegative gravity anomaliesassociated with the deep seatrenches of island arcs.They indicateda deficiencyof masslarger than that missing from the trench.Vening Meinesz (1952) tried to accountfor the deficiency as a consequenceof a symmetricdown-buckling of low-density plastic mantle beneaththe trench due to lateral compressionnormal to the arc. Othersclaimed that the deficiencywas due not to compressionbut to the majorextension that alsoproduced the topographictrench. Still othersfelt the negativeanomaly could be explainedby a large accumulationof low densitysediments beneath part of the trench.The latter explanationwas moreor lesscorrect, but was only a part of the storythat eventuallywould be revealed. When, as describedearlier, it was proposedon other groundsthat a trenchis a prominentfeature of an arc as a consequenceof a Full FledgedPlate Tectonics 95 and presumablyhigh density,slab of lithosphere,that new model seemed incompatiblenot only with the tectonicmodels that the interpretersof gravity data alone had proposedbut also with just the raw gravity data alone. How, it was asked,could adding a large quantity of high density material to a regionin the form of a deep slabproduce a negativerather than a positiveanomaly? Eventually the matter was resolvedwhen it was recognizedthat the narrow negative anomaly was superimposedon a much broader and hence less apparent positive anomaly. The positive anomaly spreadover almostthe entire region of the arc and was attribut- able to the high density,and mostly deep, slab. The narrower and more prominent negative anomaly near the trench was interpretedas a consequenceof the shallow accretionarywedge of sedimentsscraped from the top of the slaband plasteredagainst the inner wall of the trenchas the slab was subductedand descended.So, after someearly and misplacedconcern, the .... • gravity ..... l•,• ass,•,-i•,•l with the deep •o• •o,•hos •^,oro fully reconciledwith the down-goingslab model, and gravityjoined the list of disciplinesproviding support for the plate tectonicsconcept. Lurking behindthe scenesas the plate tectonicsmodel evolvedwas the suspicionthat the whole thing wouldn't work simplybecause the rheology of earth materialsjust wouldn't permit the earth to behavein the matter suggested.That view had a long standingbase of supportbecause Harold Jeffreys,the extraordinaryBritish geophysicist with a seriesof major scientific achievementsto his credit, had opposedWegener's theory of continental drift on those grounds,and he stood firm on his argumentsand opposedplate tectonicson similargrounds. When Holmesin the 1930s,and otherslater, had proposedthe idea of huge convectioncells in the mantle, they had done so in very sketchyfashion and with no firm quantitative basis. In 1969,Don Turcotte,an aerospaceengineer from Cornellturned geo- physicistwhile on sabbaticalleave at Oxford,and Ron Oxburgh,a geologist there, publisheda quantitativestudy that showedthat convectionin the mantle as postulatedfor the plate tectonicsmodel was possibleand likely. That study (Turcotteand Oxburgh, 1968) seemedto satisfymany of the skepticsand set the stagefor additional, more detailed, studiesby them and others.Just how flow in the interior takesplace to accommodateor to drive the movementof the platesremains a somewhatcontroversial subject at this writing however,but that someappropriate kind of movementcan occurseems no longerin doubt as a result of suchstudies and of the com- plementaryobservations based on glacialrebound and otherphenomena. The earlieststages of the plate tectonicsrevolution were basedprimar- ily on Euler's simpletheorem of sphericalgeometry, on evidencefrom 96 Full FledgedPlate Tectonics disciplinesof geophysics,geo/paleomagnetism and seismology,and on evidencefrom marine geology.Evidence from more conventionalland- basedgeology, except as generalbackground information, was not a key part of the early action.Of course,that had to changeif the new theoryof earth dynamicswas to be truly global,because land-based geology consti- tutesa major and very significantfraction of the observationalevidence of earthscience. A globaltheory had to be compatiblewith that importantevi- dence.And changeit did, at first largelythrough the effortsof JohnDewey and JackBird who, in a remarkablesynthesis entitled "Mountain Belts and The New Global Tectonics"published in JGRof May 1970,showed how, and what kinds of, geologicevidence could be used to demonstratehow mountainbelts around the world were productsof plate evolution,revealing in the processnew informationabout the historiesof mountainbelts and the processesthat formedthem. The Dewey and Bird paperwas widely read and very influential.Skeptics who had scoffedat nascentplate tec- tonicsbecause it was "all geophysicswith no geology"were forcedto mod- ify their positionsor confronta risingtide of pro-platetectonics geologists. In spite of their use of the term "new global tectonics"in the title, which was apparently designed to parallel "Seismologyand the New Global Tectonics,"Dewey and Bird often used the equivalentterm "plate tectonics"in the text. Sofar asI know,that was oneof the first papersto use that now widely-acceptedterm. Justhow and where the term "plate tectonics"originated seems not widely-known.I have evenbeen erroneously creditedwith originatingit myself.It is my impressionthat the term was not invented by someonewell-known for early work on the subject.It seemsto have been used sporadicallyat first until eventually,and rather suddenly,it becamethe acceptedname for the subject.In any case"plate" is probablya better term than "thin lithosphericblock," or even "tabular block" or some other variant, so "plate tectonics"it now is, everywhere from a modernnewspaper article on a destructiveearthquake to a text on the historyor philosophyof scienceto a highly technicalarticle in the best of scientificjournals. Other geologistspushed the subjectfurther and accountedfor once enigmaticfeatures of continentaland oceanicgeology under the new paradigm. The term "subduction"was born and understandingof the processes associatedwith a down-goinglithospheric slab refined. Ophiolites were identified as onceparts of the seafloor broughtto the near-surfaceof continentsas a result of convergentplate margin processes.Petrologists and geochemistsentered the sceneas a paper by RobertKay and others(1970) showedhow the rocksof the oceaniccrust were formed through partial melting of the mantle.A long controversybegan over how water and FullFledged Plate Tectonics 97 mentssubducted by the down-goingslab of lithospherein islandarcs influencedthe compositionand generationof magmasand lavas,but the basic elementsof the subductionmodel have withstoodchallenge. Island arcsgot other kinds of attentionas well. Karig (1970),in a study of oceanfloor relief and sedimentdistribution for the region behind the Tonga-Kermadectrench, showed that the evolutionof an arc, a convergent and henceone might think mostly a compressionalfeature, could include extensionin parts of the system,resulting in an inter-arcbasin behind the main volcanicarc. What at first seemeda contradictionto the large scale view of arcs as convergentfeatures was reconciledwith that view by appealingto secondarypatterns in the flow of the asthenosphereas it acco- modatedthe penetratinglithosphere and alsomoved magma to volcanoes at the surface.That secondaryflow actedon the surfacerocks so as to produce local extension behind the main arc. Minear, Toksozand othersmade quantitativenumerical studies of the thermalhistory of the down-goingslab of lithosphereand its surroundings and showed, as earlier intuitive suggestionsand crude calculationshad suggested,that for a reasonableconvergence rate at least part of the slab remainedcooler than adjacentasthenosphere to depths of nearly 700 km. Whether the slablost its identity completelyat that depth would becomea matter of continuingcontroversy but the conceptof the slabas a body that maintainslower temperaturefor a long period seemsfirmly established. The enormousjob of trying to recreate,with plate tectonicsas the basic paradigm, the geologicalhistories of rocksand rock masseson all scales includingthat of the continentsbegan. For the recentpast, and going back to the splitting of the supercontinentPangaea about two hundred million yearsago, the taskwas not overly difficult, for abundantmagnetic data on the age of the oceanfloor, the geometricconfigurations of shorelines,and the fit of rock formationsacross areas once intact but now brokenapart and separatedby spreadingoceans could be used.In fact, many early conclusions about jigsaw-like fits of continentsby Wegener and other early drifterswere shown to be compatiblewith the new ideasof plate dynamics.One particularlyimportant and revealingadvance came with the recognition that large segmentsof the continentswere a consequenceof the accretionat various times of many smaller terranesthat had been trans- ported separatelyacross long distancesand then plasteredagainst a continental mass. For, pre-Pangaeantime, data are more sparse.No old ocean floor remainsexcept perhaps beneath a few inland seassuch as the Caspian or where it has been thrust on to continentalcrust; it has long sincebeen sub- ducted.Geologic data on older rockssince reworked and partially 98 Full FledgedPlate Tectonics ated by more recentevents must be piecedtogether to provide a segment of historythat is plausibleand consistentwith otherparts of globalgeolog- ical historyfor all time. This subject,of course,is still beingdeveloped and has practicalas well as purely scientificimportance. The reconstructions that result are often useful, for example,in explorationfor mineralsand hydrocarbons.On the lesstechnical side, the geologichistory of earth in toto prevails as solid earth science'smost important contributionto the intellectualsphere as humansstrive to understandhumans and the meaning of human existenceon earth and in the universe. With that very brief and sketchyaccount of the accomplishmentsand goalsof earthscientists as the effectsof thenew paradigmwere felt throughout the scienceas background, let us turn now to somedevelopments in seismologythat cameabout because of the flourishingof the conceptof platetectonics.Bear in mind that, in the 1970s,the scopeand diversityof the activi- tiesin platetectonics-related seismology, and the numbersof scientistsactive in the subject,were alreadymuch greater than they had beenin the mid and late-1960s.Plate tectonicswas gettingincreasing attention, not only within earth sciencecircles, but throughoutmuch of the restof the scientificcommunity and parts of the non-scientificcommunity as well. It had become newsworthyand scientistswho were alreadyinvolved were calledupon to spreadthe news. Scientistswho publishpapers with overlappingcontent in differentsci- entificjournals are often subjected to criticismand discouraged from doing so. However,as a resultof the NGT paperin JGR,Lyrm, Bryan and I were invited and encouragedby editorsand othersto write updatedversions of that paperfor thejournals of otherscientific societies and variousother less tech- nicalpublications. Invitations to lectureon the subjectwere frequentand they camefrom a wide varietyof organizations.Never knownfor the charismatic presentationof an outstandingpopularizer of science,nevertheless I spoke on invitationto audiencesat colleges,universities, local chapter meetings of pro- fessionalsocieties, international scientific meetings, committee meetings, large corporations,museums, Rotary clubs, and an off-beatquasi-religious institution. I oncegave consecutive lectures to an audienceof BoyScouts and then to a groupof scientistsat theNational Academy of Sciencesand used the same setof slides!The textwas worded differently in eachof the abovecases but the basiccontent of all the lectureswas the same-theessentials of plate tectonics and the relationof that new conceptto seismology. For those readers who are not familiar with custom in scientific circles and who may have the mistakenimpression that suchlecturing is a finan- cial bonanzafor the lecturer,I hastento point out that, with the exception of a coupleof talks at industrial laboratories,those science lectures Full FledgedPlate Tectonics 99 providedas a publicservice by my universityand resultedin no feeswhat- soeverfor the lecturerother than travel expenses! The pressof speakingengagements and writing activitiesthat followed the NGT paper, and the press of additional administrativeduties as I becamechairman of Columbia'sgeology department forced me to curtail my researchsomewhat during the late 1960s.Meanwhile the NGT paper openednew opportunitiesfor the authorsand, in 1971,I left Columbia and Lamontto becomechairman of the geologydepartment at Cornell where therewas a rare opportunityto rebuild a departmentof a major research universityaround a new focus.Bryan Isacks, Muawia Barazangiand George Hade also moved from Columbiato Cornell, and shortlythereafter Jack Bird,Dan Karig and BobKay, all knownfor theirearly contributions to plate tectonics,joined the Cornell department.Don Turcotte,already at Cornell, transferredfrom aerospaceengineering to geologyand the departmenttook on a forward-looking,plate tectonicsflavor that made it stand out from departmentselsewhere that were not in a positionto undertakesuch major rebuilding.The reorienteddepartment attracted many fine graduatestu- dentsand developeda varietyof importantresearch projects, some of which are discussedin the following.At Cornell during the 1970s,with Sidney Kaufmanand Larry Brown,I initiatedthe COCORPproject which is based on the kind of strategythat was successful in bringingabout plate tectonics. COCORPis discussedin moredetail in the next chapter. Lynn Sykesmoved into the facultyposition that I vacatedat Columbia and,working with variousstudents and postdocs, produced a seriesof fine paperson seismotectonics.Some followed the themeof someof his previous work but were appliedto partsof the world previouslyunstudied in this manner,and somebroke new ground.Among other things,Lynn became interestedin earthquakeprediction, and his backgroundin plate tectonics helpedin the developmentof the seismicgap theoryof earthquakeprediction. The seismicgap theoryis basedon the idea that, as lithosphericplates convergeat, say,an islandarc, the intermittentmotion along segments of the arcas representedby individualshallow thrusting earthquakes must eventually even out alongthe entirearc. Thus that segmentof the arc that has gonelongest without havinga majorshock is the part of the arc mostlikely to havethe next one. This concept has enjoyed some success, and maybe the most reliablemethod for predictionat the presenttime, but neither it nor any othermethod yet hasthe capabilityof providingin advancethe precise time, location,and size of an impendingearthquake. That goal seismologists,and the inhabitantsof earthquake-proneareas, continue to seek. After the NGT paper,Lynn turned a part of his attentionto the seismological aspectsof the comprehensivenuclear test ban treaty, still 100 Full FledgedPlate Tectonics

, activelypursued in internationaldiplomatic circles, and becameone of the moreprominent seismologists in thatactivity in whichplate tectonics plays a modestrole, thoughnot a decisiveone. BryanIsacks, at the time of the NGT paper alreadymore heavily into the studyof arcsand their deepearthquake zones than mostanyone else, continuedto probethat subjectin the followingyears. Working with Sykes, Molnar, Barazangiand others,and as mentionedearlier, he showedhow the spatialpatterns of focalmechanisms varied from arc to arc depending upon the maximum depth and extentof the deep earthquakezone. The observationsindicated that the slabswere sinking,as a consequenceof their high density,until they eventuallyencountered resistance at depth. Contortionsin the deep earthquakezones apparently indicated deformation of the slabat depth,perhaps as a consequenceof asthenosphericflow or resistanceto settling. The Tonga-Fiji project moved to Cornell with Bryan and me. Eventually,under Bryan's leadership, it first shiftedwest of Fiji to the New Hebridesarc where strain,tilt and levelingmeasurements as well as seismic observations were made, and then to the Andes of South America, anotherconvergent plate structurebut in this caseone involvingthe margin of a continent.All of the studiesunder this projectprovided information of higherresolution on the configurationand dynamicsof down-going slabsin thesearcs, but revealedno evidenceincompatible with the basic subductionmodel of plate tectonics.Isacks' work in the Andesresulted in much improved understandingof the formationof the centralAndean Plateauand the Bolivianorocline. Recently, the projecthas evolved so as to revealthe roleof longterm erosion of topographyon the tectonicsof mountain belts.Surprisingly, it is turning out that the prolongedeffects on sur- facerelief that resultfrom non-uniform spatial patterns of precipitationand erosionin mountainousregions can, when integratedover millenia,be a factorin geodynamics.Climate, in otherwords, may influencethe way in which the tectonicplates move becausemasses at the surfaceare slowly redistributedthrough erosion. Seismologicalresearch of the 1970srelated to platetectonics took place againsta backgroundof activitymuch different from that of the mid 1960s. Seismologists,and earth scientistsof all kinds, almosteverywhere were enteringthe competitionto makenew contributionsto the understanding of tectonicson a local,regional, and globalscale. Studies and information proliferatedand indoctrinationinto the subjectwas no longera simplematter of learninga few concepts,a little marine geology,and the latestnews from geomagnetism.Innovative ideas about the consequencesof platetec- tonicsand testsof themwere springing up in mostof the specialtiesof FullFledged Plate Tectonics 101 science.Doing somethingnew and informativebecame easier in one sense as topicsnot yet probedunder the plate tectonicsparadigm caught attention, but quickly becameharder in another senseas the most obvious opportunitieswere seizedupon by a crowd of investigators.And therewas no longerthe smellof an undiscoveredparadigm in the air; ratherthe 1970s marked the early phaseof problem-solvingunder a new paradigm. Seismologists,including especiallythose bypassedor left out of the excitementof the 1960s,were aggressiveabout developingnew topicsfor study,or new angleson old topics.The island arcsand deep earthquake zonescontinued to receiveseismological attention. For example,it was proposedthat the down-goingslab had achievedits shapeas a consequenceof successiveshearing of portionsof the slab as it descended.That concept, however, turned out to be inconsistent with the data on focal mechanisms. It wasproposed that prevailingcurrents of globalscale in the asthenosphere resultedin consistentlygreater dips in slabsdipping east than in thosedip- ping west,the evidenceon slabdips coming,of course,from the configuration of the deep seismiczones. That matteris not yet fully resolvedand the patternof asthenosphericflow continuesto be a controversialtopic. It was suggestedthat anomaliesin seismicwave propagationcould be detectedat depthsgreater than that of the deepestearthquakes, indicating that the slab,though aseismicin its lower reaches,had neverthelesspene- trated to thosegreater depths. Or perhaps,it was suggested,the leading portionof the slabhad descendedto a level of resistance,or buoyancy,and then drifted horizontallyaway from the deep earthquakezone. It was also suggestedthat detached,aseismic portions of the descendingslabs sink throughthe entiremantle to its baseat the core-mantleboundary. The core- mantleboundary is alsothought by someto be the sourceof the plumes that rise to the surfacecreating hotspots and other effects.Such specula- tions, though often the crux of unresolvedcontroversy, have stimulated seismologiststo gatherevidence on the existenceand configurationof het- erogeneitiesin the mantle. Currently tomographictechniques are used to seek mantle hetero- geneities,and comprehensivestudies of the interiorhave beenmade using statisticaltechniques and huge quantitiesof data. They seem to reveal zonesof higher than normal and lower than normal velocitiesand, based on their spatialdistribution, the anomalieshave been associatedwith phenomenaof the interior related to plate dynamics.However, the marginal signalto noise ratios of the data and various other factorshave left some lingeringdoubts about the reliabilityof the results. Substantialseismological effort has gone into determinationof the thicknessof the lithosphere,a parameterthat probablyvaries 102 Full FledgedPlate Tectonics with location. Some have found the lithosphericthickness to be much greaterthan average,perhaps several hundreds of km thick, beneaththe continents.Others dispute this claim and find a thinner lithospherethere. Suchcontroversies are oftenclouded by ambiguityin definitionof the term lithosphereand by disagreementover just what certainobservations tell us about the lithosphere.At one time or anotherthe lithospheremight be defined as the layer of strength,the layer of efficientpropagation of high frequencyP and S waves,the layer of high velocity overlyingthe mantle low velocitylayer, or the layer abovea particularreflecting horizon, all this in the face of the distinctpossibility that the boundary is a gradational rather than a sharp and distinctone. The caseis furthercomplicated by evidencethat continentalearthquakes mostlyoccur within only the upper 20 km or so of the crust,i.e. only the uppermost part of what we typically designate as the lithosphere. Presumablythat is the brittle portion and it overliesa more ductilelower crust and mantle. The nature and configurationof the lithospherethus remaincontroversial topics and onesbeing actively investigated. There is no seriousobjection at the presenttime, however, to the generalconcept of large thinplates of lithospherecapable of transmittingstresses over long distances. Recentmaps of stressesin the crustsupport this statementby revealingpat- terns of stress that are consistent over continent-scale dimensions. Anotherphase of earthquakeseismology that has developedsince the 1960sis the determinationof anisotropyof rocksat variousdepths in the earth.Such studies are based particularly on observationsof splittingof the shear wave. Some assumptionsare involved, but it may be that the anisotropymeasured is a consequenceof flow that alignscertain crystals. If so,there is potentiallya way to measureflow, present or past,in theasthenos- phereand perhapsresolve some major questions of earthdynamics. Almostall of the post 1960sefforts and advancesnoted in the last few paragraphsbecame feasible on a reasonablescale only becauseof the adventof seismographnetworks producing data that canbe readilymanip- ulated in digital form. Digital recordingand relateddevelopments in com- putingtechniques have brought the field of seismologythe potentialto do things oncethought beyondreach or onceunimagined. So far, however, although the capacityto manipulate and analyze data has been much enhancdby the digital revolution,the new techniqueshave not yet produceda conceptualrevolution in the science.The changehas so far been more in style and thoroughnessthan content.The conceptualrevolution may not be far off, however. Finally, seismologistsusing earthquakesas sourcesof seismicwaves havenot beenalone in seismologicalefforts to testand build uponthe FullFledged Plate Tectonics 103 versionof plate tectonics.Artificial sourceshave alsobeen much used,par- ticularlyby marine geophysicistsand industrialgeophysicists, to produce spectacularseismic reflection profiles that reveal informationparticularly aboutsedimentary basins on land and at sea.Of all geophysicaltechniques, it is the reflectiontechnique that producesby far the highestresolution and most detailedinformation on the interior.In general,the new information on the sedimentarylayers tends to fit well and complementthe new tectonics.For example,the structureof the accretionarywedges of sediments at the innerwalls of trencheshas been resolved in strikingdetail and seems to fit well the plate tectonicsstory of sedimentsscraped there from the down-goingslab. The major decollementbeneath the wedge, and separat- ing it from the slab, has been tracedby seismicreflection profiling to con- siderabledepths, and it appearsjust where, and as, the model predicted. The sedimentarybasins that are the habitat of petroleumhave becomebet- ter understood based on the relation between the new seismic data and the plate tectonicsmodel, and the basinshave been organizedinto a classifica- tion scheme based on the model. Most of the seismic reflection effort is directedtoward the searchfor hydrocarbons,but the searchfor certainmin- eralsis now beginningto rely more on seismicreflection profiling and on interpretationsbased on plate tectonics. So,in contrastto the days of turmoil and doubt of the 1960s,plate tectonicswith its componentsof sea floor spreading,subduction, and transform faulting has comeof age as the prime paradigm of earth dynamics. Neither the challengesthat have arisen to date, nor the alternativesthat have been suggested,have gained momentum, or even much attention, suggestingthat it will be sometime, if ever,before a freshnew paradigm replacesplate tectonics.Refinements arise, of course,and will continueto do so,but the basicconcept has not been shaken. In the followingchapter, I make somecomments on the strategiesand philosophiesthat led to plate tectonicsand alsoattempt some comments on fruitful activitiesfor the future, based on first-hand experiencewith the plate tectonics 10

Strategy,Philosophy, and ThingsLike That

Whatwe needis a logarithmicform Torate paradigm shifts 'gainst a norm Wecould just declarethen, Plate tectonics was a "10" And avertphilosophical storm.

ball drifting throughspace. It had a thick, inert, rockymantle that .encasedsthe 1960sa central,began, molten, theearth, metallicas humans core, perceived and a complexit,wasyet a round static pattern of land and sea coveredthe surface.That image of a relatively placid earth was aboutto changedramatically. By the end of the 1960s,the moltencore remained in that imagebut the once-inertmantle had become dynamic,still solid but neverthelessdeforming readily and carryingheat convectivelyfrom the hot coreto the coolsurface. There a remarkablythin, strong,brittle, cold, outer layer, loosely-speakinga frozen shell, enclosed the earth's interior. And the shell was neither static nor intact; it was broken into a few large platesthat moved aboutin conjunctionwith circulationin the mantlebelow. In sodoing the plate motioncaused the disturbancesthat shapedand modifiedthat thin outershell that is soimportant to us because it is the habitat of the human race. For earth scientiststhat transitionfrom a near-staticimage of earth to a much more dynamicone correspondedto an abruptparadigm shift, a fun- damentalchange in the basicframework of the science,an historicand almost certainlya once-in-a-lifetimeevent. As such it not only expanded knowledgeof earth and openedfertile new areasfor researchbut it also gave scientistsof that era a uniqueopportunity to observein personand at closequarters just how a scienceoperates before, during, and after a major upheaval. In this chaptertherefore, I focusnot so much on the basicscience of plate tectonicsbut rather on certainoperational aspects of scienceas they 106 Strategy,Philosophy, and Things Like That boreon the revolution.I do so in the hopethat provocationof somecon- templationof suchmatters may be helpful to thosescientists attempting to fomentyet anothermajor advance in the science.Specifically, in the fol- lowingI discussstyles of science,choice of frontiers,peer review, communicationof science,the roleof scientificsocieties, and some selected oppor- tunitiesand new directionsfor the field of seismologyand earthscience in general. It is somewhatpresumptuous, of course,for any individualto attempt a comprehensiveanalysis of phenomenonathat eachof us saw from a dif- ferentperspective and in a differentlight. No scientistknows precisely whatwent on in theheads of fellowscientists as the revolution developed, or evenremembers precisely what went on in his or her own head.I like to think of the participantsin the platetectonics revolution as analogousto dropletsof waterin a braidedstream. Each droplet entered the streamat a differenttime and place,and eachinteracted from time to time with other dropletsas various rivulets formed and intertwined.All were driven in the samegeneral direction and towardthe samegoal, yet eachfollowed a dif- ferentcourse and underwent different influences. Eventually all endedup neatlychanneled within a singlebroad valley as part of a moresmoothly and moreslowly flowing unidirectional river. The motion of the constituent dropletsis lesschaotic now, but none, of course,knows what stream gradi- entsor waterfallslie ahead.To attemptto describethe nature of thatstream basedon the historyof onedroplet is, of course,a riskybusiness, but at leastthe attempt may provoke others to contemplatewhat happened to the entirestream based on theirown experienceswith a part of it.

Styles of Science In previouschapters I have drawn attentionto somestyles of science thatwere important to therevolution, particularly to theseismological componentof it. Theyinclude science by hypothesistesting or whatsome might callthe hypothetico-deductive method, science by serendipityor theinduc- tive method,and scienceby synthesis,in a sensebroad application of the first two. All of thesestyles, and variationsand combinationsof them, playeda rolein theplate tectonics revolution. Science thrives on diversity.It wouldbe a seriousmistake for scienceas a wholeto curtailthe diversityof approachesin scienceand so divertattention away from any oneof these styles.However, in seekingto fomentanother revolution that advancesthe sciencein a majorway, as scientistsare wont to do, we might first askjust whatit wasthat triggered the plate tectonics revolution and so sent earth sci- enceoff in a new direction.Perhaps the sequenceof eventsthat led to suc- cessthen might serve as a guideto director prioritizeactivities in the Strategy,Philosophy, and Things Like That 107

In my opinion,and I think many otherswould agree,it was the effort to obtaindiverse and comprehensivegeophysical observations of the deep seafloor during the intervalimmediately following World War 11that triggered the revolution.In this effort, the inductive style, i.e. scienceby serendipity,prevailed. It was primarily the call of exploration of the unknown and not the desireto test a hypothesisthat drove the scientific effort.Of course,there were somehypotheses about the natureof the deep seabasins at the time. Wegener'shypothesis of drift was one, the idea that the Pacificwas a scarleft by the moon'sdeparture another. But the post WW II explorationof the oceanswas not specificallydesigned to testthose or other hypotheses.Rather the scientificgoal was probing of the great unknownearth science frontier of that era, the oceanbasins, just to seeand then to understand what was there. It was fresh new observations of those great marine basinsthat initiated the string of events of the 1960sthat would lead to plate tectonics,and therecan be little doubt aboutit. To make that conclusion,and so to focuson a broadly-basedobserva- tional program,is by no meansto minimize the importanceof the inge- nious and inspired ideas of clever individuals that moved the revolution along or to deny thoseindividuals respectand credit for their contributions. Nor is it to denigratethe significanceof the wealth of geological informationcollected earlier on dry land that becamethe basisfor part of what followed.Nor is it to makelight of early,pre-WW 11,progress in any branchof sciencethat played a role in the comingof plate tectonics.As I have tried to showin the caseof seismology,the early exploratorywork in a field is a criticaland necessarystep. Nevertheless, it seemsclear that, during that post-WWII period,the deepsea floor stoodout as the prime frontier of earth science,and observationof it, just to seewhat was there,was the thing to do. Thosewho led the effort to explorethe oceanbasins in this mannerled the scienceto major advance.I have already cited Lamont in this regardand in somedetail because of its specialrole in earthquakeseismology and becauseof my familiarity with that organization,but I hasten to add herethat otherorganizations were similarlyinspired and dedicated, notably among them ScrippsInstitute of Oceanography,Woods Hole OceanographicInstitution and Cambridge University.And, of course,it was interestin understandingthe oceansfor the military purposesof the Cold War that in part stimulatedthe funding for marine researchduring that era,but I referhere to the motivesfor the basicscientific program. That prime lesson from the plate tectonicsrevolution seems clear enough;it is neithera new one in sciencenor a surprisingone. The success of basicexploration of the oceanbasins in instigatingthe scientificupheaval is by no meansunprecedented. In science,exploration of a new 108 Strategy,Philosophy, and Things Like That almostalways produces surprising discovery. To hasten the paceto the next revolution,we need only identify the next majorfrontier and then observe it thoroughlyand in every significantway possible. Of course,to identify that frontieris no simpletask; it is one that calls for insight,vision, and good-fortune.It is a continuingchallenge to the leadersof a branchof scienceto seekout promisingmajor frontiersand to make certainthat no possibilityis bypassedor overlooked.And, of course, it is not only new discoverythat scienceseeks but also continualtesting, reevaluation,and upgradingof all scientifictheory. A little later, I shall draw attentionto what I believeis the, or at leasta, major frontierof modern earth science,the moderncounterpart of the deep oceanbasins of the post-W.W.II years.It is the entireburied continentalcrust. For the present however,let us continuethe discussionof stylesof science,and what to do about, or with, them. Sciencein the inductivestyle of explorationof the unknownnot only triggeredthe early stagesof the plate tectonicsrevolution, it alsoprovided the observational basis for much of what followed. It was the observational data basethat stimulatedthe serendipitousdiscovery that brought scientiststo the key conceptsof the revolution.It is difficultto seehow the revolution could ever have come about if observation had been conducted solelyto test specifichypotheses. Nevertheless, science by hypothesistesting with its emphasison deductionalso played an importantrole in certain aspectsof the revolution. Wegener,Hess, Vine and Matthews, Wilson, Morgan, and othersenvisioned the "way it might be" and so stimulated testingby observationthat moved the scienceahead. It is difficult to see, however, how those ideas would have caught on or how the revolution would have blossomedif those hypotheseshad been proposedin the absenceof a substantialquantity of relevantobservation. To demonstratethis point, one need only note how Wegener'sideas sputteredfor decadesuntil the post WW II observationsof the deep sea basinswere made, and the misconceptionsof scientistswho had obstruct- ed the idea of continental drift could be overruled. Or note that Ortelius madean importantand relatedhypothesis centuries earlier in the complete absenceof observationof both the seafloor and the geologyof the land surfaceand that it was immediatelyconsigned to obscurity.From these exam- plesalone, it seemsclear that the way to beginan episodeof majordiscov- ery in science,at leastan observationalscience like earth science,is through explorationof a new and importantfrontier. Once the comprehensivebase of freshnew observationsis available,a kind of interchangebetween the inductive and deductivemodes can be very beneficial,but it seemsmost conduciveto major discoveryto begin in the inductive :.

. .

JackOliver at pressconference, 1960s

MauriceEwing and Jack Oliver with seismographrecorders at Lamont Hall, about Lynn Sykes,1968

AFOSRAdvisory Committee, early 1960s--left to right: William Stauder, S.J., St. Louis University;Thomas O'Donnell, Gulf Research;James T. Wilson,University of Michigan;Hugo Benioff, California Technical University; Sidney Kaufman, Shell Development(later at CornellUniversity); Jack Oliver, Columbia University (later at Cornell Installationof seismograph station in Fiji, InstallationofLamont seismograph station at Niumate in Tonga, GeorgeHade at seismographstation in deepmine, Ogdensburg,New Jersey,about 1960

GeorgeHade and Merrill Connor, seismograph station in deepmine, Ogdensburg, New Top:Bryan Isacks, 1990s;

Left:Lynn Sykes,1992;

Bottom:Visit to Cornelland COCORPproject by Directorof NSF; left to right:James Krumhansl, RichardAtkinson, NSF, and Jack Oliverretirement party, 1993. Top row, left to right:Jim Dorman,Walter Mitronovas, Bob Metsger,Jack Oliver, Mark Langseth,Frank Press, Jim Brune, Bryan Isacks; Bottom row, left to right:Klaus Jacob, George Hade, George Sutton, Chuck Drake, Dick Edwards,John Kuo, Judy JackOliver at seismographrecorder, late Strategy,Philosophy, and Things Like That 109

I make this point at somelength in order to stressthe value and the primary importanceof the inductive style as well as the need to maintain diversity of styles of science.In modern science,there are certain factors relatedto organization,justification of effort,and funding that tend to force the sciencetoward the deductivehypothesis-testing style to the exclusion of others.More generally,it seemsthat, during the periodsof problemsolv- ing as opposedto periods of paradigm shift, hypothesistesting tends to predominate. Hypothesis-testingis usually a somewhatmore tidy kind of operationthan is observationof a frontier,and henceis more easyto organize and administer.It is also somewhateasier to justify to non-scientists becausethe goalsare clear-cutand becausea particularproject usually has a relativelyshort and finite life. Inductive,serendipity-based, science is less easyto categorizeand delimit and may seemless appealing to non-scientists.Consequently, hypothesis testing is often, too often, heavily favored by the politicalworld and the bureaucracy,by review committees,by peer reviewers and the like. The historical record of discoverydescribed here for the caseof the plate tectonicsrevolution is a firm basis for challenge of that strong favoritismfor scienceby hypothesistesting over scienceby serendipity,i.e., for the deductivestyle over the inductivestyle.. A scienceis most likely to be productiveof major discoveryif it is not controlledand fettered and directedby hypothesistest after hypothesistest. Sole reliance on hypothesistesting limits a scienceto progressat a rate dependentupon the imagi- nation of the participatingscientists. Science need not be limited in that manner.Inductive sciencefavoring serendipityoffers better prospectsfor the startlingadvances that can far surpassthe rate at which human minds can envision new discoveries. Scienceby synthesisusually involvesa combinationof the inductive and deductivestyles as applied to many and varied observations,hypotheses,and theories.Science by synthesiswas very importantin the plate tectonics revolution. In addition to the seismologicalstudies cited earlier, examplesinclude the global map of marine physiographyprepared by Heezen and Tharp, the globalmap of agesof the seafloor basedon magnetic anomaliesby Heirtzler and others, and the interrelatingby many geologistsof a wide variety of geologicalphenomena with the plate tectonic model as the basis.These few spectacularexamples from earth science are climacticones and attention-getters,but perhapsthey will highlightthe opportunityand the need for synthesisof almostall active scientificsub- jectson a continuingbasis. Later,while discussingpublication, I shall try to make a strongcase favoringmore fundingfor synthesisand for preparationand publication 110 Strategy,Philosophy, and Things Like That

, , more booksand articlesthat review subjectsthoroughly and in depth as oneway to overcomesome current problems of communicationof modern science.But let us turn now to the topic of the major frontier of earth scienceat present.This topicis a part of this bookbecause it is importantand becausethe history of plate tectonicsprovides a basisfor deliberationand decision on this crucial matter.

Frontiers of Earth Science It is my firm conviction,based on personalexperience with studyof the earth,that the buried continental crust and adjoininguppermost mantle con- stitutethe majorfrontier of earthscience at present.The positionof thispart of the earthon the list of targetsfor human explorationof earthseems anal- ogousnow to that of the oceanbasins just afterWW II. Boththe oceanbasins and the continentalcrust must be importantto the understandingof the earth if for no otherreason than that they arehuge. Furthermore, except for the surface where the geologyis everywhereknown on at least a reconnaissance basisfrom geologicalobservation, and the sedimentarybasins where petroleum explorationhas focused much attention and efforton subsurfaceexplo- ration, and a few other locationswhere drilling or other specialcircum- stanceshave produceddetailed information, the continentalcrust has been exploredonly spottilyand then only with techniquesof low or modestreso- lution.Like the floor of the deep oceansafter WW II, or like the continentof North Americaduring arrivalof the earlysettlers, the continentalcrust at this point in historyawaits exploration in detailedcomprehensive fashion. Thereare, of course,numerous other important frontiers of modernearth science.For seismolog•5for example,the core-mantleboundary is clearly one,the innercore perhaps another. I focuson the continentalcrust here however,partly becauseit is my personalfavorite and because,like the ocean basins,it has the potentialfor providing scientistswith huge quantitiesof diverseand revealingdata. Of course,I do not mean to imply that every frontier of modernearth scienceis necessarilya spatially-definedentity. For example,the existence and the nature of microbiallife at depth in the earth is clearly a frontier topic of modernbiology and geology.The role of fluids travelingthrough and interactingwith permeablerocks of the crustis anotherfrontier topic. Nor do I imply that the potentialfor further importantdiscovery concern- ing the floorsof the oceanbasins is exhausted.Students of earth scienceof this era are blessedwith the opportunityto exploremany challengingtop- ics,far too many to enumerateor discussin detail here. As explorationof the buried continentalcrust progresses, by analogy with the history of explorationof other earth frontierswe can Strategy,Philosophy, and Things Like That 111

, surpriseand revelationand eventual satisfactionas we develop a much improvedunderstanding of that part of the earthand thoseparts that inter- actwith it. In all probability,our currentconcepts of what is thereand what has happenedthere will undergo major revision.Perhaps another paradigm is in store.A major opportunityfor earth sciencecan be exploitedby forming a program for comprehensiveexploration of the continentalcrust. And the inductive style, scienceby serendipity,should be stressed.The program will flourish if, like its counterpartdirected toward the oceans after WW II, a variety of kinds of observationalmethods are employedin an effort to survey the territory comprehensivelyin both a spatial and a multi-disciplinarysense. The criticalideas will, asthey alwayshave, spring up from someone,somehow, somewhere, once those observations become availableand known.The programwill sputterfitfully if the styleof science is hypothesistesting alone. Generation of hypotheses,and hypothesistesting, is appropriateof course,but if only that style is employed,advances will be more or lesslimited by the imaginationof the scientistsinvolved. Thoseminds will havebetter and more creativeideas if stimulatedby new setsof comprehensiveand diverseobservations. On the more practical and humanistic side, the continental crust deservesto be exploredthoroughly because it is the part of the universe that humanslive on and derive their livelihoodfrom. The upper portions of the crustare accessiblenow and provide resourcesthat supporthuman life; still deeper portionswill becomeaccessible as technologyimproves with time. We need to know what resourcesare thereand how they canbe usedeffectively to supportand benefitthe humanpopulation. For the fore- seeablefuture, and then some,most humansare not going to live on, or travel to, anotherplanet. They are destinedto remain on SpaceshipEarth. They need and deserveto know fully the resourcesthat are availableto them on this planet. Thereare otherbases for justificationof intenseexploration of the continental crustin the near future. Explorationof the oceanbasins after WW II was facilitated by the availability of new instrumentsand new techniques,developed in part during the war for wartime purposes,that could be adaptedto observationof the oceanfloor. Seismicand acousticdevices, magnetometers,and navigationtools are some examples.Currently, and analogously,there are geophysicaltechniques and deep drilling methods availableand waiting to be appliedto study of the buried continentalbasement. We need only to mount the effort to apply them in widespreadand effective fashion. The wide variety of geophysicaltechniques that is availablefor exploration of the deep continentalcrust includes the seismicreflection 112 Strategy,Philosophy, and Things Like That refraction,earthquake seismology, gravimetric, electrical, electromagnetic, magnetic,thermal, and satellitepositioning techniques. There is alsoa variety of geochemicaland geologicaltechniques. By way of illustration,I dis- cusshere the seismicreflection and deep drilling techniques. Basedon the strategicalreasoning of the precedingparagraphs, at Cornell in the early 1970sSid Kaufman, a retired industry geophysicist with an eye to innovation in earth scienceand the know-how to make it happen, and I began a projectnow known as COCORP (Consortiumfor ContinentalReflection Profiling). We were soonjoined by Larry Brownas a graduatestudent soon to becomea new professorat Cornell.The essence of the COCORPproject is the applicationof the powerful seismicreflection techniqueof the petroleumindustry to explorationof the entirecontinental crust.COCORP is devotedto explorationto depthsgreater than thosenor- mally exploredin the searchfor oil, and to exploration,wherever they are found, of the crystallinerocks of the basementthat are not normally of interestto modern industry. COCORPhas surveyeda number of deep seismicreflection profiles in the US, observingin the processonly a small fraction of what eventually can and must be observed,and has demonstratedbeyond all reasonable doubt the value and efficacyof applicationof the seismicreflection technique to study of the deepbasement (Brown et al, 1986).There are clearly many large, eerie, geologicalfeatures, normally hidden from view but revealedby the seismictechnique, within the deepcrust and sometimesthe uppermantle. In someareas such as the Appalachians,COCORP data have beencritically important to our presentlevel of understandingof majortec- tonicfeatures. From the COCORPexperience and that of relatedprograms that have sprungup in the US and abroad,there can be no doubt that society will eventuallydemand, and profit from, the knowledgeof the subsur- face geologicstructure of the continentsthat the seismicreflection technique, drilling, and other meansof exploringthe crustcan provide.How and when that knowledgeis obtainedis no longer a technicalmatter; it is a politicaland financialone. Drilling is, of course,an obviousand proven techniquefor exploring, and for recoveringresources from, the earth'sinterior. It is estimatedthat some two million holes have been drilled in the United States, most relatively shallowbut many over 5 kilometersin depth. To date, most deep holeshave been drilled in the sedimentarybasins, the habitatof petroleum. The recorddepth so far, some12 km or abouta third of the thicknessof the continentalcrust, was drilled in a scientificexperiment in the SovietUnion. It is not unreasonableto hope that depthsas great as 20 km or more may eventuallybe penetratedand Strategy,Philosophy, and Things Like That 113

Widespreadexploration of the continentsby deep drilling seemsa certainty for the future, partly for purely scientificpurposes and partly, perhaps largely,because of possiblepractical benefits. Surely it is in the interest of humans to know what existsand is accessiblejust a few kilometers beneath their feet. Drilling should be done hand in hand with seismic reflectionexploration, using both conventional2D and modern,high resolution, 3D methods.Other kinds of geophysicalexploration based on the seismic,electrical, electromagnetic,gravity, heat flow, etc. methods, all shouldbe part of a coordinatedprogram for comprehensiveexploration of the continents.The stimulusand the role for sucha programcan be found in the history of explorationof the sea floor in the 50s and 60s that led to plate tectonics. Perceptivereaders may claim a possibleinconsistency between the pro- posedfuture explorationof the continentalbasement under an "organized" program,and the actualpost WW II explorationof the oceanbasins used here as a guide or model. For the most part, the explorationof the ocean basinswas not organizedabove the level of the individual researchlaboratory. That is, no one, so far as I know, tried to make certain that Scripps, Lamont,Woods Hole and the othersoperated under a coordinatedcentral plan for exploringall the oceanbasins. Instead the initiative and the deci- sionson scopeof researchand researchtargets were left to the leadersof thoseinstitutions. In fact, this approachworked well in this casebecause the researchlaboratories were sufficientlylarge and diverseso that they couldreadily handle the task.At present,there are no parallelorganizations on which to basea comprehensiveexploration program for the continents. Almost certainlythe bestway to explorethe continentswould be to form not onebut severalsuch organizations. A structurewith suchresearch units taking the lead would almostcertainly prove preferable to operationunder a singlecentral organizationwithin the Beltway.With the multi-organiza- tional researchlab structure,the apparentinconsistency referred to at the beginningof this paragraphwould vanish;history would have again provided an appropriateprecedent.

Peer Review One major difficulty with operationunder a single,central organization is that, with only a smallnumber of administratorsand advisorsin sole controlof a large program,innovation and flexibility are commonlystifled because of absence of competition from outside the organization. Judgmentsbased on peer review,rather than on the individual initiatives and insightsof the top scientists,become the controllingfactor. That is not to imply that peer review is all bad. To someextent control by peer 114 Strategy,Philosophy, and Things Like That is goodbecause, in democraticfashion, it rules out initiativesthat are clearly sub-par.Unfortunately however, peer review often, and also in democraticfashion, rules against the exceptionallygood initiatives as well. That is becausethe truly outstandingdiscoveries in sciencecome as a surpriseto even the very best of scientistsincluding, of course,the peer reviewers. Thus prior to learning of the discovery,those scientistsacting as peer reviewersmight well rejectthe proposedeffort that eventuallywould have led to the discovery. This phenomenoncan be neatly documented.It was demonstrated clearly and spontaneouslyat the American GeophysicalUnion spring meetingof 1992held in Montreal.A specialsession was held at that meeting in honorof what was called,somewhat arbitrarily, the 25th anniversary of the comingof plate tectonics.At that sessioneight scientists,all key participantsin the revolutionduring the peak of the upheavala quarterof a centuryearlier, gave presentations.All were given a minimum of guidance from otherson what to say.The eighttalks, in otherwords, were developed independentlywith virtually no coordinationamong the speakers. If my memory is correct,and I think it is becauseI was so strongly impressedby the phenomenonat the time, five of those eight speakers spontaneouslydrew attentionto some aspectof the revolution that had been unjustly delayed or otherwiseimpeded by peer review. Ideas that would turn out later to be key to the revolutionwere rejectedin proposal form or in paperssubmitted for publication.I have neverseen a moredra- matic and clear-cutdemonstration that peer review doesnot work well, or at leastas effectivelyas it might, when a truly major advancein scienceis beingmade. Remember that thosecomplaints were madeby prominentsci- entists25 years after the incidents,long after thosescientists might have beensuspected of harboringsour grapes, or of having an ax to grind, or of having a ruffled ego. I do not mean to imply that peer review shouldbe abandonedfor it is not clearthat a bettermethod for judgmentof basicsci- encecan be devised,but we must keep our eyes open and be aware that peer review may impedeprogress at someof the mostcritical times in science,specifically those times of major discoveryand paradigmshift. A somewhatrelated matter is the unfortunatetendency of bureaucrati- cally controlledfunding agencies to breakprograms into smallpieces, each piecerun by a singlemanager and eachpiece militantly protected by its con- stituency. As one meansto overcomethe deficienciesof the peer review system,I suggestthat somesort of publicationbe designedto encouragethe printing of radicalideas and hypothesesthat might not make it into the regularsci- entificjournals because of rejectionbased on peer review.A Strategy,Philosophy, and Things Like That 115 free-wheelingjournal would, of course,contain many hypothesesdoomed to failure,and the journalwould draw somederogatory comments, but, if done circumspectly,it would draw many readers,stimulate the imagina- tion, and provide a meansfor airing a radical hypothesisthat might catch on and reorganizesome branch of science.Perhaps such a "journal" will evolveas the new electronicforms of communicationcontinue to develop.

Communication of Science From the historyof plate tectonicsin the earlierpages, as well as from histories of other branches of science,it seems clear that, in order for an individual to make an impactin the world of science,just havinga goodidea is not sufficient.It is alsonecessary that the idea be communicatedappropri- ately to the milieu where the idea can catchon. As Menard (1986)has pointed out, ideasare plentifulin science.Most, includingsome very goodones, have little impact on the mainstreamof scienceat leastpartly becausethey arenot communicatedproperly. Appropriate communication of a goodidea, in otherwords, is as importantas generationof the idea. That point seems obvious,yet it is often overlooked.Wegener, for example,made his ideas widely-knownthrough oral presentationand publication.Holmes' ideas are readily available in his outstandingtext (Holmes, 1944), but his original paper on the subjectthat would becomeplate tectonicswas publishedin Transactionsof the GeologicalSociety of Glascowand henceoverlooked by Americansand otherswhose libraries did not includethat journal and who would becomeleaders in the establishingof platetectonics. Goodcommunication is alsovitally importantin otherways in science. The historyof plate tectonicsdemonstrates very clearlythat the important contributionsof many individualswere made at leastpartly becausethose individualshappened to be in the right placeand at the right time, and not necessarilybecause those particular individuals were exceptionallygifted. Beingat the right placeat the right time oftenmeant that the individualwas situatedso as to receiveand assemblethe key information,which was often split into multiple componentsand which revealedthe opportunityfor dis- coverybefore others were soinformed. Good communication not only within a disciplinebut also, and especially,across discipline boundaries was importantboth early on and as the revolutionprogressed and broadenedits scope.Good paperswritten with a minimum of jargon and thosewith the specialfigures that become "icons" became important, as noted earlier, partly becausethey communicatedinformation readilly across discipline boundaries.It is easyto make the casethat good commtmicationis not only desir- ablein science,it is essential.It is deservingof specialattention and efforton the part of the scientistto assurethat he or sheis favorablysituated 116 Strategy,Philosophy, and ThingsLike That the communicationnetwork, and that his or her papersof importanceand broadinterest are written for an audienceof non-specialists. Sometimeago, but after the transformationof earth scienceduring the 1960sand early 1970s,someone attempted to list the dateswhen various earth scientistshad "converted"from the "fixist" campto the "platetectonics" camp.At first, somethought that sucha chronologymight providea measureof the scientificability, or the expertise,or just the "savvy,"of the individualsinvolved. But it is far morelikely that the orderingof thatlist was much more dependenton the times when the significantand convincing informationhappened to reacheach individual. Nearly every earthscientist recognizedthe importanceof the new conceptonce the newsand theessence of the conceptwere conveyedto that individual.The historyof the platetec- tonicsrevolution makes the point emphaticallythat communicationis of the essencein scienceand soimplies that maintainingand enhancingthe means of active and efficientcommunication in sciencedeserve continuing and appropriateattention. At present,communication in scienceappears to be in the earlystages of a periodof majortransition. On the onehand, thereis the well-established, conventionalform of recordingand communicatingscience through the scientificpaper in the scientificjournal. This style was setin the nineteenthcen- tury and has served sciencewell, but with time it has developedsome notabledeficiencies as a consequenceof growthand increasingspecialization in science,proliferation of jargon,and a styleof presentationof information thatis not optimizedwith regardto communicationand retrievalof essential information. Evenif nothingelse were to happento affectthe realmof scientificcom- munication,the establishedsystem of publicationin scientificjournals of papers,all written in stereotypedstyle, would need reevaluationand mod- ernization.Modem scientificpapers are oftennot beingread by the intended audienceand henceare not communicatinginformation well, if at all. Too manyare written only for thesmall circle of specialistsin thatfield who know thejargon or, who throughpeer review,influence the activityand fundingin that specialty.Certainly all the informationof modem scienceis not being communicated,or stored,optimally. Often things already reported in the lit- eratureare "discovered"anew by someoneunable or unwilling to keep up with therecord of pastdiscovery. If thistrend continues and the problemscon- tinueto grow,we may somedayreach an equilibriumstate in whichsuppos- edly new scientificdiscoveries occur at, or evenless than, the rateat whichthe samescientific discoveries already made earlier are forgotten! But somethingelse is happeningin the realm of scientificcommunica- tion, and in all other forms of communication. It is the electronic Strategy,Philosophy, and Things Like That 117 a truly awesomephenomenon of our times.We have at presentseen only the beginningof what promisesto be a completeupheaval in the ways in which societystores and communicatesinformation and in which individ- ualslearn from that information.From the beginningswe canalready sense the awesomepotential of this revolutionand someof the problemsthat will be encountered.I do not pretendto clairvoyancehere and make no specif- ic predictionson how it will all turn out but one point seemsclear. The electronicsrevolution, coupled with growing deficiencyin the presentsystem, is going to force a major changein the way scientistsstore and communicate the information of science.It is already beginning to do so. Furthermore,the effectswill be profound and go well beyondthe simple processesof storageand retrievalof information. To illustratethis point with just one example,consider the matter of thoroughobservation of all of the earth'ssurface geology. This huge task has occupiedinnumerable geologists for decades,even centuries.They have walked over almostevery part of the dry land surfaceof the earth,eventually to producecomprehensive information in the form of maps,at leaston a reconnaissancebasis, of all of that surface.Now, becauseof the comingof the electronicage, that task will almostcertainly have to be redone-completely redone!The reasonit will be redoneis that the raw informationso obtainedhas been collatedand then stored,but in most casesonly stored selectivelyin interpretedform, primarily geologicmaps. Going from the raw observationsto geologicmaps was necessaryto, amongother things, organizeand reducethe amountof informationstored and transmitted. In the new electronicage, however, it is possibleto storehuge quantities of data. It may eventuallybe possibleto store,and retrieveon command, any or all of the raw observationaldata of geology,not just the interpreted and condensedforms of thosedata. From the raw databank, computers will quicklyproduce, for example,maps emphasizing those features of particular interestat the moment,not merely thoseselected by someearly map maker.Surely suchflexibility would be a boon to the science.The catchis thatthose raw observationson the earth'sgeology have not beenstored sys- tematically,nor are they in computercompatible form. Eventuallywe shall want all of thosedata in that computercompatible form and soprobably we shall have to re-observemost of the world's surfacegeology, this time recordingthose observations in a new fashionin orderto getit that way.The task is not trivial, of course, but neither is it insurmountable, and modern technologywill make it easierthan it oncewas. At presentwe face the dangerthat progressin earth scienceis being skewed away from the optimum coursebecause only a fraction of the observationaldata of our scienceare in computer compatible 118 Strategy,Philosophy, and Things Like That

Computersare making possiblethe applicationof marvelousnew techniquesfor handling,processing, and analyzingdata, but thosetechniques can only be applied to computercompatible data. However,our scienceis aboutorganization of observations,all observations,of earth.We eventually must take stepsto insurethat all importantraw observationsof earth, includingparticularly those made in the field by geologists,are computer compatible,so that the directionof the scienceis not restrictedor restrained by incompatibilityof someof our mostimportant data with new electronic devices.

The Role of Scientific Societies Let us now turn to the role of scientificsocieties in the plate tectonics revolution.Because of the broad scopeof the subject,almost all earth sci- encesocieties were affectedin somemanner by the revolution,some during its early formativestages, others later. Therecan be little doubtthat, of all scientificsocieties, the American GeophysicalUnion had the biggest impacton the growthof the plate tectonicsrevolution, and viceversa. The extraordinaryAGU meetingof 1967has alreadybeen cited severaltimes in this regardin earlier chapters.Furthermore, a disproportionatenumber of key publicationson plate tectonicsand its forerunnersappeared in AGU's Journalof GeophysicalResearch. Scientists of the discoveryera clearly thought of AGU as the place where the action in the subjectwas to be found.Why wasthis so? Was it justchance and goodfortune that AGU was focusedin just the right subjectareas so as to capturethe centralposition as thosesubjects were integratedunder the new conceptof plate tectonics?Or were the policiesof AGU differentfrom thoseof other societiesin a way that fostered the revolution? I think the answerto both of thosequestions is yes. ClearlyAGU was in the right place at the right time to capturethe plate tectonicsaction, includingparticularly the forerunners.For example,geophysical observationsof the deepsea floor were a key pieceof early information,and AGU, by welcomingpresentations of thisthen rather unusual kind of information on the earth at its meetingsand in its publications,established itself as a, if not the,place to reportsuch observations. Furthermore, through wise plan- ning,and probablysome good fortune, the AGU organizationincluded sec- tionsin most of the specialtiesthat would be importantto the revolution. And certainlynot all of thosespecialties were popularat the time theywere formed. I can recall,for example,meetings during the 1950swhen atten- danceat the sessionsof the Tectonophys,•cssection often numberedhalf a dozenat most.Later, as plate tectonics blossomed, multiple sessions of that particularsection packed the meeting Strategy,Philosophy, and Things Like That 119

Certain aspectsand policiesof AGU set it apart from other earth sci- encesocieties, and leadersof the AGU during that early era deservecredit for their foresight.For onething, AGU setout to be interdisciplinary,merg- ing geologyand physicsand chemistry.AGU covereda broadrange of spe- cialtiesand hence the audiencesit generatedwere more attractive to a speakerwith a topicof broadsignificance than was a morespecialized society. Such societies,confined to one narrow subdiscipline,attract almost exclusivelyspeakers who feel their talk has appealonly for thosespecial- ists.In the age of modern sciencewhen the biggestdiscoveries are com- monlycross-or inter-disciplinary, it seems obvious that the societywith the broaderpurview is more likely to get the mostimportant papers, unless of course,the societyis so broad that only a small fraction of its members mightbe interestedin the papereven though it might spanseveral specialties.Thus the SeismologicalSociety of America,the MineralogicalSociety of Americaand the PaleontologicalSociety were too narrowly orientedto capture the main action of the plate tectonicsrevolution, the American Associationfor the Advancementof Sciencewas too broadlyoriented, and the AmericanGeophysical Union was just aboutright. Furthermore,as bestI canrecall, AGU had a relativelyliberal policy on the subjectmatter of papers,whether printed or presentedorally. Submitted papersof somemerit that did not clearlyfit within the boundariesof a particular AGU section were nevertheless somehow shoehorned into the meeting or the journal.Some papers that presentedill-conceived hypotheses or that were basedon questionablereasoning slipped in the back door under thispolicy, and generatedsome scorn and negativecomments in the hallsof the meetingplace after the presentationof course,but in the long run the societywas not harmed by this policy and eventuallythe liberal policy broughtthe societybenefit and the scienceboth stimulationand advance. As noted earlier,and in my opinion,scientific circles should provide more rather than fewer opportunitiesfor presentationof the off-beat,long shot,hypothesis. AGU's successin capturingthe excitementof the plate tectonicsrevolution, and the failure of othersocieties to do so when they might have, demonstratesthis point. Furthermore,although I cannotcite specificexperiences from the plate tectonicsrevolution to supportit, it is also my opinionthat most modern scientificsocieties, though focused appropriately on hard science,should includesections on the historyand philosophyof their scienceas well. If appropriateactivities developed, hard scientistsmight learn strategiesand tacticsfrom historyand the analysisof it, and historiansand philosophers would have closer interaction with active scientists and hence firmer groundingin the sciencethan they might otherwise 120 Strategy,Philosophy, and Things Like That

Opportunities in Seismology And then there is the questionof what ideasthis particularhistory of seismology'srole in the plate tectonicsrevolution might provoke and so improvethe scienceof seismologyand relateddisciplines in the future.This topic is a difficult one becausein most ways the scienceof seismologyis currently in a healthy state,certainly far healthier than it was in the early days of the subjectas discussedin previouschapters. Since the middle of this century,there has been a marked increasein the numbersof seismologists,and in the numbersof exceptionallytalented and well-trainedseis- mologists.Observational and analyticalfacilities have grown remarkably. Furthermore,modern seismologistshave been aggressiveand adept at developingnew subjectareas, new approaches,and new techniques. To citejust a few of thosedevelopments, I draw attentionto currentor recentwork on topicssuch as paleoseismicity;focal mechanisms;digital data handling and processing;tomogaphy; prompt high speedtransmis- sionof information;earthquake-related geodetic studies; slow earthquakes; spatialand temporalpatterns of seismicity;and the explorationof the crust and upper mantleusing explosive and vibratorysources. Against this array of advancesand signsof vitality, it is difficult to earmarkfertile new areas for researchin the future,but a few suggestionsfollow nevertheless. First, and this will comeas no surpriseto seismologistsfor the point has often been made before,there is a strongneed to remedy the dearth of observationsfor the areas of the ocean basins by operating numerous observingstations on or below the seafloor. Some two-thirds of the earth's surfacearea, exceptfor a few island stationsand an occasionalresearch instrument on the oceanbottom, is free of seismicstations. The resulting and enormousgap in informationgrows in importanceas more and more we seek to resolvedepartures from the spherically-symmetricmodel of earth. At some time in the future we must observe the seismic motion of the entire earth's surfacein a comprehensivemanner; the oceanicareas are clearlythe sitesof the greatestdeficiency in coverage. Interactionbetween seismology and geologyon a regularbut selective basiscommonly takes place now and with somestriking examples of success.However there is potential for considerableincrease in productive interactionin the future. The caseof explorationof the continentalcrust providesa goodexample of that potential.Traditionally, for the uppermost crust,particularly the basement,the observationalbase is primarily geo- logicalin nature and the data are very numerousand extremelycomplex. As a consequence,a style of reasoningand of analysisof suchdata that might be termed the "geologicalstyle" prevails.For the lower portion of the crust,the database is primarily geophysicaland the data,in contrast Strategy,Philosophy, and Things Like That 121 the casefor the upper crust,are sparseindeed. Consequently, a "geophysical style" of reasoninginvolving numbersand modelswith geometrically simple components,is the norm. However, as more detailed, more complex,and more plentiful information on the deep crustis acquired,as it surelywill be through application of the 2D and 3D seismicreflection techniques, a situationmore like the casefor the upper crustwill develop.The "geologicalstyle" of reasoning will supersedethe "geophysicalstyle" and will be applied at increasingly greaterdepths, eventually to includeall of the crustand uppermostmantle, and perhapseventually all of the mantle.Seismologists can foreseethis developmentand move aggressivelyinto this interdisciplinaryarea by appropriatelyadjusting the tenorof the scientificmeetings of their societies and the scopeof their specializedjournals. Geologists can likewise move further into the subjectby extendingthe purview of their organizations. Continualadjustment and someoverlapping of the scopeand natureof the societieswill insure that this important interdisciplinarysubject does not fall into a crackbetween organizations, or forcethe formationof a new and overly specializedone. History shows that seismologistshave made remarkableprogress in their subjectthrough emphasis of an analyticalstyle of study of seismological data, following in generalthe reductionistmode pioneeredin physics. In this mode, data on the very complexmotions of the earth following an earthquakeare organized and generalizedby recognizingand defining characteristicphases and wave trains that are found most everywhere, regardlessof positionon the earth.Such information is, asnoted earlier,the basis for much of what we know of the earth's interior. However, there remains a storehouse of information recorded on mul- titudinousseismograms from innumerableseismograph stations that cannot be so readily generalizedbecause it is locationspecific. Such information hasoccasionally been studied in specialcases but muchof it is ignored, at leastpartly becauseto conventionallytrained seismologistsit seemstoo complexand too diverseto be handledneatly. Seismologists might take a tip here from geologistswho commonlyface large quantitiesof complex, locationspecific, data. They have developedvarious ways of handling and describingsuch information-geologic maps in all their variety provide an example.Seismologists might develop similar techniques.Nearly every stationseismologist eventually recognizes information peculiar to that station but, at present,nowhere are thoseobservations catalogued and por- trayedso that they canbe assimilatedby any seismologistanywhere. There seemsa huge opportunityhere as emphasison the earth'ssubtle 3D char- acteristicsgrows. The study by Molnar and Oliver (1969), cited earlier 122 Strategy,Philosophy, and Things Like That supportfor the plate tectonicstheory, is one very crudeexample of the use of this kind of information in this fashion. The lack of summariesor synthesesin book or map form of the kind of seismologicalinformation just referred to draws attentionby a particular exampleto a generaland widespreadshortcoming in science.It was mentioned briefly in an earlier chapter.Modern sciencehas produceda huge, virtually inestimable,amount of knowledgeof the natural world. Most of that informationis written in the form of articlesin scientificjournals and storedon the shelvesof libraries.But mostof thosejournal articlesare narrowly oriented.They concernprimarily and often exclusivelysome fragment of informationthat the author hopesmight somedaybear on much broadersubjects. However, comprehensivesyntheses of suchinformation on particularsubjects are limited. Thereis, in otherwords, and in my opinion, a cryingneed for morebooks or summariesor reviewsthat thoroughly and authoritativelysynthesize knowledge on particulartopics in a manner that is morecompact than the numerousand diversejournal articles on which sucha book canbe basedyet is far more detailedthan, say,a section of a generaltextbook. Suchbooks exist for someselected topics, of course,but not for many and not nearlyin sufficientquantity to make communicationof information in scienceas efficientas it might be. The problemarises because the kind of bookI referto will not havea sufficientlylarge audience to be profitablecom- merciallyand henceis not encouragedby privatepublishers. Funding agen- ciesfor science,though generous in their supportfor journalarticles, are typ- icallyreluctant to fund synthesesbecause the work is not "new"research.We needto encourageefforts by individualswho will reviewjournal articles in varioussubject areas and draw the essenceof knowledgeof thosesubjects into a form of review,a bookor perhapsnow a video,for thosewho needthe informationbut have not the time for a thoroughliterature review of their own. If we do not soonmount an effortto organizescientific knowledge, one that goeswell beyondthat of review articlesin journalsand the all too occa- sionalreview volume, as noted earlier sciencemay somedaydegenerate to the point where most "new" discoveriesare all rediscoveriesof phenomena onceknown, but then hiddenand forgotten! Seismographsoperating as they do for 24 hoursper day,day afterday, at numerouslocations record a huge amottntof data on not only earthquakes but alsoon the backgrottndnoise of the earth,a phenomenoncalled "microseisms."Although microseisms have beenstudied from time to time and in somedetail, this subjectis not exhausted.Furthermore it callsfor knowledge of meteorologyand oceanographyas well asseismology, because winds generate the oceanwaves that in turn generatethe microseisms.It calls Strategy,Philosophy, and Things Like That 123 knowledgeof geologybecause microseisms are mostly surface waves of such shortwavelengths that they are highly sensitiveto spatialchanges in geology. We can concludethat the subjectis not exhaustedbecause microseismic activity is not understoodsufficiently well to enableprediction of seismic backgroundnoise at any particulartime and place.This subject calls for additional attentionand for freshideas on how to study,analyze, and organize the backgroundnoise of the earth. During the electronicsrevolution, and particularlysince the early phases of the digital era, advancesin meansfor handlingand processingseismic data havebeen truly astounding,and seismologywould be far moreprimi- tive if it had failed to capitalizeon thosenewly availablecapabilities. I am a strongsupporter of thesedevelopments and do not intend the following point to detractin any way from the importanceof the new digital toolsof seismology. However, when viewing the sciencefrom the historicalperspective, it is evident that the excitementthat pervadesthe subjectof seismologyis partly a consequenceof the surprisethat is inherentin the earth'sseismo- logicalactivity. There is somethingawesome and fascinatingabout learning that the earth hasjust producedanother major earthquake,no matter how many suchevents one has observedin the past.There is somethingstimu- lating aboutworking in a subjectin which eachday might bring yet another unexpectedand information-ladenevent. Thus there is the dangerthat somethingintangible is lost when seismicmotion of the earth is recorded solelyin non-visibleelectronic form suchas tapesor discs.The excitement and the vitality of old-timeseismology depended partly on the old pen and ink recordersthat gave even the casualviewer a senseof earth as a normally quiescentbody whosepeace was occasionallyinterrupted by a huge catastrophicevent. Suchan eventbrought pain and sufferingto somebut wonder and excitementand intrigueto othersas questionsabout the location and size of the event were answeredbefore their eyesas wave forms slowly accumulatedon the recordingin full view of spectators. ! urge seismologiststo maintain,in addition to modern meansfor elec- tronicdata storage,visibly recording devices (and not just a video with the last few minutesof data on the screen)so that generationsof future seismologists,and others,scientist and non-scientistalike, can experiencethat thrill and excitementand stimulation.Let the earth,in its unpredictableyet characteristicstyle, tell them that it is dynamic,that it has ruptured once again, and that seismologyand plate tectonics,working together in the mannerwe learnedto understandduring the 1960sare at it onceagain. In the ideal world as I seeit from my biasedposition as a seismologist, every institutionof higher learningand every high schoolgiving 124 Strategy,Philosophy, and Things Like That tion in earthscience would have such a seismograph,visibly recording and prominentlydisplayed, so that all studentscan sense the adventure of mon- itoringthe unpredictable fits and starts of our planet. And then,last but surelynot least,there is the continuingopportunity to relatenew kindsof observationsto earthquakeoccurrence, as technology developsto makethose new observationspossible. The prime current exampleof suchopportunity concerns refinements in globalpositions basedon satellitenetworks which seemnow to be capableof providing long term strainand strainrate informationin seismicallyactive areas. Surelythe understanding of earthquakes, and plates, will be enhancedby this 11

Epilogue

In a worldbeset with defiances It behooves us to build more alliances Sowhy not immerse Technicaljargon in verse And integrateart with thesciences?

inally,I wouldlike to sharesome misgivings about authoring and publishinga book like this one. There is a danger,I sense, that some young scientistswill read historieslike this that portray casesof dra- matic successin scienceand feel depressedand unfortunatebecause they missedit. They may feel that as a resultof an accidentof birth that brought them along a little too late, they were not aroundfor the fun, the excitement, the satisfaction,and the accoladesof the plate tectonicsrevolution. For thosewho feel that way I have advicein the form of a singlecon- traction:"Don't." No one working in earth sciencetoday should feel that they have missedout on opportunityfor discoveryfor theresurely will be more big and exciting discoveriesin earth sciencein the future. That is clearlyevident, and a virtual certainty,because much of the earth is unob- servedby the many and variousmeans that are availableto us. There are many important observationsof earth remainingto be made. Fresh new observationsof an importantpart of earthnearly alwaysproduce surprises and discoveries.Thus if young scientiststoday get out and make those observationsin the style of earth explorersof the past, or in a still better style,they will surelyshare in future revelationabout earth. To strengthenthis point further,I note that when I was a student,well beforethe plate tectonicsrevolution, my fellow studentsand I often expe- rienceda senseof discouragementwhen we read a paper that described someearlier earth scientist'striumph. We had comealong too late and we had missedthe days of goldenopportunity, we thought.I distinctlyrecall feelingthat way when I read Gilbert'spaper (Gilbert,1928) on the geology 126 Epilogue of the "Basinand RangeProvince." Perhaps if I had comealong a generation or two earlier,I could have done somethinglike that, I thought. "Was that the lastgreat opportunity?" We askedourselves that questionover and over again.Well, of courseit wasn't and my fellow studentsand I were all dead wrong and pitifully naive to imaginethat all the big discoverieshad alreadybeen made. Plate tectonics was just ahead,and had we beensmart enoughand optimisticenough, perhaps we might havesensed that upcoming revolution,as a few of our leadersdid. In no case,however, should we have lost hope. Basedon that experiencewith those early misperceptions,and the many indicationsof promiseof future greatdiscovery about the earthcited elsewherein this book, I foreseea period of excitingdiscovery in earth sci- encefor at leasta centuryor more.If modernyoung scientistsset visionary goals and pursue them with skill, determinationand perseverance,the excitingstory of the discoveryof plate tectonicswill surelyfade into the backgroundas they report their own astonishingdiscoveries about earth. For the older generationthe era of discoveryof plate tectonicswas a joy to experienceand it remainsa joy to remember,but for the youngergenera- tion it shouldbe but a stimulusand a steppingstone. All of us - old scientists,young scientists,and the public at large - will be better off if it turns out that BIBLIOGRAPHY

Barazangi,M., and J.Dorman, 1969, "World SeismicityMap of the E.S.S.A.Coast and GeodeticSurvey Epicenter Data for 1961-1967,"Seismological Society of America Bulletin,59, pp. 369-380. Bird, J., and B. Isacks(eds.), 1972, Plate Tectonics: Selected Readings from theJournal of GeophysicalResearch, Washington, D.C.: AmericanGeophysical Union. Brown, L. D., M. Barazangi,S. Kaufman and J. Oliver, 1986, "The First Decade of COCORP: 1974-1984,"AGU GeodynamicsSeries 13, Retection Seismotogy: A Global Perspective,pp. 107-120. Bullard, E., J. Everett, and A. Smith, 1965, "The Fit of the Continents Around the Atlantic,"Philosophical Transactions of the Royal Society of London, A-258, pp. 41-51. Carder,D. S., D. Tocher,C. Bufe,S.W. Stewart, J. Eisler,and E. Berg,1967, "Seismic wave arrivalsfrom Longshot,0 ø to 27%' Butt.Seismot. Soc. Am., 57, p. 573. Cleary;J., and A. L. Hales,1966, "An analysisof the travel timesof P wavesto North Americastations, in the distancerange 32ø-100 ø, ' Butt.Seismot. Soc. Am., 56, p. 467. Coats,R. R., 1962,"Magma type and crustalstructure in theAleutian arc," in TheCrust of thePacific Basin, Geophys. Monograph 6, edited by G.A. MacDonaldand H. Kuno,p. 92, AmericanGeophysical Union, Washington,D.C. Daly, R., 1940,Strength and Structure of theEarth, Englewood Cliffs, N.J., 434 pp. Dewey,J., and J. Bird, 1970,"Mountain Belts and the New GlobalTectonics," Journal of GeophysicalResearch, 75, pp. 2625-2647. Dewey, J., 1971, "Origin and Emplacementof the Ophiolite Suite: Appalachian Ophiolitesin Newfoundland,"Journal of GeophysicalResearch, 76, pp. 3179-3206. Dietz, R., 1961,"Continent and OceanBasin Evolution by Spreadingof the SeaFloor," Nature,190, pp. 854-857. Drake,C.L., M. Ewing,and G. Sutton,1958, "Continental Margins and Geosynclines; The EastCoast of North AmericaNorth of Cape Hatteras,"Phys. Chem. Earth, 3, pp. 110-193. Elsasser,W.M., 1967, "Convectionand stresspropagation in the upper mantle," PrincetonUniv. Tech. Rep. 5. Ewing, J.,and M. Ewing, 1967,"Sediment distribution on the mid-oceanridges with respectto spreadingof the sea-floor."Science, 156, p. 1590. Ewing, J.,and E Press,1955, "Geophysical Contrasts Between Continents and Ocean Basins,"in A. Poldervaart(ed.), Crust of theEarth, Geological Society of American SpecialPaper 62. Gilbert,G.K., "Studiesof Basin-RangeStructure," U.S. Geol.Survey Prof. Paper 153, 92 p. Griggs,D., 1939,"A Theoryof MountainBuildings," American Journal of Science,237, pp. 611-650. Gutenberg,B. and C.E Richter,1941, "Seismicity of the Earth, ' Geol.Soc. Amer. Sp. Paper34, Supplementarypaper Geol. Soc. Amer. Butt. 56, pp. 603-688.1949, Princeton UniversityPress. Heezen,B., M. Tharp, and M. Ewing, 1959,The Floors of the Oceans:1. TheNorth Atlantic,Geological Society of America Special Paper 65. Heirtzler, J., G. Dickson, E. Herron, W. Pitman, and X. LePichon, 1968, "Marine MagneticAnomalies, Geomagnetic Field Reversals,and Motionsof the Ocean Floorand Continents,"Journal of GeophysicalResearch, 73, pp. 2119-2136. Herrin, E., and J. Taggart,1966, "Epicenter determinations for Longshot(abstract)," TransAm. Geophys.Union, 47, p. 164. Hess, H., 1960, "The Evolution of Ocean Basins: Report to the Office of Naval Researchon ResearchSupported by ONR Contract,Nonr, 1858,"(10), 38. Holmes,A., 1931,"Radioactivity and EarthMovements", Geological Society of Glascow Transactions,18, pp. 559-606. Holmes,A., 1944,Principles of Physical Geology, New York:Ronald. Honda, H., A. Masatsuka, and K. Emura, 1956, "On the mechanisms of the earthquakesand the stressesproducing them in Japanand its vicinity,"2, Sci.Rept. TohokuUniv., Ser. 5, Geophys.,8, p. 186. Howell, B.E, 1990,An Introductionto SeismologicalResearch, New York; Cambridge University Press. Isacks,B., J. Oliver,and L. Sykes,1968, "Seismology and the New GlobalTectonics," Journalof GeophysicalResearch, 73, pp. 5855-5899. Isacks,B., and P. Molnar, 1969,"Mantle EarthquakeMechanisms and the Sinkingof the Lithosphere,"Nature, 223, pp. 1121-1124. Karig, D., 1970,"Ridges and Trenchesof the Tonga-KermadecIsland Arc System," Journalof GeophysicalResearch, 75, pp. 239-254. Katsumata,M., 1967,"Seismic activites in and nearJapan, 3: Seismicactivities versus depth," (in Japanese),J. Seismol.Soc. Japan, 20, p. 75. Kay, R. N.J. Hubbard and P.W.Gast, 1970,"Chemical Characteristics and Origin of OceanicRidge VolcanicRocks," Journal of GeophysicalResearch, 75 8, pp. 1585- 1613. Kuhn, T., 1970a, The Structureof ScientificRevolutions (second edition), Chicago: Universityof ChicagoPress. LePichon,X., 1968,"Sea-Floor Spreading and ContinentalDrift," Journal of Geophysical Research,73, pp. 3611-3697. McKenzie,D., and R. Parker,1967, "The North Pacific:An Exampleof Tectonicson a Sphere,"Nature, 216,pp. 1276-1280. Molnar, E, and J. Oliver, 1969,"Lateral variation of attenuationin the upper nmatle and discontinuitiesin the lithosphere,"Journal of GeophysicalResearch, 74, pp.2648-2682.. Morgan, W., 1968, "Rises,Trenches, Great Faults,and Crustal Blocks,"Journal of GeophysicalResearch, 73, pp. 1959-1982. Officer, C.B., M. Ewing and P.C. Wuenschel, 1952, "Seismic Refraction Meausurements in the Atlantic Ocean." Part IV: Bermuda, Bermuda Rise and Nares Basin,Bull. Geol.Soc. America, 63, pp. 777-808. Oliver,J., and B. Isacks,1967, "Deep earthquakezones, anomalous structures in the upper mantle,and the lithosphere,"J. Geophys.Res. 72, p. 4259. Oliver,J., and B. Isacks,1968, "Structure and mobilityof the crustand mantlein the vicinityof islandarcs," Can. J. EarthSci., 5. Oliver, J.E., M. Ewing and F. Press, 1955, "Crustal Struc•re and Surface-Wave DispersionPt. IV; Atlanticand PacificOcean Basins," Bull. Geol. $oc. Amer., 60, pp. 913-946. Oliver, J., 1972, "ContributionsOf SeismologyTo Plate Tectonics,"The American AssociationofPetroleum Geologists Bulletin (reprint) 56, pp.214-225. Oliver,J., and B. Isacks,1967, "Deep Earthquake Zones, Anomalous Structures in the Upper Mantle,and the Lithosphere,"Journal of GeophysicalResearch, 72, pp. 4259- 4275. Oliver, J.E.,1993, The Incomplete Guide to theArt of Discover,New York; Columbia University Press. Pih•nan,W., and J. Heirtzler, 1966,"Magnetic Anomalies Over the Pacific-Antarctic Ridge."Science, 154, pp. 1164-1171. Romm,J., 1994, " New Forerunnerfor ContinentalDrift." Nature,Vol. 367,p.407. Rothe,J.P., 1969. "Seismicityof the Earth, 1963-1965",Paris, UNESCO. Runcorn,S., 1962, "Paleomagnefic Evidence for ContinentalDrift and Its Geophysical Cause"in S. Runcorn(ed.), Continental Drift, London:Academic Press, pp. 1-40. Sclater,J., and J. Francheteau,1970, "The Implicationsof TerrestrialHeat Flow Observations on Current Tectonics and Geochemical Models of the Crust and Upper Mantle of the Earth,"Geophysical Journal, 20, pp. 509-542. Sykes,L. R., "eismici• of the southPacific Ocean,:" Journal of Geophysical Res., 68, No. 21, November 21, 1963. Sykes,L.R., 1966, "The Seismici• and DeepStructure of IslandArcs", J. Geophys. Res., 71, 2981. Sykes,L. R., 1967,"Mechanisms of earthquakesand naNre of faultingon mid-ocean ridges,"]. of Geophys,Res., 72, p. 2131. Sykes,L., J. Oliver, and B. Isacks,1968, "Earthquakes and Tectonics,"The Sea, 4, part 1, pp.-353-420. Turcotte,D.L., and E. R. Oxburgh,1968, "A fluid theoryfor the deep structureof dip- slip fault zones,"Phys. Earth Planet Interiors, 1, p. 381. UmbgroveJ.H.E,1947, "The Pulseof the Earth."The Hague; Martinus Nijhoff. Utsu, T., "1967,Anomalies in seismicwave velocityand attenuationassociated with a deepearthquake zone 1,"]. Fac.$ci. Hokkaido Univ. Japan, $er. 7, Geophys.,p. 1. Vening-Meinesz,F.A., 1952,"Convection Currents in the Earth and the Origin of the Continents",Kon. Ned. Akad. Wet. B55, pp. 527-552. Vine, F., and D. Matthews, 1963,"Magnetic Anomalies Over OceanRidges," Nature, 199,pp. 947-949. Wegener,A., 1924,The Origin of Continentsand Oceans, London: Methuen. Wegener,A. F., 1966, The Origin of Continentsand Oceans.New York; Dover Publications, Inc. Wilson,J.T., 1965, "A New Classof Faultsand Their Bearingon ContinentalDrift," Nature,207, pp. 343-347. INDEX

A "bibleof seismicity,"Gutenberg and accretionarywedges, slabs ...... 95 Richter 1949 book ...... 20 air guns,seismic exploration ...... 27-28 Bird, Jack Aleutians,deep seismiczones ...... 83 Jack Oliver's move to Cornell ...... 99 Allen, Clarence,Philippine shocks . .82-83 "Mountain Belts and The New Global Alps, seismicity...... 86 Tectonics" ...... 96 American Association for the blocks,use for lithosphereblocks ...... 76 Advancement of Science, body waves scientific focus ...... 119 travel time ...... 16 AmericanGeophysical Union seealso P waves; S waves; seismic waves 1967meeting ...... 72-73,75, 91 Bolivianorocline, Bryan Isack'swork .100 1992meeting on 25th anniversaryof Bollinger,focal mechanisms ...... 58 plate tectonics...... 114 Bolt, Bruce, seismic studies ...... 50 role in plate tectonicsrevolution 118-119 Brown,Larry, work with JackOliver on seealso Journal of GeophysicalResearch COCORP ...... 99 amplitude,seismic transients ...... 3 Brune, Jim, seismic studies ...... 50 Andean earthquakes,Darwin ...... 2 Bucher,Walter, debate with LesterKing on Anderson, Orson continental drift ...... 36 seismic studies ...... 50 Bullard, E. seismographs...... 6 heat flow ...... 94 Andes,convergent plate structure .... 100 meetingwith Lynn Sykes ...... 57 anisotropy,vs. depthin crust ...... 102 Bullen Appalachians,COCORP studies ...... 112 nuclearexplosions ...... 30 Arabs,earthquakes ...... 1 travel time ...... 16 Arctic Ocean, T-3 field station ...... 63 seealso Jeffreys-Bullen tables arrival times, networks ...... 9 Bureau Centrale Internationale de artificialsources, earthquakes ...... 23-30 Seismologie,France, seismographs ...8 asthenosphere Byerly,earthquakes ...... 19 low velocitylayer ...... 17 seealso crust-mantle system; lithosphere- C asthenospheresystem; mantle; Moho Cal Tech. see California Institute of Atlantic Ocean Technology drift ...... 31-33 California,earthquakes ...... 18 reflection ...... 27 CaliforniaInstitute of Technology, Atlantis,myths ...... 31 seismology...... 9, 15-16 atomicage, seismology ...... 9 CambridgeUniversity ...... 107 Hill ...... 25 seismology ...... 16 B CanadianJournal of EarthSciences, Barazangi,Muawia seminalpapers ...... 72 computerizedmaps ...... 79-80 cartographers,16th century ...... 31-32 Jack Oliver's move to Cornell ...... 99 Chile, seismicity...... 86 Benioff,Hugo, seismographs...... 6, 7 Chilean1835 earthquake ...... 2 Benioff zone. see Wadati-Benioff zone Chinesewriters, earthquakes ...... 1 Berckhemer, Hans, seismic studies ..... 50 Coats,R., paper on seafloor ...... 71 Berkner Panel, networks ...... 10-11 COCORP Bermuda, reflection ...... 27 history ...... 112-113 launchof program ...... 99 deepearthquakes vibroseis ...... 29 Pacific Ocean ...... 61-62, 66-74 coda waves Wadati ...... 21 travel time ...... 17-18 Deep SeaDrilling Project, seealso seismic waves; surface waves seismic reflection ...... 29 Cold War deformation, reflection evidence .... 28-29 fundingmarine research ...... 107 Dewey,John, "Mountain Belts and The seismology...... 9-10 New Global Tectonics" ...... 96 ColumbiaUniversity Dickens, Oliver Twist character name thoughtson continentaldrift in the similarityto authorsof new global 1950s ...... 35-37 tectonics ...... 78 seealso Lamont Geological Observatory; Dietz, R., coiningof the term "sea-floor Lamont-DohertyEarth Observatory spreading"...... 38 communications in science, digitalmethods, seismography ...... 5 goodand bad ...... 115-118 Dobrin,Milton, aspart of MauriceEwing's compression,gravity measurements...94 group ...... 42 compressionalwaves. see P waves DominionObservatory Consortium for Continental Reflection Canada, networks ...... 9 Profiling.see COCORP earthquakes...... 19 continental crust Donn, Bill, as part of Maurice Ewing's frontiers in earth science ...... 110-113 group ...... 42 needfor comprehensiveexploration Dorman, Jim program ...... 111-113 computerizedseismology ...... 79-80 refraction ...... 25 seismic studies ...... 50 continentaldrift, as a forerunnerto plate transform faults ...... 55-56 tectonics ...... 31-40 Drake, Chuck continents,plate dynamics ...... 97 gravity measurements...... 94 convection, mantle ...... 95 marine seismic studies ...... 51 convection cells, Arthur Holmes ...... 70 as part of MauriceEwing's group ...42 convergentmargins dynamicearth ...... 105 deepearthquakes ...... 81, 85 partialmelting ...... 96 E core, seismic waves ...... 16 earth interior, seismic waves ...... 16-18 core, inner, frontiers in earth science . .110 earth science, frontiers ...... 110-113 core-mantleboundary, frontiers in earth earthquakeprediction, work at Columbia science ...... 110 University ...... 99 CornellUniversity, Jack Oliver's move .99 earthquakeprogram, Lamont Geological cosmicrays, instrumentation ...... 1 Observatory...... 48-49 crust, seismic waves ...... 16 earthquakeseismolog59 early daysat crust-mantlesystem LamontGeological Observatory . .41-52 nomenclature ...... 70 earthquakes seealso lithosphere-asthenosphere; as events ...... 123 mantle instrumentation ...... 1 seealso deep earthquakes D earthquakes,continental, Darwin, earthquakes...... 2 occurrencedepth ...... 102 data bases,seismograms ...... 121-123 Edwards,Dick, as part of MauriceEwing's Davidson,Maurice, geomagnetism .... 46 group ...... 42 deductivestyle, see also hypothetico-deduc- electronicpublication tive method;inductive style; serendipity changes...... 116-118 deepdrilling, continental crust .... 112-113 innovationin peer review ...... 114-115 Elsasser,Walter, theoretical model of earth Gamow,George, "The Biographyof the tectonics ...... 72, 77 Earth" ...... 34 eurekaphenomenon geodeticstudies, modern studies ..... 120 in science ...... 69 geologicalstyle, continental seealso serendipity crust ...... 120-121 Ewing,John geology explosives...... 27 interactionwith seismology.... 120-121 marine seismic studies ...... 51 supportingevidence for plate aspart of MauriceEwing's group ...42 tectonics ...... 96 Ewing, Maurice geomagnetism as directorof LamontGeological LamontGeological Observatory .... 46 Observatory;beginning ...... 42-43 oceanic crust ...... 92 earlyopposition to seafloor geophysicalobservations, World War II107 spreading ...... 44 geophysicalstyle, continental LamontGeological Observatory . .41-45 crust ...... 120-121 philosophy ...... 42 geophysicaltechniques, application to pioneeringwork ...... 25-26 continental crust ...... 111-113 asprofessor ...... 34 Gilbert,G. K. "Studiesof Basin-Range reflection ...... 27 Structure" ...... 125-126 seismographs...... 6 Gilluly,Jim, name similarity of authorsof seealso Press-Ewing seismographs new globaltectonicsto Dickens extension,gravity measurements...... 94 characters ...... 78 glacialrebound, Pleistocene ...... 16-17 globalnetworks F seismographs...... 2, 8 fault planes,San Francisco 1906 seealso standardized long-period global earthquake ...... 18 networks; World Wide Standardized faults, movement ...... 82-83 SeismographNetwork Fiji globalseisrcdc belts, patterns ...... 20-21 stations ...... 67 global synthesis,oceanic crust ...... 92 seealso Tonga-Fiji deep earthquake globaltectonics project mountain belts ...... 96 focal mechanisms useinstead of platetectonics ...... 76 globaldistribution ...... 19 grabens,sediments ...... 84 modern studies ...... 120 Greekphilosophers, earthquakes ...... 1 shocks ...... 56 Griggs,David spatialpatterns ...... 81-82,85-86, 100 mantle convection ...... 73 Foster,John, geomagnetism ...... 46-47 networks ...... 10 fracture zones groundmotion patterns ...... 54-55 earthquakes...... 2 transform faults ...... 55-56 instruments ...... 3 Francheteau,J., oceanicrocks ...... 94 vs. time ...... 13-14 free oscillations, earth interior ...... 18 Gulf of Aden, seismicity ...... 86 funding Gulf of California,seismicity ...... 86 bureaucracies ...... 114 Gutenberg,Bruno increase ...... 10 1949 book ...... 20 seismology...... 6 low velocitylayer ...... 17, 70 travel time ...... 15-16 guyots G surveysby Harry Hess ...... 37 Galitzin,seismographs ...... 6 see also seamounts H inertialmass, seismographs ...... 4-5 Hade, George innovation in science,deficiencies and engineeringstudies ...... 50, 66 solutions ...... 113-115 Jack Oliver's move to Cornell ...... 99 instruments Hamilton,Gordon, as part of Maurice history ...... 3 Ewing's group ...... 42 seealso seismographs Heacock,Jack, as part of MauriceEwing's internationalcenters, seismographs ...8-9 group ...... 42 InternationalGeophysical Year heat conductivity,earthquakes ...... 71 internationalcooperation ...... 65 heat flow, sea floor ...... 94 networks ...... 10 Heezen, Bruce worldwide seismographnetworks . .63 aspart of MauriceEwing's group ...42 internationalpolitics physiographicmap of oceanfloors36-37 seismology ...... 10 Heirtzler, Jim seealso Cold War; World War I; World geomagnetism...... 46-47 War II magneticanomalies ...... 57 InternationalSeismological Center. see Helgoland,explosives ...... 24 InternationalSeismological Summary Herron, Ellen,geomagnetism ...... 46 InternationalSeismological Summary, Hersey,Woods Hole Oceanographic London,seismographs ...... 8 Institution ...... 25 InternationalSymposium on Continental Hess,Harry, rift systems...... 37-38 Marginsand IslandArcs, Zurich .... 72 high-resolutionmethods, Irving, Ted,paleomagnetism ...... 46 deepdrilling ...... 113 Isacks,Bryan Hill, CambridgeUniversity ...... 25 deepearthquake zones ...... 65-66,100 history,dangers in publishing JackOliver's move to Cornell ...... 99 and reading ...... 125-126 LamontGeological Observatory .... 39 Hodgson,John recruitment ...... 63 earthquakes...... 19 Tonga-Fijiprogram ...... 64-66 focal mechanisms ...... 58 seealso "Seismology and the New Global Holmes, Arthur Tectonics" communication in science ...... 115 Isacks-Oliver-Sykespaper convection cells ...... 95 publication ...... 76 ideas on ocean basins ...... 38 seealso "Seismology and the New Global Honda, H. Tectonics" earthquakes...... 18-19 ISC. seeInternational Seismological seismic zones ...... 73 Summary hotspots,age and cause ...... 93-94 island arcs Houtz, Robert, marine seismic studies..51 convergentfeatures ...... 97 hydrocarbons,exploration ...... 103 deepearthquakes ...... 21 hypocenters,relocation by deep seismiczones ...... 83 Lynn Sykes ...... 54 earthquakes...... 61-62,66-70 hypotheses definition ...... 51-52 testing ...... 53-60,106 J hypothesistesting, vs. serendipity.... 109 Japan hypothetico-deductivemethod ...... 106 deepearthquakes ...... 21 earthquakes...... 18-19 I JapaneseMeteorological Agency, inductivestyle networks ...... 9 importancein science...... 107-109 jargonin science,use and misuse ..... 116 seealso deductive style; serendipity JBtables. see Jeffreys-Bullen tables Jeffreys,Harold LePichon, Xavier oppositionto plate tectonics...... 95 directionsof slip ...... 82 travel time ...... 16 geomagnetism ...... 46, 77 seealso Jeffreys-Bullen tables lithosphere Jeffreys-Bullentables, travel time ...16, 30 mobility ...... 72 JOIDESprogram, sea floor studies . .92-93 slabs ...... 95 Journalof GeophysicalResearch thickness ...... 101-102 seminalpapers .57,72-74, 76, 96, 118-119 lithosphere-asthenospheresystem seealso American Geophysical Union nomenclature ...... 70 journalsin science seealso crust-mantle system communications in science ...... 116 Love waves,discovery ...... 17-18 seealso Canadian Journal of Earth Sciences; low velocitylayer Journalof GeophysicalResearch; Gutenberg ...... 70 Transactionsof the Geological Society of mantle ...... 17 Glasgow Ludwig, Bill, marine seismicstudies ...51 Luskin,Bernie, as part of MauriceEwing's K group ...... 42 Karig, Dan island arcs ...... 97 JackOliver's move to Cornell ...... 99 M Katsumata, seismic zones ...... 73 McKenzie, D. Katz, Sam,as part of MauriceEwing's meetingwith Lynn Sykes ...... 57 group ...... 42 seismology...... 77 Kaufman,Sidney magneticanomalies COCORP ...... 112-113 confirmationby drilling data ..... 92-93 work with JackOliver on COCORP .99 sea floor ...... 38 Kay,Robert symmetryon seafloor ...... 46 Jack Oliver's move to Cornell ...... 99 mantle partialmelting ...... 96 convection ...... 95 King, Lester,South African geology .... 36 flow ...... 17 Kuhn,T., new paradigms ...... 87 heterogeneity ...... 101 Kuo, John, seismic studies ...... 50 low velocitylayer ...... 17 warping ...... 83 seealso asthenosphere; core-mantle L boundary;crust-mantle system; lithos- Lake,ideas on shearplanes ...... 45 phere-asthenospheresystem; Moho LamontGeological Observatory mantle,uppermost, frontiers in beginning ...... 42-43 earth science ...... 110-113 earthquakeseismology ...... 39 maps,computerized seismicity ..... 79-80 originand earlydays ...... 41-52 marinegeology, seismic exploration .27-28 seismograms...... 10 Matthews, D. seeVine-Matthews seealso Lamont-Doherty Earth hypothesis Observatory Maxwell, heat flow ...... 94 Lamont group, spirit ...... 51 mid-oceanridges Lamont-DohertyEarth Observatory patterns ...... 54-55 JackOliver's move to Cornell ...... 99 seismic belts ...... 37 Maurice Ewing ...... 25 Milne, seismographs...... 6 seealso Lamont Geological Observatory Minear,thermal history of lithosphere..97 lecturesin science,the new MineralogicalSociety of America,scientific paradigm ...... 98-99 focus ...... 119 Lehmann,Inge, seismicstudies ...... 50 mobilisim ...... 105 Moho, re&action ...... 25 Officer,Chuck, as part of MauriceEwing's Mohorovicic, A., refraction ...... 25 group ...... 42 Molnar, Peter, seismic wave Oliver, Jack propagation...... 89 associationwith MauriceEwing .... 44 Morgan, Jason earthquakeseismology ...... 48-49 geometricalmodel basedon Euler's LamontGeological Observatory .... 39 theorem ...... 77 aspart of MauriceEwing's group ...42 hotspotcause ...... 93-94 surface waves ...... 64 paper on globalconfiguation thoughtson continentaldrift in the of plates ...... 73 1950s ...... 34-37 mountainbelts, global tectonics ...... 96 seealso Isacks-Oliver-Sykes paper; Mueller,Stephan, seismic studies ...... 50 "Seismologyand the New Global Tectonics" N OliverTwist, character name similarityto Nafe, Jack authorsof new globaltectonics ..... 78 marine seismic studies ...... 51 Oliver-Isackspaper, 1967 study on deep seismic studies ...... 50 earthquakes...... 73-74 networks,see also global networks; stan- Opdyke,NeiI, paleomagnetism...... 46 dardizedlong-period global networks; ophiolites,as part of seafloor ...... 96 WorldWide StandardizedSeismograph organizationsin science, Network continental crust ...... 113 new globaltectonics Ortelius, Abraham, Thesaurus geometry ...... 77 Geographicus ...... 31-33 seealso global tectonics; plate tectonics oscillations,seismographs ...... 4 New Hebridesarc, deep earthquakes .100 Oxburgh,Ron, convection...... 95 New JerseyZinc Company,mine in Ogdenburg,New Jersey ...... 66 NGT. seenew globaltectonics; "Seismology P and the NewGlobal Tectonics" P waves noise, instruments ...... 6 earthquakes...... 18-19 nucleartest ban treaties,seismology ...10 travel ...... 14 nuclear tests seealso body waves;S waves;seismic detection ...... 29-30 waves seismology ...... 9 PacificOcean, earthquakes ...... 54-55 work by Lynn Sykes ...... 99-100 paleomagnetism LamontGeological Observatory .... 46 O observations ...... 35 ocean basins PaleontologicalSociety, scientific focus 119 Harry Hessproposal of hypothesis. .38 paleoseismicity,modem studies...... 120 history ...... 28 Palisadesdiabase sill, Lamont Geological new models ...... 26 Observatory ...... 47 ocean floors Pangaea,splitting ...... 97 observations ...... 35-38, 107 paradigmshift ...... 105-106 physiographicmap ...... 36-37 Kuhn'snew paradigms...... 87 oceanic crust Parker,seisinology ...... 77 convergentplate boundaries ...... 92 partial melting, oceaniccrust ...... 96 hyopthesisby Harry Hess...... 37 peerreview in science,deficiencies and partial melting ...... 96-97 solutions ...... 113-115 refraction and reflection ...... 25-30 peridotite,serpentinized, Harry Hess sediments ...... 26 hypothesison crust ...... 37 thickness ...... 26 PhilippineFault, shocks...... 82-83 physiographicfeatures, confirmation of Rothe,J.p., maps ...... 79 plateboundaries ...... 93 rupture,earthquakes ...... 18-19 Pitman,Walter, geomagnetism ...... 46 plate motion s ratesas proposed by Mauriceand John s waves Ewing ...... 4445 travel time ...... 15 velocity ...... 88-89 seealso body waves;P waves;seismic plate tectonics waves comingof age ...... 91-104 S waves,high-frequency, Tonga-Fiji deep forerunners ...... 31-40 earthquakes...... 68-69 "icons"...... 78 St. LouisUniversity, focal mechanisms .82 originof the term ...... 96 San Andreas Fault pre-1960s...... 2 SanFrancisco 1906 earthquake ...... 18 use in the late 1960s ...... 76 seismicity...... 86 plate tectonicsrevolution SanFrancisco 1906 earthquake, causes .18 development ...... 106-107 Sato, Yasuo, seismic studies ...... 50 senseof discovery...... 125-126 Scandinavia,glacial rebound ...... 16-17 Plato, tale of Atlantis ...... 31 scientificdiscovery plumes,hotspots ...... 94 plate tectonicsrevolution as Pomeroy,Paul model ...... 107-110 computerizedseismicity ...... 80 young scientists...... 125-126 seismic studies ...... 50 scientific observations, relation to Press, Frank hypothesismaking ...... 106-110 networks ...... 10 scientific societies aspart of MauriceEwing's role in plate tectonics group ...... 42, 4748 revolution ...... 118-119 seismographs...... 6 see also American Association for the Press-Ewingseismographs ...... 63 Advancement of Science;American Ptolemy,maps ...... 32 GeophysicalUnion; Mineralogical public servicein science,science Societyof America;Paleontological lectures ...... 98-99 Society;Seismological Society of publicationsin science,deficiencies and America solutions ...... 113-115 Sclater,J., oceanic rocks ...... 94 ScrippsInstitute of Oceanography.... 107 R Raitt ...... 25 radioactivity,instrumentation ...... 1 seafloor spreading Raitt,Scripps Institute of Oceanography early reflectionmisleading evidence .28 Rayleighwaves, discovery ...... 17-18 as a forerunnerto platetectonics .31-40 reflection Harry Hessproposal of hypothesis. .38 artificial sources ...... 24-25 ocean basins ...... 26 see also seismic reflection R. Dietz coiningof term ...... 38 refraction seamounts,see also guyots artificial sources ...... 24-25 sediments,data fit with plate see also seismic refraction tectonics ...... 103 Revelle, heat flow ...... 94 seismicbelts, as part of oceanridges ...37 rheology,slabs ...... 95 seismicexploration, methods ...... 27-28 Richter, C. E seismicgap theory,work at Columbia 1949 book ...... 20 University ...... 99 travel time ...... 15-16 seismic reflection rift systems,postulation by MauriceEwing, COCORP ...... 112-113 BruceHeezen and Marie Tharp . .37-38 continental crust ...... 121 resolution ...... 84 shapeof earth,effect on velocityof seismic seismic reflection and refraction, continen- waves ...... 14-15 tal crust ...... 111-112 shear waves. see S waves seismicwave propagation,descending shocks,earthquakes ...... 13 slabs ...... 101 sphericalsymmetry seismic waves earth ...... 14-15 behavior ...... 3 earth model ...... 67-68 propagation...... 89 plate tectonics...... 87 seismic record ...... 14, 16-18 standardization,seismographs ...... 7 seealso body waves;coda waves; P standardizedlong-period global networks waves; S waves; surface waves InternationalGeophysical Year ...10-11 seismiczones, deep, Japan ...... 83 see also World Wide Standardized seismicity,development of plate SeismographNetwork static earth..105 tectonics ...... 20-22 Stauder, focal mechanisms ...... 58, 82, 84 seismicity,global, support for plate Steenland,Nelson, as part of Maurice tectonics ...... 79-82 Ewing'sgroup ...... 42 seismicitymaps ...... 79-81 subduction seismograms evidence ...... 29 archives ...... 10-11 origin of the term ...... 96 data bases ...... 121-123 subduction zones seismographs downgoingslabs ...... 84 analogywith technologicalchange ...5 earthquakes ...... 61-62,66-70 availabilityin an ideal world ...123-124 first formulationof concept ...... 69 description...... 3-4 surface waves, seealso coda waves; Love diversity ...... 6 waves;Rayleigh waves globalnetworks ...... 2 Sutton,George, seismic studies ...... 50 historyof development ...... 2-5 Sykes,Lynn standardization ...... 7 1966paper at NASA Goddard timing ...... 7-8 Symposiumin New York ...... 57 see also instruments deepearthquakes ...... 65-66 SeismologicalLaboratory of CalTechand focal mechanisms ...... 58, 82 Universityof California,Berkeley, net- hypothesis...... 53-60 works ...... 9 LamontGeological Observatory..39, 47 SeismologicalSociety of America, seismotectonics at Columbia scientific focus ...... 119 University ...... 99 seismologicaltests, transform faults .... 39 Tuzo Wilson's transform faults ...... 56 seismology seealso Isacks-Oliver-Sykes paper; 1970s ...... 98-101 "Seismologyand the New Global interactionwith geology ...... 120-121 Tectonics" opportunities ...... 120-124 synthesis "Seismologyand the New Global in science ...... 75-90 Tectonics" scientificprogress ...... 109-110 paperpublication and subsequenteffects ...... 76-90,98-100 seealso Isacks-Oliver-Sykes paper T seismotectonics,work at Columbia Talwani,gravity measurements...... 94 University ...... 99 tectosphere,W. Elsasser'shypothesis ...72 serendipity teleseismicevents, seismographs ...... 6 placein science...... 61-74 terranes, continents ...... 97 vs. hypothesis-testing...... 109 Tharp,Marie, physiographicmap of ocean seealso eureka phenomenon floors ...... 36-37 Theatrum Orbis Terrarium, Abraham V Ortelius ...... 32 velocity,horizontal uniformity ...... 14 ThesaurusGeographicus, VeningMeinesz, E A. 16th century ...... 31-32 gravitymeasurements ...... 94 Thompson,George, seismic studies .... 50 plasticflow ...... 46 three-dimensional studies, vibroseis, COCORP ...... 29 continental crust ...... 121 Vine, E, meetingwith Lynn Sykes ..... 57 timing,seismographs ...... 7 Vine-Matthewshypothesis, magnetic Toksoz,thermal history of lithosphere..97 reversals ...... 38-39 Tolstoy,Ivan, aspart of MauriceEwing's viscosity,mantle ...... 17 group ...... 42 volcaniceruptions, earthquakes ...... 2 tomography volcanoes,global distribution ...... 93 mantle heterogeneity...... 101 modern studies ...... 120 W Tongaouter arc,stations ...... 67 Wadati,deep earthquakes ...... 21 Tongatrench, stations ...... 67 Wadati-Benioffzone, deep Tonga-Fijideep earthquake project .... 55 earthquakes...... 21 history ...... 61-62,66-74, 100 wave trains, surface waves ...... 17-18 Tonga-Kermadecarc, seismicity .... 54-55 Wegener,Alfred Tonga-Kermadectrench, sediment communication in science ...... 115 distribution ...... 97 "father of continental drift" ...... 33-34 topography,relation to geodynamics. .100 Wegener'shypothesis, testing ...... 107 Transactionsofthe Geological Society of Westernscholars, earthquakes ...... 1 Glasgow,communications in science.115 Wiechert,seismographs ...... 6 transcurrent faults, offsets ...... 56 Wilson, Tuzo transform faults, Tuzo Wilson's hotspots...... 93 hypothesis...... 39, 55-56 integrationof ideas ...... 73 travel time, seismic waves ...... 14-15 paperon transformfaults ...... 55-56 trench formation, extensional model ...83 sea-floorspreading ...... 38-39,47 tsunamis,generation in relationto plate Wood,seismographs ...... 6 model ...... 85 WoodsHole OceanographicInstitution 107 Turcotte, Don Hersey ...... 25 convection ...... 95 WorldWar I, explosives...... 24, 27 moveto geologyat Cornell World War II University ...... 99 geophysicalobservations ...... 107 two-dimensional studies, sea-floor studies ...... 35 continental crust ...... 121 World Wide StandardizedSeismograph Network U archives ...... 11, 43 Umbgrove,J. H. E, observations ...... 19 "Pulse of the Earth" ...... 45 seealso standardized long-period global underthrust faults, subduction ...... 69 network underthrusting,models ...... 88 Worzel, Joe Upper MantleProgram gravitymeasurements ...... 94 InternationalGeophysical Year ...... 65 marine seismic studies ...... 51 National Science Foundation ...... 66 aspart of MauriceEwing's group ...42 U.S. Coastand GeodeticSurvey Wuenschel,Paul, as part of Maurice networks ...... 9 Ewing'sgroup ...... 42 seismograms...... 11 WWSSN. see World Wide Standardized Utsu, T., seismic zones ...... 73 SeismographNetwork APPENDIX

"Seismologyand the New Global Tectonics"

Bryan Isacks,Jack Oliver and Lynn Sykes JOURNAl,OF GEOPHYSICAl.RZS•AUCH VOL. 73, No. 18, Szr?z•aszR 15, 1968

Seismologyand the New Global Tectonics

BRYAN ISACKS AND JACK OLIVER

Lamont GeologicalObservatory, Columbia University, Palisades,New York 10964

LYNN R. SYKES •

Earth Sciences Laboratories, ESSA Lamont GeologicalObservatory, Columbia University, Palisades,New York 10964

A comprehensivestudy of the observationsof seismologyprovides widely based strong supportfor the new global tectonicswhich is foundedon the hypthesesof continentaldrift, sea-floor spreading,transform faults, and underthrustingof the lithosphere at island arcs. Although further developmentswill be requiredto explain certain part of the seismological data, at presentwithin the entire field of seismologythere appear to be no ser(ousobstacles to the new tectonics.Seismic phenomena are generally explainedas the result interactions and other processesat or near the edgesof a few large mobile plates of lithospherethat spread apart at the ocean ridges where new surficial materials arise, slide past one another along the large strike-slip faults, and convergeat the island arcs and arc-like structureswhere surficiM materials descend.Study of world seismicityshows that most earthquakesare confinedto narrow continuous belts that bound large stable areas. In the zones of divergence and strike-slip motion, the activity is moderate and shallow and consistent with the transform fault hypothesis; in the zones of convergence, activity is normally at shallow depths and includes intermediate and deep shocks that grossly define the present configuration of the down-going slabs of lithosphere. Seismic data on focal mechanismsgive the relative direction of motion of adjoining plates of lithosphere throughout the active belts. The focal mechanismsof about a hundred widely distributed shocks give relative motions that agree remarkably well with Le Pichon's simplified model in which relative motions of six large, rigid blocks of lithosphere covering the entire earth were determined from magnetic and topographic data associatedwith the zones of divergence.In the zones of convergencethe seismicdata provide the only geophysical information on such movements. Two principal types of mechanismsare found for shallow earthquakesin island arcs: The extremely active zone of seismicity under the inner margin of the ocean trench is characterized by a predominance of thrust faulting, which is interpreted as the relative motion of two converging plates of lithosphere; a less active zone in the trench and on the outer wall of the trench is characterizedby normal faulting and is thought to be a surficial manifestation of the abrupt bending of the down-going slab of lithosphere. Graben-like structures along the outer walls of trenchesmay provide a mechanismfor including and transporting sedimentsto depth in quantities that may be very significantpetrologically. Large volumes of sediments beneath the inner slopes of many trenches may correspond, at least in part, to sediments scrapedfrom the crust and deformed in the thrusting. Simple underthrustingtypical of the main zone of shallow earthquakesin island arcs does not, in general, persist at great depth. The most striking regularity in the mechanismsof intermediate and deep earthquakesin several arcs is the tendency of the compressionalaxis to parallel the local dip of the seismic zone. These events appear to reflect stressesin the relatively strong slab of down-goinglithosphere, whereas shearing deformations parallel to the motion of the slab are presumably accommodatedby flow or creep in the adjoining ductile parts of the mantle. Several different methods yield average rates of underthrustingas high as 5 to 15 cm/yr for some of the more active arcs. These rates suggestthat temperatureslow enoughto permit dehydration of hydrousminerals and hence shear fracture may persist even to depthsof 700 km. The thicknessof the seismiczone in a part of the Tonga arc where very precisehypocentral locations are availableis lessthan about20 km for a wide rangeof depths. Lateral variations in thickness of the lithosphere seem to occur, and in some areas the lithospheremay not include a significantthickness of the uppermostmantle.

Lamont Geological Observatory Contribution 1234. Order of authors determined by lot. 5855 5856 ISACKS, OLIVER, AND SYKES The lengthsof the deep seismiczones appear to be a measureof the amount of under thrustingduring about the last 10 m.y. Hence,these lengths constitute another 'yardstick' for investigationsof globaltectonics. The presenceof volcanism,the generationof manytsunamis (seismicsea waves), and the frequencyof occurrenceof large earthquakesalso seem to be related to underthrustingor rates of underthrustingin island arcs. Many island arcs exhibit a secondarymaximum in activity which variesconsiderably in depth amongthe variousarcs. These depthsappear, however, to correlatewith the rate of underthrusting,and the deep maximaappear to be locatednear the leading(bottom) part of the down-goingslab. In some casesthe down-goingplates appearto be contorted,possibly because they are encounteringa more resistantlayer in the mantle.The interactionof platesof lithosphereappears to be more complexwhen all the platesinvolved are continentsor piecesof continentsthan when at least one plate is an oceanicplate. The new globaltectonics suggests new approachesto a variety of topicsin seisinologyincluding earthquake prediction, the detectionand accurate locationof seismicevents, and the generalproblem of earth structure.

INTRODUCTION [1968], who pursuedthis conceptfurther by This paper relatesobservations from the investigatingthe relative motion in plan of field of seismologyand allied disciplinesto largeblocks of lithosphere. what is here termed the 'new global tectonics.' Figure I is a block diagram illustrating some This term is used to refer in a generalway to of the principalpoints of the mobilelithosphere current conceptsof large-scaletectonic move- hypothesis.In a relatively undisturbedsection, mentsand processeswithin the earth,concepts three fiat-lying layers are distinguished:(1) that are basedon the hypothesesof continental the lithosphere,which generally includes the drift [Wegener, 1966], sea-floor spreading crust and uppermost mantle, has significant [Hess, 1962; Dietz, 1961], and transform strength, and is of the order of 100 km in faults [Wilson, 1965a] and that includevar- thickness; (2) the asthenosphere,which is a ious refinementsand developmentsof these layer of effectivelyno strengthon the appro- ideas.A comprehensiveview of the relationship priate time scale and which extendsfrom the betweenseismology and the new global tec- base of the lithosphereto a depth of several tonicsis attempted,but there is emphasison hundred kilometers; and (3) the mesosphere, data from earthquakeseismology, as opposed which may have strengthand which makesup to explosionseismology, and on a particular the lower remainingportion of the mantle and version of the sea-floorspreading hypothesis is relatively passive,perhaps inert, at present, in which a mobile, near-surface layer of in tectonic processes.(Elsasser refers to the strength,the lithosphere,plays a key role.Two lithosphereas the tectosphereand definessome basic questionsare considered.First, do the other terms somewhat differently, but the observations of seismologysupport the new terminology of Daly [1940] is retained here. globaltectonics in someform? To summarize The term 'strength,' which has many defini- briefly, they do, in general,give remarkable tions and connotations,is usedhere, following supportto the newtectonics. Second, what new Daly, in a general senseto denote enduring approachesto the problemsof seismologyare resistanceto a shearingstress with a limiting suggestedby the new global tectonics?There value.) The boundariesbetween the layers are many; at the very least the new global may be gradational within the earth. The tectonicsis a highly stimulating influence on thenospherecorresponds more or less to the the field of seismology;very likely the effect low-velocity layer of seismology;it strongly will be one of revolutionary proportions. attenuates seismic waves, particularly high- The mobile lithosphere concept is based frequency shear waves. The lithosphere and partly on an earlierstudy [Oliver and Isacks, the mesospherehave relatively high seismic 1967], but, as presentedhere, it incorporates velocitiesand propagate seismicwaves without ideas from Elsasset [1967], who independently great attenuation. developeda modelwith many similarfeatures At the principal zonesof tectonicactivity based on entirely different considerations,and within the earth (the oceanridges, the island ideas from Morgan [1968] and Le Pichon arc or island-arc-likestructures, and the ma•or SEISMOLOGY AND NEW GLOBAL TECTONICS 5857

A S T H E N 0 S P H E

M E S 0 S P H E R E

Fig. 1. Block diagram illustrating schematically the configurationsand roles of the lithosphere, asthenosphere,and mesospherein a version of the new global tectonicsin which the lithosphere•a layer of strength,plays a key role. Arrows on lithosphereindicate relative movementsof adjoining blocks.Arrows in asthenosphererepresent possible compensating flow in responseto downwardmovement of segmentsof lithosphere.One arc-to-arc transform fault appears at left between oppositely facing zones of convergence(island arcs), two ridge-to- ridge transform faults along ocean ridge at center, simple arc structure at right.

strike-slip faults) the lithosphere is discon- settled, beneath an island arc or a continental tinuous; elsewhereit is continuous.Thus, the margin that is currently active. At an inactive lithosphere is composed of relatively thin margin the lithospherewould be unbroken or blocks, some of enormoussize, which in the healed.The left side of the diagram showstwo first approximationmay be consideredinfinitely island-arc structures,back to back, with the rigid laterally. The major tectonic features are lithosphereplunging in a different direction in the result of relative movement and interaction each case and with a transform fault between of theseblocks, which spreadapart at the rifts, the structures. Whereas the real earth must slidepast one anotherat large strike-slipfaults, be more complicated,particularly at this back- and are underthrust at island arcs and similar to-back structure, this figure represents, in structures. Morgan [1968] and Le Pichon a general way, a part of the Pacific basin in- [1968] have demonstratedin a generalway and cluding the New Hebrides, Fiji, Tonga, the with remarkable success that such movement East Pacific rise, and western South America. is self-consistent on a worldwide scale and that The counterflowcorresponding to movement the movementsagree with the pattern of sea- of the lithosphereinto the deeper mantle takes floor spreadingrates determinedfrom magnetic place in the asthenosphere,as indicated sche- anomalies at sea and with the orientation of matically by the appropriate arrows in the oceanic fracture zones. McKenzie and Parker figure. To what extent, if any, there is flow of [1967] used the mobile lithosphereconcept to the adjoining upper part of the asthenosphere explain focal mechanismsof earthquakes,vol- in the same direction as the overlying litho- canism, and other tectonic features in the sphereis an important but open question,par- northern Pacific. tially dependenton the definitionof the bound- Figure I also demonstratesthese concepts ary. A key point of this model is that the in block diagram form. Near the center of the pattern of flow in the asthenospheremay largely figure the lithospherehas been pulled apart, be controlledby the configurationsand motions leaving a pattern of ocean ridges and trans- of the surface plates of lithosphere and not by form faults on the surface and a thin litho- a geometricalfit of convectioncells of simple sphere thickening toward the flanks beneath shape into an idealized model of the earth. the ridge as the new surface material cools It is tempting to think that the basic driving and gains strength. To the right of the dia- mechanismsfor this processis gravitational in- gram the lithospherehas been thrust, or has stability resulting from surface cooling and 5858 ISACKS, OLIVER, AND SYKES hence a relatively high density of near-surface sible to begin consideringthe data, but in this mantle materials. Thus, convectivecirculation process it soon becomes evident that much in the upper mantle might occuras thin blocks more detailed information on the earth is of lithosphere of large horizontal dimensions available and that the hypothesisand the earth slide laterally over large distancesas they de- model can be developedmuch further. These scend; a compensatingreturn flow takes place developmentsare presented later in the text in the asthenosphere.The processin the real as the relevant data are discussed. earth must be more complexthan this simple This paragraph gives a brief review of some model, however. The reader is referred to of the developmentsleading to the new global Elsasset[1967] for a discussionof many points tectonics. A number of contributions vital to relating to this problem. the development of the current position on Alternatively, the surfaceconfiguration might this topic are cited, but the review is not in- be taken as the complicatedresponse of the tended to be comprehensive.The literature strong lithosphere to relatively simple con- bearing on this topic is voluminous,is wide- vection patterns within the asthenosphere. spread in spaceand time, and differs in degree Thus, the basic question whether the litho- of relevance,so that a thorough documentation sphere or the asthenospheremay be thought of its developmentis a job for a historian,not of as the active element, with the other being a scientist. The hypothesisof continental drift passive, is not yet resolved. Probably, how- had a substantial impact on the field of geol- ever, there has been a progressivethinning of ogy when it was proposedin 1929 by Wegener the convective zone with time, deeper parts [1966], but until recently it had not received of the mantle having also been involved dur- general acceptance, largely because no satis- ing early geologictime. factory mechanismhad been proposedto ex- Figure 2, adapted from Le Pichon [1968] plain the movement,without substantialchange with additions, shows the plan of blocks of of form, of the continentsthrough the oceanic lithosphere as chosen by Le Pichon for the crust and upper mantle. When many new data spherical earth and indicates how their move- became available, particularly in the fields of ments are being accommodatedon a worldwide marine geology and geophysics,Hess [1962] scale.The remarkably detailed fit between this and Dietz [1961] proposed that the sea floor scheme,based on a very small number of rigid was spreading apart at the ocean ridges so blocks of lithosphere (six) and the data of a that new 'crust' was being generated there number of fields, is very impressive.The num- while older 'crust' was disappearing into the ber and configuration of the blocks of litho- mantle at the sites of the ocean trenches. The sphere is surely larger than six at present and driving mechanism for this spreading was almost certainly the pattern has changedwithin thought to be convection within the mantle. geologictime, but the present pattern must, The remarkable successwith which the hy- in general, be representative of at least the pothesis of sea-floor spreading accommodated Quaternary and late Tertiary. The duration such diversegeologic observations as the linear of the current episode of sea-floor spreading magnetic anomaliesof the ocean [Vine and is not known. Some evidence suggeststhat it Matthews, 1963; Pitman and Heirtzler, 1966], beganin the Mesozoicand has continuedrather the topography of the ocean floor [Menard, steadily to the present. Other evidence[Ewing 1965], the distribution and configuration of and Ewing, 1967] indicates that the most re- continental margins and various other land cent episodeof spreadingbegan about 10 m.y. patterns [Wilson, 1965a; Bullard, 1964; Bul- ago. This suggestionis consideredhere because lard et al., 1965], and certain aspectsof deep- it opensnew possibilitiesfor explainingcertain sea sediments[Ewing and Ewing, 1967] raised seismological observations, particularly the this hypothesisto a level of great importance configurationof the deep earthquake zones. and still greater promise.The contributionsof Other explanationsfor such evidenceare also seismologyto this developmenthave been sub- considered,however. stantial, not only in the form of generalinfor- With this one very simple version of the mation on earth structure but also in the form new global tectonicsas backgroundit is pos- of certain studies that bear especiallyon this SEISMOLOGY AND NEW GLOBAL TECTONICS 5859

x w 5860 ISACKS, OLIVER, AND SYKES hypothesis.Two specificexamples are $ykes's hypotheseson global tectonics,for example, [1967] evidence on seismicity patterns and the expandingearth and the contractingearth focal mechanismsto support the transform hypotheses,have been far less satisfactoryin fault hypothesisof Wilson [1965a] and Oliver explaining seismologicalphenomena. and Isacks's [1967] discovery of anomalous Certainly the most important factor is that zonesthat appear to correspondto underthrust the new globaltectonics seem capable of draw- lithospherein the mantle beneath island arcs. ing together the observations of seismoIogy There are many important seismologicalfacts and observationsof a host of other fields,such that are so apparent that they are commonly as geomagnetism,marine geology,geochemistry, accepted without much concern as to their gravity, and various branchesof land geology, origin; they fall into place remarkably well under a single unifying concept. Such a step under the new global tectonics.For example, is of utmost importance to the earth sciences the general pattern of seismicity,which con- and will surely mark the beginningof a new sists of a number of continuous narrow active era. belts dividing the earth's surface into a num- In the remainderof this paper, the relation- ber of stable blocks, is in accord with this ship between the new global tectonicsand the concept.In part this agreementis by design, field of seismologyis discussedfor a variety for the blocks were chosen to some extent on of topics ranging from seismicityto tsunamis, this basis,but data from other fields were used from earth structure to earthquake predic- as well. That the end result is internally con- tion. In each case what the authors judge to sistent is significant.Zones predicted by the be representative,reliable evidence from the theory to be tensional,such as oceanrifts, are field of seismologyis presented.This judgment sites of only shallow earthquakes (the thin is based on the quality of raw data and their shallowlithosphere is beingpulled apart; earth- analysis,not on the relation of the resultsto quakes cannot occur in the asthenosphere), the new global tectonics.Reasonable specula- and the general level of seismicactivity and tion is presentedwhere it seemsproper. The the size of the largest earthquakes are lower organizationof the paper is based not on the there than in the more active compressional classical divisions of seismologybut on the features. In the compressionalfeatures (the principal effects predicted by the new global arcs) large, deep earthquakesoccur and ac- tectonicsand relevant to seismology.As a re- tivity is high as the lithosphereplunges into sult of the remarkable capacity of the new the deeper mantle eventually to be absorbed. global tectonics for unification, an obvious Deep earthquakescan occuronly whereformer division of material among the sectionswas crustal and uppermost mantle materials are not completely achieved, however. now found in the mantle. Where one block of The first two sectionspresent seismological the lithosphereis moving past another along evidence that the worldwide rift system and the surface at the zones of large strike-slip island arcs are the sourcesand sinks, respec- faulting, seismicactivity is shallow, but occa- tively, for surficial material. The third section sional rather large shallowearthquakes are ob- on compatibility of movementson a worldwide served. Some zones combine thrusting and scale is closelyrelated to the first two sections. strike-slipmotion. The generalpattern of earthquake focal mechanismsis in remarkableagree- Fig. 3. (Opposite) Summary map of slip vec- ment with the pattern predicted by the move- tors derived from earthquake mechanism studies. ments of the lithosphere determined in other Arrows indicate horizontal component of direction ways and providesmuch additional informa- of relative motion of block on which arrow is drawn to adjoining block. Crests of world rift tion on this process.The depth of the deepest system are denoted by double lines; island arcs, earthquakes (about 700 km) has been rea- and arc-like features, by bold single lines; major sonablywell known,but unexplained,for many transform faults, by thin single lines. Both slip years.The mobilelithosphere hypothesis offers, vectors are shown for an earthquakes near the at this writing, several possiblealternatives to western end of the Azores-Gibraltar ridge since a rational choice between the two could not be explain this observation.Many similar points made. Compare with directions computed by Le are raisedin the remainderof this paper. Other Pichon (Figure 2). SEISMOLOGY AND NEW GLOBAL TECTONICS 5861

-4

ß o 5862 ISACKS, OLIVER, AND SYKES Evidence from seismologyon the structure of that resemblemajor fault zoneson the conti- the mantle in terms of a lithosphere, an nents. The apparent displacementsalong these asthenosphere,and a mesosphereis so volumi- fracture zones have been explained in at least nous and well known that the section on this three different ways, including simple offset of topic, the fourth section, presents primarily the ridge by strike-slip faulting IVacquiet, additional evidence of particular relevance to 1962], in situ developmentof the ridge crests the new global tectonics.The fifth section,on at separate locations accompaniedby normal the impact of the new global tectonics on faulting along fracture zones [Talwani et al., seismology,is lessdocumented by data than the 1965b], and transform faulting [Wilson, 1965a]. previoussections partly becausethe impact of Transform iaults. Although the concept of the new global tectonicsis quite recent. The simple offset tacitly assumesthe conservation natural lag in pursuing this aspect is such that of surfacearea, the growth or the destruction there has been to date relatively little emphasis of surface area is basic to the definition of the in this particular field. This section is, then, transform fault. In this hypothesisthe active somewhatspeculative and, hopefully, provoca- portion (BC in Figure 4) of a strike-slip fault tive. along which large horizontal displacementhas Few scientificpapers are completelyobjective occurredends abruptly at the crestof a growing and impartial; this one is not. It clearly favors oceanridge. The horizontal displacementalong the new global tectonicswith a strong prefer- the fault is transformed(or absorbed)by sea- ence for the mobile lithosphere version of this floor growth on the ridge; the growingridge is, subject.In the final section,however, we report in turn, terminatedby the fault. Two separate an earnest effort to uncover reliable informa- segmentsof ridgecrest can be joined (Figure4) tion from the field of seismologythat might by a strike-slipfault of this type; thesefaults provide a caseagainst the new global tectonics. are called transform faults of the ridge-ridge There appearsto be no suchevidence. This does type. not mean, however, that many of the data Wilson [1965a] recognizedthat the senseof could not be explained equally well by other hypotheses(although probably not so well by any other single hypothesis) or that further developmentor modificationof the new global tectonicswill not be required to explain some of the observationsof seismology.It merely means that, at present, in the field of seismology, there cannot readily be found a major obstacleto the new globaltectonics.

MID-OCEAN P•IDGES--T•-IE SOURCES Displacements along fracture zones. Recent studies of earthquakes have revealed several important facts about the nature of displacements on the ocean floor [Sykes, 1967, 1968]. The recognition of the worldwide extent of the mid-ocean ridge system (Figures 2 and 3) [Ewing and Heezen, 1956] led to a great interest in the significanceof this major feature to global tectonics. Although the ridge system Fig. 4. An idealized model of sea-floor spreading and transform faulting of the ridge-ridge type. appearsto be a continuousfeature on a large Hatching indicates new surface area created dur- scale, the crest of the ridge is actually discon- ing a given period of sea-floor spreading along the tinuousin a number of places (Figures 1, 2, active ridge crests BF and CE. Present seismicity and 3). These discontinuitiescorrelate with the (indicated by crosses) is confined to ridge crests and to segment BC of the fracture zone AD. intersectionsof the ridge and the major fracture Arrows denote sense of shear motion along active zones--longlinear zonesof rough topography segment BC. SEISMOLOGY AND NEW GLOBAL TECTONICS 5863 shear displacementalong transform faults of fracture zonesof the East Pacificrise (a branch the ridge-ridgetype wouldbe exactlyopposite of the mid-oceanridge system)is illustratedin that requiredfor a simple offset of the two Figure 1. segmentsof ridgecrest. He alsopointed out that Sykes also showedthat earthquakeslocated seismicactivity along transform faults should on the ridge crests (segmentsBF and CE in be confinedto the regionbetween the two ridge Figure 4) but not located on fracture zonesare crests(segment BC in Figure 4). If the crestal characterizedby a predominanceof normal zonesare beingdisplaced by simpleoffset, how- faulting. Normal faulting on oceanridges had ever, seismicactivity shouldbe presentalong long been suspectedbecause of the existenceof the entire lengthof the fracture zone. a rift valley near the crestof large portionsof Earthquake mechanisms. The first motions the ridge system [Ewing and Heezen, 1956]. of seismicwaves from earthquakesoffer a means More than fifty mechanismsolutions (Figure 3) for ascertainingthe senseand type of displace- havenow been obtained for the worldrift system ments on fracture zones. First-motion studies [Sykes,1968; Tobin and Sykes,1968; Banghat are often called'fault-plane solutions'or 'focal- and Sykes,1968]; they continueto confirmthe mechanismsolutions.' Although many earth pattern of transform faulting and normal fault- scientistshave beendisappointed by the large ing describedby Sykes. Nearly the same tec- uncertainties involved in many first-motion tonic phenomenonis observedfor each of the studies,investigations of focal mechanismswere major oceans. vastly upgraded by the installation of the Seismicity. The distributionof earthquakes World-Wide StandardizedSeismograph Net- is another key piece of seismicevidence for the work [Murphy, 1966]. Reliable calibration, hypothesisof transformfaulting. Nearly all the availability of data, high sensitivity,use of earthquakeson the mid-oceanridges are con- seismographsof both long and short periods, finedeither to the ridgecrests or to the partsof and greatergeographical coverage are someof fracture zones that lie between ridge crests the more important characteristicsof this net- [Sykes,1967]. Seismicactivity alonga fracture work, which commencedoperation in 1962. zone ends abruptly (Figure 4) when the frac- Various studiesusing data from these stations ture zone encountersa ridge; only a few earth- haveconfirmed that a doublecouple (or a shear quakeshave been detected from the outerparts dislocation)is an appropriate model for the (segmentsAB and CI)) of most fracture zones. radiationfield of earthquakes[Stander, '1967; If the transformfault theory is correct,the Isacksand Sykes,1968]. Hence,the first mo- areasof seafloor that are now boundedby the tions observedat seismographstations around outer inactive parts of fracture zoneswere once the world may be usedto determinethe orienta- locatedbetween two ridge crests; theseblocks tion and the sense of the shear motion at the of sea floor moved beyond either crest as sourcesof earthquakesin various tectonic re- spreadingprogressed. Thus, the age of deforma- gions. Additional backgroundinformation on tion becomes older as the distance from an earthquake mechanismswill be introduced in active crest increases. later sectionsas further clarificationis required. Earthquake swarms. The occurrence of Mechanismsalong world ri[t system. Sykes earthquakeswarms along the world rift system [1967] examinedthe focal mechanismsof seven- suggeststhat the crestalzone probably is char- teen earthquakesalong various parts of the acterized by submarine volcanic eruptions world rift system.In his study all the earth- [Sykes et al., 1968]. Earthquakeswarms are a quakes located on fracture zones were charac- distinctivesequence of shockshighly grouped terizedby a predominanceof strike-slipmotion. in spaceand time with no one outstanding In each case the shear motion was in the correct principalevent. Although these sequences some- sensefor transformfaulting (Figure 4), but it timesoccur in nonvolcanicregions, most of the was consistentlyopposite in senseto that ex- world's earthquake swarms are concentratedin pected for simple offset. This is an instance in areasof presentvolcanism or geologicallyrecent the earth sciencesin which a yes-or-noanswer volcanism[Richter, 1958; Minikami, 1960]. could be suppliedby data analysis.The sense Largeswarms often occur before volcanic erup- of motion(left lateral) alongone of the major tions or accompanyingthem; smaller swarms 5864 ISACKS, OLIVER, AND SYKES may be indicativeof magmaticactivity that excellentagreement with evidenceof spreading failed to reach the surface as an eruption. from magneticanomalies, ages of rocks,and the From the seismographrecords at Palisades, distributionof sediments[Vine, 1966;Heirtzler New York, Sykeset al. [1968] recognizedmore et al., 1968; Wilson,1963; Burckle et al., 1967]. than twenty swarms of earthquakesoccurring The world rift systemmust be recognizedas during the past 10 years. These swarmscom- oneof the major tectonicfeatures of the world. monly lasted a few hours or a few days. Al- It is characterizednearly everywhereby exten- though many of the larger earthquakesalong sional tectonics,sea-floor growth at its crest, the world rift system occur on fracture zones and transform faulting on its fracture zones. and are characterized by strike-slip faulting, The focal depthsand the maximummagni- nearly all the swarmsare restrictedto the ridge tudesof earthquakes,the narrownessof seismic crests (segmentsBF and CE in Figure 4) and zones,and the propagationof S• wavesalong seem to be characterized by normal faulting. oceanridges and transformfaults will be de- Swarms are commonly (but not always) asso- scribed in the sectionson worldwide compara- ciated with volcanic eruptions on islands or bility of movementsand on additionalevidence on or near the crest of the world rift system. for the existenceof the lithosphere. From a simulation of magnetic anomalies Implicationsfor continentaldrift. The simi- Matthews and Bath [1967] and Vine and larity of the earthquake mechanismsalong Morgan [1967] estimate that most of the new nearly the entire length of the ridge system surface material along the world rift system is suggeststhat transformfaulting and spreading injected within a few kilometers of the axis of have been occurring in these regionsfor ex- the ridge. In Iceland, where the rift may be tended,but as yet unspecified,periods of time. seen and studied in detail, postglacialvolcanism The distributionof magneticanomalies, palco- is confined largely to the median rift that magneticinvestigations, and the shapesof con- crossesthe island [Bodvarsson and Walker, tinental blocksthat supposedlywere split apart 1964]. The rift apparently marks the landward by spreadingfurnish a more completehistory continuation of the crest of the mid-Atlantic of the processesof sea-floorspreading and trans- ridge (Figures2 and 3). form faulting. A questionof particularinterest The lack of weatheringin rock samples,the is: Have the various segmentsof ridge grown youngages measured by radioactiveand paleon- in place; i.e., has the en echelonpattern of tologic dating of rocks and core materials, and ridgesand fracturezones prevailed throughout the general absenceof sedimentas revealed by an episodeof sea-floorspreading? bottom photographsand by reflectionprofiling Both the Gulf of Aden and the Gulf of all attest to the youthful character of the California are thought to have openedby con- crestal zones of the mid-ocean ridge [Ewing tinental drift duringthe last 25 m.y. [Hamilton, et al., 1964; Burckle et al., 1967; van Andel 1961; Laughton,1966]. If drift occurredin a,d Bowin, 1968; Dymond and De#eyes, 1968]. theseareas, the displacementsare at mosta few Thus, the occurrenceof earthquake swarms is hundred kilometers. If continental drift can be compatiblewith the hypothesisthat new surface confirmedfor these features, inferencesabout materials are being emplacedmagmatically near drift on an ocean-widescale are placed on a the axes of the ocean ridges. The large earth- much firmer basis. quake swarms(and perhapssome of the smaller Figure5 showsthe distributionof structural swarms) may be indicative of eruptionsor mag- features,earthquake epicenters, and earthquake matic processesin progressnear the ridge crests. mechanismsfor the Gulf of Aden [Sykes,1968]. Nonetheless,more work is needed to ascertain Nearly all the epicentersare confinedeither to if a causal relationship exists between the two northeast-strikingfracture zonesor to the ridge phenomena. that extends from a branch of the mid-ocean Synthesis o] data for ridges. Seismological ridge (the Carlsbergridge) near 9øN, 57øE •o evidence of various types seems to provide a the westernpart of the Gulf of Adennear 12øN, definitiveargument for the hypothesesof trans- 43øE. This ridge coincideswith the rough cen- form faulting and sea-floor spreading on the tral zonein Figure 5. As in other parts of the mid-ocean ridge system. These data are in world rift systemthe earthquakesoccurring on SEISMOLOGY AND NEW GLOBAL TECTONICS 5865

13_

o o 5866 ISACKS, OLIVER, AND SYKES

fracture zonesare mostly restrictedto the re- patterns of the island arcs. The closeassociation gionsbetween ridge crests.Mechanism solutions of the major oceandeeps with these arcs is for events 22 and 23 (numbers after Sykes obviousand suggestsexceptional subsidence in [1968]) indicatetransform faulting of the ridge- thesezones, but otherfacts are equally striking. ridge type. Nearly all the world'searthquakes in the deep If the openingof the Gulf of Aden was ac- and intermediate range, most of the world's complishedthrough a simple processof sea- shallowearthquakes, and the largestdepartures floor spreading and transform faulting, the from isostatic equilibrium are associatedwith fracture zonesshould join points in Arabia and islandarcs or arc-likestructures, as shownby in Africa that were togetherbefore the drifting Gutenbergand Richter[1954]. Volcanoes, sea- commenced.Also, the fracture zonesshould not level changes,folding, faulting, and other forms continueinto the two continentalplates. œaugh- of geologicevidence also demonstrate the high ton [1966] has shown,in fact, that thesefaults level of tectonic activity of these features.A do not continue inland. In addition, his pre- conceptof global tectonicsin which the arcs do Miocene reconstruction,in which the two sides not play an important role is unthinkable. If of the Gulf of Aden are moved togetherparallel crustalmaterial is to descendinto the mantle, to the fracture zones,juxtaposes a large num- the island arcs are suspectas sitesof the sinks. ber of older structural features on the two sides The asymmetrical structure of the arcs and of the gulf. The en echelon arrangement of the associatedpattern of earthquakeoccurrence segmentsof ridge is alsomirrored in the stepped in the mantle led many investigators(e.g., shape of the continental margins of Arabia and VeningMeinesz [1954], Beniof•[1954], Hess Africa. Hence, the present en echelonpattern [1962], Dietz [1961]) to postulate that the seemsto have prevailedsince the initial breakup structuresare the result of compressivestresses of thesetwo blocksabout 5 to 25 m.y. ago. normal to the arc and are the sites of vertical A similar pattern of en echelonridges is pres- movements in various convective schemes. Al- ent in the Gulf of California (Figure 6). Earth- thoughsuch ideas were supportedby the in- quake mechanismsfrom this region are investigationsof focalmechanisms of earthquakes dicative of a series of northwesterly striking madeby Hondae• al. [1956]and by thegravity transform faults with right-lateral displacement studies of Vening Meinsz [1930] and Hess [Sykes, 1968]. These transform faults, which [1938], later analysesby Hodgson[1957] for are arranged en echelon to the San Andreas focalmechanisms and by Talwaniet al. [1959] fault, connect individual segmentsof growing and Worzel [1965] for gravity led to different ridges in the Gulf of California. Hence, sea- conclusions.This section reviews the data and floor spreadingand transform faulting alsowere showsthat there is strongsupport for the com- responsiblefor the displacementof Baja Cali- pressive nature of island arcs and for their role fornia relative to the mainland of Mexico. If as sites where surface material moves down- thesetwo blocksare reconstructedby horizontal ward into the mantle.In particular,a variety displacements parallel to the northwesterly of evidencesupports the modelof the arc shown striking fracture zones, the peninsula of Baja in Figure 1. In this modelthe leadingedge of California is placed in the indentation or 'hitch' the lithosphereunderthrusts the arc and moves of the mainland of Mexico near 21øN, 106øW. downwardinto the mantleas a coherentbody. Thus, the two piecesappear to fit together in The proposedpredominance of strike-slipfault- this reconstruction.Wilson [1965a] has pointed ing in island arcs [Hodgson,1957] is not in out that the steppedshape of the fracture-zone- agreementwith this modelbut appears,in view ridge pattern in the equatorial Atlantic is of recentand vastlyimproved seismic data, to mirrored in the steppedshape of the coastlines be based on unreliable determinationsof focal and the continental margins of Africa and mechanisms[Hodgson and Stevens,1964]. The Brazil. extensional features of structures based on

ISLAND ARCS--THE SINKS gravity and seismicdata appear to be surficial and canbe reconciledwith, andin fact are pre- Almost anyone who glances casually at a dictedby, the new hypothesis. map of the worldis intriguedby the organized High-Q and high-velocityzones in the mantle SEISMOLOGY AND NEW GLOBAL TECTONICS 5867

/ Faults II Troughs striking NE 0 Earthquake Epicenters •"A EarthquakeMechanisms_.

Pacific Ocean

I I 115ø I10ø 105ø

Fig.6. Structuralfeatures of the Gulfof California[after Sykes, 1968]. Relocated epicentersof earthquakesfor the period1954 to 1962.Seismicity and focal mechanisms support the hypothesisof spreadingby ocean-ridge-transform-faultmechanism. beneathisland arcs. The grossstructure of an generatedby deepearthquakes in the seismic idealized island arc as shown in Figure I is zoneand propagatedalong two differentkinds basedon the resultsof Oliver and Isacks [1967]. of paths,one alongthe seismiczone and one Their study was primarily concernedwith the throughan aseismicpart of the mantle,demon- Fiji-Tongaarea. Comparisonof seismicwaves strated the existenceof an anomalous zone in 5868 ISACKS, OLIVER, AND SYKES the upper mantle. The anomalouszone was esti- each parameter on temperature is reasonable. mated to be about 100 km thick and to be This point is discussedfurther in another sec- bounded on the upper surface by the seismic tion. zone. Thus, the zone dips beneaththe Tonga Bending of lithospherebeneath an island arc. arc at about 45ø and extendsto depths of al- The evidence supporting the model in which most 700 km. The zone is anomalous in that the lithosphereplunges beneath the island arc attenuation of seismic waves is low and seismic is varied. To explore this point further, consider velocities are high relative to those of the first the configurationof the upper part of the mantle at comparabledepths elsewhere.Recent lithosphere in the vicinity of an island arc studies of the Japanese arc [Wadati et al., (Figure 7a). Seismic refraction studies of a 1967; Utsu, 1967] have confirmedthe existence number of island arcs have been made. In- of sucha structurefor that region.Similar zones variably they show the surface of the mantle, appear to be associatedwith other island arcs which is shallow beneath the deep ocean, deep- [Oliverand Isacks, 1967; Cleary,1967; Molnar ening beneath the trench, as suggestedby Fig- and Oliver, 1968]. ure 7a. Althoughsome authors suggest that the The presenceof a high-velocityslab beneath mantle merely deepens slightly beneath the an island arc introducesa significantazimuthal islands of the arc and shoalsagain behind the variation in the travel times of seismic waves. arc, evidencefor such a structureis incomplete. Such variations with respect to source anom- Mantle velocities beneath the islands, where alies are shownby Herrin and Taggart [1966], determined, are low, and there is no case for Sykes [1966], Cleary [1967] and are indicated which the data could not be interpreted as sug- by the data of Carder et al. [1967] all for the gestedin Figure 7a (see, e.g., Badgley [1965] case of the Longshot nuclear explosion.With and OJ•ceret al. [1959]). In fact, the difficulty respect to station anomaliessuch variations are experiencedin documentingthe modelin which shown by Oliver and Isacks [1967], Utsu the mantle is merely warped beneaththe islands [1967], Cleary and Hales [1966], and Herrin is evidenceagainst this model.The main crustal [1966] from data from earthquakes. These layer as determined from seismic refraction effects must therefore be taken into account as studies seems to parallel the surface of the sourcesof systematicerrors in the locationsof dipping mantle beneath the seawardslope of earthquakesand the constructionof travel-time the trench. In some interpretationsthe crustal curves.The large anomaly of Q associatedwith layer thins beneath the trench; in others it the slab must play a very important role in the thickens or remains constant. Perhaps these are Q structureof the mantle, especiallyfor studies real variationsfrom trench to trench, but the based on body waves from deep earthquakes. data are not always definitive. Studies in which this effect is ignored [e.g., Thinning of the crust has beeninterpreted by Teng, 1968] must be reassessedon this basis. Worzel [1965] and others as an indication of Oliver and Isacks associated the anomalous extension,and there is considerableevidence in zone with the layer of low attenuationnear the the structure of the sediments on the seaward surface to the east of Tonga. In their interpre- slopes of many trenches supporting the hy- tation of the data, they correlatedlow attenua- pothesisof extension (see, e.g., Ludwig et al. tion with strengthto arrive at the structureof [1966]). Figure 8, one of Ludwig's sections Figure I in which the lithosphere,a layer of acrossthe Japan trench, demonstratesthis point strength, descendsinto the mantle. This con- dramatically. Several graben-likestructures are figuration suggeststhe mobility of the litho- seen on the seaward slope of the trench. Al- sphereimplied in Figure I and describedin the though such evidencefor extensionhas been introduction. Based on current estimates of cited as an argument against sea-floorspread- lithospherevelocities and other parameters,the ing and convectionon the basisthat down-going down-goingslab would be much coolerthan its currents at the sites of the ocean deeps would surroundingsfor a long time interval. Although causecompression normal to the arcs,the argu- there is little evidencesupporting a direct rela- ment loses its force when the role of the litho- tion between low attenuation and strength, an sphereis recognized.All the evidencefor exten- indirect relation based on the dependenceof sion relatesonly to the sedimentsand crust,i.e., SEISMOLOGYAND NEW GLOBAL TECTONICS 5869

50 I00 km :5 50km 300 2_50 200 150 I00 50 0 LOCAL TENSION

S CRUST S S CRUS

LITHO- SPHERE $PHERE 0

ß ß ß ß ß ß ß ß U ASTHENOSPHERE ß:: ASTHENOSPHERE ß ß km

Fig. 7a.

3 50 km 300 250 200 150 100 50 0 50 IOOkm LOCAL TENSION S ½ CR UST S S CRU• L I T H 0 S P HERE $ P H E RE sø' LOCAL 0 TENSION '•"/' I00'

ASTHENOSPHERE ASTHENOSPHERE •5( '%/ km

Fig. 7b. Figure7 showsvertical sections through an island are indicating hypothetical structures and otherfeatures. Both sectionsshow down-going slab of lithosphere,seismic zone near surface of slaband in adjacentcrust, tensional features beneath ocean deep where slab bends abruptly andsurface is free.(In bothsections, S indicates seismic activity.) (a) A gapin mantleportion of lithospherebeneath island are and circulation in mantle associated with crustal material of •heslab and with adjoiningmantle [Holmes, 1965]. (b) The overridinglithosphere in contact with the down-goingslab and bent upward as a resultof overthrusting.The relationof the bendingto thevolcanoes follows Gunn [1947]. No verticalexaggeration.

the upperfew kilometers oœ the lithosphere.For activitybeneath the seaward slope of thetrench the modelspictured in Figure7 in whicha thick is, in general,infrequent and apparentlyof stronglayer bends sharply as it passesbeneath shallowdepth. The focalmechanisms that have the trench,extensional stresses are predicted been determined for such shocks indeed indi- near the surface on the convex side of the bend cate extensionas predicted,i.e. normal to the even tho,gh the principalstress deeper in the trench, the axis of bending(Stauder [1968] lithospheremay be compressional.Earthquake and T. Fitch and P. Davis, personalcommuni- 5870 ISACKS, OLIVER, AND SYKES

M d

v x\__4

Approximate Horizontal Scale I , , , , I , , , , I 0 50 I00 k rn

Fig. 8. Seismic reflection profile across the Japan trench extending easterly along 35øN from point M near Japan to point N [after Ludwig et al., 1966]. Vertical scalerepresents two- way reflection time in seconds (i.e., I sec ---- I km of penetration for a velocity of 2 km/sec). Note block faulting along seaward slope of trench demonstrating extension in crust and inclu- sion of sedimentsin basement rocks. Also note shoaling of oceanic basement on approaching trench as suggestedby work of Gunn [1937]. Vertical exaggeration•25: 1. cartons). Stauder demonstratesthis point very tion of underformed sediments depends on well in a paper on focal mechanismsof shocks the ratio of rate of sediment accumulation of the Aleutian arc. to rate and continuity of underthrusting, The extensional features also suggest a and such data must be evaluated for each area mechanismfor includingand transportingsome with these factors in mind. The South Chile sedimentswithin the down-goingrock layers. trench, for example,has a large sedimentation As implied by Figure 7a, sedimentsin the rate but no associateddeep earthquakes,sug- graben-like features may be carried down to gestinglittle or no recentthrusting. The results some depth in quantities that may be very of Schollet al. must be consideredin this light. significantpetrologically, as suggestedby Coats The Itikurangi trench (east of northern New [1962]. Probably not all the sedimentscarried Zealand), another exampleof a partially filled into the trench by motion of the sea floor or by trench, is also thought to be in a zone of low normal processesof sedimentationare absorbed convergencerate (see Le Pichon [1968] and in the mantle,however. There are largevolumes Figure 2). of low-densitymaterial beneaththe inner slope Underthrusting beneath island arcs. The on the island side of most trenches [Talwani shallow earthquakesmentioned above that indi- and Hayes, 1967] that may correspondto sedi- cate extensionnormal to the arc occur relatively ment scrapedfrom the crust and deformedin infrequently and appear always to be located the thrusting. Unfortunately, the structure of beneath or seaward of the trench axis. The these low-density bodies is not well explored, earthquakes that account for most of the seismic and,in fact, the very difficultyof exploringthem activity at shallow depths in island arcs are lo- may be an indicationof their contortednature, cated beneath the landward slope of the trench which results-fromgreat deformation. and form a slab-like zone that dips beneaththe The above arguments apply to trenchesthat island arc [Fedotov et al., 1963, 1964; Sykes, are relatively free of fiat-lying sediments,such 1966; Hamilton, 1968; Mitronovaset al., 1968]. as the Japan or Tonga trenches.The occurrence This point is illustrated in Figure 9, which of substantial quantities of fiat-lying unde- showsa vertical sectionthrough the Tonga arc. formed sediments in some other trenches has Note that, for a wide range of depths,foci are been cited as evidenceagainst underthrusting confined to a zone 20 km or less in thickness. in island ares [Scholl et al., 1968]. Aeeumula- In the focal mechanisms of the shallow SEISMOLOGY AND NEW GLOBAL TECTONICS 5871

Volcanic Coralline Arc Arc 0 • ,: ::• , , Water Depth, km i0 I ,. 0 + 200

Distance from Niumate, Tonga, km -400 -200 0 + 200 o

200

400 e / ' 550 /

...--•- • .... 6O0 i I

• , • - 350 - 500 - 250 km I . •J•. 9. ¾ertic•! •ectJonoriented perpendicular to the Tonka arc. CJrc]e•represent earth- quake8proiected from •Jth•n 0 to 150km north of the •ectJon;triangle8 correspond to eYent8 projectedfrom within 0 to 150km southof the section.All shocksoccurred during 1965 while the Lamont network of stations in Tonga and Fiji was in operation. Locationsare based on data from these stations and from more distant stations. No microearthquakesfrom a sample of 750 eventsoriginated from within the hatchedregion near the station at Niumate, Tonga (i.e., for $-P timesless than 6.5 sec).A verticalexaggeration of about13:1 wasused for the insertshowing the topography[after Raitt et al., 1955]; the horizontaland vertical scalesare equalin the crosssection depicting earthquake locations. Lower insert shows enlargement of southernhalf of sectionlor depthsbetween 500 and 625 km. Note small thickness(less than ~20 km) of seismic zone for wide range of depths. 5872 ISACKS, OI,IVER, AND SYKES shocksalong the slab-like zone of the Tonga- that of the deep seismiczone below it, appears Kermadee are, Isacks and Sykes [1968] find to be confinedmainly to the crust and to be consistentevidence for underthrustingof the secondaryto the activity along the main seismic seaward block beneath the landward block. zone. The Niigata earthquake of June 16, Abundant evidence for a similar process for 1964, appears to be located in such a secondary various island arcs of the North Pacific is zone of the North Honshu arc. The mecha- found by Stauder [1962, 1968], Udias and nism of this earthquake [Hirasawa, 1965] indi- Stauder [1964], Stauder and Bollinger [1964, cates that the axis of maximum eompressive 1966a, b], Aki [1966], and Ichikawa [1966]. stress is more nearly horizontal than vertical Critical evaluationof focal mechanismdata by and trends perpendicular to the strike of the Adams [1963], Hodgson and Stevens [1964], North Honshu arc. It is interesting that this Stauder [1964], and Ritsema [1964] showsthat stress is also perpendicular to the trend of the generalizationthat strike-slip faulting is Neogene folding in North Honshu [Matsuda predominant in island arcs is based on unre- et al., 1967]. These resultsmight indicate some liable data and possiblesystematic errors in the compresslye deformation of the overriding analyses.The recent data, greatly improved in platesin the modelsof Figure 7. quality and quantity, indicatethat, in fact, dip- Deep earthquakes:the down-goingslab. The slip mechanismsare predominantin island arcs. shallow seismiczone indicated by the major The thrust fault mechanisms characteristic of seismic activity is continuouswith the deep shallowearthquakes in island arcs thus appear zone, which normally dips beneath the island to reflect directly the relative movementsof the arc at about 45 ø. The thickness of the seismic convergingplates of lithosphereand the down- zone is not well knownin most cases,but it ward motion of the oceanic plate. The com- appears to be lessthan about 100 km and some patibility of these motions as determined by evidencesuggests that it may, at least in some focal mechanismdata with the worldwidepat- areas, be less than 20 kin. Figure 9 illustrates tern of plate movements is discussedlater and this point for a sectionthrough the Tonga arc. is shown to be excellent. Althoughthe surfaceapproximating the distri- Considerableevidence for underthrustingin bution of hypocentersmay be describedroughly the main shallow seismic zone exists in other as above, it is clear that significantvariations kinds of observations.Geodetic and geologic from this simplepicture exist and are important. studies of the Alaskan earthquake of 1964 For example,the over-all dips may vary from [Parle'in,1966; PlaCket,1965] stronglysupport at least 30ø to 70ø , and locally the variation the concept of underthrusting.Geologic evi- may be greater, as suggestedby the data in dencealso indicatesthe repeated occurrenceof Figure 9. For the Tonga-Kermadeearc, the such thrusting in this arc during recent time number of deepevents is large and the zonecan [Plafl•'er and Rubin, 1967]. Data from other be definedin somedetail [Sykes,1966]. Sykes arcs on crustal movements are voluminous and was able to show,as a result of a marked curva- have not all been examined in light of the ture of the northern part of the Tonga arc, a hypothesesof the new global tectonics.In fact, clear correlationbetween the configurationof in many arcs the principal zone of underthrust- the deep seismiczone and surface features of ing would outcrop beneath the sea and im- the arc, thereby demonstratingthe intimate portant data would be largely obscured. One relationship between the deep and the shallow important point can be made. It is well knowa processes. [Richter, 1958] that vertical movements in For most arcs,however, the numberof deep island arcs are of primary importance.This con- events, particularly since the World-Wide trasts with the predominantlyhorizontal move- Standardized SeismographNetwork has been ment in such zonesas California, where strike- in operation,is relatively small; therefore,the slip faulting predominates. deep zonescannot be definedas preciselyas one Other shallow activity in island arcs. In might desire.Nevertheless, sufficient information some island arcs there is appreciable shallow is availableon the patternof seismicactivity so seismicactivity landward of the principal seis- that the conceptof the mobilelithosphere can mic zone. This activity, which is distinct from be tested in general,and it must be assumed SEISMOLOGY AND NEW GLOBAL TECTONICS 5573 that subsequentdetailed studies of other island than the compressionalaxes, these axes are not arcs may reveal contortionsin the seismiczone randomlyoriented. The axisof tensiontends to comparablewith the contortionsalready found be perpendicularto the seismiczone; the null in Tonga-Fiji. axis, parallelto the strike of the zone.These Focal mechanisms. The simple underthrust- generalizationsare shownschematically in Fig- ing typical of the shallow earthquakesof the ure 11. The slip planesand slip directionsare principal zonesdoes not, in general,persist at thus systematicallynonparallel to the seismic great depths.For shocksdeeper than about 100 zones;the orientationsare thereforedifficult to k•n the orientation of the focal mechanisms reconcilewith a simpleshearing parallel to the varies considerablybut exhibitscertain clear-cut seismiczone as suggestedby the commoncon- regularities.To understandthese regularities, it cept of the zoneas a large thrust fault. Sugi- is important to recall what is determinedin a mura and Uyeda [1967] soughtto reconcilethe focal mechanism solution. The double-couple observationswith that conceptby postulating solution,which appears to be the best repre- a reorientation of crystalline slip planes per- sentation of most earthquakes, comprisestwo pendicularto the axisof maximumcompressive orthogonalnodal planes, either of which may stress,such that in the caseof horizontalcom- be taken as the slip plane of the equivalent pressionthe slip planes would tend to be vertical. shear dislocation.Bisecting these nodal planes are the axis of compression,P, in the quadrants STRIKE OF of dilatational first motions and the axis of ,ARC tension, T, in the quadrants of compressional first motions. The axis formed by the intersection of the nodal planes is the null, or B, axis parallelto which no relative motiontakes place. If one nodal plane is chosenas the slip plane, the poleof the other nodalplane is the direction of relative motion of the slip vector. It is important to realizethat the primary information PLUNGE given by a double-couplesolution is the orientation of the two possibleslip planes and slip vectors.The interpretation of the double-couple mechanisms in terms of stress in the source regionrequires an assumptionabout the failure process.The P, T, and B axescorrespond to the maximum,minimum, and intermediateaxes of compressirestress in the medium only if the Fig. 10. Orientations of the axes of stress as shear dislocation is assumed to form parallel givenby the double-couplefocal mechanism solu- to a plane of maximumshear stress in the me- tions of deep and intermediateearthquakes in the dium,i.e., a planethat is parallelto the axisof Tonga arc, the Izu-Bonin arc, and the North intermediate stress and that forms a 45 ø angle Honshu arc. Open circlesare axes of compression, to the axes of maximum and minimum stress. P; solid circlesare axesof tension,T; and crosses are null axes, B, all plotted on the lower hemi- Pattem.• o1• /ocal mecha•isrns/or deep earth- sphere of an equal-areaprojection. The data, se- quakes. The most striking regularity in the lected from available literature as the most reli- orientation of the double-couplefocal mecha- able solutions, are taken from Isacks and Sykes nisms of deep and intermediate earthquakesis [1968], Honda et al. [1956], Ritsema [1965], and Hirasawa [1966]. The data for each of the the tendencyof the P axesto parallel the local three arcs are plotted relative to the strike of the dip of the seismiczone. Figure I0 illustrates arc (Tonga arc, N 20øEß Izu-Bonin, N 15øW; this point for the three zones (Tonga, Izu- North I-Ionshu arc, N 20øE). The dips of the zones Bonin,and North Itonshu) for whichreliable vary between about 30' and 60ø , as indicated by data are most numerous.This figure also shows the dashed lines in the figure. Note the tendency of the P axes to parallel the dip of the seismic that, althoughthe orientationof the axes of zone and the weaker tendency for the T axes to tension and the null axes tend to be less stable be perpendicularto the zone. 5874 ISACKS, OLIVER, AND SYKES

0 • // TRENCHAXIS

200 KM

400

6øøI Fig. 11. Vertical sectionsperpendicular to the strike of an island arc showingschematically typical orientations of double-couple focal mechanisms. The horizontal scale is the same as the vertical scale.The axis of compressionis representedby a convergingpair of arrows;the axis of tension is representedby a diverging pair; the null axis is perpendicularto the section.In the circular blowups,the senseof motion is shownfor both of the two possible slip planes. The features shown in the main part of the figure are based on results from the Tonga arc and the arcs of the North Pacific. The insert shows the orientation of a focal mechanismthat could indicate extensioninstead of compressionparallel to the dip of the zone.

Alternatively, Isacks and Sykes [1968] show would, therefore,be expected,especially near that, if it is assumedthat the slip planes form parts of the zone with complexstructure. at angleswith respect to the axis of maximum The important feature of the interpretation compressive stress that are not significantly presentedhere is that the deep earthquake different from 45 ø, then a very simple interpre- mechanismsreflect stressesin the relatively tation can be made on the basis of the model strong slab of lithosphereand do not directly of Figure 1. In this interpretation the axis of accommodatethe shearingmotions parallel to maximum compressivestress is parallel to the the motion of the slab as is implied by the dip of the seismiczone (i.e., parallel to the simplefault-zone model. The shearingdeforma- presumedmotion of the slab in the mantle), tionsparallel to the motionof the slabare pre- and the axis of least compressivestress is per- sumablyaccommodated by flow or creepin the pendicularto the zone or parallel to the thin adjoining ductile parts of the mantle. dimension of the slab. Do the stressesin the slabvary with depth? The tendency for the compresslyeaxes to be In particular, the axis of least compressive more stable than the other two axes can be stress,the T axis,may be parallel to the dip interpreted to indicate that the differencebe- of the zone if the material at greaterdepths tween the intermediate and the least principal were sinkingand pulling shallowerparts of the stresses is less than the difference between the slab [Elsasset,1967]. In the Tonga,Aleutian, greatestand the intermediateprincipal stresses. and Japanesearcs, the focalmechanisms indicate In general,the stressstate may be quite vari- that the slabis undercompression parallel to its able owing to contortionsof the slab, as sug- dip at all depths greater than about 75 to 100 gestedby Figure 9. Possiblylarge variability km. In thesearcs, therefore, any extensionin in the orientations of the deep mechanisms the slab must be shallower than 75 to 100 lma. SEISMOI,OGY AND NEW GLOBAL TECTONICS 5875 Very limited evidence from the Kermadec ures7a and 11 and supportedby the data shown [Isacks and Sykes,1968], New Zealand (North in Figure 9. Although catastrophic phase Island) [Adams, 1963], South Amcrica,• (A. R. changesmay be ruled out as direct sourcesof Ritsema, personalcommunication), and Sunda seismic waves on the basis of the radiation (T. Fitch, personal communication)arcs sug- pattern of the waves, the possibility remains gests,however, that mechanismsindicating ex- that the stressesresponsible or partly respon- tensionof the slab, as shownin the insert of sible for shear failure may result from some- Figure 11, may exist at intermediatedepths in what slowerphase changes. some arcs. Further work is required to dis- Seismic activity versus depth. Frequencies tinguish such mechanismsfrom the under- of earthquakesversus depth for several island thrusting type of mechanismcharacteristic of arcs are shown in Figure 12. There are two earthquakesat shallowdepths or from complex main results emergingfrom these analyses.(1) mechanismsrelated to changesin structure or In all island arcs studied the activity decreases contortions of the slab. in the upper 100 to 200 km approximately Processo) • deep earthquakes. The idea that exponentially as a function of depth with a deep earthquakesoccur in downgoingslabs of decay constant of about 100 km [Sykes, 1966]. lithospherehas important implicationsfor the (2) At greater depths the seismic activity in problem of identifying the physical processre- many (but not all) island arcs increasesrela- sponsible for sudden shear failure in the en- tive to the exponentialdecay extrapolatedfrom vironment of the upper mantle. That deep shallowerdepths, and the seismicactivity shows earthquakes are essentially sudden shearing a fairly well-definedmaximum in some depth movements and not explosive or implosive range in the upper mantle. The variation of changes in volume is now extensively docu- seismicactivity with depth is thus grosslycor- mented (see Isacks and Sykes [1968] for refer- related with the variation of seismic focal ences). Anomaloustemperatures and composi- mechanismswith depth and supports the gen- tion might be expected to be associatedwith eralization that deep earthquake mechanisms the down-goingslab, either or both ot• which have a different relationship to the zone than may accountfor the existenceof earthquakesat shallow mechanisms do. In this correlation the great depths. earthquakesthat define the shallow exponential Several investigators [Raleigh and Paterson, decay in seismicactivity are characterizedby 1965; Raleigh, 1967] concludedthat dehydra- the underthrusting type mechanisms,whereas tion of hydrous minerals can release enough the deeper earthquakesappear to be related to water to permit shear fracture at temperatures the stressesin the down-goingslab. between about 300ø and 1000øC. Although There is an approximate correlation of the Griggs [1966] and Griggs and Baker [1968], decreasein seismicactivity versus depth with assumingnormal thermal gradients, suggested a similar general decreasein seismicvelocities, that these reactions would not take place for Q, and viscosity in the upper 150 km. These depths greater than about 100 km, rates of effectsmay be related to a decreasein the differ- underthrusting as high as 5 to 15 cm/yr sug- encebetween the temperatureand local melting gest that temperatureslow enough to permit temperature. Thus, the decreasein activity with these reactions to occur may exist even to depth may correspond to an increase in the depthsof 700 km. Certainly a re-evaluationof ratio of the amount of deformationby ductile theseprocesses is in order. flow to that by suddenshear failure. An impli- The lowest temperatures, the largest tem- cation of this interpretation is that the ex- perature gradients, and the largest composi- ponential decay constant of 100 km may tional anomalies would probably be most roughly indicate the thicknessof the overthrust marked near the upper part of the slab, i.e. plate of lithosphere.This interpretation is illu- the part correspondingto the crust and upper- strated in Figure 7b, in which the overriding most mantle in the surficial lithosphere.Thus, plate of lithosphere is in 'contact' with the the seismicactivity associatedwith theseanom- down-goingplate along the seismic zone. As aliesmight be expectedto concentratenear the shown in a later section,the assumptionthat upper part of the slab, as is suggestedin Fig- the depth distribution of shallow earthquakes 5876 ISACKS, OLIVER, AND SYKES

[oo

I0 \ - 6 - • _

5 •• \ TONGA-FIJI-KERMADEC -

2 - KURIL-KAMCHATKA AN I

_- , "'A / i ; - 0.6 _ :k,.. t/ I ,

0.3 ½ o -- ALASKAi! /• CENTRAL •AMERICA •' .•- 0.2 - i AND MEXICO -

0.1 CARIBBEAN

.06 I ! i I I i I I I • I I I I I I I I I I I I 0 I00 200 500 400 500 600 700 Depth, km Fig. 12. Number of earthquakesper 25-km depth intervals as function of depth for several island arcs.Except for Japan, data are from Sykes [1966]. Data for Japan expressedas percent- age of events per 50-km depth intervals [Katsumata, 1967]. Since the various curves were not normalized for the sample lengths and for the lower limit of detectability in each area, only the relative shapesand not the absolutelevels of the various curvesshould be comparedwith one another. The number of earthquakes per unit depth within the upper 200 km of all these island arcs is approximatelyproportional to exp (--Z/100), where Z is the depth in kilometers.Peaks in activity below 200 km appear to fluctuateboth in amplitude and in depth among the various arcs. of the mid-ocean rift system yields a measure rectly beneath and behind the arc as shown in of the thickness of the lithosphere is not an Figure 7b. Oliver and Isacks [1967] and Molnar unreasonable one. and Oliver [1968] showthat high-frequency$, Several lines of evidence do not, however, does not propagate acrossthe concaveside of support the existenceof thick lithospheredi- island arcs, which probably indicatesthat the SEISMOLOGY AND NEW GLOBAL TECTONICS 5877

uppermostmantle there has low Q values. This result is in agreementwith the low P, velocities generally found beneath islandsof many arcs. Also, the active volcanismand high heat flow characetristic of the concave side of island arcs [Uyeda and Horai, 1964; Sclater et al., 1968] o[ suggestthat the lithospheremay be thin there. These data are qualitatively fitted by the model shown in Figure 7a. One implication of this figure is that at least part of the shallowearth- quake zone might not result from the contact 400- between two pieces of lithospherebut might instead indicate an embrittlcd and weakened zone formed by the downward moving crustal 600 materials [Raleigh and Paterson, 1965; Griggs, 1967]. In this case the exponential decay in 35 52 5.3 57 74 '•6 81 86 activity might reflect changesin the properties 80C of the earthquake zone as a function of depth. Fig. 13. Depth range of maxima in the seismic Thus, both models in Figure 7 must be re- activity (numbers of earthquakes) as a function tained for the present. of depth in island arcs and arc-like structuresfor which data are sufficiently numerous. The data Although in some arcs such as the Aleutians are from Gutenberg and Richter [1954], Katsu- or Middle America the exponential decay in mata [1967], Sykes [1966], and listings of earth- activity appears to be the only feature present quakes located by the USCGS in the preliminary in curves of activity versus depth, most arcs determination of epicenters (PDE). The numbers exhibit a more or less well-defined maximum in at the bottom of the figure give the rate (in centimeters per year) of convergence for the arc activity in the mantle, as illustrated in Figure as plotted in Figure 2. Note that the maxima 12. The approximate ranges of depth of these occur over a wide range of depths and that the maxima are shown in Figure 13 for several depths appear to correlate, in general, with the island arcs. The main point of this figure is to calculated slip rate. show that the depths of these maxima vary considerablyamong the various arcs and do not cared no earthquakeswith a depth greater than appear to be associatedwith any particular level 690 km during the period 1961-1967. These of depth in the mantle, contrary to general depths are near the region of the mantle in opinion. As shown in Figure 13, the depths of which gradients in the variation of seismic the deep maxima are approximately correlated velocitiesmay be high [Johnson,1967]. Ander- with the rates of convergencein the arcs as cal- son [1967a] arguesthat this region corresponds culated by Le Pichon. As will be shown later to a phasechange in the material. These depths (see Figure 16), the correlation is considerably may therefore be in some way related to the better betweenthe rate of convergenceand the boundary of the mesosphereas shown in Fig- length of the zonemeasured along the dip of the ures 1 and 14 and as discussed in the next sec- zone. Thus, the simplestexplanation, one direct tion. The second feature is the absence of consequenceof the model of Figure 1, is that maxima around 300 kin. Thus, in a worldwide the deep maxima are near the leadingparts of compositeplot of activity versusdepth, a mini- the down-goingslabs. mum in activity near this depth generallyap- Two features of the distributions shown in pears. Figures 12 and 13 may be related to certain Downward movemen• of lithosphere in the levels of depth in the mantle. Although the mantle: some hypotheses. Although the con- length of the seismiczone measuredalong the cept is so new that it is difficult to make dip of the zone exceeds 1000 km for several definitive statements, a brief speculative dis- cases,no earthquakeswith depths greater than cussionis in order to emphasizethe importance 720 km have ever been documented. The U.S. of these resultsin global tectonics.Figure 14, Coast and Geodetic Survey (USCGS) has lo- four hypotheticaland very schematiccross sec- 5878 ISACKS, OLIVER, AND SYKES tions of an island arc, illustratessome points that should be considered. Figure 14a shows a case in which the lithosphere has descendedinto the mantle beneath an island arc. In this model the lithospherehas not been appreciably modified with regard to its potential for earthquakesand the length of the submergedportion, l, and hencethe depths //////////// of the deepest earthquakesare dependenton Fig. 14a. Length l is a measureof the amount the rate of movementdown dip and the dura- of underthrustingduring the most recent period of tion of the current cycle of sea-floorspreading sea-floor spreading. and underthrusting. In Figure 14b the leadingedge of the descending lithospherehas encounteredsignificant resistance to further descent and has become distorted. In this situation, the depth of the deepestearthquakes in sucha zonedepends on the depth to the mesosphere.Sykes' [1966] analysisof the relocationsof earthquakesin the Tonga arc (seealso Figure 9) revealsthe pres- Fig. 14b. Lithosphere is deformed along its enceof contortionsof the lower part of the seis- lower edge as it encountersa more resistant layer mic zone which might indicate a phenomenon (the mesosphere). similar to that pictured in Figure 14b. This model has the interesting consequencethat a cycle of sea-floor spreading might be termi- nated or sharply modifiedby the bottomingof the lithosphereat certainpoints. Figure 14c indicatesschematically that the depths of the deepestearthquakes might de- pend on modificationof the lithosphereby its environment.In this model the depth of the //////////// deepestshocks depends on the rate of descent and the rate of modificationsor absorptionof of rate of underthrusting and time constant for the lithosphere. assimilation of slab by upper mantle. In Figure 14d the lower portion of the des- cendingpart of the lithosphereis not connected with the upper portion, possiblybecause it has pulled away as a result of a large density contrast between the sinking part of lithosphere and the surroundingmantle. Another possibility is that the lower piece representsa previous episode of movement, so that the break between the pieces then representsa period of //////////// quiescencein the surface movements. For ex- Fig. 14d. A piece (or pieces) of the lithosphere ample, the Spanish deep earthquake of 1954 becomes detached either by gravitational sinking [Hodgsonand Cock, 1956] and the very deep or by forces in the asthenosphere. earthquakesbeneath the North Island of New Zealand [Adams, 1963] might indicate isolated piecesof lithosphere.A variant of Figure 14d Figure 14 shows four possible configurationsof is the case in which movements of the ductile an underthrust plate of lithosphere in island arcs. Solid areas indicates lithosphere; white area, material of the asthenosphere,movements that asthenosphere; hatched area, mesosphere. could be quite different from the movements SEISMOLOGY AND NEW GLOBAL TECTONICS 5879 of the surficiallithospheric plates, could deform for extension and compressionalfeatures of the slabs and possibly break off pieces. For island arcs. The resultssuggest that there are example, the marked contortionsof the deep two basic types of focal mechanisms.The first seismic zone of Tonga may be explained by type is apparentlyconfined to shallowdepths such deformation.Thus, althoughthe evidence and directly accommodates,and thereforeindi- is at present only suggestive,such evidenceis cates the direction of, the movementsbetween important becauseof the implications of the the platesof lithosphere.The secondtype indi- hypotheseswith regard to the dynamics of cates stressand deformationwithin a plate of the system within the asthenosphere.Various lithosphereand includes,besides the deep and combinationsof the effects illustrated in Fig- intermediateearthquake mechanisms, the nor- ure 14 may alsobe considered. mal-faulting mechanismsat shallow depths Lateral terminationso• island arcs. The dis- beneaththe axis of the trench and, possibly, cussionsabove are based on, and apply largely the shallow earthquake mechanismslocated to, the structure of an island arc taken in a landward of the underthrust zone. The deep vertical section normal to the strike of an arc. earthquakezones may provide the most direct The three-dimensionalconfiguration of the arc source of information on the movement of mate- must also be considered.The plate model of rial in the asthenosphereand on the basic tectonicsprovides, in a simple way, for the questionof the relative importanceof the litho- termination of an island arc by the abrupt or sphericand asthenosphericmotions in driving gradual transitionto a transformfault, by a the convectivesystem. The global pattern of decreasein the rate of convergenceto zero, or motions between the plates, derived in part by somecombination of these.In the first case from the shallowfocal mechanism in islandarcs, the relative movement that is predominantly providesa severetest of the hypothesisof plate normal to the zone of deformation changesto movementsin generaland providesin particular relative movementthat is predominantlyparal- key evidencefor the conclusionthat island arcs lel to the zone. In the second case the pole are the major zonesof convergenceand down- governing the relative motion between the ward movements of the lithospheric plates. plates may be locatedalong the strike of the This evidence,including observationsof direc- feature. Isacks and Sykes [1968] describewhat tions as well as rates of movement,is discussed may be a particularly simple caseof the first in the next section. possibility.The northern end of the Tonga are COMPATIBILITY OF MOVEMENTS ON A appearsto end in a transformfault that strikes WORLDWIDE SCALE approximatelynormal to the are. In this case evidence is also found for a scissorstype of In this sectiondeformations along the world faulting in which the downgoingPacific plate rift system and along island arcs and major tears away from the part of the plate remain- mountain belts are examined for their internal ing at the surface. consistencyand for their global compatibility. Summary o• data on island arcs. The litho- The major finding is that these displacements sphere model of an island arc thus gives a can be approximatedrather precisely by the remarkably simple account of diverse and im- interactions and the relative movements of portant observedfeatures of island arcs. The large plates of lithosphere,much of the defor- existence and distribution of earthquakes in mation being concentratedalong the edges of the mantle beneath island arcs, the anomalous the platesand relatively little deformationbeing transmissionproperties of deep seismiczones, within the individual plates themselves.It has and the correspondencesin the variations of long been recognizedthat recent deformations seismicactivity and the orientationsof the focal of the earth's surface are concentrated in nar- mechanismsas functions of depth are all in row belts. These belts, which largely coincide agreementwith the conceptof a cooler, rela- with the major seismiczones of the world, tively strong slab moving through a relatively includethe world rift system,island arcs, and ductile asthenosphere.The bend in the slab re- such island-arc-like features as active mountain quired by this movement provides a simple belts and active continental margins. These means of reconciling the conflicting evidence major tectonicfeatures do not end abruptly; 5880 ISACKS, OLIVER, AND SYKES they appear to be linked togetherinto a global data were taken from $tauder [1962, 1968], tectonic scheme. Stauder and Udias [1963], $tauder and Boll- Continuity o)• seismicbelts and distribution o)• inger [1964, 1966a, b], Harding and Rinehart seismicactivity. Figure 15, a compilationof [1966], Ichikawa [1966], Sykes [1967, 1968], about 29,000 earthquake epicenters for the Banghat and Sykes [1968•, Isacks and Sykes world as reported by the U.S. Coast and Geo- [1968], and Tobin and Sykes [1968]. Although detic Survey for the period 1961 to 1967 [Bara- no attempt was made to ensure that the collec- zangi and Dotman, 1968], showsthat most of tion of mechanismsolutions represented all the the world's seismicactivity is concentratedin reliable previouswork, nonetheless,the data are rather narrow belts and that these belts may be thought to be representative; no attempt was regardedas continuous.Thus, if global tectonics made to select the data by criteria other than can be modeledby the interactionof a few large their reliability. In some cases such as the plates of lithosphere, this model can account aftershocksof the great Alaska earthquake of for most of the world's seismicactivity as effects 1964 and the aftershocks of the large Rat at or near the edges of the plates. Figure 15 Islands earthquake of 1965 [Stauder and Boll- also shows that the earthquakes occur much i•ger, 1966b; Stauder, 1968], only representa- more frequently, in general, in the zones of tive solutionswere included,since the numberof convergence,the arcs and arc-like features,than solutions for these regions was too large to in the zones of divergence,the ocean ridges. depict clearly in Figure 3. Solutions cited as Along the ocean ridges, where the less compli- reliable by Ritsema [1964] as well as those of cated processesof tectonics are apparently Honda et al. [1956], for example,were not used occurring, the zones are narrow; on the con- becausethey pertain to subcrustalshocks. tinents, where the processesare apparently From each mechanism solution used one of more complex, the zones are broad, and dis- two possibleslip vectors was chosenas indica- tinctive features are not easily resolved. Deep tive of the relative motions of the two interact- earthquake zones,indicated in Figure 15 only ing blocks of lithosphere.For each slip vector by the width of epicentralregions behind arcs, the arrow depicts the relative motion of the correspond to the zones of underthrusting. block on which it is drawn with respectto the Thus, all the major features of the map of block on the other side of the tectonic feature. seismic epicentersare in general accord with Since the double-couplemodel (or shear dis- the new global tectonics.No other hypothesis location) appears to be an excellentapproxi- has ever begun to account so well for the dis- mation to the radiationfield of earthquakes,it tribution of seismicactivity, which must rank is not possibleto choosefrom seismicdata alone as one of the primary observationsof seismol- one of the two possibleslip vectorsas the actual ogy. The details of the configurationof the motion vector (or alternatively to chooseone seismicbelts of Figure 15 are discussedfurther of two possiblenodal planes as the fault plane). in other sectionsof this paper. Nevertheless,the choiceof one vector is not Slip vectors. Figure 3 illustrates the distri- arbitrary but is justified either by the orienta- bution of these major tectonic features and tion of the vectors with respect to known tec- summarizesazimuths of motion as indicatedby tonic featuressuch as fracture zonesor by the the slip vectorsdetermined from variousstudies consistencyof a set of vectorsin a givenregion. of the focal mechanisms of shallow-focus earth- For earthquakeslocated on such major trans- quakes. Deep and intermediate earthquakesas form faults as the oceanicfracture zones,the well as shallow earthquakeswith normal fault- S•n Andreas fault, and the Queen Charlotte ing mechanismsnear trenches were not repre- Isl;mdsfault (Figure 3), one of the slip vectors sented in this figure, since these mechanisms is very nearly parallel to the transform fault on are not •hought to involve the relative displace- which the earthquake was located. Observed ments of two large blocksof lithosphere.Earth- surfacebreakage and geodeticmeasurements in quake mech,qnismswere included in Figure 3 someearthquakes, the alignmentof epicenters only when, by careful examinationof the first- alongthe strike of major transformfaults, the motion plots, we could verify that the slip linearity of fracture zones, and petrological vectorswere reasonablywell determined.These evidencefor intense shearing stressesin the SEISMOLOGY AND NEW GLOBAL TECTONICS 5881 5882 ISACKS, OLIVER, AND SYKES vicinity of fracture zonesconstitute strong evi- ruled out in the new global tectonics,rapid ex- dencefor making a rational choicebetween the pansion of the earth is not required to explain two possible slip vectors [Sykes, 1967]. The the large amountsof new materialsadded at the choiceof the other possibleslip vector (or nodal crests of the world rift system. This approxi- plane) for many of the oceanicfracture zones mate equality of surfacearea is, however,prob- would indicatestrike-slip motion nearly parallel ably maintained for periods longer than thou- to the ridge axis. On the contrary, earthquakes sandsto millionsof years,but minor imbalances along the ridge crest but not on fracture zones very likely could be maintained for shorter do not containa large strike-slipcomponent but periods as strains within the plates of litho- are characterizedby a predominanceof normal sphere. More exact knowledge of these im- faulting. balances could be of direct interest to the Evidence yor motions oy lithosphericplates. problem of earthquake prediction. One of the most obviousfeatures in Figure 3 is Figure 3 suggeststhat nearly all the east- that the slip vectors are consistentwith the west spreadingalong the East Pacific rise and hypothesis that surface area is being created the mid-Atlantic ridge is taken up either by along the world rift system and is being de- the island arcs of the westernPacific or by the stroyed in island arcs. Along the mid-Atlantic arc-like features bordering the west coasts of ridge, for example,slip vectors for more than Central and South America. Much of the north- ten events are nearly parallel to one another south spreading in the Indian Ocean is ab- and are parallel to their neighboringfracture sorbedin the Alpide zone,which stretchesfrom zones within the limits of uncertainty in either the Azores-Gibraltar ridge acrossthe Mediter- the mechanism solutions (about 20ø ) or the ranean to southern Asia and then to Indonesia. strikes of the fracture zones. Relative motions in the southwest Pacific. Morgan [1968] and Le Pichon [1968] showed Le Pichon's computed directions of motion in that the distribution of fracture zones and the the Tonga and Kermadec arcs of the south- observed directions and rates of spreading on west Pacific agree very closelywith mechanisms ocean ridges as determined from geomagnetic we obtained from a special study of that region data could be explainedby the relative motions [Isacks and Sykes, 1968]. Mechanismssouth of of a few large plates of lithosphere.They deter- New Zealand alongthe Macquarie ridge [Sykes, mined the poles of rotation that describethe 1967; Banghat and Sykes,1968] indicatea com- relative motion of adjacent plates on the globe. bination of thrust faulting and right-lateral Our evidence from earthquake mechanisms strike-slip motion. These data suggestthat the and from the worldwide distribution of seismic pole of rotation for these two large blocks is activity is in remarkable agreementwith their located about 10 ø farther south than estimated hypothesis.Although their data are mostly from by Le Pichon [1968]. Althoughthe Pacificplate ridgesand transform faults, earthquakemecha- is being underthrust in the Tonga and Kerma- nisms give the relative motions along island dec arcs and in northern New Zealand,this plate arcs as well as along ridges and transform is apparently being overthrust along the Mac- faults. quarie ridge (see also Summerhayes[1967]). Le Pichon used data from ocean ridges to In this interpretation the Alpine fault is a infer the direction of motion in island arcs. His right-lateral transform fault of the arc-arc type predicted movements (Figure 2), which are that connectstwo zones of thrusting with op- based on the assumption of conservation of posing dips. Computed slip vectors for this surface area and no deformation within the region alsoindicate a componentof thrust fault- plates of lithosphere,compare very closelywith ing either along the Alpine fault itself or in mechanism solutions in a number of arcs. This other parts of the South Island of New Zea- agreementis a strong argument for the hy- land. Wellman's [1955] studies of Quaternary pothesisthat the amount of surfacearea that is deformation, which indicate a thrusting com- destroyedin island arcs is approximatelyequal ponent as well as a right-lateral strike-slipcom- to the amount of new area that is created along ponent of motion along the Alpine and asso- the world rift system. Thus, although modest ciated faults, seem to be in general accordwith expansionor contractionof the earth is not this concept. SEISMOLOGY AND NEW GLOBAL TECTONICS 5883

Likewise,the Philippine fault appearsto con- larger strike-slipcomponent along the Aleu- nect a zoneof underthrustingof the Pacificfloor tian are as the longitudebecomes more westerly. near the Philippine trench with a region of Displacementsin the Kurile, Kamchatka,and overthrustingwest of the island of Luzon near Japanesearcs are nearly pure dip-slip and the Manila trench. Also, the existenceof a deep representunderthrusting of the Pacificplate seismiczone in the New Hebrides are that dips beneath the arcs. toward the Pacific rather than away from the The systemof great east--westfracture zones Pacific as in the Tonga are [Sykes, 1966] is in the northeastern Pacific (Figure 2) appar- understandableif the Pacificplate is beingover- entlywas formed more than 10 m.y. ago.Since thrust in the New Hebrides and underthrust in that time the directionsof spreadingchanged Tonga (Figure 1). The ends of thesetwo arcs from east-west to their present northwest- appear to be joined togther by one or more southeastpattern [Vine, 1966]. The morewest- transform faults that pass close to Fiji, but erly strike of the slip vectorsalong the Juan additional complicationsappear to exist in this de Fuca and Gorda ridges is consistentwith area. the hypothesisthat the area to the east of North Pacific. The uniformity in the slip di- these ridgesrepresents a small separateplate rections and the distribution of major faults [Morgan, 1968; McKenzieand Parker, 1967] along the margins of the North Pacific (with that was underthrust beneath the coast of perhaps somesystematic departure in the slip Washingtonand Oregonto form the volcanoes directionsoff the coast of Washingtonand Ore- of the Cascaderange. A few earthquakesthat gon) indicatethat only two blocksare involved have been detected in western Oregon and in the major tectonics[Tobin and Sykes,1968; Washingtonare apparently of a subcrustal Morgan, 1908; McKenzie and Parker, 1967]. In origin [Neumann, 1959; Tobin and Sykes, this scheme,the San Andreasfault, the Queen 1968]. The tectonicsof this block appearsto Charlotte Islands fault, and a seriesof north- be quite complex.Internal deformationwithin westerly striking faults in the Gulf of Cali- this block, as indicated by the presenceof fornia are interpreted as major transform earthquakes[Tobin and Sykes,1968], and the faults [Wilson, 1965a, b]. The observedrates small number of subcrustal events previously of displacementalong the San Andreas as de- mentionedsuggest that the tectonicregime in termined geodetically [Whitten, 1955, 1956] this regionis beingreadjusted. are very similar to the rates determinedfrom Depths of earthquakes,volcanism, and their the seismicity by means of a dislocationmodel correlation with high rates of underthrusting. [Brune, 1968]. These rates are in close agree- The depths of the deepestearthquakes, the ment with the rates inferred from magnetic presenceof volcanism,and the occurrenceof anomaliesfor the region of growing ridges at tsunamis(seismic sea waves) seemto be related the northwestern end of the San Andreas fault closelyto the presentrates of underthrustingin [Vine and Wilson, 1965; Vine, 1966]. Esti- isIand arcs. In the series of arcs that stretches mates of the total amount of offset along the from Tonga to the Macquarieridge the depths San Andreas [Hamilton and Meyers, 1966] of the deepest earthquakes generally decrease are comparableto the amount of offset needed from about 690 km in the north to less than to close the Gulf of California and to the width 100 km in the south [Hamilton and Evison, of the zonesof northeasterlystriking magnetic 1968]. The rates of underthrusting computed anomalies off the coast of Oregon and Wash- by Le Pichonalso decrease from north to south. ington [Vine and Wilson, 1965]. Volcanic activity, which is prominent north of Thus, the present tectonics of much of the the SouthIsland of New Zealand,dies out when North Pacific can be related to the motion of the deepestpart of the seismiczone shoalsto the Pacific plate relative to North America depths less than about 100 to 200 km. Simi- and northeastern Asia. The slip vectors strike larly, the depth of deepest activity in the northwesterly along the west coast of the Aleutian arc and the number of volcanoes United States and Canada, represent nearly (Figure 2) appear to decreasefrom east to pure dip-slip motion in southern Alaska and west as the rate of underthrusting decreases. the easternAleutians, and have an increasingly In this arc the rate of underthrustingdecreases 5884 ISACKS, OLIVER, AND SYKES becausethe slip vectors becomemore nearly in Hawaii [Gutenberg and Richter, 1954]. parallel to the arc. Volcanismand the depth Other generatingmechanisms, such as volcanic of seismicactivity increasespectacularly as the eruptionsand submarineslumping, cannot be slip vectorschange from a predominanceof excludedas the causativeagents for at least strike-slip motion in the western Aleutians to some tsunamis. largely dip-slip motion in Kamchatka, the Gutenberg [1939] argued that the epicen- Kurile Islands,and Japan. Most of the world's ters of severalearthquakes generating tsunamis active volcanoesare located either along the were located on land. Although the epicenter, world rift system or in regionsthat contain which representsthe point of initial rupture, intermediate-depth earthquakes. Hence, the may be locatedinland, the actual zoneof rup- latter volcanoesare presumablylocated in the ture in great earthquakesis now known to regionswhere the lithospherehas been under- extend for several hundred kilometers. In many thrust to depthsof at least 100 kin. of the great earthquakesassociated with island Tsunamis. Although some prominent seis- arcs and with active continental margins at mologists(e.g., Gutenberg [1939]) have argued leasta portion of the zoneof rupture is located that some of the largest tsunamis (seismic in water-coveredareas. Hence, the location of sea waves) are generatedby submarineslides, the epicenteritself is not an argumentfor the many other geophysicistsmaintained that sud- absenceof significantvertical displacementsin den dip-slip motion along faults during large nearby submarine areas. Utsu [1967] has earthquakesis the principal generatingmecha- shownthat, as a result of the anomalouszones nism for most of the world's more widespread in the mantle beneathisland arcs, locationsof tsunamis.L. Bailey (personalcommunication), shallow shocks based on teleseismic data are followinga lecture by Sykes, suggestedthat usually landward of the actual location.Both tsunami generationmight correlatewith areas improved locationsbased on a better knowl- of large dip-slip motion. This suggestionmay edge of seismicwave travel times in anoma- have considerable merit. lous regions such as island arcs and rapid Figure 2 showsthe epicentersof earthquakes determinations of focal mechanismsmay be for which tsunamis were detected at distances of substantial value in tsunami warning sys- of 1000km or greater [Heck, 1947; Gutenberg tems. and Richter, 1954; U.S. Coast and Geodetic Length o] seismiczones in island arcs. The Survey, 1935-1965]. The distancecriterion was systematicchanges in the seismiczones in the used to eliminate waves of more local origin, Aleutians and in the southwestPacific suggest some of which actually may be related to that the lengths of zones of deep earthquakes seichesor to submarine slides. Although this might be a measure of the amount of under- compilationundoubtedly is not completeeven thrusting during the last several million years. for the present century, the conclusionsdrawn Using the maps of deepand intermediate-depth here should not be seriously affected by the earthquakesprepared by Gutenbergand Richter choice of data. [1954], Oliver and Isacks [1968] estimated Since most of the earthquakes generating the area of these zones and divided the total tsunamis are located in regions that are char- area by the length of the world rift system acterized by a high rate of dip-slip motion (about two great circles) and by 10 m.y., (Figure2), the two phenomenaappear to be which is the duration of the latest cycle of eausallyrelated in many eases.Although most spreadingbased on data from ocean-floorsedi- of the world's largest earthquakes occur in ments and from magnetics [Ewing and Ewing, these areas and are characterized by a large 1967; Vine, 1966]. They obtained an average componentof thrust faulting,large earthquakes rate of spreading for the entire rift system of along major strike-slipfaults near the coasts 1.3 cm/yr for the half-velocity. This value is of southeast Alaska and British Columbia have reasonablefor the average velocity of spread- not generatedsea wavesthat were observable ing along the world rift system. at distancesgreater than about 100 km. An The hypothesis that the lengths of deep earthquakelocated off the coastof California seismic zones are a measure of the amount of in 1927, however,generated a wave detected underthrusting during the past 10 m.y. is SEISMOIZ)GY ANI) NEW GL()BAL TECT()NICS 5885 examinedin greater detail in Figure 16 and in underthrusting. Since the calculated slip rates Table 1. Figure 16 illustrates the lengths of and some of the measured lengths may be the seismic zones in various arcs and the cor- uncertainby 20% or more, the correlationbe- respondingrates of underthrustingas calculated tween the two variablesis, in fact, surprisingly by Le Piehah [1968] from observedvelocities good. of sea-floor spreading and the orientation of Although six points fall well above a line of fracture zones along the world rift system. In unit slope, which representsan age of 10 m.y., nearly all casesthe regions with the deepest all but one of the lengths are within a factor earthquakes (and hence the longest seismic of 2 of the lengths predicted from the hy- zonesas measuredalong the zonesand perpen- pothesis that these zones represent materials dicular to the arcs) correspondto the areas underthrust during the last 10 m.y. Three of with the greatest rates of underthrusting; re- the more discrepantpoints, which are denoted gions with only shallow- and intermediate- by crosseson the figure, representa small num- depth events are typified by lower rates of ber of deep earthquakes that are located in

1500

Japan -Manchuria

I• i•I SouthAmerica • Marianos+ Ryukyu

_qlooo New Hebrides•l ß 12øS- 17ø S LLI Indones•a•l Philippines + Z Tonga•11 PaLau

NO XDeepest Events Kurile-Kamchatka New Zealand • X Spanish Deep 2•_ Earthquake

1.1.1500 Karmadec 03 New Zealond•l North Island •1 S. Alaska + E. Aleutians

•ICentral America i- Andaman Arc z outh Sandwich

New ZeaLand-South Is + Macquar•e R•dge ,leut•a 170øE

o 5 IO CALCULATED SLIP RATE J_ ARC (CM/YR) Fig. 16. Calculated rates of underthrusting [Le Pichon, 1968], and length of seismiczone for various island arcs and arc-like features (solid circles), for several unusual deep events (crosses),and for Southern Chile (diamond). The solid line indicates the theoretical locus of points for uniform spreadingover a 10-m.y. interval. 5886 ISACKS, OLIVER, AND SYKES

TABLE 1. Comparison of Estimated Slip Rates for Three Island Arcs Values are in centimetersper year.

Slip Rate* Assuming Slip Rate from Length SeismicZone Shallow-Focus Calculated Slip Rate, Measures Amount Seismicity, after Data from Ocean Ridges, Underthrust during Island Are Brune [1968] after Le Pichon [1968] Last 10 m.y.

Tonga 5.2 7.6 8.4 Japan 15.7 8.8 16.3 Aleutians 3.8 6.1 4.0

* Correctedfor azimuth of slip vectorusing data of Le Pichon[1968]. Rates givenfor the three methods are the magnitudesof the total slip vectors and not the magnitudesof the dip-slip componentsused in Figure 16. unusual locationswith respect to the more ac- of the normal oceaniclithosphere. Hence, this tive, planar zones of deep earthquakes.They value is not an unreasonablefirst approximation include the unusual deep Spanish earthquake to the time constant for assimilation of the of 1954, three deep earthquakesin New Zea- lithospherein island arcs. It shouldbe recog- land, and a few deep events under Fiji that nized, however, that the time constant for appear to fall between the deep zones in the ocean ridges may be uncertain by more than Tonga and New Hebridesarcs. a factor of 2 because of scatter in heat flow Other estimates o/ slip rates in island arcs. data. At present,it doesnot seempossible to Slip rates for Tonga, Japan, and the Aleutians ascertain which of the two alternative pro- (Table 1) were calculated (1) from the disloca- posals(Figure 14a or 14c) governsthe lengths tion theory usingdata on the past occurrenceof of seismic zones in island arcs. earthquakes[Brune, 1968], (2) from the ob- With the exception of the data for south- served spreadingvelocities along ocean ridges ern Chile (the diamond in Figure 16) there assumingthat all of the spreadingis absorbed are no points lying significantlybelow the 10- by underthrustingin island arcs [Le Pichon, m.y. isochron. A reduction in slope at the 1968], and (3) from the lengthsof the seismic higher spreadingrates might occurif the litho- zones assumingthese lengths are a measure of spherewere suddenlymodified at a givendepth the amount of underthrusting during the last throughwarming or a phasechange that would 10 m.y. For each arc the three rates are within prevent deeper earthquakesfrom occurring. a factor of 2 of one another. These processesapparently have not led to The 10-m.y. isochron. Two hypothesesmay significantdecreases in the lengthsof the seis- explain the correlationbetween the length of mic zones comparedwith those predicted by seismic zones and the computed rate of under- the 10-m.y. isochronin Figure 16. thrusting (Figure 16). One theory, which was Southern Chile. In southern Chile between mentionedearlier, assumesthat the presentseis- 46ø and 54øS,the absenceof observabledeep mic zones in island arcs were created during activity,the near-absenceof shallowseismicity, a recent episode of spreading which began and the presence of a sediment-filled trench about 10 m.y. ago (Figure 14a). In the other [Ewing, 1963; Hayes, 1966] are in obvious hypothesis,10 m.y. is regardedas the approxi- conflict with the predicted length of the seis- mate time constant for assimilation of the mic zone. Le Pichon'somission of two very lithosphereby the upper mantle (Figure 14c). active features,the West Chile ridge and the Since the zone of anomalouslyhigh heat flow Galapagosrift zone (Figure 15) in his analysis on ocean ridges appears to be confinedto re- may, however,explain this discrepancy.Wilson gions less than about 10 m.y. old [McKenzie, [1965a] argued that since Antarctica is almost 1967; Le Pichon and Langseth, 1968], 10 m.y. surroundedby spreadingridges, the coast of is a reasonable time constant for the creation South America south of its juncturewith the SEISMOLOGY AND NEW GLOBAL TECTONICS 5887 West Chile ridge at 46øS shouldnot be seismi- or the typical islandarcs, occupiesa very broad cally active.An alternativeexplanation for the region (Figure 15). Spreading in the Indian near-absenceof shallowseismicity is that activ- Ocean is apparently being absorbedin several ity duringthe past70 yearsof instrumentalseis- subzoneswithin the Alpide belt. This conclu- inology is not representativeof the long-term sion is supported by the widespreaddistribu- activity. This explanationis, however,difficult tion of large shallow earthquakesand by the to apply to the hypotheticaldeep seismiczone relatively small number of intermediate- and that is predicted from Le Pichon's calcula- deep-focusearthquakes [Gutenberg and Rich- tions. Hence, we feel that either Wilson'spro- ter, 1954]. posal or some other explanationis neededto Although, in principle, it seems reasonable account for the tectonics of southern Chile. to describe the tectonics of Eurasia by the Departuresfrom the 10-rn.y.isochron. Sev- interaction of blocks of lithosphere,it is not eral factors could explain the six points that yet clear how successfulthis idea will be in fall well above the 10-m.y. isochronin Figure practice becauseof the large number of blocks 16: (1) Somepieces of lithospherebecame de- involved. The interaction of blocks of litho- tached from the main dipping zones of deep sphereappears to be much more complexwhen activity by active processes,such as gravita- all the blocks are continentsor piecesof con- tional settling of slabs of lithosphereor continents than when at least one is an oceanic vection currents in the asthenosphere(Figure block. In addition to activity in the Alpide 14d). (2) The initiation of underthrustingand belt, earthquakesin East Africa, northern Si- spreadingwas not simultaneousin all regions. beria, and western North America (including (3) Someor all anomalousdeep eventsmay be Alaska) are more diffusedin areal extent (Fig- related to previous episodesof underthrusting. ure 15). In contrast, seismic zones associated (4) The computed slip rates are not correct. with oceanridges, island arcs, and many active (5) If the thermal time constant for assimi- continental margins appear to be extremely lation of the lithosphere is generally about 10 narrow and well defined. Several factors may m.y. (Figure 14c), the anomalouspoints may explain these differences:(1) The lithosphere represent pieces of lithosphere with anoma- may be more heterogeneousin some or all con- lously high time constants so that they are tinental areas and, hence, may break in a not completelyassimilated. more complex fashion. (2) Old zonesof weak- Although some of the computed spreading ness in continental areas may be reactivated. rates may be in error (particularly the values (3) Becauseof its relatively low density, it for Indonesia and South America may be in may not be possibleto underthrust a block of error becausethe magnetic data in the Indian continental lithosphere into the mantle to Ocean are poor and becausethe West Chile depths of several hundred kilometers. This ridge and the Galapagosrift zone (Figure 14) third point is supportedby the relatively large were not included in Le ?ichon's analysis), it areas of older continental rocks. is difficult to argue that theserates are greatly As much of the sea floor appearsto be rela- in error for each of the six anomalouspoints. tively young, plates of oceanic lithosphere The possible ramifications of the various al- probably do not contain a large number of ternative proposalsshould be exciting enough zones of weakness, whereas the continental to encouragefurther study of these new ap- plates, which are older, seem to contain many proaches to the distribution of deep earth- zones of weakness. Except for a few earthquakes. quakes along the extensionsof fracture zones, Interaction of continental blocks of litho- the ocean basins are extremely quiet seismi- sphere. The Alpide belt, which comprises cally. Continents, however, exhibit a low level much of the Mediterraneanregion, the Middle of activity in many areas that do not appear East, and large parts of central and southern to be undergoingstrong deformation.This con- Asia, was not included in Figure 16 because trast in activity does not appear to be an it is very difficult to define the total amount artifact of the detection system but rather of underthrusting.Seismic actvity in the Alpide appears to be related to the different charac- zone,unlike that alongeither the oceanridges ter of the oceanicand continentallithosphere. 5888 ISACKS, OLIVER, AND SYKES One exceptionto this rule is the near-absence African rift valleys;if spreadingfrom the east- of observable activity in Antarctica. Unlike ern flank of the East Pacific rise near Mexico the other continents,Antarctica is almost sur- is absorbed only a few hundred kilometers rounded by ocean ridges that are probably from the rise in the Middle America trench, migrating outwardly [Wilson, 1965a]. Hence, it is difficult to imagine how materials on the antarctic plate may not be subjected to the western flank of the rise travel more stressesas large as the stressesoccurring in than 10,000 km before they are absorbed other continents. in the trenches of the western Pacific. A1- Summaryof evidencefor globalmovements. though some of the fracture zoneson ocean The concentrationof seismicactivity and the ridges may be as long as 1000 km, some consistencyof individual mechanism solutions of these zones are less than 50 to 100 km for mid-oceanridges, for island arcs, and for apart. It is almost impossibleto believe that most of the world'sactive continentalmargins these long narrow strips reflect the shapeof indicate that tectonic models involving a few independent convection cells. These and sev- plates are highly applicable for these areas. eral similar dilemmasmay be resolvedif a Since individual mechanism solutions in a layer of greater strength (the lithosphere) given regionare highly coherent,each solution overliesa regionof much lesserstrength (the may be regarded as an extremely pertinent asthenosphere). datum. It is not necessaryto resortto complex The configurationof the asthenosphereand statisticalmethods for analyzing the relative of the various pieces of lithospheremay in motionsof theselarge blocks.For these re- someways be analogousto blocksof ice float- gions much of the scatter in many previous ing on water. Althoughthe surfacepattern of mechanismstudies appears to be largely a re- the ice may be very complex,the pattern of sult of poor seismicdata and errors in anal- convectionin the water below or in the air ysis.Although the conceptof interactingblocks abovemay be very simple,or it may be com- of lithospheremay not be as easilyapplied to plex and of a completelydifferent character the complexinteractions of continentalblocks than that of the motionsof the ice. This anal- in suchareas as the Alpide belt, a consistent ogymay be relevantto the morecomplex non- (but complex)tectonic pattern may yet emerge symmetricaltectonic pattern exhibitedby the when the pattern of seismicityis well defined earth's surface.A lithosphericmodel readily and when a sufficientlylarge numberof high- accountsfor the asymmetricalstructure of is- qualitymechanism solutions is available.Thus, land arcs and for the symmetricalconfigura- the new global tectonicsamong other things tion of oceanridges. It alsoexplains the rather appearsto explainmost of the major seismic smoothvariations in rates of spreadingin a zonesof the world, the distributionand con- given ocean. figurationof the zonesof intermediateand deep In this modelthe ridgesand islandarcs are earthquakes,the focal mechanismsof earth- dynamicfeatures, and hencethey appearto quakesin many areas,and the distributionof move with respectto one another.Thus, a a largenumber of the world'sactive volcanoes. regionof extensionaltectonism may be located between two spreading ridges. What is de- ADDITIONALEVIDENCE FOREXISTENCE OFLITI-IO- manded in this model, however, is that sur- sr•ERE,AST•ENosr•sRS, ANDMssosr•sRE facearea be conservedona worldwidebasis. Complextectonic patterns near the earth's To what extent the motionsof large platesof surface. A number of objectionshave been lithosphere are coupled to motions in the raised to models of sea-floorspreading that asthenosphereremains an unsolvedproblem. involve simplesymmetrical convection ceils ex- Depths of earthquakesalong the world rift tending from great depths to the surfaceof system. The existenceof a layer of strength the earth. They include such statements as: near the earth's surfaceis compatiblewith the Sincethe Atlantic and Indian oceansare both observationthat all earthquakeson the ocean spreading,the tectonicsof Africa shouldbe of ridgesare apparently of shallowfocus (i.e., a compressionalnature rather than of an ex- lessthan a few tensof kilometersdeep). Using tensional nature, as reported for the East a dislocationmodel, Brune [1968] estimated SEISMOLOGY AND NEW GLOBAL TECTONICS 5889 that the zone of earthquake generation for [1968] observed that ridges with spreading some oceanicfracture zonesappears to be less rates higher than about 3 cm/yr are typified than 10 km in vertical extent. by fairly smooth relief, a thin crest, and the The San Andreas fault of California appears absenceof a median rift valley. Ridges with to be a major transform fault connectinggrow- lower spreadingrates are, however,character- ing ocean ridges off the west coast of Oregon ized by high relief, a thick crust, and a well- and Washingtonwith spreadingridges in the developedmedian valley. van Andel and Bowin Gulf of California (Figure 3). Observedseis- [1968] suggestthat materials undergoingbrit- mic activity along the San Andreas system is tle fracture may be thin at sites of faster confinedto the upper 20 km of the earth, and spreading because higher temperatures might much of this activity is confinedto the upper occur at shallower depths. The crest of the 5 or 10 km [Press and Brace, 1966]. Most East Pacific rise is characterizedby heat flow estimatesof the depths of faulting in the great anomaliesthat are broader and apparently of San Franciscoearthquake of 1906, the Imperial greater amplitude than the anomalies associ- Valley earthquakeof 1940, and the Parkfield ated with ridge crestsin the Atlantic and In- earthquake of 1966 are within the upper 10 to dian oceans[Le Pichon and Langseth, 1968]. 20 km of the earth [Kasahara, 1957; Byefly Thus, higher temperatures beneath the East and DeNoyer, 1958; Chinnery,1961; McEvilly Pacificrise may suppressthe buildup of stresses et al., 1967; Eaton, 1967]. Since displacements large enough to generate observable seismic of at least 100 km (and probably 300 to 500 activity. km) probably have occurred along the San If the occurrence of earthquakes and the Andreas fault [Hamilton and Meyers, 1966], presenceof a median rift valley correlate with it is almost certain that deformationby creep high strength and their absences(in spite of without observableearthquake activity must large deformations)correlate with low strength, be occurring at depths greater than 20 km. the lithospheremay be very thin or almost Thus, an upper layer of strength in which nonexistent near the crest of the East Pacific earthquakes occur associatedwith a zone of rise. The lithospheremust be present,however, low strength below seemsto be demandedfor within a few tens of kilometers or less of the California and for other parts of the world crest to account for the much higher seismic rift system. activity alongfracture zones.It is possiblethat Variations in thickness of the lithosphere seismicactivity along the crest of the East Pa- along the world rift system. Although earth- cific rise may occur mainly as small earthquakes on the mid-Atlantic, mid-Indian, and quakes that either are not detected or are Arctic ridgesoccur along the ridge creststhem- rarely detected by the present network of selvesas well as along fracture zones,observ- seismographstations. A microearthquakestudy able seismicactivity along much of the East using either hydrophonesor ocean-bottomseis- Pacific rise is concentratedalmost exclusively mographscould furnish important information along fracture zones [Menard, 1966; Tobin about the mode of deformation at the crest and Sykes, 1968]. With the exceptionof the of thesesubmarine ridges. Gorda ridge off northern California,which ap- High-frequency S• waves crossing ocean pears to be spreadingrelatively slowly,earth- ridges. Molnar and Oliver [1968] studied S• quakesare only rarely recordedfrom the crest propagation in the upper mantle for about of most of the East Pacific rise. Since evidence fifteen hundred paths; they did not observe from magnetic anomaliesindicates a relatively any high-frequencyS• waves for ocean paths fast rate of spreading (greater than 3 cm/yr that either originateat or crossan active ridge for the past i m.y.) along much of the East crest. Their observationsalso suggestthat the Pacific rise [Heirtzler et al., 1968] and since uppermostmantle directly beneath ridge crests earthquakesare presenton the boundingtrans- is not included in the lithospherebut that the form faults, much of the crest of this rise ap- uppermost mantle must be included in the pears to be spreading by ductile flow with lithosphere beyond about 200 km of the crest little or no observableseismic activity. in order to explain the propagationof high- Menard [1967] and van Andel and Bowin frequency $• from several events on some of 5890 ISACKS, OLIVER, AND SYKES the larger fracture zones.The distance for any mic energy in earthquakes;island arcs and given ridge may dependon the spreadingrate, similar arcuate structures contribute more but the data were inadequate to confirm this than 90% of the world's energy for shallow assumption.The thinning or near-absenceof earthquakesand nearly all the energyfor deep lithospherein a zonenear the ridge crestwould earthquakes [Gutenberg and Richter, 1954]. be compatible with the magmatic eraplace- The relatively small amount of seismicactivity ment of the lithospherenear the crests of on the ridge systemand its largely submarine oceanridges and with a gradual thickeningof environmentprobably explain why the signifi- this layer as it cools and moves away from cance and the worldwide nature of this tec- the crest. This thinning may also explain the tonic feature were not clearly recognizeduntil maintenance of near-isostatic equilibrium over about 12 years ago. ocean ridges [Talwani et al., 1965a]. Although Richter [1958] lists more than 175 Thickness o• the lithosphere •rom heat flow very large earthquakes (magnitude M greater anomalies. If the spreading rates inferred than or equal to 7.9) as originatingin arcs and from magnetic anomalies are accepted, the arc-like features, he reports only 5 events of calculated heat flow anomaly over ridges for this size for the world rift system. Whereas a model of a simple mantle convectioncurrent the largest event on the ridge system was of that extends to the surface is not compatible magnitude 8.4, the largest known earthquakes with the observedheat flow anomaly [Langseth from island arcs were of magnitude 8.9. This et al., 1966; Bott, 1967; McKenzie, 1967]. differencein magnitude correspondsto about These observationsare, however, compatible 8 times as much energy in the larger shocks. with the computed heat flow for a simple Most (and perhaps all) of the earthquakeson model of cooling lithospheric plate approxi- the ridge system with magnitudesgreater than mately 50 km thick [McKenzie, 1967]. In this about 7 appear to have occurredalong major model the heat flow anomaly results from the transform faults. cooling of the lithosphere after it is emplaced Thus, the maximum magnitudesof earth- magmatically near the axis of the ridge. Be- quakes during the last 70 years for various cause of the scatter in the heat flow data and tectonic features may be summarized as fol- because of the simplifying assumptionsused lows: island arcs, 8.9; major fracture zones, to obtain the estimate of 50 km, the value for 8.4; ridge crestsin the Atlantic, Indian, and the thickness either may be uncertain or may Arctic oceans,about 7; most of the crest of vary by a factor of 2 or more. The occurrence the East Pacific rise (with the exceptionof the of earthquakes as deep as 60 km beneath Gorda ridge), few (and possibly no) events Hawaii [Eaton, 1962] may also be used as an larger than 5. The relative frequenciesof very estimate of the thickness of the lithosphere. large earthquakesand possiblythe upper lim- Orowan [-1966] tried to fit the observed heat its to the sizes of earthquakes appear to be flow pattern for ridges with models in which related to the area of contact between pieces the crust is stretched on the ridge flanks and of lithosphere that move with respect to one the rate of spreadingis a function of distance another. On the crest of the East Pacific rise from the ridge. His model does not appear to the zone of contact between brittle materials be compatible either with the symmetry of is either absent or is confined to a thin sur- magnetic anomaliesclose to the ridge or with ficial layer; at the crests of other ridges the the narrow width of the seismic zones on most lithosphere appears to be somewhat thicker. ocean ridges. For fracture zones the thickness of the litho- Maximum sizes o[ earthquakes. The sizes sphere may be as great as a few tens of kilo- of the largest earthquakesalong ocean ridges meters; thus, the maximum magnitudes may and along island arcs may be related to the be limited by the length of the fracture zone thickness of the lithosphere in each of these and by the thickness of the lithosphere. In tectonic provinces.The world rift system (in- island arcs the thicknessof the lithospheremay cluding California, southeastAlaska, and east be greater than that on the ridges becausethe Africa) accountsfor lessthan 9% of the world's temperatures beneath ridges are higher and earthquakesand for less than 6% of the sets- perhapsbecause some material is added to the SEISMOLOGY AND NEW GLOBAL TECTONICS 5891 bottom of the lithosphere as it moves away estimates represent real differences in thick- from a ridge crest. Since the dip of seismic ness or merely differencesin the relations be- zonesin island arcs usually is not vertical, the tween thicknessand the various physical pa- area of contact between plates of lithosphere rameters measured. may be increased by this factor alone. For The asthenospherein the three-layeredmodel example, the zone of slippage in the great shownin Figure I also roughly coincideswith Alaska earthquake of 1964 appears to dip the low-velocityzone for S waves [Gutenberg, northwesterly at a shallow angle (about 10ø) 1959; Dotman et al., 1960; Anderson, 1966; and to extend for about 200 km perpendicular Ibrahim and Nuttli, 1967], regions of either to the Aleutian island arc [PlaCket, 1965; low velocity or nearly constant velocity for Savage and Hastie, 1966]. Since the dip is P waves[Lehmann, 1964a, b] and of low den- shallow, this zone does not extend more than sity or nearly constantdensity [Pekeris, 1966], about 40 km below the surfaceat its deepest a region of high attenuation for seismicwaves, point. Thus, it is not unreasonablethat much particularly S waves [Andersonand Archam- greater amountsof elastic energy can be stored beau, 1964; Anderson, 1967b; Oliver and and releasedalong island arcs than along the Isacks, 1967], and a low-viscosityzone in the mid-ocean ridges. upper mantle [McConnell, 1965]. Although Since single great earthquakes apparently these observationsyield very little information have not involved a rupture that extended about the actual mechanisms of dissipation, along the entire length of the larger island arcs they are consistentwith the hypothesis that [Richter, 1958], some other factor appearsto the asthenosphereis a region of low strength limit the length of rupture and hencethe maxi- boundedbelow by the mesosphere,a region of mum sizes of earthquakes.Tear faults (i.e., greater strength. The simplest explanation for transform faults of the arc-arc type) within these phenomenais that the closestapproach the upper thrust plate may divide island arcs to melting (or partial melting) occurs in the into smaller subprovincesthat are not com- asthenosphere.That relative displacementsof pletely coupledmechanically. the earth's surfacemay be modeledby a series Some properties o• the lithosphereand as- of moving plates is a strong argument for a thenosphere. Thus far we have defined the region of low strength (the asthenosphere)in lithosphereas a layer of high strengthand the the upper mantle. Nevertheless, the physical asthenosphereas a layer of low strength.Since properties and configuration of the astheno- the rheologicalproperties of the mantle are sphereand the lithospheremay vary from place not at all well understood,we have purposely to place. In fact, variations of these types seem usedthe term strength in a generalsense with- to be required to account for many of the com- out being more specificabout the actual mecha- plexities of the outer few hundred kilometers nismsof deformation.Thus, it is not at all cer- of the earth. The evidence for such lateral tain that varioustechniques for measuringthe variations is now so strong as to demand re- thicknessand the presenceof the lithosphere evaluation of studies of velocity and Q struc- necessarilyyield comparablevalues since these ture basedon modelsthat consistonly of con- methods probably sample different physical centric sphericalshells. parameters.It is encouraging,however, that estimatesof the thicknessof the lithosphere CONFLICTING SEISMOLOGICAL EVIDENCE from analysesof heat flow anomalies,the ab- A•D SOME PROBLEMS senceof earthquakes(in spite of large defor- The new global tectonics,in one form or mation), the requirementsof isostasyand the another, has been remarkably successfulin ex- maintenanceof mountains[Daly, 1940], the plaining many gross features and observations amplitudesand wavelengthsof gravity anoma- of geology,but its developmentmust be con- lies [McKenzie, 1967], the propagationof tinued in an effort to establisha theory that high-frequency$, waves, and a transition in is effective throughout the earth sciencesat the types of earthquake mechanismsin island levels of increasinglygreater detail. Many arcsare about100 km or less.It is not clear, difficult problemsremain. A quantitative under- however, to what extent the variations in these standing of the processesin the mantle that 5892 ISACKS, OLIVER, AND SYKES result in the observed surface features is ur- does not readily fit into the new global tec- gently needed.The roles of the earth's initial tonics at present does exist, however.For ex- heat,gravitational energy, radioactive heat, and ample, the locationsof the Spanishdeep earth- phase changesmust be understood.The me- quake and certain deep shocksin New Zealand chanics of flow in the mantle is little known. and Fiji are not easily explained by current The history of sea-floorspreading through geo- thinking, as has been pointed out above. The logic time and the relation between the vast patterns of minor seismicity,including an oc- quantities of geologic observations and the casionallarge earthquake in certain continental hypothesesmust be worked out. A vital un- areas such as the St. Lawrence Valley and the answeredquestion is why island arcs and ocean Rocky Mountains and broad regional scatter- ridges have their particular characteristicpat- ing of epicenters as in east Africa, are not terns. These grand questions are of concern simply explainedas yet, nor is the almost com- to seismologists,not only becausethere is gen- plete lack of earthquakes in the oceaniclitho- eral interest in the hypothesesbut also because sphere, which presumablycan and does trans- the solutionsmay very well be dependent on mit stress over large distances.The occurrence evidence from seismology.In the absenceof of large active strike-slip faults that trend all but the most preliminary attempts to solve alongthe arc structure,such as the Alpine fault these problems,however, it is difficult to spec- in New Zealand,the Philippinefault, the North ulate on the direction and force of the evi- Anatolian fault in Turkey, and the Atacama dence. At the present there appears to be no fault in Chile, ofters some diffculties.Tentative evidence from seismologythat cannot eventu- explanationshave been proposedfor the Alpine ally be reconciledwith the new globaltectonics and Philippine faults; therefore, this matter in some form. Some traditional views that once may be resolved.Perhaps related to this prob- would have appeared contradictory (such as, lem is the occurrence in island arcs of occasional for example, the assignment of substantial earthquakeswith large strike-slip components rigidity to the entire mantle over a long time along faults subnormal to the arc. The tec- interval because of efficient propagation of tonics of the arc can hardly be as simple as shear waves with periods of about 10 sec) that implied by Figure 1. have long been discountedon the basis of in- A generalproblem involving seismology con- formation on glacial rebound, gravity, and, cerns contrasts between continental and oce- more recently, creep along long strike-slip anic areas [MacDonald, 1964, 1966]. Some faults, such as the San Andreas fault. That studies [e.g., Brune and Dotman, 1963; Tok- deep earthquakes generate shear waves and sSz and Anderson,1966] show corresponding radiate waves in a pattern characteristicof a structural differencesto depths of severalhun- shear dislocation can hardly be taken as evi- dred kilometers.Are such differencescontrary dence for strength throughout the mantle in to the new global tectonics?Might the obser- that depth range in light of a hypothesisthat vations be equally well satisfied by other suggeststhat the mantle in the deep earth- models that are compatiblewith requirements quake zonesis much different from the mantle of the new hypotheses?A seriousdifficulty at comparabledepths elsewhere. may arise here if the.y cannot.Comparable and The arc-like patterns of the active zone is related problems arise in other disciplines. but one of the problemsof a subjectthat might Observationsuggests that heat flow per unit be termed 'lithosphere mechanics.' Others are area is about the same under the oceans as the shape of the deep zones,which sometimes under the continents. If it is assumed that the appear to be near-planar after turning a rather heat flux is largely due to radioactive decay sharp corner near the surface; the distribu- and that radioactivity is heavily concentrated tion of stress and strain in the lithosphere, in certain continental rocks, lateral hetero- not merely in active seismicareas but through- gcneity of the upper mantle is required and the out the world; the interaction of lithosphere mixing predicted by the new tectonicsmay and asthenosphere;and flow in the astheno- be so great as to destroy such heterogeneity. sphere. Perhaps the amount of radioactivity in the Some specificevidence from seisinologythat earth has been highly overestimated,as has SEISMOLOGY AND NEW GLOBAL TECTONICS 5893 occasionallybeen suggested[Verhoogen, 1956], be carefullyand thoroughlyconsidered for indi- and much of the heat lost at the surface is cationsof new directionsfor, and new attitudes transportedfrom the deep interior by convec- toward, seismologicalresearch. This section is tion. More accurate determination of deep largely speculativeand someof the points may structure by seismologicaltechniques is thus seem farfetched. If, however, a basic under- relevant to the problem of the amount of standing of global tectonicsis imminent, even radioactivity in the earth and its distribution. the most imaginative and wisest forecastof its Seismologyis alsovitally linked with heat flow effect on seismologyis likely to be too conserva- in the island arcs, where low heat flow values tive. Nor is seismologyalone in this regard, for appear to correlatewith zonesof descending all branchesof geologyand geophysicsrelated lithospherebut anomaloushigh values appear to the earth's interior will be comparably over the deep seismiczones, and at the ocean affected.The assumptionthat a major advance ridges,where high heat flow may correspond in our understandingof global tectonics has with low strength and hence low seismicac- been achievedis tacit in the following. tivity. Seismicity is an important branch of seis- Other disciplinesare also involved. The new mologyand one that will be strongly affectedby structures based mainly on seismologicalinfor- the impact of the new global tectonics.Such mation must be tested against gravity data. basic questions as why earthquakes occur The record in the ocean sedimentssurely pro- largely in narrow belts separatedby large stable vides the most completehistory of the ocean blocks, why these belts are continuous on a basins and hence must provide insight into worldwidescale, why they branch, why certain seismologicalprocesses. Perhaps the crucial details of their configurationare as they are, evidenceon the validity of new globaltectonics why intermediateand deep earthquakesoccur will come from coresof the entire sedimentary in someareas and not others are in the process column of the ocean floor. of being answeredtoday. There is every indica- In petrologyan interestingquestion concerns tion that the relation between seismicactivity the origin of the belts of andesiticvolcanoes. and geologywill soonbe understoodto a much Coats [1962] proposed that ocean crust and greater extent than contemplated heretofore. sediments were thrust into the mantle under Improved accuracy in hypocenter location re- the Aleutian arc and subsequentlyerupted with sulting from a better knowledgeof velocity mantle rock to account for the petrology of structure will facilitate this development. It those islands. Can all the andesites of the active will also assist in solving seismology'schief tectonic belts be explainedin this manner? Are problemof a politicalnature, distinguishingbe- there any young (lessthan 10 m.y.) andesiresof tween earthquakesand undergroundnuclear ex- the same type that are not associatedwith plosions.The self-consistentworldwide pattern active deep earthquake zones and arcs? Can of focal mechanisms will also be valuable here. this sort of information be used to identify Perhaps location of new seismographstations ancient arcs ? in the proper relation to high-Q zoneswithin The importantproblems arising from the new the earth will result in improved detection global tectonics are many; some are crucial. capability. Important questions still remain. Evidence from seismology against the new For example,why have the earthquake belts globaltectonics appears, however, to lack force. and the associated tectonic belts assumed their present configuration? What can be learned EFFECTS OF •'EW GLOBAL TECTONICS ON about paleoseismicityand its relation to mod- SEISMOLOGY ern seismicity?How can the pattern of minor That seismologyis providin.g abundant and seismicity,including the occasionalmajor earth- important information for testing the new quake outsidethe establishedseismic belts, be global tectonics is demonstratedin other sec- fit into the new global tectonics? tionsof this paper and elsewhere.To date,most Closely related to the distribution of seismic of the seismologicalwork related to this topic events in time and space and to tectonics is has beenso directed.The counteringimpact of the focal mechanismof the earthquake.Data the new globaltectonics on seismologymust also of qualityand quantityfrom modernobserving 5894 ISACKS, OLIVER, AND SYKES stationscan provide information for determina- a basic understandingof the earthquake phe- tion of focal mechanismsthat severelytest the nomenonis established.The new globaltectonics new hypothesesand are crucial to their devel- offers great promise for such an achievement. opment. It appears that with only modest It has already provided a theory that predicts advances in technique properly documented over-all strain ratesin tectonicareas throughout earthquakes will provide reliable detailed in- the world. It suggestsa means for predicting formationon tectonicactivity. Focalmechanism, the maximum size of an earthquakein a given stressdrop, slip, and orientationof principal region. It provides a framework in which to stresses should be available from individual relate measurementsof distortion, such as earthquakes,and singly or cumulatively these strains,tilts, and sea-levelchanges, with obser- data will be integratedwith the tectonicpattern. vations on the mechanismof the earthquake.It Implicit in the new global tectonicsis a new has establishedthe continuity of active zonesand attitude toward mobility of the earth's strata. has shownthat apparentlyinactive segments of Measurementsof fault creepand other forms of otherwiseseismic belts must, indeed,be active, earth strain over recent short intervals of time either by subsequentearthquakes or by creep. are basedon a variety of measuringtechniques Refinementsand further developmentsmust be that includegeodetic surveying, tide and strain anticipated. gaging, and measurementsof earthquake slip. Seisinologyhas long been the principalsource Various methodsof field geologygive data ex- of information on the structure of the earth's tending over varying but greater intervals of interior and is likely to continuein that role, time. When all these data are reduced to veloci- with or without the new global tectonics.The ties that describethe motion of one point in the new hypotheseswill, however,certainly stimu- earth relative to another point located in an late radicallynew approachesto the exploration adjacent relatively underformedblock, values of the earth'sinterior. A commonand powerful are obtainedthat are in the same range as the technique of seismologyinvolves the use of velocity valuesassociated with sea-floorspread- simplifiedearth modelsfor predictionof certain ing determinedthrough analysisof geomagnetic observedeffects. The new global tectonicscalls data. This rangeis from about 1 to perhaps10 for an entirely new kind of model. Layered cm/yr or more. These valuesare considerably modelsin which the shellsare sphericallysym- higher than the values usually assumedin the metric are now outdatedfor many areasof the past by most earth scientistsconsidering de- earth. Modelsbased on the effectof spreading formation of this type; hence, new attitudes and growth of the lithosphereat the rifts and toward old problemsmust be anticipated. A underthrustingat the island arcsmust be tested prime examplein seismology,cited above,is the against observation. The conventional division associationof the lengthalong the dip of a deep of the earth's surface into oceanic and conti- seismiczone with the amountof relatively re- nental areas, or oceanic,shield, and tectonic cent underthrustingin a region. areas,requires a new look whenthe lithosphere, With the new global tectonics,interrelation- with its lateral variations, is involved. New ships on a large, perhaps worldwide, scale can models in which the age and the thicknessof be predictedand perhapsobserved. Thus, major the lithosphere,as well as other properties,are seismicactivity in one area could be related to taken into accountare required. that in an adjoining,or perhapsdistant, area In the new tectonics,information on prop- associatedwith the same lithosphericunit or erties and parameters such as attenuation, units, for, althoughthe propagationtime for the strength, creep, viscosity,and temperatureis effect may be long, stressmay be transmitted critically needed. Further efforts must be made over large distancesthrough the lithosphere.Of to understandthe relationbetween such prop- specialinterest is the new insight into the sub- erties and the seismicproperties that are meas- ject of earthquakeprediction, even prevention. ured in a more straightforwardmanner. At the Although an empirical method of prediction island arcs, where cold exotic materials are could by chancebe found in the absenceof an descendinginto the mantle, nature is perform- understanding of the process, an effective ing the experimentof subjectingwhat are nor- method is much more likely to be achievedif mally near-surfacematerials to the highertern- SEISMOLOGY AND NEW GLOBAL TECTONICS 5895 peraturesand pressuresof the asthenosphere,an are shapingthe surficial features of interest to experimentin many respectsmuch like the ones classicalgeology. Even if it is destined for being performedin variouslaboratories. As our discard at some time in the future, the new techniques for measuring the compositionof global tectonicsis certain to have a healthy, the material and the dynamic and environ- stimulating,and unifying effecton all the earth mental parameters of this situation improve, sciences. this experimentshould yield importantinforma- Acknowledgments. We thank Professor L. A. tion on many topics. Of great interest to seis- Alsop and Drs. Neff Opdyke and Walter Pitman mologistsis the long-standingproblem of the for critically reading the manuscript. M. Bara- mechanismof deep earthquakes.The radiation zangi, J. Dorman, W. Elsasser, R. Hamilton, X. patterns of seismic waves from many events Le Pichon, P. Molnar, W. J. Morgan, and W. Stauder, S. J., furnished preprints of their papers presenta reasonablyconsistent pattern but one prior to publication. W. Ludwig of Lamont pro- that cannot readily be explained in detail by vided an original tracing of the seismic reflection existing hypotheses.The variation of seismic profile of the Japan trench that was used in the activity with depth is another sourceof infor- paper. We appreciate and acknowledge discus- mation. Are the focal mechanisms of earth- sions with many colleagues including J. Ewing, D. Griggs, D. McKenzie, and W. Pitman. Messrs. quakes at depth really as much like those in R. Laverty and J. Brock assistedin preparing the near-surface brittle materials as they seem? data for computation, and Mrs. Judith Healey What is the role of water, of other interstitial assistedin editing and preparing the manuscript. fluid, of partial melting? Are phasechanges in Computing facilities were made available by the the exotic mantle materials important, perhaps NASA Goddard SpaceFlight Center, Institute for SpaceStudies, New York. This study was partially not as sources of seismic waves but as con- supported under Department of Commerce Grant centrators of stress that leads ultimately to E-22-100-68G from the Environmental Science rupture? ttow are the seismicobservations of Services Administration, through the National focal mechanism,spatial and temporal distribu- Science Foundation grant NSF GA-827, and through the Air Force Cambridge ResearchLab- tion, energy,etc., related to the material of the oratories, Office of Aerospace Research, under mantle and what can we learn about that mate- contract AF19(628)-4082 as part of the Advanced rial? ttow are these data related to the general ResearchProjects Agency's Project Vela-uniform. configuration of the seismic zone and of the [REFERENCES geology of the island arc? These are some of the topics on which this experimentmay pro- Adams,R. D., Sourcecharacteristics of somedeep vide information. New Zealand earthquakes, New Zealand J. Geol. Geophys.,6, 209, 1953. Surely, the most striking and perhaps the Aki, K., Earthquakegenerating stress in Japan most significanteffect of the new global tec- for the years 1961to 1963obtained by smooth- tonics on seismologywill be an accentuated ing the first motion radiation patterns, Bull. interplay between seismologyand the many EarthquakeRes. Inst. Tokyo Univ., 44, 447, 1966. other disciplinesof geology. The various dis- Anderson,D. L., Recent evidence concerningthe ciplines which have tended to go their sepa- structure and compositionof the earth's mantle, rate ways will find the attraction of the unifying Phys. Chem. Earth, 6, 1, 1966. conceptsirresistible, and large numbers of re- Anderson,D. L., Phase changesin the upper freshing and revealing interdisciplinarystudies mantle, Science,157, 1165, 1957a. Anderson, D. L., Latest information from seismic may be anticipated. For example, the geomor- observations,in The Earth's Mantle, edited by pho]ogy of an area of raised beachestakes on T. F. Gaskell, p. 355, Academic Press, New new light for thoseinterested in paleoseismicity; York, 1967b. the tectonic significance of a feature of the Anderson, D. L., and C. B. Archambeau, The anelasticityof the earth, J. Geophys.Res., 69, oceanfloor is determinedby its seismicityand 2071, 1964. by the mechanismof the earthquakes; the Badgley, P. C., Structural and Tectonic Principles, petrology of a volcano of an island arc is re- 521 pp., Harper and Row, New York, 1965. lated in a meaningfulway to the seismicactivity Banghar, A., and L. R. Sykes, Focal mechanism below; the worldwidephenomena of seismology of earthquakes in the Indian Ocean and adjacent areas, J. Geophys.Res., 73, in press, provide crucial evidenceon the basicprocesses 1968. of the earth's interior that have shaped and Barazangi, M., and J. Dorman, World seismicity 5896 ISACKS, OLIVER, AND SYKES map of ESSA Coast and Geodetic Survey epi- Hawaii, in The Crust o/the PacificBasin, Geo- center data for 1961-1967, Bull. Seismol. Soc. phys.Monograph 6, editedby G. A. Macdonald Am., 53, in press, 1968. and H. Kuno, p. 13, American Geophysical Benloft, H., Orogenesisand deep crustal structure: Union, Washington, D.C., 1962. Additional evidencefrom seismology,Bull. Geol. Eaton, J.P., Instrumental seismicstudies, in the Soc. Am., 65, 385, 1954. Parkfield-Cholame, California, earthquakes of Bodvarsson,G., and G. P. L. Walker, Crustal drift June-August 1966--surface geologic effects, in Iceland, Geophys.J., 8, 285, 1964. water-resourcesaspects, and preliminary seismic Bott, M. H. P., Terrestrial heat flow and the data, U.S. Geol. Surv. Pro•. Paper 579, p. 57, mantle convection hypothesis, Geophys. J., 14, 1967. 413, 1967. Elsasser, W. M., Convection and stress propaga- Brune, J. N., Seismicmoment, seismicity, and rate tion in the upper mantle, Princeton Univ. Tech. of slip along major fault zones, J. Geophys. Rept. 5, June 15, 1967. Res., 73, 777, 1968. Ewing, J., and M. Ewing, Sediment distribution Brune, J., and J. Dorman, Seismic waves and on the mid-oceanridges with respectto spread- earth structure in the Canadian shield, Bull. ing of the sea floor, Science,156, 1590, 1967. Seismol.Soc. Am., 53, 167, 1963. Ewing, M., Sediments of ocean basins, in Man, Bullard, E. C., Continental drift, Quart. J. Geol. Science, Learning, and Education, p. 41, Rice Soc. London, 120, 1, 1964. University, Houston, Texas, 1963. Bullard, E., J. E. Everett, and A. G. Smith, The Ewing, M., J. Ewing, and M. Taiwan/, Sediment fit of the continentsaround the Atlantic, Phil. distribution in the oceans: The mid-Atlantic Trans. Roy. Soc. London, A, 253, 41, 1965. ridge, Bull. Geol. Soc. Am., 75, 17, 1964. Burckle, L. H., J. Ewing, T. Saito, and R. Leyden, Ewing, M., and B. Iteezen, Some problems of Tertiary sediment from the East Pacific rise, antarctic submarine geology, in Antarctica in Science,157, 537, 1967. the International Geophysical Year, Geophys. Byerly, P., and J. DeNoyer, Energy in earth- Monograph 1, edited by A. Craw et al., p. 75, quakes as computed from geodetic observations, American Geophysical Union, Washington, in Contributions in Geophysics in Honor o• D. C., 1956. Beno Gutenberg,edited by H. Benloft et al., p. Fedotov, S. A., A.M. Bagdasarova,I. P. Kuzin, 17, Pergamon Press, New York, 1958. R. Z. Tarakanov, and D. Yu. Shmidt., Seismicity Carder, D. S., D. Tocher, C. Bufe, S. W. Stewart, and deep structure of the southern part of the J. Eisler, and E. Berg, Seismic wave arrivals I•urile Island arc, Dokl. Akad. Nauk SSR, from Longshot, 0 ø to 27ø, Bull. Seismol. Soc. 153(3), 668, 1963 (English translation, p. 71). Am., 57, 573, 1967. Fedotov, S. A., I. I. Kuzin, and M. F. Bobkob, Chinnew, M. A., The deformation of the ground Detailed seismological investigation of Kam- around surface faults, Bull. Seismol. Soc. Am., chatka in 1961-1962,Izv. Akad. Nauk SSSR, 51, 355, 1961. Set. Geofiz., 9, 1360, 1964. Cleary, J., Azimuthal variation of the Longshot Griggs, D. T., Reflections on the earthquake source term, Earth Planetary Sci. Letters, 3, 27, mechanism, in Proceedings o[ the Second 1967. United States-Japan Conference on Research Cleary, J., and A. L. Hales, An analysis of the Related to Earthquake Prediction Problems, travel times of P waves to North American sta- edited by R. Page, p. 63, National Science tions, in the distance range 32ø-100ø, Bull. Foundation, Washington,D.C. (also Japan'So- Seismol. Soc. Am., 56, 467, 1966. ciety for Promotion of Science,Tokyo), 1966. Coats, R. R., Magma type and crustal structure Griggs,D. T., and D. W. Baker, The origin of in the Aleutian arc, in The Crust of the Pacific deep-focusearthquakes, in Propertieso] Matter, Basin, Geophys. Monograph 6, edited by G. A. p. 11, John Wiley, New York, in press,1968. Macdonald and H. Kuno, p. 92, American Geo- Gunn, R., A quantitative study of mountain physical Union, Washington, D.C., 1962. building on an unsymmetricalearth, J. Frank- Daly, R. A., Strength and Structure of the Earth, lin Inst., 224, 19, 1937. 434 pp., Prentice-Hall, Englewood Cliffs, N.J., Gunn, R., Quantitative aspects of juxtaposed 1940. ocean deeps, mountain chains, and volcanic Dietz, R. S., Continent and ocean basin evolution ranges,Geophysics, 12, 238, 1947. by spreading of the sea floor, Nature, 190, 854, Gutenberg, B., Tsunamis and earthquakes,Bull. 1961. Seismol. Soc. Ann., 29, 517, 1939. Dorman, J., M. Ewing, and $. Oliver, Study of Gutenberg, B., Physics o] the Earth's Interior, shear velocity distribution by mantle Rayleigh 240 pp., AcademicPress, New York, 1959. waves, Bull. Seismol. Soc. Am., 50, 87, 1960. Gutenberg,B., and C. F. Richter, Seismicityo] Dymond, J., and K. Defteyes, I(-Ar ages of deep- the Earth, 2nd ed., Princeton University Press, sea rocks and their relation to sea floor spread- Princeton, N.J., 1954. ing (abstract), Trans. Am. Geophys. Union, 49, Hamilton, R. M., and A. W. Gale, Seismicity 364, 1968. and structure of the North Island, New Zealand, Eaton, J.P., Crustal structure and volcanism in paper presented at 64th Annual Meeting of SEISMOLOGY AND NEW GLOBAL TECTONICS 5897 SeismologicalSociety of America, Tucson, Ari- and upper-mantle structure from long period S zona, April 1968. waves, Bull. Seismol. Soc. Am., 57, 1063, 1967. Hamilton, R. M., and F. F. Evison, Earthquakes Ichikawa, M., Mechanism of earthquakes in and at intermediate depths in southwest New Zea- near Japan, 1950-1962,Papers Meteorol. Geo- land, New Zealand Y. Geol. Geophys., 10, 1319, phys. Tokyo, 16, 201, 1966. 1968. Isacks, B. L., and L. R. Sykes, Focal mechanisms Hamilton, W., Origin of the Gulf of California, of deep and shallow earthquakes in the Tonga Bull. Geol. Soc. Am., 72, 1307, 1961. and Kermadec island arcs, in preparation, 1968. Hamilton, W., and W. B. Meyers, Cenozoic Johnson, L. R., Array measurementsof P veloci- tectonics of the western United States, Rev. ties in the upper mantle, J. Geophys. Res., 72, Geophys., 4, 509, 1966. 6309, 1967. ttarding, S. T., and W. Rinehart, Preliminary Kasahara, K., The nature of seismic origins as seismologicalreport, in The Parkfield, Cali- inferred from seismologicaland geodetic obser- fornia Earthquake of June 27, 1966, p. 1, U.S. vations, 1, Bull. Earthquake Res. Inst. Tokyo Government Printing Office, Washington,D.C., Univ., 35, 473, 1957. 1966. Katsumata, M., Seismic activities in and near Hayes,I). E., A geophysicalinvestigation of the Japan, 3' Seismic activities versus depth (in Peru-Chile trench, Marine Geol., 6, 309, 1966. Japanese), J. Seismol. Soc. Japan, 20, 75, 1967. Iteck, N. YI., List of seismicsea waves,Bull. Langseth, M. G., X. Le Pichon, and M. Ewing, Seismol.Soc. Am., 37, 269, 1947. Crustal structure of the mid-oceanic ridges, 5, Heirtzler, J. R., G. 0. Dickson,E. M. Herron, Heat flow through the Atlantic Ocean floor and W. C. Pitman, III, and X. Le Pichon, Marine convection currents, J. Geophys. Res., 71, 5321, magneticanomalies, geomagnetic field reversals, 1966. and motions of the ocean floor and continents, Laughton, A. S., The Gulf of Aden, Phil. Trans. J. Geophys. Res., 73, 2119, 1968. Roy. Soc. London, A, 259, 150, 1966. Herrin, E., Travel-time anomaliesand structure Lehmann, I., On the travel times of P as deter- of the upper mantle (abstract), Trans. Am. mined from nuclear explosions, Bull. Seismol. Geophys.Union, 47, 44, 1966. Soc. Am., 54, 123, 1964a. Herrin, E., and J. Taggart,Epicenter determina- Lehmann, I., On the velocity of P in the upper tions for Longshot(abstract), Trans. Am. Geo- mantle, Bull. Seismol.Soc. Am., 54, 1097, 1964b. Le Pichon, X., Sea-floor spreadingand continental phys. Union, 47, 164, 1966. drift, J. Geophys. Res., 73, 3661, 1968. Hess, H. It., Gravity anomaliesand island arc Le Pichon, X., and M. G. Langseth, Heat flow structurewith particular referenceto the West from the mid-oceanic ridges and sea-floor Indies, Proc. Am. Phil. Soc.,79, 71, 1938. spreading,J. Geophys.Res., 73, in press,1968. Itess, H. It., History of the oceanbasins, in Ludwig, W. J., J. I. Ewing, M. Ewing, S. Mu- PetrologicalStudies' A Volume in Honor of rauchi, N. Den, S. Asano, H. Hotta, M. Haya- A. F. Buddington,edited by A. E. J. Engel kawa, T. Asanuma,K. Ichikawa, and I. Noguchi, et al., p. 599, GeologicalSociety of America, Sediments and structure of the Japan trench, New York, 1962. J. Geophys.Res., 71, 2121, 1966. Hirasawa,T., Sourcemechanism of the Niigata MacDonald, G. J. F., The deep structure of con- earthquakeof June 16, 1964,as derived from tinents, Science,143, 921, 1964. body waves,J. Phys. Earth, 13, 35, 1965. MacDonald, G. J. F., Mantle properties and con- Hirasawa,T., A least-squaremethod for the focal tinental drift, in Continental Drift, Spec. Publ. mechanism determinations from S wave data, 2, 9, edited by G. D. Garland, p. 18, Royal Society Bull. Earthquake Res. Inst. Tokyo Univ., 44, of Canada, Ottawa, 1966. 919, 1966. Matsuda, T., K. Nakamura, and A. Sugimura, Hodgson,J. H., Nature of faultingin large earth- Late Cenozoic orogeny in Japan, Tectonophys- quakes,Bull. Geol.Soc. Am., 68,611, 1957. ics, 4, 349, 1967. Hodgson,J. H., and J. D. Cock, Direction of Matthews, D. M., and J. Bath, Formation of faulting in the deep focusSpanish earthquake magnetic anomaly pattern of mid-oceanic ridge, of March 29, 1954, Tellus, 8, 321, 1956. Geophys. J., 13, 349, 1967. Hodgson,J. H., and A. E. Stevens,Seismicity and McConnell, R. K., Isostatic adjustment in a lay- earthquakemechanism, Res. Geophys.,2, 27, ered earth, Y. Geophys.Res., 70, 5171, 1965. 1964. McEvilly, T. V., W. H. Bakun, and K. B. Casa- Holmes, A., Principles o] Physical Geology, day, The Parkfield, California, earthquakes of Ronald Press, New York, 1965. 1966, Bull. Seismol.Soc. Am., 57, 1221, 1967. Honda, H., A. Masatsuka,and K. Emura, On the McKenzie, D. P., Some remarks on heat flow and mechanismsof the earthquakes and the stresses gravity anomalies, J. Geophys. Res., 72, 6261, producingthem in Japan and its vicinity, 2, 1967. Sci. Rept. Tohoku Univ., Set. 5, Geophys.,8, McKenzie, D., and R. L. Parker, The North 186, 1956. Pacific: An example of tectonics on a sphere, Ibrahim, A., and O. W. Nuttli, Travel-timecurves Nature, 216, 1276, 1967. 5898 ISACKS, OLIVER, AND SYKES Menard, H. W., Sea floor relief and mantle con- mantlesilicate minerals, Geophys. J., 14, 45, vection, Phys. Chem. Earth, 6, 315, 1965. 1967. Menard, H. W., Fracture zones and offsetsof the Raleigh,C. B., andM. S. Paterson,Experimental East Pacific rise, J. Geophys.Res., 71, 682, 1966. deformationof serpentiniteand its tectonic Menard, I-I. W., Sea-floor spreading, topography, implications,J. Geophys. Res., 70, 3965, 1965. and the secondlayer, Science,157, 923, 1967. Richter,C. F., ElementarySeismology, 768 pp., Minikami, T., Fundamental researchfor predict- W. I•. Freemanand Co., San Francisco, 1958. ing volcanic eruptions, 1, Earthquakes and Ritsema,A. R., Somereliable fault planesolu- crustal deformations originating from volcanic tions,Pure Appl. Geophys. 59, 58, 1964. activities, Bull. Earthquake Res. Inst. Tokyo Ritsema,A. R.,The mechanism of some deep and Univ., 38, 497, 1960. intermediateearthquakes in the region of Japan, Mitronovas, W., L. Seebet, and B. Isacks, Earth- Bull. EarthquakeRes. Inst. TokyoUniv., 45, quake distribution and seismic wave propaga- 39, 1965. tion in the upper 200 km of the Tonga island Savage,J. C., and L. M. Hastie,Surface deforma- arc (abstract), Trans. Am. Geophys. Union, 49, tion associatedwith dip-slipfaulting, J. Geo- 293, 1968. phys.Res., 71, 4897,1966. Molnar, P., and J. Oliver, Lateral variation of Scholl,E. W., R. Von Huene,and J. B. Ridlon, attenuation in the upper mantle and discon- Spreading of the ocean floor: Undeformed tinuities in the lithosphere, in preparation, 1968. sedimentsin the Peru-Chiletrench, Science, Morgan, W. J., Rises, trenches, great faults, and 159, 869, 1968. crustal blocks, J. Geophys.Res., 73, 1959, 1968. Sclater,J. G., V. Vacquier,J.P. Greenhouse,and Murphy, L. M., Worldwide seismic network, in F. S. Dixon,Melanesian subcontinent presenta- ESSA Symposium on Earthquake Prediction, tion and discussionof recentheat flow measure- February 1966, p. 53, U.S. Government Printing ments(abstract), Trans. Am. Geophys.Union, Office, Washington, D.C., 1966. 49, 217, 1968. Neumann, F., Crustal structure in the Puget Stauder,W., S wavestudies of earthquakesof the Sound area, IUGG Assoc.Seismol. Set. A., Tray. North Pacific, 1, Kamchatka, Bull. Seismol. Sci., 20, 153, 1959. $oc. Am., 52, 527, 1962. Officer, C. B., J. Ewing, J. YIennion, I). Yiarkrider, Stauder,W., A comparisonof multiplesolutions of and I). Miller, Geophysical investigations in the focalmechanisms, Bull. Seismol.Soc. Am., 54, eastern Caribbean: Summary of 1955 and 1956 927, 1964. cruises,Phys. Chem. Earth, 3, 17, 1959. Stauder,W., Earthquakemechanisms, Trans. Am. Oliver, J., and B. Isacks, Deep earthquake zones, Geophys.Union, 48, 395,1967. anomalous structures in the upper mantle, and Stauder,W., Mechanismof the Rat Island earth- the lithosphere,J. Geophys.Res., 72, 4259, 1967. quakesequence of February4, 1965,with rela- Oliver, J., and B. Isacks, Structure and mobility tion to islandarcs and sea-floorspreading, of the crust and mantle in the vicinity of island J. Geophys.Res., 73, 3847,1968. arcs, Can. J. Earth Sci., 5, in press, 1968. Stauder,W., S. J., and G. A. Bollinger,The S- Orowan, E., Age of the ocean floor, Science, 154, wave project for focal mechanismstudies, 413, 1966. earthquakesof 1962,Bull. Seismol.Soc. Am., Parkin, E. J., Alaskan Surveys to Determine 54, 2198, 1964. Crustal Movement, Part 2, Horizontal Displace- Stauder,W., S. J., and G. A. Bollinger,The S- ment, 15 pp., U.S. Department of Commerce, wave project for focal mechanismstudies, Washington, I). C., 1966. earthquakesof 1963,Bull. Seismol.Soc. Am., Pekeris, C. L., The internal constitution of the 56, 1363, 1966a. earth, Geophys.J., 11, 85, 1966. Stauder,W., S. J., and G. A. Bollinger,The focal Pitman, W., and J. YIeirtzler, Magnetic anomalies mechanismof the Alaskaearthquake of March over the Pacific-Antarcticridge, Science,154, 28, 1964,and of its aftershocksequence, J. Geo- 1164, 1966. phys.Res., 71, 5283,1966b. Plafker, G. P., Tectonic deformation associated Stauder, W., and A. Udias, S-wave studies of with the 1964Alaska earthquake, Science, 143, earthquakesof the North Pacific,2, Aleutian 1675, 1965. Islands,Bull. Seismol.Soc. Am., 53, 59, 1963. Sugimura,A., and S. Uyeda,A possibleanisot- Plafker, G. P., and M. Rubin, Vertical tectonic ropyin the uppermantle accounting for deep displacementsin south centralAlaska during earthquakefaulting, Tectonophysics, 5, 25, 1967. and prior to the great 1964earthquake, J. Geo- Summerhayes,C. P., New Zealandregion vol- sci., OsakaUniv., 10, 53, 1967. canismand structure, Nature, 215, 610, 1967. Press,F., and W. F. Brace,Earthquake predic- Sykes,L. R., The seismicityand deepstructure tion, Science,152, 1575,1966. of islandarcs, J. Geophys.Res., 71, 2981, 1966. Raitt, R. W., R. L. Fisher,and R. G. Mason, Sykes,L. R., Mechanismof earthquakesand Tongatrench, Geol. Soc. Am. Spec.Paper, 62, natureof faultingon the mid-oceanicridges, 237, 1955. J. Geophys.Res., 72, 2131,1967. Raleigh,C. B., Plasticdeformation of upper Sykes,L. R., Seismologicalevidence for trans- SEISMOLOGY AND NEW GLOBAL TECTONICS 5899

form faults, sea-floor spreading,and continental ridge between 22ø and 23ø north latitude and drift, in Proceedingso/NASA Symposium, His- the tectonics of mid-ocean rises, J. Geophys. tory o] the Earth's Crust, edited by R. A. Res., 73, 1279, 1968. Phinncy, Princetou University Press, Princeton, Vening Meinesz, F. A., Maritime gravity surveys in press, 1968. in the Netherland East Indies, tentative inter- Sykes, L. R., B. Isacks, and J. Oliver, Earthquake pretation of the results, Koninkl. Ned. Akad. swarms, volcanism, and sea-floor spreading, in Wetenschap.Proc., B, 33, 566, 1930. preparation, 1968. Vening Meinesz, F. A., Indonesian Archipelago: Talwani, M., and D. E. Hayes, Continuous gravity A geophysicalstudy, Bull. Geol. Soc. Am., 65, profiles over island arcs and deep sea trenches 143, 1954. (abstract), Trans. Am. Geophys. Union, 48, 217, Verhoogen, J., Temperatures within the earth, 1967. Phys. Chem. Earth, 1, 17, 1956. Talwani, M., X. Le Pichon, and M. Ewing, Vine, F. J., Spreading of the ocean floor: New Crustal structure of the mid-oceanic ridges, 2, evidence, Science,154, 1405, 1966. Computed model from gravity and seismic re- Vine, F. J., and P.M. Matthews, Magnetic fraction data, J. Geophys. Res., 70, 341, 1965a. anomalies over ocean ridges, Nature, 199, 947, Talwani, M., X. Le Pichon, and J. R. Heirtzler, 1963. East Pacific rise: The magnetic pattern and the Vine, F. J., and W. J. Morgan, Simulation of mid- fracture zones,Science, 150, 1109, 1965b. ocean ridge magnetic anomalies using a random Talwani, M., G. YI. Sutton, and J. L. Worzel, injection model (abstract), Program 1967 Ann. Crustal section across the Puerto Rico trench, Meeting, Geol. Soc. Am., November 1967. J. Geophys. Res., 64, 1545, 1959. Vine, F. J., and J. T. Wilson, Magnetic anoma- Teng, T., Attenuation of body waves and the Q lies over a young oceanic ridge off Vancouver structure of the mantle, J. Geophys. Res., 73, Island, Science,150, 485, 1965. 2195, 1968. Wadaft, I(., T. YIirono, and T. Yumura, The ab- Tobin, D., and L. R. Sykes, Seismicity and tec- sorption of transverse waves in the upper tonics of the northeast Pacific Ocean, J. Geo- mantle under Japan (abstract), Gem Assembly phys. Res., 73, 3821, 1968. IUGG Abstr. Papers,2, IASPEI-154, 1967. ToksSz, M. N., and D. L. Anderson,Phase veloci- Wegener, A., The Origin of Continents and ties of long-period surface waves and structure Oceans,4th ed., 246 pp., Dover, New York, of the upper mantle, 1, Great-circle Love and 1966. R•yleigh wave data, J. Geophys.Res., 71, 1649, Wellman, Y[. W., New Zealand Quaternary tec- 1966. tonics, Geol. Rundschau,43, 257, 1955. Udias, A., and W. Stauder, Application of nu- Whirten, C. A., Measurements of earth move- merical methods for S-wave focal mechanism mentsin California,Cali]. Div. Mines Bull., 171, determination to earthquakes of I•amchatka- 75, 1955. Kurile region, Bull. Seismol.Soc. Am., 54, 2049, Whitten, C. A., Crustal movement in California 1964. and Nevada, Trans. Am. Geophys. Union, 37, If. S. Coast and Geodetic Survey, United States 393, 1956. Earlhquakes(annual publications) ESSA, Rock- Wilson, J. T., Evidence from islands on the ville. Maryland, 1935-1965. spreadingof oceanfloors, Nature, 197,536, 1963. Utsu, T., Anomalies in seismicwave velocity and Wilson, J. T., A new class of faults and their attenuation associatedwith a deep earthquake bearing on continentaldrift, Nature, 207, 343, zone 1, J. Fac. Sci. Hokkaido Univ. Japan, Set. 1965a. 7, Geophys.,3, 1, 1967. Wilson, J. T., Transform faults, oceanic ridges, Uycda, S., and K. Ylorai, Terrestrial heat flow in and magnetic anomalies southwest of Vancouver Japan, J. Geophys. Res., 69, 2121, 1964. Island, Science,150, 482, 1965b. Vacquier, V., Magnetic evidence for horizontal Worzel, J. L., Deep structure of coastal margins displacementsin the floor of the Pacific Ocean, and mid-oceanic ridges, Colstoa Papers, 17, in Continental Drift, edited by S. K. Runcorn, 1965. p. 135, Academic Press. New York, 1962. van Andel, T. H., and C. O. Bowin, Mid-Atlantic (Received May 21, 1968.)