been about 700 000 years since the time the Earth's magnetic field “ ‘more than just an inconvenience to gators. In the November issue of Research Letters,. they an intriguing, if rather spocule- ease for the thesis that many, 8 all, polarity reversals of the churning in our planets liq: uldiron outer core are caused by the ct of large extraterrestrial objects to Barth. oid oF cometary nucleus hitting Tofts enough dust to set off thing like a “nuclear winter.” cold persists long after the dust auise of the increased reflec: ‘the snow-covered continents, nurse ofa few centuries, enough ial ocean water is transported lat ice caps to drop the sea 10 meters and thus reduce ent of inertia of the solid outer ‘the Earth (crust and mantle) in a million, doesn’t sound like much," us, “But when we realized ranslatos into a full radian of en mantle and core in f Inertia of the crust and \ddenly” decreased, the ar- they begin spinning fas- ‘solidiron inner core al the hearth, The 2300-km-thick tiquid outer core that separates e from the inner core thus ‘velocity shear, which in the bout & thousand years de- he pattern of convective flows randomized flow pattern to being twists and tan- ‘caReOWNTROGEN Search &: CiseomEEY asteroid impacts trigger geomagnetic reversals? we ED HEIGHT ABOVE SEEIVENT BAS in) Organic carbon/nitrogen rato (upper curve) and carbon-13 isotopic depletion (ower ‘curve, as functions of depth (age) i ne exposed Katewa lke sadmentn Kachmi, show Sharp peaks romarkably eolnedemt wih tae gacmagnetic reversals ast 2 mion Years The peaks are beleved toindieato spo! unusual cols. Ears at op inleate Tmognetesiatgraphy ofthe sediment black (wnt) for peniods of norm reversed) ‘Jeomagnatic polariy. Dates give estimated ages of reversal boundaries ard tog of sample in rrions of yous gles up the pre-existing maznetie eld fines, rapidly destroying the dipole field. Thus, until a new stable dynamo configuration establishes itself, per- hhaps 10 000 years later, the geomagnet- ic field is dominated by short-range hhigher-multipole components, so that ‘one sees very little field at the Earth's Surface, To the extent that the new dynamo has little or no memory of the old, its chance of having the same polarity as its predecessor is roughly 5050, One would thus expect, on average, one reversal for every’ two ‘major extraterrestrial impact events. Even when there is no ultimate rever- ‘al, the impact will have produced an “excursion,” a 10 000-year hiatus with fa much reduced surface geomagnetic cs to test his hypothesis against some of the competing models of spontaneous geomagnetic reversal ‘in 1905, Albert Binstein ranked the problem of geomagnetism as one of the five most important unsolved problems in physics. ‘The problem is still “un- solved” today, in the sense that there is no detailed consensus as to how the Earth's dynamo works. Theories that attribute reversal to a bistability aris- ing from the simultaneous action of two competing effective dynamos in the liquid-core flow, for example, prediot sudden, spontaneous “snap” trans tions from one polarity to the other. ‘They would preclude the long, multi. pole-dominated hiatus anticipated by PHYSICS TODAY / FEBRUARY 1987 17Muller and Morris. It has recently ‘become fashionable to think about planetary dynames as chaotic fluid systems. Though one might expect that the spontaneous reversals charac- teristie of chaotic systems would ‘ways he sudden, dynamo theorist Wi Tem Malius (MIT) told us that chaotic models can exhibit extended pauses during reversal ‘What doos the geological record say? Sedimentary layers and once molten rocks record the polarity ofthe Barth's field at the time they were laid down or solidified, by the orientation of their ferromagnetic components. JP. Valet and coworkers at the Centre des Fai bles Radicactivités in Gif sur Yvetie, near Paris, have recently published sea floor core analysis® displaying. @ {geomagnetic reversal with extraordin- Ary temporal resolution. Cores were analyzed every few centimeters in region where the sedimentation rato ‘was roughly 10 cm every thousand ‘years. One sees clearly that the rever: ‘sl is preceded by a sudden drop in the Gipole field intensity in Toss than 2000 years, followed by a delay of about 20000 years before the reversal finally comes. Both the rapid decay and subst. quent hiatus, Muller and Morris writ, “are predicted by our model, and incon” sistent with [many] previous reversal models.” Tektites, those small glassy objocts of varied shape and mysterious proven- ‘ance, were prized as ornaments by our CroMagnon ancestors, Tektites have primarily been found littering four ‘extensive “strewnfields"—both on land fand in sea floor sediment. In recent ‘years it has finally become the consen- ‘sus among geologists that tektites have their origin in the melting and splatter- ing of surface material when large extraterrestrial objects fall to Earth with great impact. Large impact ‘craters have indeed been associated with three of the four known tektite Strewnfields For the last 20 years, geologist Bill Glass and his colleagues at the Univer- sity of Delaware have been document- ing what appeared to be an extraordin- ary train of coincidences: Each of the three “toktite events” associated with craters ceoms to have occurred at just about the time of a geomagnetic rever- sal. Examining dozens of soa floor cores from the vast Australasian strewnfield, for example, Glass und company found that the microtektite abundance as a function of stratum epth in each core always peaked unambiguously within 20 cm of the magnetic boundary marking the last reversal, 700000 years ago. (See the figure on page 19) ‘That is to say, the ‘Australasian tektite event (for which 18 PHYSICS TODAY / FEBRUARY 1967 there are several candidate eraters on the Asian mainland) must have oc curred within a few thousand years (either way) of the latest geomagnetic reversal. The third-most recent revor- sal—some 950000 years ago—is simi- larly seen to be coincident with the Ivory Coast strewnfeld, which is asso- ciated with the Bosumptwi crater in Ghana. Before Muller and Morris came along, we had no plausible expla- nation for Glass's provecative observa tion that such striking coincidences hold for three of the four known tektite strewnfields. ‘The 24-km-diameter Ries crater in southern Germany, associated with the Cucchoslovak tektite strewnfield, pro- Vides the most striking temporal coinci- dence ofall, pointed out ten years ago by Jean Pohi at the Technical Universi- ty of Munich. ‘The magnetization of the ies erater’s fallback breccias, partial- ly melted by the impact, tells us the polarity of the geomagnetic fold at the moment of impact, 14.8 million years ‘ago. The first sediments in the newly {ormed crater, however, indicate the opposite polarity. In other words, there ‘mist have been a reversal within the few thousand years between the impact, forming the eater and the laying down, of the first finegrained sediment. The ‘odds are 100 to 1 against this being a random coincidence. Climate, If the Muller-Morris expla- nation of these coincidences is correct, fone should also find evidence for the abrupt climatic change that, in their scenario, couples the extraterrestrial impact to the subsequent disruption of the Barth’s magnetic field. Very strike ing evidence of this kind was reported a few months ago by RV. Krishna. ‘murthy and his coworkers at the Ah- medabd Physical Research Leborato- ry in India® The Karewa plateau in the Vale of Kashmir was once a vast lake. ‘The geomagnetic reversals of the last few million years are precisely documented in the magnotostrati raphy ofits deep sediment, now conve. hiently exposed. Examining the carbon-nitrogen ra- tio of the organic component of this same sediment, Krishnamurthy and company found striking, narrow peaks rising an order of magnitude above background at precisely the times (epths) of three geomagnetic rever- sala, including the most recent. The nitrogen concentration is taken to be a messure of the protein-rich plankton ‘and algae population ofthe lake water, so that C/N hecomes something of a paleothermometer, peaking when vni- sual cold reduces this population. (See the figure on page 17) The C/N findings are corroborated in the Karewa sediment by the coincident, if tess spectacular, peaking of the ratio C/G, anther paleoth mneter fn the calculation of Muller an Morris, a seslovel drop of 10 meters in ‘few centuries = sufient to tig ‘Some 300 geomagnetic reversals hay Deen recorded for the last. 170 mili years, with a conspicuous dearth int interval from 110 millon to 75 mil years before the present. For this 170-million-year period, about abrupt sea-level drops of 16-200 mi hhave been documented. (See the fi ‘on page 20) This is thought tobe o ‘small fraction of the steep drops actually ocurred, and they have ‘yet been timed with sufficient preci io say how individual drops rela individual reversals. But when same dearth of occurrences in period around 100 million years that characterizes the dsebut During that period the seas significantly warmer than they wer ‘more recent epochs. All this seem to be abrupt enough for ‘purposes of Muller and Morris, ears ago, Christopher Doake o British Antarctic Survey did, suggest that changes in the sspin rate due to the onset or ‘on of ice ages might trizger Buthe did not propose a mechani such reversals, ‘The dynamo. ‘There being no sus on how the Earth's dynamo in detail, Muller and Morris base their argument on quite considerations, not depending validity of any specific geom ‘model. A dynamo is a system of c ductors moving in a magnetic field such a way that the current tht induced provides @ positive fe sufficient to cause an initially field to grow exponentially —up equilibrium limit. ‘Thisis easi plished if one allows topolo ‘complex asa homopolar Faraday mo. But the problem of p dynamo theory is that one mt do with a much simplerDEPTH BELOW SERFLGOR im) boundary, 1200 kim from the center of ‘the Barth, to the solid-mantle bound- Coriolis foreo and the magnetic field ‘eel, The challenge to the theorists is ‘ sabe this highly nonlinear flow blem. Computer simulation nowa- ys seems to offer the greatest prom: . Dynamo action requires that the etic analog of the Reynolds num- r excbed some eritical value. Be- this dimensionless parameter is tional to the product of the flow velocity and linear dimen: n ofthe system, a planetary dynamo ellnigh impossible to simulate in a Morris point out, id hit the arth with a kinetic rey of 10 ergs—enough, they con- to initiate climatic sequence that would lower the sea level by 10 8 and thus spin up the crust and tle by a part per million in a time {or iostaticadjustment of the went of inertia of the Earth as a © The inner boundary of the ntle at the equator would speed up bout 0.03 em/see. _The diurnal eireuiating angular city of the liquid will increase with ee from the inner-care boundary it matches that of the mantle Jong time the spin of the small nner core will continue to lag nical viscosity of the hot abundanes (red cuvos) asa function of depth (age) in tn soa-foor cores (wth “iferent,varable sedimentation rates) from the Australasian svownflaVerteal bars show estatlrapry for cach core: black (white) or normal (reversed) goomagnet poly, kts abundance for each core shows a srking peak ver close to the tme of the last get reversal, 0.7 millon years ago, suggesting halt reversals elated tothe red the microtekites. Bar at ek shows geomagnetic line connecting the 0 7-illonyear-old reversal ‘ouncary onthe diferent cores. (Data of 8. lags of al) liquid metal is far too small to sustain such a velocity shear field across a radial distance of several thousand that couples the liquid metal to the existing field lines, the bulk of the liquid) would largely ignore the speedup of the mantle above it. If, on the other hand, the electromag- netic coupling hetween care and mai tle is too strong, the spun-up mantle ‘would transfer angular momentum to the liquid core so quickly that a shear velocity field would disappear in short order. Extrapolating the surface Held to the core-mantle boundary, Muller and Morris argue that the electromag. netic coupling is neither too strong 10f too weak to maintain the shear velocity field their scenario requires. Dynamo theorist Friedrich Busse (UCLA and University of Bayreuth), however, points to the millisecond variations of the length of day we see on a scale of decades. These observations suggest to ‘Busse that the core-mantle coupling is probably stronger than Muller and ‘Morris would like “On the other hhand,” Busse told us, “their scenario is rot out of the question, It's the most plausible proposal anyone has come up ‘with—but it's marginal Tn pre-impact equilibrium, the inter. faction of radial convection transport with the Coriolis and Lorentz forces will have established a set of convective fiow cells, For a simplified heuristic picture, one might invoke Busse's re- cent magnetohydrodynamic computer simulations. Looking down at the ‘equatorial cross section of the Earth, fone would see perhaps half a dozen convection vortices arrayed around the liquid-core annulus, carrying lower density liquid away from the inner core and denser liquid back down. A round trip takes perhass a thousand years. ‘The buoyancy may be of thermal or gin, or it may be due to phase segrea- ton as liquid solidifies at the innercore boundary. Adjacent vortical cells will have opposite senses of rotation. “This is a particularly simple solution, voked as a basis for discussion. For different parameters one gets less star ble, perhaps even chaotic, solutions. Wien a sudden speeding up of the ‘mantle imposes a shear ow on this nicely ordered convection flow, the disruption is mortal. The 0.03 cmn/see speed-up ofthe mantle after a 10-neter sea-level drop, Muller emphasizes, is quite comparable to the pre-existing flow velocities. Orderly buoyant trans: port depends on columns of liquid with ‘nest, monotonic density variation with distance from the center, The shear field now converts the more or less vertical columns into extended curving ares, stretching to the east as they encounter the faster shear flow near the mantle. Asthe shape of the convec- tive cells is severely distorted by the shear flow over a few hundred years, ‘the density distribution no longer sup. ports the previous flow, and the convec- tive engine becomes ‘ineffective. “A blob of highdensity liquid will now be confused to ‘find lower-density liquid both above and below it,” Morris ex: plains. ‘The convective velocities fall, ‘and the source of Kinetic energy for the dynamo is interrupted ‘As 2 now flow pattern establishes itself it tangles the previous magnetic field lines so badly that the dipole is destroyed in a few hundred years—a hundred times faster than the free decay time of the dynamo, Extinetions. Muller is no stranger to theories that invoke extraterrestrial dei ex machina. ‘Three years ago he was one of the principal proponents of the idea* that the great xtinctions— for example, the disappearance of the dinosaurs 64 million years ago—occur with a roughly 30-million-year period- icity caused by “Nemesis,” a supposed binary sister of our Sun. ‘The present paper does not invoke this contraver- sial hypothesis, but David Raup (Uni versity of Chicago) has pointed out® that the frequency distribution of the last 300 reversals suggests something like the 30-million-year periodicity he hhad earlier proposed for the mass extinctions. Nemesis is imagined to do its deadly work by perturbing the Oort cloud of 10" comets orbiting quietly in the outer reaches of our Solar System, thus greatly increasing the rate of serious collisions with the Earth every 80 million years or so. PHYSICS TODAY / FEBRUARY 1087 19