Christopher T. Russell $ Institute of Geophysics and Planetary Physics University of California Los Angeles, California, U.S.A. Originally published in Critical Problems of MagnetosphericPhysics, edited by E. R. Dyer, 116, IUSTP Secretariat,Washington, D.C., 1972. Introduction The earth's magnetic field presents an obstacle to the solar windflow about 50 earth radii (Re) across (transverse to the flow)and about 1000 Re long (in the direction of flow). Asillustrated in Figure 1, the supersonic solar wind is deceleratedand deflected by a collisionless shock wave, called the bowshock, which stands in front of the magnetic cavity. The regionof shocked solar wind plasma surrounding the cavity is called themagnetosheath. Although there is yet much to be learned aboutboth the earth's bow shock and magnetosheath, these regions willnot be discussed in depth during this symposium. The magnetic cavity, generally called the magnetosphere,effectively shields the earth and its atmosphere from the directinflux of the solar wind, so that there is very little flow ofmass or momentum across the magnetopause, the boundary of themagnetosphere. However, the departure from perfect shielding,although small, is extremely important. It is generally believedthat auroral particles and Van Allen belt particles originate inthe solar wind, and the magnetotail would not exist without theflow of momentum from the solar wind into the magnetosphere. Unfortunately, we know very little about the internal structureof the magnetopause, nor of its cousin, the polar cusp, a regionto which the magnetosheath plasma has direct access. As I will discuss in more detail below, the long extension of themagnetosphere in the antisolar direction, the magnetotail, actsas a storage reservoir for the energy released during substorms. This energy is stored principally in the magnetic field. In thecenter of the tail is the plasma sheet in which the energydensity of the particles is comparable to that of the field. Located within the plasma sheet is the current sheet, called theneutral sheet, responsible for the reversal of the tail fieldfrom the north to south lobe. This current sheet is seldomneutral, i.e., has zero magnetic field, and is not always a welldefined subregion within the plasma sheet. The plasma sheet isshown to be field aligned in Figure 1 but this has not beenproven experimentally. A neutral point is postulated to exist insome theories but if it is ever present within the orbit of themoon it must be a transient phenomenon. The magnetic flux in thetail changes with time. Magnetic flux is added through fluxtransfer from the daysid The plasmasphere is an important subregion within themagnetosphere in which the mass density is dominated by coldparticles (T~2000 K). It is thought to control waveparticl @@ instabilities and, hence, affect the loss and transport ofradiation belt particles. The radiation belt particles in turnare energized by various processes occurring during substormswhen the magnetic field in the tail is changing rapidly and theconvection electric fields are enhanced. cAs we can see from this brief introduction and as will becomeclearer during the following papers, the various regions of themagnetosphere and the physical processes occurring therein arenot independent. It is difficult or even impossible to studyeffectively any magnetospheric phenomenon in isolation. Thus, any successful investigation ofthe magnetosphere must be comprehensive. The proposedInternational Magnetospheric Study is directed towards this goalbut every effort must be made to ensure that it is comprehensive. In the following sections we will discuss the consequences of anopen magnetosphere, in particular flux transport and the growthphase of substorms, then examine a model for the expansion phaseof substorms, and finally review what still needs to be done inthese areas. The Open Versus the Closed Magnetosphere Strictly speaking, magnetic field lines have no ends. However,it is convenient and generally unambiguous to speak of fieldlines which are closed and open. A closed field is one which iffollowed through space from the surface of the earth, returns tothe surface of the earth within a finite path length. Otherwise afield line is open. As evidenced by the existence of the radiation belts andconjugate wave phenomena, the majority of magnetospheric fieldlines are closed. The field lines from the polar cap whichextend into the tail do not contain radiation belt particles norconjugate wave phenomena. However, this alone does not provethese lines are open, since other properties of the tail may beresponsible for their absence. In situ measurements of the tail magnetic field do not showwhether the tail is open or closed, either. This is because theratio of the length to the width of the tail (of the order of 20)implies that, if the tail is open the magnetic field strengthperpendicular to the magnetopause in the tail, is of the order ofone half gamma. This is unresolvable by present techniques. Insitu, measurements of the dayside magnetopause have beeninterpreted to indicate that at times, some magnetospheric fieldlines are open (Sonnerup and Cahill, 1967; 1968; Sonnerup, 1971),but the evidence is far from clear (Aubbry et al., 1971).0* + + + c6Figure 2 shows the Dungey reconnection model of the magnetosphere (Dungey, 1963) forsouthward and northward oriented interplanetary fields. As sketched here, there is quite adifference in topology when the interplanetary field switches from southward to northward.When the interplanetary is southward there are field lines with two, one and no feet on theearth, but when the interplanetary field is northward, field lines either intersect the earth twiceor not at all. Thus, in the Dungey model the magnetosphere is open for southward interplanetary fields and closed for northward interplanetary fields. Another difference between thenorth and south models, is the role played by the merging of the magnetospheric and interplanetary fields. For southward interplanetary fields flux is added to the tail by this merging;for northward interplanetary fields flux is removed from the tail. Either process is a steadystate one, however, since flux is removed from the tail for southward fields by merging at a(,_*_*neutr Observations of solar particles provide the best evidence to date that the tail field lines areopen to the solar wind. However, both solar electron data (West and Vampola, 1971) andsolar proton data (Evans and Stone, 1971; Morfill and Scholer, 1972) indicate that the tail isopen at all times, regardless of the sign of the northsouth component of the interplanetarymagnetic field. Thus, the Dungey model for northward fields sketched in Figure 2 is notconsistent with these results. The necessary modification to it is simple, however. We justtake note of the fact that it is highly improbable that the same field line will connect to boththe north and south tail neutral points. This is particularly true if the interplanetary magneticfield is not exactly north. Figure 3 illustrates the time history of a field line merging at the north tail neutral point. Thefield line Bb in the interplanetary medium is convected to the magnetopause where it, B' b',merges at the north tail neutral point N with the north lobe tail line' E forming two lines b'E and B''. The field line B' does not intersect the earth and is therefore removed fromthe tail. The line b E is attached to the earth and to the solar wind. It is dragged down thetail, b' E, and replaces the lost line B''. Thus, the merging of the magnetospheric andinterplanetary fields involves no transfer of flux from the dayside magnetosphere to the tail orvice versa. The neutral point in the center of the tail can only exist if there is another neutralpoint at infinity (Dungey, 1963). However, it must exist to separate the closed and open lines.If reconnection took place at this neutral point, the tail would be depleted in the steady state,thus this neutral point must evolve to a Ytype neutral point when the interplan For every line Bb there is another line Aa which merges with a south lobe field line. Forsimplicity this line has not been drawn in Figure 3; its merging history is strictly analogous tothat of Bb. We note the following properties of bE and AE, the merged field lines. First, thefield line connected to the north lobe is connected to the interplanetary medium south of themagnetosphere and the field line connected to the south lobe is connected to the interplanetarymedium north of the magnetosphere. Secondly, the lines bE and AE will probably convectaround the magnetosphere on opposite sides. If we let be the angle about the earth sun linein solar magnetospheric coordinates where =0 along the Ydirection, and /2 along the Zdirection (Russell, 1971), then AE lines will tend to slip to the dawn side and bE lines willtend to slip to the dusk side for /2vv and vice versa for 0vv/2. Figure 4 shows the history of field line motion external to the magnetosphere for merging inthe presence of a northward interplanetary field. In a realistic situation, i.e., c/2, themajority of the A and the majority of the B lines would be on opposite sides of themagnetosphere. We note that the models illustrated in Figures 2 through 4 do not depend onthe merging mechanism, but rather only on geometrical considerations given that mergingtakes place. The strength and importance of these effects, however, does depend on themechanism. Flux Bookkeeping(,_*_*The major practical importance of the magnetotail lies in its role as the reservoir of energy forsubstorms. During the growth phase of substorms, the field normal to the neutral sheetdecreases (Fairfield and Ness, 1970; Russell et al., 1971) and the plasma sheet thins (Hones etal., 1971). These are both indications that the number of closed field lines in the tail isdecreasing before the sudden energy release. Thus, the tail energy state must be determinedby the number of open lines in the tail, and to know the potential of the tail for energy releasewe must make an accounting of the transfer of flux from open to closed field lines and viceversa. Conceptually, we can do this by observing the flow in the ionosphere across the lastclosed field lines, or measuring the area of the ionosphere enclosed by the last closed fieldlines. If the interplanetary field merges with the dayside magnetosphere, then flux is openedup and flows into the polar cap, loosely defined here as the area in th In order to study the effect of the solar wind magnetospheric interaction on the polar cap, orto use polar cap observations to study the magnetospheric interaction with the solar wind, it isdesirable to map the polar cap into the tail. Such a map is complicated by the existence orpossible existence of electric fields parallel to magnetic fields and by nonpotential electricfields. However, there can be little doubt that the magnetopause maps into the last closedfield line in the polar cap and that field lines in the plasma sheet in the center of the tail reachthe earth near midnight. Similarly motions along the tail, i.e., parallel to the magnetic fieldshould have no effect in the polar cap, but motions around the tail axis should map into localtime motions in the polar cap. Finally, vertical motions through the tail, e.g., from themagnetopause to the plasma sheet should map into motions across the polar cap. We mustremember, however, that such a map is qualitative and thus can serve only to describe ver Figure 5 schematically illustrates the possible flows over the polar cap. If there is daysidemerging, closed field lines open up and are convected into the polar cap near noon. The polarcap increases in area. If flux is reconnected at the neutral sheet, open flux closes and isconvected out of the polar cap near midnight. The polar cap then decreases in area. Althoughneither process can proceed alone in steady state, since the polar cap cannot expand or shrinkindefinitely, these two processes can act simultaneously in steady state. If the interplanetaryfield is southward, the expected polar cap flow is essentially directly over the polar cap. Thisis simply because the north polar cap is connected to the solar wind north of themagnetosphere and the south polar cap to the south solar wind. In this situation newlyreconnected field lines are piled on top of old field lines, resulting in a sinking of field linesinto the lobes of the tail. Connection of the interplanetary field to tail field lines which are already open, does not causeflux to enter or leave the polar cap. However, it does cause a convective flow. This flowshould be concentrated at the edge of the polar cap for a northward interplanetary field sincethere is no piling on of tail flux but rather a redistribution of flux on the surface of the tail.(,_*_*The flow should go from noon to midnight since the south polar cap is connected to the solarwind north of the magnetosphere and vice versa for the north polar cap. As illustrated inFigure 5, this flow may have a dawndusk asymmetry, if the azimuthal angle, , about theearthsun line is not /2, i.e., if the interplanetary field is not exactly northward. Theasymmetry is, of course, reversed in the opposite hemisphere. We note that there should be adistributed return flow in the antisolar direction in this case, and that this process is a steadystate process because the polar cap area is left unchanged. We note in passing, that thec Finally, the bottom two panels show the case for dayside merging with northward interplanetary fields. If this can occur, the north tail field lines will be connected to the solar windsouth of the magnetosphere as in the previous case and thus the flow will tend to occur alongthe edge of the polar cap. As in the previous case, also, there may be a dawnduskasymmetry in the flow. However, as in the case of a southward directed interplanetary field,neither dayside merging nor neutral sheet reconnection can form a steady state process alonebecause the former increases the area of the polar cap and the latter decreases it. Measuring the Dayside Merging Rate Experimentally we cannot measure the size of the polar cap or the total flux in the tail with asatellite because the boundaries of these regions can change shape as well as scale. Thus, asingle point measurement of the last closed field line or the magnetopause in the tail isessentially useless. Another method would be to measure the flow velocity through the lastclosed field line boundary. Such a measurement would require a knowledge of the orientationof this boundary as well as one position on it. Ideally, we need the integrated flow across theboundary over a wide segment of the boundary. Typically polar satellite orbits cross thisboundary at right angles to it, giving a flow velocity over a rather limited region of space. Another method would be to measure the flow velocity across a closed path normal to the flow(e.g., a dawndusk satellite orbit) and find the net flow in the polar cap and the net flowexterior to the polar cap. Ideally these should be equal and opposite and in turn equal to themerging rate if the polar cap was not increasing in size. Unfortunately, such a measurementwould require magnetospheric processes to remain steady for the order of an hour, which is anunusual situation during interesting periods. Further, such a measurement requires at least atwo component measurement of the electric field with both high accuracy and precision. Another method for determining the merging rate is to measure the magnetic field normal tothe magnetopause. This has been done for the dayside magnetopause, but with muchambiguity. The problem with this method is that the magnetopause moves while the normalcomponent is being measured. A dual satellite measurement would be useful in this regard byenabling spatial and temporal variations to be removed, as well as providing a measure of thevelocity and thickness of the boundary. However, the total flux merged per unit time depends(,_*_*on the dimensions of the merging region, as well as the local merging rate, and this can onlybe obtained statistically by this approach. We note that this technique is not useful forstudying the component normal to the magnetopause in the tail because the expected size ofthis component (about l/2nT) is at about the limit of accuracy of present day magnetometers. The purpose of measuring the dayside merging rate is to determine what controls this rate andthe functional form of this control. It is expected theoretically (cf., Petschek, 1964) andobserved experimentally that the merging rate is controlled to a major degree by the magnitudeof the northsouth component of the interplanetary magnetic field. It is, therefore, importantthat this quantity be measured in the course of any comprehensive magnetosphericinvestigation. At present, we have only hints about the form of this relationship. On the one hand, Arnoldy(1971) found a high correlation between the hourly integrated southward component andgeomagnetic activity, if he assumed no interaction of the magnetosphere with the solar windwhen the interplanetary field was northward. Russell and McPherron (1972) in a study of thesemiannual variation of geomagnetic activity, found that if the southward component wereresponsible for the semiannual variation, then the interaction assumed by Arnoldy wasconsistent with the observed variation, but, the interaction proposed theoretically by Petschekin which the merging rate varies slowly and continuously as the interplanetary field directiongoes from north to south, was inconsistent with the observations. On the other hand, theconvection patterns measured on Injun 5 Cauffman and Gurnett, 1971) and OGO6 (Heppner,1972a) often resemble the sum of the bottom two convection patterns of Figure 5, which weredrawn for dayside reconnection with Despite the large gaps in our understanding of the solar wind magnetosphere interaction,attempts have been made to predict the orientation of the interplanetary field using polar capmagnetograms. Svalgaard (1968) and Mansurov (1969) noted that there was a broad minimumfor a few hours near local noon in the vertical component of a station near the northern poleY!such as Thule (magnetic latitude 86.8o) and an increase in the horizontal component of a lowerY"latitude station such as Godhavn (magnetic latitude 77.5o) during an "away" sector in theinterplanetary magnetic field, and the opposite in a toward sector. FriisChristensen et al.(1971) used this effect to predict rather successfully the sector structure for 1969. Furtherstudies, however, (FriisChristensen et al., 1972) have shown that the eastwest component ofthe interplanetary magnetic field rather than the radial component controls the polar capcurrent system. This, as we have mentioned above, is a geometrical consequence of the solarwind flow around The Growth Phase of Substorms A substorm is a sequence of auroral zone phenomena which occur repeatedly with roughly thesame behavior. There are auroral substorms, magnetic substorms, xray substorms, etc. Thesehave been described by Akasofu (1964). We are concerned, however, with magnetosphericsubstorms (Parks et al., 1968; McPherron et al., 1968; Coroniti et al., 1968), of whichauroral substorms, magnetic substorms, etc., are simply manifestations. The magnetosphericsubstorm can be divided into three stages: the growth phase when the energy reservoir in thetail is being filled; the expansion phase when the energy is being most rapidly released; andthe recovery phase when the currents set up during the expansion phase are decaying.Although there is some controversy as to whether the growth phase can be unambiguouslyrecognized in ground based data, data taken in the magnetosphere show the growth phase quiteclearly. Starting on the dayside, the growth phase is seen as an inward motion, or erosion, of themagnetopause, caused by a southward turning of the interplanetary field. A case study of suchan event was first carried out by Aubry et al. (1970). During this event, the solar windvelocity and number density remained roughly constant, while the magnetopause movedY4inwards over 2 Re. This is equivalent to the transport of 1016 Maxwells to the tail whichwould increase the flux content of the tail by about 15%. This sequence of events studied on a single pass of OGO5 is supported by statistical studies ofMeng (1970) and Fairfield (1971). Meng examined the position of the magnetopause as afunction of the AE index. He found that distant magnetopause encounters invariablycorrespond to quiet times, whereas, abnormally earthward encounters could occur at eitherquiet or disturbed times. Fairfield examined the magnetopause position as a function of thesign of the northsouth component of the magnetosheath field. He found that when themagnetosheath field was southward, the magnetopause was on the average 1 Re closer to theearth than when it was northward. This transport of flux has some important consequences in the tail. In the region where the tailboundary is parallel to the undisturbed solar wind, the magnetic flux density is determined bythe normal pressure of the solar wind which is usually found to be constant in this process.Thus, the area of the tail must increase to accommodate the additional flux. However, thestress of the solar wind on the tail equals the energy density of the magnetic field times thearea of the tail cross section. Siscoe and Cummings (1969) have proposed that the tail currentsystem must move closer to the earth in order that the force between the tail current systemand the earth's dipole be strong enough to balance the additional stress. It is very difficult to verify experimentally this proposed inward motion of the tail currentsystem, because the field measured locally at a satellite is also determined by distant currents.However, satellite observations do show the existence of taillike fields quite close to the earthin the night hemisphere preceding substorms, with moderate distortions from a dipolar field at6.6 Re (Coleman and Cummings, 1971), and almost complete taillike character at 8 Re(,_*_*(Russell et al., 1971; McPherron et al., 1972). On the other hand, since the daysidemagnetosphere shrinks and the distant tail cross section expands, the flaring angle of the tail orthe angle of the tail boundary to the undisturbed solar wind, will increase in the near earthregion. This, in turn, compresses the tail flux requiring larger currents in the near tail region,which would also increase the taillike character of the midnight magnetospheric field. Thisincrease in the tail field in the region from 10 to 30 Re behind the earth during the Another phenomenon occurring during the growth phase is the thinning of the plasma sheet(Hones et al., 1971). The plasma sheet is defined by and best studied with measurements ofthe low energy plasma in the tail (E~1 kev). However, the plasma sheet has only beenextensively studied this way at 18 Re. Fortunately, the motion of the plasma sheet may bestudied by observing its signature in the magnetic field (Fairfield and Ness, 1970; Meng etal., 1971; Russell et al., 1971); the energetic electrons (Meng et al., 1971), or energeticprotons (Buck et al., 1972). It is observed to drift with apparent velocities from 4 to 20km/sec during the growth phase. We note that this thinning and the subsequent expansion ofthe plasma sheet during the expansion phase need not represent convective motions, but in factmay represent the loss and acceleration of plasma sheet particles. In summary, the essential features of the growth phase in the magnetosphere are: a transportof flux from the dayside magnetosphere which causes the dayside magnetosphere to shrink; anincrease in the strength of the near tail current systems due to the increased flaring angle ofthe tail, and possibly an inward motion of the current system; and a thinning of the plasmasheet. The Onset of the Expansion Phase of Substorms A priori, the expansion phase of substorms could be initiated by a trigger in the solar wind, anevent in the distant tail, in the near tail or in the ionosphere since all these regions areobserved to participate to some degree in the substorm process. However, there is evidence toindicate that the "disturbance" causing the expansion phase is initiated in the near tail andpropagates tailward. This evidence stems primarily from a comprehensive study ofobservations made in the near tail region during a single pass of the OGO5 satellite. The firstpiece of evidence is the simultaneity of events in the near tail with the onset of the groundsignature of substorms (McPherron et al., 1972). This, however, essentially determines onlythe field line on which the disturbance is initiated and the substorm could be triggeredanywhere along that field line from the ionosphere to the equator. The evidence for theequator rather than the ionosphere, is that during this pass the thinning of the plasma sheetproceeded until t The model that best explains these observations is illustrated in Figure 6. During the growthphase, the plasma sheet thins. Satellites such as OGO, denoted by the X and VELA by thestar, see evidence of this thinning. At the onset of the expansion phase, a Xtype neutral pointforms near the earth. This may form on closed field lines in the plasma sheet or the plasmasheet may disappear over a limited interval of local time so that merging is initiated on thefirst open field lines. OGO might leave the plasma sheet temporarily at this time only torapidly reenter the plasma sheet. VELA need not leave the plasma sheet until much later asthe merging region passes by VELA. This explains the variable timing of the thinning eventsrelative to expansion phase onsets seen at the VELA satellite. Eventually, the expandingplasma sheet on the earthward side of the neutral point overtakes VELA. There are several good reasons for invoking a moving neutral point model such as this one.First, there should exist a pressure imbalance across the merging region because the flow isimpeded by the earth on one side, but unimpeded on the other. Second, since the averagefield is northward in this region, any neutral point formed close to the earth must havetransient behavior and third, southward magnetic fields are observed prior to plasma sheetexpansions (cf. Figures 6 and 9 of Fairfield and Ness, 1970). However, a conclusive test ofthis model awaits the observation of this southward component with the appropriate time delayby two suitably spaced satellites in the tail, such as on a motherdaughter mission. Future Requirements The models of the interaction of the solar wind with the magnetosphere which we havepresented here are only qualitative models. If we are to make them quantitative as we must, ifwe ever hope to predict or control magnetospheric phenomena, we need to continue ourobservational program. This includes new satellite missions as well as continued groundbasedobservations. The most important requirement is a solar wind monitor as close to the earth aspossible to minimize uncertainties in arrival times of solar wind events. This argues for aearth orbiting probe. If this is done, then two spacecraft would be needed for 100% solarwind coverage. The extra cost of two launches would be repaid in the security of a redundantspacecraft, and the added reliability in measuring the orientation of discontinuities when bothspacecraft are in the solar wind. The second most important new mission is an eccentric orbiting spacecraft with an apogee ofabout 15 Re to make measurements of the magnetopause and near tail region. The height ofapogee should be kept to a minimum to maximize the duration of magnetosphericmeasurements. Apogee and perigee should occur as close as possible to the earth's equator,although the inclination of the orbit may be arbitrary. This will maximize observation timenear the neutral sheet and at low latitudes on the magnetopause. This mission should be amotherdaughter type. ( ,_*_*Thirdly, a similar mission with apogee well above the earth's equator to probe the polar cuspshould be flown. Since the polar cusp moves, and its dimensions are quite important( ,_*_*parameters, this should also be a motherdaughter mission. Low altitude polar orbiters can also make a fundamental contribution to magnetosphericstudies and should not be overlooked. They can be used to map the convective motions overthe pole and probe the length of the tail via studies of solar particle events. A dual launchwith spacecraft 180 degrees apart would be quite useful in this regard. Finally, although we have stressed observations in the distant magnetosphere, and the growthphase and the onset of the expansion phase of substorms, we must remember that we need alsoto understand the consequences of the solar wind interaction and substorm processes deepwithin the magnetosphere. Synchronous orbiters are good, convenient platforms for suchstudies, and may be ideal locations for future magnetospheric monitors. However, at thepresent we know little about the magnetosphere just inside and just outside of the synchronousorbit. Thus, serious consideration should be given to a plasmaspheric explorer, perhaps with a24 hour orbit, but with an apogee of about 10 Re and a perigee around 3 Re to probe the innermagnetosphere.* x-++ In conclusion, we have just finished a decade of in situ magnetospheric exploration. Thisexploratory phase has provided us with qualitative models for most magnetosphericphenomena. The task ahead of us, in particular during the International MagnetosphericStudy, is to make these models quantitative. If we are to achieve this goal, we must have acomprehensive set of synoptic measurements on the ground, in the magnetosphere, and, mostimportantly, in the near earth solar wind. Acknowledgments Discussions with my colleagues, especially M.G. Kivelson and R.L. McPherron, have beenvery helpful in preparing this review. During the preparation of this report the authorreceived support from the National Aeronautics and Space Administration under contract NAS59098. References Akasofu, S. I., Polar and Magnetic Substorms, D. Reidel, Dordrecht, Holland, 1968. Arnoldy, R.L., J. Geophys. Res., 76 (22), 51895201, 1971. Aubry, M.P. and R.L. McPherron, J. Geophys. Res., 76, 4831, 1971. Aubry, M.P., C.T. Russell and M.G. Kivelson, J. Geophys. Res., 75, 7018, 1970. Aubry, M.P., M.G. Kivelson and C.T. Russell, J. Geophys. Res., 76, 1673, 1971. Buck, R.M., H.I. West, Jr. and R.G. D'Arcy, Jr., Satellite studies of magnetosphericsubstorms on August 15, 1968: 7, OGO5 energetic proton observationsspatial boundaries,submitted to J. Geophys. Res., 1972. Cauffman, D.P. and D.A. Gurnett, J. Geophys. Res., 76, 6014, 1971. Coleman, P.J., Jr. and W.D. Cummings, J. Geophys. Res., 76, 51, 1971. Coroniti, F.V., R.L. McPherron and G.K. Parks, J. Geophys. Res., 73, 1715, 1968. ) ,**Dungey, J.W., The structure of the exosphere or adventures in velocity space, in Geophysics,The Earth's Environment, edited by C. DeWitt, J. Hieblot, A. Lebeau, Gordon and Breach,New York, 1963. Evans, L.C., E.C. Stone and Meng, C.I., J. Geophys. Res., 75, 3252, 1970. Meng, C.I., S.I. Akasofu, E.W. Jones and K. Kawasaki, J. Geophys. Res., 76, 7584, 1971. Morfill, G. and M. Scholer, Configuration and reconnection of the geomagnetic tail deducedfrom solar particle observations, submitted to J. Geophys. Res., 1972. Parks, G.K., F.V. Coroniti, R.L. McPherron and K.A. Anderson, J. Geophys. Res., 73,1685, 1968. Petschek, H.E., Magnetic field annihilation, in AASNASA Symposium on the Physics ofSolar Flares, 425, NASA SP50, 1964. Russell, C.T., Cosmic Electrodyn., 2, 184, 1971. Russell, C.T. and R.L. McPherron, The semiannual variation of geomagnetic activity,submitted to J. Geophys. Res., 1972. Russell, C.T., R.L. McPherron and P.J. Coleman, Jr., J. Geophys. Res., 76, 1823, 1971. Svalgaard, L., Sector structure of the interplanetary magnetic field and daily variation of thegeomagnetic field at high latitudes, Geophys. Pap. R6, Danish Meterol. Inst., Charlottenlund,1968. Siscoe, G.L. and W.D. Cummings, Planet. Space Sci., 17, 1795, 1969. Sonnerup, B.U.O., J. Geophys. Res., 76, 6717, 1971. Sonnerup, B.U.O. and L.J. Cahill, Jr., J. Geophys. Res., 73, 1757, 1968. Unti, T. and G. Atkinson, J. Geophys. Res., 73, 7319, 1968. West, H.I., Jr. and A.L. Vampola, Phys. Rev. Letters, 26, 458, 1971. DISCUSSION Ness: In your first figure and in your discussion you referred to the possible existence of asingle neutral line which, if it exists, probably occurs beyond the distance to the Moon.Recently Schindler and Ness (1971) have shown that magnetic field data from Explorer 34 areconsistent with multiple neutral lines inside the orbit of the Moon. What is the basis for yourbelief that only one neutral line beyond the Moon can exist? ),**Russell: In my paper, I stated that if a neutral point exists inside the orbit of the Moon, itmust be a transient phenomenon. I should have said that if a single neutral point exists, itmust be a transient phenomenon. The work of Schindler and Ness (1972) has shown thatmagnetic field observations in the tail are consistent with the existence of many neutral pointsseparated by northward fields so that the spatial average of the cislunar field is northward.Frankly this result is so new and unexpected in the light of previous theories about the tail,that I remain skeptical. Fortunately, the proposed motherdaughter mission can provide adefinitive test of the two competing models, i.e (1) a moving single neutral point versus (2)multiple stationary points. If the two spacecraft A and B always encounter the neutral points inthe same sequence, e.g. A, then B; A, then B, etc., then they are passing through multiplepoints. However, if the two spacecraft encounter neutral points in the sequence A, then B; B,the Ness: The sequence in which neutral points are observed will not, however, uniquely separatea moving single neutral point from a moving ensemble of neutral lines. Only if there were astationary configuration which was traversed by the satellite would your suggested test bevalid. Since you have emphasized the transient feature of the neutralsheet structure, I doubtwhether any method other than that of statistical consistency (as used by Schindler and Ness)will shed light on this important problem. Burrows: I refer to your discussion of convection around the morning and evening auroralzones and across the polar cap: your diagram showed convection preferentially on the morningside for daytime merging and neutral sheet reconnection. Do you mean to imply anypreference for morningside convection as a result of any properties of the interplanetary field? Russell: The choice of the asymmetry in the bottom two panels of Figure 5 was arbitrary. Astronger flow on the morning side, as sketched, would be expected in the northern polar capwhen the solar magnetospheric Y component of the interplanetary field is positive and in thesouthern polar cap when it is negative. Willis: The observational evidence for "erosion" of magnetic flux from the dayside of theYearth is that the magnetopause moved earthwards about 2RE while the solar wind momentumflux remained constant but after the direction of the interplanetary magnetic field changed fromnorthward to southward. Are these observations consistent with the classical pressure balanceY!condition at the magnetopause, in which the gas pressure on the magnetopause varies as p=pSTY"cos 2 v, and an increasing magnetic intensity with decreasing distance from the earth? Russell: Yes, these observations are consistent with the classical pressure balance condition. If magnetic flux cannot return to the dayside hemisphere to replace the eroded flux, then themagnetopause changes its shape as it changes its position. Such a situation would arise eitherif the rate of reconnection in the tail were less than the dayside merging rate or if theionosphere impeded the flow from the nightside. The former situation has been treated for atwodimensional magnetosphere by Unti and Atkinson (1968); the latter situation is presentlyunder study by Coroniti and Kennel.),**Figure Captions Figure 1. The noonmidnight cross section of the magnetosphere. The size of the polar cusphas been exaggerated; the neutral point has been drawn at 30 Re although it must commonlybe beyond the orbit of the moon; and the plasma sheet boundary is not necessarily fieldaligned. Figure 2. The Dungey model of the magnetosphere. The letter N denotes a neutral point. Arrows indicate the direction of plasma flow. The model is highly qualitative and inparticular no attempt has been made to draw these diagrams to scale (after Dungey, 1963). Figure 3. A modified Dungey model of the magnetosphere illustrating merging when theinterplanetary field is not exactly northward. In this situation field lines are expected to mergewith only one of the two neutral points on the surface of the tail. Only merging at the northneutral point is illustrated. Other field lines merge at a south neutral point (not shown) in ananalogous manner. The sequence of field lines Bb, B'b, etc. indicates the time history of onefield line or a set of field lines at one particular time. The dashed portion of b''E indicatesthat this field line is out of the magnetic meridian. It is carried around the magnetosphere bythe solar wind. Figure 4. The connection of interplanetary field lines to the magnetopause for northwardinterplanetary fields. "A" field lines have merged with the south lobe of the magnetotail and"B" field lines have merged with the north lobe. Lines A and a, B and b, etc. wereconnected before merging with the tail field. If merging took place on the dayside fornorthward fields, the field lines a, B, and B' would not be present and field lines A and bwould be the products of the breaking of a single interplanetary field line. This figure hasbeen drawn to emphasize its consistency with Frank's model of the magnetosphere (see figure7, Frank and Gurnett, 1971). Figure 5. Qualitative polar cap convection patterns due to merging of the interplanetary fieldwith the dayside magnetospheric field lines, and with the tail field lines and due to merging ofopen tail field lines across the neutral sheet. Plus signs indicate the area enclosed by the lastclosed field lines (LCF) is increasing; minus signs indicate it is decreasing. Dawnduskasymmetries in the flow may arise when there is a nonzero eastwest component of theinterplanetary magnetic field. The middle panel indicates the situation expected when theinterplanetary field has a negative solar magnetospheric Y component. Figure 6. A model for the onset of the expansion phase of a magnetospheric substorm. The xmarks the position of the OGO5 satellite during a well documented substorm on August 15,1968. The asterisk marks a typical location of a VELA satellite during a substorm. At timeA, the plasma sheet is thinning but thins faster near the earth. At time B, an xtype neutralpoint has formed near OGO, and OGO5 leaves the plasma sheet. The neutral point movesaway from the earth and OGO is enveloped in the expanding plasma sheet. Finally, at D, theneutral point has moved far down the tail enveloping VELA in the plasma sheet also.