C. T. Russell and B. R. Lichtenstein
Institute of Geophysics and Planetary Physics, University of California, Los Angeles, California 90024
Abstract. Magnetic field measurements made on board the Apollo 15 and 16 subsatellites in low-altitude lunar orbit are used to study the occurrences of compressions of the magnetic field over the terminator regions when the moon is in the solar wind. These compressional disturbances occur at or just in front of the terminator, but only when specific selenographic regions appear at the lunar limb. The occurrence rate, strength, and position of the compression relative to the limb are different for each source region. Where overlap of limb compression occurrence statistics and lunar magnetic field maps exist, the limb compression occurrence rates have a high degree of correlation with the magnetic field strength. Thus the most probable cause of limb compressions is the deflection of the solar wind by the lunar surface magnetic field as the solar wind flows past the lunar terminators.
The basic nature of the interaction of the solar wind with the moon was evident from the initial results of the experiments aboard the lunar orbiter Explorer 35. The MIT plasma probe [Lyon et al., 1967] measured undisturbed interplanetary proton flux on the sunward side of the moon and the instrument noise level directly behind the moon. The boundary of the low-flux region was not sharp, but rather there was a gradual decline in flux as the lunar shadow was entered. The early measurements from the Ames Research Center (ARC) magnetometer [Colburn et al., 1967] and the Goddard Space Flight Center (GSFC) magnetometer [Ness et al., 1967] showed that the principal perturbations of the solar wind magnetic field were restricted to a lunar wake region downstream of the moon and that they consisted of an enhancement of the magnetic field magnitude directly behind the moon bounded by dips in the magnetic field strength. No evidence of detached bow shock upstream of the moon was detected by either the plasma probe or the magnetometers. These results were interpreted by the above authors and Michel  to mean that the solar wind plasma flowed undeflected onto the sunward lunar surface, the principal perturbations of the solar wind plasma and magnetic field being associated with the creation of a plasma void behind the moon.
Figure 1 shows the orbit of Explorer 35 about the moon and magnetic field observations on one pass through the optical shadow. In region 1, upstream of the moon, the magnetic field is apparently unaffected by the near presence of the moon, while in region IV, behind the moon, an enhanced field is seen, bounded by dips (labeled III) as the satellite enters and leaves the solar wind shadow. Subsequent studies by Ness et al. , Taylor et al. , and Colburn et al.  confirmed these initial results and revealed occasional small increases, labeled II, in the field magnitude upstream from the dips in the field strength. These features were referred to as 'penumbral increases' by Ness et al.  and 'external maxima' by Colburn et al. . Despite their small size at Explorer 35 and their only occasional presence the explanation of this phenomenon has served as the inspiration for much theoretical work on the solar wind-moon interaction and will provide the main focus of this paper.
|Fig. 1. Sample orbits of Explorer 35 and the Apollo subsatellites showing the location of the various features associated with the interaction of the solar wind with the moon. The Explorer 35 orbit is shown projected onto the ecliptic plane, while the subsatellite orbit has been rotated into the ecliptic plane. The inclinations of the orbits to the ecliptic plane are 11° for Explorer 35, 28° for Apollo 15, and 10° for Apollo 16. The magnetic signature of the interaction is shown in the bottom panel from a typical pass of the Explorer 35 [Whang and Ness, 1970].|
First, we would like to discuss nomenclature. The term external maxima seemed too general to us, while the term penumbral increase was inspired by the particle-kinetic approach, which is now generally out of favor. Further, the occasionally used term 'limb shock' seems inappropriate because there is no evidence that the plasma is shocked on passage through the features. Rather, since these features occur over the lunar limbs at low altitudes and have their source there and since they have been shown to consist of compressions of both the field and the plasma [Siscoe et al., 1969], we prefer and will use throughout this paper the term 'limb compression.'
Early theoretical studies of the steady state solar wind-moon interaction were based on two different premises. One group [Whang, 1968a, b, 1969, 1970; Dryer, 1968; Wang, 1971; Catto, 1974] assumed that the absence of a lunar bow shock implies that for an object with a characteristic scale length of a lunar diameter the solar wind does not interact as a continuum but rather as an aggregate of individual particles requiring a kinetic theory description. The other group [Johnson and Midgley, 1968; Michel, 1968a, b; Wolf, 1968; Spreiter et al., 1970; Beers, 1972] assumed that a hydromagnetic description is appropriate and that the absence of a detached bow shock is due to the neutralization (and thereby effective absorption) of the solar wind plasma incident on the lunar surface.
Attempts to explain limb compressions by means of the particle-kinetic approach have been unsuccessful. The small positive and negative increases in |B| shown in Whang's [1968b] results apparently were due to an artifact of his numerical calculations [Schwartz et al., 1969]. His later treatment [Whang, 1969, 1970] shows no such compressions except when an ad hoc perturbation above the lunar limbs is included in the boundary conditions. A recent more thorough treatment of the problem by Catto , which includes no ad hoc perturbation, fails to reproduce the compressions. Catto suggests that this is perhaps due to some of the approximations used.
The first attempts to explain the source of the limb compressions from the fluid point of view were made by Hollweg  and Schwartz et al. . These authors postulated that the magnetic pressure needed to deflect the solar wind slightly at the lunar limbs is produced by a version of the unipolar generator bow shock mechanism originally hypothesized for the moon by Sonett and Colburn  and Johnson and Midgley . The limb compressions were thus interpreted to be manifestations of a weak lunar limb shock. It should be noted that this mechanism is equivalent to the Tozer and Wilson  modification of Gold's  field line accretion hypothesis for formation of a lunar bow shock. More recently, Horning and Schubert  have examined this mechanism, and they found that it is viable only for conductivities much above those found in lunar samples. Hollweg  proposed an alternative induction mechanism based on eddy currents induced in lunar conducting islands by time-varying magnetic fields. However, a statistical study of the magnetic field perturbations in the lunar wake by Colburn et al.  does not support either steady state or transient induction as a source of lunar limb compressions.
Many other mechanisms for locally deflecting some of the solar wind flow have been proposed since these early theories. The most popular of these theories is the possible deflection by lunar remanent magnetic fields [Barnes et al., 1971; Colburn et al., 1971; Mihalov et al., 1971; Sonett and Mihalov, 1972; Lichtenstein et al., 1973; Russell et al., 1973; Sharp et al., 1973]. Siscoe and Goldstein  have shown how the solar wind would be deflected for different field geometries. Other mechanisms include photoelectron pressure over highland regions [Criswell, 1973a, b] and deflection by the lunar atmosphere [Siscoe and Mukherjee, 1972; Wallis, 1974]. Finally, Perez-de-Tejada  has suggested that limb compressions are due to a thermoviscous effect in a magnetohydrodynamic boundary layer.
The Explorer 35 observations do not unambiguously determine which of the above mechanisms is correct. However, they do limit the possibilities. First, the limb compressions consist of a compression of both the magnetic field and the plasma and are accompanied by a slight deflection, 3°-6° of the solar wind away from the moon [Siscoe et al., 1969]. This indicates that when a limb compression is present, the plasma is deflected at the limbs before its expansion into the lunar wake and is a point in favor of the fluid deflection models and against the particle-kinetic hypotheses and the boundary layer hypothesis.
Limb compressions are not always present. Their occurrence is not simply a function of selenographic longitude [Whang and Ness, 1972]. However, when latitude is included in the correlation, the occurrence rate is found to be controlled by the selenographic region at the lunar limb [Mihalov et al., 1971; Sonett and Mihalov, 1972]. This argues for the mechanisms which invoke a deflection whose strength is a function of lunar surface properties. Thus the lunar magnetic field is a strong candidate, but the lunar photoelectron density and neutral atmospheric density could also vary with position. Finally, the amplitude of limb compressions decreases with distance from the moon and increases with increasing plasma β [Whang and Ness, 1972].
In this paper we use magnetic field observations over the lunar limbs obtained from more than 2500 circumlunar orbits of the Apollo 15 and 16 subsatellites. We examine where the limb compressions occur in relation to the lunar limbs, in relation to specific selenographic regions, and in relation to the lunar remanent magnetic field as measured at 100-km attitude on the same satellites. These results can be interpreted in no other way than that properties of the lunar surface control the occurrence of limb compressions as reported by Mihalov et al.  and Sonett and Mihalov . Further, they lend strong support to the hypothesis that limb compressions are caused by the deflection of the solar wind at the limbs by the lunar remanent magnetic field.
The Apollo subsatellite magnetometer consists of two flux gate sensors mounted orthogonally at the end of a 1.83-m boom and an electronics unit housed in the main body of the spacecraft. A detailed description of the magnetometer has been given by Coleman et al. [1972a, b]. The vector components of the magnetic field are constructed from the measurement of the amplitude and sign of the component of the field along the spin axis and the amplitude and phase of the field in the spin plane. The amplitude in the spin plane is determined on the ground by Fourier analysis of high telemetry rate data (one sample per second) and by analog techniques on the spacecraft for the low rate stored data (1 sample every 12 or 24 s). The phase is determined by measuring the time delay between the crossing of solar direction and the positive-going zero crossing of the magnetic field. During eclipse the time of the positive-going zero crossing is telemetered to earth, and the times of crossings of the solar direction are computed from an empirical model of the change in spin rate as the subsatellite cools during eclipse together with the phase of the sun both before and after eclipse. Data are generally available for the entire orbit, except for a period of about 10 min during each orbit when stored data are being transmitted.
The Apollo 15 subsatellite was launched on August 4, 1971, and operated successfully for 7 months before a failure in the telemetry system prevented further transmission of the magnetometer data. The orbit of the subsatellite was inclined 29° to the moon's equator, was at an average altitude of close to 100 km, and had a period of 2 hours as compared to the approximately 12-hour period of Explorer 35. The shorter orbital period enabled us to follow the evolution of limb compressions as the moon rotated in relation to the solar wind direction. The subsatellite orbit has been drawn to scale in Figure 1.
Typical magnetic field signatures for complete lunar orbits when the moon was in the solar wind, the magnetosheath, and the lobes of the geomagnetic tail are shown in Figure 2. The enhancement in the cavity or shadow region is present in both the magnetosheath and the solar wind but is typically as shown here, i.e., larger in the magnetosheath. One well-developed limb compression is present on the sunrise limb crossing in the solar wind. As is evident from these data, it is difficult to detect limb compressions in the background of large temporal fluctuations usually observed in the magnetosheath. The field enhancement in the magnetotail is due to the lunar field. From data such as these, lunar field maps have been created [Russell et al., 1974, 1975].
|Fig. 2. Magnetic field strength versus time for subsatellite orbits while the moon was in the solar wind (top), the magnetosheath (middle), and the geomagnetic tail (bottom). The region labeled shadow in this figure and subsequent figures corresponds to the period when the satellite was optically shadowed from the sun by the moon.|
Before proceeding to discuss the observations in more detail, we should define some of the terms we will be using. First, we use the term 'shadow' as in Figure 2 to refer to the region in which the subsatellites are optically shadowed by the moon from the sun. Second, the term 'terminator' is used as a short form of 'solar wind terminator,' the great circle on the moon perpendicular to the aberrated flow direction of a 400 km s-1 solar wind. Third, the term 'limb' refers to the region of the moon near the terminator. Finally, for purposes of calculating limb compression occurrence rates for specific regions on the moon, we have assigned the occurrence or nonoccurrence of a limb compression on each orbit to the position on the moon's surface of the intersection of the orbit plane with the solar wind terminator.
Limb compressions are highly variable in amplitude, shape, and occurrence. As is illustrated in the top panel of Figure 3, compressions can be absent at both limbs, or they can be present at either limb, as is illustrated in the middle two panels, or at both limbs, as is illustrated in the bottom panel. Such behavior might be expected if the limb compressions were a general feature of the interaction whose occurrence was modulated by the orientation of the external magnetic field. However, these data do not support such control. The upstream interplanetary magnetic field measured on the same spacecraft at local noon was roughly along the Archimedean spiral direction for both orbits 1965 and 280 and was between the Archimedean spiral and flow direction for the other two orbits. Table 1 lists the upstream field direction for each of these four orbits.
|Fig. 3. Magnetic field strength versus time for subsatellite orbits while the moon was in the solar wind. This shows the four possible types of behavior.|
TABLE 1. Magnetic Field Direction in Solar Ecliptic Coordinates Measured Near Local Noon for the Orbits Shown in Figure 3
We note that although the overall interaction does not appear to be controlled by the direction of the interplanetary field, the size of individual limb compressions does exhibit such control. The detailed discussion of this point is beyond the scope of this paper and will be deferred until a later publication.
While limb compressions may appear at neither limb, either limb, or both limbs, they do not do so randomly. Figures 4-8 illustrate this point. When limb compressions are absent, they tend to remain absent for many orbits, as is illustrated in Figure 4, where they are absent above both limbs. We note, however, that the lunar limb region is almost always magnetically disturbed. Figure 5 shows a set of limb compressions occurring at the eclipse limb. This set of orbits shows a gradual evolution of the size and shape of the limb compression. Figure 6 shows a sequence of eclipse limb crossings which do not show such a gradual evolution, yet limb compressions are present on each successive orbit. We note that the absence of any limb compression at sunrise on all these orbits rules out the possibility that limb occurrences are simply controlled by the value of some scalar parameter of the solar plasma such as temperature or Mach number, since such a control would require a 2-hour modulation of this parameter in phase with the satellite orbit to reproduce the effect seen here.
|Fig. 4. A sequence of orbits illustrating the persistence of no limb compressions at either limb.|
|Fig. 5. A sequence of orbits illustrating the persistence of an eclipse limb compression.|
|Fig. 6. A sequence of orbits illustrating the persistence of eclipse limb compressions, sunrise compressions being absent.|
Figure 7 shows a similar example at sunrise. Again there is no limb compression at eclipse despite the presence of a very large compression at the sunrise terminator. This sequence is rather stable in form when it is compared to that shown in Figure 6. Finally, Figure 8 shows a sequence of limb compressions at both terminators. Note the relative stability of the shape of the sunrise compressions compared to the eclipse compressions.
|Fig. 7. A sequence of orbits illustrating the occurrence of sunrise limb compressions, eclipse compressions being absent.|
|Fig. 8. A sequence of limb compressions occurring at both terminators.|
In summary, the limb compressions can occur at neither limb, either limb, or both limbs. When they do occur, they persist for many revolutions of the spacecraft, an indication that the occurrence is not simply controlled by the modulation of some scalar parameter of the solar plasma. Further, the direction of the interplanetary magnetic field does not simply control the appearance of limb compressions. Finally, they exhibit both slow evolution from orbit to orbit and at times rather abrupt changes from orbit to orbit.
THE LOCATION OF LIMB COMPRESSIONS
Although it is evident from Figures 3 to 8 that limb compressions occur near the crossing of the lunar terminators, since the satellite is approximately 100 km above the lunar surface, the precise geometric relationship is not clear from these figures alone. The lunar terminator crossing occurs about 6 min before (after) the subsatellite enters (leaves) the solar wind shadow. To find this relationship, we must first define criteria for observing a limb compression. We have taken the relatively conservative criteria that the maximum magnetic field must exceed 1.5 times the upstream field and that the upstream field must exceed 4 g . This in turn limits all compressions included in this study to enhancements over background greater than 2 g . Further, we restrict the area of concern to the region within 45° upstream of satellite optical eclipse or sunrise. Finally, we do not use any data when it is apparent that the interplanetary magnetic field is unsteady, such as near interplanetary sector boundaries or at discontinuities in the magnetic field.
We have defined five positions on the limb compressions. The first is the position at which the first unambiguous break from the upstream field occurs. The second is the location of the maximum field. The third is the location at which the field returns to its upstream value. We note that this position is the same as that referred to by Whang  and Whang and Ness  as the lunar Mach cone. The fourth is the position of the minimum following the compression, and the fifth is the position where the field recrosses the upstream field value. If any of these positions could not be accurately defined on any orbit, we measured and used only those locations that could be so defined.
Figure 9 shows the average location of these positions together with the standard deviation of the mean for all 269 compressions meeting our criteria in the Apollo 15 data. All sunset terminator crossings are in the northern hemisphere between 22° and 28° N latitude, and all sunrise terminator crossings are in the southern hemisphere between 22° and 28° S. The sunset compressions start at roughly 9' in front of the terminator and maximize over the terminator, with a following minimum directly downstream of the solar wind terminator. We recall that we have compensated for the aberration of a 400 km s-1 solar wind in this analysis. Over the other limb the compressions on the average begin 20° before the terminator and maximize roughly 9° in front of it, a minimum following nearly 5° ahead of the solar wind shadow. We attribute the difference between the two limbs to the fact that the source of the limb compressions is stronger in the southern latitudes over which we have flown than in the northern latitudes. We will examine this hypothesis in more detail in a later section.
|Fig. 9. The location of limb compressions relative to the solar wind terminator. Angles are measured in the orbit plane with ±90' at the two crossings of the terminator. Position I is the beginning of the field increase. Position 2 is the maximum. Positions 3 and 5 mark the points where the field strength crosses its upstream value, and position 4 marks the intervening minimum. The error bars give the probable error in the mean.|
Having defined the location of limb compressions relative to the lunar limbs, we are now ready to investigate the claim of Mihalov et al.  and Sonett and Mihalov  that the limb compression occurrence rate is modulated by some property of the lunar surface at the limb. To do this, we assume that the region associated with the appearance or nonappearance of limb compressions is the region beneath the satellite at the crossing of the solar wind terminator. We note that strong limb compressions seen at sunrise actually have an average location 9° in front of the terminator. We should keep this fact in mind when we associate large limb compressions with source regions on the lunar surface.
Having assigned every appearance and nonappearance of a limb compression to a specific selenographic position, we calculated the rate of occurrence in 5° longitude by 5° latitude bins. There were 120 of these bins having five or more observing periods assigned to them. The rate of occurrence of limb compressions in these bins is shown in Figure 10. The shaded regions indicate the fraction of the regions in the 0-20% interval with exactly 0% occurrence and the fraction of the regions in the 80-100% interval with 100% occurrence.
|Fig. 10. The number of selenographic regions as a function of limb compression occurrence rates. Shaded regions denote 0% occurrence on the left and 100% occurrence on the right. Each occurrence of a limb compression was assigned to a 5' X 5o box at the terminator. Only boxes with five or more occurrences were used in making this graph.|
It is obvious that the distribution of limb compression occurrence rate is not a normal distribution. The mean occurrence rate is 23% with a standard deviation of 34%. However, instead of a most probable rate of occurrence near the mean rate and a decreasing rate on either side of the peak, there are two maxima at high rates and low rates. Thus there appear to be two types of regions, those with low occurrence rates and those with high occurrence rates.
Figure 11 shows the occurrence rate for overlapping 10° longitude by 10° latitude bins for the Apollo 15 and 16 data as a function of longitude for the four latitude bands covered by the two satellites. We required five or more observing periods before calculating an occurrence rate with the Apollo 15 data and four or more with the Apollo 16 data. This figure demonstrates that these high occurrence rate regions are far from being randomly distributed about the lunar surface. Limb compressions occur more frequently over the lunar far side. However, not all far side regions are associated with limb compressions. In fact less than 50% of the far side regions are associated with limb compression occurrence.
|Fig. 11. Limb compression occurrence rate versus selenographic longitude for the four Apollo 15 and 16 limb crossing tracks. The latitude of the limb crossings changed only slowly during these missions.|
Figure 12 shows these data in a different format. The upper panel shows the occurrence rate superimposed on a sketch of the lunar mare. The bottom panel shows the location of the limb crossing for which a limb compression was observed. While there are several limb compressions close to the edges of mare basins, only two limb compressions are clearly over mare material. By far, most limb compressions occur over highland material.
|Fig. 12. Limb compression occurrence rate versus selenographic position for Apollo 15 and 16. The top panel shows the rate superimposed on a sketch of the mare. The bottom panel shows the positions associated with the limb compressions.|
Finally, Figure 13 illustrates the repetitive nature of limb compression occurrences. Here the occurrence rates for the Apollo 15 data are divided into three groups, each group containing data from 2 months for lunar longitudes from 140' to 280'. Roughly the same behavior is seen in each successive lunation. Gradual changes in the pattern over the 6 months may be due to the slow equatorward progression of the latitude of the terminator crossings from 28° to 22° S latitude.
|Fig. 13. Limb compression occurrence rate from 140° E to 280° E for lunations 1 and 2, lunations 3 and 4, and lunations 5 and 6 from the Apollo 15 data.|
In summary, on the average, limb compressions have their maximum amplitude above or in front of the terminator crossing. The limb compressions observed in the northern hemisphere on Apollo 15 maximize over the terminator, while those in the south maximize 9° in front of the terminator. When ordered by the selenographic location of the terminator crossing, a strongly nonrandom pattern emerges. Certain regions have high occurrence rates, while most have low occurrence rates. There are three such high occurrence rate regions in the Apollo 15 data: the 30° -110° E region in the north and the 140° to -165° E and -165° E to -115° E regions in the south. We note that the source region of the four examples chosen by Sonett and Mihalov [19721 to illustrate the repetitive nature of limb compressions corresponds to the second of the three regions found in the Apollo data. This fact further emphasizes the agreement with the conclusions of the Ames group.
It is evident from Figures 3 to 8 that the eclipse limb compressions tend to be somewhat weaker than the sunrise compressions. Further, as we noted above, the eclipse compressions occurred directly over the terminator, while the sunrise compressions occurred on the average 9° ahead of the terminator. Since the eclipse data correspond to the northern hemisphere only, it is apparent that at least one of the three source regions has properties different from those of the other two. To quantify these differences, we have calculated the average strength and average position of the limb compressions in each of these three regions. These are shown in Figure 14 for the northern source region, labeled N, the southern region from 140° to -165° E, labeled VDG, and the southern region from -165° E to -115° E, labeled A. The reason for the last two labels is that the peak occurrence rates of these two regions lie close to the craters Van de Graaff and Apollo, respectively.
|Fig. 14. The average angle ahead of terminator at which peak compression is observed versus average strength of limb compression for the three main source regions. Error bars represent the probable error of the mean.|
As was expected, the northern region produces limb compressions that on the average are smaller and closer to the terminator than those of the two southern regions. However the N region is not as different from the VDG region as the VDG region is from the A region. The average amplitude in the A region is 22% greater than that in the VDG region, while the average amplitude in the VDG region is only 13% greater than that in the N region. Further, while the N region compressions peak 1.9° ahead of the terminator and the VDG region compressions peak 3.7° ahead of the terminator, the A region compressions peak 13.7° ahead.
In summary, the different source regions have distinct properties. Especially striking is the difference between the VDG region and the A region in the south, whose separation into two different regions might have seemed arbitrary upon examination of Figures 11 and 12.
CORRELATION WITH THE LUNAR SURFACE MAGNETIC FIELD
Maps of the remanent surface magnetic field have been constructed from the Apollo 15 and 16 subsatellite magnetometer data obtained in the geomagnetic tail [Russell et al., 1974, 1975] see also the frontispiece of the Proceedings of the Fifth Lunar Science Conference. These maps overlap regions of good limb occurrence statistics in only two regions, in the north from -50° to 10° E and in the south from 130° to 190° E. Figure 15 shows the limb compression occurrence rate and the field strength at 100-km altitude over these two regions. The total field strength is computed from the square root of the sum of the squares of the filtered field components averaged in 1° ´ 1° bins. This average total field was then found for overlapping 10° intervals for all data between 23° and 29° N and between 23° and 29° S. The error bars are the most probable errors in the means. The bars on the occurrence rates indicate the variation in rate corresponding to a change of one observation from occurrence to nonoccurrence and vice versa. Field strengths of 0.1 g or less probably represent the noise level of the mapping technique.
|Fig. 15. The limb compression occurrence rate and lunar magnetic field strength at 100 km for the two regions where limb compression statistics overlap the field maps computed from data obtained in the geomagnetic tail. The error bars in the top panels represent the change due to an error in one identification. The error bars in the bottom panels are the probable errors in the mean.|
Examining first the two left-hand panels, we see that there were never any limb compressions in this region and that the field strength is at background. On the other hand, in the right-hand panels we see a sizeable magnetic field rising well above background and then returning to it at the same time the limb compression rate increases. The limb compression rate rises again, however, even though the field drops, probably because of the lessening coverage of the limb compression track with mapped fields. The orbit track turns equatorward at roughly 180° and completely misses the limb compression region by 200° E. To check this possibility, we recomputed the limb compression occurrence rate from 23° to 25° S, which more closely corresponded to the field map from 180° to 200° . This is shown as the dashed line in Figure 15. It more closely follows the trend in the field.
We note that the peak in the limb compressions occurs 10° ± 5° to the east of the peak of the field. Since the assigned position of each limb compression is the terminator and since Figure 14 shows that the compressions in this region occur 40 ahead of (to the west of) the terminator, the limb compressions have their maxima on the average 6° ± 5° to the east (downstream) from the peak surface field. It is improbable that such excellent agreement is a coincidence. These results in themselves do not prove that limb compressions are caused by deflection of the solar wind at the limbs by the lunar surface magnetic field. However, there is no other mechanism known to the authors that can account for this correlation.
Finally, we note that these results combined with those of Figures 12 and 14 suggest that the A region has probably stronger and more extensive magnetization than the Van de Graaff region. Further, since the peak compression strength occurs 14° ahead of the terminator, the source of this magnetism lies not over the Apollo crater but closer to the Van de Graaff region, probably in the neighborhood of the crater Rumford at 190° E, only 20° to the east of Van de Graaff.
Limb compressions occur 23% of the time in the Apollo 15 subsatellite data. However, this occurrence is far from random, being controlled by the appearance of certain selenographic regions near the lunar terminators beneath the subsatellite orbit. Over some regions, limb compressions were never observed. Regions of nonoccurrence were both in mare and in, highland material, but all regions of frequent occurrence were over highland material. The source strengths of the limb compressions, as measured by their average amplitude, and the position of occurrence relative to the limb differed markedly from region to region. Finally, in the regions where comparison is possible, the occurrence rate of limb compressions correlates well with the surface field strength measured when the moon is in the geomagnetic tail. This suggests that limb compressions are simply due to the deflection of the solar wind by the lunar surface field near the lunar terminators.
Acknowledgments. The authors wish to thank G. L. Siscoe, B. E. Goldstein, C. P. Sonett and their coinvestigators P. J. Coleman, Jr., and G. Schubert for many useful discussions. This work was supported by the National Aeronautics and Space Administration under NASA grant NGR 05-007-351. Institute of Geophysics and Planetary Physics publication 1099-15.
The Editor thanks D. S. Colburn and Y.-C. Whang for their assistance in evaluating this paper.
Barnes, A., P. M. Cassen, J. D. Mihalov, and A. Eviatar, Permanent lunar surface magnetism and its deflection of the solar wind, Science, 172, 716-718, 1971.
Beers, B. L., Numerical calculation of the lunar wake in an MHD model, Phys. Fluids, 15,1450, 1972.
Catto, P. J., A model for the steady-state interaction of the solar wind with the moon, Astrophys. Space Sci., 26, 47, 1974.
Colburn, D. S., R. G. Currie, J. D. Mihalov, and C. P. Sonett, Diamagnetic solar-wind cavity discovered behind moon, Science, 158, 1040, 1967.
Colburn, D. S., J. D. Mihalov, and C. P. Sonett, Magnetic observations of the lunar cavity, J. Geophys. Res., 76, 2940, 1971.
Coleman, P. J., Jr., G. Schubert, C. T. Russell, and L. R. Sharp, The particles and fields subsatellite magnetometer experiment, Apollo 15 Preliminary Science Report, NASA Spec. Publ. SP-289, 22-1, 1972a.
Coleman, P. J., Jr., G. Schubert, C. T. Russell, and L. R. Sharp, Satellite measurements of the moon's magnetic field: A preliminary report, Moon, 4, 419, 1972b.
Criswell, D. R., Photoelectrons and solar wind/lunar limb interaction, Moon, 7, 202, 1973a.
Criswell, D. R., Photoelectrons and lunar limb shocks, in Photon and Particle Interactions With Surfaces in Space, edited by R. J. L. Grard, p. 443, D. Reidel, Dordrecht, Netherlands, 1973b.
Dryer, M., Hypothesis for absence of lunar bow and wake shock waves, J. Geophys. Res., 73, 3583, 1968.
Gold, T., The magnetosphere of the moon, in The Solar Wind, edited by R. J. Mackin and M. Neugebauer, p. 381, Pergamon, New York, 1966.
Hollweg, J. V., Interaction of the solar wind with the moon and formation of a lunar limb shock, J. Geophys. Res., 73, 7269, 1968.
Hollweg, J. V., Lunar conducting islands and formation of a lunar limb shock wave, J. Geophys. Res., 75,1209, 1970.
Horning, B. L., and G. Schubert, Steady state asymmetric planetary electrical induction, J. Geophys. Res., 79, 5077, 1974.
Johnson, F. S., and J. E. Midgley, Notes on the lunar magnetosphere, J. Geophys. Res., 73, 1523, 1968.
Lichtenstein, B. R., C. T. Russell, and P. J. Coleman, Jr., Magnetic measurements of the solar wind interaction with the moon, in Proceedings of Symposium on Photon and Particle Interactions With Surfaces in Space, edited by R. J. L. Grard, p. 471, D. Reidel, Dordrecht, Netherlands, 1973.
Lyon, E. F., H. S. Bridge, and J. H. Binsack, Explorer 35 plasma measurements in the vicinity of the moon, J. Geophys. Res., 72, 6113, 1967.
Michel, F. C., Shock wave trailing the moon, J. Geophys. Res., 72, 5508,1967.
Michel, F. C., Magnetic field structure behind the moon, J. Geophys. Res., 73, 1533, 1968a.
Michel, F. C., Lunar wake at large distances, J. Geophys. Res., 73, 7277, 1968b.
Mihalov, J. D., C. P. Sonett, J. H. Binsack, and M. D. Moutsoulos, Possible fossil lunar magnetism inferred from satellite data, Science, 171, 892, 1971.
Ness, N. F., K. W. Behannon, C. S. Scearce, and S. C. Cantarano, Early results from the magnetic field experiment on lunar Explorer 35, J. Geophys. Res., 72, 5769, 1967.
Ness, N. F., K. W. Behannon, H. E. Taylor, and Y. C. Whang, Perturbations of the interplanetary magnetic field by the lunar wake, J. Geophys. Res., 73, 3421, 1968.
Pérez-de-Tejada H., On the continuum fluid approach to the solar wind-moon interaction problem, J. Geophys. Res., 78, 1711, 1973.
Russell, C. T., P. J. Coleman, Jr., B. R. Lichtenstein, G. Schubert, and L. R. Sharp, Subsatellite measurements of the lunar magnetic field, Proc. Lunar Sci. Conf. 4th, 3, 2833, 1973.
Russell, C. T., P. J. Coleman, Jr., B. R. Lichtenstein, G. Schubert, and L. R. Sharp, Apollo 15 and 16 subsatellite magnetometer measurements of the lunar magnetic field, Space Res., 14, 629, 1974.
Russell, C. T., P. J. Coleman, Jr., and G. Schubert, The lunar magnetic field, Space Res., 15, in press, 1975.
Schwartz, K., C. P. Sonett, and D. S. Colburn, Unipolar induction in the moon and a lunar limb shock mechanism, Moon, 1, 7, 1969.
Sharp, L. R., P. J. Coleman, Jr., B. R. Lichtenstein, C. T. Russell, and G. Schubert, Orbital mapping of the lunar magnetic field, Moon, 7, 322, 1973.
Siscoe, G. L., and B. Goldstein, Solar wind interaction with lunar magnetic fields, J. Geophys. Res., 78, 6741, 1973.
Siscoe, G. L., and N. R. Mukherjee, Upper limits on the lunar atmosphere determined from solar wind measurements, J. Geophys. Res., 77, 6042, 1972.
Siscoe, G. L., E. F. Lyon, J. H. Binsack, and H. S. Bridge, Experimental evidence for a detached lunar compression wave, J. Geophys. Res., 74, 59, 1969.
Sonett, C. P., and D. S. Colburn, Establishment of a lunar unipolar generator and associated shock and wake by the solar wind, Nature, 216, 340, 1967.
Sonett, C. P., and J. D. Mihalov, Lunar fossil magnetism and perturbation of the solar wind, J. Geophys. Res., 77, 588, 1972.
Spreiter, J. R., M. C. Marsh, and A. L. Summers, Hydromagnetic aspects of solar wind flow past the moon, Cosmic Electrodynamics, 1, 5, 1970.
Taylor, H. E., K. W. Behannon, and N. F. Ness, Measurements of the perturbed interplanetary magnetic field in the lunar wake, J. Geophys. Res., 73, 6723, 1968.
Tozer, D. C., and J. Wilson, The electrical conductivity of the moon's interior, Proc. Roy. Soc., Ser. A., 296, 320, 1967.
Wallis, M. K., Perturbation of the solar wind by the lunar atmosphere, J. Geophys. Res., 79, 275-279, 1974.
Wang, C. P., Interaction of solar wind with the moon and possibly other planetary bodies, AIAA J., 9, 1148-1153, 1971.
Whang, Y. C., Interaction of the magnetized solar wind with the moon, Phys. Fluids, 11, 969, 1968a.
Whang, Y. C., Theoretical study of the magnetic field in the lunar wake, Phys. Fluids, 11, 1713, 1968b.
Whang, Y. C., Field and plasma in the lunar wake, Phys. Rev., 186, 143, 1969.
Whang, Y. C., Two-dimensional guiding-center model for the solar wind-moon interaction, Solar Phys., 12, 489, 1970.
Whang, Y. C., and N. F. Ness, Observations and interpretation of the lunar Mach cone, J. Geophys. Res., 75, 6002, 1970.
Whang, Y. C., and N. F. Ness, Magnetic field anomalies in the lunar wake, J. Geophys. Res., 77, 1109, 1972.
Wolf, R. A., Solar wind flow behind the moon, J. Geophys. Res., 73, 4281, 1968.
(Received October 3, 1974; accepted August 5, 1975.)