We have identified an anti-sunward-directed, Petschek-type, reconnection exhaust during a crossing of the heliospheric current sheet, HCS, by the ACE spacecraft. The exhaust in this relatively rare HCS event was a region of accelerated plasma flow filling the field reversal region and was bounded by Alfven waves propagating in opposite directions along recently merged field lines. Heliospheric magnetic field lines disconnected from the Sun by the reconnection process were identified in and surrounding the exhaust by the disappearance of the solar wind electron strahl there and by the interpenetration along the field of sunward-streaming halo electrons from opposite sides of the exhaust. These observations demonstrate that strahl dropouts observed in the vicinity of the HCS at least at times are signatures of magnetic disconnection from the Sun.
 Magnetic reconnection describes a process in which pairs of magnetic field lines merge to produce topological changes in the field [see, e.g., Hones, 1984]. We have recently obtained direct evidence for local, quasi-stationary, magnetic reconnection in the solar wind using plasma and magnetic field data obtained by the Advanced Composition Explorer, ACE [Gosling et al., 2005]. The prime evidence consists of intervals of accelerated or decelerated plasma flow observed within magnetic field reversal regions that we interpret as encounters with (generic) Petschek-type reconnection exhausts [e.g., Petschek, 1964] that are bounded by Alfven waves. Compared to the surrounding solar wind, the exhausts are typically characterized by accelerated plasma flow, increases in proton density and (apparent) proton temperature, and decreases in magnetic field strength. Electron temperatures and magnetic field orientations within the exhausts typically are intermediate between values observed on opposite sides of the exhausts. All six reconnection exhausts identified in our initial survey of the ACE data occurred within interplanetary coronal mass ejections, ICMEs, at relatively abrupt shears (field rotations ranging from 98° to 162°) in the heliospheric magnetic field, HMF, separating regions of low proton beta and high Alfven speed. The apparent proton temperature increases within the exhausts were largely the result of counterstreaming solar wind proton beams of comparable density; together with the intermediate solar wind electron temperatures observed there, these counterstreaming ion beams demonstrated magnetic connection from one side of an exhaust to the other.
 We have now developed better techniques for recognizing reconnection exhausts in the ACE data set and as a result have identified 42 exhaust events in approximately 7 years of data. All of the newly identified exhausts had characteristics nominally similar to the 6 events previously described, the most distinguishing signature being accelerated or decelerated plasma flow within a field reversal region. Of these 42 events, at least 26 were associated with ICMEs, including 6 events that occurred at the leading edges of ICMEs. At least 12 of the events were encounters with sunward-directed exhausts. Only one event was associated with a crossing of the heliospheric current sheet, HCS, separating open field lines of opposite magnetic polarity. Our purpose here is to present and discuss detailed observations of this singular HCS event, which occurred on 17 September 1998 and which produced HMF lines disconnected from the Sun.
 Before proceeding to the observations we note that solar wind electron spectra almost ubiquitously comprise two populations with a breakpoint near ∼60 eV at 1 AU: 1) a marginally collisional thermal core population and 2) a nearly collisionless suprathermal population [e.g., Feldman et al., 1975; Phillips and Gosling, 1990]. Further, the suprathermal electrons can usually be split into two components: 1) a relatively intense beam known as the strahl [e.g., Rosenbauer et al., 1977] that is directed outward from the Sun along the HMF and that carries the solar wind electron heat flux; and 2) a roughly isotropic component that we call the halo. We have suggested that the halo electrons largely originate well beyond 1 AU, with the anti-sunward-directed portion of the halo resulting from mirroring of the sunward-directed portion at locations sunward of the observation point [Gosling et al., 2001]. Because of their nearly collisionless nature and relatively high speeds, the suprathermal electrons are effective tracers of HMF topology.
 ACE was launched 25 August 1997 into an orbit about the L1 Lagrange point ∼.01 AU upstream from Earth in the solar wind. For the period of interest here, the Los Alamos Solar Wind Electron Proton Alpha Monitor, SWEPAM, on ACE [McComas et al., 1998] provided 3-dimensional measurements of solar wind ion and suprathermal electron velocity distribution functions, f(v), every 64 s and 3-dimensional measurements of the thermal electron f(v) every 128 s. Measurements at any electron energy were acquired in one spacecraft spin of ∼12s. The overall plasma measurement cadence was barely sufficient to resolve the structure of the 17 September 1998 reconnection exhaust to be described here. However, the Magnetic Field Experiment, MAG, on ACE [Smith et al., 1998] provided measurements of the magnetic field vector, B, at a cadence of 24 vectors/s and adequately resolved the structure of the exhaust. Here we use 16 s averages of the magnetic field data in r, t, n coordinates where the +r-direction is radial outward from the Sun, the +t-direction is in the direction of Earth's motion about the Sun, and the +n-direction completes a right-handed system.
Figure 1 provides an overview of the 17 September 1998 HCS crossing. The electron strahl is the relatively intense beam that initially peaked at a pitch angle, PA, of 0° and later at 180°; the halo is the more isotropic distribution that spread across all PAs. The change in strahl flow polarity from 0° to 180° at ∼0319 UT on 17 September coincided with flips in the r and t components of the HMF and thus signaled a crossing of the HCS. The transverse component of the field, Bt, was dominant on both sides of the HCS whereas the normal field component, Bn, was very small. Therefore, B was nearly in the ecliptic plane oriented nearly transverse to the radial (from the Sun) direction, with the fields on opposite sides of the HCS being nearly anti-parallel.
Figures 2 and 3 provide expanded views of the HCS crossing. Figure 2 illustrates that: 1) the field rotation associated with the crossing occurred within an ∼176 s interval (from ∼03:17:33 to ∼03:20:29 UT) at a time of low flow speed; 2) the field reversal region separated plasmas with distinctly different flow velocities, V; 3) the field lingered at an intermediate field orientation (Bt ∼ +4.5 nT) approximately midway through the field reversal region; i.e., the HCS actually contained two separate current sheets bounding a region of intermediate field orientation; 4) accelerated plasma was observed within the field reversal region, the acceleration being primarily in the −t direction with a smaller acceleration in the +r direction; and 5) the changes in V and B were anti-correlated at the leading edge of the field reversal region but were largely correlated (particularly in the dominant t components) at the trailing edge of the region. Since Alfven waves propagating parallel (anti-parallel) to B produce anti-correlated (correlated) variations in B and V, the above indicate that the field reversal region was bounded by Alfven waves propagating away from the Sun along the HMF.
Figure 3 illustrates that: 1) the field magnitude dipped within the field reversal region, as is characteristic of reconnection exhausts in the solar wind previously discussed; 2) a heat flux decrease occurred near the time of entry into the field reversal region, persisted briefly after the reversal, and coincided with a decrease in the number density of suprathermal electrons directed outward away from the Sun along the HMF; and 3) the thermal electron temperature within the field reversal region and shortly thereafter was intermediate between temperatures observed on opposite sides of the region. Those intermediate electron temperatures suggest that the thermal electrons found within the field reversal layer interpenetrated into the layer from opposite sides along B; they thus constitute strong evidence for magnetic connection across the HCS.
Figure 4 shows the evolution of the suprathermal electron PA distributions, PADs, in the vicinity of the HCS. As is apparent also in Figure 1, the strahl was more strongly peaked and the halo was less intense before the HCS crossing than after. The strahl disappeared within the field reversal region and was considerably reduced in intensity for a couple of minutes immediately thereafter. That disappearance and reduction correspond to the decreases in suprathermal electron number density and electron heat flux obvious in Figure 3. A halo population was present within the HCS, but at 03:17:25 (near the time of entry into the field reversal region) and at 03:19:33 UT it was distinctly anisotropic, the PAD between 0° and 90° being essentially identical to that at those PAs after the HCS crossing and the PAD between 90° and 180° being essentially identical to that at those PAs prior to the crossing. The observations clearly indicate that the halo PADs in the immediate vicinity of the HCS layer resulted from the interpenetration along B of sunward-streaming halo electrons from opposite sides of the exhaust. Together, the combination of interpenetrating halo electrons and the absence of the strahl in the near vicinity of the HCS demonstrates that field lines within that region were not only connected to the surrounding solar wind on both sides but also were totally disconnected from the Sun.
3. Interpretation and Discussion
 The observations presented in the previous section can consistently be interpreted as resulting from ACE's encounter with a (generic) Petschek-type reconnection exhaust associated with a solar wind reconnection site that initially was positioned sunward and to the west (i.e., in the +t-direction) of ACE. Figure 5 provides an idealized sketch of the field line geometry associated with the encounter as well as sketches illustrating the evolution of the suprathermal electron PADs in the vicinity of the exhaust. Merging of open field lines creates closed field lines sunward and disconnected field lines anti-sunward of the reconnection site. Field line kinks associated with newly merged field lines propagate both parallel and anti-parallel to B into the surrounding solar wind at the respective Alfven speeds there. Dashed lines marked A1 and A2 in Figure 5, which pass through the kinks on different merged field lines in the anti-sunward exhaust, define the boundaries of that exhaust and the region of accelerated plasma flow. Anti-correlated changes in V and B should occur along A1 and correlated variations in V and B along A2, as observed in the 17 September 1998 event. Plasma entering the anti-sunward exhaust along merged field lines is accelerated primarily in the eastward (−t) direction, again as observed. When the Alfven Mach number of the solar wind flow is >2 (the usual situation in the solar wind near 1 AU and specifically so in the 17 September 1998 event), the reconnection site and the field and plasma in both exhausts are carried away from the Sun by the solar wind flow.
 As indicated in Figure 5, a spacecraft successively encounters: 1) field lines of away (from the Sun) polarity not yet merged at the reconnection site, 2) merged field lines of away polarity that lie outside the region of accelerated flow, 3) the accelerated flow region (i.e., the reconnection exhaust) containing the field polarity reversal and bounded by Alfven waves propagating parallel and anti-parallel to B respectively on merged field lines, 4) merged field lines of toward (the Sun) polarity that lie outside the exhaust region, and 5) field lines of toward polarity that have not yet merged. As shown by the PAD sketches in Figure 5, at points 1 and 5 a spacecraft should observe combinations of electron strahls flowing parallel (anti-parallel) to B and nearly isotropic halos, the sunward-directed portions of which are shown as cross-hatched. Consistent with the 17 September 1998 data, the halo intensity after the field reversal region is shown as being greater than that before the reversal. From points 2 to 4 the field lines are disconnected from the Sun so that the strahl is absent there (for simplicity we assume here that, upon merging, it disappears from those field lines instantaneously) and the halo is asymmetric, consisting of electrons that originally were all sunward-directed on opposite sides of the HCS, again essentially as observed in the 17 September 1998 event. We note that the “model” also accurately predicts the sense of the observed asymmetry. From the observation that the electron heat flux decrease extended beyond the boundaries of the exhaust, we surmise that the SWEPAM measurement cadence was sufficient to resolve the merged field lines lying outside the reconnection exhaust (in the so-called separatrix layer [e.g., Sonnerup et al., 1981]) on the trailing edge of the event and perhaps at the leading edge as well.
 The ACE observations of the 17 September 1998 crossing of the HCS provide strong evidence for reconnection in the solar wind at the HCS and aptly demonstrate the nature of suprathermal electron PADs on the disconnected HMF. We emphasize, however, the relative rarity of this event – it is the only crossing of the HCS separating open field lines that we are aware of in which ACE detected an accelerated plasma flow within the magnetic field reversal region. The HCS was also a relatively thin current sheet (∼5.3 × 104 km thick) separating relatively strong magnetic fields that were nearly anti-parallel to one another. (The above thickness corresponds to that of the reconnection exhaust where sampled by ACE; the HCS must have been considerably thinner at the actual reconnection site.) Usually the HCS transition as observed by ACE occurs on a time scale of tens of minutes and the fields on opposite sides are neither strong nor nearly anti-parallel.
 Reconnection in the solar wind near 1 AU usually does not affect the balance of magnetic flux in the heliosphere since both exhausts are carried away from the Sun by the super-Alfvenic flow. In order for reconnection to affect the overall flux balance it must occur in regions where the flow is sub-Alfvenic, e.g. close to the Sun. Reconnection near the base of the HCS in the corona should affect the heliospheric magnetic flux balance and the resulting disconnected field lines should also be detected far from the Sun as electron heat flux dropouts, HFDs, in which the strahl vanishes and the suprathermal electron number density decreases substantially. Indeed, HFDs of this nature are relatively common in the vicinity of the HCS near 1 AU and have been interpreted as possible evidence for magnetic disconnection from the Sun [e.g., McComas et al., 1989]. However, Lin and Kahler  showed that anti-sunward streaming, solar energetic electrons are frequently observed within HFDs, implying that many strahl disappearances are not associated with magnetic disconnection from the Sun (no solar energetic electrons were present during the 17 September 1998 event). Others have argued that HFDs are primarily the result of scattering en route from the Sun [e.g., Fitzenreiter and Ogilvie, 1992; Crooker et al., 2003; Pagel et al., 2005]. On the other hand, the present set of observations clearly demonstrates that HFDs in which the strahl disappears and the halo persists at least at times are signatures of magnetic disconnection from the Sun.
 In closing, we note that many of the reconnection exhausts observed by ACE, including the 17 September 1998 event, have also been detected by other spacecraft. We intend to use multi-spacecraft observations of these events to obtain information on the 3-dimensional nature of the exhausts and their overall spatial extents and to obtain estimates on the time intervals over which reconnection in these events is quasi-stationary.
 Work at Los Alamos was performed under the auspices of the U.S. Department of Energy with support from NASA as a part of the ACE program and an SR&T grant. Work at Southwest Research Institute and the University of New Hampshire was supported by the NASA/ACE program.