Beyond a few tens of solar radii, the solar wind outflow becomes faster than the Alfven speed (a characteristic speed in a plasma) and the solar wind and frozen-in magnetic field can never contract back to the Sun. In addition, while low-altitude arcades are often observed below expanding CMEs, coronagraph observations of the high-altitude portions of the corona virtually never show evidence of magnetic structures contracting back toward the Sun. As a consequence, the only reasonable method of reducing the IMF magnitude in interplanetary space seems to be via reconnection between oppositely directed, previously open field lines [ McComas et al., 1989, 1991; McComas, 1994]. The top left inset in Figure 4 schematically displays such a sequence of events. Compression of the oppositely directed magnetic field regions across a helmet streamer ultimately leads to reconnection between previously open field lines. The topological effect is to 1) create closed field loops that can return to the Sun, and 2) release disconnected U-shaped field structures (D in Figure 2) into interplanetary space. An example of such a coronal disconnection event was shown by McComas et al. [1991] using Solar Maximum Mission coronagraph images from 1 June 1989. In that sequence of events a large coronal streamer appears to disconnect, creating the appearance of bright loops at lower altitudes and a large, disconnected U-shaped structure above it. Subsequent images show the U-shaped structure accelerating outward from the corona, leaving behind an arcade of bright loops indicative of the newly returned magnetic field lines.
The remainder of Figure 4 shows evidence for a possible coronal disconnection as far back as the 16 April 1893 solar eclipse. This figure, adapted from Cliver et al. [1989] by McComas [1994], shows sketches (data in 1893) made in sequence from Chile, Brazil, and Senegal, indicating the outward motion of a large U-shaped structure. While originally interpreted as a comet, Cliver et al. [1989] reinterpreted this sequence as evidence of a fully detached CME (e.g., a plasmoid). While this interpretation may be correct, the sketches do not display any leading edge of a CME. Rather, the observations suggest that a coronal disconnection event occurred.
A statistical study of three months of Solar Maximum Mission (SMM) coronagraph observations [ Hundhausen, 1993] was conducted to assess how common coronal disconnection events are [ McComas et al., 1992c]. The first three months of 1988, in which an initial SMM survey [ St. Cyr and Burkepile, 1990] had found no obvious disconnections, was resurveyed specifically to look for such events. Of the 53 transient events during this interval, 6 (11%) showed some evidence of disconnection in more than one frame and 13 (23%) showed a single frame with an outward ``U'' or ``V'' structure. While these latter structures may not be significant in light of the fact that there are ambiguities in interpreting images from optically thin media, McComas et al. [1992c] concluded that magnetic disconnection events on previously open field lines (above helmet streamers) may be far more common than previously appreciated.
An alternate location for some disconnection events is on the
back side of rising CMEs. The only previous published example of
this was provided by Illing and Hundhausen [1983].
Recently, however, Webb and Cliver [1994, submitted
manuscript] examined CME observations for evidence of
disconnection closely following the rise of the expanding CME
loops. These authors find that 
10% of all CMEs they examined
show some evidence of disconnection. This number is certainly
consistent with the 
30% of all CMEs in interplanetary space
that show magnetic cloud- or flux rope-like rotations suggestive
of reconnection as shown in Figure 3 and at least partial
disconnection from the Sun. It seems highly likely that both
processes, partial disconnection of CMEs and reconnection on
previously open field lines above helmet streamers, occur
regularly in the corona.
Impulsive, low-energy (<10 keV) solar electron events called ``high coronal flares'' [ Cliver and Kahler, 1991] provide another independent line of evidence for magnetic reconnection above helmet streamers. These electrons are inferred to have been accelerated in the corona at heights above the solar surface >0.5 solar radii [ Potter et al. 1980]. Reconnection above helmet streamers is the most likely location for such high-altitude reconnection [ Cliver and Kahler, 1991]. Lin [1985] found that such events occur roughly once every three days, which suggests that the formation of detached magnetic structures may be a relatively common feature of the solar corona.
As discussed above, energetic electrons streaming along the local IMF provide good indications of the magnetic topology in interplanetary space. Since the normal unidirectional heat flux indicates simple magnetic connection to the Sun along open field lines and counterstreaming halo electron events indicate closed magnetic structures, intervals devoid of these energetic electrons and the electron heat flux they carry would be logically interpreted as magnetic structures that are disconnected from the Sun and open to the outer heliosphere at both ends (D in Figure 2) [ McComas et al., 1989]. This is true because the hot halo electrons run outward along the field far faster than the disconnected structure moves outward with the bulk solar wind.
Intervals devoid of any beaming halo electrons, coined ``heat flux dropouts'' or HFDs, have been found using data from the International Sun-Earth Explorer (ISEE)-3 solar wind electron experiment [ McComas et al., 1989]. While HFDs were identified purely on the basis of the electron distributions, they correlate extremely well with field reversals of the interplanetary magnetic field (current sheet crossings) in general, and crossings of the heliospheric current sheet (sector boundary crossings), in particular [ McComas et al., 1989; 1992d]. This association is naturally explained if HFDs arise from reconnection between oppositely directed fields above helmet streamers, at the base of the heliospheric current sheet.
In an independent analysis of HFDs, Lin and Kahler [1992] examined the electron distributions at higher energies (2-8.5 keV); higher energy particles are superior tracers of magnetic topologies compared to 0.1-1 keV halo electrons, whenever they are present. These authors found that of the 25 events on the original HFD list [ McComas et al., 1989], nine exhibited strong unidirectional streaming at these higher energies, indicating that the field lines were probably still connected to the Sun. A weak unidirectional streaming was found for 13 of 25 of the identified HFDs, leaving in doubt the connectivity of these events. Finally, two of the cases exhibited clear dropouts even at these higher energies, indicating that at least these events were probably real disconnections.
In addition to the topological explanation for HFDs, enhanced coulomb collisions have been suggested as another possible explanation for HFDs [ McComas et al., 1989; Lin and Kahler, 1992]. While scattering may account for some HFDs identified with the lower energy electrons, Lin and Kahler [1992] showed that it cannot account for all of the HFD observations. Finally, the possibility of reconnection out in interplanetary space, rather than down in the corona, was examined as another possible source of HFDs [ McComas et al., 1994]. These authors showed a case study where reconnection ahead of a faster moving CME driver appears to have converted normal open field lines into a pair of structures: a tongue and a disconnected U-shaped structure, indicated by an HFD. It is worth noting, however, that this sort of reconnection in interplanetary space can not in any way affect the balance of opening and closing of field lines. HFDs appear not to be unique indicators of coronal disconnection events and the consequent return of magnetic flux to the Sun since both reconnection out in interplanetary space and other processes (possibly enhanced collisions) can also create HFD signatures.
The only suggested method that I am aware of for reducing the magnetic field magnitude in interplanetary space is through reconnection between oppositely directed open field lines, inside of the distance where the solar wind becomes super-Alfvenic, most likely at the base of the heliospheric current sheet above helmet streamers. This process returns previously open magnetic lines to the Sun as closed arches and can also account for the observations of 1) coronal disconnection events, 2) high coronal flares, and 3) some HFDs.