The appearance of the comet was reported as most unusual; the object appeared as a dense, linear bar about 1 arc minute long and had a fainter, wispy tail. (A circle is divided into 360 degrees, each degree into 60 minutes, and each minute into 60 seconds. The word arc is added to denote an angular measure rather than time. The diameter of the Moon is near 30 arc minutes, for example, while the apparent diameter of Jupiter when closest to Earth is 50 arc seconds.) The comet's brightness was reported as about magnitude 14, more than a thousand times too faint to be seen with the naked eye.
The existence of this object was soon confirmed by James V. Scotti of the Spacewatch program at the University of Arizona. The International Astronomical Union's Central Bureau for Astronomical Telegrams immediately issued Circular No. 5725 reporting the discovery as a new comet, giving it the provisional designation of 1993e (the fifth comet discovered or recovered in 1993). Scotti reported at least five condensations in a long, narrow train about 47 arc seconds in length and about 11 arc seconds in width. He also discovered dust trails extending 4.20 arc minutes to the east and 6.89 arc minutes to the west and tails extending about 1 arc minute from elements of the nuclear train. Bureau director Brian G. Marsden noted that the comet was some 4 arc minutes from Jupiter and that its motion suggested that it could be near Jupiter's distance from the Sun.
By March 27, Marsden had enough positions to attempt to derive possible orbits. One elliptical solution gave a close approach to Jupiter in July 1992. Also on March 27, Jane Luu and David Jewitt took an image with the 2.2-m telescope on Mauna Kea in Hawaii that showed as many as 17 separate sub-nuclei strung out like pearls on a string 50 arc seconds long. Figure 4 shows an early image taken by Scotti on March 30, 1993. This long exposure (440 seconds on a CCD detector) brings out the faint detail of the debris field, though it overexposes the individual nucleus fragments. Figure 5 is an image from the Hubble Space Telescope (HST), taken by Harold A. Weaver and collaborators on July 1, 1993 (before the HST repair mission), that clearly shows at least 15 individual fragments in one image frame of the train.
In IAU Circular No. 5744, dated April 3, 1993, Marsden used positions covering a period of 17 days (including two prediscovery positions from March 15) and was able to report that no orbit of very long period (near parabolic) was possible. The orbit had to be an ellipse of rather small eccentricity relative to the Sun and relatively short period. Because it was not at all obvious where the center of mass of this new comet lay, most observers were just reporting the position of what appeared to be the center of the train. This made an accurate orbit (or orbits) difficult to determine. Marsden suggested that a very close approach to Jupiter in 1992 continued to be a distinct possibility. The orbit he chose to publish was one with the comet at least temporarily in orbit around Jupiter.
By May 22 Marsden had almost 200 positions of the center of the train. In Circular No. 5800 he reported on an orbit computed May 18 by Syuichi Nakano that showed the comet approaching within 120,000 kilometers (74,600 miles) of Jupiter on July 8, 1992, and approaching again, this time within 45,000 kilometers (28,000 miles) of the center of Jupiter, on July 25, 1994. Marsden noted that this distance was less than the radius of Jupiter. In other words, the comet, or at least parts of it, could very well hit Jupiter.
By October 18, 1993, Paul W. Chodas and Donald K. Yeomans were able to report at the annual American Astronomical Society's Division of Planetary Sciences meeting that the probability of impact for the major fragments of Shoemaker-Levy 9 was greater than 99%. The fragments apparently would hit over a period of several days, centered on July 21.2, on the night side of Jupiter at latitude 44° S and longitude 35° past the midnight meridian, according to available observations. Unfortunately, this is also the back side of Jupiter as viewed from Earth. The 1992 approach to Jupiter that disrupted the comet was calculated to have been at a distance of 113,000 kilometers (70,200 miles) from the planet's center and only 42,000 kilometers (26,100 miles) above its cloud tops. Furthermore, they found that the comet had been in a rapidly changing orbit around Jupiter for some time before this, probably for at least several decades. It did not fragment during earlier approaches to Jupiter, however, because these were at much greater distances than that of 1992.
After recovery of the comet on December 9, following the period during which it was too near to the Sun in the sky to observe, Chodas and Yeomans found that the probability was greater than 99.99% that all the large fragments would hit Jupiter. The encounter period was now centered on July 19.5, and orbits for individual fragments were uncertain by about 0.03 days (1 s). The impact site moved closer to the limb of Jupiter, near 75° from the midnight meridian and only a few degrees beyond the dark limb as seen from Earth; however, all pieces would still impact on the back side. The 1992 approach that split the comet is now calculated to have occurred on July 7.84 and only 25,000 kilometers (15,500 miles) above the clouds. These data now cover a much longer time base and are based upon calculations for individual fragments. They are unlikely to change significantly in the future. The comet probably approached Jupiter no nearer than about 9 million kilometers (5.6 million miles) in the orbit prior to that of 1992.
In a comprehensive paper prepared for The Astronomical Journal, Zdenek Sekanina, Chodas, and Yeomans report on the details of the breakup of Shoemaker-Levy 9 as calculated from the positions, motions, and brightness of the fragments and debris. They used data from Jewitt, Luu, and Chen taken in Hawaii, Scotti in Arizona, and Weaver's Hubble Space Telescope (HST) observing team. For example, the 11 brightest fragments as measured with the HST-- visual (V) magnitude 23.7-24.8 or about 15 million times too faint to be seen by the naked eye--had the brightness one would expect from spheres 4.3 down to 2.5 kilometers (1.6 miles) in diameter, assuming a normal cometary reflectivity for the fragments (about 4%). Of course, the fragments are not spheres, since tidal disruption tends to occur in planes perpendicular to the direction of the object causing the disruption (Jupiter), and since comets generally are not spherical to begin with. Nevertheless, adding up the sizes of these 11 fragments, the other fragments not precisely measured, and all of the debris making up the trails and tails, suggests that the original comet must have been at least 9 kilometers (5.6 miles) in average diameter, and it could have been somewhat larger. This was a good-sized comet, about the same size as Comet Halley.
When comets split, the pieces do not go flying apart at a high velocity, each to immediately go into its own independent orbit. The escape velocity from a non-rotating spherical comet 5 km in radius with a density of 0.5 g/cm^3 (half that of water) is 2.65 meters/second (6.5 mph). If suddenly freed of gravity and molecular bonds, a particle at the equator of that 10-km body, assuming a rotation period of 12 hours, would depart with a velocity of only 0.72 m/s (1.6 mph) relative to the center of the comet. Some comets appear to rotate more rapidly than once per half day, while many, such as Halley, rotate more slowly. In any case the centrifugal force on unattached pieces of material lying on the surface of a rotating comet is not normally sufficient to overcome the gravity holding them there. Pieces do not fly off of the nucleus spontaneously. Even when the tidal forces overcome self-gravity, the pieces separate slowly and they interact gravitationally. More important, the pieces bang into one another, changing their velocities and perhaps fragmenting further.
In the case of Shoemaker-Levy 9, Sekanina, Chodas, and Yeomans estimate that although fragmentation probably began before closest approach to Jupiter, dynamic independence of the pieces didn't occur until almost two hours after closest approach. For a period of at least two to three hours, collisions dominated the dynamics of all but the largest pieces, with each small grain suffering some 10 collisions per second and the bigger pieces being subjected to many times this number of low-velocity impacts by the small particles. All of this converted the original rotational velocities of the bits and pieces of 0 to 3 kilometers per hour (0-2 miles per hour) into a random "equilibrium" velocity distribution, with some smaller pieces having velocities several times their original velocity. Once the pieces stopped hitting one another, each continued to move in its own independent orbit determined mainly by the gravity of Jupiter and the Sun. The pressure of light from the Sun also had a significant effect upon the smallest particles, creating a broad dust tail just as happens in a normal comet. There has been no evidence of the presence of gases from Shoemaker-Levy 9, either direct spectroscopic evidence or motion of the dust particles that cannot otherwise be explained. This is not to say that there are no gases, only that there is no evidence for them. The only direct evidence we have that Shoemaker-Levy 9 is really a comet and not an asteroid is the fact that it broke up so easily! Asteroids are not thought to be so fragile.
It is unlikely that the exact circumstances of the breakup of Shoemaker-Levy 9 will ever be known with certainty. However, the physical model needed to reproduce the train (of individual large fragments), the trails (of debris on either side of the train), and the tails (of very small particles in the anti-Sun direction) in many images like those shown in Figures 4 and 5 does set limits on the separation time, sizes, and velocities of the pieces and particles making up each element. The model of Sekanina, Chodas, and Yeomans, the most complete at this writing, suggests that the original comet cannot have been much smaller than 9 kilometers (5.6 miles) in mean diameter, that it probably was rotating quite rapidly (perhaps once in eight hours), and that the breakup, as defined by dynamical independence from collisions and limited mutual gravitational effects, was not completed until about two hours after the closest approach to Jupiter. The comet nucleus was probably not very spherical or the debris trails on either side of the train of nucleus fragments would be nearly equal in length, which they are not. After the collisions ceased, the motion of the largest fragments was dominated by Jupiter, with those fragments closest to Jupiter at breakup remaining closest and therefore moving with a shorter period in accordance with basic mechanics. The fragment that started nearest to Jupiter will be the first to return to Jupiter and hit the planet.
All of the large fragments were soon strung out in nearly a straight line that pointed at Jupiter, and they remained so until they collided with the planet. H. J. Melosh and P. Schenk have offered the intriguing suggestion that linear chains of craters observed on Jupiter's satellites Ganymede and Callisto are the product of impacts by earlier comets fragmented by Jupiter.