R EPORTS The average of all known rotation periods is of the order of 10 (10). Radar and Optical Observations Among unambiguously determined rotation periods, the shortest period is 136 min for of Asteroid 1998 KY26 1566 Icarus (11). Asteroid 1998 KY26 would need a bulk density of the order of 39,000/P2 1 2 1 Steven J. Ostro, * Petr Pravec, Lance A. M. Benner, gcmϪ3 (12), or about 340 g cmϪ3, for it to R. Scott Hudson,3 Lenka S˘arounova´,2 Michael D. Hicks,1 consist of pieces held together just by their David L. Rabinowitz,1 James V. Scotti,4 David J. Tholen,5 mutual gravitational attraction. This aster- Marek Wolf,6 Raymond F. Jurgens,1 Michael L. Thomas,1 oid’s rapid spin thus reveals it to be a mono- Jon D. Giorgini,1 Paul W. Chodas,1 Donald K. Yeomans,1 lithic body bound by tensile strength alone (13). Randy Rose,1 Robert Frye,1 Keith D. Rosema,1 1 1 With the calculated period known, the Ron Winkler, Martin A. Slade echo spectra could be inverted (14) to esti- mate the asteroid’s shape. The spheroidal Observations of near- asteroid 1998 KY26 shortly after its discovery reveal model (Fig. 3) has a mean diameter of 26 a slightly elongated spheroid with a diameter of about 30 meters, a composition m/cos(␦). In the absence of prominent varia- analogous to carbonaceous chondritic meteorites, and a of 10.7 tions in echo bandwidth over a 54° sky arc, it minutes, which is an order of magnitude shorter than that measured for any is unlikely that any of our observations were other solar system object. The rotation is too rapid for 1998 KY26 to consist within a few tens of degrees of a pole-on of multiple components bound together just by their mutual gravitational view. The Doppler-based model and the attraction. This monolithic object probably is a fragment derived from cratering range-resolved data therefore bound 1998 Յ Յ or collisional destruction of a much larger asteroid. KY26’s effective diameter: 20 m Deff 40 m. The population of with a diameter During 6 through 8 June 1998, we ob- Whereas the asteroid’s relative topograph- range of 10 to 100 m is thought to number served the asteroid with the Goldstone X- ic relief is subdued, roughness at centimeter- ϳ107 in Earth-crossing orbits and ϳ109 in the band (8510 MHz, 3.5 cm) radar, using wave- to-decimeter scales is revealed by the ratio of main belt (1, 2). However, only a few dozen forms that provided various degrees of reso- echo power in the same sense of circular objects of several decameters size have been lution in time delay (range) and Doppler fre- polarization as transmitted (the SC sense) to discovered, and little is known about their spin quency (radial velocity). During 2 through 8 that in the opposite (OC) sense. A perfectly states, shapes, compositions, surface character- June, we observed the asteroid photometri- smooth surface would reflect echoes with istics, or interiors, much less their origins and cally with the Ondr˘ejov (Czech Republic) SC/OC ϭ 0. Asteroid 1998 KY26’s mean collisional histories. Spacecraft observa- 0.65-m telescope, the Steward 0.9-m Space- value of SC/OC, 0.5 Ϯ 0.1, exceeds 90% of tions of several much larger objects (3) help to watch telescope (Arizona), the Mauna Kea the values measured for near-Earth asteroids constrain the collisional evolution of those ob- 0.61-m telescope (Hawaii), and the Table (15). Because the asteroid is spinning too fast jects and the asteroid population as a whole (4), Mountain 0.6-m telescope (California), ob- to retain any loose particulate material (a but even for those objects, definitive informa- taining CCD (charge-coupled device) light- regolith) except perhaps near the poles, most tion about interior configuration (and hence curves and broadband colors (7). of its surface is exposed bare rock that has collisional history) is lacking: Are they co- The asteroid’s rapid spin rate was re- been roughened, probably at least in part by herent solid rocks or gravitationally bound, vealed by extreme Doppler broadening of bombardment (16). multicomponent agglomerates? spatially resolved radar echoes and shortly Our yields an estimate of Here, we report observations that provide thereafter was measured precisely by Fourier 1998 KY26’s mean, absolute V magnitude, a detailed look at a solar system object small- analysis of time series formed from disk- H ϭ 25.5 Ϯ 0.3 (17). That estimate and our er than 100 m. The discovery of 1998 KY26 integrated optical brightness measurements. Deff interval bound the asteroid’s visual albe- Յ Յ was announced on 1 June 1998 (5), 1 week The echoes’ bandwidth B (in hertz) satisfies do (18): 0.05 pv 0.37. Cluster analysis of ϫ ϭ ␦ before it passed 2.10 lunar distances (8.061 B 5.945 D cos( )/P, where P (in minutes) values of pv and visible to near-infrared col- 105 km) from Earth. Part of a recently recog- is the instantaneous apparent spin period, D ors of hundreds of asteroids has defined tax- nized, potentially abundant subpopulation of (in meters) is the width of the plane-of-sky onomic classes (19) that may correspond to small near-Earth asteroids (NEAs) (1), it is projection of the asteroid’s pole-on silhou- different mineralogical compositions, most of more accessible to a spacecraft rendezvous ette, and ␦ is the subradar latitude. B is at which have meteoritic analogs. Our color than any of the other ϳ25,000 known aster- least 11 Hz for all our echo spectra (Fig. 1). indices (B-R ϭ 0.083 Ϯ 0.070, V-R ϭ oids with secure orbits (6). Waveforms with time-delay resolution of 125 0.058 Ϯ 0.055, and R-I ϭ 0.088 Ϯ 0.053, ns (19 m in range) produced echoes confined with solar colors subtracted) reveal a neutral 1Jet Propulsion Laboratory, California Institute of Tech- to a single range cell, placing an upper limit reflectance spectrum that excludes all classes nology, Pasadena, CA 91109–8099, USA. 2Astronomical of 19 m on the asteroid’s visible range extent except six (B, C, F, G, D, P) that are associ- Institute, Academy of Sciences of the Czech Republic, and an upper limit of 40 m on the asteroid’s ated with mixtures of carbonaceous material 3 CZ-25165 Ondr˘ejov, Czech Republic. School of Electri- physical range extent (8). Therefore, the radar and mafic silicates, and one (M) that corre- cal Engineering and Computer Science, Washington Յ State University, Pullman, WA 99164–2752, USA. 4Lu- data require P 22 min. Our spectra show no sponds to NiFe metal or to mixtures of metal nar and Planetary Laboratory, University of Arizona, prominent asymmetries or features and only and spectrally neutral silicates. Our pv inter- Tucson, AZ 85721, USA. 5Institute for Astronomy, Uni- subtle bandwidth variations, precluding more val encompasses all values for asteroids with versity of Hawaii, 2680 Woodlawn Drive, Honolulu, HI precise radar estimation of P. However, anal- unambiguous B, G, or M classifications and 96822, USA. 6Astronomical Institute, Charles University Prague, V Holes˘ovic˘ka´ch 2, CZ-18000 Prague, Czech ysis of our finest-time-resolution photometric about half of the values for asteroids with Republic. light curves (Fig. 2) yields an unambiguous unambiguous P, F, D, or C classifications. ϭ Ϯ *To whom correspondence should be addressed. E- estimate of the period, P 10.7015 0.0004 The asteroid’s composition is also con- mail: [email protected]..gov min (9). strained by (i) comparison of its radar reflec-

www.sciencemag.org SCIENCE VOL 285 23 JULY 1999 557 R EPORTS Ϯ ␲ 2 tivity with values for asteroids whose taxo- The asteroid’s OC radar cross section (25 10 section divided by Deff /4) in the OC and 2 ϭ ϩ nomic classes are known and (ii) implications m ), its SC/OC ratio, and our Deff interval total-power (T OC SC) polarizations: ␴ Յ ␴ Յ Յ ␴ Յ of its reflectivity for its surface bulk density. bound its radar albedo ˆ (that is, radar cross 0.012 ˆ OC 0.11 and 0.018 ˆ T 0.17. ␴ The means and standard deviations of ˆ T for radar-observed main-belt asteroids (20) are Ϯ ASTEROID 1998 KY26 0.10 0.06 for nine objects classified B, G, F, or P; 0.17 Ϯ 0.05 for eight C objects; and 1998 JUNE 8 0.32 Ϯ 0.12 for four M objects. Among NEAs Goldstone echo power spectra ␴ for which ˆ T has been measured, it equals 0.10 and 0.18 for the two C objects and 0.63 for the sole M object. Thus, 1998 KY26’s radar albedo 0° 30° 60° is consistent with B/C/F/G/D/P classification but not with M classification. ␴ For a smooth sphere, ˆ sc would equal zero ␴ and ˆ OC would equal R, the smooth-surface reflection coefficient, which for materials of asteroidal interest depends primarily on sur- face bulk density d. For a target with a non- spherical shape or with moderate surface roughness at scales much greater than the 90° 120° 150° ϭ␴ wavelength, one could write R ˆ OC/g, where plausible values of the backscatter gain g are between 1.0 and 1.5. With wavelength- scale roughness, some echo power would be shifted to the SC polarization, and only part of the OC power would arise from smooth- surface echoes. For 1998 KY26, such rough- ness is present and the shape is not perfectly 180° 210° 240° spherical. In this situation, the upper bound on the total-power albedo can be taken as an upper bound on R, and the corresponding value of d(R) can be taken as an upper bound on the surface’s average bulk density. The most widely used empirical relation for d(R) (21) predicts d(R ϭ 0.17) ϭ 2.8gcmϪ3. 12 Meteoritic analogs of M-class asteroids in- clude irons and enstatite chondrites, which 8 270° 300° 330° have mean specific gravities of 7.6 and 3.6 g Ϫ3 4 cm , respectively. C/B/G/F asteroids appear analogous to CI1/CM2/CM3 carbonaceous

Echo power 0 chondrites (22), which are primitive, un- (std. deviations) melted, volatile-rich samples of solar nebula 4 OC 30 20 10 010 20 30 condensates (23) and whose mean specific SC Ϫ3 Doppler frequency (Hz) gravities (2.2 to 2.9 g cm ) bracket our upper limit on d, supporting identification of Fig. 1. Our strongest echo spectra from June 8, sorted into rotation-phase bins 30° wide. Our 1998 KY26 as carbonaceous chondritic. (D/P analyses used 15° bins. Each spectrum is a one-dimensional image equivalent to a scan of radar brightness measured through a slit parallel to the asteroid’s projected apparent pole. The frequency asteroids lack meteoritic analogs and may resolution, 1 Hz, corresponds to a slit width of 1.8 m/cos ␦, with ␦ the subradar latitude. The resemble .) rotation phase origin is at 1998 June 8.2600. Our optical and radar measurements reveal

Fig. 2. Composite light-curve constructed for P ϭ 10.7015 min from clear-filter photometry obtained at Ondr˘ejov on 26.4 June 2.0 (ϫ) and June 3.0 (ϩ) and at on June 5.3 (F). Brightness is in visual ( V) magnitudes, scaled to corre- spond to geocentric and heliocentric distances of 1 AU and an Earth-asteroid-sun angle of 28.1°. The rotation phase origin is at 1998 June 5.3045. The noise level of individual 26.7 measurements, about 0.1 magnitude, reflects the asteroid’s faintness and our short integration times (20 to 50 s). The

light-curve data are represented well by a fourth-order Fou- Brightness rier series (solid curve). 27.0

27.3 0 0.2 0.4 0.6 0.8 1.0 Rotational phase

558 23 JULY 1999 VOL 285 SCIENCE www.sciencemag.org R EPORTS Fig. 3. Asteroid 1998 KY26’s 97.2 or 145.8 min [D. I. Steel, R. H. McNaught, G. J. shape estimated from least- Garradd, D. J. Asher, A. D. Taylor, Planet. Space Sci. squares fitting of a 124-ver- 45, 1091 (1997)]. tex polyhedral model to the 12. A. W. Harris, Lunar Planet. Sci. 27, 493 (1996). 13. The minimum required tensile strength T for an echo spectra. The effective object of bulk density ␳ is of the order of ␲2␳D 2/P2 resolution of the model is 3 [D. L. Turcotte and G. Schubert, Geodynamics Appli- m. The model solution de- cations of Continuum Physics to Geological Problems pends on the subradar lati- (Wiley, New York, 1982), pp. 85–86, 204]. For this tude ␦, which is not known asteroid, T is less than 104 dyne cm–2, which is Ͻ10–4 but probably was at least a times values for rocks [ J. Suppe, Principles of Struc- few tens of degrees from tural Geology (Prentice-Hall, Englewood Cliffs, NJ, the pole. This figure shows 1985), p. 155]. Thus, 1998 KY26 could be held to- gether by very meager bonds. the solution for an equatori- ␦ϭ 14. D. L. Mitchell, R. S. Hudson, S. J. Ostro, K. D. Rosema, al view ( 0). The frac- Icarus 131, 4 (1998). tional topographic relief is 15. L. A. M. Benner et al., ibid. 137, 247 (1999); L. A. M. fairly insensitive to ␦. The Benner et al., ibid. 130, 296 (1997). model’s mean diameter is 16. For nominal relative velocities of 5 to 10 km s–1, milli- 26 m/cos␦. Principal axes of meter-sized (ϳ0.01 g) would create pits inertia are labeled x, y, and z with diameters of the order of 1 cm, probably surround- (the pole). ed by a spall zone in which rock has been shattered or removed, or both [D. E. Gault, F. Horz, J. B. Hartung, Proc. Third Lunar Sci. Conf., Suppl. 3,inGeochim. Cosmochim. Acta 3, 2713 (1972)]. The flux of 1-mm meteoroids, of the order of 10–12 m–2 s–1 in the inner solar system [E. Grun, H. A. Zook, H. Fechtig, R. H. Giese, Icarus 62, 244 (1985)] but perhaps an order of magni- tude higher in the main belt [D. S. McKay, T. D. Swindle, R. Greenberg, in Asteroids II, R. P. Binzel, T. Gehrels, M. S. Matthews, Eds. (Univ. of Arizona Press, Tucson, AZ, 1989), pp. 617–642], would have roughened much of the asteroid in 107 years. 17. We conservatively assume a slope parameter, G ϭ 0.15 Ϯ 0.2, for the asteroid’s brightness as a function of solar phase angle. ϭ 18. We use the relation: log pv 12.247–2logDeff – 0.4 H, where Deff is in meters [E. F. Tedesco, G. J. Veeder, an unelongated, monolithic, several-decameter- References and Notes J. W. Fowler, J. R. Chillemi, Technical Report PL-TR- wide, carbonaceous-chondritic body in a rapid 1. D. L. Rabinowitz et al., Nature 363, 704 (1993). 92-2049 (Phillips Laboratory, Hanscom AFB, MA, 1992)]. 2. D. L. Rabinowitz, E. Bowell, E. Shoemaker, K. Mui- spin state. This combination of characteristics is 19. D. J. Tholen and M. A. Barucci, in Asteroids II,R.P. nonen, in Hazards Due to and Asteroids,T. Binzel, T. Gehrels, M. S. Matthews, Eds. (Univ. of unique among known objects and sets bound- Gehrels, Ed. (Univ. of Arizona Press, Tucson, AZ, Arizona Press, Tucson, AZ, 1989), pp. 298–315. ary conditions on theories for the collisional 1994), pp. 285–312. 20. C. Magri et al., Icarus, in press. 3. Spacecraft flybys have imaged four asteroids, all with and rotational evolution of individual asteroids 21. J. B. Garvin, J. W. Head, G. H. Pettengill, S. H. Zisk, J. average overall dimensions between 14 and 58 km. and the population as a whole. Asteroids the Geophys. Res. 90, 6859 (1985). 4. R. Greenberg, M. C. Nolan, W. F. Bottke, R. A. Kol- 22. D. T. Britt and L. A. Lebofsky, in Encyclopedia of the size of 1998 KY26 are expected to have life- voord, J. Veverka, Icarus 107, 84 (1994); R. Greenberg Solar System, P. R. Weissman, L.-A. McFadden, T. V. times against catastrophic disruption of 107 to et al., ibid. 120, 106 (1996); J. Veverka et al., Science Johnson, Eds. (Academic Press, San Diego, CA, 1999), 8 278, 2109 (1997). 10 years (24); carbonaceous chondrites are the pp. 585–605. 5. G. V. Williams, Electron. Circ. 1998-L02 23. J. Lewis, Physics and Chemistry of the Solar System weakest meteorites, so the short end of that (1998). (available at http://cfa-www.harvard.edu/iau/ interval probably applies here. Therefore, this (Academic Press, San Diego, CA, 1997). services/MPEC.html) 24. P. Farinella, D. Vokrouhlicky´, W. K. Hartmann, Icarus object may be a nonprimordial collision frag- 6. Using simple expressions [E. M. Shoemaker and E. F. 132, 378 (1998); E. V. Ryan and H. J. Melosh, ibid. ment derived from cratering or destruction Helin, in Asteroids: An Exploration Assessment,D. 133, 1 (1998). Morrison and W. C. Wells, Eds. (NASA Conf. Publ. of a larger asteroid (25). Many asteroids 25. It has been suggested that tidal disruption of rubble- 2053, Washington, DC, 1978), pp. 245–256] for the pile asteroids during close planetary encounters larger than a few hundred meters are accelerations needed to leave low-Earth orbit and might generate large numbers of several-decameter- rendezvous with an asteroid, we find that for 1998 thought to be porous, nearly strengthless ϭ ϭ ϭ sized objects [W. F. Bottke Jr., D. C. Richardson, P. KY26 (a 1.23 AU, e 0.20, i 1.5°), the total Michel, S. G. Love, Planet. Space Sci. 46, 311 (1998)], “rubble piles” (26). The size distribution of required change in velocity (⌬V) is 4.2 km s–1. A lunar –1 but such a process seems unlikely to produce ultra- monolithic subunits in rubble-pile asteroids rendezvous requires 6 km s . rapid fragment rotation. is unknown, but the existence of 1998 7. We used B, V, R, and I color filters of the Johnson- 26. S. G. Love and T. J. Ahrens, Icarus 124, 141 (1996). KY26 suggests that they extend up to sizes Cousins system, with respective effective wave- 27. A. Fujiwara and A. Tsukamoto, ibid. 48, 329 (1981). lengths of 440, 550, 650, and 800 nm [M. S. Bessell, 28. Other laboratory experiments indicate that collision- of at least several decameters. Publ. Astron. Soc. Pac. 102, 1181 (1990)]. ally ejected fragments initially possess excited (non– How did this asteroid’s fast spin originate? 8. The physical range extent typically is about twice the principal axis) spin states [I. Giblin and P. Farinella, Laboratory impact experiments (27) have visible range extent [R. S. Hudson and S. J. Ostro, Icarus 127, 424 (1997)]. Our data show no evidence Science 270, 84 (1995); A. K. Andrews, R. S. Hudson, shown that rapid rotations are common among for non–principal axis rotation, but this is not surpris- D. Psaltis, Opt. Lett. 20, 2327 (1995)]. ing given that the time scale for rotational relaxation spall fragments thrown out from the surface 9. Asteroid 1998 KY26’s period was determined unam- of an object with 1998 KY26’s size and spin period layer surrounding the impact site and that these biguously in independent data from three different [A. W. Harris, Icarus 107, 209 (1994)] is only of the 5 ϳ ϫ 7 fragments acquire their spins in the impact- stations (Ondr˘ejov, Table Mountain, Spacewatch). order of 10 years, much less than the 3 10 year Observational techniques and data reduction proce- dynamical lifetime of objects in Earth-crossing orbits. generated shear field (28). However, the appli- dures were similar to those described by P. Pravec, L. 29. E. Asphaug et al., Nature 393, 437 (1998). cability of experiments on basaltic centimeter- S˘arounova´, and M. Wolf [Icarus 124, 471 (1996)] and 30. Part of this research was conducted at the Jet Pro- sized targets to carbonaceous chondritic targets by D. Rabinowitz [ibid. 134, 342 (1998)], and used an pulsion Laboratory, California Institute of Technolo- implementation of the Fourier method developed by gy, under contract with NASA. Work at Washington 10,000 times larger is not known. Novel ap- A. W. Harris et al. [ibid. 77, 171 (1989)]. State University was supported in part by NASA’s proaches to computer simulation of collisions 10. Rotation periods of near-Earth asteroids and similar- Planetary Geology and Geophysics Program. Work at (29) may shed light on the source of 1998 sized, main-belt asteroids average about 12 and 7 Ondr˘ejov was supported by the Grant Agency of the hours, respectively [D. F. Lupishko and M. DiMartino, Academy of Sciences of the Czech Republic. KY26’s rotation and on how unique it might be Planet. Space Sci. 46, 47 (1998)]. among similar-sized collision fragments. 11. The period of 1995 HM has been suggested to be 13 May 1999; accepted 14 June 1999

www.sciencemag.org SCIENCE VOL 285 23 JULY 1999 559