for the average cheilostome species to generate a progressively greater skeletal mass than the average cyclostome species. This could result from a gradual trend toward relatively larger colony sizes within cheilostomes (17 ), a greater number of colonies per cheilostome species, or both. These data suggest that multiple mea- sures of biotic change through time are necessary for a rich understanding of bio- spheric evolution. As in contemporary bi- otic communities, taxonomic diversity cap- tures only limited aspects of the complexity of the biota (18). Multiple measures of biotic systems provide greater insight into history and processes and a better basis for predicting future biodiversity. Reference and Notes 1. P. W. Signor, Annu. Rev. Ecol. Syst. 21, 509 (1990); J. J. Sepkoski Jr., Paleobiology 19, 43 (1993); in Global Events and Event Stratigraphy in the Phanerozoic, O. H. Walliser, Ed. (Springer-Verlag, Berlin, 1995), pp. 35–51; M. J. Benton, Science 268, 52 (1995). 2. J. J. Sepkoski Jr. and M. L. Hulver, in Phanerozoic Diversity Patterns, J. W. Valentine, Ed. (Princeton Univ. Press, Princeton, NJ, 1985), pp. 11–39. 3. W. I. Ausich and D. J. Bottjer, Science 216, 173 (1982); G. J. Vermeij, Evolution and Escalation (Prince- ton Univ. Press, Princeton, NJ, 1987); J. J. Sepkoski Jr., Paleobiology 14, 221 (1988); R. K. Bambach, ibid. 19, 372 (1993). 4. C. W. Thayer, in Biotic Interactions in Recent and Fossil Benthic Communities, M. J. S. Tevesz and P. O. McCall, Eds. (Plenum, New York, 1983), pp. 480 – 625. 5. Counts of individuals are inappropriate for skeletalized colonial organisms because of modularity and fragmen- tation, and the sizes of sexually mature bryozoan colo- nies range over several orders of magnitude. Therefore, skeletal mass was taken as an appropriate proxy for abundance or biomass [W. I. Ausich, Ohio J. Sci. 81, 268 (1981); G. Staff, E. N. Powell, R. J. Stanton Jr., H. Cum- mins, Lethaia 18, 209 (1985)]. We stress that this proxy is potentially subject to taphonomic bias and is strictly empirical. Competitive dominance in the once-living community requires different bases of inference. 6. S. J. McNaughton and L. L. Wolf, Science 167, 131 (1970); R. H. Whittaker, Communities and Ecosystems (Macmillan, New York, ed. 2, 1975); J. H. Brown and B. A. Maurer, Nature 324, 248 (1986); E. O. Wilson, The Diversity of Life (Belknap, Cambridge, MA, 1992). 7. J. B. C. Jackson, Paleobiology 14, 307 (1988); S. L. Wing, L. J. Hickey, C. C. Swisher, Nature 363, 342 (1993); M. L. Droser, D. J. Bottjer, P. M. Sheehan, Geology 25, 167 (1997). 8. T. L. Phillips and R. A. Peppers, Int. J. Coal Geol. 3, 205 (1984); D. W. Krause, Univ. Wyoming Contrib. Geol. Spec. Pap. 3 (1986), pp. 95–117; S. Lidgard and J. B. C. Jackson, Paleobiology 15, 255 (1989); B. E. Berglund, H. J. B. Birks, M. Ralska-Jasiewiczowa, H. E. Wright, Eds., Paleoecological Events During the Last 15,000 Years (Wiley, Chichester, UK, 1996); R. Lupia, P. R. Crane, S. Lidgard, in Biotic Response to Global Change: The Last 145 Million Years, S. J. Culver and P. F. Rawson, Eds. (Chapman & Hall, London, in press). 9. F. K. McKinney and J. B. C. Jackson, Bryozoan Evolution (Unwin Hyman, Boston, 1989). 10. P. D. Taylor and G. P. Larwood, in Extinction and Survival in the Fossil Record, G. P. Larwood, Ed. (Clar- endon, Oxford, 1988), pp. 99 –119. 11. All bryozoan fragments retained on 0.5-mm and larg- er screens were picked from 60 entire disaggregated 0.3- to 3.0-kg samples or, where bryozoans were extraordinarily abundant, from subsamples; frag- ments were sorted on the basis of clade, and total mass for each clade was determined directly or cal- culated on the basis of subsamples. Species that produce colony fragments typically smaller than 0.5 mm have not been included in most previous studies of bryozoan species diversity [summarized in (13)]. We have excluded such small size fragments from our calculations of relative skeletal mass as well; these fractions are subject to winnowing in many recent bryozoan habitats and are rarely derived from the ecologically dominant taxa. Nonetheless, small colo- nies of cyclostomes and cheilostomes may be numer- ous in some environments (E. Håkansson, personal communication). Our data were supplemented by data in O. Berthelsen, Danmarks Geol. Unders. 83,1 (1962) and A. H. Cheetham, Smithsonian Contrib. Paleobiol. 6, 1 (1971). Some variation in the data inevitably results from variable amounts of cement or of matrix or shell fragments. Cyclostomes tend to have thinner walls than do cheilostomes, and raw cheilostome mass was weighted by 1.26 on the basis of a thin-section determination of the ratio of skel- eton to cement plus adherent material in control sam- ples [33 cheilostomes (X 5 0.62, SD 5 0.147) and 35 cyclostomes (X 5 0.51, SD 5 0.151) from four repre- sentative collections]. Additional “noise” in the data may be due to different taphonomic responses of cy- clostome and cheilostome bryozoans in different envi- ronments. However, dissolution and abrasion rates of mineralized bryozoans are dependent upon diverse fac- tors that cut across clade assignment [A. M. Smith, C. S. Nelson, P. J. Danaher, Palaeogeogr. Palaeoclimatol. Palaeoecol. 93, 213 (1992); A. M. Smith and C. S. Nelson, in Biology and Palaeobiology of Bryozoans, P. J. Hayward, J. S. Ryland, P. D. Taylor, Eds. (Olsen & Olsen, Fredensborg, Denmark, 1994), pp. 177–180; in Bryozo- ans in Space and Time, D. P. Gordon, A. M. Smith, J. A. Grant-Mackie, Eds. (National Institute of Water & At- mospheric Research Ltd., Wellington, New Zealand, 1996), pp. 213–226. 12. J. J. Sepkoski Jr., J. Paleontol. 71, 533 (1997). 13. S. Lidgard, F. K. McKinney, P. D. Taylor, Paleobiology 19, 352 (1993). 14. The skeletal mass of cyclostomes was 70% in middle Danian and 62% in late Danian. 15. G. R. Upchurch, in Mass Extinctions Process and Evi- dence, S. K. Donovan, Ed. (Columbia Univ. Press, New York, 1989), pp. 195–216; S. J. Fowell and P. E. Olsen, Tectonophysics 222, 361 (1993); R. P. Speijer and G. J. Van der Zwan, in Biotic Recovery from Mass Extinc- tion Events, M. B. Hart, Ed. ( The Geological Society, London, 1996), pp. 343–371. 16. M. Foote, Palaeontology 34, 461 (1991); Univ. Mich- igan Mus. Paleontol. Contrib. 28, 101 (1991); Proc. Natl. Acad. Sci. U.S.A. 89, 7325 (1992); Paleobiology 19, 185 (1993); ibid. 21, 273 (1995); Science 274, 1492 (1996). 17. The mean size of encrusting cyclostomes is 4.7 mm 2 and of cheilostomes is 10.1 mm 2 on shell debris in the Mainstreet Limestone Member of the Grayson Formation (Albian), Roanoke, Texas; the mean size of encrusting cyclostomes is 5.8 mm 2 and of encrusting cheilostomes is 51.8 mm 2 on shell debris in the northern Adriatic off Rovinj, Croatia (F. K. McKinney, unpublished data). The two faunas are qualitatively judged to represent colony size of contemporaneous faunas, and the major change has been an increase in mean size of cheilostome colonies. 18. K. J. Gaston, Ed., Biodiversity: A Biology of Numbers and Difference (Blackwell, Oxford, 1996). 19. L. Wilkinson, SYSTAT: The System for Statistics (Sys- tat, Evanston, IL, 1990). 20. Supported by the NSF (DEB 9306729 to S.L.; EAR 9117289 to F.K.M.), U.S.-U.K. Fulbright program (F.K.M.), National Geographic Society (F.K.M.), Petro- leum Research Fund of American Chemical Society (F.K.M.), NASA (NAGW-1963 to J.J.S.J.), and Global Change and the Biosphere Programme of the Nation- al History Museum/University College London (P.D.T.). We thank S. Hageman, R. Lupia, and two anonymous reviewers for evaluating the manuscript and numerous colleagues who served as guides to field localities. 16 April 1998; accepted 26 June 1998 In Situ Observations of a High-Pressure Phase of H 2 O Ice I-Ming Chou,* Jennifer G. Blank,² Alexander F. Goncharov, Ho-kwang Mao, Russell J. Hemley A previously unknown solid phase of H 2 O has been identified by its peculiar growth patterns, distinct pressure-temperature melting relations, and vibra- tional Raman spectra. Morphologies of ice crystals and their pressure-temper- ature melting relations were directly observed in a hydrothermal diamond-anvil cell for H 2 O bulk densities between 1203 and 1257 kilograms per cubic meter at temperatures between –10° and 50°C. Under these conditions, four different ice forms were observed to melt: two stable phases, ice V and ice VI, and two metastable phases, ice IV and the new ice phase. The Raman spectra and crystal morphology are consistent with a disordered anisotropic structure with some similarities to ice VI. The manifold ways in which the water mol- ecules may link through hydrogen bonding give rise to a remarkably rich phase diagram (1–5). Enhancing this complexity is the exis- tence of both proton-ordered and -disordered forms as well as metastable crystalline and amorphous phases (3, 6 ). Though evidence for additional phases in the system has been obtained in the past [for example, (7 )], infor- mation about them has been very sparse, if not controversial, because previous studies have relied principally on quench techniques or limited in situ probes (7–10). Here we document the existence of another H 2 O phase from in situ microscopy and Raman spectros- copy at 0.7 to 1.2 GPa. The phase exhibits an I. Chou, 955 National Center, U.S. Geological Survey, Reston, VA 20192, USA. J. G. Blank, A. F. Goncharov, H. Mao, R. J. Hemley, Geophysical Laboratory and Center for High Pressure Research, Carnegie Institu- tion of Washington, 5251 Broad Branch Road, NW, Washington, DC 20015, USA. *To whom correspondence should be addressed. ²Present address: Department of Geology and Geo- physics, University of California, Berkeley, CA 94720, USA. R EPORTS www.sciencemag.org SCIENCE VOL 281 7 AUGUST 1998 809