This dissertation has been microfilmed exactly as received 69-22,155

JONES, Lois Marilyn, 1934- THE APPLICATION OF STRONTIUM ISOTOPES AS NATURAL TRACERS: THE ORIGIN OF THE SALTS IN THE LAKES AND SOILS OF SOUTHERN VICTORIA LAND, ANTARCTICA.

The Ohio State University, Ph.D., 1969 Geology University Microiilms, Inc., Ann Arbor, Michigan THE APPLICATION OP STRONTIUM ISOTOPES AS NATURAL TRACERS: THE ORIGIN OF THE SALTS IN THE LAKES AND SOILS OF SOUTHERN VICTORIA LAND, ANTARCTICA

DISSERTATION Presented in Partial Fullfillment of the Requirements for the Degree Doctor of Philosophy in the Graduate School of The Ohio State University

by Lois Marilyn Jones, B. Sc,, M. Sc

The Ohio State University

1969

Approved by

Adviser Department of Geology PLEASE NOTE: Several pages contain colored illustrations. Filmed in the best possible way. UNIVERSITY MICROFILMS ACKNOWLEDGMENTS

I wish to express my gratitude to Professor Gunter Faure, whose advice, discussions, and encour­ agement have been indispensible in bringing this study to a satisfactory conclusion. This study also could not have been possible without the generosity of numerous people who willingly contributed valuable samples utilized in this investi­ gation. Dr. Derry D. Koob contributed the excellent series of samples from depth profiles of Lakes Vanda and Bonney and the sample from Don Juan Pond. Mr. R. E. Behling and Dr. K. R. Everett contributed the samples of soil from the vicinity of the Meserve Glacier and from Taylor Valley, as well as samples of salts and meltwater. Mr. H. J. E. Montigny collected the soils and water samples from the floor of Wright Valley and the extensive suites of rock samples from the basement complex of the valley. Mr. Gerald Holds- worth provided the samples of ice, neve, meltwater, and salts from the Meserve Glacier. Samples of the KcKurdo volcanics were donated by the following: Dr. Peter Anderton; Dr. G. H. Denton (Yale University); Dr. H. H. Gair (Geological Survey Office, New Zealand); Mr. J. Kovach; Dr. V. H. Minshew; Mr. J. G. Murtaugh;

ii Miss Tonia Sledzinska (Victoria University of Wellington,

New Zealand); and Dr, S, B, Treves (University of

Nebraska), Mr. David Greegor provided samples of the bottom sediment of Lake Vanda.

1 would also like to acknowledge the following people who have maintained an interest in this study and who have contributed to its present conclusion through numerous discussions: Dr. Colin Bull;

Dr. K. H. Everett; Mr. R. E, Behling; and Mr. G. Holds- worth. Dr. R. J. Fleck and Dr. S. E. White read the manuscript and made many helpful suggestions.

During the past two years I have been able to devote full-time efforts to this investigation through awards of an NDEA Title IV Fellowship during the

I9 6 7 -I9 6 8 academic year and the John A. Bownocker

Fellowship for the 1 9 6 8 -1 9 6 9 academic year. Financial assistance through the National Science Foundation

Grant No. GA-713 Is also gratefully acknowledged.

ill VITA

September 6 , 1934 .... Born, Berea, Ohio

1955...... B. Sc., Chemistry, The Ohio State University, Columbus, Ohio

1959 * M. Sc., Analytical Chemistry, The Ohio State University Columbus, Ohio

1958-1961 ...... Research Analytical Chemist, E. I. duPont de Nemours, Wilmington, Delaware

1961-1963 * ...... Lecturer, Department of Chemistry, Memorial University of Newfoundland, St. John’s, Newfoundland

1963-1965 ...... Analytical Chemist, U. S. Geological Survey, Menlo Park, California

1966-1967 • ...... • Research Assistant, Department of Geology, The Ohio State University, Columbus, Ohio 1967-1968 ...... NDEA Title IV Fellow, Geochemistry, The Ohio State University, Columbus, Ohio

1 9 6 8 -1 9 6 9 • ...... John A. Bownocker Fellowship, Department of Geology, The Ohio State University, Columbus, Ohio

iv PUBLICATIONS

Jones, L. M., 1959, 2,4-Dihydroxybenzenearsonic acid as a spectrophotometric reagent for the determination of iron (III): M. Sc* Thesis, Department of Chemistry, The Ohio State University. Radtke, A, S., and Jones, L. M., 1966, Strontium- bearing todorokite from SoganliyurQk, Turkey: U, S. Geol. Survey Prof. Paper 550-C, p. C158-C161.

Jones, L. M., Faure, G., and Montigny, R. J. E., 1967, Geochemical studies in Wright Valley: Antarctic Jour, of the U. S., v. 3, P* 114. Jones, L* M * , and Faure, G., 1967, Origin of the salts in Lake Vanda, Wright Valley, southern Victoria Land, Antarctuca: Earth Planet. Sci. Letters, v. 3 , p. 101-106.

Faure, G., Jones, L. M., Eastin, R., and Christner, M., 1967, Strontium isotope composition and trace element concentrations in Lake Huron and its principal tributaries: Report No. 2, Laboratory for Isotope Geology and Geochemistry, Water Resources Center and Department of Geology, The Ohio State University, 1 0 9 pp. Jones, L, M., and Faure, G., 1968, Age of the Vanda porphyry dikes, Wright Valley, southern Victoria Land, Antarctica: Earth Planet. Sci. Letters, v. 3 , p. 321-324. Jones, L. M,, and Faure, G., 1968, Origin of the salts in Taylor Valley, Antarctica: Antarctic Jour, of the U. S., v. 3, P* 177-178.

Faure, G., and Jones, L. M*, 1969* Anomalous strontium in the Red Sea brines, ijj Hot Brines and Recent Heavy Metal Deposits in the Red Sea, E. T. Degens and D. A. Ross, Editors, in press. Jones, L. M., and Faure, G,, Strontium isotope geochem­ istry of the Great Salt Lake Basin: in preparation.

v TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS...... ii VITA ...... iv TABLES ...... x FIGURES ...... xiv Chapter I. INTRODUCTION : STATEMENT OF THE PROBLEM . 1 II. PHYSIOGRAPHY OF THE ICE-FREE VALLEYS . . 4 A. Geology and Geomorphology of Antarctica ...... 4 B. Ice-Free Region of Southern Victoria L a n d ...... 11 C. Physiography of Wright Valley . . 12 D. Physiography of Taylor Valley . . 22 E. Climate of the Ice-Free Valleys . 27 F. Geology of Wright and Taylor V a l l e y s ...... 43 III. PHYSICAL AND CHEMICAL PROPERTIES OF LAKE VANDA, WRIGHT V A L L E Y ...... ?C A. Introduction ...... ?0 B. D e n s i t y ...... 52 C. Temperature...... 57

vi Chapter Page D. Source of the Heat ...... 61 E. Chemical Composition of Water in the Onyx R i v e r ...... 64 F. Chemical Composition of Water in Lake Vanda •»•*»»••*••• 82 Introduction Concentration profiles of the principal cations Concentration profiles of the princioal anions Layered structure of Lake Vanda Chemical composition of Lake Vanda The pH profile of Lake Vanda Cause of the high salinity in Lake Vanda Trace elements in Lake Vanda G. Geochemical History of Lake Vanda . 137 H. Source of Salts in Lake Vanda: Previous Investigations 139 Introduction Evaporites Concentration ratios Deuterium content U234/U238 activity ratio Conclusions IV. THE ISOTOPIC COMPOSITION OF STRONTIUM IN WRIGHT VALLEY ...... 150 A. Introduction ...... 150 B. Isotope Geochemistry of Strontium . 151 C. Lake Vanda ...... 154 D* Onyx River ...... 158 E. Ross Sea ...... 160 F. McMurdo Volcanics * ...... 164

vii Chapter Page G. Soil and Bedrock ...... 169 H. The Source of Strontium andOther Elements in Lake V a n d a ...... 175 V. PHYSICAL, CHEMICAL, AND ISOTOPIC CHARACTERISTICS OF DON JUAN POND, WRIGHT VALLEY ...... 183 A. Introduction ...... 183 B. D e n s i t y ...... 138 C. Temperature 18R D. Chemical Composition ofBrine • . . 139 E. Mineralogy of Salts Forming in Don Juan Pond ...... 194 F. Isotopic Composition ofStrontium . 196 G. Conclusions ...... 197 VI. PHYSICAL AND CHEMICAL PROPERTIES OF LAKE BONNEY, TAYLOR VALLEY ...... 200 A. Introduction ...... 200 B. D e n s i t y ...... 209 C. Temperature ...... 208 D. Source of the H e a t ...... 213 E. Chemical Composition ofLake Bonney 216 Introduction Concentration of the major cations Concentration of the major anions Layered structure of Lake Bonney The pH profile of Lake Bonney Chemical Composition of Lake Bonney Trace element concentrations in Lake Bonney

viii Chapter Page F. Geochemical History of Lake Bonney •••■* ...... 251 G. Chemical Composition of"Taylor Red C o n e " ...... 257 VII. THE ISOTOPIC COMPOSITION OF STRONTIUM IN TAYLOR VA L L E Y ...... 264 A. Introduction ...... 264 B. Lake Bonney ...... 265 C. "Taylor Red Melt" ...... 269 D. "Suess Pond" ...... 270 E. Lake Fryxell ...... 272 F. S o i l ...... 273 G. Conclusions ...... 276 VIII. THE CASE FOR CHEMICAL WEATHERING IN WRIGHTVALLEY ...... 279 A. Introduction ...... 279 B. Salt Accumulation, Chemical Weathering, and Age ofthe Salts • 280 C. Chemical Weathering in Wright V a l l e y ...... 292 D. Conclusions...... 303 APPENDIX ...... 304 BIBLIOGRAPHY ...... 325

lx TABLES

Table Page

1. Approximate mean monthly temperatures at McMurdo Station and South Pole Station • • 29 2. Mean and extreme mean temperatures at various sites in the ice-free valleys • • • • 3° 3* Air temperatures on ice surface of Lake Bonney, Taylor Valley ...... 32 4. Number of occasions of wind from different directions, Lake Vanda, Wright Valley . • . . 33 5. Air temperature and relative humidity for the eastern part of Wright Valley • ••»•• 35 6. Mean wind velocities at Lake Vanda, Wright Valley ...... 37 7. Mean temperatures under different wind conditions, Lake Vanda ..••••••••• 33 8. Summer cloud cover for ice-free valleys and for Hut Point, Ross Island * ...... 39 9- Days and total amount of snowfall in the ice-free valleys and for Hut Point, Ross Island...... 41 10. Chronology of the Wright Valley area .... 44 11. Density profile of Lake Vanda, Wright Valley 53 12. Partial chemical analysis of meltwater from Wright Valley ...... 67 13* Ratios of Ca/Mg, Ca/Na, Ca/K, and Na/K for meltwater in Wright Valley . . • ...... 71

x Table Page

14* Partial chemical analysis of n6ve, ice, and meltwater from the Meserve Glacier ...... 76 15* Concentration ratios of Ca/Na, Ca/fcg, Ca/K, and Na/K for n6v£, ice, and meltwater from the Meserve Glacier ...... 79 16, Concentration of calcium, strontium, sodium, and potassium in Lake Vanda, Wright Valley . 85 17* Chemical composition of Lake Vanda • . * . . 86 18. Cation composition of Lake Vanda, expressed as mole per cent ...... 110 19* Anion composition of Lake Vanda, expressed as mole per cent ...... 113 2 0 . Concentration of the major dissolved constituents of sea water, with selected chemical ratios calculated from these values. 114 21. Profile of pH of Lake Vanda ••.*••••. 123 22. Concentration ratios of the principal cations for the lower Onyx River, meltwater stream of the Meserve Glacier, and upper two meters of water of Lake Vanda ...... 128 23. The isotopic composition and concentration of strontium from a depth profile of Lake Vanda ...... 155 24. Isotopic composition of strontium in meltwater from Wright Valley ..*••••• 159 25* Isotopic composition of strontium in the Ross Sea, Meserve Glacier, and Lyttleton Harbor, Hew Zealand ••••..••••••••••. 161 26. Isotopic composition and concentration of strontium of McMurdo volcanics, Victoria L a n d ...... 167 27. Sr87/Sr8* ratios of salts and soils from the floor of Wright V a l l e y ...... 170

xi Table Page

28. Sr8?/Sr06 ratios of the major units of the bedrock, Wright Valley ...... 174 29* Chemical composition of Don Juan Pond and its influent streams ..... 190 30* Local variation in the chemical composition of Don Juan Pond, expressed as equivalent per c e n t ...... 191 31* Density profile of Lake Bonney, Taylor V a l l e y ...... 206

32. Temperature profile of Lake Bonney, Taylor Valley •••••*.*•••••••• 210 33* Concentrations of the principal cations in Lake Bonney ...... 218 34* Concentrations of the principal ions in Lake Bonney ••••••* ...... 219 35. The pH profile for the east lobe of Lake Bonney ...**.••• ...... •• 236 36. Molal concentrations and the salt composition for the major cations of Lake Bonney .... 238 37* Molal concentrations and the salt composition for the principal anions of Lake Bonney . . . 239 38. Partial chemical analysis of water-soluble salts from the "Taylor Red Cone" •••.•• 261 39* Isotopic composition and concentration of strontium from a depth profile of Lake B o n n e y ...... 267 40. Sr®7/Sr86 ratios of "Taylor Red Cone", "Suess Pond", Lake Fryxell, and water-soluble salts of Taylor Valley ...... 271

41. Sr8?/Sr ratios of water-soluble salts and total salt-free soil, and the lithic compo­ sition of moraine samnles, Meserve Glacier . 298 xii Table Page

42. Sr8?/ S r 86 ratios of salt encrustations from boulders in the vicinity of the Meserve Glacier ...... 302 43. Analyses of the isotopic composition of strontium of the Eimer and Amend SrC03 standard ...... 317 44. Analysis of U. S. G. S. standard rocks G-l and W-l for calcium, magnesium, sodium, and potassium ...... 319

xiil FIGURES

Outline map of Antarctica ...... 5 Wright Valley, view east toward Ross Island and McMurdo Sound ...... 14 Wright Valley, view east toward Ross Island and McKurdo Sound ...... 15 Western Wright Valley, view west toward the Inland ice plateau ...... 16 North Fork, Wright Valley, view toward the east ...... 17 View north toward Victoria Valley system from over Wright Valley • ...... 18 Longitudinal profile of Wright Valley • • 21 Taylor Valley, view east toward Ross Island and McMurdo Sound ...... 23 Taylor Valley, view west toward inland ice plateau ...... 24 Eastern Taylor Valley, view east toward Ross Island and McMurdo Sound ...... 25 Ablation from a snow-free ice surface 62 m above sea level near Mawson ...... 42 Lake Vanda, Wright Valley, view west toward North Fork and Wright Upper Glacier 51 Density profile of Lake Vanda, Wright V a l l e y ...... 55

xiv Figure Page

14. Temperature profile of Lake Vanda, Wright Valley ...... 58 l?. Onyx River, Wright Valley, view toward the west ...... 65 16. Onyx River, Wright V a l l e y ...... 66 17* Location of water, ice, and soil samples in Wright Valley ...... 69 1 8 . Variation of the Na/K ratio with distance along the Onyx River, Wright Valley .... 74 19. Concentration of sodium along a depth profile, Lake Vanda, Wright Valley ...... 89 20. Logarithmic plot of the sodium concentration along a depth profile of Lake Vanda .... 92 21. Logarithmic plot of the potassium concentration along a depth profile of Lake V a n d a ...... 94 22. Logarithmic plot of the calcium concentration along a depth profile of Lake Vanda .... 97 23* Logarithmic plot of the chloride concentration along a depth profile of Lake V a n d a ...... 100 24. Concentrations of calcium and strontium in Lake V a n d a ...... 106 25. Concentrations of sodium and potassium in Lake V a n d a ...... 107 26. Concentrations of calcium and sodium in Lake Vanda ...... • 109 27. Variation of the Na/K ratio along a depth profile of Lake Vanda ...... 117 2 8 . Variation of the Ca/Sr ratio along a depth profile of Lake V a n d a ...... 120 xv Figure Page

29. Relative deuterium concentration along a depth profile of Lake V a n d a ...... 144 30. The Sre?/Sr86 ratio and strontium concentration from a depth profile of Lake V a n d a ...... 156 31. Comparison of the isotonic composition of strontium in Lake Vanda and possible sources of its salts ...... 162 32. Basaltic cinder cone on the Loop Koraine, eastern Wright Valley ...... 165 33. Distribution of Sr8?/Sr86 ratio measurements of the KcMurdo volcanics, Victoria Land . • 168 34. Western portion of Wright Valley, view west toward the inland ice plateau ..... 184 35. South Fork, Wright Valley ...... 185 36. Don Juan Pond, Wright Valley, view from edge of pond westward up the South Fork . . 186 37. Lake Bonney and Taylor Glacier, Taylor Valley, view toward the northwest ..... 202 38. Density profile of Lake Bonney ...... 207 39. Temperature profile of Lake Bonney ...... 209 40. Concentration profile of magnesium for Lake Bonney ...... 221 41. Concentration profile of calcium for Lake Bonney ...... 223 42. Concentration profile of sodium for Lake Bonney ...... * 225 43. Concentration profile of potassium for Lake Bonney 226 44. Concentrations of sodium and potassium in Lake Bonney ...... 231

xvi Figure Page

45* Concentrations of magnesium and sodium in Lake Bonney ...... 232 46. Concentrations of calcium and strontium in Lake Bonney ...... 234 47. Mole per cent of sodium, magnesium, and calcium in Lake Bonney ...... 240 48. Depth profile of the Mg/Ca concentration ratio of Lake Bonney ...... 242 49. Concentration ratio of Ca/Sr for a depth profile of Lake Bonney ...... 246 50. Concentration ratio of Na/lC for a depth profile of Lake Bonney ••••• 248 51. Change in channel width of Lake Bonney since 1903 • •* • • • • . * ...... 255

92. "Taylor Bed Cone” ...... 258 53. "Taylor Bed Cone", a close-up ...... 2 59 54. Location of water and soil samples from Taylor Valley ...... 266 55. Isotopic composition and concentration of strontium along a depth profile of Lake Bonney ...... 268 56. Variation of the Sr8?/Sr86 ratio along a longitudinal profile of Taylor Valley • . . 278 57- Salt accumulations in an excavation in the outer moraine, Meserve Glacier ...... 281 58. Cross-section of salt-riven schist ...... 282 59* Cavernously-weathered coarse-grained diabase boulder, Wright Valley ...... » 283 60. Ventifacts of Ferrar Dolerite, Wright V a l l e y ...... 284 xvii Figure Page

61. Lake Vanda, Wright Valley, view to the west ...... 288 62. Aerial view of the Meserve Glacier ...... 290 63. Basaltic cinder cone, western edge of the accumulation basin of the Meserve Glacier . 291 6 4 . Pit in the middle moraine, west side of the Meserve Glacier ...... 293

65* Meserve Glacier, view to the s o u t h ...... 296 66. Location of soil samples from moraines of the Meserve Glacier ...... 297

xviii CHAPTER I

INTRODUCTION STATEMENT OF THE PROBLEM

The ice-free valleys of southern Victoria Land are located in the Transantarctic Mountains west of Ross Island, Antarctica, In these valleys, the average annual temperature is well below the freezing point of water and precipitation is extremely low. Nevertheless, lakes are present and soils have begun to form since degla­ ciation. These lakes and soils are characterized by the presence of large amounts of salts. The salinity of 3ome lakes, such as Lake Vanda and Don Juan Pond in Wright Valley and Lake Bonney in Taylor Valley, is several times greater than that of sea water. In addition, salts occur as surface efflorescences and as lenses and cement within the soil. The source of the large quantities of salts in the lakes and soils of the ice-free valleys is controversial. Possible sources that have been suggested include: (1) trapped sea water; (2) wind-transported marine salts; (3) volcanic activity and associated hot springs; 1 2

(4) leaching of evaporite beds in the local sedimentary rocks; and (5) chemical weathering of local soil and bedrock. The parameters used in previous attempts to link the salts with a specific source have failed because they are continually being modified by the chemical and physical processes occurring in the lakes and, therefore, lead to conflicting conclusions. In order to determine the origin of the salts, a new parameter was needed that could unambiguously Identify a specific source for the salts. The isotoDic composition of strontium in the salts meets the necessary requirements, because: (1) the isotopic composition of strontium of each of the possible sources is distinctive and differs significantly from that of the other sources, and (2) the isotopes of strontium are not measureably fractionated in natural processes such as are occurring In the ice-free valleys. Therefore, this study was initiated to ascertain the applicability of strontium isotopes as natural tracers, and to identify thereby the source(s) of the salts in the lakes and soils of the ice-free valleys in Antarctica. Lakes Vanda and Bonney were studied in detail because they are the largest and most unusual of the Antarctic lakes. Both lakes are perennially ice-covered and are meromictic. At depth the water is highly saline and has surprisingly high temperatures. Lake Vanda has a maximum density of 1.10 g/ml at a depth of 67 m, and a maximum recorded temperature of +28°C at the bottom of the lake. Lake Bonney has a maximum density of 1.20 g/ml at a depth of 32 m, and reaches a temperature of +8°C at about the middle of the depth profile. In order to determine the origin of the salts in the two lakes, measurements were made of the isotopic compo­ sition of strontium in water samples collected at different depths from the surface to the bottom of the lakes. The results of these analyses were then compared to icotopic compositions of strontium in sea water, basalts of the McMurdo volcanic province, and the strontium In water-soluble salts from the soils in Wright and Taylor Valleys. From these comparisons the principal sources of the strontium in the two lakes could be clearly identified. In addition to measurements of the isotopic compo­ sition of strontium in the lakes, chemical analyses of the brines have been made to provide information that can be used with the isotopic studies to develop a model for the geochemical evolution of the lakes. The combination of isotopic and chemical analyses of the brines in Lakes Vanda and Bonney permit the formulation of a more valid model than has been possible before. CHAPTER XI

PHYSIOGRAPHY OF THE ICE-FREE VALLEYS

A. Geology and Geomorphology of Antarctica

Antarctica Is a roughly circular land mass about 4500 km In diameter, including the continental shelf. Its coastal outline Is broken by the narrow curving Antarctic Peninsula and by two deep embayments, the Ross Sea and the Weddell Sea (figure 1). The continent is usually divided into two major portions: East Antarctica, which lies chiefly In longitudes east of Greenwich, south and east of the Weddell Sea and south and west of the Ross Sea and makes up about three- quarters of the total area, and West Antarctica, which lies predominantly in the western hemisphere. Several reviews of the geology and geomorphology of Antarctica are in the literature (e.g., Adle, 1962) Ford, 1964; Gunn, 1963; Harrington, 1965; Nichols, 1965 and 1966; Warren, 1965), and only a general description follows. The area of Antarctica is about 14 million square km. More than 953* of its area is blanketed by the 4 5 i

SEA

ROSS

SHELF PACIFIC

ROSS' V iW ISLAND {\l

ANTARCTICA 100 HH.II

Figure 1. Outline nap of Antarctica 6

Antarctic Ice Sheet. The ice sheet of East Antarctica is deeply embayed south of the Indian Ocean by the Amery Ice Shelf and by the Lambert Glacier, with a drainage system extending well over 1000 km Inland towards the Weddell Sea. The pattern of ice flow outward is also considerably affected by the Transantarctic Mountains, a range that borders East Antarctica from northern Victoria Land to Coates Land on the Weddell Sea. This range extends for 5000 km from near Cape Adare In Victoria Land, through the Queen Maud Range to the Horllck, Thiel, and Pensacola Mountains and the Shackleton Range. Only two other major mountain systems project through the ice cap: the range near Pronning Maud Land, stretching Intermittently In an arc to the neighborhood of Liitzow-Holm Bay, and the ranges that flank the Lambert Glacier. A Precambrlan shield underlies most of East Antarctica. No general tectonic pattern can be deduced from available field observations, which are limited to isolated exposures mainly at the continental margin. Descriptions of regions 3000 km apart are remarkable in their similarity. In general, thick sequences of quartzltic, politic, and calcareous sediments have been migmatlzed and often remobilized and metamorphosed to granulltes, Invaded by oharnockitic granites and aplltes, and later invaded again by a series of younger granites 7 and charnockites of Paleozoic age (Gunn, 1963)* (Refer to age compilations by Picciotto and Coppez, 1963, 1964} Webb, 1963b; and Webb and Warren, 1965*) The Transantarctic Mountains probably occupy the site of the Ross Geosyncline. The Ross Geosyncline extended at least from Oates Land for 4200 km to the Weddell Sea* The age of the sediments, based on paleontological evidence, range from at least Late Precambrian to mid-Cambrian time* Folding occurred in Precambrian time during the Mawson Orogeny, before deposition of the Cambrian sediments* Folding also occurred in late Cambrian-early Ordovician time during the Ross Orogeny. At this time at least much of the exposed shield was subjected to igneous intrusions and regional metamorphism or to uplift and cooling, which is reflected in K-Ar dates (Picciotto and Coppez, 1963, 1964} Webb, 1963b; and Webb and Warren, 1965)* Although the range of dates is large, from 400 to 560 m.y., there is a concentration of dates in the interval of 450 to 500 m.y. This probably represents the period of the most Intense orogenic activity during the Ross Orogeny* A period of unknown length of erosion of the basement complex followed the Ross Orogeny. The length of this Interval has been estimated to extend from Silurian to Early Devonian time (Gunn, 1 963). This erosion resulted in a flat to gently undulating surface of great extent, that is found not only In the Transantarctic Mountains but elsewhere In East Antarctica. This surface was first described by Debenham (1921) from the Kukri Hills near McMurdo Sound and subsequently named the Kukri Peneplain (Gunn and Warren, 1962), Flat-lying to gently-dipping continental or near-shore sediments unconformably overlie the basement complex extensively in East Antarctica. These sediments have been named the Beacon Group. In Victoria Land these rocks are called the Beacon Sandstone Group and those elsewhere have been given local names (Harrington, 1958). The greatest thickness known for the Beacon Group is approximately 2600 m for a section In the Beardmore Glacier area (Barrett, 1968), The predominant lithology is pale yellow or buff sandstone, commonly cross-bedded. Siltstone is present In all sections, and conglomerates, limestones, calcareous sandstones, pyroclastlcs, black fissile shale, and coal measures occurr in many areas (Hamilton and Hayes, 1963; Warren, 1965; and Zeller, et al.. 1961). Although there has been considerable controversy concerning the depositlonal environment of the Beacon rocks, recent studies using the isotoplc composition of strontium In carbonate rocks suggest that the majority of the sediments were laid down under non- marine conditions (Barrett, £& 4l*, 1968). The age of the Beacon Group ranges from Lover Devonian (Boucot, Al* * 1963) to at least Upper Jurassic (Plumstead, 1962). During the Jurassic Period large quantities of diabase were intruded into the basement complex and the Beacon Group (McDougall, 1963)* These rocks have been named the Ferrar Dolerites (Harrington, 1958) and the contemporaneous associated lavas and pyro- clastics have been named the Kirkpatrick Basalts (Grindley, 1963). Although some tlllites containing volcanics have been found overlying the Beacon Group in Victoria Land (Gunn and Warren, 1962), no major rock units younger than the Jurassic sediments or the Ferrar dolerites are known in East Antarctica except for some late Tertiary- Quaternary basic volcanics (the McMurdo volcanics). These are located mainly on Ross Island, at Gaussberg, at Mirnyy, and sporadically in the Transantarctic Mountains from Cape Adare south to Mount Early in the Scott Glacier area. These volcanics have been dated by the K-Ar method, giving an age for the Gaussberg basalts of 20 m.y* (StarIk, al.. 1961) and 22.1 and 27*3 m.y. for basalts found in the vicinity of Mount Early (V. H. Mlnshew, 1969, personal communication). K-Ar dates for basalts from Wright and Taylor Valleys, southern Victoria Land, range from 2 to 4 m.y. (Armstrong, al., 1968} and Denton and Armstrong, 1968). A K-Ar date of anorthoclase indicates an age of 0.68 m.y. for the kenyte at Cape Royds, the 10 westernmost part of Ross Island (Treves, 1967)* Mount Erebus on Ross Island Is the only known active volcano in East Antarctica. Vest Antarctica Is usually divided Into three main regions: Marie Byrd Land, Ellsworth Land, and the Antarctic Peninsula. It is accepted that West Antarctica is a Paleozoic, Mesozoic, and Cenozolc orogenlc or mobile zone (Harrington, 196?). Adle (1962) states that the crystalline shield of East Antarctica extends into parts of West Antarctica, where It is found only to the south of the Marguerite Bay region* Geophysical studies Indicate the presence of a normal continental crust of 30-38 km thickness in Marie Byrd Land (Woollard, 1962). The ice cover of Marie Byrd Land is broken by several mountain ranges that are relatively short compared to the Trans­ antarctic Mountains. These ranges include the Edsel Ford and Executive Committee Ranges and the Ellsworth Mountains* The Antarctic Peninsula, including the Scotia Arc, is a long, narrow, curving chain of mountains that extends into the fold mountain belt of Patagonia (Ford, 1964). A deep submarine trench borders the convex side of the Scotia Arc, and the outermost islands are subject to earthquake and volcanic activity. The most recent volcanic activity occurred on Deception Island in

February, 1969* 11

B. Ice-Free Region of Southern Victoria Land

Host of the Transantarctic Mountains are Ice-covered and the main valleys are occupied by outlet glaciers. However, in southern Victoria Land, for about 150 km, extending from the Convoy Aange south to the Koettlitz Glacier, there are areas that remain almost entirely free of snow and Ice. This region of about 4000 square km constitutes the largest accessible Ice-free area in Antarctica (see maps in back pocket). The main valleys are oriented In an east-west direction and were carved by outlet glaciers from the inland Ice plateau. Various mechanisms have been proposed to explain local deglaciation of the Antarctic continent and these are summarized by Bull and others (1962). These mechanisms Include volcanlsm, settling of dust particles, underground coal fires, and heating by radio­ activity in the basement rocks. Bull and others (1962) have shown that the deglaciation was caused by a decrease in the surface level of the Inland ice plateau with consequent emergence of high rock thresholds at the western ends of the valleys, cutting off the supply of plateau Ice to the valley glaciers (see maps in back pocket). Changes in the climate of these valleys have slowly caused the disappearance of much of the glacial 12 Ice in the valleys and most of the nev£ fields in the separating ridges (Bull, 1962). The most extensive of the exposed Ice-free areas Includes three major valley systems and the intervening ridges. Wright and Taylor Valleys are relatively simple east-west valleys, while the Victoria Valley system consists of five Interconnected valleys (see maps in back pocket). A series of smaller valleys, ranging in length up to about 12 km occurs along the western edge of the Koettlitz Glacier. Some ice-free areas also occur on the eastern flanks of the Convoy Range about 30-40 km north of Victoria Valley. Trunk glaciers occur at the heads of the ice-free valleys. Alpine glaciers flow down the flanks of the intervening ridges. Some valleys open onto the Ross Ice Shelf, such as Taylor and Miers Valleys, but WTight and Victoria Valleys are Isolated from the coast by the Wilson Piedmont Glacier, which reaches an altitude of more than 600 m.

C. Physiography of Wright Valley

The largest of the ice-free valleys is Wright Valley, extending for more than 50 km in an east-west direction (see maps in back pocket). Although field parties from the expeditions of Scott and Shackleton in the early 1900’s had carried out investigations in Taylor Valley 13 and in the valleys to the south and along the coast of McMurdo Sound, the large Ice-free areas of Vfright Valley and the Victoria Valley system were unknown to them. Aerial photographs taken during the 1955-1956 and 1956-1957 field seasons show these valleys (Bull, fit al., 1962; and B. M. Gunn, 1969, personal communication). Field parties from several nations, mainly the United States, New Zealand, and Japan, have been working in Wright Valley. Wright Valley is an elongated basin bounded on the north by the Olympus Range, which attains a maximum height of more than 24-40 m (figures 2*5)* The Asgard Range forms the southern boundary of the valley and attains altitudes greater than 2400 nu Wright Valley is connected to the Victoria Valley system to the north by Bull Pass, which cuts through the Olympus Range about mid-way along Wright Valley (figure 6). A large alluvial fan has formed at the base of Bull Pass where it opens onto Wright Valley (figure 3). Alpine glaciers extend down the valley walls; the majority are on the south wall, and these decrease in size away from McMurdo Sound, until about mid-valley. None are found in the western half of the valley. At the western end of Wright Valley, ice flowing eastward from the inland ice plateau is channelled between nunataks and subglacial extensions of the Asgard and Olympus Ranges (figure 4). This ice flows over high rock shelves flanking Mount Fleming and coalesces to form 14

Figure 2. Wright Valley, view east toward Ross Island and McMurdo Sound at top of photograph. BP Bull Pass RI Ross Island D Dais N North Fork IR Insel Range S South Fork L Labyrinth V Lake Vanda M Meserve Glacier W Victoria Valley Me McMurdo Sound * - Wilson Piedmont Glacier McK McKelvey Valley 15

Figure 3 . Wright Valley, view east toward Ross Island and McMurdo Sound at top of photo. Olympus Range is at left and Asgard Range along right. B Bar tley Glacier M Meserve Glacier BP Bull Pass Me McMurdo Sound C Clark Glacier 0 Onyx River D Denton Glacier HI Ross Island G Goodspeed Glacier T Taylor Valley II Hart Glacier WL Wright Lower Glacier L Loop Moraine WP Wilson Piedmont Glacier 16

Figure 4-. Western Wright Valley, view west; toward the inland ice plateau at top of photo.

B Bartley Glacier 0 Onyx River BP Bull Pass S South Fork D Dais T Taylor Glacier H Hart Glacier Tu Tunnel in Meserve Glacier L Loop Moraine V Lake Vanda M Meserve Glacier WU Wright Upper Glacier N North Fork 17

Figure 5. North Fork, Wright Valley, View is toward the east, with part of Lake Vanda visible at center. Alpine glaciers on the south wall of the valley are in the background. The Dais is at right and the Olympus Range at left. (Photo courtesy of D, D. Koob) Figure 6. Aerial view north toward Victoria Valley System from over Wright Valley. BP Bull Pass VI Lake Vida BV Balham Valley Vs Lake Vashka IR Insel Range W Victoria Valley McV McKelvey Valley 19 the 10 km-long Wright Upper Glacier at an altitude of 1400 a (Bull, al., 1962). At the base of the Wright Upper Glacier there is an area of about 18 square kilometers called the Labyrinth (figure 2). The topography consists of anastomosing incised channels cut into terrain consisting mainly of Ferrar dolerite. The origin of this unusual topography has been attributed tos (1) subglacial dissection (Cotton, 1966; and Gunn and Warren, 1962); (2) cata­ strophic erosion, either of fluvial (Smith, 1965) or volcanic (Warren, 1965) processes; and (3 ) differential salt weathering of the dolerite along a regional jointing pattern (A* T. Wilson, letter to C. Bull, dated April 16, 1968). East of the Labyrinth is the Dais, an elongated, flat-topped feature composed mainly of granite. It separates Wright Valley Into two parallel branches, the North Fork and the South Fork (figure 6). The floors of these smaller valleys contain thick moraine material over a length of 8 km. To the vest they rise steeply and merge into the Labyrinth. Wright Valley is separated from McMurdo Sound to the east by the Wright Lower Glacier, which is an outlet of the Wilson Piedmont Glacier (figure 3)* T^e ice has a maximum width of about 20 km here and its surface attains ao a maximum altitude of about 500 m. The exact sub-glacial topography is not known, but a gravity traverse suggests that there Is a bedrock threshold of the valley about 300 m in height (Bull, I960; figure 7). At the foot of Wright Lower Glacier, at about 300 m altitude, is a large, shallow lake (figure 3)* Although this lake probably is frozen to its base during the winter, the air temperatures In the summer are sufficiently high to melt some of the snow and Ice. Meltwater is also added to this lake from Wright Lower Glacier and by inter­ mittent streams from the Clark Glacier. When sufficient water is present, a stream, the Onyx River, forms and flows to the west, away from the coast, In a braided channel. The Onyx River is joined along its course by meltwater streams issuing from the valley glaciers and patches of accumulated snow and ice. The Onyx River flows westward for a total distance of approximately 30 k™* eventually emptying into Lake Vanda, whose present surface is 123 m above sea level. Lake Vanda occupies a closed bedrock basin situated between Bull Pass and the Dais (figure 4). Alluvial and glacial debris cover the floor of Wright Valley and extend up the valley walls. Moraines are prominent ir. certain locations along the valley floor and around the alpine glaciers. The periglacial features m eters 500- 2504 750- SL exaggerationabout (from 14X Bull, I960), 16 iue7 Longitudinalprofile ofFigure Wright7.Valley; vertical rE Lake Vanda 7* I62#E oe Gtaciei Lower Wright erc threshold bedrock I63#E idot Gl. Piedmont Wilton u r have been aptly described by McCraw (1967b) and Nichols (1963, 1966).

D. Physiography of Taylor Valley

Taylor Valley is an ice-free valley trending in an east-west direction roughly parallel to Wright Valley, which lies to the north* It extends for 35 km from the base of Taylor Glacier eastward to McMurdo Sound (see maps in back pocket and figures 8, 9» and 10)* The northern boundary of the valley Is the Asgard Range, whose highest peaks range from 2000 to over 2400 m. The Kukri Hills border the valley to the south* At the western end of the valley, the highest peak Is over 2200 m* Summit elevations gradually descend as the coast is approached* Taylor Valley is narrowest at the western end, where it is about 6 km wide. It widens eastward to about 12 km among the more subdued mountains near the sea* The valley opens directly onto McMurdo Sound (figures 8 and 10) and neither an intervening bedrock threshold nor a glacier prevents direct access to the sea as in Wright Valley* The floor of the valley undulates, thereby creating several basins with internal drainage. Most of these basins are occupied by small lakes. Taylor Glacier occupies the westernmost part of the valley (figures 8 and 9)* It is an active outlet glacier 23

Figure 8. Taylor Valley, view east toward Ross Island and McMurdo Sound at top of photograph* Black areas near Sollas Glacier are basaltic cones and patches. B Lake Bonney M Matterhorn Glacier C Calkin Glacier Me McMurdo Sound F Lake Fryxell R Rhone Glacier Fr Ferrar Glacier RI Ross Island - H Hughes Glacier S Sollas Glacier L LaCrolx Glacier WP Wilson Piedmont Glacier 24

Figure 9* Taylor Valley, view west toward inland ice plateau at top of photograph. B Lake Bonney M Matterhorn Glacier C Calkin Glacier R Rhone Glacier Fr Ferrar Glacier S Sollas Glacier H Hughes Glacier T Taylor Glacier L LaCrolx Glacier 25

Figure 10. Eastern Taylor Valley, view east toward Ross Island and McMurdo Sound at top of photograph. Strand lines are visible on the southeast wall of valley. C Lake Chad M McMurdo Sound Ca Canada Glacier MP Marble Point Cw Commonwealth Glacier HI Ross Island F Lake Fryxell S Ross Ice Shelf Fr Ferrar Glacier WP Wilson Piedmont Glacier H Howard Glacier 26 flowing to the east from the Inland ice plateau* At the base of Taylor Glacier is Lake Bonney, composed to two lobes separated by the Bonney Riegel. Other lakes lie to the east along the valley floor. The principal ones are Lakes Chad and Fryxell (figure 10)* Most of the floor of the valley is blanketed by morainal material deposited by the Taylor Glacier, the alpine glaciers that extend down the valley walls to the north and south, and former glaciers that extended into the valley from McMurdo Sound* Prominent ridges are observed at the eastern end of Taylor Valley, particularly on the south wall (figure 10). These have been variously Interpreted as marine strand lines or lateral moraines (R* F. Black, 1968, and C* Bull, 1969* personal communications)* Detailed studies of the glacial geology completed during the 1968-1969 field season by G. H. Denton (1969, personal communication) Indicate they are strand lines, consisting of well-sorted sand* These ridges were formed by a large, deep (over 300 m deep) proglacial lake dammed back into Taylor Valley by a glacier moving inland from McMurdo Sound. Some of these strand lines extend to the west at least as far as Lake Bonney. Bedrock in Taylor Valley is very similar to that in Wright Valley, which is discussed later in this chapter. However, several differences appear to be that 27 the Asgard raetasedlments In Taylor Valley contain more marble beds than are found In Wright Valley. McMurdo basalts are more prevalent In Taylor Valley than In Wright Valley. These volcanics are readily visible in figures 8 and 9. They occur as scoraclous olivine basalt cones, several of which have been over-ridden and scattered by the alpine glaciers. More than 20 distinct cones have been counted. In addition to the visible scoriaceous material, a source of scoria also may exist somewhere to the north under the Canada Glacier (McCrav, 1962).

E. Climate of the Ice-Free Valleys

The Antarctic continent may be divided conveniently Into three broad climatological zones: (1 ) the interior Antarctic plateau, (2) the Antarctic slopes, and (3 ) the Antarctic coast (Weyant, 1966). The climate of the coastal regions are affected by latitude, by proximity to steep slopes, and by the nature of the underlying surface (i.e., ice or exposed bedrock). In general, coastal areas are warmer and have more precipitation than the slopes or the plateau. The ice-free areas have resulted, however, from a combination of meteorological and topographic factors. The local climate of these areas is markedly different from that of snow-covered 28 areas and are characterized by dryness of the air and low precipitation (Weyant, 1966)« Satellite photographs of southern Victoria Land taken during late winter readily show snow-free areas in these valleys, although the winter temperatures are far below freezing. The only terrestrial plants now living in the valleys are mosses, lichens, and algae (Llano, 1962), and these are restricted to a few areas with favorable micro­ climates. Bacteria may be abundant locally, but they are limited to the presence of moisture (£• Kudolph, 19691 personal communication). Approximate mean monthly temperatures for McMurdo Station (table 1) show a marked difference in the temperature regimen between the coastal regions and the interior. Mean annual air temperature at McMurdo Station is about -17°C (Pewe, I960, 1966), and only for a few days does the air temperature rise 2-3 degrees above 0°C (P6w6, i960). The ice-free areas west of McMurdo Sound are about 2-3 degrees warmer than McMurdo Station in the summer (Pewl, I960). Not enough data are available to allow worthwhile estimates of daily or weekly means. Mean temperatures for half-monthly periods during the austral summer have been measured by Bull (1966) at Lakes Vanda, Vi^a, Vashka, and Balham, Bull Pass, and Marble Point (table 2). They are thought to be accurate within * 1°C. TABLE 1* Approximate mean monthly temperatures at McMurdo Station and South Pole Station (from Tedrow and Ugolini, 1966$ based on data from 1956 to 1963).

McMurdo South Month Station Pole

January -4°C -29°C February -9 -40 March -19 -55 April -21 -56 May -23 -57 June -24 -57 July -27 -59 August -29 -57 September -24 -59 October -20 -51 November -9 -38 December -4 -29 30

TABLE 2. Mean and extreme mean temperatures for two-week periods at various sites In the Ice-free valleys, southern Victoria Land (from Bull, 1966).

Mean Mean Period Mean Max. Min. Location

Oct* 16-31 -23.7°C -15.9°C -31.6°C e. Lake Vlda Nov. 1-1? -14.8 -7.8 -21.8 e. Lake Vlda

Nov. 16-31 -6.6 -1.8 -11.3 Bull Pass -7.2 -4.5 -10.0 Marble Point Dec. 1-15 (1961) -4.8 -0.9 -8.7 Lake Vashka (1959) -0.2 3.6 -2*9 Lake Vashka -5.9 -3.5 -8.2 Marble Point Dec. 16-31 -3.2 0.6 -7.0 Balham Lake 0.7 4.1 -2.7 Lake Vashka -2.7 -0.7 -4.7 Marble Point -0.8 1.6 -3.2 Lake Vanda Jan. 1-15 -1.3 1.9 -4.5 w. Lake Vanda -2.7 0.6 - 5.9 Lake Vashka -0.1 2.1 -2.3 Marble Point 2.2 4.9 -0.5 Lake Vanda Jan. 16-31 - 4 . 4 0.1 -8.9 v. Lake Vida -2.5 1.3 -5.8 Lake Vashka -1.4 1.0 -3.8 Marble Point 0.3 3.3 -2.8 Lake Vanda Feb. 1-14 -7.4 -1.7 -13.1 v. Lake Vlda 31 Temperature measurements for Taylor Valley are listed in table 3 . Although the Ice-free valleys are warmer than the coastal regions in the summer, they are also colder In the winter* At Lake Vida a minimum thermometer showed a minimum temperature of -62°C for the winter of i9 6 0 . The lowest temperature recorded at Hut Point on Ross Island in this period was -51°C (Bull, 1966). For an appreciable number of days during the summer, the temperature rises sufficiently above freezing to melt snow and ice, form meltwater streams and pools, and create moats at the edges of lakes* The highest recorded air temperature is 1 2 .3 °C at Lake Vashka in Barwick Valley (Bull, 1966). Although the daytime air temperatures do not get more than a few degrees above freezing, soil temperatures have been measured as high as 24°C at Marble Point (Kelly and Zumberge, 1961) and 27°C at Hallett Station (Rudolph, 1963). Wind direction in Wright Valley is controlled almost completely by the trend of the valley. The wind blew either from south of west or from north of east during the summer of 1956-1959 at Lake Vanda (table 4). On some occasions when the wind was westerly at Lake Vanda, easterly winds were observed in the valley more than 20 km to the east. No westerly winds have been recorded in the eastern portion of Wright Valley (Bull, 1966). Westerly winds encountered near Lake Vanda are 32

TABLE 3* Air temperatures on Ice surface of Lake Bonney, Taylor Valley (from Angino and others, 1962a).

Thermometer height 4 ft above 2 In above Ice surface ice surface Time Date (1961) Max.°C Min. °C Max. °C Min. °C Read

31 October 3.3 -6.1 -6.1 -8.9 1830 1 November 3.9 -15.3 4.4 -10.6 1530 2 ii 2.2 -7.8 8.3 -6.1 1630 3 N 5.6 -16.1 5.6 -12.2 1630 4 II 4.4 -18.3 0 -13.3 1655 5 II 1.7 -16.1 -6.1 -11.7 1805 6 If 3.3 -16.1 -6.1 -17.8 1930 7 K 2.2 -17.2 -6.7 -13.3 2000 8 If 3.9 -15.0 -16.1 1830 9 M 3.9 -15.0 -2.8 -5.6 1930 10 m 2.2 -13.9 -2.2 -15.6 1780 28 u 18.3 -14.4 -ll.l 1730 29 n 7.8 -7.8 • -3.3 1800 30 « 1.7 -5.0 - 0 1830 1 December -5.0 -2.2 1830 2 ii 16.7 -6.1 — -1.7 2030 3 ii -2.8 -6.1 - -0.6 2230 ti 4 1.7 -7.2 - -7.2 2230 16 ii 23.9 -6.7 - -7.2 1400 18 n 20.0 -3.3 - -5.6 1000 33

TABLE 4* Number of occasions of wind from different directions, Lake Vanda, Wright Valley, December 1958 to January 1959 (Bull, 1966)

Time Direction 0001 0600 1200 1800

Easterly 30 24 18 21 Westerly 8 9 14 14 Calm or <5 kt 3 6 5 5 Indet erminate 11 13 15 11 shallow katabatic winds, originating on the ice plateau west of Wright Valley. The air masses remain in close contact with ice until they reach the eastern end of the Wright Upper Glacier at an elevation of 1300 m. While in contact with the ice, the temperature of the air mass remains low and its relative humidity high. During its rapid descent into the valley, the air is heated by adiabatic compression and the specific humidity does not Increase appreciably. Relative humidity of the westerly winds consequently is low. During a six-week period of observation, the relative humidity averaged 45% at a mean temperature of 2.2°C; extremes were about and about 60>6 (Bull, 1966). Easterly winds at Lake Vanda blow from McMurdo Sound. For December 1958 and January 1959* wind at Marble Point was dominantly southeasterlyf mean temp­ erature was about -2.7°C and relative humidity about 75J* (Bull, 1966). As the air mass moves westward, it is cooled by adiabatic expansion in crossing the Wilson Piedmont Glacier, with an elevation of 600 m, and often becomes saturated. Thus, low cloud cover over the glacier is prevalent when the wind is easterly. Representative air temperature and relative humidity data from the eastern part of Wright Valley are In table 5* Although mean annual humidity for the valleys is not known at the present time, the value for Wright Valley may be 3?

TABLE ?• Air temperature and relative humidity for the eastern part of Wright Valley. 0*8 km from Wright Lover Glacier (Ugollni, 1965)

Temperature, °C Relative Humidity Date Max. Min. Max. Min. 7 November -7.0 -8.0 70 30

8 November -5.5 -11.0 65 32 22 December 1.5 -7.0 53 25 19 January 0.0 -5.0 70 30 36 estimated. Since crystals of CaCl2*6H20 occur at the edges of Don Juan Pond in the South Fork, it has been deduced that the mean relative humidity is less than 40£ (Bell, 1966). Considering the frequent katabatic winds that blow off the ice plateau, Ragotzkie and Likens (1964) have estimated that the relative humidity at Lake Vanda would be about 13£« Wind velocities at Lake Vanda for the 1958-1959 observation period are in table 6. Air temperatures at Lake Vanda were higher with westerly winds than with easterly winds (table 7)• For both easterly and westerly winds, diurnal variation was about 3°C (Bull, 1966). Summer cloud cover for the ice-free valleys is consistently lower than for coastal parts of McMurdo Sound (table 8). Low cloud cover is associated with easterly winds, especially in Wright Valley* The high cloud cover usually originates in the south or southwest (Bull, 1966). Precipitation in the ice-free valleys Is light and falls as snow, although on one occasion rain has been observed at Lake Vanda (R* E, Behllng, 1969, personal communication). Most precipitation is derived from the east* At the coastward ends of the mountain ranges bordering the valleys, cirques are occupied by small glaciers. At the inland end of the ranges nearly all 37

TABLE 6. Mean vlnd velocities in knots observed at Lake Vanda. Wright Valley, Summer 1958-1959 (Bull, 1966)

Time 0001 0600 1200 1800

All observations 11.4 8.0 11.3 14.7 With easterlies 11.5 8.2 11.7 18.3 With westerlies 12.8 10.7 16.0 13.4 38

TABLE 7* Mean temperatures (°C) under different wind conditions. Lake Vanda, Wright Valley, Summer 1958-1959 (Bull, 1966)

Time 0001 0600 1200 1800

All observations -0.8 -0.8 2.2 1.6

Westerly winds 1.4 1.0 3.3 4.1

Easterly winds -1.4 -1-7 1.3 -0.1 39

TABLE 8. Sumner cloud cover for Ice-free areas and for Hut Point, Ross Island (Bull, 1966)

Total Low Period Year Place Cloud Cloud Oct. and Nov. 1961 Victoria Valley 4 4 18

Oct. and Nov. 1961 Hut Point 59 Dec. 1961 Victoria Valley 63 44 Dec. 1961 Hut Point 72 Dec. 1959 Victoria Valley 38 20

Dec. 1959 Hut Point 85 Dec. 1958 Wright Valley 67 33 Dec. 1958 Marble Point 60 30 Dec. 1958 Scott Base 62 Jan. 1962 Victoria Valley 4 0 22 Jan. 1962 Hut Point 76

Jan. I960 Victoria Valley 38 18

Jan. I960 Hut Point 4 3 Jan. 1959 Wright Valley 46 17 Jan. 1959 Marble Point 56 15 Jan. 1959 Hut Point 58 Feb. (part) 1962 Victoria Valley 28 16 Feb. 1962 Hut Point 74 40 the cirques are empty. Thus, meltwater that enters Lake Vanda is derived almost exclusively from the alpine glaciers in the eastern part of the valley and from Wright Lower Glacier. The actual amount of precipitation in the ice-free valleys can only be estimated at present. Available data on the amount of snowfall in the valleys and at Hut Point, Ross Island, are given in table 9* The ice-free valleys receive less total precipitation than the coastal areas. On the Wilson Piedmont Glacier at 450 m, annual snow accumulation is about 6 cm of water. At the western end of Wright Valley, annual snowfall is less than this and most of it is removed quickly by deflation and subli­ mation (Ugolini and Bull, 1965)* Although most of the surface of Antarctica can be classed as accumulation area (net accumulation of snow each year), the ice-free valley regions represent net ablation areas. Annual net ablation as high as 5° cm of water has been measured along the coast of East Antarctica, but the ablation (sublimation and melting) rate falls off rapidly with increasing altitude of the ice surface. At a height of 456 m, the ablation rate is only about one-third of the sea level rate (Mellor, 1961). The ablation rate also varies markedly during the year (figure 11). 4 1

TABLE 9. Days and total amount of snowfall in the ice- free areas and at Hut Point , Ross Island (Bull, 1966)

Days Amount Snow Period Year Place (Water Equiv.) Oct, 22- 1961 Victoria Valley 2 0.8 cm at Nov. 30 1000 m altit. Nov, 1961 Hut Point 0.3 cm Dec, 1961 Victoria Valley 6 1.4 cm Dec. 1961 Hut Point 1.1 cm

Dec, 1959 Victoria Valley 8 1.0 cm Dec. 1959 Hut Point 2 . 5 cm Dec. 10-31 1958 Wright Valley 5 1.3 cm at 1200 m altit.

Dec. 10-31 1958 Marble Point 7 4.0 cm Jem, 1962 Victoria Valley 6 0.6 cm at 600 m altit.

Jan. 1962 Hut Point 1.5 cm Jan. i960 Victoria Valley 7 Jan. 1959 Wright Valley 3 0,3 cm Jan* 1959 Marble Point 0 4 2

•IFM AM JJuAuS OND Month

Figure 11. Ablation from a snow-free ice surface 62 m above sea level near Mavson, 6?°36 S, 62 ?2 E (froe Mellor, 1961).

i 43 The annual ablation rate for Lake Fryxell, an Ice- covered lake situated in lower Taylor Valley at 22 m is 30 to 40 cm (Henderson, &£ &1*, 1966). Angino and others (1965) has estimated the sublimation rate of Lake Vanda to be about 75 * 15 cm per year. Lake Balham, in the Victoria Valley system, had dropped 170 cm in level between the summers of 1958-1959 and 1961-1962 (Bull, 1966)* Once the altitude of the ice plateau dropped sufficiently that rock thresholds began to limit the flow of outlet glaciers, strong katabatic winds, low rates of precipitation, and relatively high ablation rates have served to create the ice-free areas. These factors, plus a high net balance of radiation, combine to maintain the existence of the ice-free valleys.

P. Geology of Wright and Taylor Valleys

The geology of Wright and Taylor Valleys has been studied in detail by several investigators (e.g., Angino, et &!•, 1962b$ Gunn and Warren, 1962; Hamilton and Hayes, I960; Harrington, 1958; Haskell, al.. 1965; McKelvey and Webb, 1962; and Webb, 1963a). A brief outline of the geology is presented here to describe the type of bedrock in the ice-free valleys. The rock formations of Wright Valley, as defined by McKelvey and Webb (1 9 6 2 ), are listed in table 10. The TABLE 10. Chronology of the Wright Valley area, after McKelvey and Webb (1962). Names in parentheses are equivalent units in Taylor Valley.

AGE GROUP FORMATION

Quaternary Upper Tertiary McMurdo Volcanics

Victoria Orogeny Cretaceous Jurassic Ferrar Dolerites Mid-Mesozoic (Jurassic) Beacon Sandstone Mid-Paleozoic (Devonian) Kukri Peneplain Lover Paleozoic Victoria Intrusives Vanda Lamprophyre and Porphyry Vida Granite (Irizar Granite) Wright Intrusives Theseus Granodiorite (Granite Harbor Loke Microdiorite Intrusive Complex) Dais Granite (Larsen Granodiorite) Upper Cambrian Olympus Granite-Gneiss Ross Orogeny Cambrian Skelton Group Precambrian Asgard Formation *5 geology of Taylor Valley Is similar to that of Wright Valley. Although the same units occur in Wright and Taylor Valleys, they have been assigned different names. Where this occurs, the alternate names given to these units by Haskell and others (1965) are included in the parentheses to correlate then with the units defined earlier by Gunn and Warren (1962). The geology consists of a basement complex of metasediments, granite-gneiss, granite and associated dikes unconformably overlain by the Beacon Group of sedimentary rocks. The basement complex and Beacon rocks have been intruded by dikes and sills of diabase. The oldest rocks of Wright Valley includes more than 5000 m of tightly-folded metamorphosed sedimentary rocks named the Asgard Formation (McKelvey and Webb, 1962). This formation is the equivalent of the Skelton Group elsewhere in Victoria Land (Gunn and Warren, 1962), and consists of interbedded marbles, hornfels (sic), and schist, with a complex history. According to Gunn and Warren (1962), a minimum of 7600 m of limestones, dolomites, calcareous mudstones, and rare quartzltes and arkoses were deposited in a shallow, marine environ­ ment (Boss Geosyncline). Deposition of at least 2750 m of calcareous sandstones, slltstones, and arkoses may have preceded or followed deposition of the limestones. The sediments .^ay ha^e undergone folding along 46 general north-south axes during the Mavson Orogeny 600 to 800 m.y. ago, and the Boss Orogeny 400 to ??0 m.y. ago (Angino and Turner, 1964). The folding was accompanied by emplacement of Olympus Granite-Gneiss and by regional raetamorphism. In Wright and Taylor Valleys, the metasediments are flanked on the east and west by the granite-gneiss. Although the contact is generally gradational, it is thought to be intrusive because inclusions of Asgard schist occur within the gneiss and because the alignment of the inclusions is parallel to the schistosity of the Asgard Formation (McKelvey and Webb, 1962). The age of the Olympus Granite-Gneiss is not certain. Zircons from this unit in Victoria Valley (figure 6) have been dated by the U-Pb method (Deutsch and Grogler, 1966). Their data, when plotted on a concordia diagram, indicate an age of 610 m.y. Accordingly, the Olympus granite- gnelss is probably late Precambrian to early Cambrian in age. The Asgard Formation and the Olympus granite-gneiss in both Wright and Taylor Valleys are in turn flanked by the Dais granite; this is equivalent to the Larsen granodiorite. In Wright Valley, this unit forms the coastal and inland limits of the exposVd basement complex. It has a coarse foliation generally parallel to the Asgard Formation and the Olympus granite-gneiss. 47 Its exact relationship to the Olympus granite is not clear. Loke microdiorlte intrudes Olympus granite-gneiss. The Dais granite, Olympus granite, Loke microdiorite, and Asgard metasediments are intruded by dikes of Theseus granodiorite. The Victoria intrusives, as defined by McKelvey and Webb (1962) comprise all intrusives of the basement complex younger than the Theseus granodiorite. The Vida granite is the principal representative of the Victoria intrusives. It is the equivalent of the Irizar granite of Taylor Valley. In Wright Valley it forms light-colored sill-like bodies particularly evident in the western part (see figures 2 and 3)* The Vanda lamprophyre and porphyry dikes intrude all other rocks of the basement complex. Vanda porphyry dikes have been found to be 470 m.y. old by the Rb-Sr method (Jones and Faure, 1967)* One sample of these dikes had been dated by Deutsch and Webb (1964), who reported an anomalous date of 1000 m.y. Work by Jones and Faure (1967), however, indicated that the 1000 m.y. date may be the result of contamination of the dikes during intrusion. Rubidium-strontium analyses of Olympus granite- gneiss, Dais granite, and Vida granite indicate an age of 490 * 20 m.y. for all three units (L. M. Jones, unpublished studies). This suggests that either all three units were emplaced within a relatively small 4 8 interval of time or that the Olympus granite-gneiss and the Dais granite are older, but their dates reflect intrusion of the younger Vida granite. The latter interpretation is supported by the zircon date of 610 m.y. for the Olympus granite (Deutsch and Grogler, 1966). The Beacon Group, which lies unconformably on the basement complex, attains a maximum thickness of more than 1550 m in the region of Wright and Taylor Valleys (Webb, 1963a). The rocks are mainly subgreywacke, arkose, and orthoquartzite, which were deposited in fluvial, lacustrine, and paludal environments. In the Beacon Valley region to the southwest, the Beacon Group rocks consist largely of light-colored medium- to coarse­ grained sandstone (Hamilton and Hayes, 1963). Rocks of the Beacon Group and the basement complex were intruded by dikes and sills of Ferrar dolerite during the Jurassic period (McDougall, 1963). Individual sheets of diabase often attain thicknesses up to 460 m, with total thicknesses up to 1550 m (Hamilton and Hayes, 1963). Regional block faulting (the "Victoria Orogeny") occurred during upper Tertiary and Quaternary time (Gunn and Warren, 1962), giving rise to the present day mountains of Victoria Land. As a result of this uplift, the peneplain surface, Beacon Group, and Perrar dolerite sil]s dip gently (6-10°) to the west and

southwest. Volcanic activity occurred during the Quaternary in the ice-free valleys. Mount Erebus, which dominates Ross Island, is presently active. This region of the Transantarctic Mountains is part of a larger region of recent volcanic activity that extends as far north as Cape Adare and southward to Mount Early in the Scott Glacier region. CHAPTER III

PfflfflSAI. MM PROPERTIES 2E U £E IMMA, WMW VALLSX

A. Introduction

Lake Vanda occupies a closed U-shaped bedrock basin in the western portion of Wright Valley, about 47 km from the coast (figures 2, 4, and 12). It is about 8*5 km long, 2*4 km wide, and attains a maximum depth of about 67 m (Angino, ei fil*» 1965; and Torii, si al.. 1967). The lake is perennially ice-covered, with a range in ice thickness of 3*4 m (Armitage and House, 1962) to about 4*2 m (Angino, si si*» 1965). During winter, the ice cover does not increase in thickness appreciably. During summer, air and ground temperatures rise suffi­ ciently so that a meltwater moat forms at the edges of the lake. The width of the moat varies considerably, ranging up to 10 m. Where the Onyx River enters the lake at its eastern end, an ice-free area up to 0*5 square kilometer may form If seasonal temperature is sufficiently high

(C* Bull, 1969, personal communication). Lake Vanda has been more extensive in former times. 50 51

Figure 12* Lake Vanda, Wright Valley, view west toward North Fork and the Wright Upper Glacier. Bull Pond is in the foreground, through which the Onyx River flows before discharging Into Lake Vanda. Strand lines are visible on the north edge of the lake. 52 A well-defined series of horizontal, raised strand lines extend up to about 56*5 m higher than the present lake level (Nichols, 1965)* These are shown in figure 12. Less well-formed strand lines extend to about 70 m above the lake (C. Bull, 1969* personal communication). When the lake was about 2 a higher than present, it extended for some distance into the North Fork of Wright Valley (figure 5), A series of water samples along a depth profile of Lake Vanda was collected by Dr. Derry D, Koob (Department of Botany, The Ohio State University) during the 1965-1966 austral summer, A Van Dorn sampling apparatus was used for the collecting. These samples were obtained at one location at the center of the main lobe of the lake, where the maximum depth was 60 m. Unless otherwise indicated, physical, chemical, and Isotopic data for Lake Vanda used in this discussion pertain to this series of water samples and were determined by the writer,

B. Density

Density of the water samples from Lake Vanda was determined by weighing a known volume of water. The method is described in Appendix A. All densities were normalized to 20°C for comparison within the series of samples (table 11). TABLE IX. Density profile of Lake Vanda, Wright Valley

Depth helov Density, surface, m g/ml at 20°C

4 0.9983 0.9981 5 0.9986 6 0.9978 7 0.9987 8 0.9984 9 0.9986 10 0.9981 11 0.9984 12 0.9985 13 0.9982

14 0.9991 15 0.9990 20 0.9987 25 0.9990 30 0.9988

35 0.9985 40 0.9992 1.0001 46 0.9996 48 1.0009 50 1.0227 52 I.0306 54 1.0490 56 1.0636 58 1.0737 54 The water in Lake Vanda Is density stratified (figure 13). The density of the water Immediately below the ice to a depth of about 14 m is essentially constant and Identical to that of pure water. At 14 m, the density Increases slightly, but then remains constant to about 44 m. There Is a second small increase in density to 48-50 m depth. The density then increases continually to 60 m, the bottom of the lake, where a density of

1*08 g/nl was measured. A density of 1.10 g/ml has been recorded for the maximum depth of 67+ m (Torii, et al., 1967). The increase in density with depth is directly related to the amount of dissolved salts in the water. Although no numerical values of salinity have been calculated, the salinity at maximum depth has been estimated to be about four times that of sea water

(Yamagata, gi * 1967). Lake Vanda is also compositionally stratified, with the chemocline pattern coinciding with that of the density profile. Concentrations of cations and anions not only change with depth, but the overall composition of dissolved salts changes with depth. Figaro 13* Density profile Lake Vanda, Wright Valley „ DEPTH, meters 0) m * 57

C. Temperature

The temperature profile of Lake Vanda Is shown In figure 14* The temperature data were obtained by Torii and others (1967)* Two sharp chemoclines are indicated, one beneath the Ice cover and the other at about 50 m depth, A third, less pronounced thermocllne occurs at 14 nu Although a minor change at 66 m was reported by Angino and others (1965), its presence has not been confirmed by later investigators (including Hoare, 1966; Hagotzkle and Likens, 1964; Torii, e£ al*, 1967; and Wilson and Wellman, 1962). The temperature of the uppermost thermocllne is about 2°C, the intermediate one is about 7°C, and the lowermost, about 15°C. Locations of the thermoclines closely correspond to those of the density-layer boundaries (compare figures 13 and 14). Temperature data obtained from early spring to late summer suggest that the lake is thermally stratified throughout the year. The horizontal distribution of temperature is surprisingly uniform, even near the edges of Lake Vanda (Hoare, 1968; Ragotzkle and Likens, 1964; and Torii, et al.. 1967)* Although Angino and others (1965) reported the existence of a thermal dome in the western portion of the lake, this has not been confirmed by other investigators. There is a slight tendency for the ?8

Figure 14 • Temperature profile of Lake Vanda, Wright Valley (from Torll and others, 1967)* DEPTH, meters 20 40 30 60 50 0 5 EPRTR, *C TEMPERATURE, 10 15 20 25 60 5, 6, 7* and 8°C isotherms to dip a meter or less in the central region of the lake. However, below 17 m all Isotherms are horizontal within limits of detection (Ragotzkle and Likens, 1964). Upward bending of the upper isotherms at the edges may be due to solar heating of shallow water near shore. Despite the inverted temperature structure (i.e., temperature Increasing with increasing depth), the brine layers are dynamically stable because of the increasing salinity and density of the water with depth. Temperature measurements made at small depth Intervals above 50 m have revealed the presence of numer­ ous smaller thermoclines (Hoare, 1966, 1968; and Shirt- cliffe and Calhaem, 1968). They can be traced for at least 4 km horizontally and apparently can retain their identity for a year or longer (Hoare, 1968). The dense water at the bottom of Lake Vanda is considered to be non-convective (Hoare, 1966, 1968; Wilson, 1964; and Wilson and Wellman, 1962). The upper layers are described as convective cells, and the fine thermal structure reported by Hoare (1968) is attributed to dynamic equilibrium within these layers. 61

D. Source of the Heat

The source of the heat In Lake Vanda is still controversial. Possible sources are: (1 ) high geo­ thermal gradient in Wright Valley (Angino, gt al., 1965; Goldman, e£ &1., 1967; Nichols, 1962; and Hagotzkle and Likens, 1964); (2) discharge of thermal springs (Angino, et al.. 1965; and Nichols, 1962); (3) absorption of solar radiation (Angino, et al., 1965; Hoare, 1966; and Wilson and Wellman, 1962); (4 ) chemical heating (Wilson and Wellman, 1962); and (5) biological activity within or at the bottom of the lake (Wilson and Wellman, 1962). Since Lake Vanda lies within the McMurdo volcanic province and basaltic cinder cones are to the east of the lake in Wright Valley, postulation of a high geo­ thermal gradient is reasonable. However, the geothermal gradient has not been determined for any location in the ice-free valleys, and measurements of the temperature gradients in sediments at the bottom of Lake Vanda are contradictory and therefore inconclusive (Ragotzkie and Likens, 1964; and Wilson and Wellman, 1962). Nichols (1962) also suggested that the heat may result from pyroclastics, sublacustrine flows, fumaroles and meltwater heated by volcanic activity. The inflow of hot springs reported by Angino and others (1965) has not been confirmed. Wilson and Wellman (1962) reported 62 that there is no geological evidence for such thermal activity in the area surrounding the lake. Contribution of heat to Lake Vanda by biological activity cannot be evaluated at the present time. The activity is limited principally to algae and bacteria detected throughout the lake (Goldman, fi£ fii* t 1967). There is a remarkably Increased level of photosynthesis in the warm bottom waters (productivity data are given in detail by Goldman, e£. al.. 1 967). Presence of organic material in the bottom sediments is indicated by relatively large amounts of H 2S (E. E. Angino, 1967, personal communication). Four samples of water below $8 m were found to have H 2S concentrations of 1.3&, 1*53* 3*40, and 27*4 ppm (Yamagata, fit fil., 1967). A trace of H 2S was also detected by the writer in a sample of lake sediment four months after collection. This sediment had been collected in about 30 m of water in the eastern lobe of the lake by D. Greegor in February, 1 9 6 8 * Precipitation of salts has been considered to be a source of heat but negative heats of solution of the expected salts rule out any heat contribution by this source (Wilson and Wellman, 1962). There is no evidence for other chemical reactions that could produce the observed heat. P. Galkin (1969, personal communicat inn) has observed that calcium sulfate is present in indurated sediments of Lake Vanda. I. Friedman (1969, written communication) has collected "strontium-carbonate precipitate" (perhaps calcium carbonate?) from these sediments. This suggests that absence of solid salts at the bottom of Lake Vanda reported by Wilson and Wellman (1962) may refer to loose sediment only, while below this material relatively large quantities of salts may exist but were not detected by gravity sampling techniques previously used. Absorption of solar radiation within the lake may be an important source of heat (Hoare, 1968; and Wilson and Wellman, 1962). There is continual daylight during the Antarctic summer. The ice cover of the lake, with the exception of the surface layer, is relatively free of air bubbles and is therefore transparent to solar radiation. Hence, a large fraction of incident radiation penetrates the ice and enters the water (Goldman, ££ al.. 1967). The water is so clear that some light can pene­ trate to the bottom of the lake. Thus, heat at depth could be attributed to deep penetration of solar radi­ ation, absorbance mainly in the bottom layer, and reten­ tion of heat by the non-convective nature of this layer. This mechanism has been shown to be the source of heating in Lake Bonney, Taylor Valley by Shirtcliffe (1 9 6 4 ) and Shirtcllffe and Benseman (1964). In summary, the actual source of heat in Lake Vanda 64 is still not known. Absorption of solar radiation and/or a high geothermal gradient are the most likely source(s). Until the geothermal gradient under the lake has been determined, this problem remains unanswered.

E. Chemical Composition of Water in the Onyx River

The Onyx River is the major supply of water to Lake Vanda (figures 15 and 16). It flows intermittently and only during the warmest part of the summer, from early December to mid-January. The partial chemical composition of five samples of the river are given in table 12, and the sample locations are shown in figure 17• Concentrations of the principal ions in the Onyx River vary considerably, some within an order of magni­ tude. An extensive chemical study of the Onyx River has not yet been carried out and, therefore, the seasonal variation of concentrations of major ions in the water is not known. On the basis of available data, the concen­ tration of the major ions appear to Increase downstream (table 12). Since the samples were collected over a period of several days, this characteristic has not been

definitely established. Concentrations of the principal cations of water of the Onyx River are relatively high compared to the mean 65

Figure 15. Onyx River, Wright Valley, view toward the west. Lake Vanda is beyond the bend in the valley. Asgard Range is at the left and center background. (Photograph courtesy of D. D. Koob.) 66

Figure 16. Onyx River, Wright Valley, flowing from left to right toward Lake Vanda beyond. Asgard Range is in the background. (Photograph courtesy of R. J. E. Montlgny.) TABLE 12. Partial chemical analyses of aeltvater from Wright Valley. Concentrations are given in ppm.

Date, Saaple No. 1966 Location Ca Mg sr Na K

DV-66-002 Nov. 18 Foot, Lover Wright Glacier 2.5 0.21 - 2.3 0.72

DV-66-012 Dec. 5 Onyx River, below Denton 2.6 0.19 - 7.6 0.59 Glacier

DV-66-013 Dec. 5 Onyx River, between Denton 4.2 0.78 - 9.7 0.99 and Goodspeed Glaciers

DV-66-009 Dec. 1 Onyx River, below Meserve 15.7 1.15 0.0194 30.2 3.50 Glacier

DV-66-010 Dec. 4 Onyx River, below Bull Pass 38.1 3.07 - 30.5 4.30

DV-66-016 Dec. 7 Onyx River, between Bull 43.0 3.05 - 44.6 6.20 Pass and Lake Vanda

DV-66-005 Nov. 25 Bartley Glacier 3*3 0.30 - 3.4 0,48 DV-66-OO6 Nov. 28 E. Lake Vanda, pond on ice 2.1 - - 2.18 1.03

DV-66-007 Nov. 28 N. Lake Vanda, pond on ice 2.3 - - 2.80 1.08

DV-66-011 Dec. 5 Stream below Clark Glacier 6.2 - 10.2 0.95

O' TABLE 12. Continued

Sample No. ^9^6*______Location______Ca Mg______Sr Ha K

DV-66-017 Dec. 12 Stream, Meserve Glacier 5.3 1.38 * 8.3 0.80

DV-66-025 Dec. 19 Stream, Meserve Glacier 2.2 0.47 - 4.3 0.72

2-GH (1968) Stream, Meserve Glacier 3.51 - - 1.21 0.44

V-4 Jan. 18 Lake Vanda, 4 m depth 38.5 2*97 0.141 31.9 9.98 v-5 Jan. 18 Lake Vanda, 5 m depth 64.? 5.01 0.226 45.1 12.5

e - - Rivers, World Average 15 4.1 - 6.3 2.3

^Livingstone, 1963*

O' 00 Figure 17* Location of water, ice, and soil samples in Wright Valley. 70 composition of river waters of the world (Livingstone, 1963). Values of the corresponding cation concentrations from the compilation of the world average composition of rivers are listed in table 12. Cation ratios for Ca/Kg, Ca/Na, Ca/K, and Na/K are given in table 13, With the exception of sample DV-66-OI3 , Ca/Mg ratios of the Onyx River are similar to those in the upper layer of Lake Vanda* Data for the 4 and 5 m depths of Lake Vanda have been added to tables 12 and 13 for comparison. The relationships concerning the Ca/Na and C a / K ratios are inconclusive. The N a / K ratios for the Onyx River are high, ranging from 7»1 to 1 3 , while those values for the 4 and 5 m depths of Lake Vanda are 3*2 and 3 .6 , respec­ tively. This suggests that there has been some modi­ fication of the relative sodium and potassium contents, if Indeed water just under the ice cover represents recent inflow from the Onyx River. Na/K ratios for the two samples of meltwater from the surface of the ice cover of Lake Vanda are 2.1 and 2.6 (table 13). These values, compared to the higher values for the Onyx River, suggest that as water froze at the base of the ice cover, sodium was excluded to a greater degree than was potassium. This process is discussed in more detail in the following section. It is not possible with the presently available samples to explain the Na/K ratios in the upper waters of Lake Vanda, compared to TABLE 13* Ratios of Ca/Mg, Ca/Na, CaA» and N a A for aeltester in Wright Valley.

Sample No. Location Ca/Mg Ca/Na C a A N a A

DV-66-002 Foot, Lower Wright Glacier 11.9 1.1 3.5 3.2

DV-66-012 Onyx River, below Denton 14.7 0.37 4.8 13 Glacier

DV-66-013 Onyx River, between Denton and 5.4 0.43 4.3 9.8 Goodspeed Glaciers

DV-66-009 Onyx River, below Meserve 13.6 0.52 4.5 8.7 Glacier

DV-66-010 Onyx River, below Bull Pass 12.4 1.3 8.8 7.1 DV-66-016 Onyx River, between Bull Pass 14.1 0.97 6.9 7.2 and Lake Vanda

DV-66-005 Bartley Glacier 11.9 0.97 6.9 7.1

DV-66-006 E. Lake Vanda, pond on ice - 1.0 2.0 2.1

DV-66-00? N. Lake Vanda, pond on ice - 0.82 2.1 2.6

DV-66-011 Stream below Clark Glacier _ 0.61 6.5 11 TABLE 13, Continued

Sample No. Location Ca/fcg Ca/Na Ca/K Na/K

DV-66-017 Stream, Meserve Glacier 3.84 0.64 6.6 10

DV-66-02? Stream, Meserve Glacier 4.67 0.51 3.1 6.0 2-GH Stream, Meserve Glacier - 2.8 8.0 2.8

V-4 Lake Vanda, 4 m depth 13.0 1.2 3.9 3.2 v-5 Lake Vanda, 5 n depth 12.8 1.4 5.2 3.6

- Rivers, World Average* 3.66 2.4 6.5 2.7

* . Livingstone, 1963* 73 those of the Onyx River. The Na/K ratios for the Onyx River appear to decrease downstream (figure 18). At the eastern end of the valley the Na/K was 13f decreasing to 7*2 at the base of Bull Pass. However, the random sampling schedule limits the interpretation of these data. If this decreasing trend in the Na/K ratio does continue downstream, the value of the river just before it discharges into the lake might be close to the Na/K ratio in the upper water of Lake

Vanda• The Meserve Glacier lies about midway in the drainage basin of Lake Vanda, as shown in figure 17. Since the relative amount of salts in precipitation tends to decrease Inland, the chemical composition of the ice from the upper surface of the Meserve Glacier might be considered to represent to a first approximation an average composition of ice and snow in the drainage basin of the lake. A series of samples of meltwater, neve, and ice from the Meserve Glacier were obtained from G, Holds- worth. Neve samples were collected within the accumu­ lation basin of the glacier. Ice samples were collected both from the surface and within the glacier. A tunnel ?4 m In length has been excavated normal to the eastern edge of the Meserve Glacier about 300 m above the snout (figure 4). Horizontal drifts were cut opiiBA M i n —JOO Distance Distance in km from Wright Low* Glocior

o CM O M o

Figure 18. Variation of the Na/K ratio with distance along the Onyx River, Wright Valley. 75 normal to the main tunnel at the 20, 48, and 54 m posi­ tions. At the base of the glacier is a layer of "dirty" ice, designated the "amber" layer by Holdsworth. Its thickness is fairly uniform, varying between 45 and 60 cm. At the entrance to the tunnel, there is a wedge of clear ice under the amber layer, about 40 cm thick at the outside edge. It narrows under the glacier until it disappears approximately 10 m from the cliff edge. Bands of dirty ice within the clear ice overlying the amber layer were encountered during the tunneling, and some of these bands were sampled in addition to the basal amber layer and clear Ice. Samples were selected from the surface, from clear ice at depth, and dirty bands. Meltwater samples were collected near the tunnel entrance from a stream the flows near the base of the glacier during the summer. Partial chemical analyses of neve, ice, and melt­ water samples from the Meserve Glacier are given in table 14. It can be seen that the cation content of these samples are low compared to that of the Onyx Elver (table 12). Ice samples collected from the amber layers have higher salt contents than clear ice from both the top surface and the base of the glacier.

Major cation concentrations are low in the three TABLE 14* Partial chemical analyses of neve, ice, and meltvater from the Meserve Glacier. Concentrations are in ppm.

Sample Ho. Description Ca Mg Na K

1 Neve, accumulation basin 0.29 _ 0.61 0.15

14 Nevl, accumulation basin 0.48 - 0.54 0.45

M-ll Neve, accumulation basin O.29 - 0.79 0.29

DV-66-017 Stream, base of glacier 5.3 1.38 8.3 0.80

DV-66-025 Stream, base of glacier 2.2 0.47 4.3 0.72 2-GH Stream, base of glacier 3.51 - 4.4 0.55 3 Ice, surface of glacier 0.30 - 1.21 0.44

M-T Ice, surface of glacier O.38 0.24 0.43 0.15 10 Ice, clear, base O.36 - 0.96 0.34

11 Ice, clear, base 0.33 - 1.40 0.85 12 Ice, clear, base 1.03 - 1.64 O .36

48-100-150 Ice, clear, base 0.33 - 0.33 0.01

5 Ice, clear, near amber layer 1.48 • 1.32 0.41 TABLE 14. Continued

Sample No. Description Ca Mg Na K

13 Ice, clear, near amber layer 0.91 - 0.80 0.13 K-B Ice, amber layer 1.9 0.66 4.36 0.69

A- 3*4 Ice, amber layer 2.43 0.66 8.20 1.24

6 Ice, amber layer 5.08 - 9.6 1.50

7 Ice, amber layer 6.76 - 10.0 1.82 8 Ice, amber layer 6.94 - 7.7 1.68

9 Ice, amber layer 4.90 - 5.3 1.50

48-0-20 Ice, amber layer 5.19 - 9.1 1.65

48-20-45 Ice, amber layer 5.20 - 3-8 1.50 78 samples (Dy-6 6 -0 1 7 , DV-66-025, and 2-GH) of the meltwater stream Issuing directly from the glacier (table 14), In this series of samples, the calcium concentration ranged from 2.2 to 5>3 ppnu The range for sodium was 1.21 to 8.3 ppm, and for potassium, 0.44 to 0.80 ppm. These concentrations are also similar to those In the ice samples listed in table 14, in particular, those from the amber layers. The salt content of clear Ice and neve Is less than that in the amber layers. Since the meltwater appears to be derived mainly from clear ice, while its salt content is similar to that of amber ice, the stream had to pick up additional solute sometime between melting and sampling. If the solutes in the meltwater stream were derived directly from the ice, concentration ratios might indicate this relationship. Concentration ratios of the principal cations of stream samples, neve, and ice are given in table 15* Comparison of these ratios, how­ ever, provides inconclusive evidence for the Increase in salinity from clear ice to the meltwater stream. Since the meltwater had been in contact with soil and bedrock prior to sampling, the additional salts may be derived from salt accumulations on the ground. The concentration of the principal cations in the Onyx River, especially in those samples collected down- valley from the Meserve Glacier, are much higher than TABLE 1?. Concentration ratios of Ca/Na, Ca/Mg, Ca/K, and Na/K for neve, Ice, and meltvater from the Meserve Glacier.

Sample No* Description Ca/Na Ca/Mg Ca/K N a A

1 N6ve, accumulation basin 0.48 - 1.9 4.1

14 N6v6 , accumulation basin 0.89 - 1.1 1.2

N-II N6v6 , accumulation basin O.38 - 1.0 2.7

DV-66-017 Stream, base of glacier 0.64 3.8 6.6 10

DV-66-025 Stream, base of glacier 0.51 4.7 3.1 6.0

2-GH Stream, base of glacier 0.79 - 6.4 8.0

3 Ice, surface of glacier 0.25 - 0.68 2.8 M-T Ice, surface of glacier 0,88 1.6 2.5 2.9

10 Ice, clear, base 0.38 - 1.1 2.8

11 Ice. clear, base 0.24 - 0.39 1.7

12 Ice, clear, base 0.63 - 2.9 4.6

48-100-150 Ice, clear, base 1.0 - 33 33

5 Ice, clear, near amber layer 1.1 - 3.6 3.2 TABLE 15* Continued

Sample Ho* Description Ca/Na Ca/Mg Ca/K Na/K

13 Ice, clear, near amber layer 1.14 - 7.0 6*2 M-B Ice, amber layer 0.44 2.9 2.8 6.3 A-3.4 Ice, amber layer 0.30 3.7 2.0 6.6

6 Ice, amber layer 0.53 - 3.4 6.4

7 Ice, amber layer 0.68 - 3.8 5.5 8 Ice, amber layer 0.90 - 4.1 4.6

9 Ice, amber layer 0.92 - 3.3 3.5

48-0-20 Ice, amber layer 0.57 - 3.1 5.5

48-20-45 Ice, amber layer 1.37 3.5 2.5

00 O 81 those of the glacial meltwater stream fed by the Meserve Glacier (table 12). The Increase In concentration of the major cations Is considerable. Calcium concentrations are higher in the Onyx River by almost an order of magni­ tude. The sodium content is higher in the river by factors of 4-40, and potassium is higher by factors of 4-10. Because no samples were collected between the glacier and the Onyx River, an Increase in salinity of the melt- water stream with distance could not be demonstrated. If the chemical composition of the snow and ice of the Meserve Glacier is representative of snow and ice from Wright Valley (and this is reasonable on the basis that meltwater issuing directly from the Wright Lower Glacier and Bartley Glacier have correspondingly low concen­ trations, as listed in table 12), then another source must contribute the high concentrations of these ions in the Onyx River. A sample of the meltwater stream of the Goodspeed Glacier collected about 200 m below the glacier con­ tained higher cation concentrations than those in the snow and ice of the nearby Meserve Glacier (table 12). Assuming that the composition of ice in the Goodspeed Glacier is very similar to that of the Meserve Glacier, it appears that there has been an abrupt increase In solute content in the water after it flowed over the 82 bedrock and soil for a distance of only 200 m. This suggests that the meltwater stream had dissolved salts from the soil, since these salts are found everywhere in the valley. On the basis of these relatively few samples, it is suggested that as snow and ice melt, their salt content is increased upon contact with soil and bedrock, which contain lenses and encrustations of salts. These streams feed into the Onyx River, whose total salt concentration Increases downstream, perhaps by additional dissolution of salts found in and around its stream bed. By the time the Onyx River finally discharges into Lake Vanda, the salt content has attained its maximum level. In order to resolve the problem of this increase in the salt content of the Onyx River and the apparent increase of salinity downstream, a series of water samples should be collected along the length of the river and along one or more tributary streams during a small time interval.

P. The Chemical Composition of Water in Lake Vanda

1. Introduction

Concentration of the principal cations and anions in Lake Vanda increase with depth in a manner similar to the density and temperature profiles. New analyses for calcium, strontium, sodium, and potassium are reported here for samples collected at intervals from the surface to the bottom of the lake. Concentrations of major ions and trace elements reported by other investigators have been included to provide a more complete summary of the chemical composition of the lake. Lake Vanda is compositionally stratified, as well as being thermally and density stratified. Variation In the chemical composition with depth has been expressed here in terms of mole % of the constituent ions and concentration ratios of Ca/Sr and Na/K« Chemical aspects of the lake have been made more complete by inclusion of the pH profile and available analyses of trace elements. A mechanism for concentration of salts in Lake Vanda has been developed, based on freezing out of salts during the formation of ice on the lake. Lake Vanda has been divided into four layers on the basis of the chemical, density, and temperature profiles. This layered structure has been used to develop a tenative history for the brines of Lake Vanda, based on subtle changes in the past climate of Wright Valley. 84

2. Concentration, profiles of principal cations

■ m Concentration of the dissolved salts In Lake Vanda Increases with depth, directly reflecting the downward Increase in density (figure 13)* The concentrations of calcium, strontium, sodium, and potassium along the depth profile collected by D. D. Koob are shown in table 16. Calcium was determined by titration with EDTA at pH 12, using Hydroxy Naphthol Blue as indicator. Strontium was analyzed by isotope dilution using Sr enriched spike. Both sodium and potassium were deter­ mined by flame photometry. Analytical methods used in these analyses are given in Appendix A. More complete chemical analyses of a depth profile from both anglno and others (1965) and Yamagata and others (1967) are given in table 17* These analyses are included to provide a more complete summary of the chemical composition of the lake. Inclusion of this additional information with that obtained from the samples collected by D. D. Koob is Justified because (1) layering of the water is believed to be stable with respect to time and extends horizontally throughout the lake, and (2) comparison of the analyses in table 16 with those of table 17 shows that concentrations at a given depth are similar for the elements determined, particularly for 85

TABLE 16. Concentration In ppm of calcium, strontium, sodium, and potassium in Lake Vanda, Wright Valley.

Depth b e l o w C a + a Sr+t l f a + K+ surface

4 m 3 8 . 5 0.141 31.9 9.9 8 5 64 . 5 0.226 45.1 1 2 . 5 6 66.6 0.228 45*5 12. 6 7 67.7 • 45.8 12. 8 8 68.0 0.228 46.0 14. 1

9 68 . 6 47.0 15.7 10 76.2 0.256 49.5 16.1 — 11 78.3 5 M 17.8 12 89.2 0.298 62.8 17. 9 13 106.2 - 6 9.5 19.2 14 169.4 108 33.1 1 5 170.4 0.536 101 31.4 2 0 1 72.6 0.539 1 0 5 29 . 2 25 170.8 0.552 1 0 5 2 8 . 5 30 171.3 0.553 102 2 9 . 5 35 1 72.2 0.553 1 0 4 26.9 40 27 5.° 0.886 37. 8 44 4 4 8 . 0 1.283 183 4 0. 7 46 6 00.1 1.655 2 2 0 41.1 48 8 1 5 . 0 2.281 223 39.6

50 6,680 18.50 1 , 5 5 0 276 52 8,890 • 1 , 9 2 0 309 54 13i§50 39.16 3 , 1 0 0 4 2 1 56 17;690 50.90 3 , 9 0 0 4 5 9 58 20,140 58.97 5,ioo 60 22,180 67.05 5,420 684 TABLE 17• The chemical composition of Lake Vanda, Wright Valley, Data from Anglno and others (196?); values in parentheses () from Yamagata and others (1967). Concentrations are in ppm. n Depth _ + CO o belov Mg+* Ca+a Ha K+ C l" * hco 3“ surface

(3«6)m (1.7) (6.4) (2) (1.8) (21,4) (2.2) ♦.? 17 6 2 <2 8 0 10 (8.5) (32.7) (19) (7.2) (105) (10) (? 12 44 30 9 134 4 39 9 12 46 30 10 149 8 39

12 13 g4 1? 11 164 8 41 15 22 51 15 260 16 49 (15.8) (6^.1) (34) (13.4) (210) (17) 47 181 99 29 564 12 66 21 49 182 99 30 578 24 77

24 48 182 100 30 576 24 78 27 50 182 100 30 584 28 77 30 182 100 30 586 24 78 (31.5) (44.3) (184) (64) (28.8) (573) (24) 33 47 181 101 30 571 32 74 36 48 182 100 30 577 32 77 39 50 182 100 30 576 44 77 42 70 264 127 35 827 36 79 (42) (3 0 .8 ) (123) (33) (13.0) (199) (16) 4? 95 345 150 40 1,060 40 81 TABLE 17* Continued

Depth below M g * 1 c . " lfa+ K+ Cl* H C O 3" star face

48 ■ 173 614 228 53 1,910 80 84 51 91? 2,730 746 142 8,590 200 88 (52) (2,329) (7,918) (1,655) (406) (23,650) (187) 54 2,190 7,150 1,610 281 21,500 400 116 57 3,780 11,700 2,450 389 36,600 770 139

60 6,190 16,700 3,750^ 515 51,000 1,880 182 (10,120) (1,794) (419) (30,680) (288) W 24,254 6.761 766 75,870 770 126 88 samples from greater depth* Minor discrepancies between the three concentration profiles are probably a result of both sampling technique and analytical problems. Because analyses presented in this work were done in duplicate on the same material from which the lsotopic data were obtained, they are more pertinent to the objectives of this study and have been utilized wherever possible. It should be noted that these samples were collected at smaller depth Increments than any previously analyzed. a. Sodium concentration The concentration of sodium in Lake Vanda ranges from 32 ppm immediately under the ice to 5420 ppm at 60 m, the maximum depth reached by the series of samples analyzed in this study* The concentration profile of sodium is shown in figure 1 9 * The concentration remains relatively low, between 30 ppm and 220 ppm from the bottom of the ice to a depth of about 48 n. Between 48 and 5° the concentration of sodium abruptly increases and continues to increase steadily with depth from about 220 ppm to 5420 ppm at the bottom of the lake* The shape of the concentration profile parallels the density profile of the lake (compare figures 13 and 19)* When the concentration of sodium is plotted linearly versus depth (figure 19), small variations are masked due to the great range of concentration throughout the profile. Figure 19* Concentration of sodium in ppm along a depth profile, Lake Vanda, Wright Valley. 90

20fr

30 tu 40

50

60

2000 40000 SODIUM, ppm 91 On this curve, only one chemocllne is observed at a depth of about 50 When the logarithm of the concentration Is plotted against depth, more detail emerges. The log concentration-depth profile of sodium Is shown in figure 20* Now two chemocllnes In addition to the one at a depth of 50 ® are evident; one at a depth of 4 to 5 m, and the other at a depth of about 14 m. These chemo- clines have been marked in figures 19 and 20 by horizontal dashed lines. The shape of this concentration profile is very similar to that of the temperature profile of the lake (figure 14); I.e., the depths of the chemocllnes coincide with those of the thermocllnes. Although the temperature structure Is inverted In Lake Vanda, the density increases with depth and thus preserves the layering. b. Potassium concentration The concentration of potassium ranges from about 10 ppm just below the Ice cover of Lake Vanda, to 634 ppm at a depth of 60 m. The depth profile of the logarithmic concentration of potassium Is shown in figure 21. Three chemocllnes are evident and are indicated on the graph. These occur at depths of about 4.5* 14, and 50 m, and are identical to depths of the principal chemocllnes estab­ lished by the sodium analyses (see figure 20). Figure 20. Logarithmic plot of the sodium concentration in ppm along a depth profile of Lake Vanda. 40 20 50 60 DEPTH. 30 meters 100 O IM ppm SODIUM, 1000 10000 93 9*

Figure 21. Logarithmic plot of the potassium concentration in ppm along a depth profile of Lake Vanda. 95

K>0 POTASSIUM, ppm 96 c. Calcium concentration The concentration of calcium Is PP® at the base of the ice cover of Lake Vanda, and increases downward to 2 2 , 1 8 0 ppm at a depth of 6 0 m. The logarithmic concentration profile of calcium is shown in figure 2 2 . The shape of this profile is very similar to that of sodium and potassium, and the three principal chemo- clines for calcium occur at Identical depths as the chemocllnes for the two alkali metals. The only apparent differences in the curves occur between depths of 40 to 50 m. Over this depth interval, the change in concentration for sodium (from 147 to 223 PP®) and potassium (from 3 7. 8 to 39* 6 ppm) Is relatively small, while that of calcium Is large, ranging from 2 7 5 to 815 ppm.

d. Strontium concentration The concentration of strontium ranges from 0.141 ppm Just under the Ice to 67*05 ppm at a depth of 60 m. The concentration profile of strontium will be discussed in Chapter IV. It is noted here that the profile is similar to that of calcium and the three major chemo- clines occur at depths Identical to those of the elements Just considered. 97

Figure 22. Logarithmic plot of the calcium concentration in ppm along a depth profile of Lake Vanda. 10

20

X30(— h- o. LU Q

40

50

60 I 99 e. Magnesia® concentration The concentration of magnesium varies from 4 ppm under the ice to 7*684 ppm at a depth of 66 m (table 17)* The concentration profile for magnesium is very similar to those of other principal cations, and therefore has not been plotted.

3. Concentration profiles of the principal anions

The concentration of chloride ranges from less than 2 # ppm under the ice to about 7 6 , 0 0 0 ppm at a depth of 66 m (table 17). The concentration profile plotted from chloride analyses of Angino and others (1965) is shown in figure 23. Three chemocllnes are indicated on the graph* The uppermost chemocllne occurs between 4*5 end 6 m depth, the second at a depth of about 15 m, and the third at about 50 m depth. These positions in general agree with those determined in this study. Although this series of samples was collected at a different locality and during a different summer (1 9 6 1 -1 9 6 2 ) than for the series collected by D. D. Koob in January, 1966, the chemocllnes occur at Identical depths. Concentration profiles for both sulfate and bicarbonate ions are similar in shape to the chloride Figure 23. Logarithmic plot of the chloride concentration In ppm along a depth profile of Lake Vanda* Data are from Anglno and others (1965)* Depth t 0 4 0 6 0 3 0 5 20 10 2 l lg ppm log Cl, 3 4 101 102 ion concentration profile, tut these curves are not as well-defined (table 17). This Irregular shape of the profiles could be a result of biological activity within the lake and/or analytical problems, either in method or, in the case of bicarbonate ion, because of the time elapsed between sampling and analysis. Since total sulfate was measured, time effects will be negligible. However, time Is an Important factor in the analysis of carbonate species. If much time lapses between collec­ tion and analysis, differences will occur in the relative proportion of each species, mainly through contact with the atmosphere and biological activity within the sample. In any case, it is clear that the depths of the chemocllnes identified from the data of other investi­ gators are essentially Identical to depths of the three principal chemocllnes indicated from analyses of the major cations in the previous section.

4. Layered structure of Lake Vanda

Lake Vanda may be divided conveniently into four discrete layers on the basis of (1 ) chemocllnes, (2 ) thermoclines, and (3 ) density changes. These three boundaries occur at identical depths 1 4.5, 14, and 50 m. The resulting four layers are numbered 1 to 4 from the 103 top of the lake downward. Each layer is characterized by (1 ) concentrations and relative proportions of major elements, (2 ) temperature, and (3 ) density. Layer 1 Is the uppermost layer, extending from the base of the ice cover to a depth between 4 and 5 m. This Is the thinnest layer, being less than 1 m In thickness. In Layer 1, temperature and salt content are Increasing, and density varies only with the ambient temperature since the salinity is too low to cause a detectable change within this layer. Layer 2 extends from a depth of less than 5 m to 14 n, This layer Is characterized by constant density (0.998-0.999 g/al at 20°C) and by relatively constant concentrations of the principal Ions. From about 5 to 10 m, the concentrations and temperatures remain constant, but between 10 and 14 m, a slight increase in both parameters was noted. Because this change is small compared to the full range of concentrations and temper­ ature throughout the profile, Layer 2 does not warrant further subdivision* Layer 3 occupies a depth Interval from 14 to 50 m. In this region, profiles for temperature, major Ion concentration, and density are similar. In crossing the boundary line at 14 a from Layer 2 to Layer 3, there Is a relatively small but abrupt increase In these parameters. Below 15 m, numerical values of these three parameters 104 remain essentially constant down to a depth of about 38m* There is a gradual increase between 38 and 40 m of density, temperature, and ion concentrations until a depth of about 50 m, which has been taken as the base of Layer 3.

Layer 4 has bean assigned to depths between about 50 m and the bottom of Lake Vanda* The top of this layer has been identified mainly on the basis of the shape of the density and concentration profiles, where a very abrupt Increase in density and salt content occurs. This Increase is of greater magnitude than that observed at the boundary between Layers 2 and 3* The major thermocline is also located at the upper boundary of Layer 4 (figure 14). The nature of the boundaries between layers is not known at the present time* The very dense Layer 4 is considered to be a non-eonvectlve layer, while Layers 2 and 3 are thought to be convective (Hoare, 1966, 1968; Vilson, 1964; and Wilson and Wellman, 1962)* Layer 1 may be the result of recent discharge of the Onyx River. Boundaries between Layers 1 and 2 and 2 and 3 are probably the result of mixing of layers of differing density, temperature, and salinity, if these layers are convectlng. The transition from Layer 3 into the denser Layer 4 has possibly been modified by upward diffusion of ions (Wilson 105 and Wellman, 1962). Craig (1966) has stated that the chemocllne at 50 m cannot result primarily from diffusion. He maintained that the concentration gradient Is the same for the principal ions In Lake Vanda, although their diffusion coefficients are quite different* The chemo­ cllne represents mixing "by convection and turbulent diffusion" (Craig, 1966)* In order to attempt to resolve the controversy of the nature of the boundary at 50 m, concentratlon-concen- tratlon plots of Ca-Sr, Na-K, and Ca-Na were constructed* These plots will compare relative proportions of a pair of Ions with + 2 charge, a pair with +1 charge, and a pair of which one Ion has a charge of + 2 and the other +1* If physical mixing is the dominant process occurring across the chemocllne, a linear relationship will result* A similar curve will result, however, from a diffusion process, where the diffusion coefficients of the two ions involved are very similar* The concentration concentration plot of Ca-Sr is shown In figure 24. This curve Is linear, and thus the chemocllne at 50 a may be either a result of physical mixing of two brines or of diffusion of two Ions with essentially Identical diffusion coefficients* The Na-K concentration plot Is shown In figure 25* This curve Is not linear, thereby indicating that diffu­ sion Is significant across the boundary between in Lake Vanda. Lake in Sr, ppm 0 5 20 0 4 0 3 0 6 0 7 Figure Figure 0 24-. Concentrations of calcium and strontium strontium and calcium of Concentrations 5 t p * 03 I0 * ppm at C 10 15 20 106 n ae Vanda* Lake in ppm 0 0 4 0 0 6 200 Figure 25. Concentrations of sodium and potassium potassium and sodium of Concentrations 25. Figure 2000 o ppm No, 0 0 0 4 7 0 1 108

Layers 3 and 4. On this basis, the linear curve for the Ca-Sr plot Is perhaps the result of the upward diffusion of two ions with similar diffusion coefficients. The Ca-Na concentration plot is shown In figure 2 6 . This curve also Indicates that diffusion night be a major process in establishing the chemocllne. Calcium has been depleted relative to sodlun, suggesting that there has been a more rapid migration upward in the lake of sodium ions. Thus, it appears that the dominant process across the chemocllne at a depth of 50 ® Is diffusion and not physical mixing. Therefore, in attempting to establish the length of time diffusion has occurred, calculations of Wilson (1964) based on diffusion of chloride ion may be valid, in spite of the objections of Craig (1966).

5. The chemical composition

Calcium chloride (CaCl3) constitutes most of the soluble salt In Lake Vanda, with lesser amounts of MgCl2 + and NaCl, and minor amounts of salts represented by K , HCO3-, and S04\ The mole % composition of the principal cations at selected depths are listed in table 1 8 . It is readily seen that the relative proportions of cations change with depth. Calcium constitutes about 35% of the total moles of cations at a depth of 4 m, increasing to 109

6

4

Q- o 2

O 0 5 10 IS 20 Ca, ppm * I03

Figure 26. Concentrations of calcium and sodium in Lake Vanda. TABLE 18. The cation composition of Lake Vanda, Wright Vailay, expressed as mole par cant.

Depth below Ion Molality x 10"5 Mole % surface

Ca 0.960 3 5 . 2 Sr 0.00161 0 . 0 0 5 9 Mg 0.122 4 . 4 8 Na 1 . 3 8 7 50.9 K 0.256 9 .4 Ca 1 . 6 0 5 3 9 . 2 5 S r 0.00258 0 . 0 0 6 3 Mg 0.206 5.04 Ha 1.961 4 8 . 0 K 0 . 3 2 0 7.83

10 Ca 1 . 8 9 7 3 5 . 7 Sr 0 . 0 0 2 9 2 0 . 0 0 5 5 0 .8 4 4 1 5 . 9 K 2 . 1 5 5 4 0 . 5 K 0 .4 1 2 7 .74 2 0 Ca 4*30 3 7 . 1 Sr 0 . 0 0 6 1 5 0 .0053 Mg 1 . 9 7 4 1 7 . 0 5 Na 4 . 5 6 5 39.3 K 0 . 7 4 6 6 . 4 4

30 Ca 4 . 2 6 5 37.3 Sr 0.00630 0 . 0 0 5 5 1 . 9 7 4 1 7 . 2 5 Na 4 .4 4 38.8 K 0 . 7 5 5 6.60 40 Ca 6 . 8 5 3 9 . 8 Sr 0 . 0 1 2 9 0 . 0 0 7 5 Mg 3•00 1 7 . 2 111 TABLE 18* Continued

Dapth balow Ion Molality z 10" 3 Mola % surfaca

44 Ca 11.17 43.2 Sr 0.0147 0.0057 Me 5*68 21.95 Na 7.955 30.8 K 1.04 4.02

50 ca 166.3 51.2 Sr 0.211 0.0062 Mg 83*9 25-8 Na 67.45 20.7 K 6.95 2.14 56 Ca 4 4 0 . 5 49.5 Sr 0.580 0.0065 Mg 269.4 30.3 Na 169.0 19.0 K 11.72 1.33 60 Ca 551.0^ 50.5 Sr 0.765 0.0070 Mg 289.0 26.43 Na 235.5 23.. 5 K 17. 5 1.60 6 6 * Ca 584 47.5 Mg 3 3 0 26.9 Na 294 , 23.9 K 19.6 1.6

D a t a froa Anglno and othars, 1 9 6 5 112 51% at a depth of 50 a* Magnesium increases from about 4.5% under the Ice to about $0% at a depth of 56 a. Relative proportions of both sodium and potassium decrease with depth. Sodium decreases from 51% at a depth of 4 m to a minimum of 19% at 56 a depth, and potassium decreases from 9*4% at 4 m depth to a minimum of 1.3J* at 56 m. Composition of the anions are given In table 19 ; the mole % values are calculated from data reported by Anglno and others (1965)* The relative proportions of the major anions also change with depth. Chloride Ion is the principal anion at all depths; its proportion ranges from about 85% at a depth of 6 a below the surface of the lake to almost 100% at 66 a depth. Proportions of both blear* bonate and sulfate ions are decreasing with depth. Of these, bicarbonate is the more important, ranging from 14.4£ at 6 m depth to less than O.ljf at a depth of 66 m. The chemical composition of the major dissolved constituents in sea water are In table 20; the dominant chemical compound Is sodium chloride. The chemical composition of sea water differs significantly from that of the brine* In Lake Vanda, whose dominant salt is calcium chloride. If the brines In Lake Vanda have formed from sea water, then chemical and/or physical processes must have altered the composition of the marine salts to that present in the lake today.

Don Juan Pond in the South Pork of Wright Valley has TABLK 19* Tha anion composition of Lako Vanda, Wright Vallay, axprassad as mola par cent*

Dapth b a l o v Ion Molality x 10"3 Mola % •orfaca

6 a Cl 3.78 84*7 so4 0*042 0.94 BCOj 0.64 14*4 12 Cl 4*63 86.0 S04 0.0833 1.58 HCUJ 0*872 12.5 24 Cl 16.75 91.6 so* 0*25 1.37 HCOj 1*28 7.02 36 Cl 16.75 91.4 1*82 Heij Ml4*203 6.88 45 Cl 29.5 ^ 94.5 0*416 1*36 “ ft, 1.44 4.25 51 Cl 249.3 98.7 2*08 0*82 hc 8s 1.44 0.57 57 Cl 1060 99.0 8*01 0.75 Sft, 2*28 0*21 60 Cl 1480 98.5 19.6 1.31 hc S3 2*98 0.20 66 Cl 2260 99.6 so* 8*01 0.35 HCO) 2 .05 0.09

*Valass calculated f r o a data of Anglno and othars, 1965* TABLE 20* Concentration of the aajor dissolved constituents of sea water . with selected cheaical ratios calculated froa these values*

Concentration Ion ppa

Wa+ 10,556 M*+a 1,272 Ca+a 400 K+ 380 Sr+* 8 Cl" 18,980 S 0 4“ 2,649

h c o 3" 140

NaA 27.8 C a/M a 0.0379 CaA 1 . 0 5 CaAg 0*314 Ca/Sr 50.0 Sr/Ca 0.020

M f/C a 3.18

M f/M a 0 . 1 2 1 M«A 3-35 C1/S04 7 . 1 6 5

•froa Mason, 1966 115 a chemical composition similar to that of Lake Vanda because calcium chloride is the major salt in the brine, this suggests a similar source for the salts in both lakes. In Lake Vanda, the relative proportions of magnesium and sodium are greater than in Don Juan Pond. However, this could Indicate that since Don Juan Pond Is very shallow today and may have been completely dry in the past, only the most soluble of the principal chlorides will occur in the dense brines, i.e., calcium chloride. The salt content of Lake Vanda is less and a greater amount of other chloride salts can be dissolved in the water. Don Juan Pond has been considered in more detail in Chapter V. Inspection of tables 18 and 19 shows that Lake Vanda is compositionally stratified. If the salts represented in each layer of the lake have been derived from different sources, or if chemical and physical processes have latered the composition of the salts, a more detailed study of the composition might provide additional infor­ mation of the chemical history of the lake. Concen­ tration ratios of ions have been utilised in this portion of the study. Various physical and chemical processes will have differing effects on given ions. Pairs of ions that are chemically similar will be affected to a lesser degree than a pair of chemically different ions. For 116 example, differences in the chemical regime of the lake could be enhanced by investigating the interrelationship of the Ca/Na or Ca/K ratio rather than the Ca/Sr ratio. Similarly, the Na/K ratio would be less affected than the Ca/K ratio. For water with a salt content as high as Lake Vanda, the variation in the relative amounts of chemically dissimilar ions is readily seen in table 18, and so ratios of only chemically similar ions have been considered here. The concentration ratio profile for Na/K in Lake Vanda is shown in figure 27. This ratio throughout the lake is less than the Na/K ratio in sea water, indicating that the brines of Lake Vanda are enriched in potassium relative to sea water. ?he ratio does not remain constant throughout the lake, suggesting that either the source of the salts has not remained the same, or that the chemical composition has been modified during the history of the lake. The N a / K ratio increases with depth, especially below the 14 m depth, where the ratio Increases from about 3 to more than 8. There are Inflections and reversals in the concentration ratio profile, and these occur at the boundaries of the layers established on chemical, density, and temperature data. It is not clear why the shape of this curve exhibits the unusual shape in Layer 2. The concentration ratio profile for Ca/Sr in Lake 117

Figure 27. Variation of the Na/ K ratio along a depth profile of Lake Vanda. 1X8

0

10

20

E 30

40 Na/K In ••a watar * 27.8

5 6 7 8 Na/K Vanda Is shown in figure 2 8 . This ratio at all depths is greater than the C a / S r of sea water (50.0), thereby indicating the brines of the lake are enriched in calcium relative to sea water. The ratio varies with depth throughout the lake, suggesting that either the source of these salts has not remained the same, or that the chemical composition has bean modified. It is not clear why the C a / S r profile exhibits the reversal at a depth of about 20 m. The Ca/Sr ratio decreases below the 50 m cheao- cline to the bottom of the lake. The Ca/Sr ratio changes from a value of 360 to 330 at the bottom. Since this is caused by a decrease in the relative proportion of calcium with respect to strontium, precipitation of calcium salts, such as carbonate and sulfate, may be occurring at the bottom of the lake. Calcium sulfate has been identified In samples of bottom sediments of Lake Vanda (P. Calkin, 1969, personal communication). It is probable that calcium sulfate is being precipitated at depth in the lake at the present time. Calclte also has been identified in x-ray diffraction studies of the bottom sediments (Angino, al-, 1965). If precipitation of gypsum and calcite is occurring at the bottom of Lake Vanda, the apparent increase in strontium content could suggest exclusion of this ion, resulting in its Increase relative to calcium in the bottom waters. 120

Figure 2 8 . Variation of the Ca/Sr ratio along a depth profile of Lake Vanda* 121

20

u> €> «>

3 0

h i o 40!

5 0

6 0

280 320 360 Ca/Sr 122

6. The pH profile of Lake Vanda

The pH profiles of Lake Vanda as determined by several investigators C D. D. Koob, 1967, personal communication) Angino, at a l . . 1965) and Yamagata, at a l , . 1967) are listed In table 21. There is consid­ erable discrepancy in pH values at a given depth* The greatest difference is for a depth of 57 m» for which Angino and others (1965) report a pH of 6*7, while Yamagata and others (1 9 6 7 ) report a pH of 8*4-8*6. The pH of the brines ranges from very slightly acid (pH 6) to slightly alkaline (pH 9)* Why the variation Is so large is not known* It could be due to real differences within the water, these changes occurring horizontally or with time. Actual dates of the collec­ tion of all samples listed in table 21 are not known. 0* D. Koob collected the first series (pH1) on January 18, 1966, and the measurements were made in the field as soon as the sample was brought to the surface. The second column represents measurements made by Angino and others (1965) on a series collected January 1, 1962* Presumably the data were obtained in the field, but this has not been ascertained* The series of samples collected by Yamagata and others (1967) were acquired in December, 1965* 123

TABLE 21. Profile of pH of Lake Vanda, Wright Valley

Depth below pH1 pH* pH3 pH4 pH5 surface

3.5 ■ 7.3 8.0 3.6 7.1 4 9.10 4 . 5 7 . 5 5 9 . 0 9 8.4 8 . 4 8.4 6 9.08 7 . 4 7 9 . 0 7 8 9 . 0 6 9 9.0 4 7 . 4 1 0 9.02 1 1 8.9 6 12 8.9 0 7 . 5 1 3 „ 8 . 7 6 - 1 3 . 5 8.3 9 . 0 1 4 8 . 6 0

1 5 8 . 4 4 7 . 4 15.5 8.2 9.0 16 8.22 XZ 8.21 1 8 8 . 2 0 7 . 4

1 9 8 . 2 0 2 0 8 . 2 0 8.5 8.3 2 1 8.20 7 . 4 2 2 8.20 23 8 . 2 0 2 4 8.20 7 . 4 2 5 8.20 2 7 7 . 4 3 0 8.21 7. 3 3 1 . 5 8.6 8.6 8.4 124 TABLE 21* Continued

Depth belov pH1 pH* pH3 pH* pH5 star face

33 7.3 36 7.3 38 8.2 39 7.3 42 7.3 8.6 8.4 - 45 7.3 8.5 46 51 6.9 52.5 8.6 7.3 54 6.7 57 6.7 8.4 8.6 60 6.7 62 8.6 63 8.4 6.8 65.5 8.6 66 6.1 66. 5 6.2

1D. D. Koob, 1966, personal communication, These determinations do not extend belov 3 0 m. 2Angino and others, 1 9 6 ?* 3Yamagata and others, 1967( profile V3 . ♦Yamagata and others, 1967* profile V£. * Yamagata and others, 1967* profile V6. 1 2 5 The method employed In the Measurement of the pH may have a considerable effect on the reported value. Ideally, the sample should be measured la situ. If the measurements are not made at the time of collection, pH will change, mainly as a result of absorption of C02 from the atmosphere. This might be the reason why the results In column pHa are lower t h a n data in the other columns. Also, the pH will change in time even though the sample is stored In a tightly sealed bottle, usually as a result of biological activity, leakage around the cap, or, in the case of a plastic bottle, diffusion of CO2 through the walls. Prom inspection of table 21, in spite of the variation of pH measurements at a given depth, It appears that the pH decreases a few meters above the bottom of the lake and the water remains very slightly acid down to the basal sediments. This is reasonable since H2S has been noted in samples collected at depths of 58 ■ or lower (Angino, £& a l . . 1965; and Yamagata, £& 1967)* The relatively neutral waters found in Lake Vanda account for the negligible quantities of carbonate ion (C03“) detected by Angino and others (1965). 126 7. Cause of the high salinity

Comparing the chemical composition of the Onyx River belov Bull Pass to that of the 4 and 5 m depths of Lake Vanda (table 12), it is evident that the composition of the 4 m level is very similar to that of the Onyx River, while that of the 5 m level is not. As an example, calcium concentrations of two samples of the Onyx River between Bull Pass and Lake Vanda were 3 8 . 1 and 43*0 ppm. Calcium concentrations in the lake Just belov the ice cover is 3 8 . 5 ppm (Layer 1), while at the 5 ■ level, calcium Increases to 6 4 . ? ppm (the top of Layer 2). A similar relationship holds for magnesium. The same two Onyx River samples contain 3.05 and 3.07 ppm magnesium. Compared to 2 . 9 7 ppm at a depth of 4 m and 5*01 ppm at 5 m, the magnesium content of the lower Onyx River is similar to the magnesium content In Layer 1 of Lake Vanda. Similar increases in concentration from depths of 4 to 5 m in the lake and the similarity between the 4 m concen­ trations and those of the lower Onyx River also apply for both sodium and potassium ions. This information strongly suggests that the Onyx River has been the major source of water of Layer 1 of Lake Vanda. In order to substantiate this hypothesis, concentration ratios of the major oatlons were studied. If unmodified Onyx River water is present Just under the ice, these ratios should be similar to those of the river near the point where It discharges into the lake. No water samples were collected at the mouth of the Onyx River because the river disappears Into gravel deposits here. One sample was collected from the river below Bull Pass, about 3 km east of the lake, and a second one approximately midway between Bull Pass and Lake Vanda, about 1.5 km east of the lake. Ratios of Na/K, Ca/K, Hg/K, Ca/fta, and Ca/Mg for these two river samples are given In table 22* Ratios of these Ions In the Onyx River more closely resemble those of the meltwater stream of the Meserve Glacier than the upper water of Lake Vanda. This was unexpected, since, on the basis of chemical analyses, water of the Onyx River below Bull Pass Is very similar to that of the 4 a level of Lake Vanda. Two explanations are possible for the apparent dlscrepencles of the ratios In table 22 with the theory that the values of a given ratio should be similar for the stream from the Meserve Glacier, Onyx River, and upper water of Lake Vanda. First, It is obvious that there are very few samples available for study. Also, the samples were collected over a period of the entire summer, and, If seasonal variation In concentration of the meltwater does exist, 128

TABLE 22* Concentration ratios of the principal cations for the lover Onyx River, aeltvater stress of the Meserve Glacier, and upper 2 n of vater of Lake Vanda.

MaA CaA MgA CaA» CaA*

Meltwater, Meserve Glacier

DV-66017 10 6.7 1.4 0.64 3.8 DV-66-025 6.0 3.1 0.6? 0.51 4.7 2-G H 7.9 6.3 - 0.80 -

Vater, lover Onyx River

DV-66-010 7.1 8.8 0.71 1.25 12.4 DV-66-016 7.2 6.8 0. 5 4 0.91 12.8

Vater, Lake Vanda

4- meter depth 3.2 3.9 3.0 1.2 13. 0 5 meter depth 3.6 5.2 4.2 1.4 12. 9 1 2 9 the data In table 22 are not representative. The Onyx River was not sampled near its mouth, a requirement in a study of this nature. To improve the comparison that has been attempted here, a greater number of samples is required, and several series of samples should be collected throughout the season, each series collected during a short interval of time. Second, the possibility exists that variations in the concentration ratios listed in table 22 are real, and the particular time of collection has no effect on the final interpretation. In this situation, which is reasonably speculative, it is possible that chemical processes, such as selective dissolution of salts, might have altered the relative proportions of the principal Ions during the course of flow of the meltwater streams. Because concentration ratios of the principal ions in the meltwater stream of the Meserve Glacier resemble those ratios in the Onyx River to a greater degree than those in the upper waters of Lake Vanda, this modification in the chemical composition of the Onyx River or that of Layer 1 in the lake has occurred either In the distance between the lowermost sample of the Onyx River and Lake Vanda, a nd/or in Layer 1 itself. It Is known that when ice forms from sea water, most of the dissolved salt in the water is excluded from the 130 Ice and concentrated in any restricted water belov* Such a process may be occurring at the base of the Ice cover of Lake Vanda* If so, the salt content of the loe should be less than that of the underlying vater* This hypothesis can be teste^by comparing the salt content of the Ice on Lake Vanda to that of Layer 1 or the Onyx River* No samples of Ice of Lake Vanda were available for analysis* Two water samples were collected from the surface of the Ice cover* Concentrations of the principal Ions In these vater samples should approximate that of the Ice, unless these Ions are derived from airborne material or from accumulation of salt by ablation of ice. These effects may be very Important, and, therefore, the salinity of the ponds on the ice surface of the lake is probably an upper limit of the salinity of the ice of Lake Vanda. A partial chemical analysis of these two ponds on the ice surface of Lake Vanda are given in table 12 (samples DV-66-006 and DV-66-007). As expected, concen­ trations of the cations are much lower than those in the Onyx River or of the water below the ice In the lake. For example, the calcium content of these two ponds on the Ice was about 2*2 * 0*1 ppm, while that in the Onyx River belov Bull Pass and In Lake Vanda Is always greater 131 than 38 ppm* These ponds contained 2.18 and 2*80 ppm sodium, while the sodium content of the river and lake was not found In any sample to be less than 31 ppm. The process of salt exclusion during freezing of water suggests that the salt content of the brines of Lake Vanda may be established Just under the ice cover of the lake* The thickness of Ice cover on the lake has remained fairly constant at an average of about 4 m , at least since 1961, when the lake was first studied. The lake level in this time probably has not varied greatly, and a balance between inflow and ablation is suggested for the present time (Angino, gl*, 1965)* If the lake level Is constant, salts that are being added to the lake by the Onyx River must be accumulating, since there is no outlet* As the surface of the ice sublimes, a balance in the thickness of the ice cover is achieved by formation of new ice at the base. This ice presumably forms from water recently added by the Onyx River. As this water freezes, salt In the water is excluded from the ice, and the salinity of water below the ice Increases. If there is an annual balance between inflow and sublimation plus evaporation, there will be some seasonal changes in lake level and salinity in the water Just below the ice cover* It is suggested that by the end of winter, Layer 1 no longer exists, or if it does, it Is thinner and less well-developed at that time than indicated by this sample profile* When the Onyx River begins to flow, the amount of water added to the lake is greater than the loss during the short summer. During the time Layer 1 reaches its maximum thickness and the lake level presumably rises. In the series of samples discussed In this study, Layer 1 should be about at its maximum thickness, since the samples were collected in late summer (January 18). Beyond this date, there are few days when the Onyx River flows* From data presented here, it is estimated that the thickness of Layer 1 is about 50-100 cm* Once the Onyx River ceases to flow, the net change in lake level is loss by sublimation, and this continues throughout the course of the long winter. During this time water below the Ice becomes more saline and Layer 1 gradually disappears as Its salinity approaches that of Layer 2* At this stage the summer is approaching, the Onyx River begins to flow again and Layer 1 is regenerated. This process may explain the presence of Layer 2 in Lake Vanda, which extends from 4.5 ■ to about 14 a below the surface. Throughout this layer, concentrations of the major ions remain constant (tables 16 and 17)* If Layer 2 has resulted by the proposed mechanism of salt concentration, the principal cations have been concen- 133 trated by the following factors from Layer 1 to Layer 2* calcium, 1.8} strontium, 1.6; sodium, 1.4-} and potassium, 1.4. These factors are sufficiently consistent to allow some estimates on the rate of sublimation. In the calculation of rate of ablation on the basis of Increase In concentration of major Ions in Layer 2 from Layer l v concentration of these Ions by a factor of 1.6 during the course of one year was assumed. Assumptions that no appreciable quantity of salts are occluded in the Ice and a maximum thickness of 100 cm for Layer 1 were also made. On these bases, 3&£ of the water In Layer 1 would have to be lost to produce a salinity similar to that found In Layer 2. Therefore, if Layer 1 is Indeed about 100 cm thick by the end of the summer, then the annual ablation rate is about 38 cm. An ablation rate of 38 cm per year Is In reasonably good agreement with other estimates for the Ice-free valleys (Chapter II). On the premise that these assumptions are roughly valid, then the level of Lake Vanda should be increasing at the present time at a rate of about 60 om per year* The sample profile of Lake Vanda was collected during a relatively warm summer, and the quantity of meltwater was greater than usual (D. 0. Koob, 1969, personal communication). During colder summers, the Onyx River will discharge less water Into the lake. However, 134 sublimation vlll occur at about the same rate with the possible consequences that the lake level would decrease, Layer 1 may not be present, Layer 2 would Increase in concentration, and Layer 2 would decrease in thickness. More data are needed to test this hypothesis* Information is needed on the total discharge of the Onyx Blver and Its annual variation, annual rate of ablation of the Ice cover, and thicknesses of the Ice cover and Layer 1, and changes of these thicknesses during the course of several years. If an appreciable quantity of low density water Is being discharged at the bottom of Lake Vanda, the density stratification of the lake would be disturbed by the upwelllng of this water* Extensive studies of the lake involving numerous depth profiles have failed to show any disturbances of the stratification of the brines. Consequently, at present there Is no evidence that could be Interpreted as supporting a hypothesis that water is being discharged through the lake bottom* Similarly, the numerous temperature profiles measured for Lake Vanda have failed to disclose the presence of temperature anomalies which might be caused by discharge of hydro- thermal fluids* 135

8. Trace, elements in Lake Vanda

The few trace element analyses that have been reported for Lake Vanda are given here to complete a chemical summary of the lake. Some of the analyses have been used to support a particular source of salts In the lake.

Anglno and others (1965) have reported the following lithium analyses for the depths Indicated: 4.0 ppm, 42 m$ *1.0 ppm, 45 m? *1.0 ppm, 48 m; 4 ppm, 51 ■ ! 14 ppm, 54 mf 20 ppm, 57 mj and 32 ppm, 60 m. Armitage and Anglno (196?) have analysed a depth profile for nitrogen and phosphate. Nitrogen concen­ trations ranged from 0 ppm to 0.1 ppm, and the concen­ tration of phosphate ranged from 0.07 ppm to 3 ppm. The trend of the curve in the depth profile is erratic and does not duplicate the shape of concentration profiles of the major ions. Since these elements are essential for living processes, their concentrations are probably governed by biological activity In the lake. Nitrite (NO2") and phosphate analyses reported by Yamagata and others (1 9 6 7 ) have concentration ranges of 2.0-28 ppb and 0 .0 3 -1 . 3 0 ppb, respectively. The shape of these profiles are also erratic, further suggesting the dependence of the concentration on biological activity. 136 The silica content of a profile of Lake Vanda varies erratically with depth; this also suggests a biological effect on its concentration. Anglno and others (196?) have reported silica concentrations in the range of 8*110 ppm, and Yamagata and others (1967) report a range of 1.34-10.1 ppm. Boron analyses have been reported by Yamagata and others (1967) for the following depths: 0.032 ppm, 1? m; 0.042 ppm, 30 m; 1.7 ppm, 4? m; and 2.4 ppm, 63 m. For the 63 m sample, they reported concentrations of rubidium of 170 ppb and cesium of ?9 ppb. For this sample, the concentrations of sodium, potassium, and chloride were 3? , 6 2 0 ppm, 2,714 ppm, and 116,000 ppm, respectively. Uranium has been determined in three samples from Lake Vanda by Thurber and others (1968). The uranium concentrations at depths Indicated were: 0.?2 ppm, 10 m; 0.80 ppm, 50 m; and 13*0 ppm, 57 m. The U a34/U23® activity ratios for these three samples were 2.? +- 0.2, 3*3 * 0.5, and 4.5 * 0.2, respectively, Boswell and others (1 9 6 7 ) have reported the following trace metal analyses of a sample of "bottom water" from Lake Vanda: Zn, 5400 ppm; Pb, <30 ppm; Bi, 6.1 ppm; Fe, 490 ppm; Mi, 44 ppm; and Mo, 2.1 ppm. 137

G. Geochemical History of Lake Vanda

A tenative history for Lake Vanda can be developed if it Is assumed that each layer represents a period of time during the lifetime of the lake, and each boundary represents a relatively short term event* This interpre­ tation implies that increasingly older water is encountered in going from the top to the bottom of the lake. Evidence from Lake Bonney in Taylor Valley indicates an increase of 9*2 m in the level of that lake since 1903 (Chapter VI). If this change in level was due to a change In climate, then Wright Valley should also have been affected. If Lake Vanda responded to this suggested climatic change that appears to have begun just prior to 1903 in a similar manner to that of Lake Bonney, then Layer 2 might represent water added since that time* Since Layer 2 is about 9*5 m thick, and 63 years have elapsed between 1903 and the year D* D. Koob collected this series of samples, the level of Lake Vanda has increased by an average of about 15 cm per year* If the ablation rate has remained constant at about 40 cm per year since 1903, and the chemical composition of the Onyx River has not changed in that time, then the average annual gross increase of Lake Vanda has been about 55 cm per year* The rate of Increase in the level of Lake Bonney 138 appears to be decreasing at the present time, and if the suggested climatic cause of this Increase can be extrapo­ lated to Lake Vandaf the rate of level increase for this lake also must be decreasing. Thus, the estimate of 55 cm per year of gross Increase In lake level is only a mean value over the interval of 1903-1966. Similar assumptions can be made for the origin of Layer 3, but conclusions become even more tenuous. Layer 3 probably represents a period of time when less dense water flowed onto a layer of water with higher salinity. The densest brine may represent the remnant of a former, larger lake, or perhaps it is the result of water flowing onto a dried lake bed. It Is possible that Lake Vanda has dried up one or more times in the past, but there is no direct evidence for this. The date of this inflow of water over the densest brine layer has been estimated using the diffusion rate of chloride ion (wilson, 1964). Calculations provide a date of 1,200 years ago, at which time Wilson (1964) claims there was a change in climate of the ice-free valleys. Later, more refined calculations by Roberts (1965) indicated a more probable date of 984 years ago. This degree of exactness Is not justified, but It does confirm the magnitude of time obtained by Wilson, 1964). Assuming the date calculated by Roberts (1965) to 139 be correct and that water In Layer 3 began to flow over Layer 4 at that time, Layer 3 formed at an annual net rate of approximately 4 cm/year. Since Layer 3 and the boundary between Layers 3 and 4 have been well-preserved, perhaps Lake Vanda has always been ice-covered or It became ice-covered soon after the fresh water of Layer 3 began to flow into the lake basin. After Layer 3 attained its present thickness of 36 m, the average annual discharge increased, resulting in the formation of Layer 2 .

H. Source of Salts in Lake Vanda Previous Investigations

1. Introduction Possible sources of the salts in Lake Vanda are: (1 ) evaporlte beds (?) in local sedimentary rocks; (2 ) sea water, by direct evaporation or wind transport; (3 ) volcanic activity; and (4) chemical weathering of local soil and bedrock. Identification of a specific source for the salts has been attempted by several investigators, who have used such parameters as (1 ) con­ centration ratios, (2 ) deuterium content, and (3) U237 U 238 activity ratio.

2. Evaoorltes Leaching of evaporltes in Beacon Group sedimentary 140 rocks has been suggested as a possible source of the salts In Lake Vanda (Nichols, 1962). In the Ice-free valleys, Beacon Group rocks occur only at the highest elevations, and these rocks constitute a quantitatively unimportant portion of bedrock. No evaporlte beds were located in the course of a study of sedimentary rocks In the region of Wright Valley and Taylor Glacier by Webb (1963a). He did find some sedimentary carbonate concretions within the Aztec Slltstone. The nodules, however, are a minor component of the slltstone, and the slltstone Itself Is a relatively thin unit (up to 22 m thick) within the Beacon Group sequence. Some anhydrite cement has been Identified in a sand­ stone unit near the top of the Beacon Group sequence in the Nlmrod-Beardmore-Axel Heiberg Glaciers region (Grlndley, ai-» 1964) about 750 km south of the Ice- free valleys. Until evaporites have been found In the region of Wright Valley, this source should remain speculative.

3. Concentration ratios Concentration ratios of major ions and some trace elements have been used to advocate either a marine, volcanic (hot spring), or bedrock origin for the salts In Lake Vanda. These ratios are probably not reliable because they are modified by processes such as precip- 141 itatlon of salts and biological activity. Since calcium predominates over magnesium, the salts in Lake Vanda cannot have been derived by direct evapo­ ration of sea water. In sea water, the magnesium content is higher than that of calcium (table 20). When sea water evaporates, calcium sulfate precipitates early, enriching the remaining solution In magnesium. Thus, If the brines In the lake were remnants of sea water, magnesium is expected to be present in higher concen* tratlons than calcium. In addition, the relatively low Na/tC ratio and high Ca/Sr ratio (figures 27 and 28) of the lake exclude a direct link to a marine source for the salts. Anglno and Armltage (1963) have attributed the calcium and magnesium content of Lake Vanda to dissolution of "dolomitic" marble beds by meltwater. An x-ray diffraction analysis of one specimen of marble from the Asgard Formation, however, indicated that it contained less than 10% dolomite. In addition, marble is quanti­ tatively unimportant in the bedrock of Wright Valley. Therefore, it is unlikely that a large portion of the magnesium in Lake Vanda has been derived from this source. Anglno and others (1965) showed that the HC03/C1 and S04/C1 ratios approximate those of sea water only in certain strata of the lake. They attributed chloride 142 Ion to Influx of thermal vaters and leaching from "chloride rich soils In the area." Sulfate Ion was also assigned a thermal spring origin. Ll/fta ratios In Lake Vanda were found to be higher (0.004-0*009) than the Ll/Na ratio of sea water (0.9? x 10"*). These ratios were cited in support of a hot spring origin of the salts. Wilson and Wellman (1962), however, have found no geologic evidence for hot spring activity in the area surrounding the lake. Prom analyses of the trace metals Zn, Pb, Bi, Fe, Mn, and Mo in the "bottom water" of Lake Vanda, Boswell and others (1967) concluded that the lake must have received Its salt contents from glacial meltwater and not from sea water. They based this conclusion on ratios such as Zn/Na, Zn/K, Mn/Na, Zn/Mn, and metal/total solids, which are quite different from corresponding ratios for sea water. They felt that if sea water had been origin­ ally present, "extensive chemical alteration" of the salts was responsible for changes in the ratios of the trace

elements. B/Cl concentration ratios of Lake Vanda range from 1.4 x 10"+ to 7.2 x 10~* (Yamagata, 1967). These ratios are lower then th*: B/Cl ratio of the ocean (2.4 x 10~4’), but they may not be sufficiently different to exclude a marine origin on this basis. Because of the difficulties in the analysis of boron, 143 any Identification of a source for the salts on the basis of the B/Cl ratio is probably not reliable. Concentration ratios of alkali metals suggest a non-marine origin. It can be seen in figure 27 that at all depths of Lake Vanda, the Na/K ratio (maximum 8.5) Is lower than the Na/K ratio of sea water (27.8). Rubidium and cesium analyses at the 63 m depth of the lake (Yamagata, £i al*, 1967) also suggest a non-marine origin for the salts. The K/Rb ratio for the lake (4.3 x 10”3) is higher than the K/Bb ratio of sea water (3.0 x 10” 3), but the K/Cs ratio of the lake (1.2 x 10~4) is lower than the corresponding marine ratio (3.9 x 10"4) The Rb/Cs ratio for the 63 m sample of the lake is 2.9, which is considerably lower than the Rb/Cs ratio for sea water (1 3 .0 ).

4. Deuterium content The deuterium content of Lake Vanda has been determined by Ragotzkle and Friedman (1965). A depth profile is shown in figure 29. The lake has a low deuterium content compared to sea water and the lake is stratified with respect to that isotope. Deuterium concentration is expressed as 6D, measured relative to the "standard mean ocean water" (SMOW)i

(D/B)aampie - (D/H)s m o w 3DJC x 100 (D/^ s m o w 14-4

Figure 29* Relative deuterium concentration along a depth profile of Lake Vanda. Bata are from Ragotzkie and Friedman (1965)* The chemoclines are Indicated by the horizontal lines. Dtpth, 0 2 0 4 30 10 0uwlmwo % ow 0#utwrlumsw 8 “ 145 146 the range of the values of 6D In I«ke Vanda Is from •26.5 to -29.45*. Belov a depth of 40 m, the depletion of deuterium appears to be the greatest, with a minimum 0D of -29 . 45*. Comparing these analyses with the tfD value of -0 .1 )^ for Antarctic sea water, Ragotzkie and Friedman (1965) concluded that water in Lake Vanda could not have been derived from the sea* The deuterium content of the Wright Lower Glacier, presently the principal source of water of Lake Vanda, is expected to be higher than that of Wright Upper Glacier due to the lower elevation and proximity to the Ross Sea of the former* The of -25*95* for the Onyx River is probably representative of this ice mass* Although the 6D value is not known for the Wright Upper Glacier, snow 300 km west and southwest of Wright Valley was found to have a ffD of -34.3#* Since the $D value of the bottom waters of Lake Vanda is lower than the ffD value of the Onyx River, Ragotzkie and Friedman (1965) have suggested that the lake origin­ ally received water from both Wright Upper Glacier (low 5D value) and Wright Lower Glacier (high 6D value). The combination of water from these two glaciers was considered to have produced the low ffD value of bottom water of the lake* They suggested that the Wright Lover Glacier later became the only source of water, explaining the higher ffD values for the upper part of the lake. 147 Craig (1966), however, concluded that the saline water in Lake Vanda is of marine origin despite the deuterium analyses. He stated that sea water "will evaporate until the activity of the water equals the relative humidity, at which point evaporation stops, but isotopic exchange continues until the water is in Isotopic equilibrium with the atmospheric vapor. Thus, evaporating sea water will reach the lsotoplc composition of local precipitation, so that the lsotoplc ratios do not exclude a marine origin." Therefore, on the basis of deuterium measurements alone, the source of the water, and thus of the salts, In Lake Vanda remains inconclusive.

5* XJa3*/qa3e activity ratio The activity ratio of IT234/tJ238 has been used to identify the source of the salts in Lake Vanda (Thurber, et al., 1968). The U 23+/Q23® activity ratio was determined at the 10, 5 0 , and 57 m depths of the lake, and was found to be 2.5, 3*3, and 4.5, respectively. Since U 234 is a daughter of U23®, and its half-life is relatively short (*1/a “ 2.48 x 10* years) compared to the half-life of U 23® (T = 4.5 x 10* years), the 1/2 activity ratio of the two Isotopes will approach a value of unity at secular equilibrium, providing the system has been closed for a period of a number of half-lives of U234. k state of disequilibrium between U234 and U23®, 148 however, has been found In materials derived by weathering processes (Thurber, 1962). Anomalously high activity ratios of U 2 3 4 / U 2 38 are caused by preferential leaching of U234 during chemical weathering. The decay of the parent U 23* nuclide apparently disrupts the crystal 2) A lattice and thereby the U nuclide is easier to remove from the crystal during weathering. Therefore, water draining rocks containing uranium may exhibit U 234/tJ238 activity ratios significantly greater than unity. For example, an activity ratio of 2.30 ± 0.01 was measured for a sediment sample from Great Salt Lake (Thurber, 1962). The U 234/U238 activity ratio for sea water is 1.1? * 0.03 (Thurber, 1962) and this appears to be a constant characteristic of sea water. The anomalously high U234/U23® activity ratios (2.5-4.5) for Lake Vanda are therefore not unusual. Because of the significant difference between the ratios in the lake and the ratio for marine uranium (1.15), the uranium in Lake Vanda cannot be derived principally from a marine source. On this basis, Thurber and others (1968) have attributed the uranium in the lake to a local bedrock source.

6. Conclusions It is readily seen that the previously considered chemical and lsotoplc parameters used to identify the 149 source of the salts In Lake Vanda lead to conflicting conclusions. While one ratio may suggest a marine origin, another may suggest a volcanic or bedrock origin. More­ over, ratios of a given pair of elements may vary from one layer of the lake to another and may suggest different sources for adjacent layers of brine. Of the many parameters that have been considered, the U234/0238 activity ratio probably is the most reliable. Because uranium is chemically dissimilar from all the principal ions constituting the salts in the lake, however, extrapolation of conclusions reached using this ratio cannot be Justified to include the bulk of the salts. Therefore, another parameter is needed to link the salts in Lake Vanda unambiguously to a particular source. To solve this problem, a parameter is needed that is not affected by chemical and physical processes occurring in the lake and which has unique values for each of the several possible sources. The lsotoplc composition of strontium appears to satisfy these requirements. An investigation of the applicability of using strontium Isotopes as natural tracers was undertaken. Development of this method and the results which link the salts in the lake to a specific source are described in the following chapter. CHAPTER IV

THE ISOTOPIC COMPOSITION OF STRONTIUM IN WRIGHT VALLEY

A* Introduction

The salts In Lake Vanda have been attributed to various sources, including: Cl) evaporation of trapped sea water $ (2) wind-transported marine salts; (3) volcanic activity and associated hot spring activity; (4) leaching of evaporlte beds (?) in the local sedimentary rocks; and (?) chemical weathering of local soil and bedrock* Parameters previously used to identify the source(s) of these salts are continually being modified by chemical and physical processes* Thus, conflicting conclusions have resulted and the source of the salts occurring in the lakes and soils of the ice-free valleys has not yet been explained. A new method is needed that can unambiguously identify one or more specific sources of the salts. It is necessary that this new parameter be Independent of physical and chemical processes that occur within the lakes and soils. This parameter must also be distinctive

for each of the possible sources of the salts. The 1?0 151 lsotoplc composition of strontium meets these requirements. Therefore, this Investigation of the applicability of using strontium isotopes as a natural tracer vas under* taken.

B. Isotope Geochemistry of Strontium

Naturally-occurring strontium consists of four stable isotopes* Sr8*, Sr86, Sr87, and Sr88. At this high mass range, these Isotopes are not measureably fractionated In natural processes. Thus, variations In the lsotoplc composition of strontium result exclusively from the beta decay of naturally-occurring Rb87 to Sr87 according to the reaction:

37Rb87— » 3e Sr87 + p•’+*' + 0.27 Mev idlere p“ is a negatively charged beta particle emitted from the nucleus and v Is a neutrino. The equation for radioactive decay

N = N0*"Xt Cl)

states that the number of parent atoms N remaining from an Initial number of parent atoms N0 Is a function of the decay constant X and the length of time t that the system has been closed. Growth of a daughter D as the parent decays can be represented by 152 D * N0 - N (2) or

N 0 = D + N (3) Rearranging equation (1)

N 0 = (4) and substituting equation (3) into equation (4)

D = N(eXt - 1) (5)

Prom equation (5)* the number of radiogenic Sr87 atoms resulting from decay of a given number of Rb87 atoms within a system closed since time t is

*Sr87 * Rb87(eAt - 1) (6) where A is the decay constant of Rb87* The decay constant of Rb87 most often used has been determined by Aldrich and others (1956), who obtained a value of 1*39 x 10-11 yrs"1. Not all Sr87 in a sample is radiogenic; some Sr87 Is inherited. Thus, the total amount of Sr87 can be represented by

Sr87+ _ = Sr87 + *Sr87 (7) to tax where Sr87 represents the amount of Sr87 inherited in the system and *Sr87 is the amount of radiogenic Sr87 formed within the system. Substituting equation (6) into equa- 153 8 7 tion (6) Into equation (7)« the total Sr content in a sample Is represented by

Sr?7. , = Sr®7 + Hb®7 (ext - 1) (8)

This equation relates the total amount of Sr87 In a closed system to the amount that was Inherited and the amount formed since the system became closed t years ago. It is more convenient to measure Isotope ratios, and since the number of Sr86 atoms does not change in a closed system* equation (8) can be stated with respect to Sr86t

(9)

Thus, the total Sr87/Sr88 ratio in a sample is a function of the Initial Sr®7/Sr86 ratio, the Rb/Sr ratio, and the length of time the system has been closed. The isotopic compositions of strontium of the relatively young McMurdo volcanlcs, the Ross Sea, and granitic rocks of Precambrian and Paleozoic age of the basement complex of Wright Valley are significantly dis­ tinct and can be readily distinguished from one another. Therefore, the lsotoplc composition of strontium in Lake Vanda should Identify its principal source. Sr®7/Sr®* ratios were measured for a suite of samples of bedrock and salts of Wright Valley, McMurdo volcanlcs 154 from Ross Island and Victoria Land, and in a series of water and Ice samples from the Onyx River, the Ross Sea, and Meserve Glacier, These ratios were then compared to the average Sr87/Sr86 ratio of Lake Vanda to ascertain the principal source of the salts In the lake*

C, Lake Vanda

The lsotoplc composition of strontium In Lake Vanda was determined for 13 samples of brine from the depth profile collected by D* D, Koob In January, 1966. Two samples were analyzed In duplicate* The results are listed In table 23 and shown in figure 30* The column & Includes values for the standard deviation of the mean for a number of sets of mass scans made during the course of measurement of the isotopic ratio* The concentration profile of strontium for Lake Vanda is also listed in table 23 and shown in figure 30* The concentration of strontium, like that of other major ions (see Chapter III), varies with depth, ranging from 0.141 ppm under the ice cover of the lake to 67*05 ppm for a bottom sample 60 m below the surface. Chemocllnes for strontium occur at depths identical to those of the other principal ions, i.e., at depths of 4.5, 14, and 50 m. On the other hand, the lsotoplc composition of strontium was found to be constant within experimental 155

TABLE 23• The lsotoplc composition and concentration of strontium from a depth profile of Lake Vanda, Wright Valley.

Depth below Sr 87/Sr8‘* Sr* ppm surface

4 a 0.7149 0 . 0 0 0 2 0 . 1 4 1 5 0 . 2 2 6 6 0 . 2 2 8 8 0 . 2 2 8 9 0.7150 0 . 0 0 0 1 10 0 . 2 5 6 12 0 . 2 9 8 14 0.7149 0.0004 1 5 0 . 5 3 6 20 0.7151 0 . 0 0 0 2 0 . 5 3 9 25 0.7146 0 . 0 0 0 2 0 . 5 5 2 30 0.7147 0.0003 0 . 5 5 3 3 5 0.7149 0.0006 ° * | 5 3 40 0.7147 0. 0 0 0 4 0 . 8 8 6 44 0.7150 0.0002 1 . 2 8 3 46 1 . 6 5 5 48 0.7150 0. 0 0 0 2 2 . 2 8 1 0.7150 0.0003 50 1 8 . 5 0 52 0.7148 0.0003 54 3 9 . 1 6

56 0.7149 0 . 0 0 0 6 5 0.90 58 5 8 . 9 7 60 0.7154 0.0002 6 7 . 0 5 0.7150 0 .0002

Average * 0.7149 * 0 • 00 0 1

*Normalized to Sr*6/Sr •• ■ 0.1194 e 156

Figure 3 0 . The Sr 7/ Sp ratio and strontium concentration in ppm of water samples from a depth profile of Lake Vanda. DEPTH, Sr^Sr86 0.1 m a r ppm. Srf 157 1 5 8 error throughout the entire depth of Lake Vanda (table 23). The average Sr87/Sr88 ratio for the lake Is 0.7149 * 0.0003 at the 99% confidence level. Because precision of the measurement Is * 0.0005, all ratio determinations for Lake Vanda fall vlthin this range and therefore may be considered identical. The fact that the strontium Is isotopically homogeneous is consistent with the interpretation that the source of the strontium in each of the layers of Lake Vanda has remained the same during the lifetime of the lake.

D. Onyx River

The Onyx River constitutes the only major supply of water to Lake Vanda at the present time. The isotoplc composition of strontium was determined on three samples of water from the Onyx River and for one sample from the "Goodspeed Stream," a tributary of the Onyx River. All four samples were collected during the 1 9 6 6 - 1 9 6 7 field season; the sample locations are shown in figure 17.

The Sr87/Sr88 ratios of the Onyx River and "Good­ speed Stream" samples are listed In table 24* The average Sr87/Sr87 ratio for these four samples is 0.7146 * 0.0002. All four values are Identical to each other within experimental limits. The average Sr87/Sr86 ratio for these meltwater streams is very similar to the 159

TABLB 24. Isotopic coaposltlon of strontioa In aeltvater froa Wright Valley

_ 87 /a 84* Saaple No. Description Sr /Sr 0 -

D V - 6 6 - 0 0 9 Water, Onyx River 0.7148 0.0004 DV-66-010 Water, Onyx River 0.7144 0.0004

DV-66-016 Water, Onyx River 0.7147 0 . 0 0 0 5 D V - 6 6 - 0 1 5 Water, Goodspeed streaa 0.7145 0. 0 0 0 5

Average ■ 0.7146 * 0.0002

^Corrected for fractionation assnalng Sra*/Sr8a * 0.1194. 160 average Sr87/Sr88 ratio of the strontium dissolved In

lake Vanda (0.7149). On the basis of these dataf it is apparent that the Onyx River is presently supplying to the lake strontium of the same isotoplc composition as that found throughout the lake.

E. Ross Sea

If Lake Vanda and the Onyx River contain strontium predominantly derived from a marine source, then the isotopic composition of the strontium should be identical to that of sea water. The Sr87/Sr86 ratios of two samples of water from the Ross Sea (McMurdo Sound) and one from Lyttleton Harbor, New Zealand, were measured. The results are given in table 25* The average isotoplc composition of strontium of the samples from the Ross Sea (0.7^94) and Lyttleton Harbor (0.7095) agree with previous analyses of marine strontium from other locali­ ties (Faure, et & 1 ., 1967$ Faure, 1965; and Hurthy and Belsex, 1968). When the average Sr87/Sr88 ratio of Lake Vanda (0.7149) Is compared to the average Sr*7/Sr88 ratio of water from McMurdo Sound (0.7094), It Is clear that the lake contains strontium significantly different from that of the sea. These results are Illustrated In figure 31. Strontium In 161

TABLE 25. Isotoplc composition of strontium In ths Ross Sea, Meserve Glacier, end Lyttleton Harbor, New Zealand

Sample No. Locality Sr87/Sr84* cr*

DV-66-001 McMurdo Sound 0.7095 0.0001 DV-66-026 Cape Evans 0.7093 0.0003

Average = 0.7094 * 0. 0001

DV-67-002 Ice, upper surface, 0.7090 0.0005 Meserve Glacier

DV-67-001 Lyttleton Harbor 0.7095 0.0003

m A / A A Corrected for fractionation assuming Sr /Sr * 0 .1194. 162

— i— '— i— i— r » i ■" Water, Lake Vanda ...... •M Sediment, Lake Vanda ...... 1 Water, Onyx River ...... H Soil, Wright Valley ...... H Ice, Meeerve Glacier ...... * Water, Ross Sea ...... * Volcanic s, Ross Island ■ ■ 1 14 8 Victoria Land ___ 1___ i___ 1___ • i . i 0.704 0.708 0.7W Sr*TfSr**

Figure 31. Comparison of the isotopic composition of strontium in Lake Vanda and possible sources of its salts. 163 Lake Vanda contains more radiogenic Sr87 than does sea water. Because on-shore winds prevail In the eastern part of Wright Valley during certain periods of the day, it Is possible that marine salt derived from the Ross Sea is being transported inland by these winds. To ascertain whether or not marine strontium was being aerially transported inland into Wright Valley, the Sr87/Sr88 ratio was determined for a sample of Ice from the Meserve Glacier, about midway between Lake Vanda and Wright Lower Glacier (figure 17). Approximately 10 kg of ice from the upper surface of the glacior was obtained from G. Holds- worth. The Sr87/Sr8* ratio for this sample is 0.7090, and is therefore Identical within experimental limits to the average Sr87/Sr88 ratio of the Ross Sea (0.7094). Thus, strontium in the ice on the surface of the Meserve Glacier was probably transported Into Wright Valley from the Ross Sea by easterly winds. Strontium was precipitated and eventually was incorporated into the ice of the glacier. Strontium from the upper Ice of the Meserve Glacier is the only marine strontium in Wright Valley that can be identified by its characteristic isotope composition. Its concentration In the upper Ice is less than 0.22 ppb and is therefore so low that when the ice melts and the meltwater comes in contact with salts in the soil, the 164 marine strontium is mixed with such large quantities of strontium from the soil that this marine component losses Its identity. This observation can be extended to the whole of Wright Valley. Marine strontium apparently exists throughout the valley, transported Inland mainly by easterly winds, but can only be identified in snow and ice that has not come into contact with the water- soluble salts of the bedrock and soil. Although some strontium of marine origin is dissolved in the melt- water, it is completely masked by strontium derived by weathering of silicate minerals.

P. McMurdo Volcanics

Wright Valley is located within the McMurdo volcanic province. Volcanic activity is indicated in the valley by cinder cones near the Bartley and Meserve Glaciers and Loop Moraine (figure 32). The possibility that volcanic activity has been an important source of the strontium in Lake Vanda can be tested by comparing the Sr8?/Sr86 ratios of the volcanics to the average Sr®7/Sr06 ratio of the lake.

Sr8V S r e* ratios were measured for 31 volcanic rocks, mainly basalts, from the McMurdo volcanic province of Victoria Land. Samples were collected from 16?

Figure 32. Basaltic cinder cone (?) on the Loop Moraine, eastern Wright Valley, The Loop Moraine extends across the background. (Photograph courtesy of Ft. E. Behling) 166 Ross Island and from the Transantarctic Mountains, as far north as Cape Hallett southward to Brown Peninsula. Basalts from Wright and Taylor Valleys have been dated by the K/Ar method, with dates ranging from about 2 to 3*9 m.y. (Armstrong, ££ &!•* 1968} and Denton and Armstrong, 1968). A K/Ar date of 0.68 * 0.14 m.y. has been obtained on anorthoclase from a kenyte flow on Ross Island (Treves, 1967). The Sr87/Sr84 ratios for these volcanic rocks are listed in table 2 6 . The strontium concentrations in ppm for these volcanics are also listed. The average Sr87/Sr88 ratio is O.7 O4 O. This value is in good agreement with analyses of strontium isotopes in basaltic rocks from other localities (Faure and Hurley, 1963$ Hedge, 1966; and Powell, al., 1965). A histogram of the Sr07/Sr88 ratios for the McMurdo volcanics is shown in figure 33* The range Is from 0.7028 to 0.7054. Inspection of figure 33 suggests that these analyses represent a normal distribution. A chi-square test (Krumbeln and Grayblll, 1965) was applied to these measurements. It showed that the probability is greater

than 9 0 % that the data represent a single parent population of normal distribution. This statistical analysis confirms what can be deduced by inspection of the histogram. It is clearly seen that the Sr87/Sr86 ratio of the 167 TABLE 26. Isotoplc composition and concentration In p|£n of strontium of McMurdo volcanics, Victoria Land

Sample No. Locality Sr87/Sr8**

257 Ross Island 0. 7 0 4 5 0 . 0 0 0 2 747*5 258 Ross Island 0 . 7 0 4 0 0 .0004 1 0 1 1 259 Ross Island 0.7042 0.0004 1009 m Wright Valley 0.7043 0.0003 7 2 2 . 0 2 68 Cape Rallett 0.7037 0.0003 1150 269 Cape Hallett 0 .7042 0.0003 1301 270 Cape Hallett 0 .7054 0.0003 0 4 6.6 272 Ross Island 0 . 7 0 4 5 0 .0003 596.4 273 Ross Island 0 . 7 0 4 5 0 . 0 0 0 6 7 1 2.0 2 75 Dailey Islands 0 . 7047 0 . 0 0 0 4 8 3 2.6 277 Ross Island 0 . 7 0 2 8 0.0003 4 3 3.6 281 Ross Island 0 . 7 0 4 4 0 . 0 0 0 5 1276 282 Ross Island 0 . 7 0 3 2 0.0003 7 82.7 287 Ross Island 0 . 7 0 4 4 0 .0003 9 40.9 288 Inaccessible Island 0 . 7 0 4 7 0 . 0 0 0 5 9 08.8 301 Mount Overlord 0.7044 0.0004 577.9 302 Mount Overlord 0 . 7 0 4 7 0 . 0 0 0 2 1 0 0.7 304 Lover Campbell 01. 0 . 7 0 3 4 0.0003 7 3 8 . 0 3 0 5 Lover Campbell 01. 0 .7041 0 . 0 0 0 5 1135 327 Taylor Valley 0 .7048 0.0003 1 0 9 0

3 4 1 Dromedary Platform 0 .7032 0.0003 1215 342 Hovchln Glacier 0 . 7 0 3 3 0.0003 810.3 343 Taylor Valley 0 . 7 0 3 8 0.0002 775.0 344 T a y l o r Valley 0 . 7 0 3 5 0 . 0 0 0 4 9 72.4 346 Taylor V a l l e y 0.7039 0.0005 1077 2 1 001 Brovn Peninsula 0 .7036 0 . 0 0 0 3 912.4 2 1 058 Brovn Peninsula 0 .7044 0 . 0 0 0 3 1146 2 1 2 4 8 Black Island 0 . 7 0 3 7 0 .0003 1004 D V - 6 6 - 1 0 6 Wright V a l l e y 0 . 7 0 3 5 0.0003 6 7 3 . 4 DV-66-1?*0 Taylor Valley 0 . 7 0 3 5 0 . 0 0 0 2 1 0 0 5 DV-66-lt12A Wright V a l l e y 0 . 7 0 4 2 0 .0003 708.4

Average * 0 . 7 0 4 0

^Corrected for fractionation assuming Sr87/Sr88 * 0.1194. 168

10

0*- A » o m M s ♦ m A A O 2 2 § ! o e • • » * • o o 1O o O o o Sr87/Sr**

Figure 33* Histogram of Sr87/Sr88 ratio measurements of the McMurdo volcanics, Victoria Land. 169 McMurdo volcanics (0.7040 * 0*0003 at the 99flf confidence level) is significantly lower than the Sre7/Sr®8 ratio for Lake Vanda (0.7149)* This excludes the possibility that the bulk of the strontium in Lake Vanda could have had a local volcanic source.

G. Soil and Bedrock

Since the bulk of strontium in Lake Vanda has not been derived from either a volcanic or marine source, local soil and bedrock in Wright Valley was considered as a possible source. The Isotoplc composition of stron­ tium was determined for 16 samples of soils from the valley. These were collected mainly from or near the valley floor between Lake Vanda and Wright Lover Glacier; the sample localities are shown in figure 17. In some of the samples, two Sr8 ?/Sr8t ratio measurements were made: one analysis on the salt removed by a water or dilute acid leach, and the other on the total salt-free soil. The water leach is to be preferred because (1 ) water leaching does not readily attack the unweathered silicates, and (2 ) it simulates conditions likely to occur in the environment of the ice-free valleys when meltwater flows through the soil. The analytical procedure used is described in Appendix A. The Sr87/Sr®* ratios for the salts and total soils are given in table 27. Three facts are immediately TABLE 27 • Sr87/Sr86 ratios of salts and soils from the floor of Wright Valley,

Sample Bo. Location Sr87/Sr8 w

DV-66-201-V9 18 m from W edge, L. Vanda 0.7144 0,0004 water leach

DV-66-201-V16 100 a from N.edge, L. Vanda 0.7148 0.000? water leach

DV-66-202 E side, Meserve Glacier 0.7144 0.0003 water leach

DV-66-203 Clark Valley 0.7119 0.0002 water leach

DV-66-204 £• Wright Valley 0.714? 0.0002 water leach

DV-66-20? E. Wright Valley 0.714? 0.0004 water leach

DV-66-206 near Onyx River, between 0.7144 0.0003 HC1 leach, pH 3-4 L. Vanda and Bull Pass

DV-66-207 near Onyx River, 0.7148 0.0003 HC1 leach, pH 3-4 east of Bull Pass

DV-66-207 0.714? 0,0003 HC1 leach, 0.? N

DV-66-207 a 0.7148 0.0003 Total soil

DV-66-208 near Onyx River, below 0.7141 0.0002 HC1 leach, pH 3-4 Bartley Glacier TABLE 27, Continued

Sample No* Location Sr87/Sr86*

V-62 Sediment, L, Vanda, 62 m 0.7148 0.0004 total sample

DV-68-201 Sediment, L. Vanda, 30 ffl 0.7149 0.0003 total sample

VLG-0-12 base, Wright Lover Glacier 0.7149 0.0003 water leach

VLG-0-12 u 0.7144 0.0004 total sample

BP-0-12 Bull Pass 0.7148 0.0003 water leach

WVM-9-2-12 near Meserve Glacier 0,714? 0.0003 water leach

WVM-23-2-9 ti 0,7142 0.0003 water leach n W M - 2 3 - 9 - 1 9 0.7142 0.0004 water leach

WVM-24-7-36 tt 0.71?7 0,0003 water leach

^1 Corrected for fractionation assuming Sr8V S r 88 = 0,1194. 172 evident by examination of data In the table. First, all the Sr87/Sr86 ratios of samples along the floor of Wright Valley are concentrated within a very narrow range, 0.7141-0.7149, with two exceptions. These two samples (DV-66-203 and WVM-24-7-36) are not included in the average$ they were collected In the "Clark Valley" and high on the south wall of Wright Valley, respectively. These samples are probably not as representative of the bedrock of Wright Valley as samples collected from the valley floor, and they have been omitted for this reason. Second, the average Sr87/Sr86 ratio is 0.7146. This value is identical to the average Sr®7/Sr8 * ratio of the Onyx River (0.7146) and Lake Vanda (0.7149)* Third, the Sr87/Sr88 ratios of water-soluble salts are identical to those of the total salt-free soil, within the precision of the measurements. Two samples of the bottom sediment from Lake Vanda were dissolved and their Sr®7/Sr86 ratios were measured. One sample, V-62, was collected in the center of the main lobe of the lake at a depth of about 62 m. This is at the same locality where the series of water samples was collected for this study. The other sample, D V - 68-201, was collected from the smaller eastern lobe at 30 m depth by D. Qreegor In February, 1968. The Sr®7/Sr88 ratios were found to be 0.7148 and 0.7149 173 for samples V-62 and DV-68-201, respectively (table 27)* These ratios for the bottom sediments are identical to the average Sr87/Sr88 ratio of Lake Vanda (0.714-9) within experimental limits* These observations justify the conclusions that (1) strontium in the salts in Wright Valley has been derived from local soils and bedrock and (2) strontium in these salts has been dissolved by the Onyx River and transported to Lake Vanda. The Isotopic composition of strontium was measured on a suite of samples from most of the major units of 87 . 86 the bedrock in Wright Valley; the Sr /Sr ratios are given In table 28. The selection of these samples was biased to provide the greatest possible range in chemical composition (and thus the Sr8V s*88 ratio) within each suite. Therefore, these analyses do not represent what could be considered an average value for each rock type, but rather should be regarded as an indication 8 7 8 6 of the range of the Sr 7/Sr ratio within each unit. In spite of the biased sampling, it is noteworthy that of the 23 Sr8?/Sr88 ratios given in table 28, 16 samples are in the range 0.712-0.718. It is reasonable to conclude from these measurements that a value of 0,715 represents an average of the Sr8?/Sr88 ratios for the bedrock of Wright Valley. 174

TABLE 28. Sr87/Sr88 ratios of the major units of the bedrock, Wright Valley.

Unit Sample No. Sr87/Sr86*

Olympus Qranite- DV-66-108 0.7171 0.0007 Gneiss DV-66-126 0.7207 0.0002 DV-66-135 0.7146 0.0003 DV-66-138 0 . 7 1 5 5 0.0003 Dais Granite DV-66-115 0.7174 0.0004 D V-66-118 0 . 7 1 8 5 0.0003 D V - 6 6 - 1 2 3 A 0 . 7 2 8 5 0.0003 DV-66-123B 0.7280 0.0004 DV-66-134 0.7154 0.0003 DV-66-154 0 . 7 2 3 5 0.0005 Vida Granite DV-66-136 0.7147 0.0002 D V - 66-142 0.7161 0.0003 D V - 6 6 - 1 4 8 A 0.7146 0.0003 DV-66-148B 0 . 7 1 4 9 0.0003 DV-66-149 0 . 7 1 4 5 0.0003 DV- 6 6 - 1 5 7 A 0.7232 0.0004 DV - 6 6 - 1 5 7 B 0. 7 2 0 5 0.0002 Ferrar Dolerlte 310a 0.7121 0.0007 368 0.7134 0.0003 DV-66-105 0.7116 0.0002 DV-66-155 0. 7 1 2 5 0.0002 DV-66-160 0.7118 0.0003

Asgard Formation DV-66-177 0 .7088b marble

*Corrected for fractionation assuming Sr88/Sr88 ~ 0.1194 aSample from Taylor Valley bratio determined by R. J. E. Montigny 175 The observation that the Sr87/Sr84 ratio of the water-soluble salts is identical to the Sr87/Sr8* ratio of the total salt-free samples is unexpected, k difference in the Sr87/Sr86 ratios would be expected due to differences in susceptibility to weathering of various minerals. The more resistant minerals, such as mica and potassium feldspar, have high Rb/Sr ratios and contain strontium with a relatively high Sr87/Sr88 ratio. More readily weathered minerals, such as carbonate, pyroxene, and plagioclase, have low Rb/Sr ratios and therefore have correspondingly lower Sr87/Sr88 ratios.

H. The Source of Strontium and Other Elements in Lake Vanda

The isotopic composition of strontium of Lake Vanda is compared with those of the possible sources of salts in figure 31* It is readily seen that the water in the lake contains more radiogenic strontium than does sea water or the McMurdo volcanics. On the other hand, Sr87/Sr88 ratios of water-soluble salts from the floor of Wright Valley and the total soil material are iden­ tical to the average Sr87/Sr88 ratio of Lake Vanda. Because the Onyx River contains strontium of the same isotoplc composition as both Lake Vanda and the soils and salts along the valley floor, it is apparent that 176 the salts are forming from the soils, are dissolved by meltwater, and are transported by the Onyx River to Lake Vanda* Thus, the bulk of the strontium in Lake Vanda has been derived from the local soil and bedrock; any volcanic or marine component is negligible* The nature of the bedrock in Wright Valley does not lend itself to an easy explanation of either the cation or anion content of Lake Vanda. Most of the exposed bedrock consists of granite, granite-gnelss, and dolerite. These rocks are generally poor in both calcium and chloride* Some approximate calculations can be made to provide an estimate of the quantity of bedrock which must weather to provide the amount of calcium and chlorine In Lake Vanda* The volume of Lake Vanda was found to be l .?2 x 1012 liters by graphic integration, using data of Angino and others (1965). Graphic inte­ gration was also used to determine the total weight of calcium and chlorine In the lake; these quantities were 1.02 x 1011 g calcium and 3*12 x 1011 g chlorine* Using these values, it can be estimated that if Lake Vanda were to become chemically homogeneous, Its calcium and chloride concentrations would be about 700 ppm and 2000 ppm, respectively. In these calculations, the bedrock of Wright Valley 177 was assumed to be composed of a high-calclum granite. The average calcium and chlorine contents for this type of rock has been estimated to be 2?,300 ppm and 130 ppm, respectively (Tureklon and Wedepohl, 1961). The density of this granite uas assumed to be 2*7 g/cm3. It was also assumed that all the calcium and chlorine are released In soluble form during chemical weathering. On this basis, all the calcium In Lake Vanda could be derived from the chemical weathering of 4.0 x IO13 g of hlgh-calcium granite. This is equivalent to a cube of granite 250 m on edge. To account for all the chlorine in the lake, a total of 2.4 x IO15 g of high-calclum granite would have to weather completely, and this represents a cube of granite 960 m on an edge. The weight of the hlgh-calcium granite required to provide all the chlorine in Lake Vanda is almost 60 times greater than that needed to supply all the calcium in the lake. If It can be assumed to a first approximation that all the calcium has been derived from a bedrock source, then perhaps only about one- slxtleth of the chlorine present has been derived by weathering of the local bedrock. The remainder of the chloride may have originated from the ocean, both as spray and in the s n o w that falls in the basin and on the surrounding glaciers.

If most of the calcium in Lake Vanda has been 178 derived from the local bedrock, an estimate of the rate of chemical weathering can be made. The area of the drainage basin of the lake is estimated to be about 200 square kilometers. A layer of rock 7*5 cm In thick­ ness must have weathered to provide the calcium in the lake. Assuming that the last major glacial event occurred about one million years ago, a weathering rate of 7*5 x 10"5 mm/year is obtained. This is a reasonable value, and it appears possible to explain the calcium content of the lake in this manner. On the other hand, it is also reasonable to consider that most (>98^ by weight) of the chloride ion in Lake Vanda has been derived directly or indirectly from the sea. The Ross Sea is separated from Wright Valley only by the Wilson Piedmont Glacier (figure 3)* Easterly (onshore) winds prevail in this part of the valley for several hours each day at least during the summer. These winds carry marine salts that are precipitated with the snow on the Wilson Piedmont Glacier, the surrounding mountain ranges, and the valley floor. Chloride is the principal anion found in the ice, and chloride ion has been detected in snow at the South Pole, at least 1600 km from the coast. Analyses of the snow at the South Pole give a chloride ion content of about 0.027 ppm (Wilson and House). This chloride ion 179 was thought to be of marine origin. Presumably the concentration of chloride Ion increases as the coast Is approached from the South Pole. Angino and others (1964a) have reported a chloride content of about 14 ppm for an Ice sample from the Wright Valley side of the Wilson Piedmont Glacier. Since this glacier is the major source of the meltvater of the Onyx River, it is perhaps the major supply of chloride ion to the Onyx River and eventually to Lake Vanda. If a chloride ion concen­ tration of 14- ppm and a density of 0.9 g/em3 are assumed to be representative for the ice of the Wilson Piedmont Glacier, then 2.2 x 1013 kg of the ice could provide the chloride ion estimated to be dissolved in Lake Vanda at present. This amount of ice is equivalent to a cube of ice 2900 m on edge. On the basis of these calculations, it seems realistic to consider the ice, and ultimately the sea, as the principal source of the chloride ion in Lake Vanda. The melting of the large quantity of ice required to release the chloride ion presently dissolved in the lake is more reasonable than the assumption of complete weathering of 2.4 x 1018 kg of a hlgh-calcium granite. The relative amounts of calcium in Lake Vanda that could have been derived from both & marine source and local bedrock can be estimated by assuming that all the chloride Ion In the lake originated from a marine source. It was further assumed that there is no fractionation between calcium and chloride Ion during aerial transport from the sea, so that for the chloride ion in the lake, there is a corresponding amount of calcium present derived from the ocean. The Ca/Cl ratio in Lake Vanda along a depth profile remains relatively constant at about O.32. The Ca/Cl ratio for sea water is 0.021, indicating that there is more calcium in the lake than can be accounted for by a marine source. Since frac­ tionation of marine salts does occur, these calculations probably establish an upper limit to the amount of marine calcium expected to be present in the lake. Calcium has been enriched at least 15*2 times over that expected from a marine source. Thus, it appears that about 94jf of the calcium in Lake Vanda has been derived from sources other than the sea. Therefore, on the basis of the above calculations, it appears that the bulk of the chloride ion In Lake Vanda can reasonably be assigned a marine origin, while most of the calcium seems to have been derived from other sources, probably bedrock. Although exact proportions of the salts cannot be assigned to given source(s), the Ca/Cl ratio of the lake may provide the most reliable limits. Even if some calcium salts have 181 precipitated in the lake, an upper limit of about 6?6 calcium of marine origin may be realistic for the lake. If at least 5% of the calcium in Lake Vanda is of marine origin, then possibly about 5% of the strontium in the lake may be derived from the sea and 95?^ from the bedrock. Assuming that the average Sr87/Sr8t ratio for the strontium derived from bedrock is 0.7150 and the Sr®7/Sr86 ratio for sea water is 0.7093, the ratio of a mixture of strontium in proportions 9 respectively, will be 0.714-8. Thus, the presence of 5# marine strontium is compatible with the data of Lake Vanda. If a total of 10% marine strontium was present, then the Sr®7/Sr86 ratio of the mixture would become 0.7144. The difference between 0.7148 and 0.7144 could possibly be detected according to the precision of the measurements. Therefore, the value of 10% appears to be an upper limit for a marine component of the strontium in Lake Vanda. Once the source of the strontium has been Identified, some speculation can be made concerning the source of the other cations. Since the geochemical behavior of calcium is similar to that of strontium, the same conclusion may be extended to this ion; i.e., the bulk of the calcium in Lake Vanda has been derived from the local bedrock. The extrapolation of the source of strontium to that of 182 magnesium Is more tenuous. Magnesium behaves geochemi- cally less like strontium than does calcium. Since the bulk of the strontium and calcium are locally derived, perhaps a considerable proportion of the magnesium is also derived from the local bedrock. The source of the alkali metals is more difficult to identify. If most of the chloride ion has a marine source, it is reasonable to postulate that appropriate proportions of sodium and potassium have also been derived from that source. In summary, the bulk of the strontium in Lake Vanda is derived from the local bedrock. Most of the calcium and possibly the magnesium are probably derived from the same source. Much of the sodium, potassium, and chloride ion in the lake, on the other hand, may be derived from the sea, a result of aerial transport by easterly winds. CHAPTER V

PHYSICAL, CHEMICAL, AND ISOTOPIC CHARACTERISTICS OF DON JUAN POND, WRIGHT VALLEY

A* Introduction

Don Juan Pond Is located in the South Fork of Wright Valley (figures 4, 34— 36)* It was discovered in early October, 1961 by a group of scientists under the leadership of G. II. Meyer during the course of a field reconnaissance by helicopter (Meyer, e£ ai.*, 1962). Don Juan Pond was readily detected by the glint of light reflected by open water, although It was early spring and the ambient air temperature was -24°C. Don Juan Pond is situated in a valley cut largely into granitic rocks, while the higher ridges and peaks surrounding the valley are capped by Beacon sandstone and thick sills of Ferrar dolerite. The pond has formed behind moraines which block both ends of the South Fork of Wright Valley and permit no drainage outlet*

The size of Don Juan Pond probably varies season­ ally or annually. During the summer of 1961-1962, its

183 184

Figure 34. Western portion of Wright Valley, view west toward the inland Ice plateau at top of the photo­ graph* C Possible cinder cone S South Fork D Dais T Taylor Glacier DJ Don Juan Pond V Lake Vanda L Labyrinth WU Wright Upper Glacier N North Fork 18?

Figure 3?, South Fork, Wright Valley. Don Juan Pond Is located beyond the bend in the fork. (Photograph courtesy of D. D. Koob) 186

Figure 3 6 . Don Juan Pond, Wright Valley, view from edge of pond westward up the South Fork. Moraines forming the west boundary of the lake basin are in the center. (Photograph courtesy of D. D. Koob) 187 aeral extent was estimated by Meyer and others (1962) as approximately 200 m wide and 700 m long, with an average depth of 11 cm. Yamagata and others (1967) reported the pond dimensions In December, 1963, as only about 100 m wide, 300 m long, and a depth of 10 cm. This difference may be due to large accumulations of salts at the edges of the pond, which make it difficult to ascertain the actual dimensions. Because the pond is very shallow, a relatively small change in depth could result in a marked change In its area. The altitude of the pond has been estimated to be 136 m above sea level and the altitude of a possible drainage outlet to the east is 322 m above sea level (Tedrow, ££ &1*» 1963). The lake basin is in a position to receive potential runoff from a large area of the South Fork. At the present time, its drainage basin Is restricted to the small area defined by moraines to the east and west of the pond. Its only present surface sources of water are two small streams that drain the (ice-cored?) moraine to the west and the small amount of snow that falls directly into the small catchment area. At some time in the past the pond has been more extensive. This is suggested by strand lines, now poorly defined, which occur about 10 m above the present water level (Meyer, et al., 1962). 188

B. Density

The density of the brine in Don Juan Pond is unusually high. Previously determined values of the density are 1.2514 g/ml (Keyer, e£ , 1962) and 1.351 and I.38O g/ml (Yamagata, ££ fli., 1967). One sample from the edge of the pond was collected by D. D. Koob during the 1 9 6 4 - 1 9 6 5 field season. Its density at 20°C was 1.2774 g/ml. (See Appendix A for the analytical method used for the density determination.) From these data, it is clear that densities of brine in the pond are not constant. This variation in density of the brine is caused by differences in its chemical composition, discussed later in this chapter. Presumably there is no density stratification in the pond due to its depth of only 10 cm, its lack of ice cover, and the presence of continual, strong wind. It is possible that variations in density and chemical composition occur at edges of the pond where small pockets of brine are isolated or have only limited acess to the main body of water.

C. Temperature

The temperature of the water in Don Juan Pond should vary considerably during the year* Because of tho shallow 189 depth of the pond and lack of Ice cover, the temperature of the brine should be dependent upon the temperatures of surrounding air and underlying bedrock and soil. No simultaneous recordings of air, soil, and water temper­ ature have been reported. The only documented temper­ atures are the ambient (sic) temperature of -24°C by Keyer and others (1962) on 11 October 1961, and a water temperature of +10.6°C on 21 November 1961 (Tedrow, et al.i 1963). Because of the extremely high salt content, no ice has been seen as yet on Don Juan Pond. The freezing point of brine in the pond ranges from -48°C (Meyer, e£ , 1962) to -57°C (Tedrow, gt & 1 . , 1963)* The variation in the freezing point is probably due to differences in the salt content of the samples. Thus, the pond may not freeze, even during the Antarctic winter, except possibly during exceptionally cold periods. (A minimum temperature of - 6 2 ° C has been recorded at Lake Vida, Victoria Valley (Bull, 1966)).

D. Chemical Composition of Brine

Chemical analyses of brine from Don Juan Pond and meltwater discharge are given in tables 29 and 3 0 , Inspection of these tables shows that CaCl2 is the dominant salt, with lesser amounts of NaCl, MgCl2, and KC1. A semi-quantitative spectrographic analysis of the TABLE 29* The chemical composition of Don Juan Pond and its influent streams. Concentrations are expressed in ppm.

Don Juan Pond Pond Inflow Ion la 2* 3° 4d 5* 6 e 7e Ca 70,010 114,000 108,000 123,900 130, ?oo 26.3 527.9 Kg - 1,000 1,200 1,320 1,820 2,?90 18.? 44.2 Sr 82?. 1 ------Ra 17,250 11,?00 10,900 4,110 2,160 36 65 K 1?0 160 220 150 230 2.1 6.5 Fe - 23.7 ---- - Kn — c.o? - -•- Cl — 212,000 190,300 229,400 2?0,?00 61.1 1,090 SO. _ 11 160 0 0 ?2.9 80.? HCO, - 49 ----- NO, - 12.7 -- - -- S - 0 -- -- -

CaA-g 70 9? 83 68 50 1.42 11.9 Na/K 115 72 ?o 27 9.3 17 10 C1/S04 - 19,300 1,190 — 1-15 13-5 a. This work, collected by D. D, Koob. b. Meyer and others, 1962. c. Calculated from Tedrow andothers, 1963, assuming reported values are given in weight per cent. d. Torii and Ossaka, 196?. e. Yamagata and others, 1967. 191

TABLE 30. Local variation in the chemical composition of Don Juan Pond, expressed as equivalent per cent of the dissolved solids.

Time of Collection CaCl2 MgCla NaCl KC1 SrCl2

October 19611 90.42 1.57 7.95 0.065 November 196l2 90.23 1.80 7.88 0.093 December 19623 94.89 2.30 2.75 0.058 December 19633 95.42 3.12 1.38 0.086 December 1964 D1 96.99 1.96 0.95 0.092 D2 96.70 1.92 1.27 0.096 D3 94.77 2.15 2.99 0.099 D8 96.83 2.07 1.00 0.096 January 19654 80.37 1.89 17.23 0.087 0.432

1 Calculated from data reported by Meyer and others, 1962. z Calculated from Tedrow and others, 1963. 3 Yaraagata and others, 1967. 4 This work, collected by D. D. Koob. 192 trace elements in the brine was reported by Tedrow and others (1963). The brine in Don Juan Pond is beige-to-brown (sic) in color. This has been attributed in part to pyritic particles in suspension (Meyer, £t , 1962). Reported pH values cover a narrow range, 5*1 to 5.4 (Tedrow, et al.. 1963i and Meyer, ££.al>9 1962, respectively). The pH probably remains stable since there is little or no decay of organic matter, and the main controlling factor of pH is the CO2-HCO3" buffering action. The only organisms identified in Don Juan Pond are three types of bacteria (Bacillus megaterlum. Micro- coccus. sp., and Corynebacterium sp.) and a single yeast species, Sporobolomvces (Meyer, &1., 1962). This microbial population is restricted to Don Juan Pond and its environs and is not found in other lakes and ponds in Wright Valley. The principal source of water to the pond is small streams draining the ice-cored (?) moraine to the west (figure 36). Migration of water through the glacial debris may be an important source of water, but no information is known about this factor at the present time. Chemical analyses of two streams that drain the western part of the South Pork and discharge into the pond are given in table 29 (columns 6 and 7)* Ion concentrations in the meltwater stream in column 6 are 193 somewhat higher than expected compared to the salt content of the ice in the Meserve Glacier (table 14), The salt content of the other stream (column 7) is unusually high for a meltvater stream, even when compared to that of the Onyx River and the world average composi­ tion of river waters (table 12). Although no discharge information for these streams is reported, the latter stream possibly supplies most of the salts entering Don Juan Pond today. The dissolved salt content of this stream is mainly calcium chloride, with lesser amounts of sodium and magnesium chlorides and sulfates. This may explain the calcium chloride content of the pond, but there is a discrepency in the sulfate values. The weight ratio of C1/S04 for this stream is 13.5, while for two samples of brine from the pond, this ratio is 19,300 and 1190 (columns 2 and 3 , respectively). This indicates a marked decrease in relative sulfate content in the pond brines and may be explained by precipitation of calcium sulfate. Crystals of calcium sulfate are reported from the edges of the pond (Torli and Ossaka,

1965). There is a noticeable difference in the Ca/Mg ratios of the two streams (1.42 and 11.9) compared to those in Don Juan Pond (range 50 to 95)* These ratios (table 29) indicate a preferential loss of magnesium in the pond, 194 although no magnesium-containing salts have been detected. Information concerning the Na/K ratio is inconclusive (table 29). In summary, the unusual chemical composition of Don Juan Pond can be accounted for on the basis of the composition of the influent waters modified by reason­ able chemical processes. The ultimate challenge, however, lies in explaining the unusual chemical composition of the water discharged Into the pond.

E. Mineralogy of Salts Forming in Don Juan Pond

A new mineral, antarcticite (CaCl2*6H20), has been reported from Don Juan Pond (Torii and Ossaka, 1965). Crystals of this mineral occur in the water and at the bottom of the pond. The occurrence of this mineral is surprising for two reasons. First, It is a deliquescent compound, at least under ordinary conditions. The very low relative humidity of the Ice-free valleys probably accounts for the stabilization of this substance. Secondly, the fact that calcium chloride crystallises at all in a natural environment is unusual. Solubility of pure CaCl2*6H20 Is 279 g/100 ml H20 at 0°c, and

536 g/100 ml Ha0 at 20°C (Handbook of Chemistry and Physics, 41st Edition, 1959)* Thus, in order for the 19? salt to crystallize, the salinity of the pond brines must be unusually high. For the brine sample collected by Tedrow and others, (1963) (see tables 29 and 30)» a freezing point of -57°C was reported. Tedrow and others (1963) also state the freezing point of a saturated solution of CaCl2 is -?5°C, at which temperature the solution contains 29.83# CaCl2. They attributed the 2°C difference to the presence of othor salts, namely NaCl, MgCl2 , and KC1, In the pond. On a total weight basis, this sample contained 29.78# CaCl2. Thus, brines in Don Juan Pond should freeze only at the extreme lowest winter temperatures, although this has not been observed. The solubility of CaCl2*6H20 should only be exceeded in the pond either when the water temperature is very low, e.g., about -57°C, or at the edges, where pockets of water become Isolated from the main body of the pond and the relative rate of evaporation is higher. The fact that CaCl2*6H20 occurs "in the water and at the bottom of the pond" (Torii and Ossaka, 1965) is surprising and requires confirmation. If the limit of solubility of the hexahydrate compound is reached only at exceedingly low temperatures, how could It occur in the water, during the summer, when the pond should be undersaturated with respect to the hexahydrate? Until 196 more information is available, this problem remains unresolved. Crystals of both calcium sulfate and sodium chloride have been described from the salt encrustations at the edge of Don Juan Pond. Since both anhydrite and gypsum have low solubilities, (CaS04, 0.209 g/100 ml H20 at 3°°c? CaS04, 0.241 g/100 ml H20 at 0°C; Handbook of Chemistry and Physics, 41st Edition, 1959)* and calcium concen­ tration of the brine is very high, little sulfate ion could be expected to remain in solution* This is con­ firmed by the analyses in table 29«

F. Isotopic Composition of Strontium

Strontium from one brine sample of Don Juan Pond was isotopically analyzed; the Sr87/Sr®* ratio is 0.7183* which indicates that the strontium in the pond is more radiogenic than that in Lake Vanda (0.7149), which lies about 5-6 km to the northeast. Comparison of this Sr8?/Sr06 ratio to those of the Ross Sea (0.7094) and the KcKurdo volcanics (0*7040) readily indicates that the bulk of strontium in the pond cannot be derived from marine or volcanic sources. Bedrock underlying the South Fork of Wright Valley is mainly Vida granite, with some Beacon Group rocks and Ferrar dolerite at the higher elevations. No 197 McMurdo volcanics are In the western part of Wright Valley, although figures 2 and 34 show a dark, possibly conical, feature (designated 11C*') on the upper glacial terrace to the south of Don Juan Pond, The exact nature of this feature has not yet been established. It was not possible to obtain soil samples from the South Fork, so that the relative contribution of strontium from each bedrock unit is not known, Vida granite is expected to be the major parent material of

8 7 t 88 the soil. The Sr /Sr ratios of seven samples of the Vida granite from Wright Valley have been measured (table 28); the ratios range from 0.714? to 0 .7232, with an average of 0.7170, The Sr87/Sr8* ratios of five samples of Ferrar dolerite from Wright and Taylor Valleys range from 0.7116 to 0.7134, with an average of 0,712? (table 28). No Beacon rocks from southern Victoria Land have been analyzed. From the available Sr87/Sr86 ratios, it seems likely that the strontium in Don Juan Pond has been derived principally from the local bedrock and not from marine or volcanic sources,

G. Conclusions

The Sr87/Sr84 ratio of brine in Don Juan Pond suggests that strontium in the pond was derived from the 198 local bedrock, which consists predominantly of Vida granite. Since calcium and strontium are chemically similar, this conclusion may be extended to the calcium present in the pond. If both the calcium and strontium were derived from the bedrock, it is surprising that the concentrations of both sodium and potassium in the brine are so low, compared to the alkaline earths (table 29)* To appreciate the magnitude of the chloride content of J>on Juan Pond in relation to the chlorine contained in igneous rocks, an estimate has been made of the volume of granite necessary to weather to account for the chloride in the brine. In making this calculation, it was assumed that: (1) the bedrock consisted of a high- calcium granite containing 130 ppm chlorine (Turekian and Wedepohl, 1961); (2) the dimensions of the pond are 700 m long, 200 m wide, and 11 cm deep; (3) the density is 1.2514 g/cm3; (4) chloride ion of the brine is 212,000 ppm; (5) average density of the bedrock is 2.7 g/cm3; and (6) all chloride ion present in solution has been derived by complete weathering of the rock and none has precipitated as antarcticite. With these assumptions, the amount of chloride ion in solution in Don Juan Pond is equivalent to that contained in a cube of high-calcium granite 420 m on an edge. This is an underestimate because not all the chlorine present in 199 the rock Is necessarily released during weathering and because not all of the chlorine which Is released Is still present in solution In the pond* CHAPTER VI

PHYSICAL AND CHEMICAL PROPERTIES OF LAKE BQHNEY. TAYLOR VALLEY

A* Introduction Lake Bonney is situated in an enclosed basin in the westernmost part of Taylor Valley, about 30 km from the coast (see maps in back pocket and figures 8-10). The lake occupies the lowest part of the valley. The altitude of the surface of the lake is probably less than 100 m above sea level, but the exact altitude is In question at the present time. Reported altitudes include values of 38 ® (Angino, al.. 1964b), ?6 m (Yamagata, al*, 1967), and 98 m (U. S. G. S., Taylor Glacier topographic map, in back pocket). Lake Bonney extends almost 6 km in an east-west direction; the western end of the lake reaches almost to the eastern end of the Taylor Glacier (figures 8 and 9), an outlet glacier of the inland ice plateau. Although scientists had carried out investigations In Taylor Valley as early as the first Antarctic expedition of Scott in 1903, the first limnologlcal reconnaissance

200 201 was not done until the summer of 1961-1962 by Armltage and House (1962). Lake Bonney is composed of two lobes separated by the Bonney Rlegel (figures 8, 9, and 37)* From measurements taken during the 1961-1962 Antarctic summer (Angino, * 1964b), the western lobe is about 2.0 km in length and 612 m wide. The eastern lobe is 3.8 km long and 864 m wide. Maximum recorded depths of the lake are 21 m for the western lobe and 33 m for the eastern lobe (Yamagata, al* * 1967)* The two lobes of Lake Bonney are presently connected by a channel that was 42.3 ® wide in February, 1964 (Shirtcliffe, 1964) and 11 m deep. The size of the lake has fluctuated considerably in the past. It was much more extensive, as indicated by strand lines that surround the lake. Some of these beaches have been observed as much as 300-400 m above the present lake level (G. H. Donton, 1969) personal communication). These strand lines represent high water levels of a much larger proglacial lake dammed to the east by a glacier that moved inland from McKurdo Sound. Lake Bonney may be a remnant of this larger, older lake, whose strand lines can be traced as far as the eastern end of Taylor Valley. At least two large proglaclal lakes apparently have occupied the west end of Taylor Valley. Four major stages of glaciation are recognized in this valley (Pewe, 1966). 202

Figure 37. Lake Bonney and Taylor Glacier, Taylor Valley, a view toward the northwest. Bonney Riegel separates the two lobes of the lake. "Taylor Red Cone" is visible at the northwest edge of the lake. Rhone Glacier extends from the upper right. (Photograph courtesy of D . D . Koob) 203 Prom oldest to youngest, these are the McMurdo, Taylor, Fryxell, and Koettlltz Glaciations* When Ice of the Taylor Glaciation began to recede, the large Glacial Lake Washburn formed, dammed to the west by the Inland ice and to the east by the ice edge retreating toward McMurdo Sound. This lake was probably at least 300 m deep. Only the uppermost strand lines have been pre­ served. During the waning stages of the younger Fryxell Glaciation, Glacial Lake Llano formed behind the Suess or Canada Glacier. Well-preserved strand lines of this former lake are evident above Lake Bonney, especially on the north wall of the valley. Pewe (1966) suggests that Lake Bonney is the remnant of Lake Llano. Strand lines of Lakes Washburn and Llano can readily be distinguished from each other, as the degree of weathering on older strand lines is more intense than that of the deposits of Lake Llano (G. H. Denton, 1969* personal communication). The highest altitude attained by the youngest lake is 310 m above the present lake level (Denton and Armstrong, 1968). Lake Bonney has also been increasing in size since the turn of the century. In 1903 the width of the channel connecting the two lobes was only 5.2 m, while in 1964, it was 42.3 wide (Shirtcliffe, 1964). This Increase in channel width represents an increase of 9.2 m in the lake level between 1903 1964. The 204 exact cause of this increase Is not known. The lake is perennially ice-covered except for a narrow moat about the edge that forms during the warmest part of the summer. The average thickness of the ice cover is about 4 m, ranging from 3*5 to 4.3 m (Hagotzkie and Likens, 1964), This is similar to the thickness of ice on Lake Vanda. This observation prompted Angino and others (1964b) to suggest that this thickness of ice represents a state of equilibrium between thermal conductivity of the ice, rate of sublimation, and temperature of the ice-free valleys. Meltwater is supplied to Lake Bonney during summer mainly from the Taylor Glacier and from alpine glaciers surrounding the lake basin to the north and south (figure 8). No estimates of the quantity of discharge into the lake are presently available. A series of water samples at one-meter intervals from a depth profile of Lake Bonney were obtained by D. D. Koob during the 1965-1966 field season. These samples were collected close to the center of the eastern lobe of the lake; the maximum depth reached was 30 m. The chemical, physical, and isotopic measurements reported here were obtained from this series of samples. Some additional data of other investigators have been included to provide a more complete discussion of the lake. These data are acknowledged in the text and tables. 20? B, Density

Density determinations for Lake Bonney are given in table 31 and are plotted in figure 38* Density increases with depth in the lake in a regular manner. On the basis of the density determinations, five discrete layers can be recognized. These layers are shown in figure 38 and have been numbered 1 through 5 from the top down. Layer 1 extends from the ice-water interface at an average depth of 4 m below the surface of the lake and extends to a depth of 8 m. The density of this layer is 0.999 g/ml and is similar to that of distilled water at the same temperature. Layer 2 extends from above 9 m to a depth of about 10.5 m. The density of this layer is 1.008 g/ml. Layer 2 is only 2 m thick and the density difference between it and Layer 1 is small but signifi­ cant. For these reasons, and because previous investi­ gators had collected samples at larger depth intervals, the presence of this layer has not been previously reported. Layer 3 extends from a depth of 10.5 m to 17.5 m. This layer is characterized by a large charge in density. The density varies from a little more than 1.008 g/ml to about 1.155 g/ml* Layer 4 extends from a depth of 17.5 m to about 28 m. The density of brine in this layer increases slowly downward, ranging from about 206

TABLE 31. Density profile of Lake Bonney, Taylor Valley

Depth below Density, g/ml surface at 20°C

4 m 0.9992 5 0.9987 6 0.9994 7 0.9994 8 1.0023 9 1.0080 10 1.0075 11 1.0283 12 1.0559 13 1.0775 14 1.1103 15 1.1259 16 1.1410 17 1.1509 18 1.1613 19 1.1643 20 1.1699 21 1.1703 22 1.1744 23 1.1744 24 1.1750 25 1.1787 26 1.1764 27 1.1791 28 1.1820

29 1.1820 30 1.1932 DEPTH BELOW SURFACE (meters) 25 20 30 fdniylyr ae niae y h oiotl lines. horizontal the by indicated are layers density of Valley. All densities are normalized to 20 C. Boundaries Boundaries C. 20 to normalized are densities All Valley. Figure Figure 1.00 38 . Density profile of Lake Bonney, Taylor Taylor Bonney, Lake of profile Density . EST (/l a 20°C 0 2 at (g/ml) DENSITY LOS 1.10 LI5 1.20 7 0 2 208

1.157 g/ml at the top of the layer to 1*182 g/ml at the base* Layer 7 extends from about 28 m to the bottom of Lake Bonney. In this layer, the density increases from 1.182 g/ml to 1.193 g/nl' Where the lake has been sampled to a depth of 32 m, a density of 1.20 g/ml has been reported (Tamagata, et al.. 1967)* The density stratification of Lake Bonney is apparently permanent. Boundaries of the major density layers determined by several Investigators over the past few years have been Identical within the limits of measurement* The Increase of density with depth is directly related to the amount of salts dissolved in the water. On the basis of conductivity measurements by Anglno and others (1964b), density layering throughout the lake is horizontal, except at the edges where seasonal temperature variations are encountered.

C* Temperature

The temperature profile for the eastern lobe of Lake Bonney Is shown in figure 39- This curve was plotted from data obtained by D* D. Koob (table 32). Temper­ atures were measured using a Precision Scientific oxygen analyzer equipped with a thermometer. The location in the lake of the temperature profile is identical to that of the water samples used in this study, 209

« €> «

2 5

3 0

-2 0 2 4 6 8 TEMPERATURE, °C

Figure 39- Temperature profile of the eastern lobe of LaKe Bonney, Taylor Valley. Data obtained by D. D. Koob, 1965-1966. Position of the density boundaries have been included. 210

TABLE 32. Temperature profile of Lake Bonney, Taylor Valley

Depth below surface Temperature °C

4 ffl 0.11 5 1.17 6 2.39 7 3.67 8 4.78

9 5.56 10 6.11 11 6.67 12 7.11 13 7.33 14 7.39 7.33 16 7.11 17 6.78 18 6.22

19 5.72 20 5 . U 21 4.50 22 3.83 23 3.22 24 2.50 25 1.77 26 1.0? 2Z 0.44 28 -0.28

29 -0.89 30 -1.39

Measurements made by D. D. Koob, 1966. 211 The temperature of water In contact with the base of the ice cover Is +0.11°C. It is probable that the water and the basal ice are in close temperature equi­ librium. The temperature Increases with depth to a maximum of +7*39°C at a depth of about 14 m below the surface. Below 14 m the temperature decreases contin­ ually to the bottom of the lake, where, at a depth of 30 m, the temperature was -1.39°C* It Is interesting to note that the maximum temperature Is reached in the middle of Layer 3, the layer in which the density is rapidly changing. The temperature profile of Lake Bonney Is markedly different from that of Lake Vanda (compare figure 39 with figure 14). In Lake Bonney, the maximum temperature occurs at approximately half the maximum depth of the lake, while in Lake Vanda the maximum temperature is reached at the bottom. Because of the great difference in the shapes of the temperature profiles, the two lakes may be heated by different processes. Water temperatures in the western lobe of Lake Bonney are lower than those of the eastern lobe. This difference may be due to the influence of the Taylor Glacier and meltwater streams from the nearby alpine glaciers. For the western lobe, maximum temperature Is attained at a depth of 9 m. Minimum temperature occurs 212 at the bottom of the lake, and it is lower than the minimum temperature of the eastern lobe. Representative maximum and minimum temperatures that have been recorded for the western lobe are; +1.1°C and -3.0°C (Torii, at al., 1967) and +2°C and -4.5°C (Angino, at » 1964b). Extensive temperature measurements by Angino and others (1964b) suggest that heat flows from the eastern lobe across the shallow, narrow channel to warm the upper waters of the western lobe. This transfer of heat results in the maximum temperature being attained at a depth of 9 m and a general decrease of temperature with increasing distance from the connecting channel. Iso­ therms in Lake Bonney extend nearly horizontally through­ out most of each lobe (Hoare, et £l., 1964). At the edges of the lake the isotherms tend to curve upward, probably as a result of heat flow from the bedrock. Angino and others (1964b) described a reversal in the temperature at the base of a profile taken from the eastern past of the eastern lobe. They suggested that it indicates the presence of a thermal spring in this region. However, its presence has not been substantiated by other investigators working in the same locality. The thermal structure of Lake Bonney appears stable, as the temperature profiles obtained by various investi­ gators are essentially identical. From examination of 213 the data of Angino and others (1964b), a seasonal variation of temperature apparently exists, involving the entire lake. It does not appear that the increase is much more than 1°C at all depths during the course of the summer. This increase is probably real, as their data suggest that the temperature could be measured to within *0.0?°C.

D. Source of the Heat

Various sources of the heat in Lake Bonney have been considered. These include: (1) solar radiation; (2) high geothermal gradient; (3) inflow of thermal waters; (4) biological activity; and (?) chemical or radioactive heating. Angino and others (1964b) suggested that geothermal heating is the principal source of the heat in Lake Bonney. Since Taylor Valley lies within the McMurdo volcanic province and there is evidence of recent basaltic activity in the valley (figure 8), postulation of a high geothermal gradient is reasonable. However, by inspection of the temperature profile of the lake (figure 39), it is readily seen that hest is being lost to the bottom of the lake. Therefore, a high geothermal gradient is not responsible for the heat. 214 Influx of thermal waters into Lake Bonney at depth has also been suggested as a possible source of heat by Angino and others (1964b). This suggestion is based on a temperature profile obtained In the east end of the eastern lobe. At this location, the temperature increased slightly In the bottom 3-4 m. Otherwise, the profile was Identical to all others measured for the eastern lobe. They suggested the possibility that the thermal waters are less dense than the bottom brine. These warmer waters then Mmay rise as large 'bubbles' or envelopes from a source at depth and then spread out to form long, discontinuous layers of warm water•" However, their thermal rise at the bottom of the lake has not been confirmed by other Investigators and this explanation remains in doubt. Biological activity and chemical and radioactive heating are minor sources of heat and can therefore be neglected (Hoare, fii al., 1964). Absorption of solar radiation within Lake Bonney probably is the principal source of heat. Continual daylight occurs during the summer months. The ice cover of the lake Is relatively free of air bubbles and a large fraction of the incident radiation penetrates the Ice and enters the water (Goldman, aX** 1967)* Energy is then absorbed by the water and entrapped within 215 the body of the lake by high density layers below a depth of 11 m. Assuming that the temperature profile in Lake Bonney was a result of trapping of solar radiation by a dense, saline layer, several investigators have attempted to explain the temperature distribution on a theoretical basis (Hoare, si a 1., 1964; Shlrtcliffe, 1964; and Shirtcliffe and Benseman, 1964). The mathe­ matical treatment of Shirtcliffe and Benseman (1964) assumed that the lake is in a steady state, no con­ vection occurs, and that the structure of the lake is that of a layer of fresh water overlying a layer of highly saline water. Taking into account factors including energy, time, thermal diffusivities, and thermal conductivity, their calculated temperature profile Is almost identical to the profile obtained from field measurements. Later, more refined calcu­ lations by Shirtcliffe (1964) produced a curve that more closely fitted the temperature profile of the lake. Thus, absorption of solar radiation appears to be the major source of heat in Lake Bonney 216

E. Chemical Composition of Lake Bonney

1, Introduction

Both concentration and relative proportions of the principal ions dissolved in the water of Lake Bonney change with depth. The major cations, magnesium, calcium, strontium, sodium, and potassium, have been analyzed and concentration profiles of these ions have been plotted In this study. Analyses of major anions by other investi­ gators have been included to summarize the total ion composition of Lake Bonney. A layered structure may be established for the lake on the basis of density and concentration profiles of the principal Ions. The nature of the chemoclines has been considered. A discussion of the pH profile and reported trace element analyses complete the chemical study of Lake Bonney. A tenative history of Lake Bonney is proposed based on the chemistry and some measurements of physical parameters of the lake.

2. Concentration of the major cations

Concentrations of the principal cations in Lake Bonney increase with depth in a manner directly reflecting the density profile of the lake. The concentrations in ppm of magnesium, calcium, strontium, sodium, and 2 1 7 potassium were determined on the series of samples collected by D. D. Koob during the 1965-1966 field season} the analyses are given In table 33* The analytical procedures that vere used are described in Appendix A. Other Investigators have analysed lake vater from localities close to that vhere Koob collected the above series. Corresponding analyses by Angino and others (1964b) are given in table 34} values reported by Yamagata and others (1 9 6 7 ) are Indicated by parentheses in the same table. The analytical results of this study are in satis­ factory agreement with those of Angino and others (1964b) and Yamagata and others (1967)* Discrepancies that do exist may be due to sampling technique or differences in analytical methods. The greater differences between the analyses are observed in the upper part of the lake. In this region the Ion concentrations are more variable than In the lower part of the lake. Small differences in the depth from which samples vere collected may cause corre­ sponding dlscrepencies in the chemical analyses. Analyses of the major anions by Angino and others (1964b) and Yamagata and others (1967) are given in table 34 to provide a more complete summary of the chemical compo­ sition of Lake Bonney. Hew analyses of the major cations reported here will be used in preference to those 218

TABLE 33. Concentrations In ppm of the principal cations In Lake Bonney, Taylor Valley

Depth below V 2 c +a Sr+a Na+ 1C+ surface

4 m 80.5 77.3 0.7345 452 29.5 5 42.1 86.6 0.5827 268 22.6 6 92.2 56.9 0.5949 52.2 7 135.5 79.7 0.8124 689 44.9 8 367.2 125 1.750 1,550 90.3 9 902 250 3.026 3,410 210 10 1,08? 220 3.502 3,130 245 11 4,288 458 7.066 7,460 504 12 8,629 12.28 13,700 900 13 12,660 858 15.95 17,800 1,180 14 18,200 1,142 21.84 23,200 1,410 15 20,680 1,346 24.98 25,800 1,620 16 22,870 1,373 33.69 30,100 1,990 17 24,030 1,470 32,100 2,150 18 25,180 1,471 37.19 34,600 2,510 19 25,550 1,478 _ 36,200 2,430 20 26,000 1,489 38.89 37,700 2,570 21 26,070 1,477 — 38,100 2,510 22 26,550 1,461 38.70 39,600 2,650 23 26,400 1,524 39.37 39,000 2,570 24 26,520 1,514 38.93 38,900 2,670 - 25 26,860 1,527 39.92 40,500 2,720 26 26,690 1,545 39.02 40,000 2,620 27 26,750 1 529 39.41 40,000 2,620 28 26,850 1,512 38.21 41,900 2,840

29 26,150 1,459 37.02 43,400 2,710 30 25,790 1,434 35.88 47,300 2,690 TABLE 34. Concentrations in ppm of the principal ions in Lake Bonney, Taylor Valley *

Depth belov Mg+2 Ca+2 Ua+ K+ Cl" S04a h c o 3" surface

5 a 52 68 284 22 592 200 36 (64.4) (74.4) (335) (25.6) (734) (160) 8 163 80 740 54 1,660 120 60 (10.5) (1 ,762) (343) (3,920) (417) (12,420) (473) 11 909 250 2,880 231 .7,540 380 168 14 12,800 1,190 16,600 1,300 66,200 925 656

(14.5) (16,700) (1,260) (16,780) (1,540) (82,470) (2,480) 17 24,600 1,470 32,100 2,360 124.000 868 382 (19) (25,47b) (1.430) (33,300) (2,950) (129,700) (2,560) 20 25,500 1,520 35,100 2,610 134.000 3,090 290 (22) (25,290) (1.430) (32,600) (2,750) (131,600) (2,480)

(25) (27,270) (1,600) (38,800) (3,730) (142,800) (2,710) 26 27,000 1,640 39,300 2,930 144,600 3,140 143 (28.5) (26,970) (1,810) (33,900) 3,090 (144.500) (2,760) 29 26,300 2,310? 41,200 2,900 143.000 3,380 133 32 24,200 1,650 51,400 2,840 162.000 3,320 100 (26,030) (1,540) (36,000) (2,870) (154.500) (2,950)

Data are from Angino and others (1964b) and, in parentheses, Yamagata and others (1967)* 220 determined by other Investigators. All analyses in this study were performed in duplicate, except for the strontium measurements. The isotopic composition of strontium was determined on these same samples. Therefore, these analyses are more pertinent to the objectives of this study. a. Magnesium concentration Magnesium is the predominant alkaline earth In Lake Bonney. Its concentration Increases with depth and ranges from 80,5 ppm below the ice to 25*790 ppm at 30 m depthm the maximum depth of the sample profile. The profile for magnesium is shown in figure 40, The concen­ tration remains less than 400 ppm to a depth of more than 8 m. Here there is an abrupt increase to 900-1100 ppm. The concentration of magnesium then rises rapidly, and continues to Increase to about 25*000 ppm at a depth of 17*5 m. It then slowly increases to about 27,000 ppm at a depth of 28 m. In the last 2 m there is a slight decrease in the concentration of magnesium, and at 30 m, the concen­ tration is less than 26,000 ppm. The bottom samples were analysed in triplicate to comfirm that the decrease was Indeed present. A similar decrease in the magnesium content at the bottom of the lake seemed to be suggested in the analyses reported by Angino and others (1964b) and DEPTH, meters 20 aeBne. ocnrto s npm lyr boundaries lines. layer ppm; in horizontal is the by Concentration indicated are Bonney. Lake 0 3 25 iue4. ocnrto rfl o ansu for magnesium of profile Concentration 40. Figure 25 5 g pm 10* » ppm Mg, 10 15 20 221 * 222

Yamagata and others (1967). In the depth Interval of 28-30 m, the density of the brine is still increasing, suggesting that the concentrations of one or several cations continue to increase to compensate for the decrease in magnesium concentration. b. Calcium concentration The concentration profile of calcium (figure 41) is very similar to that of magnesium. From a depth just below the ice at 4 m to Just below 8 m, the calcium content remains essentially constant at about 80-100 ppm. In the interval 8.5-10.5 m, there is an abrupt Increase to 235 ppm. In the interval 10.5m to about 17.5 the concentration of calcium is increasing rapidly, from about 250 ppm to 1470 ppm. The concentration then remains relatively constant, between 1470 ppm and 1545 ppm to a depth of 28 m. From this depth to the bottom of the lake at 30 m, the concentration decreases slightly, from 1545 ppm to 1460 ppm.

c. Strontium concentration The concentration profile of strontium is very similar to that of magnesium and calcium. The concen­ tration ranges from about 0.6 ppm at the upper level of the lake and Increases to almost 40 ppm at 26-27 m depth, decreasing to 36 ppm at the bottom. The concen­ tration profile is discussed in detail in Chapter VII. 223

E

I— CL 111 O 2 0 -

2 5

3 0

0 1000 2000 CALCIUM, ppm

Figure 41. Concentration profile of calcium for Lake Bonney. Concentration is in ppm$ layer boundaries are indicated by the horizontal lines. 224 d. Sodium concentration The concentration profile of sodium is shown in figure 42. The sodium content In Lake Bonney increases from 4?0 ppm just under the ice cover to a maximum of 47»3°0 ppm at a depth of 30 m. Abrupt changes in the concentration occur at depths of 8.5 and 10. 5 m. For sodium, however, it is difficult to place a limit on the exact region where the concentration is changing rapidly, beginning at a depth of 10.5 m. In the case of the alkaline earths, the depth at which the concen­ tration becomes relatively constant following the rapid Increase below 11 m Is readily identified. The change in concentration in the sodium profile below 11 m is not as pronounced as the change in the profiles of the alkaline earths. Below a depth of 27 m, an abrupt increase In the sodium content (figure 42) accounts for the increase in the density in this region, although the concentrations of the alkaline earths are decreasing. e. Potassium concentration The concentration profile of potassium is shown in figure 43. The potassium content ranges from 23 ppm near the top of the lake to a maximum of 2,840 ppm at 28 m depth. The shape of the depth profile is similar to that of sodium down to about 28 m depth. Below 28 m, 225

ppm No i I04

Figure 42. Concentration profile of sodium for Lake Bonney, Concentration is in ppm; layer boundaries are indicated by the horizontal lines. aeBne. ocnrto i i pm lyr boundaries layer lines. ppm; in horizontal is the by Concentration indicated are Bonney. Lake

Figure Figure Depth, 0 3 5 2 20 10 o 5 O 43 Cnetain rfl fptsim for potassium of profile Concentration . K , ppm * I03 K, ppm 2 226 - o - - - - o 3 227 the concentration of potassium remains essentially constant down to the bottom of the lake.

3. Concentration of major anions

Inspection of anion concentrations in table 34 shows that the major anion In Lake Bonney is chloride, with lesser amounts of sulfate and bicarbonate. An insufficient number of analyses have been reported by either Angino and others (1964b) or Yamagata and others (1967) to provide a depth profile with enough detail to detect the fine structure that was seen in the cation profiles (figures 40-43)# There are enough data, however, to indicate that the over-all concen­ tration profile is roughly similar to that of the major cations, especially that of sodium. (The analyses of the 28.5 m and 32 m samples of Yamagata and others (1967) may be suspect. While the density reported by them is increasing, cation concentrations are decreasing.) The concentration of chloride Ion increases from about 600 ppm at a depth of 5 n to a maxiumum of about 162,000 ppm at 32 m depth. There appears to be an Increase in concentration near the bottom of the lake corresponding to increases in density and sodium concentration. The concentration profile of sulfate (table 34) 228 is less well-defined than that of chloride but the shape of the profile Is similar to that of chloride and cation profiles. The concentration of sulfate ranges from about 160 ppm at a depth of 5 m to a maximum of 3t380 ppm at 29 m depth.

4. Layered structure of Lake Bonney

Lake Bonney can be divided into 5 layers on the basis of density and concentration profiles. These layers have been numbered 1-5 from the top of the lake

Just under the ice cover downward. The boundaries of these layers have been shown in the profiles of density, temperature, and concentration.

Layer 1 extends from the base of the ice cover at a depth of 4 m to a depth of about 8.5 m* This layer is characterized by relatively constant density and concentrations of principal ions. Layer 2 extends from a depth of 8.5 m to 10.5 m. This layer is characterized by constant density and salinity, but both density and salinity are higher than in Layer 1. The exact nature of the boundary here is not clear. Samples have not been collected at sufficiently small increments in the 7-11 m

interval to define the boundary with better precision.

Layer 3 extends from a depth of 10.5 m to about

17*5 m* In this depth interval, both density and 229 salinity are Increasing rapidly downward. This layer probably represents a mixing or diffusion layer. Layer 4 extends from a depth of 17.5 m to about 27-28 m* This layer is characterized by relatively constant density and salinity. Layer 5 extends from a depth of 28 m to the bottom of the lake. In this layer, density, sodium, and chloride Ion concentrations Increase downward, while the potassium concentration remains constant, and the concentration of the alkaline earths decrease. Since concentrations of the alkaline earths in Layer 5 arc decreasing, these ions may be precipitating at the bottom of Lake Bonney. Compounds expected to precipitate are carbonates of calcium, magnesium, and strontium, and the sulfates of calcium and strontium. Crystals of gypsum have been identified from a sample of the bottom sediment of the lake by Hubert and Angino (1967). They also report the presence of "carbonate shells", which constituted 6% of their sample. These probably are of marine origin and have been transported from McMurdo Sound to the western end of Taylor Valley by a glacier moving inland. Since the shells have not dissolved, the bottom waters must be saturated with respect to calcium carbonate. The nature of the transition zone (Layer 3) between upper, relatively fresh water and the bottom brines Is of Interest. Shirtcliffe (1964) and Shirt­ cliffe and Benseman (1964) assumed that It is a diffusion boundary. Because their calculated temperature profile coincided with the profile measured in the lake, the diffusion may well be the dominant process occurring in this zone. To ascertain whether this transition zone is governed by diffusion or mixing processes, concentration- concentration curves were plotted. If mixing is the dominant process, a linear relation will result. The same relationship will result, however, If rates of diffusion of the two ions involved are similar. When the concentration of a doubly-charged Ion is plotted against the content of a singly-charged ion, greater differences in the diffusion coefficients are expected than in cases with ions of identical charge. The relationship between the concentrations of sodium and potassium are shown in figure 44. This curve suggests that the transition zone is a result of mixing, but the same result is possible If the dif- fusivlties of the two singly-charged ions are similar. The magnesium concentration is plotted against the sodium concentration in figure 4?; a deviation from a linear curve results. The shape of this curve suggests that mixing has been supplemented, probably by diffusion. Magnesium has been relatively concentrated with respect in Lake Bonney. Lake in ppm K * 10 O 2 3 Figure Figure 0 4-4. ocnrtoso sdu n potassium and sodium of Concentrations p N x 10 x Na ppm 4 231

232

4

9

o Z

o o Mg, ppm * I04

Figure 45- Concentrations of magnesium and sodium in Lake Bonney, 233 to sodium except at the lover concentrations. This possibly Indicates a more rapid migration upward in the lake by sodium ions. The calcium-strontium concentration curve is shown in figure 4-6. The presence of two straight lines are indicated, the lower one by samples above a depth of 15 no, and the upper line represents samples from 16 m depth and below. The upper line can be shifted to an extension of the lower line by increasing the calcium concentration for samples below a depth of 16 m. This is equivalent to saying that, assuming a constant chemical composition for the inflow to the lake, the water below 16 m has been depleted in calcium with respect to strontium. This may be due to the precipi­ tation of calcium salts, such as gypsum, with a propor­ tionately smaller amount of strontium having been romoved with this calcium. The calcium-strontium plot does not indicate the process occurring at the tran­ sition zone. It does seem to show the over-riding of a brine layer with water containing a higher Ca/Sr ratio than that of the deeper water. However, the break in the curve of figure 46 may indicate that all the water above the depth interval 1J-16 m has been fresh water added to Lake Bonney relatively recently. From the concentration-concentration plots, it in Lake Bonney, Lake in Sr, ppm 40 20 30 0 Figure Figure 6 4 . Concentrations of calcium and strontium strontium and calcium of Concentrations . a p * I02Ca, * ppm 6 eters m 16 5 eters m 15 5 2 0 2 234 235 appears that the boundary separating the dense bottom brines from the upper water has resulted from the upward diffusion of Ions, The nature of the boundary between Layers 1 and 2 cannot be ascertained until a series of samples collected over smaller Intervals has been analyzed.

5* The p H profile of Lake Bonney

Measurements of pH along a depth profile In the eastern lobe of Lake Bonney are given in table 35- These measurements were made in the field by D. D. Koob during the 1965-1966 field season at the time the water samples were collected. A Beckman Model N pH meter was used for the measurements. Variation within about three pH units, from 6 to 9, is indicated along the profile. Greatest variation is observed in the upper waters, particularly above a depth of 18 m and may be due to biological activity.

6. Chemical Composition of Lake Bonney

The salts dissolved in Lake Bonney consist mainly of magnesium and sodium chlorides, with minor amounts of calcium and potassium chlorides (table 3^). The molality of the principal cations in samples collected at depths of 6 , 10, 15* 2 0 , 25, and 30 m in Lake Bonney have been TABLE 35. The pH profile for the east lobe of Lake Bonne/*

Depth below pH surface

4 m 8.9 5 8.6 6 8.0 7 7.37 8 7.0

9 6.85 10 7.95 11 7.35 12 7.00 13 6.5 14 6.2 1? 6.08 16 6.1 17 6.15 18 6.4

19 6.55 20 6.75 21 6.75 23 7.35 25 7.4 26 7.45 27 7.48 29 7.45 (31) 7.32

*Data obtained in the field by D. D. Koob, 1965-1966. 237 computed and are given in table 3 6 . Relative mole percentages are also given in table 36, assuming that magnesium, calcium, strontium, sodium, and potassium constitute the bulk of the cations in the lake. The molar composition of the principal anions in Lake Bonney is given in table 37* The variation of the mole % of the anions is not as great as that for the cations. Chloride ion constitutes more than 95 mole % at all depths except for the upper part of the lake. At a depth of 6 m, chloride constitutes about 86 mole %y sulfate ion about 11 mole %9 and bicarbonate ion 3 mole %* Below this depth, relative amounts of anions other than chloride become negligible. The pH values of the lake suggest that bicarbonate will be the dominant ion of the carbonate species. Thus, the quantities of both COj- and C02 can be neglected in computing the anion compo­ sition of the lake. Therefore, salts in Lake Bonney can be considered to consist predominantly of sodium chloride and magnesium chloride, with relatively minor amounts of potassium, calcium, and strontium chlorides and sulfates. Lake Bonney is thus compositionally stratified in addition to being stratified in terras of density or salinity. In figure 47, it is readily seen that the relative proportion of sodium and magnesium are changing 238

TABLE 3 6 . Molal concentrations and the salt composition, expressed as mole per cent of the total, for the major cations of Lake Bonney

Depth below Ion molality mole % surface x 103

6 m Mg 3.793 10.44 Ca 1.420 3.91 Sr 0.00679 0.02 Na 29.97 82.48 K 1.148 3.16

10 Mg 44.63 23.17 Ca 5.489 2.85 Sr 0.0400 0.021 Na 136.2 70.70 K 6.266 3.25

15 Mg 850 .7 41.53 Ca 33.58 1.64 Sr 0.285 0.01 Na 1,122 54.79 K 41.43 2.02 20 Mg 1,069 ^ 38.03 Ca 37.15 1.32 Sr 0.444 0.02 Na 1,640 58.30 K 65.73 2.34 25 Mg 1,105 37.14 Ca 38.10 1.28 Sr 0.456 0.02 Na 1,762 59.22 K 69.57 2.34

30 Mg !,°6l n 32.91 Ca 35.78 1.11 Sr 0.410 0.01 Na 2,057 63.83 K 68.80 2.13 239

TABLE 37. Molal concentrations and the salt composition, expressed as mole per cent of the total, for the principal anions of Lake Bonney Depth below Ion molality mole % surface x 103

5 m Cl 16.7 86.22 S04 2.08 10.74 hco3 0.59 3.05 8 Cl 46.83 95.45 so. 1.25 2.55 HCO j 0.98 2.00 11 Cl 212.7 96.94 so. 3.96 1.80 HCO 3 2.75 1.25 14 Cl 1867.5 98.92 so. 9.63 0.51 HCO 3 10.75 0.57 17 Cl 3498 99.56 so. 9.04 0.26 HCO 3 6.26 0.18 20 Cl 3780 99.03 so4 32.2 0.87 HCO 3 4.75 0.12 26 Cl 4079 99.15 SO. 32.7 0.79 HCO 3 2.34 0.06 32 Cl 4570 99.21 so. 34.56 0.75 HCO 3 1.64 0.04

* Data calculated from concentrations reported by Anglno and others (1964b). 240

o No

4 2 0 4 0 6 0 100 Mole %

Figure 47. Mole per cent of sodium, magnesium, and calcium in Lake Bonney. 241 with depth in the lake. It is difficult, however, to ascertain from table 36 the variation in proportions of elements that are chemically similar. One method of investigating these changes is by use of concentration ratios, and concentration ratio profiles for Mg/Ca, Sr/Ca, and Na/K have been selected for this study. The concentration ratio profile for Mg/Ca is plotted in figure 4 8 , which also shows the boundaries of the layers of the lake. The Mg/Ca ratio increases with depth in a manner similar to the density and major ion profiles. The changes in the Mg/Ca ratio generally coincide with boundaries of the layers defined on the basis of density and concentration. Since the Kg/Ca ratio of each layer is different, two possibilities are suggested. First, the source of the water comprising each layer has differed, and with each source, composition of the salts has differed. For example, water in Layer 4 may have been derived from a source where the Mg/Ca ratio in the dissolved salts was relatively high. Water represented in Layers 1 and 2 was derived from sources whose calcium concentration was proportionately greater than the magnesium in the source responsible for the salt composition of Layer 4. A more likely explanation is that precipitation of gypsum accounts for the increase in the Mg/Ca ratio. tration ratio of Lake Bonney. Lake of ratio tration Dtpth 5 2 0 3 20 10 15 5 0 iue4. et rfl o te gC concert' Mg/Ca the of profile Depth 48. Figure O 5 Mg/Ca 10 15 242 243 This Implies that the source of the salts has been the same during the Interval of time represented by water presently In the lake. From table 37 It Is seen that there Is a relative depletion of sulfate ion from about 11 mole % at the top of the lake to 0.5 mole % at 14 m, and remains less than 1 mole % to the bottom of the lake. This depletion of sulfate Ion may be considered a result of the precipitation of gypsum. These observations can serve to propose a history of Lake Bonney, assuming the chemical composition of the inflow has remained constant. Layers 4 and ? represent the saline remains of an earlier, larger lake# As the salinity increased, eventually the water became saturated with respect to CaS04. Magnesium sulfate is very soluble in water (71 g MgS04.#7H 20 per 100 g water at 20°C; Handbook of Chemistry and Physics, 1959, 4lst edition) and therefore, this compound will not precipi­ tate at levels of magnesium and sulfate ions encountered in the densest levels of the lake. As the concentration of salts Increased, gypsum began to precipitate, sulfate was depleted relative to chloride, and the Mg/Ca ratio increased. Gypsum continued to precipitate (and may still be precipitating at the present time) until the upper layers of water were added to the lake. Chemical composition of this younger water may closely represent the original composition of meltwater that fed the older 2 4 4 lake nov represented by Layers 4 and 5* Therefore, It Is reasonable to expect that gypsum deposits of consid­ erable thickness sure at the bottom of Lake Bonney. An estimate of the expected thickness of gypsum has been made, assuming that the chemical composition of melt- water discharging Into Lake Bonney has remained the same during the Interval of time represented by the water presently In the lake. It was also assumed that there has been negligible precipitation of magnesium salts In the lake. The Increase In the Mg/Ca ratio in Layer 4 over the Mg/Ca ratio in the upper waters should represent how much calcium, as gypsum, has precipitated. At 4 m, magnesium and calcium concentrations are 80.3 and 77*3 ppm, respectively. At 28 m, the magnesium concentration is 26,8^0 ppm. If the Mg/Ca ratio at this depth was identical to that at 4 m, than the corresponding calcium concentration would be 25,780 ppm* Since the present calcium content at 28 m is 1512 ppm, a total of 24,270 ppm per kg of brine has been removed; this calcium is equivalent to 104.3 g CaS04*2H20 per kg brine. It was further assumed that this dense layer was 15 m deep (from a depth of 17 m to 32 m). For a column of brine 15 m x 10 cm x 10 cm with a density of 1.19 g/ml* & total of 1.861 x 104 g of gypsum is required to be removed by precipitation. This is equivalent to 186.1 g gypsum 2 4 ? per square cm at the bottom of the lake. Assuming that the density of gypsum is 2.32 g/cm3 (Handbook of Chemistry and Physics, 41st Edition, 1959), a total consolidated thickness of 80.2 cm of gypsum would result. Thus, It is reasonable to expect that there may be at least about one meter of gypsum at the bottom of Lake Bonney. Precipitation of calcium and magnesium carbonate may also occur in the bottom layers of brine, but these compounds are expected to be relatively minor components of the bottom sediments. The concentration of carbonate species in Lake Bonney is low throughout the lake (less than 400 ppm at any given depth, table 34), and the low acidity of the water suggests that the concentration of carbonate ion, COj", is very low. The Sr/Ca concentration ratio profile for Lake Bonney is shown in figure 49* There is an increase in the Sr/Ca ratio with depth, and the shape of the curve is similar to the Mg/Ca concentration profile (figure 48). This shape of the Sr/Ca ratio profile may represent essentially diffusion between two layers of water of different chemical composition, the water at depth having a higher strontium content relative to the calcium concentration than is in the upper water of the lake. If the source of the salts has remained the same during the history of the present lake, then a preferential depth profile of Lake of Bonney. depthprofile DEPTH, meters 5 2 20 30 Figure 49. Concentration ratio of Sr/Ca for a for Sr/Ca of ratio Concentration 49.Figure -o O j OI Sr/Ca 0.02

0.03 246 247 concentration of strontium relative to calcium is indicated, perhaps due to the precipitation of gypsum with the exclusion of strontium. The Na/^C concentration ratio profile of Lake Bonney is shown in figure JO. This plot tends to magnify relatively small fluctuations in the analyses of sodium and potassium. A constant Na/K ratio is perhaps suggested for the lake. This implies that the source of salts to Lake Bonney has remained the same throughout the history of the lake and the Mg/Ca and Sr/Ca ratios have been altered by chemical processes. It is not clear whether the abrupt change in the Na/K ratio below a depth of 28 m is real or not. Although the scale of the abscissa in figure JO is expanded, scatter of the Na/K ratio above a depth of 1J m is still considerable. Some of this scatter is undoubtedly due to analytical procedures, but that may not account for all of the fluctuation. Another possible explanation is the effect of biological activity that can exhibit a preference for particular alkali metals. Since most of the phytoplankton productivity occurs under the more favorable light and temperature conditions of the upper 1J m (Goldman, al., 1967), these organisms may be affecting the Na/K ratio. 248

III 20

25

3 0 -

12 14 16 N a / K

Figure ?0. Concentration ratio of Na/K for a depth profile of Lake Bonney. 249 7. Trace element concentrations in frato foaneg Some trace element analyses of water from Lake Bonney are available. Armitage and Angino <1967) have determined manganese, nitrogen (as N02**7), phosphorous (as total PO*"*?), and iodine (as I 0 4."’?), They did not indicate the exact anion species determined. In a depth profile, the following ranges were reported: "IO*11, 0.0-4.5 ppm; "N", 0.0-1.0 ppm; "P04"9 0 . 0 - 1 . 0 ppm; and Mn, 0 . 0 - 8 . 0 ppm. Variation with depth within these concentration ranges is erratic, and the shape of the depth profile changed during the course of the summer. A maximum concentration for each of these elements was observed at a depth of 14-15 m. This is also the depth at which maximum productivity occurs, suggesting a close dependence of these profiles with biological activity of the lake. Two measurements of the uranium concentration of Lake Bonney have been made (Thurber, Al*» 1968). At depths of 10 and 27 m, the concentrations of uranium 234 238 were 2 and 47.5 ppb, respectively. The U /U activity ratios of these 10 and 27 m samples were 2,8 - 0.2 and 3.9 * 0.1, respectively. Additional trace elements were measured for "bottom waters" of Lake Bonney by emission spectral 250 analysis (Boswell, £& Ai*» 1967)* The elements and their respective contents were reported as follows: Zn, 150 ppm; Pb, <30 ppm; Bi, 7*1 ppm; Fe, 64O ppm; Kn, 23 ppm; and Mo, 8.7 ppm. Boron analyses for a depth profile in Lake Bonney have been reported (Angino, £t §,1*» 1964b). At a depth of 11 m, the boron content is 2*0 ppm, at 14 m, 14 ppm, and then throughout Layers 4 and 5, the concentration range is 30 to 38 ppm. Angino and others (1964b) also report analyses for IO*"*, PO*"3, and NOj", but apparently these results are superceded by their later analyses (Angino and Armitage, I967). Yamagata and others (1 9 6 7 ) also report a few boron analyses for Lake Bonney. In one profile, three analyses have been reported: at 5 m, the boron content was 0.074 ppm; at 14 m, 17 ppm; and at 30 m, 3*8 ppm. These values are markedly lower than those of Angino and others (1964b). It is not known whether the difference between the two sets of data is real or not. This writer suspects that the problem may be analytical, because boron Is one of the more difficult elements to determine accurately. A rubidium and cesium analysis has been reported for a sample from a depth of 3° m in Lake Bonney (Yamagata, fii Aik*9 1967). The rubidium concentration 2 ? 1 was found to be 605 ppb and the cesium concentration was 81 ppb. For this particular sample, sodium, potassium, and chloride concentrations were 35*620 ppm, 2,714 ppm, and 116,000 ppm, respectively.

F. The Geochemical History of Lake Bonney

A history of Lake Bonney may be proposed on the basis of the chemical compositions of the brines. The dense brine below 17*5 m may represent a residue of the earlier, more extensive Glacial Lake Llano. This larger lake was gradually reduced in volume by an Imbalance of sublimation and evaporation over discharge. The predicted deposits of gypsum possibly could have formed principally at this stage. The discharge may have become limited to summer meltwater streams Issuing from alpine glaciers surrounding the drainage basin of the present Lake Bonney. This meltwater Is essentially fresh and would over-ride the denser water, the remnant of Lake Llano. This hypothesis requires that the lake be ice- covered at this time since the chemocline is preserved. If the lake were not Ice-covered, it is quite possible that mixing could have occurred by wind action upon open water. The strand lines above the lake are composed of laminated, well-sorted, sand-size material (G. H. Denton, 1969* personal communication)* These 252 beaches of the Lake Llano stage suggest that the lake had been open water, at least for part of the year. These beaches were probably produced by wave action more Intense than that observed in the narrow moat at the present time. Concentration of the salts in Lake Bonney probably occurred in the same manner as in Lake Vanda. When water at the base of the ice freezes, a large proportion of the salts is excluded from the ice. Thus, salts accumulate at the base of the ice cover. Convection within this layer will tend to produce a constant concen­ tration, as shown by the profiles of the principal ions. The proposed mechanism suggests that each year meltwater is added with a salt content less than that in Layer 1. As the ice sublimates from the surface of the lake and additional ice forms at the base, salts are excluded and remain in the upper water* This now brine will sink until its density is that of the water it displaces. Thus, the salt content of Layer 1 will increase in time, if all present conditions remain the same, I.e., if the lake level is constant or decreasing. Little is known about the age of Lake Bonney. The present lake is po-sibly the remnant of a larger lake that was present after the maximum of the Fryxell Glaci­ ation (Peve, 1966). The age of this glaciation in Taylor Valley is not yet known. Onset of a later, relatively minor glaciation in Taylor Valley probably occurred at least 34,000 years ago (tenative date assigned by G. H. Denton, 1969)* This glaciation did not extend over the region presently occupied by Lake Bonney. Therefore, it is possible that the lake is older than 34,000 years. An interesting observation can be made about the most recent history of Lake Bonney. Between the years 1903 and 196d, the lake is known to have undergone an increase in level of 9*2 m (Shirtcliffe, 1964). There­ fore, it may be significant that in the series of sample collected during the 1969-1966 field season, the major chenocline occurs almost exactly 9*2 m below the ice cover (figures 40-43)* Water added to Lake Bcnney since about 1903 could have been added in two ways: (1) rapidly, implying a catastrophic source, such as a volcanic eruption, or (2) relatively slowly, by a short-term change in climate, accompanied by an increased discharge into the lake. On the othor hand, an advance of the front of Taylor Glacier would also tend to raise the level of the lake. However, photographic evidence compiled by Pewe and Church (1962) indicates that the front of Taylor Glacier has remained in about the same position since at least 1911. An Indication of the rate of increase of the lake level s^nce 1903 can be obtained by examination of the width of the channel that connects the two lobes of Lake Eonney (figures 9 and 37). Three measurements of this channel have been made. In 1903* its width was 5.2 m; in 1911* the width had increased to 31 and by 1964, it had increased to 42.3 m (Shirtcliffe, 1964). If it is assumed that the walls s u r r o u n d * L a k e Bonney have constant slope, then the rate of change in channel width reflects the rate of change of the lake itself. This serves only as a rough estimate, but it nay indicate whether the addition of water occurred slowly or rapidly within the span of a very few years. The channel is plotted versus time in figure 51. This plot suggests that the change in the width of the channel was rapid in the interval of about 1903-1911. Since that time the width has been increasing at a decreasing rate. This suggests that water above the major chemocline in Lake Bonney was added very rapidly during 1903-1911, and since then, the rate of discharge of meltwater has decreased. The amount of water added annually to the lake appears to be greater than the amount lost by sublimation, since extrapolation of the curve in figure 51 to 1969 suggests that the channel width could be widening at the present time. If this is true, then the level of Lake Bonney should still be C 30 ony ic 1903* since Bonney Channel Width 40 20 Figure 51. Change in channel width of Lake Lake of width channel in Change 51. Figure 90 90 90 I960 1940 1920 1900 Y«ort A. D. 255 Increasing. This interpretation implies I.hat there was a short-terir climatic event, resalting in. increased discharge of meltwator into Lake Bonney. Shirtcliffe (1964) has calculated an age of 6C years (before 1°C^) for an event, when a layer of fresh water 10 m deep flowed onto the surface of a saline, conveoting lake (Lake Bonney shortly prior to 19^3). lie considered factors including insolation, geothermal gradient, and diffusivities, and assumed that once the fresh water was added to the lake, the bottom layer (Layers 4 and 5 of this study) became non-convecting. The age of 60 years was then calculated from the analysis of the concen­ tration gradient of chloride ions of the major chemc- cline. This date agrees remarkably well with the information derived from a consideration of the widening of the channel. The only other dates for Lake Bonney cited in the literature are C-14 dates of mummified seal carcasses found in the vicinity of the lake. These dates range from 300 to 1,250 years B.P. (samples determined by E. A. Olsen and W* 2. Broecker, Lament Geological Observ­ atory; quoted in Angir.o and Armitage, 1963). Prom these C-14 dates, Angino and Armltage (1963) suggest that the climate in Taylor Valley has been cold and dry for at least 30Ci years. Later, this minimum length of the cold, dry climate was extended to include the maxi mum C-14 date of about 1,200 years B.P. (Angino, at al., 196413).

G. Chemical Composition of "Taylor Red Cone"

"'he "Taylor Red Cone" was a reddish-yellow Ice cone built from br-tne discharged from Taylor Glacier at its terminus (figures 37* ?2, and 53)« This phenomenon is of particular interest because it is possible that substantial amounts of salts may have been added to lahe Penney by similar discharges of brine in the past. The exact tine of development of the "Taylor led Cone" is not known. Tt was not seen on aerial photo­ graphs taken luring the 1958-1959 field season. A siral1 reddish-yellow stain had been seen from a 31 stance in November, 1961. By November, 1962, the oor.c had apparently attained its maximum dimensions (Black, gt al.. 196?). Tt had disappeared completely by the ln^?-196B field season. According to Black and others (1965), brine emerged mainly from cne orifice about 20 n above the base of the ice front, estimated to be 30-35 m high. The cone was ■'0-15 m wide at the base and 20 m high. The total volume of the discharge, not all of which may have been salt- rich, was estimated by Black and others (1965) to be about 3,000-6,000 cubic meters. "he red color of the "Taylor Red Gone" was due to 2 58

Figure 52. "Taylor Red Cone." Taylor Clacier extends across the center of the photograph. Lake Bonney is in the left foreground. (Photograph courtesy of D. D. Koob) 259

Figure 53. "Taylor Red Gone,” a close-up. (Photograph courtesy of D. D. Koob) 260 iron oxides (Black, £t , 1965)* Locally "flowers"

(I.e., masses of crystals) rose about one cm above the cur face. 'They were white, light gray, and various shales of yellow, tan, orange, ana red. Shallow brine

o'--!:: were formed on the -uter limits <~-f ‘he cone. A partial study of the crystallized salts indicated that aragonite was the principal compound. The presence ~f calcium and magnesium chlorides was also indicated. 1artial chemical analyses ^f water-soluble material f r ' tha ice cone are given hi table 3®« The sample ■j-'l looted by D. D. Koob was meltwotor. The material analyzed by black and others (196?) was a composite of salt, brine, and ice; only their wutei sclable analysis is ,’vc;. here. Concentrations of tho principal cntl ^i.s ,f tho 7 tn and 2<+ m samples from Lake bonney have been included for comparison in table 3^«

Tt is not known whether salts are being discharged

periodlcully Into Lake Bonney in this fashion, or whether this "Taylor Led bone1' was an unique event.

Cince so few samples bad been collected, a reliable

determination of the chemical co:.pDsiti of Ihe

saline JisehurgG Is not possible.

a s* *lur event may have occi. red some time bef-rc

1??R, and only the later stages were J- nvos'.l gated

(hamilton and others, 1°62). In deeowber, 19?p, 261

TABbii 3^. Partial chemical analyses of water-soluble salts from the "Taylor lieu Cune." Concentrations aru In ppi:;.

I n 2b lake Bonney 7 motors S'* meters

- +2 ud 3l,5°o 010 7°.7 l,5li +2 *-*e 17,500 294 135.5 26,520

» ! . + nia 32P,C00 4 ti?0 629 32,900 + K 6,100 1P0 44.9 2,67c ll < j r> 0 h c o 3“ 9,coc 66,50C

01 ~ 536,400

i-ti/Ca c. 56 1.3 1.7 17-5 hu/K 53. 3 23.2 15.3 H .6 ar.laoV. and others (1965). b,. nJiiHi [- X 6 collected by D. 3. K-jeb, January, 1966. 262 liar; 11 Lon and cUiers (1?6.?) collected saline waters and sediment from the surface of an ice platform famed at the base along the north edge of the Taylor ulawl or . This is the come locality -where the ''Taylor del done" forme I later. Small saline po^ls and salt encrustations were observed on the surface of this pla? form, which was triangular in plan, 400 x 4CC x 60 feet, A partial chemical analysis of a brine from cue of the p'-ols on the surfaco of the platform indicated the following anion eoueer.trations in .eight %: 1C3=, 5.4^5 HC03" 6.1^} Cl", 6.0^; and S04“, 79/^ (iiaiui 1 ton, et iii,., 1962). Although neithei calcium n~r magnesium were determined, the chemical compos!Mon of this brine was stated to be primarily sodium sulfate, since sodium accounted for 85 equivalent % of the total ani-n content. Galt crust was petrographically examined and was founl to contain about 95# halite, possibly with some occl dod cubes of sylvite. dalcite and gypsum were recognized, and presence of polyhalite (K2S04*r!gS04* 2Gu304»2H20) and epsomite (l'igSC4.*7H20) was indicated. The source of the salts on this ice platform is not known. This platform appears to have been present when the "Taylor Bed Cone" was visited by Black and others (1965). Their photograph of the cone shows a 263 surface onto which the cone formed that is similar to the surface of the'platform photographed by Hamilton and others (1962), It is possible that salts on the surface of Die ice platform accumulated slowly over a period of many years. These salts, however, nay also be the remnants of an earlier event similar to that resulting in the building of the "Taylor Red Cone." By the summer of 1966-1967, little of the cone was left, and now the only trace that remains Is the salt accumulations on the ice platform. SHATTER VII

THE ISCTOPIC C O I i r O G J CF f/MOVTIUK IN TAYLOR VALLEY

A* Introduction

The salts found l.n i^ake Bonney have "been attributed various sources, including: (1) sea water, either dir -ot-ly by evaporation or by wind transport; CP) volcanic activity; and (3) chemical weathering of local soil and bedrock* In the past, various chemical and physical parameters have been used in attempts tx, link the salts in Lake Bonney with one or more sources* These studies have involved comparison of concentration ratios of various ions, both major and trace ions, in the lake to corresponding ratios in the possible sources. Activity ration of uranium isotopes (U23*/U238) have also been used. These previous studies, however, have resulted In conflicting conclusions, with no specific source(s) bei ig unambiguously identified as the principal source of the salts in Lake Bonney. 264 26? In this study, the isotopic composition of strontium of Lake Bonney was measured in a series of water samples collected at different depths. The Sr87/SrS^ ratio of the salt in the lake was then compared with the Sr87/Sr86 ratios of possible sources of the salts. This

On. fl / required measurement, of Sr f/St ratios of the doss Sea, KcMurdo volcanics, and salt from the local soil. Unfortunately, the collection of soil samples from Taylor Valley was not as complete as that of Wright Valley. Only four samples were obtained. These soils represent the entire length of Taylor Valley and only one pertains specifically to the drainage basin of Lake Bonney, No samples of meltwater from streams discharging into Lake Bonney were available. The sample locations are shown in figure 54.

B. Lake Bonney

The isoto/ic composition of strontium in Luke Bonney was determined for 7 samples of brine from a suite of samples collected at different depths by D* D* I'oob during the 1965-1966 field season. Three samples were analysed in duplicate. Results of the Sr S 7 /Sr~n 6 ratio measurements aro given in table 39 and shown in figure 55- Tho concentration of strontium, like that of the other major ions (Chapter VI), varies with depth, ranging from 5 km.

• water sample ▲ soil sample

4 T A Y L 0 R 4 Lake Fryxetl Q Lake Bonne$ )/ McMurdo 9 Sound RP

Figure 5^* Location of water and soil samples from Taylor Valley, 266 267

TABLE 39* The isotopic compositiDn and concen­ tration oT strontium from a depth profile of Lake Bonney

Lop til below 3ra?/Sr86* 3= Grt ppm surface

0.7345 5 0.7130 0.0003 0.5927 o 0.5949 7 0.8124 R 0.7130 0.0002 0.7150

9 3.026 10 3.502 11 7.066 12 0.7131 0.0004 12.22 13 15.95 14 21.94 15 24.98 16 0.7130 0.0002 33.69 0.7133 0.0002 18 37.19 20 0.7129 0.0002 38.89 0.7129 0.0005 22 33.70 23 32-37 24 0.7130 0.0003 33.93 0.7129 C .0005 25 39. °2 26 39.02 27 39.41 28 33.21 29 0.7133 0.0002 37.02 30 35.88

average = 0.7130 & Corrected for fractionation assuming Sr8V 3 r 86 = 0.1194. 0 3 5 2 20 DEPTH, meters CM m f totu ln et poieo ae Bonney. Lake of profile depth a along strontium of Figure 55* Isotopic composition and concentration concentration and composition Isotopic 55* Figure TOTU, ppm STRONTIUM, 20 0 4 0 3 268 269 0*73? Ppm at a depth of 4 m to a maximum of 39*9 ppm at a depth of 2? m. The strontium content then decreases to 35.9 ppm at the bottom of the lake at a depth of 30 m below the surface. Chemoclines for strontium occur at identical depths as those of the other principal ions; i.e., 3.5, 10.5* and 14 m. In contrast, the isotopic composition of strontium was constant within experimental error throughout the entire depth profile of Lake Bonney (table 3° and figure 55). The Sre7/Srti6 ratio at the 997

C. "Taylor Bed Kelt"

The isotopic composition of strontium In salts o° the "Taylor Bed Cone" was of particular interest because the possibility that this phenomenon may be contrib­ uting significant amounts of salts to Lake Bonney. One sample of the meltwater from the "Taylor Red Cone" was 27C available for analysis, and the Sr87/SrC6 ratio was found to be 0,7136 (table 40). This ratio Is very similar to the average Sr8?/Sr86 ratio for Lake Bonney (0,713°)• Although the Sr87/Sr86 ratio for the "Taylor Red Kelt" is slightly higher than the average value for the lake, the difference is probably not significant and, for the present, the isotopic composition of strontium fa . m the cone may be considered identical to that in Lake Bonney, The Sr87/3r88 ratio of the strontium In the "Taylor Red Cone" indicates that the strontium discharged into Lake Bonney during the time the cone existed was essentially of identical isotopic composition to the strontium presently in the lake. However, the contrib­ ution of significant quantities of salts to the lake by phenomena similar to the "Taylor Red Cone" is not known.

D, "Suess Pond"

The isotopic composition of strontium was measure! for water from a pond between the Suess and LaCroix Glaciers. (The name "Suess Pond" is unofficial, but it serves to identify it for this purpose.) The Sr87/Sr86 ratio was found to be 0.7112 (table 4C). This ratio is intermediate between the Sr87/Sr86 ratio of Lake Bonney (0,713°) and that of the Ross Sea (0#7°94). Therefore, 271

TABLE 40. Sra7/SrB6 ratios of "Taylor Red Cone," "Suess Pond," Lake Fryxell, and water-soluble salts from soils of Taylor Valley

Gambia No. Description Sr8?/Sr86*

TRM-1 Water, "Taylor Red Cone" 0.7136 C.C002 DV-66-019 Water, "Suess Pond" 0.711? 0.0CC3 OV-6«-001 Water, Lake ^ryxell 0 .70°0 0.0004 jY-6 ft-2 01 Soil, near Lake Bonney, 0.7136 0.0003 water leach UV-68-201 Soil, near Lake Bonney 0.7136 0.0003 total salt-free soil 0 0 0 O • i)V- 63-202 Soil, near LaCroix 0.7125 Ui Glacier, water leach LV-6R-203 Soil, near Canada 0.7101 0.0003 Glacier, water leach

DV-6P-204 Soil, near Lake Fryxell, 0.70R9 0.0003 water loach 0 .70?2 0.0006

*Coirected for fractionation assuming Sr®*/Sr88 = 0.1194. 272 this pond is probably receiving salts from a source that is not the same as that of Lake Bonney, nor has the bulk of the strontium been derived from a marine source,

E, Lake Fryxell

The isotopic composition of strontium was deter­ mined for a sample of water collected from the moat at tho edge of Lake Fryxell, The Sr8?/Sr88 ratio was 0,7090 (table iO) and is significantly different from that of Lake Bonney (0 ,7130), but essentially identical to that of the Ross Sea (0,7094), The altitude of Lake Fryxell is only about 22 m above sea level (Henderson, si al.f 1966) and the lake is less than 5 km from Mcl'uxdo Sound, Although the saline waters In Lake Fryxell could represent the remnants of Glacial Lake Llano (Tewe, 1966), it is also reasonable to expect that much of the salt in the lake could be of marine origin. The highest strand lines at nearby Marble Point (figure 10) are about 2C m above present sea level (Nichols, 1965, 1966). Thus, it is possible that the eastern portion of Taylor Valley has been flooded mainly from the sea. Salts in Lake Fryxell may also be derived from the sea by wind transport, since onshore winds prevail at the eastern end of the 273 valley, Chemical parameters, such as IICC3/C1, S04/C1, K/Ra, and Ca + Mg/lC + Na, used by An^lno and others (1962), have indicated that the salts in Lake Fryxell may be derived from (1 ) ocean spray, (2 ) efflorescences in soil, and (3 ) thermal waters. They considered relict sea water as a source of the salts as improbable. The SrC7/SrQ* ratio of water from the lake indicates that the bulk of the strontium has been derived directly or indirectly from the sea. This conclusion can be extended reasonably to include most of the calcium, magnesium, and chloride ion, as well as sodium and potassium. Differences in ratios of major ions from those of sea water are probably a result of precipitation of insoluble compounds such as alkaline earth carbonates and sulfates.

F. Soil

Four samples of soil were collected from Taylor Vaaiey by K. R. Everett and R. E. Behling. These were obtained from (1) the area east of Lake Bonney; (2 ) the valley floor at the base of the LaCroix Glacier; near the foot of the Canada Glacier; and (4.) near the edge of Lake Fryxell, Water-soluble salts were

f t m o / removed from these soils and the Sr /Sr ratios were 274 measured (table 40),

The Sr8?/Sr86 ratio Tor the water-soluble Traction of salts in the soil sample from the vicinity of Lake Bonney was found to be 0 .7136. The Sr87/Sr66 ratio for the total salt-free soil was also 0 .7136* 67 8 6 This value is similar to the average Sr /Sr ratio of Lake Bonney (0.7130)• The strontium in the water- soluble salt is significantly different from strontium f^und In the 3oss Sea (0.7094) and. McMuido volcanics (0,70-10). The bulk of the strontium appears to have been derived mainly from a source other than the sea or volcanic activity. Since strontium in Lake Bonney 8 7 has a higher radiogenic Sr content than that of either the linss Sea or McMurdo volcanics, it is probable that the bulk of the strontium in the lake has been derived fron local bedrock. A soil sample was collected near the foot of the LaCroix Glacier, located to the east of Lake Bonney (figure 9). Keltwater of this glacier discharges into i-uke Bonney. The Sr87/Sr86 ratio of water-soluble salts in this soil was 0.712?. This value is not significantly different from the Sr8?/Sr86 ratio of Lake Bonney. Strontium of this composition is probably being trans­

ported to Lake Bonney by meltwater streams. Water-soluble salt from a soil sample collected near the foot of the Canada Glacier has a Sr®7/Sr86 ratio 27? of 0*7101. This strontium Is significantly less radio­ genic than that in Lake Bonney. Meltwater from this glacier does not enter Lake Bonney. Instead, it enters La^e Fryxell to the east and Lake Chad to the west (figure 10), ’While strontium in salts of most of the previously discussed soils in Taylor and Wright Valleys had 87 . 66 Gr /Sr ratios similar to that indicated by the parent material, this soil from the eastern part of Taylor Valley did not. Parent lithology of this sample was similar to that on the floor of Wright Valley; i.e., It was composed mainly of granites, gneisses, and dolorite. Volcanisn is not as prevalent in the eartern part, of Taylor Valley as it is farther to the west (compare figures 8 and 9 with 1 0 ) . Thus, It does not appear that the decrease In the Sr°7/Sr86 ratio is due mainly to the presence of volcanics. This sample, however, comes from a locality nearer to McMurdo Sound than those samples from the LaCroix Glacier and Lake Bonney. Since there is no bedrock threshold In Taylor Valley, onshore winds can carry precipitation further upvalley. Also, the elevaticn of the valley floor here 1s less than ?0 m above sea level. Therefore, it Is possible that most of the salts In the oclls In the eastern part of Taylor Valley have been derived fr^m 2 7 6 the sea, either by direct evaporation of sea vater or by aerial transport* The fourth soil sample was collected near Lake Fryxell* Duplicate determinations of the water-soluble material gave Sr87/Sr88 ratios of 0*7089 and O.7092 (table 40). Thus, the lsotopic composition of strontium in salts and soils near Lake Fryxell is Identical to strontium in both the lake and the Boss Sea. This suggests that salts found in both the soil and Lake Fryxell have been derived from a marine source*

0* Conclusions

The Sr87/Sr88 ratios for samples of the lakes and soils of Taylor Valley decreases from the western end of the valley to the east* The Sr87/Sr88 ratios for Lake Bonney, soil from the lake basin, and "Taylor Red Cone" were found to be 0*713°* 0 *7136, and O.7 1 3 6 , respectively. A few kilometers to the east, salt near the LaCroix Glacier has a Sr87/Sr86 ratio of 0.7125. "Suess Pond," east of the LaCroix Glacier, contains strontium with a Sr87/Sr8* ratio of 0,7112* Salts in soil from near the Canada Glacier, still farther to the east, contain strontium with a Sr87/Sr88 ratio of 0*7101* Finally, farthest to the east and nearest to McMurdo Sound, the strontium in Lake Fryxell and salts In the nearly soil had the lowest Sr8?/Sr8* ratios measured, 0,7090 and 0,7091, respectively. This gradual change In the Sr8?/Sr88 ratio Is shown graphically in figure 56. The Sr87/Sr86 ratio decreases as McMurdo Sound Is approached. By the time Lake Fryxell is reached, all the strontium in the soluble salts Is identical to marine strontium. u. w

0.714

0.713-

C0 0.712-

• 0.711-

0.710- kin

Figure ?6» Variation of the Sr07/Sr86 ratio along a longitudinal profile of Taylor Valley.

to -a 03 CHAPTER VIII

THE CASE FOR CHEMICAL WEATHERING Ift.WRIQHT TIMET

A. Introduction

Evidence was cited In Chapter IV for the occurrence of chemical weathering In Wright Valley, This was based on the observation that the Isotopic composition of strontium of the water-soluble salts within the soil Is essentially Identical to that of the parent material. This appears to be unpredicted on the basis of present knowledge of chemical weathering. Therefore, a more extensive study of the Isotoplc composition of strontium In salts was undertaken. This study was centered around the Meserve Glacier for three reasons: (1) the lithology of the soils varies greatly, ranging wholly or partially from basalt to lolerite to granite and :;chlst; (2) the area of study Is located more than 500 m above the valley floor, thereby decreasing the possibility of mixing of the salts by meltwater; and (3) a large number of soil samples were made available for analysis by R. E. Behling and K. R. Everett. ?so

B. Salt Accumulation, Chemical Weathering, and Age of the Salts

Large accumulations of salts are present in the ice-free valleys. These occur as lenses and cement within the soil (figure 57), and as encrustations on the surfaces of pebbles (figure 58). From a study of the isotopic composition of strontium in the salts along the floor of Wright Valley and in Lake Vanda (Chapter IV), it was concluded that most of the strontium, calcium, and possibly the magnesium were derived from the local bedrock, while most of the chlorine and perhaps sodium were of marine origin. The most common soils of the ice-free regions are ahumlc, structureless, and coarse-textured. Because of the arid conditions, Antarctic soils are desert types. Most soils are saline, the pH is relatively high (8-9), and, where sufficient water has been available, migration of salts has taken place (Ugolini and Bull, 1965). Soil profiles are weakly developed In some ice-free areas (e.g., Claridge, 1965? and KcCraw, 1967a). Early investigators had considered physical weathering to be the predominant process of soil formation (figures 59 and 60). A study of a quartz Figure 57* Salt accumulations in an excavation in the outer moraine, Meserve Glacier, Wright Valley. (Photograph courtesy of R. E. Behling) 2 8 2

Figure 58# Cross-section of salt- riven schist* The white patches are salt accumulations* Sample collected from Wright Valley by R. E. Behling. Flgur e 59• Caverno usly-veathered coarse-grained diabase boulder, Wright Valley. (Photograph courtesy of R« B. Behltng) 2 84

Figure 60* Ventifacts of Ferrar Dolerlte, Wright Talley* (Photograph courtesy of G. Holdsvorth) 28? diorlte at Marble Point east of Wright Valley suggests that physical processes may predominate over chemical weathering (Kelly and Zumberge, 1961)* Although the diorlte appeared to be strongly chemically weathered, the bulk chemical and mlneralogical composition of material from various "stages" remained essentially constant throughout the observed weathering sequence. Since the climate at Marble Point Is different from that of Wright Valley (Chapter II) and the diorlte samples were collected near the shore of McMurdo Sound, the conclusions of Kelly and Zumberge (1961) perhaps cannot be extrapolated to weathering phenomena in Wright Valley. More recent studies, however, Indicate that chemical weathering has been or is now occurring at a rate greater than was originally believed (Claridge and Campbell, 1968; Tedrow and Ugolini, 1966). Many investigators cite the occurrence of salts on and within the soils as evidence of chemical weathering (e.g., Campbell and Claridge, 1967; Claridge, 1965; Claridge and Campbell, 1968; McCraw, 1967a; Tedrow and Ugolini, 1966). Clay minerals are present in the soils of the ice-free valleys. Claridge (1965) concluded that clays are actively forming at the present time and that they are derived by slow hydration of micas. Many soils 286 contain up to l-2Jf clay, consisting principally of vermlculite, montmorlllonite, and some chlorite. Chemical weathering has undoubtedly been respon­ sible for the release of some elements from the local bedrock. However, the age of these salts is controversial* Two possibilities exist: (1) the salts are "fossil salts" formed during a period when the climate in Wright Valley was more amenable to chemical weathering; or (2 ) the salts are young, possibly forming at the present time but at a slow rate. G, Faure (personal communication) considered it likely that the salts in the ice-free valleys are a relict feature of a former warmer and more humid environment, perhaps during the Tertiary Period. Part of his argument is based on the presence of salt accumulations in till underlying the Meserve Glacier and in the basal layers of ice, where chemical weathering is not occurring at the present time because the ice temperature is less than 0°C at all times. If glaciation of the Antarctic continent were initiated about 20 million years ago (Rutford, at fli*, 1968) and has been continuous since that time (C. Craddock, 1968, personal communication), then these salts, if they are "fossil salts," may be at least 20 million years old. From studies of the mass balance of the Keserve 2P7 Glacier, Bull and Carneln (1968) have shown that "an advance of 400 meters (by the glacier) ... could be produced very quickly by a very slight (increase) in the annual accumulation (rate)." The glacier advances by a dry-calving process, which tends not to destroy or rework debris and salts previously exposed on the surface. Thus, it is possible that the salts presently observed under the ice have been over­ ridden in relatively recent times. Sorting studies of unconsolidated material under the Meserve Glacier by G. Holdsworth (1969* personal communication) indicated that this material has been wind-blown. These data suggest that the Meserve Glacier has retreated a distance at least as far as the position of the tunnel (the sampling sites). Since one of the materials studied was McMurdo basalt, the glacier appears to have retreated before the volcanics were erupted. Then, with readvance of the glacier, the wind­ blown material was overridden, presumably by the dry- calving mechanism described by Bull and Carnein (1968). There is evidence in the ice-free valleys of a former, wetter, and probably warmer, climate. For example, the lakes were at higher levels, as indicated by elevated strand lines (figure 61). Glaciers have been more extensive than present as Indicated by their Figure 61, Lake Vanda, Wright Valley. View is to the vest; the North Fork Is in the center background. The moat is visible at the edges of the lake and the strand lines are visible to the north. (Photograph courtesy of R. E. Behling) moraines, Geomorphic investigations by Calkii, (1964) in Victoria Valiey revealed the presence of many mud­ flows in addition to much larger solifluction lobes, which transported more material than those now active in the area, Mercer (1968) cited evidence from lake sediments and solifluction flows in the Reedy Glacier area in the Transantarctic Mountains about 1200 km south of Wright and Taylor Valleys, that the climate was approximately 6 tc 10°C warmer than present. Mercer (1969) suggested that a high sea level, dated at about 120,000 years B. P*, represents the most recent time that this increase in temperature is likely to have occurred, A series of three well-defined moraines surround the Meserve Glacier (figure 62), (For reference, these will be referred to as the "outer" (oldest), "middle," and "inner" (youngest) moraines.) The moraines on the west side (to the right of figure 62) are darker due to the presence of KcKurdo basalt. This basalt has origi­ nated from two cinder cones in the accumulation basin of the glacier. One of these cones is shown in figure 63* It has been suggested that additional cones exist under the ice (R. E. Behling; G. II, Denton, 1969; R. J. E. Kontigny, 1967; personal communications). On the east side of the Meserve Glacier, McKurdo Figure 62. Aerial view of the Meserve Glacier, Wright Valley, looking south. The cinder cones are in the accumulation basin hidden behind the peak at right. (Photograph courtesy of G. Holdsworth) 2 9 1

Figure 6 3 . Basaltic cinder cone, western edge of the accumulation basin of the Meserve Glacier, Wright Valley. View is to the south. (Photograph courtesy of R* E. Behling) 2°r? volcanies are incorporated within only the two inner moraines. Thus, these inner moraines have been deposited since the basalt was erupted. All the McMurdo volcanies that have been dated in Wright and Taylor Valleys fall within the limits of 2-4 million years (Armstrong, e£ al». 1968; Denton and Armstrong, 1968), Although basalt from the area surrounding the Keserve Glacier has not been dated, it is reasonable to suggest that the age of 2-4 million years would also apply to these volcanies. Then, the age of the outer moraine is greater than 2-4 million years, and the inner moraines are younger than 2-4 million years. Basalt incorporated within one of the inner moraines has apparently undergone chemical decomposition (K, R, Everett and R. E , Behling, 1968, personal communication; shown in figure 64). Thus, chemical weathering has occurred since that moraine was deposited, i.e., sometime since the basalt was erupted, perhaps 2-4 million years ago,

C, Chemical Weathering in Wright Valley

In Wright Valley there is isotopic equilibrium of the strontium between water in the Onyx River and water- soluble salts and parent material from the valley floor (Chapter IV), This conclusion has been confirmed 293

Figure 64. Pit in the middle moraine, vest side of the Meserve Glacier, Wright Valley. Sample WVM-22. The light-colored patch is a weathered remnant of McMurdo basalt. (Photograph courtesy of R. £• Behling) 29* by Dasch (1969)* who also found that the Isotopic composition of strontium in weathering products is largely unaffected by weathering processes. It Is interesting that the materials analyzed by Dasch (1969) did not originate in a polar desert environment like that of the ice-free valleys. Instead, his samples were collected in warmer, more humid regions such as Hong Kong, Connecticut, New Hampshire, and Georgia. The isotopic composition of strontium was measured during this study for sediment and water from Lake George, Ontario. One sample of each was collected during the summer, 1968. The Sr8?/Sr86 ratio of the water was 0.7184, while that of the sediment was 0.7288. The strontium In the water is clearly not in isotopic equilibrium with strontium in the sediment. This isotopic disequilibrium between the water and sediment may be due to differential weathering of those minerals which have low Rb/Sr ratios. Hart and Tilton (1966) have analyzed the isctopic composition of strontium in sediment and water from Lake Superior. The Sr07/Sr88 ratio of the sediment was 0-739* while that of the water was 0.718. The strontium in this water is clearly not in equilibrium isotopieally with strontium In the sediment, a conclusion identical to the results obtained for Lake George. They attributed 2ofr this difference in the Sr8?/Sr88 ratios to differences in susceptibility to weathering of various minerals. The more resistant minerals, such as mica and potassium feldspar, have high Rb/Sr ratios and contain strontium with a relatively high Sr87/Sr86 ratio. More readily weathered minerals, such as carbonate, pyroxene, and plagioclase, have low Rb/Gr ratios and therefore have correspondingly lower Sr87/Sr86 ratios. A suite of samples was selected from the moraines of the Kecerve Glacier (figure 65) to provide a wide range in lithology. The sample locations are shown in figure 66 and the lithology of these samples represented by pebbles greater than 1/4 inch diameter is given in table 41* The salts were removed by water leaching, using the procedure described in Appendix A. Two samples (WVM-22 and WVM-23) were analyzed at two different 87 . 86 horizons* For many of the samples, the Sr /Sr ratio was also determined on the salt-free material (table 4 1). Three observations may be made by inspection of table 41* (1) The Sr87/Sr88 ratio of the soluble salt and lithology appear to be essentially constant within a given profile (WVM-22 and WVM-23). (2) The Sr87/Sr66 ratio of the water-soluble salt is generally similar to the Cr&7/Sr86 ratio of the salt-free soil. (3) The Gr87/Gr86 ratio of both the water-soluble salt and the 2 9 6

Figure 65. Meserve Glacier, Wright Talley. View is to the south. One of two cinder cones is visible in the southernmost region of the accumulation basin. 297

WVM-i

M 1 0 J WVM-8 \ \ | WVM-I9

WVM-6#'

8WVM-24

M “i

Figure 66. Location of soil samples from moraines of the Meserve Glacier, Wright Valley. TABLE 41. Sr /Sr ratios cf vr* ter-soluble salts and total salt-free soil, and the lithologic composition of moraine samples, Keserve Glacier, Wright Valley

b Sample Sr87/Sr86a Sr87/Sr86a Per cent Composition (cm depth) salt salt-free soil basalt dolerite basement ft'*—' 1 0.7121 0.7127 0.0 100.0 0.0 Ci ^r 1

WVM-2 0.7146 0.7152 0.0 6.2 93.8 (5.5-27.5) 0.7137 0.7137 0.0 5.2 94.3 (3-16) WVK-8 0.7130 0.7138 5.2 7.7 87.1 (2-12.5) ww:-9 0.7145 - 2.2 2.C 95.8 (2-12)

WVK-19 0.7091 0.7046 89 10 1 (15-2C) TABLE 41* Continued

Sample Sr87/Sr86a Sr87/Sr86a Per cent Composition^ (cm depth) salt salt-free soil basalt dolerite basement

WVM-22 rfc e r> (8-28) 0.7080 - 77" 20 3“ (106-109) 0.7066 0.7047 d d a WVM-23 (2-9) 0.7142 - 9.9 13.8 76.3 (9-19) 0.7142 - 1 5 . 0 6.1 78.9 WVK-24 (9-36) C.7157 - 0.0 12.7 87.3 KAB 0.7099 0.7107 34.3 65.7 0.0 aCorrected for fractionation assuming Sr8*/^6® = 0.1194

^Pebbles greater than 1/4 inch diameter c Pebble crunt approximate ^Pebble count not available at present time; sarnie highly indurated, but appears to consist of basalt with so./.e dolerite 6t>C 300 parent material reflect the lithic composition of the till (i.e., soils with a high basalt content have somewhat lower SrG7/Sr86 ratios than soils of low basaltic content). because the Sr87/Sr86 ratios of the water-soluble salts are similar to the ratios of the parent material, it is reasonable to conclude that these salts have formed ija situ. If the salts were derived from a source that is not local, or if they are relict salts, then the strontium would be more isotoplcally homogeneous. Since the salts appear to have formed during and since the moraines of the Meserve Glacier were deposited, then it follows that chemical weathering has occurred during and since deposition, i.e., in the case of the inner two moraines, sometime following eruption of the Mcl’urdo basalts. Although the age of the basalts in the vicinity of the Meserve Glacier is not known at present, basalts from near the neighboring Bartley Glacier have K-Ar ages of 3*7 m.y. (Denton and Armstrong, 1968). All K-Ar dates presently available for these basalts In './right and Taylor Valleys are within the Interval 2 to 4 mil lion years, making it reasonable to suppose that the basalt at the Meserve Glacier is of similar a^e. The salts in these moraines, therefore, have formed in relatively recent times. 301

Farther evidence of relatively recent chemical weathering can he obtained from the Sr87/Sr86 ratios of salt encrustations of glacially-dropped boulders. These surficial salt deposits are fragile and probably could not have survived glacial transport. Therefore, these salts were undoubtedly deposited on the boulders after they were dropped by the ice. Four Sr8?/Sr86 ratio analyses of these salt encrustations are given in table 42. Samples M-10 and M-ll are boulders of Ferrar dolerite and schist or gneiss, respectively. These boulders are located near the entrance to the accumu­ lation basin of the Meserve Glacier (figure 66) and occurred about 30 c® from each other (K. R. Everett, 1968, personal communication). The Sr87/Sr86 ratio of the salt from the dolerite boulder (M-10; 0,7136) is identical to that of dolerite collected nearby (0,7130* The Sr87/Sr®6 ratio of the salt from the gneissic boulder (M-ll; 0.7160) is similar to the values obtained for many samples of the basement complex (table 28), Because of the proximity of the boulders to each other, the 3alts probably have not been brought from an outside source, but have formed jji situ since the boulders were dropped by the glacier. No relative age can presently bo given for the time these boulders were deposited. The Sr87/Sr8* ratios for salts from the surfaces 302

TABLE 42* Sr87/Sr06 ratios of salt encrustations from boulders in the vicinity of the Meserve Glacier, Wright Valley.

Sample No, gr8 7/gj,8 6* cr Rock Type

M-10 0,7136 0.0003 Ferrar dolerite M-ll 0.7160 0.0003 schist or gneiss Boulder A 0.7133 0.0004 Ferrar dolerite

ST-1 0.7195 0.0004 Granite-gneiss

*Corrected for fractionation assuming Sr06/Sr08 = 0,1194 303 of Boulder A and ST-1 (table 42) further suggest the possibility of local formation of salts. Both of these boulders were located under the Meserve Glacier about 54 m from the edge of the ice cliff. They were located about one meter from each other in a cavity formed by bedrock and till. Boulder A was composed of dolerite, which is reflected by the Sr8?/Sr86 ratio of 0.7138 for the salt encrustation. ST-1 is a salt sample from a boulder of granite-gneiss. The Sr8?/Sr86 ratio of this salt (0.7195) is similar to the isotopic composition of strontium measured for the Olympus granite-gneiss and other basement rocks (table 28).

D. Conclusions

On the basis of the Sr87/Sr66 ratio measurements of salts in the moraines and on boulders in the vicinity of the Meserve Glacier, it is concluded that most of these salts have formed situ. By extrapolation of age determinations of nearby basalts similar to those in the region of the Meserve Glacier, it is further concludeu that these salts have formed within the last 2-4 million years and are possibly forming at the present time. APPENDIX A ANALYTICAL PROCEDURES

A* Introduction

All chemical and isotopic data presented in this study were determined by the writer except for values of other investigators where noted. The analytical procedures followed in this work are given below. They consist of essentially two types: (1) those necessary for the preparation of samples for mass spectrometric analysis, and (2) chemical procedures used in the analysis of water and rock samples. Several satisfactory schemes for the chemical analysis of geologic materials are available (e.g., Bennett and Hawley, 1965? Hlllebrand, £& 1953; Kolthoff and Sandell, 1952$ Langmyhr and Graff, 1965; Peck, 1964; Shapiro and Brannock, 1962$ and Vincent, I960). Some methods have been modified for special needs that arose during the course of this investigation. All reagents used in the analyses were of A.C.S. reagent grade where possible. All hydrochloric acid was distilled as the constant-boiling mixture In a Vycor 304 305 apparatus. Nitric acid used in the mass spectrometric analyses was also distilled In Vycor glass. All water was double distilled and further purified by deminerali­ zation with a mixed-bed ion exchange column.

B. Preparation of Hock Samples

Fresh, unweathered rock was used in all analyses. To obtain as representative a sample as possible, a minimum of 50 g was taken if sufficient material was available. Rock samples were ground in a steel mortar and pestle to -60 mesh. After grinding, the sample was rolled at least 25 times using a large sheet of coated paper.

C. Preparation of Water Samples

For the majority of the water samples, there was no treatment prior to the analyses. In a few samples, a fine sediment had settled out, and in these cases, the water was carefully decanted. Where the sediment could not be removed by decantation, or if the samples were charged with an algal bloom, the water was filtered through a Millipore apparatus using a prefilter cut from S+S 589 Black Ribbon filter paper, backed by two circles of S+S 589 Blue Ribbon filter paper. This filtration 3 0 6 appeared to be sufficient to prevent the return of the algal bloom*

D. Preparation of Samples for Hass Spectrometrie Analysis

1. Introduction

Strontium used In the lsotoplc determinations had to be separated from samples prior to measurement. Two types of measurements were made in the course of this study: (1) strontium lsotoplc ratio (SrIR) analyses, which yield the Sr89/Sr86 ratios, and (2) strontium isotopic dilution measurements, which yield the concen­ tration of strontium in the sample. Since isotopically- splked solutions were used in the Isotope dilution measurements, contamination in both types of analyses was kept to a minimum by using equipment exclusively allocated to each. Strontium was separated using ion exchange columns charged with Dowex 50W-X8 (200-4-00 mesh) cation exchange resin. The sample was placed on the column and eluted with hydrochloric acid (sp. gr. = 1.035). Position of the strontium band on the column was monitored by means of a Geiger counter which detected carrier-free Sr89

tracer that was added to the sample just before It was 30? placed on the column* As the strontium was eluted, 15 ml fractions were collected in polyethylene beakers. One or more beakers containing maximum activity were evaporated nearly to dryness in Vycor dishes, transferred

to 5 ml Pyrex or Vycor beakers, and taken to dryness. A drop of concentrated perchloric acid was added to destroy organic material. After the perchloric acid was evap­ orated, the beaker was stored until time for measurement. When the sample was ready for measurement, It was taken up in a fraction of a drop of 1 M nitric acid using a glass syringe. The strontium was placed on the tantalum filament of the mass spectrometer by warming the filament and slowly feeding the solution containing the strontium onto the wire ribbon.

2. Preparation of samples for Sr87/Sr66 ratio measurement a. Rock samples A quantity of sample containing at least 100 pg strontium was dissolved in a covered Teflon dish using an acid mixture consisting of 15 ml concentrated hydro­ fluoric acid, 3 ml 9 U sulfuric acid, and 3 ml 1 M nitric acid. After the sample was dissolved, the solution was heated to fumes of SO 3 and cooled. The resulting slurry was dissolved in water containing a few milliliters of dilute hydrochloric acid, followed 3 08 by warming if necessary. The solution was filtered through a fine filter paper (e.g., S+S 589 Blue Hibbon or S+S 576). A few drops of the Sr®9 tracer were added, mixed well, and the sample was placed on the column, b. Water samples A sufficient volume of water was taken that contained at least 50-100 pg strontium. If the volume was about 30 ml or less, it was acidified with about 5 ml dilute hydrochloric acid. If necessary, it was filtered through a fine paper. The Sr®9 tracer was added and the sample was placed on the ion exchange column, c. Salt samples If the material was mainly water-soluble, the sample (approximately 1 g in most instances) was dissolved in water, filtered through fine paper, and acidified with a few milliliters of dilute hydrochloric acid. If the final volume was greater than about 30 ml, it was evaporated to about 10 ml in a Teflon dish. Tracer was added and the solution was placed on the ion exchange column. Where the salt(s) contained only slightly water- soluble species (e,g., calclte), the water was acidified with hydrochloric a d d to pH 3-4. If time was available, the material was suspended in water and solution was 309 allowed to take place by the adsorption of C02 from the atmosphere* This procedure was preferred as It simulated the conditions found In the Ice-free valleys, i.e., melt- water flowing over soil containing salt encrustations. After the salt was dissolved, the solution was filtered, acidified, charged with tracer, and added to the ion exchange column. d. Soil samples For many of the soil samples, two measurements of the lsotoplc composition of strontium were desirable. The first measurement determined the Sr8?/Sr86 ratio of water-soluble salt within the soil, and the second determined the Sr8?/Sr8* ratio of the total, salt-free soil. In some cases, soil samples weighed several kilograms. These samples were quartered until a representative sample of approximately 100-400 g was obtained, the actual amount depending on the salt content of the soil* The samples was placed in a large (£00 ml) Teflon dish and 100-250 ml water was added with stirring. (In the initial phases of this study, dilute hydrochloric acid was used to dissolve the salt. However, this treatment leached Iron, and possibly strontium, from

the soli as well as the salt, and therefore, this practice was discontinued.) If salts were water-soluble, 3 1 0 the sample was allowed to stand for 7-10 days. The solution was filtered through a fine paper, evaporated in a Teflon dish, and, If some Insoluble material had formed upon evaporation, the solution was reflltered. The filtrate was acidified with hydrochloric acid, tracer added, and the sample placed on the resin column. The salt-free soil was rinsed with water two or more times by decantation. It was then stirred with about 200 ml water and continually adjusted to pH 2-3 with hydrochloric acid. The solution was decanted, rinsed with water two additional times, and air dried. The total soil sample was ground to -60 mesh. This material was rolled and quartered and finally a representative sample of approximately 0 .5-1.0 g was obtained. This was dissolved and treated as described under the preparation of rock samples.

3. Preparation of samples for isotopic dilution analysis of strontium

Strontium concentrations in water and rock samples were determined in a sample by measuring the change in its isotopic composition following addition of a known amount of strontium enriched in one of 3 isotopes. Solutions containing such isotopically- 311 enriched species are called "spike” solutions.

Theoretical treatments of the lsotoplc method of analysis has been presented in detail elsewhere (Crouch and Webster, 1963* Hamilton, 1965? and Web3ter, I960). "Normal" strontium has the following composition: Isotope % Abundance Sr®4 0.560 Sr86 9.861

Sr87 6.985 Sr88 82.593 Strontium of this isotopic composition has an atomic weight of 87.6159. To a weighed sample containing "normal" (natural) strontium, a known quantity of a "spike" solution enriched in Sr88 was added. "Spike" strontium used in this study had the following lsotoplc composition! Isotope % Abundance Sr84 0.014 Sr88 97.644 Sr87 0.654

Sr88 1.685 The atomic weight of this strontium is 85.949. The quantitative estimation of strontium by the lsotoplc dilution method is as follows: Let:

N = total number of strontium atoms present in a known amount of sample S = total number of strontium atoms present In a known amount of "spike" solution

Ajj86, Ajj88 = abundances of Sr86 and Sr88, respectively, in the sample

86 03 86 R a Ag , Ag = abundances of Sr and Sr , respectively, in the "spike" R = measured Sr88/Sr88 ratio of the mixture of strontium of the sample and "spike" The measured Sr86/Sr88 ratio "R" can then be expressed as

(AH“ x N) + (As86 X S)

R ~ (Ajj88 x N) + (Ag88 x S)

After the value "R" has been determined, the above equation is solved algebraically for the (N/S) atomic ratio. This atomic ratio is transformed into a weight ratio by multiplication with a weight factor, the ratio of the atomic weight of strontium In the sample to the atomic weight of the "spike" strontium. The weight ratio is then solved for "N", the weight of strontium in the sample, by substituting the appropriate value for "S", the weight of "spike" strontium added. In order to determine the amount of "spike" strontium that should be added to obtain the optimum results, the strontium concentration must first be estimated in the sample. This is usually done for rock samples by an x-ray fluorescence method, such as that 313 described by Champion and others (1966) or Solter (1966). For water samples, the method of assuming that the stron­ tium content was roughly 1% of the calcium concentration measured by titration was found to be satisfactory in the majority of samples. From a plot of (N/S) versus (R), the optimum quantity of "spike" has been calculated to be about 1.88. In the early stages of this study, "spike" solution was added with a pipet. This often required the pipetting of very small volumes and the error was relatively large, especially when the concentration of "spike" solution was high (on the order of 50 pg Sr/ml). Later, "spike" was prepared containing about 5 ME Sr/g solution, and volumetric errors were eliminated by the weighing of the solution. The "spike" strontium solution was calibrated in a similar manner as the determination of strontium In a sample. A known quantity of the "spike" solution was mixed with a weighed amount of "shelf" strontium. This "shelf" solution contained strontium of "normal" isotopic composition and its concentration was accurately known by careful preparation of the solution from a dried sample of pure strontium nitrate. The lsotoplc compo­ sition of the mixture of "spike" and "shelf" solutions was determined, and the concentration of the "spike" 314 solution was calculated using the above equation. Once the appropriate amount of "spike" strontium has been estimated for a given amount of sample, the sample was accurately weighed, and an accurately weighed amount of "spike" strontium was added. If the total volume was greater than 30 ^ for the water samples, it was evaporated in a Teflon dish before the tracer was added and the mixture placed on the ion exchange column. If the volume of "spike" solution exceeded 10 ml in the case of the rock samples, it was evaporated prior to the addition of the Hf-H2S04-HN03 mixture. The sample was then treated In a similar manner as described above for the measurement of the Sr87/Sr86 ratio.

4. Mass spectrometric measurements

All lsotoplc analyses were made on a six-inch 60° sector, single filament solid source instrument (Nuclide Corporation Model 6-60-S). The ion current was amplified by a vibrating reed electrometer (Cary Model 31), and recorded on a ten-inch strip chart .6 recorder. Operating pressures varied from 4 x 10 to 2 x 10" 7 torr (1 torr = 1 mm Hg) for the Sr87/Sr88 ratio measurements. Runs were continued at the higher pressures only if resolution of the peaks was completely satis­ factory. Resolution was continually monitored by observing whether the baseline between the 87 and 88 mass peaks was 315 "flat" and on "zero." Throughout each run, the baseline was checked continually and adjusted to "zero" between the 86 and 87 mass peaks. Since the 88 mass peak must be recorded on a less sensitive scale than the other peaks In the Sr87/Sr86 ratio measurements, a baseline cor­ rection was made about every 12 scans In the earlier runs, but later It was read during every other scan to minimize error Introduced by short term Instrumental fluctuations. Strontium lsotoplc dilution runs were made at a pressure less than 4 x 10* 7 torr. The mass range was continually scanned during the course of a run, and scans were averaged In sets of six. Where possible a minimum of ten sets were obtained for the Sr87/Sr88 ratio measurements. Precision of the sets for the lsotoplc dilution runs dictated the number of sets obtained for these analyses. Tantalum ribbon <0.020" x 0.001") was used as filament wire. When a new filament was required, it was precleaned in the mass spectrometer at a filament current of 2.0 amps. If no peak signals were detected In the 84-88 mass range, the filament was presumed to be clean with respect to strontium and rubidium. The isolated strontium was dissolved in a fraction of a drop of 1 £ HNOj and drawn into a syringe with a Vycor glass tip. As the filament was warmed slightly, the sample 316 was gradually fed onto the wire, Reproducibility of the Sr87/Sr88 ratio analyses was checked periodically by analyzing an interlaboratory strontium isotope standard (SrC0 3 , Elmer and Amend, lot 492327)i results of 18 replicate analyses are given in table 43* Since there is fractionation of strontium isotopes in the course of a mass spectrometric deter­ mination, values for the Sr87/Sr86 ratio were normalized using a Sr86/Sr88 ratio of 0.1194 (Faure and Hurley, 1963), since this ratio is considered to be invariant in nature. The average value for the Elmer and Amend strontium carbonate standard obtained in this laboratory agree well with results obtained by other investigators for the same standard. These values also attest both to the accuracy and reproducibility of the measurements for the standard and all other measurements made on the mass spectrometer.

E. Determination of the Density of Saline Waters

Densities of the lake and pond samples were deter­ mined by measuring the weight of a known volume of brine. Glass-3toppered volumetric flasks (25 ml) were calibrated with water and the volume corrected to 20°C. The flasks were rinsed with water and acetone and air dried before use each time. Although the flasks could have been rinsed with the saline solution, wastage of the sample was 317

TABLE 4 3 . Analyses of the lsotoplc composition of stron­ tium of the Elmer and Amend SrC03 standard (lot no. 492327)

n 8 6 /n 8 0 Date Run Sr87/Sr86# Sr /Sr Analyst

1 0.7084 0.1177 Faure % f / & 4 0.7080 0.1185 Chaudhurl 16 0.7084 0.1176 Chaudhurl 0.7089 0.1177 Chaudhurl 11/8/65 36 0.7074 0.1177 Chaudhurl 12/7/65 55 0.7083 0.1182 Faure 12/28/65 66 0.7084 0.1174 Kovach 5/25/66 192 0.7082 0.1180 Chaudhurl 6/I3/66 206 0.7084 0.1189 Montigny 8A2/66 261 0.7083 0.1173 Montigny and Jones

373 0.7082 0.1179 Kontigny 5^ 0/67 449 O.7090 0.1184 Fenton 5/25767 469 0.7087 0.1182 Jones 6/19/67 0.7085 0.1181 Jones 12/22/67 m 0.7081 0.1182 Jones 7/29/68 1298 0.7070 O.II83 Eastin 8/ 30/68 1385 0.7083 0.1193 Jones 3/20/69 1590 0.7081 0.1188 Eastin

Average: 0.7083 0.1181 * 0.0004 (2#-) *0.0005 (2*9

#Corrected for fractionation assuming Sr86/Sr88 = 0.1194. 318 kept to a minimum by using a clean, dry flask for each determination* After the calibrated volumetric flask was filled to the mark with brine and weighed to * 0*001 g, all temperature corrections were made and the densities normalized to 20°C, using the coefficient of expansion of pure water.

F. Determination of Calcium in Water Samples

An aliquot of the sample was weighed or pipetted, buffered to pH 12, and the calcium was titrated complexo- metrlcally with EDTA using Hydroxy Naphthol Blue A.R. as a visual indicator. About 25 ml were usually taken for the titration, or a sufficient amount so that at least 5 ml of 0.005 M EDTA were necessary for the titration. When titrant volumes were expected to be low, a micro- buret was used for the titrations. In samples where the indicator precipitated, interfering ions were complexed with TEA (triethanol amine) or potassium cyanide. This method was checked by analyzing the U.S.G.S. standard rocks G-l and W-l; the results are listed in

table 44. The results are satisfactory, especially for W-l; results for G-l are slightly lower than the recommended values. This could be due to the low 319

TABLE 44. Analysis of U.S.G.S. standard rocks G-l and W-l for calcium, magnesium, sodium, and potassium

Sample CaO MgO Na20 K a0

G-l (A) 1.29 0.34 3.33 5.31 5.55 G-l (B) 1.22 0.40 3.39 5.48

G-l (C) 1.31 0.39 3.38 5.58 3.35 Recommended valuest

Ingamells and Suhr (1963) 1.36 0.35 3.29 5.52 Fleischer (196?) 1.39 0.41 3.32 5.45

W-l (A) - 6.46 2.20 0.65 2.21 0.67 W-l (B) 10.91 6.56 2.13 0.61 2.17 0.63 W-l (C) 10.93 6.64 2.22 0.63 10.99 2.22 0.65

Recommended values:

Ingamells and Suhr (1963) 10.92 6 . ?4 2.15 0.63 Fleischer (1965) 10.96 6.62 2.07 0.64 320 concentration of calcium and the small quantity of sample taken for analysis. In any case, the method apparently is suitable for the analysis of water samples.

G. Determination of Magnesium in Water Samples

The magnesium content of water samples Is usually determined indirectly with EDTA. First, total calcium and magnesium content is determined by titration with EDTA at pH 10 using Eriochrome Black T indicator. The calcium concentration is then determined on a second aliquot of the sample by titration with EDTA at pH 12. The difference between the two titrations is a measure of the magnesium concentration in the sample. In some water samples analyzed from Lake Huron, particularly those in which the magnesium content was low compared to the calcium concentration, negative amounts of magnesium have been obtained (L. M. Jones, unpublished data). Therefore, magnesium was determined by EDTA titration only when the Mg/Ca ratio was greater than about 0.5* This applied mainly to the samples from Lake Bonney. When magnesium could not be determined with an EDTA titration, a spectrophotometrlc method was used. This method was adapted from the Titan Yellow method Meyrowitz (1964) developed for the microanalysis of

magnesium in silicate minerals. This method offers four 321 chief advantages over the EDTA titration for water samples: (1) the method Is direct; (2) small quantities of samples are required; (3) it is relatively free of interferences; and (4) low levels of magnesium may be determined. Instead of the sample being a solution of dissolved silicate material, an aliquot of the water sample was taken for analysis. Otherwise, the procedure used is essentially identical to that developed by Meyrowitz (1964). The procedure was checked by analyzing the U.S.G.S. standard rocks G-l and W-l; the results were found to be satisfactory (table 44).

H. Determination of Sodium in Water Samples

Sodium in the water samples was determined using flame photometry. This technique provides a rapid, sensitive, quantitative method that is essentially free of spectral interferences of other elements normally encountered. In this study a Beckman DU-2 spectrophotometer equipped with a flame photometric attachment was used. The hydrogen-oxygen fuel system was operated at 4 pounds hydrogen pressure and 10 pounds oxygen pressure. The unresolved yellow doublet of sodium (589.0/0.6 mp) was selected for all measurements of this element. 322 The effect of various cations and anions on the sodium emission vas studied. No major interferences were anticipated (Dean, i960), and this was verified by experiment (L. M. Jones, unpublished studies). The only preliminary treatment given to the water samples was filtration or decantatlon shortly prior to measurement. This filtration was done to prevent the blocking of the capillary by suspended matter. Samples were then read directly unless the concentration of sodium was too high for the concentration range of the standards. In this case, an aliquot was taken and the sample diluted to an appropriate volume with water. An important interference was found to be self­ absorption. At low concentrations (less than 10 ppm) the calibration curve is linear, but above this sodium content the self-absorption factor increases. If the sample contained more than 2$ ppm an aliquot was diluted. The procedure was checked using the U.S.G.S. standard rocks G-l and W-l; the results are in table 44. Compared to the recommended values of Na2 0 of other analysts, the results obtained in this study are slightly higher in both samples by about 0.05£* This is probably due to mutual-cation-enhancement (Dean, I960). A common method of eliminating this problem is the use of an Internal standard. However, the Beckman DU-2 is not 323 designed for this use, and, in the case of sodium, analyses agreed satisfactorily with the recommended values,

I, Determination of Potassium In Water Samples

Potassium was determined In the water samples by flame photometry. As in the case of sodium, this procedure provides a rapid, sensitive, and quantitative method for analysis of this element. However, the analysis for potassium is not as interference-free as that for sodium* Because the Beckman DU-2 spectropho­ tometer is not designed for the use of an Internal standard, the analysis of potassium is more difficult. The resonance doublet of potassium at 766 and 769 mp was used in all analyses. This is the strongest of the potassium lines. The next strongest line (the unresolved doublet at 404.4 mp) was briefly investigated, but was found to be more susceptible to interference by low level concentrations of other ions than the red doublet. Emission of the red doublet of potassium at the 2 ppm concentration level was found to be unaffected by the presence of ?0 ppm calcium, 10 ppm magnesium, 1 ppm iron, and 1 ppm aluminum. The presence of sodium, however, even at very low concentration levels, caused a marked enhancement of the potassium emission. 324 From a study of the effect of sodium, it was found that at sodium concentrations less than 2000 ppm, the enhancement of the potassium emission changed greatly with a change in the sodium concentration. However, beyond this concentration there is no further enhancement of the potassium emission. In all potassium analyses of this study, sample solutions were made to contain more than 2000 ppm sodium by the addition of solid reagent grade sodium chloride. This eliminated the effect of variations of the sodium content of individual samples. Standard potassium solutions were also treated in a like manner. The procedure was checked by analyzing the U.S.G.S. standard rocks G-l and W-l; the results are given in table 44. It is seen that the procedure yielded results in good agreement with the analyses obtained by other analysts. BIBLIOGRAPHY

Adie, R» J., 1962, The geology of Antarctica, la H. Wexler, M. J. Rubin, and J* E* Caskey, Jr., Editors, Antarctic Research, Am. Geophys. Union, Geophys. Monograph No* 7, Publ. No. IO3 6 , p. 26-39. Aldrich, L. T., Wether111, G. W., Tilton, G. R., and Davis, G. L., 195&, Half-life of Rb87: Phys. Rev., v. 103, p. 1045-1047* Angino, E* S., and Arraltage, K. B., 1963, A geochemical study of Lakes Bonney and Vanda, Victoria Land, Antarctica: Jour. Geol., v. 71, p* 89-95 * -- — , Armitage, K. B., and Tash, J. C., 1962a, Air temperatures from Taylor Glacier Dry Valley, Victoria Land, 1961: Polar Record, v. 11, p. 283-284. — , Armitage, K. B., and Tash. J. C., 1964a, Ionic content of Antarctic ice samples: Polar Record, v. 12, p. 407-409.

Armitage, K. B., and Tash, J. C., 1965, A chemical and limnologlcal study of Lake Vanda, Victoria Land, Antarctica: Univ. Kansas Sci. Bull., v. 45, P* 1097-1118. ---- — , Armstrong, K. B . , and Tash, J. C., 1962, Chemical stratification in Lake Fryxell, victoria Land, Antarctica: Science, v* 1 3 8 , p* 34-36. ------, Armstrong, K. B., and Tash, J. C., 1964, Physicochemical limnology of Lake Bonney, Antarctica: Limnol. Oceanogr. v. 9, p* 207-217* j and Turner. M. D., 1964, Antarctic orogenic belts as delineated by absolute age dates, in, R. J. Adie, Editor, Antarctic Geology: First Internet. Symp. Antarctic Geology Proc., Capetown, 1963, New York, John Wiley and Sons, Inc., p. 551-556.

325 326 Angino, E. E. , Turner, M. D., and Zeller, E. J., 1962b, Reconnaissance geology of lower Taylor valley, Victoria Land, Antarctica! Geol. Soc. America Bull., v. 73, P- 1553-1561. Armitage, K. B., and Angino, E. E . , 1967, Minor elements of Lakes Bonney and Vanda, Antarctica: manuscript. Armitage, K. B. , and House. H. B, , 1962, A llmnologlcal reconnaissance in the area of KcMurdo Sound, Antarctica: Llmnol. Oceanogr., v. 7, p. 36-41. Armstrong, R. L., Hamilton, W., and Benton, G. H., 1968, Glaciation in Taylor Valley, Antarctica older than 2.7 million years: Science, v. 159, p. 187-189. Barrett, P. J., Faure, G., and Lindsay, J. F., 1968, Environment of deposition and provenance of some Beacon rocks, Beardmore Glacier area. Antarctica, la Geochronol ogy of the Transantarctic Mountains, Report No. 3, The Ohio State University, Lab. for Isotope Geology and Geochemistry, p. 1-17. Bell, R. A. I., 1966, A seismic reconnaissance in the McMurdo Sound region, Antarctica: Jour. Glaciology, v. 6, p. 209-221. Bennett, H., and Hawley, W. G., 1965, Methods of silicate analysis, New York, Academic Press, 334 pp. Black, R. F., Jackson, M. L., and Berg, T. E., 1965, Saline discharge from Taylor Glacier, Victoria Land, Antarctica: Jour. Geology, v. 73, p. 175-181. Boswell, C. R., Brooks, R. R., and Wilson, A. T., 1967, Some trace elements in lakes of McMurdo Oasis, Antarctica: Geochlm. et Cosmochim. Acta, v. 31, P. 731-736. Boucot, A. J., Caster, K. E., Ives, D., and Talent, J. A., 1963, Relationships of a new Lower Devonian Terebratulold (Brachlopoda) from Antarctica: Bull. American Paleont., v. 4 6 , p. 81-151. Bull, C., i9 6 0 , Gravity observations in the Wright Valley area, Victoria Land, Antarctica: New Zealand Jour. Geol. and Geophys., v. 3 , p. 543-552. 327 Bull, C., 1962, Development of ice-free valleys in south Victoria Land, Antarctica (abstract): Geol Soc. America Spec. Paper No. 68, p. 309-310* , 1966, Climatological observations in ice-free areas of southern Victoria Land, Antarctica, in M. J. Rubin, Editor, Studies in Antarctic Meteorology, American Geophys. Union, Antarctic Research Series, v. 9, p. 177-194. , and Carnein, C. R., 1968, The mass balance of a cold glacier: Meserve Glacier, south Victoria Land, Antarctica: manuscript. ------— , MeKeIvey, B. C.? and Webb, P. N., 1962, Quaternary glaciations in southern Victoria Land, Antarctica: Jour. Glaciology, v. 4, p. 63-78. Calkin, P. E., 1964, Geomorphology and glacial geology of the Victoria Valley system, southern Victoria Land, Antarctica: The Ohio State University, Institute of Polar Studies Report No. 10, 66 pp. Campbell, I. B., and Claridge, G. G.C., 1967, Site and soil differences in the Brown Hills region of the Darwin Glacier, Antarctica: New Zealand Jour. Geol. and Geophys., v. 10, p. 567-577* Champion, K. P., Taylor, J. C., and Whittem. R. N.. 1966, napid x-ray fluorescence determination of traces of strontium In samples of biological and geo­ logical origin: Anal. Chem., v. 3 8 , p. 109-112* Claridge, G. G. C.. 1965* The clay mineralogy and chemistry of some soils from the Ross Dependency, Antarctica: New Zealand Jour. Geol. and Geophys., v. 8, p. 186-2 2 0 . ------« and Campbell, I. B., 1968, Soils of the Shackleton Glacier region. Queen Maud Range, Antarctica: New Zealand Jour. Geol. and Geophys., v. 11, p. 171-218. Cotton, C. A., 1967* Antarctic scablands: New Zealand Jour. Geol. and Geophys., v. 9, p* 130-132. Craig, H., 1966, Origin of the saline lakes in Victoria Land, Antarctica (abstract): Trans. American Geophys. Union, v. 47, P* 112-113* 328 4 Crouch, E. A* C., and Webster, R. K., 1963, Choice of the optimum quantity and constitution of the tracer used for lsotoplc dilution analysis! Jour. Cham. Soc., v. 1 8 , p. 118-131. Dasch, E. J.. 1969, Strontium isotopes in weathering profiles, deep-sea sediments, and sedimentary rocks^ Ph. D. dissert., Yale Univ., 94 pp. Dean, J. A., I960, Flame photometry, New York, McGraw- Hill Book Co., Inc., 354 pp. Debenham, F., 1921, The sandstone, etc., of the McMurdo Sound, Terra Nova Bay, and the Beardmore Glacier regions! Nat. Hist. Report Brit. Antarct. (Terra Nova) Exped. 1910, Geol. 1 (4a), p. 103-119. Denton, G. H., and Armstrong, R. L., 1968, Glacial geology and chronology of the McMurdo Sound region: Antarctic Jour, of the U. S., v. 3, P. 99-101. Deutsch, S., and Grbgler, N., 1966, Isotopic age of Olympus Granite-gneiss (Victoria Land - Antarc­ tica): Earth Planet. Scl. Letters, v. 1, p. 82-84. — -, and Webb, P. N.. 1964, Sr/Rb dating on basement rocks from Victoria Lana; Evidence for a 1000 million year old event, R. J. Adie, Editor, Antarctic Geology: First Internet. Symp. Antarctic Geology Proc., Capetown, 1 963, New York, John Wiley and Sons, Inc., p. 557-562. Faure, G., Crocket, J. H., and Hurley, P. M., 1967, Some aspects of the geochemistry of strontium and calcium in the Hudson Bay and the Great Lakes: Geochim. et Cosmochim. Acta, v. 3 1 , p. 451-461. ------, and Hurley, P. M., 1963* The isotopic composition of strontium in oceanic and continental basalts: Application to the origin of igneous rocks: Jour. Petrology, v. 4, p. 31-50. ------, Hurley, P. M.. and Powell, J. L., 1965, The isotopic composition of strontium in surface water from the North Atlantic Ocean: Geochim. et Cosmochim. Acta, v. 29, p. 209-220. 329 Fleischer, M., 1965, Summary of new data on rock samples G-l and W-l, 1962-196?: Geochim. et Cosmochim. Acta, v. 29, p. 1263-1283. Ford, A. B., 1964, Review of Antarctic geology: Trans. American Geophys. Union, v. 45, P* 363-381. Goldman, C. R., Mason, D. T., and Hobble, J. E., 1967, Two Antarctic desert lakes: Llmnol. Oceanogr., v. 12, p. 295-310. Grindley, G. W., McGregor, V. R., and Walcott, R. I., 1964, Outline of the geology of the Nimrod-Beardmore- Axel Heiberg Glaciers region, Ross Dependency, in R. J. Adie, Editor. Antarctic Geology: First Internet. Symp. Antarctic Geology Proc., Capetown, 1963, New York, John Wiley ana Sons, Inc., 206-219. Gunn, B. M., 1963, Geological structure and stratigraphic correlation in Antarctica: New Zealand Jour. Geol. and Geophys., v. 6, p. 423-443. ------— , and Warren, G., 1962, Geology of Victoria Land between the Mawson and Mulock Glaciers, Antarctica: New Zealand Dept, of Sci. and Industrial Res., New Zealand Geol. Survey Bull. No. 71, 157 PP. Hamilton, E. I., 1965, Applied chronology, New York, Academic Press, 267 pp. Hamilton, W., Frost, I. C., and Hayes, P. T., 1962, Saline features of a small ice platform in Taylor Valley, Antarctica: U. S. Geol. Survey Prof. Paper 45 O-B, p. B 73-B 75- — — — — , and Hayes, P. T., 1963, Type section of the Beacon Sandstone or Antarctica: U. S. Geol. Survey Prof. Paper 456-A, 18 pp. Harrington, H. J., Nomenclature of rock units in the Ross Sea region, Antarctica: Nature, v. 182, p. 290. -, 1965. Geology and morphology of Antarctica, in P. Van Oye and J. Van Mleghem, Editors, Biogeography and ecology in Antarctica, The Hague, Dr. W. Junk, Publ., from Monographiae Blologicae, v. 15, p. 1-71. 330 Hart, S. R., and Tilton, G. R., 1966, The isotope geochemistry of strontium and lead in lake Superior sediments and water, ia J. S. Steinhart and T. S. Smith, Editors, The earth beneath the continents, American Geophys. Union, Geophys. Monograph No. 1 0 , p. 127-137. Haskell, T. R., Kennett, J. P., Prebble, W.M., Smith, G., and Willis, I. A. G., 196?, The geology of the Middle and Lower Taylor Valley of South Victoria Land, Antarctica: Trans. Roy. Soc. New Zealand, y. 2 , p. 169-1 8 6 . Hedge, C. E., 1966, Variations in radiogenic strontium found in volcanic rocks: Jour. Geophys. Res., v. 71, p. 6119-6126. Henderson, R. A., Prebble, W. M., Hoare, R. A., Popplewell, K. B., House, D. A., and Wilson, A. T., i960, An ablation rate for Lake Fryxell, Victoria Land, Antarctica: Jour. Glaciology, v. 6 , p. 129-133. Hillebrand, W. P., Lundell, G. E. P., Bright, H. A., and Hoffman, J. J., 1953, Applied inorganic analysis, Second Edition, New York, John Wiley and Sons, Inc., IO34 pp. Hoare, R. A., 1966, Problems of heat transfer in Lake Vanda, a density stratified Antarctic lake: Nature, v. 210, p. 787-789. ■ , 1 9 6 8 , Thermohaline convection in Lake Vanda, Antarctica: Jour. Geophys. Res., v. 73* p. 607-6 1 2 . . Popplewell, K. B., House, D. A., Henderson, R. A., Prebble, W. M., and Wilson, A. T., 1964, Lake Bonney, Taylor Valley, Antarctica: a natural solar energy trap: Nature, v. 202, p. 886-888. Hubert, J. P., and Anglno, E. E., 1966, Petrography of bottom sands from Lakes Vanda and Bonney, Victoria Land, Antarctica: Trans. Kansas Acad. Sci., v. 69, p. 233-237. Ingamells, C. 0., and Suhr, N. H., 1963* Chemical and spectrochemlcal analysis of standard silicate samples: Geochim. et Cosmochim. Acta, v. 27, p. 897-910. 331 Jones, L. M. and Faure, G,, 1968, Age of the Vanda porphyry dikes in Wright Valley, southern Victoria Land, Antarctica: Garth Planet, Sci, Letters, v, 3, p, 321-324, Kelly, W, C, , and Zumberge, J. H. , 1961, Weathering of a quartz diorite at Marble Point, McMurdo Sound, Antarctica: Jour. Geology, v. 69, p* 433-446, Kolthoff, 1, M,, and Sandell, E, B,, 1952, Textbook of quantitative Inorganic analysis, Third Edition, New York, MacMillan Co,, 759 pp. Krumbeln, W. C,, and Grayblll, F, A., 1965, An intro­ duction to statistical models in geology, New York, McGraw-Hill Book Co*, 475 PP* Langmyhr. F, J,, and Graff, P* R,, 1965, A contribution to the analytical chemistry of silicate rocks: A scheme of analysis for eleven main constituents based on decomposition by hydrofluoric acid: Norges Geologlske Under sokelse, Publ. No, 230, Oslo, 128 pp, Livingstone, D, A*, 1963, Chemical composition of rivers and lakes. Chapter G, Data of Geochemistry, Sixth Edition, U. S, Geol, Survey Prof. Paper 440-0, 64 pp. Llano, G, A,, 1962, The terrestrial life of the Antarctic: Scientific American, v, 207, No. 3, p. 212-230. McCraw, J* D,, 1962, Volcanic detritus in Taylor Valley, Victoria Land, Antarctica: New Zealand Jour. Geol, and Geophys., v. 5, P* 740-745* — ---- 1967a. Soils of Taylor Valley, Victoria Land, Antarctica, with notes on soils from other localities in Victoria Land: New Zealand Jour. Geol. and Geophys,, v. 10, p. 498-539* — — , 1967b, Some surface features of McMurdo Sound region, Victoria Land, Antarctica: New Zealand Jour. Geol, and Geophys., v. 10, p* 391-417*

McDougall, I,, 1963, Potassium-argon age measurements on dolerltes from Antarctica and South Africa: Jour. Geophys, Res,, v. 68, p. 1535-1545* 332 McKelvey, B. C., and Webb, P. N., 1962, Geological investigations In southern Victoria Land, Antarctica; Part 3 - Geology of Wright Valley: Nev Zealand Jour. Geol. and Geophys., v. 5, p. 143-162. Mellor, M., 1961, The Antarctic ice sheet: Cold Regions Res. and Engineering Lab., Hanover, N. H. Report No. 1-81, 50 pp. Mercer, J. H., 1968, Glacial geology of the Reedy Glacier area, Antarctica: Geol. Soc. America Bull., v. 79, p. 471-486. --- — 1969, Antarctic ice and Sangamon sea level: internet. Assoc. Scientific Hydrology Publ. No. 79, p. 217-225. Meyer, G. H., Morrow, M. B., Wyss. 0., Berg, T. E., and Littlepage, J. L., 1962, Antarctica: the micro­ biology of an unfrozen pond: Science, v. 138, p. 1103-1104. Meyrovltz. Ft., 1 9 6 4 , The direct spectrophotometric microdetermination of high-level magnesium in silicate minerals. A Clayton Yellow procedure: American Mineralogist, v. 491 p. 769*777• Murthy, V. R., and Belser, E., 1968, Strontium isotopes in ocean water and marine sediments: Geochlm. et Cosmochlm. Acta, v. 3 2 , p. 1121-1126. Nichols, R. L«, 1962, Geology of Lake Vanda, Wright Valley, South Victoria Land, Antarctica, in H. Wexier, M. J. Rubin, and J. E. Caskey, Jr., Editors, Antarctic Research, American Geophys. Union, Geophys. Monograph No. 7, Publ. No. IO3 6 , P. 47-52. — t 1963, Geologic features demonstrating aridity of McMurdo Sound area, Antarctica: American Jour. Sci., v. 261, p. 20-31. ------— 1965, Antarctic interglacial features: Jour. Glaciology, v. 5* p. 433-A49. -- , 1966, Geomorphology of Antarctica, in J. C. F. Tedrow, Editor, Antarctic Soils and Soil Forming Processes. American Geophys. Union, Antarctic Research Series, v. 8, p. 1-46. 333 Peck, L. C., 1964, Systenatic analysis of silicates: 0. S. Geol* Survey Bull. 1170, 89pp. p£w£, T. L., I960, Multiple glaciation in the McMurdo Sound region, Antarctica - A progress report: Jour. Geology, v. 68, p. 498-514. ——-— 1966, Quaternary climatic variations in Antarctica as suggested by glacial fluctuations, in Symp. Pleistocent and Post-Pleistocent Climatic variations in the Pacific Area, Bishop Museum Press, p. 57-82. ------and Church, R. E., 1962, Glacier regimen in Antarctica as reflected by glacier-margin fluctuation in historic time with special reference to McMurdo Sound: 1. A. S. H. Comm, of Snow and Ice Publ. No. 58* P* 295-305* Picciotto, E., and Coppez, A., 1963, Bibliographic des mesures^d’ages absolus en Antarctique: Ann. Soc. Geologlque de Belgique, v. 85, p. 263-308. -, 1964, Bibliography of absolute age determi­ nations in Antarctica (addendum), Iq R. J. Adle, Editor, Antarctic Geology: First Internet. Symp. Antarctic Geology Proc., Capetown, 1963, New York, John Wiley and Sons, Inc., p. 563-569* Plumstead, E. P., 1 9 6 2 , Fossil floras of Antarctica. With an appendix on Antarctic fossil wood: Trans- Antarct. Exped., 1955-1958, Sci. Report No. 9, 154 pp. Powell, J. L.. Faure, G., and Hurley, P. M., 1965* Strontium-87 abundance in a suite of Hawaiian volcanic rocks of varying silica content: Jour. Geophys. Res., v. 70, p. 1509-1513* Ragotzkle, R. A., and Friedman, I., 1965, Low deuterium content of Lake Vanda, Antarctica: Science, v. 148, p. 1226-1227. — — — . and Likens, G. E., 1964, The heat balance of two Antarctic lakes: Limnol. Oceanogr., v. 9, p. 412-425. Roberts, C. D., 1965, Estimation of elapsed time since a certain climatic change for Lake Vanda: Nature, v. 207, p. 626-627* 334 Rudolph, E. D., 1963, Terrestrial vegetation of Antarctica: Past and present studies, la G* Tedrov, Editor, Antarctic Soils and Soil Forming Processes, American Geophys. Union, Antarctic Research Series, v. 8, p, 109-124. Rutford, R. H., Craddock, C., and Bastien, T. W., 1968, Late Tertiary glaciation and sea-level changes In Antarctica: Palaeogeography, Palaeoclimatology, and Palaeoecology, v. 5* p. 15-39* Shapiro, L., and Brannock, W. W., 1962, Rapid analysis of silicate, carbonate, and phosphate rocks: U. S. Geol. Survey Bull. 1144-A, 56 pp. Shlrtcliffe, T. G. L., 1964, Lake Bonney, Antarctica: cause of the elevated temperatures: Jour. Geophys. Res., v* 6 9 , p. 5257-5268. — and Benseman, R. F., 1964, A sun-heated Antarctic lake: Jour. Geophys. Res., v. 69, P. 3355-3359. -— - and Calhaem, 1. M*. 1968, Measurements of temperature and electrical conductivity in Lake Vanda, Victoria Land, Antarctica: New Zealand Jour. Geol. and Geophys., v. 11, p. 976-981. Smith, H. T. U., 1965, Anomalous erosional topography in Victoria Land, Antarctica: Science, v. 148, p. 941-942. Solter, D. D.. Jr., 1966, Distribution of rubidium, strontium, zirconium, and iron of Porphyry Mountain and age of the Silver Plume granite; M. Sc. thesis, The Ohio State University, 113 PP* Starik, I. E., Krylov, A. 7a., Ravich, M. G., and Sllln, Yu. 1., 1961, The absolute ages of East Antarctic rocks: Ann. New York Acad. Sci., v. 9 1 t P. 576-582. Tedrov, J* C. F., and Ugolinl, F. C., 1966, Antarctic soils, m J. C. F. Tedrov, Editor, Antarctic Soils and Soil Forming Processes, American Geophys. Union, Antarctic Research Series, v. 8, p. 161-177- >, Ugolinl, F. C., and Janetschek, H., 1963, An Antarctic saline lake: Nev Zealand Jour. Sci., v. 6, p. 150-156. 335 Thurber, D. L., 1962, Anomalous U234/U238 In natures Jour. Geophys. Res*, v. 6 7 , p. 4518-4520. ------, Anglno, E. E., and Armltage, K* B., 1968, The radioelement concentrations in Lake Bonney and Lake Vanda, Antarctica: manuscript. Toril, T.. and Ossaka, J*, 1965, Antarctlcites a nev mineral, calcium chloride hexahydrate, discovered in Antarctica: Science, v. 149, p. 975-977* Torli, T., Yamagata, N*, and Cho. T.« 196?, Report of the Japanese Summer Parties in dry valleys, Victoria Land, 1963-1 965* II- General description and water temperature data for the lakes: Antarctic Record (Japan), No. 28, p. 2225-2238. Treves, S. B., 1967, Volcanic rocks from the Ross Island, Marguerite Bay, and Mt. Weaver areas, Antarctica: J. A. R. E. Sci. Reports Spec. Issue No. 1, p. 136-149. Tureklan, K. K., and Wedepohl, K. H., 1961, Distribution of the elements in some major units of the earth’s crust: Geol. Soc. America Bull., v. 72, p. 172-192. Ugolinl, F. C., 1965, Soil investigations in the Lower Wright Valley: Natl. Acad. Sci. - Natl. Res. Council Publ. No. 1287, p* 55-61. — — — — , and Bull. C., 1965, Soil development and glacial events in Antarctica: Quaternarla (Rome), v. 7, p. 251-269. Vincent, E. A., I960, Analysis by gravimetric and volu­ metric methods, flame photometry, colorimetry, and related techniques, £& A. A. Smales and L. R. Wager, Editors, Methods in Geochemistry, New York, Interscience Publ. Inc., p. 33-80. Warren, C. R . , 1965, Wright Valley: conjectural volcanoes Science, v. 149, p. 658. Warren, G.. 1965* Geology of Antarctica, i& T. Hatherton, Editor, Antarctica, New York, Frederick A. Praeger P. 279-320. 336 Webb, P. N.. 1963a, Geological Investigations in southern Victoria Land, Antarctica* Part 4* Beacon Group of the wight Valley and Taylor Glacier region: New Zealand Jour. Geol. and Geophys., v. 6, p. 361-387. — -, 1963b, Isotope dating of Antarctic rocks: A summary - 1: New Zealand Jour. Geol. and Geophys., v. 5, p. 790-796. ----, and Warren, G., 1965, Isotope dating of Antarctic rocks: A summary - 2: New Zealand Jour. Geol. and Geophys., v. 8, p. 221-230. Webster, R. K., I960, Mass spectrometrlc Isotope dilution analysis, £n A. A. Smales and L. R. Wager, Editors, Methods iniSeochemistry, New York, Interscience Publ. Inc., p. 202-246. Weyant, W. S., 1966, The Antarctic climate, ifl M. J. Rubin, Editor, Studies in Antarctic Meteorology, American Geophys. Union, Antarctic Research Series, v. 9, p. 47-59. Wilson, A. T., 1964, Evidence from chemical diffusion of a climatic change in the McMurdo dry valleys 1,200 years ago: Nature, v. 201, p. 176-177* — -- — , and House, D. A., 1965, Chemical composition of South Polar snow: Jour. Geophys. Res., 70, p. 5?1?-5518. ------} and Wellman, H. W., 1962, Lake Vanda: and Antarctic lake: Nature, v. 196, p. 1171-1173* Woollard, G, P., 1962, Crustal structure in Antarctica, in H. Wexler, M. J. Rubin, and J. E. Caskey, Jr., Editors, Antarctic Research, American Geophys. Union, Monograph No. 7, Publ No. 1036, p. 53-73* Yamagata. N., Torii, T., and Murata, S., 1967, Report of the Japanese Summer Parties in ary valleys, Victoria Land, 1963-1965. V. Chemical composition of lake waters: Antarctic Record (Japan), No. 29, p. 2339-2361. Zeller, E. J., Angino. E. E. , and Turner, M. D. , 1961, Basal sedimentary section at Windy Gully, Taylor Glacier, Victoria Land, Antarctica: Geol. Soc. America Bull., v. 72, p. 781-786. I'MTEl) STATES DEPAWT.MK.vr <>E THE'. INTEKIDK (AEOUAT VI, SL'U V'KV

163" ••• 1*1 : > I*-

' ~ rJfiC : * £ % & - . ..

' WteHfl N*it»wh\ . citMtn iJjU:'; *V. . - A f ■' ~ '’ r / i ( ' 5T

A $ -G*

+1

A/ Lizardi Fool t m > / ■ ■ -‘^TrTATi A Tf' :* rt0‘ A frrwM^ 15' '/ooc- I Hafk4&A

.l*' U \1 A A

■ Pk

\ V'

< o W s|| 73

/DOo; kv m M j

\V»V Hotback HOI TV ii X . ^ ROSS ISLAND

165 166 Ml

147

4M MT 11*

44*

47* ij

II* 117 675 44*

;.io WJ y AdHwpjnguw roo

MeDwu^t t a e l JMMtf ptniuin tvQktry uo, | ' \ 450 \ 7M { RqrwiM Bm

117

1)1 M l

471 171 :W: 4*7 171

114

11*

w 411

i lilwl 14JJ.1 !iU!i! i i t j i I i i i i'i'i'i i i i i i A l ^Iryclusion Hill

! . « , ; l / ~_J>#chyte Hill M O U N t r**7“ A in

923

nj

977 * i 7 j m 914

M l

IM

141

*49 M l

110

2T* 114 C«p« Tenny*on W )\ V

4 37

nca aer f l i M l n,. lai.i1 it i**l • « ^

r. Hogback HjM B«rh«

f \x Mourtt m m . V Weyaiu

y;HARPOR

1# keC

lititfrf f * J f

rf if

Crrw***'* ■ • **t

* \ O ?> *V ■* \ v I, 'jg0f v - .>4J -*ff£ m \ 17} m \ «*

< 3 411

•» / 411 V V ■ 1*0 ; ; \ :— ^ - V r • 444 ^ / SOUND M C M u/ R D O \ f Mf Cab* Royds . 174 \ •01 494] Bwtagei 47S ) ■ ■■.> ’j- . ‘ 'mt -. / V I— .. / 744 - V 7 4H 741 :-'jh \ ✓ t ■*<%> r i X < \ \

ail: \ ✓♦v.y.;. ! vM-v 734 fSpttar Potnt : : : r y-i»I w \ \ . M9 XvX-X^

xu, : ' ..... uU tTh« Strand * \ '^Mora>naa

: X;;, x-:;:; X ' X;;i,X:u:x ::x-::XV:: ;,\V. x A * ''vy/yyy?yy

\ae t.

nCapa : xxx iCbPfiOMrtd.; A bbott»6 « Pk0r

1440

I6t0 MT TERRA NOVrf MOUNT jpKfefcUS 1400 f950

m nookery . ISM* \ >H Wllli»m*on W linar Rock V • $ AOIlie penguin rookeries 4301 . 'IK « \

penguM

'4- " 1341-t MOUNT TERROR Tti» Knoll M l Cape Croziar Bomb P o t t

C rf*nn%*» !. 3301

. 4IOX Mapped edited and published by the U S Geological Survey in cooperation with the National Science Foundation

Control by U S Geological hur-.ey and U t' Naval O cean­ ographic Oflice

Compiled m 1962 from U S Navy tnmetrogon aerial photographs taken 1956-1960

Hydrographic data by U S Naval Oreynographir Office This information is not intended lor navgaTonal purposes Names approved by U S Board on Geographic Names

Absence ot crevasse symbol does not rocessanly indicate a crevasse *ree area Contour Imes

(def i n»t£) (approjtimjt?) [conjectural)

Clift or escarpment

Steep slope

Small hillock intersected elevations

Photoftra mmetnc elevations

Ice th ic k n e ss SCALE 1 250,000 10 15 20 Stain!* Miles

10 \b 20 25 30 K'tomPttrs _J= 10 15 Nautical MNes E3 mbeH Confofmal Conic Proiection— Standard Parallels 76°40' and 79°20' CONTOUR INTERVAL 200 METERS— DATUM IS MEAN SEA LEVEL £ leva t ionr5. depths, and ice thickness m motors

LEGEND

icf area tee Nee area

Astro control station

Survey control stations A S rg n al

Traverse route with spot height

M oraine

Glacier

Crevasses CVruutta 0 0 Coastline under ice .320 .«* ice shelf

3T30- ,JIM± Icebergs

.S3 Frf5t or bay ice

OR SALE BY U S. GEOLOGICAL SURVEY. WASHINGTON, D. C. 20242 LOCATrON [.MARRAM iOL A I ‘s' (}“ + AS' *0

ruANf.Lih 'SLANL? j v s 1 W' 4* I wo*

ROV ISLAK \ ■ u r ,*

'i. . ..In'**],|o

NEW EAlAlD r.-if .] r v .j M i{ . « *•

*f s' 1 8 0 a F*'•T

l-MiS\ IS I.A .1) A. -'I '' I11' "I !r x . M) !:'(!' M" K I -1'1 K) ■ i ]t /<-['. 1 V* NlDk TAYLOR GLACIER

158' 1^9° \I 1 ^ Crtra*

'I \ I 1 i lr 13

1 I 1

I I 1 I I

1

1 1 1

1

I

1

I Ct)NVl#injN XAlf *-"-n f,-rF WiUc

ICiQOU 3000

Horseshoe Mtr>.

tot Nunatak

2140 W % V'.'

y fit

vs Win McKe'ljeV

rcules

i

,6 00

Finger Mountain

rkrrud Mtn v r vM *r V w P l q d a r a Sorre 0* KM* 1 • ”?lL^ V Joiqi-.^F f ■ ' Of f>TtiiJ Irrtf>11* Cai/Bmtlsh Rkks

Run tarn

GV^V , Mapped, edited, and published by the U S Geological Survey m cooperation with the National Science Foundation

Control by U 5 Geological Survey Compiled in 1962 from U S Navy tnmetrogon aerial photographs taken 1947-1959

Names approved by U S Board on Geographic Names

Absence of the crevasse symbol does not necessarily indicate a crevasse free area / 23207'

\

158“ 159“ 160°

SCALE 1 250 000 ,'m W(j!e v

3 ' . i k .

Lambert Conforma Conic Projection — Standard Parallels 76°40' and 79°20‘ CONTOUR INTERVAL 200 METERS — DATUM IS MEAN SEA LEVEL

L.EGFND kf Ice* tr*v ari‘d Ac,■ ru control station <> !-.Jh < . n .t»»: S'jrvey control s ta t- o n s 1041 ASignai

<*f;pr o *• m AI ej .III r id*e»M» routf *ith spot he ght

.Con.f { t ji a I j M o r a 1 ne

Oif I Oi1 prTiE? f\ G' a( ■ ?r

'itPeP Sk»PPr (It** asses CVmiMi Smau rniior.ils, 4 (. nast I1 nf undpf 1 • e I n ter sec tert ^ .320 .320 Ice fjh e ff

PriiilogrammeliiL elevations .2130 *: ■ 2130± !•' etwtflb o

FOR SALE BY U S GEOLOGICAL SURVEY. WASHINGTON. D. C. 20242 ■mia Mtn H r / ' ? S pire

N J VQfOt l.trt'rhdi »> /«//% M f 4 £ L«*vefw*Js* Ri>chb

Puntain

5V^W, ,

f nvn*itf .y {*■ ‘

Mt Feathe T abta

Monastery Nunatak

LO'-' A f -ON D.AGHAM n; f ' l :■: ■• N. rji > i(l ( ' - <0- -M •' ,; i a .!

f ^AN Hi , • Si AMi n frns iSl Af, ANTARCTICA

'■' N't . f Wr ■' ^ i ►

r-- j• < r^. ■ i.-i , nt'T ri,+ ‘ m .r,ri: M ,i [■ > ■' Aii1' : AUSTRALIA

A f

T . \ \ i ,(.H'' i i '' [ !■;IV A.VI'A K' "I If A ST -I ]}/.:■ S 1 /Oo t 1 ■■ \ X*j I'MTKI) STATES DEPARTMENT OK THE INTERIOR MOUNT DISCOVERY ■ \ . \ T A l< < 'T l'' \

\ ‘i.jIMHM) R | \ O VX\\ A IS1'A ' !'■ .A.KJI -

I f h On' Cap* Sp#«c*r-SroJth

Ml Hefoc 7A2

Mourn , Hayward

^M^Nipha^

Point

CQNWKS'ON St Ait

ikoi V

1 3000 .isickK.

<

< ? * o i joou

* 2 0 0 0

I 'GUU Alpha Bluff

162 ° 0 0 '

I > M f ■ ■ 'N "lA'-jWAM

IW ic 1 *. ■ -T! 'a" ■' \ - * \ \ :■ !

\

i... ; Photogra^^c < uTipi ation

, • O hi’, |- , •-• • , • ■ ■ • -i ' -j r ■ 1 -Till S [ 165" 166°

SCALE 1:250.000 ■ ?0 S(iTjr# M e* t*h 30 K.lomH*rS i .... JO l b Naiif'Ca F= -=l

Lambert Conformal Conic Projection— Standard Parallels 76°40r and 79 ° 20' CONTOUR INTERVAL 2 0 0 M ETERS— DATUM iS MEAN SEA LEVEL

LEGEND 'O- d'tfd I'It- tffv a FCd I I n*?s A^t'o c ont'<',i ttd'ion e E n ite i S u r v e y cont'oi s t a t i o n s TO-** ASignai .HI p p r o J.Ir*\ d tpj T 'd ^ /f s fi r o u t e a-!*- s p o t h e i g h t

on|

arpnrent Gi dCler

-lope Crevasses ['m vixnett If I lot k ',4 s CoaM 11 ne under ice c te d elevations .110 ,»0 Ice s h e l t

r a metric e:ev3Tuns 11 50 ' , JliOi icebergs o k ness aV fast or bay ice

FOR SALE BY U. S. GEOLOGICAL SURVEY, WASHINGTON. D. C 20242 ' A f ) N ( ' A' j k A N1 n; * * ■ a: : yn , !'i

A frica *0%-' a y tv b

rSOUTrl H. r«T“ N a n t a h c t i l a

.* t • . r- 1 • M . ,l

>: I D \T DIN' DA' Kk i N\,'D|-.X i,'1 'Tl- \\ s t .- (n/in

' ' / * ' /f; f ] ] A f ,