Introduction

The first bore hole through the Antarctic ice sheet was completed in January 1968, when engineers from the U.S. Army Cold Regions Research and Engineering Laboratory (CRRRL) recovered most of the 2 164 m long ice core to bedrock drilled at new “Byrd” station (lat. 80° S., long. 120° W.) (Reference Gow, Gow, Ueda and GarfieldGow and others, 1968). At this site the annual positive balance of snow is about 14 g cm⁻² and the mean annual air temperature is — 28° C (Reference Bull and QuamBull, 1971), so that very little, if any, melting occurs at the surface, and the annual balance is sufficiently large to allow the preservation of the stratigraphic record of the deposition of microparticles.

Using modifications of techniques first developed by Reference MarshallMarshall (1959,1962), an analysis has been made of the size distribution and concentration of microparticles in the size range 0.518–13.1 μm diameter, in 15 representative continuous sections, averaging a little over 1 m in length, from the “Byrd” ice core (Table I). The sections are quarter-core segments sawn from 10.8 cm diameter core. Using a Model “T” Coulter counter, the number of particles in each of 15 size ranges was measured in 500 μl samples of melt water from specimens of the core, each 2.5 cm, or less, in length. Altogether, in this study, two or more counter runs made on melt-water specimens from each of 821 samples from the ice core.

The “Byrd” Ice Core and Previous Work on it

Descriptions of the ice core have been given by Reference Gow and WilliamsonGow and Williamson (1971). In the upper 900 m of the core, crystal orientations are essentially random, but below 1 200 m the principal crystallographic axes (c-axes) of all ice crystals are orientated within 15° of the vertical. The crystal size is smaller in bands in the depth range from 1 200 to 1 800 m, but otherwise generally increases from top to bottom of the core. Below 1 800 m the c-axis orientations become more dispersed. Between 400 and 900 m depth the ice core is fragile and badly fractured, an observation of significance to our studies.

Table I

“Dirt” bands occur at intervals along the core, the majority being between 1 100 and 2 100 m. Reference Gow and WilliamsonGow and Williamson (1971) identified 25 “ash bands” (containing macroscopic particles) and about 2 000 “dust bands” (containing particles not visible to the unaided eye). Both kinds of layers have been attributed by Reference Gow and WilliamsonGow and Williamson (1971) to volcanic eruptions. The interval of maximum volcanic ash fall, between 1 100 and 2 100 m, corresponds to the interval where the δ¹⁸⁰ data (Reference Epsfein, Epstein, Sharp and GowEpstein and others, 1970; Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others, 1972) indicate maximum cooling of the Antarctic atmosphere. The concentration of major cationic constituents, Na+, K+, Ca++ and Mg++, is also largest between 1 193 and 2 100 m, probably due to increased local volcanic activity (Reference Ragone and FinelliRagone and Finelli, 1972).

Reference LambLamb (1970) noted that major historic volcanic eruptions are usually followed by periods of cooling in the lower troposphere, and Reference Rasool and SchneiderRasool and Schneider (1971), in model studies, demonstrated that increased atmospheric optical density (higher dust concentrations) would reduce surface temperatures. The variations in the ratio of oxygen isotopes ¹⁸O and ¹⁶O in the ice cores from Camp Century, Greenland, and “Byrd” station, Antarctica (Reference Epsfein, Epstein, Sharp and GowEpstein and others, 1970; Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others, 1972), together with some knowledge of the flow characteristics of ice at these locations, has allowed calculations of ages for the two ice sheets. Although the present study gives an alternative interpretation for the age of the west Antarctic ice sheet, the oxygen-isotope work shows that the most negative δ¹⁸O values represent snow that fell during the Wisconsin II glaciation, and that younger and older snow has less negative δ¹⁸O values.

Reference Hamilton and SeligaHamilton and Seliga (1972) consolidated these ideas, which point to a cause-and-effect relationship between atmospheric turbidity and temperature, and the present study is aimed at establishing the relationships more exactly, by comparing the particle concentrations in the “Byrd” core with the oxygen-isotope ratios.

Laboratory Techniques

The basic procedures for analyzing the microparticles in ice samples were established by Reference Marshall, Wexler, Rubin and CaskeyMarshall (1962), and greatly refined by Reference Bader, Bader, Hamilton and BrownBader and others (1965). Reference Taylor, Gliozzi and MellorTaylor and Gliozzi (1964) modified the techniques for use in the clean-room facilities at the Institute of Polar Studies Ohio State University, and further refinements have been described by Reference HamiltonHamilton (1967,1969). A full description of the techniques used in the present study has been given by Reference ThompsonThompson (1973).

After careful annotation of the core sections, including orientation, size and number of air bubbles and fractures, color variations and “dirt” bands, they were cut into samples, usually of 2.5 cm length, but reduced to 2.0 cm for the sections from below 1 900 m and to 1.0 cm for the 31 cm section from 180 m depth and the 76 cm section from 1 377 m.

In the clean room, samples were decontaminated by thoroughly rinsing with water, which had been demineralized in an ion-exchange resin column and sequentially filtered through “Millipore” filters of 0.45 and 0.22 μm pore size. In this process, at least 2 mm of ice was melted from the samples, which were then melted in cleaned, covered, plastic containers in a vibrating water bath. About 20 ml of water was obtained from each 2 or 2.5 cm thick sample.

Analytically pure NaCl dissolved in pure filtered water was added to the melt water to convert it to 2.27% NaCl solution. Only one-half of each quarter-core section was analyzed at a time, the second half being analyzed about 2 months later.

The microparticles in two 500 μl samples of the melt water were counted with a 15 channel Model “T” Coulter counter, set so that the average particle volume counted in adjacent channels changes by a factor of two. For channel 14, the lower threshold diameter was 0.518 μm and the mean diameter of particles counted was 0.581 μm; for channel o the upper threshold diameter measured was 13.1 μm.

Reference ThompsonThompson (1973) has described the methods of calibrating the counter and of ensuring that the calibration does not change. He also discussed sources of contamination and error. We are convinced that contamination in the laboratory is not a problem, so that the results for sound ice samples are meaningful. However, with ice samples from the fractured zone of the core, 400-900 m depth, our washing techniques are not adequate to assure removal of contaminating particles from the network of internal fractures that may have been introduced during drilling, transport or storage.

Comparison of Microparticle Concentrations and δ¹⁸O Values in Core Sections from 180 and 1 377 m Depth

The early experiments of Reference Marshall, Wexler, Rubin and CaskeyMarshall (1962) indicated a periodic variation in the microparticle concentration in ice samples from the “Byrd” station area of Antarctica. He suggested that the variations were annual and that they could be used in detecting annual layering At. the South Pole the spring and, to a lesser extent, the autumn snow precipitation is characterized by high concentrations of microparticles (Reference HamiltonHamilton, 1969), so that, at that site (where the annual positive balance is quite small, about 8 g cm⁻²) there are difficulties in determining the annual layering using only measurements of microparticle concentrations. At “Eights” station (lat. 75° 14’ S., long. 77° 10’ W.), Reference Taylor, Gliozzi and MellorTaylor and Gliozzi (1964) also detected a periodic variation in microparticle concentration which was later (Reference Bull and QuamBull, 1971) shown to be annual.

The most careful considerations of particle depositional mechanisms and patterns have been made by Reference HamiltonHamilton (1969). Earlier Reference JungeJunge (1963, p. 111–202) had shown that most natural aerosols in the troposphere range from 0.1 to 20 (μm and Reference Kumai and FrancisKumai and Francis (1962) demonstrated the importance of these aerosols in the formation of snow crystals. A relatively large microparticle is almost always present at the center of a snow crystal, with many smaller particles in the remainder of the crystal. Reference HamiltonHamilton (1969) proposed that variations in the microparticle depositional pattern can be caused by variations in the air temperature and saturation ratio because smaller particles can serve as nuclei at higher saturation ratios and lower temperatures.

Many investigators have used variations in the abundance ratio of the stable isotopes of oxygen and hydrogen in polar ice samples to determine variations in the air temperature at the time of crystallization, and hence to determine annual layering, for example, Reference Damgaard, Dansgaard, Johnsen, Moller and LangwayDansgaard and others (1969),Epstein and others (1970), and Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972).

The δ²H and δ¹⁸O variations in an ice core, however, do not necessarily provide a perfect record of near-surface air-temperature variations, because the temperature of the air mass in which the snow is formed may not be simply related to the near-surface air temperature where the snow is deposited. Furthermore, deflation and re-deposition, variations in the amount of summer and winter snowfall, and isotopie diffusion after deposition may all affect the isotopic record.

Fig. 1. Vertical profiles from the “Byrd” ice core: (a) Particle concentration for 1 cm samples; (b) δ¹⁸O values after Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972); (c) Particle concentration where the sample size has been mathematically increased to 2 cm for 15 500 year-old Antarctic ice. High particle concentrations tend to occur in ice with high negative δ¹⁸O.

In 700 year-old ice from Camp Century, Greenland, where the annual variations in the stable oxygen-isotope ratios are well marked, Reference Hamilton and LangwayHamilton and Langway (1967) obtained excellent correspondence between variations in the microparticle concentrations and those in the δ¹⁸O values, the snow in late winter or early spring containing the highest microparticle concentrations. At this site, it appears that microparticle and stable-isotope stratigraphies yield reliable estimates of ice chronology, at least in the upper part of the ice sheet where diffusion has not erased the annual δ¹⁸O variations.

According to the time scale of Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972), the ice at 1 377 m depth at “Byrd” station is approximately 15 500 years old. In Figure 1 are given the profiles of δ¹⁸O values, determined from very thinly sliced samples at 1 377 m (Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others, 1972) and of the microparticle concentrations (here defined as the number of particles larger than 0.65 μm diameter per 500 μl of melt water) in 1 cm slices of the same piece of core. In Figure 1 a, these concentrations are shown as measured on the 1 cm samples; in Figure 1c, they are smoothed, binomially, to give 2 cm running means for comparison with the δ¹⁸O variations of Figure 1b. The match is good but not perfect; part of the mismatching probably arises because some of the particle-concentration peaks are in ash bands representing local volcanic activity which obscure seasonal variations.

In the 31 cm long section from 180 m (800 years B.P.) the variations in δ¹⁸O (Fig. 2) are not sufficiently pronounced to allow the unambiguous identification of annual layering, but the microparticle-concentration profile does show periodic variations which may be related to the annual layering.

Fig. 2. Vertical profiles of particle concentration and δ¹⁸O values in 800 year old Antarctic ice. Note the small variations in particle concentration and in δ¹⁸O.

Variations in Microparticle Concentrations and Oxygen-Isotope Ratios over Millennial Time Intervals

Following the conclusions of Reference Hamilton and SeligaHamilton and Seliga (1972), on the relationships between atmospheric turbidity and temperature, representative samples from the entire “Byrd” core have been examined. The two main objectives were to determine: (i) how the microparticle concentrations and size distributions vary over the time period represented by the core, and (ii) whether these variations are related to variations in the paleotemperatures as indicated by the δ¹⁸O values. We felt that it might be possible to distinguish between local microparticles (derived from local land surfaces and volcanic eruptions) and global microparticles being carried by atmospheric circulation from distant sources, in terms of ratios of numbers of coarse and fine particles. The diameter of natural tropospheric particles ranges from 0.1 to 20 μm Reference Junge(Junge, 1963, p. 111-202) and the effect of gravity is negligible on particles smaller than 1 μm. Hence, in the “Byrd” core, the global microparticles are concentrated in the smaller particle sizes, while the local microparticles show greater concentrations in the coarser fraction. Table I gives the basic data on the 15 core sections measured. The numbers of microparticles refer to 500 μl samples.

Fig. 3. Fig. 3. Data from the “Byrd” station ice core: (a) Profile obtained from plotting the number of 0.65-0.82 μm diameter particles for the cleanest 10% of the samples from each of the core sections against depth. (b) Profile resulting from plotting Reference Epsfein, Epstein, Sharp and GowEpstein and others (1970) δ¹⁸O values for the core sections analyzed in Figure 4. (c) Average percentage of particles coarser than 1.65 μm for the sections of the “Byrd” core analyzed.

An attempt to restrict our considerations to the global particle concentrations, by disregarding ice samples containing large particles and high total concentrations of particles, has been made by examining the numbers of small microparticles (0,653-0.822 μm diameter) per 500 μl of melt water, in the cleanest 10% of the samples cut from each core section. These would presumably represent snow accumulation in the periods when the Antarctic atmosphere contained least amounts of locally derived dust or short-duration global dust clouds. In Figure 3 these concentrations are compared with Reference Epsfein, Epstein, Sharp and GowEpstein and others (1970) δ¹⁸O values for the same sections. The profiles match well except for the peak in the microparticle concentrations at 400-900 m depth. This is the fractured part of the core (Reference Gow, Gow, Ueda and GarfieldGow and others, 1968) and the high concentration of microparticles is probably contamination. The relationship between small-diameter microparticle concentrations in these cleanest samples and δ¹⁸O values is shown in Figure 4. Except for the two samples from the fractured zone, the number of microparticles increases approximately exponentially as the δ¹⁸O values become more negative.

In Figure 3c the profile is given of the coarseness of the microparticles, expressed as the average percentage, for all of the samples from a section, of the number of microparticles greater than 1.65 μm diameter, compared with the total number of microparticles. Again, a close relationship exists between this percentage and the δ¹⁸O variations; the largest percentages of coarser microparticles occur in ice with the least negative δ¹⁸O values, corresponding to warmer temperature intervals.

Fig. 4. Illustration of δ¹⁸O values plotted against the number of 0.65-0.82 μm diameter particles per 500 μl in the cleanest 10% of the samples from each core section. The length of the vertical line indicates the variance of values within each core section and the small circle indicates the average value. There is an exponential increase in the number of particles with a linear decrease in δ¹⁸O.

Chronology for the “Byrd” Core Derived from Variations in Microparticle Concentrations

Although conclusive proof is lacking that the short-period variations in microparticle concentrations are annual, the earlier work referred to above, and the detailed studies on the sections at 180 and 1 377 m give strong presumptive evidence. (In the 1973-74 Antarctic field season one of us (L.G.T.) collected samples for microparticle analysis from areas near “Byrd” station where the snow stratigraphy and chronology are well known.)

Fig. 5. The annual accumulation of ice in cm (a_i) for each core section plotted according to depth in meters as determined from particle peak counts. The solid line represents the average a_i values for the three parameters.

The a_i values are determined from variations in the following: 0.65-0.82 μm diameter particles, Δ 1.65-2.07 μm diameter particles, ʘ; particles greater than 1.65 μm in diameter, ⊡.

If however, one assumes that the cycle of microparticle concentrations is annual, the age for the ice at the bottom could be obtained by counting the peaks throughout the core. Such an analysis is not practical at present. Instead, we have counted the numbers of peaks in the profiles of microparticles of selected sizes for the 13 analyzed sections and have assumed that the thickness, a _i, of the annual accumulation, for the measured sections may be applied to the unanalyzed parts. The variations of a _i for three ranges of microparticle diameters is shown in Figure 5. Thinning by vertical strain causes a reduction in ai with increasing depth. Assuming that the balance at the surface has remained constant, the vertical strain-rate is about —28×10⁻⁵ a⁻¹, in remarkable agreement with the near-surface vertical strain-rate of —27×10⁻⁵ a⁻¹ calculated by I. Whillans (personal communication) from measurements on the “Byrd” station strain net. In fact, the ice now at depth under “Byrd” station fell as snow at the surface farther up the flow line, where the balance is now greater than at “Byrd” station (Reference WhillansWhillans, 1975).

The ages calculated from the values of a _i for the different size ranges of microparticles are shown in Figure 6, together with Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) age-depth curve calculated from variations in δ¹⁸O values. In comparing the records from “Byrd” station and from Camp Century, Greenland, Johnsen and others pointed out that, while the δ¹⁸O record is very well preserved in the Greenland core, so that their age estimates are reliable, the δ¹⁸O record from the “Byrd” core does not show well-preserved short-period cyclic variations, so that their age estimates are questionable, as they recognize.

Adjusting the time scale of Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) δ¹⁸O profile from “Byrd” station to the 27 000 year age estimate for the bottom ice from Figure 6, one obtains the profile shown in Figure 7b. This shows a much greater correspondence with the Camp Century profile (Fig. 7a) than does the original “Byrd” profile (Fig. 7c). It is tempting to conclude that the close similarity between Figure 7a and b lends support to the microparticle method of age determination and, particularly, to the assumption of an annual period for the variations of microparticle concentration.

Fig. 6. Time-depth relationships for the “Byrd” core: (a) Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) as determined from δ¹⁸O salues; (b) Established by using the minimum a_i values obtained from the particle variations; (c) Established by using the average a_i values as determined from Figure 5; (d) Established by using a_i values determined by variations in particles coarser than 1.65 μm.

Fig. 7. The “Byrd” station and Camp Century oxygen-isotope values as determined by Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972). (a) Oxygen-isotope values for the Camp Century, Greenland, ice core; (b) Oxygen-isotope values for the new “Byrd” station, Antarctica, ice core adjusted to the 27 000 year-old time scale determined from particle variations; (c) Oxygen-isotope values for the new “Byrd” station, Antarctica, ice core as determined by Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) adjusted to the time scale they derived using a_i = 21.4 cm and d = 2.5. Comparison of the three profiles thus obtained suggests that the Camp Century δ¹⁸O profile is more closely related to the “Byrd” δ¹⁸O profile adjusted to the 27 000 year-old particle time scale than the δ¹⁸O profile adjusted to the Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) time scale.

It must be emphasized, however, that the age calculation from the separation of peaks in the microparticle-concentration profiles depends on the extent of post-depositional change in the stratigraphy, as well as on the persistence of an annual cycle. In this preliminary study, a sample size of 2 cm has been used, which may be too large to resolve all of the annual variations. Experiments are now under way to test this point. Furthermore, it must be remembered that a lacuna may exist in the core. We still need dates by the ¹⁴C method.

In the meantime, some support for the estimate of 27 000 years for the age of the bottom ice comes from two sources. Reference Gow, Gow, Ueda and GarfieldBudd and others (1971, p. 149-51). from a profile for the “Byrd” flow line, calculated a maximum age of less than 40 000 years for the ice near the base of the ice sheet at “Byrd” station, and Reference HollinHollin (1962), with reasonable assumptions, showed that the ice sheet could build up from ground level in less than 15 000 years.

Discussion and Conclusions

These preliminary studies on concentrations and size distributions of microparticles, and their relationships with δ¹⁸O values, add several points which must be explained in any successful hypothesis on the causes of glaciation and climatic change, but do not themselves provide unequivocal evidence for a particular cause.

The variations in coarseness of the microparticles may be explained in terms of changing amounts of local and global components. Larger percentages of coarse particles correspond to warmer temperature intervals in the core, which may indicate that more of Antarctica was exposed, providing a larger source for “local” microparticles. Alternatively, atmospheric circulation patterns over Antarctica may change with time. At present in many areas of the continent a greater amount of cyclonic snowfall occurs in the summer than in the winter (Reference Alt, Alt, Astapmko and RopaAlt and others, 1959, p. 185-95; Reference Markov, Markov, Bardin, Lebedev, Orlov and SuyetovaMarkov and others, 1968, English translation, p. 147-248); and Reference HamiltonHamilton (1969) found, at the Pole of Inaccessibility, that the summer precipitation contains a larger number of particles greater than 3,5 μm in diameter than does the winter snowfall. Our observation on coarseness may be explained if the circulation pattern during the cold intervals of the core corresponded more closely to the present winter circulation (predominantly high-altitude convection, subsidence and katabatic surface flow); and that in the warm intervals, to the present summer circulation (more cyclonic).

As a second alternative, during the colder intervals the ice sheet should have been more extensive; large particles are favored as condensation and freezing nuclei over smaller ones (Reference Kumai and FrancisKumai and Francis, 1962). The first deposition from cyclones occurs near the coast, so that with a larger ice sheet only the small nuclei would be deposited inland near “Byrd” station.

The peak in the concentration of small microparticles occurs in the interval of 14 800-25 000 years B.P. according to the chronologies of Reference Johnsen, Johnsen, Dansgaard, Clausen and LangwayJohnsen and others (1972) or rather less than that according to our microparticle-based estimate (12 000-21 000 years B.P.) but still within the period generally accepted for the Wisconsin glaciation.

Reference Gow and WilliamsonGow and Williamson (1971) considered that large quantities of dust injected into the Antarctic stratosphere, by impeding solar radiation, could reduce troposphere temperatures by 2 or 3 deg. On the other hand, Reference LeMasurierLeMasurter (1972), from studies of the ash bands in the “Byrd” core, has shown that it was unlikely that large quantities of ash originated from local volcanoes. The small percentages of large microparticles supports this conclusion. Perhaps the small particles originated from the ash-rich explosive eruptions characteristic of the circum-Pacific volcanoes, and the small particles remained in the atmosphere long enough to reduce solar radiation sufficiently to affect world climate.

Alternatively, the Antarctic temperatures may have been reduced by some other cause; the cooling would have produced a more intense global atmospheric circulation and, hence, an increased number of small (global) particles in the Antarctic atmosphere. It appears unlikely, however, that meridional exchange, and hence global particle flux in Antarctica, could have changed by more than a factor of two, which is insufficient to explain the four-fold variations in Figure 3. Moreover, no mechanism is apparent by which increased exchange, by itself, could cause a reduction in the mean size of microparticles carried to Antarctica.

Acknowledgements

This investigation was conducted with the support of the National Science Foundation grant OPP71-04063-A02, awarded to The Ohio State University Research Foundation and the Institute of Polar Studies. Further support was provided by the Office of Research and Sponsored Programs and by the Graduate School of the University. The sections of the “Byrd” core were provided by Dr C. C. Langway, Jr., then of the U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire.

We are grateful to Mr Ian M. Whillans for reviewing an early draft of the manuscript and for providing unpublished data, to Mr John F. Splettstoesser for editorial assistance, and to Dr W. Dansgaard for his discussions of the relationships between microparticle and δ¹⁸O variations.