In January 1986, a 133 m ice core, with an estimated age at the bottom of 300-350 years, was collected (using an electromechanical drill) on Dolleman Island (70° 35.2′S, 60°55.5′ W; 398 ma.s.l.; 10 m temperature −16.75°C). The site lies on the east coast of the Antarctic Peninsula and has a continental-type climate dominated by perennial sea ice in the Weddell Sea. The core is being analysed for a range of chemical impurities, in order to assess their potential as indicators of past climate.

High-resolution (10-15 samples a−1) continuous profiles of the anionic species Cl−1, NO3− and SO42−, together with the cation Na+, have been measured on a section of the core from 26 to 71 m depth. The core has previously been dated between 0 and 32 m depth using the δ18O profile (Peel and others 1988). Lack of δ18O data for the section 32-71 m forced us to seek an alternative method of dating.

Biogenic outgassing of sulphurous gases from the ocean and subsequent photochemical oxidation contribute an excess of sulphate over that derived from the marine aerosol. We show that excess sulphate, calculated as

(concentrations in Eq. 1−1 and assuming that all measured Na+ is derived from sea salt), is highly seasonal in character, and annual horizons are well preserved over the whole of the core. This enabled us to determine the chronology to 71 m depth, and date the bottom of this section as 1844 ± 5 years.

Cl− is derived mainly from sea salt. Its profile in the core is also seasonal in character, with peaks that tend to occur in late summer, reflecting the period of minimum sea-ice extent in the Weddell Sea, and therefore maximum source area for the uptake of sea salt. From instrumental meteorological records, Limbert (1974) showed that there were three extended periods of warm or cold weather in the Antarctic Peninsula between 1903 and 1944. During the two 4 year cold periods, when the summer break-up of sea ice in the Weddell Sea is likely to have been reduced, we found that the annual flux of Cl− to the Dolleman Island snow-pack was lower than the average. Conversely, the 3 year warm period showed a peak in the values of annual flux of Cl−. We therefore propose that Cl− can be used as a palaeoclimatic indicator for sea-ice extent.

Extending our chloride data into the latter half of the nineteenth century (before the earliest continuous instrumental records for the Antarctic), we found three distinct peaks in the values of annual flux of Cl−. We suggest that the period 1850-60 was marked by a decrease in Weddell Sea ice extent (due perhaps to a warm period), followed by an extended period of increased sea ice. There were then two periods of much-reduced sea ice during (approximately) 1885-1890 and 1895-1900, with an intervening period of greatly increased ice coverage. These events are in good agreement with the warm and cold periods which Aristarain and others (1986) identified in the deuterium profile from James Ross Island.