We have shown previously (Mulvaney and others, 1988; Wolff and others, 1988) that some of the impurities in ice are localised. For samples from Dolleman Island in the Antarctic Peninsula, sulphuric acid was found at very high concentrations at the triple junctions (where three grains meet). No such localisation was found for sea salt elements, which are the other major soluble impurity. We believe that the acid is sufficiently concentrated at ice-sheet temperatures to remain liquid, forming a network of sub-micron veins through the ice.

We used a scanning electron microscope (SEM) fitted with an X-ray microanalysis system and a cold stage that holds samples below −160°C. Located at the University of Lancaster, the instrument allows frozen samples to be investigated with elemental analysis carried out at a resolution of the order of 1 micron.

Further experiments have yielded similar results for other samples from the same ice core. However, we have not yet found a method of cutting and cooling the samples that gives quantitatively reproducible data, so that it is too early to say what proportion of the acid in the sample is at the triple junctions.

Nonetheless, we have now also seen S at several triple junctions in ice from Site G in central Greenland. The sample includes part of the material from the 1783 Laki volcanic eruption. We have still to look at samples from other sites, but are reassured that the positive result is not confined to one ice core.

This work, still at a formative stage, has posed some important questions:

(1) For us there is the technical question of how we obtain reproducible quantitative results.

(2) How widespread is the phenomenon, and how much of the acid is at triple junctions? This is the next phase of studies at Lancaster, and is likely to include a study of older ice, and of temperate ice.

(3) Why is the acid at triple junctions, and why is sea salt not found there? This must be due to processes in the atmosphere or snowpack, and is likely to be related to the eutectic temperatures of impurity/water mixtures. Thus the distribution may influenced by changes in climate or chemistry. For instance, Wisconsin-age ice in Greenland is neutral, any acid having reacted with alkali dusts. How did this affect the impurity distribution?

(4) If the distribution does change as a result of a changed environment, does this affect the physical properties of the ice itself? In particular, is the presence or absence of liquid at the junctions a contributory factor to the changes in rheology between Wisconsin and Holocene ice? We are far into the realms of speculation here, but this does have the potential to be an interesting long-timescale feedback to climatic and environmental changes.