Sir

In their paper, Reference Boulton and SpringBoulton and Spring (1986) use the well- known Rayleigh model to interpret the oxygen-isotopic composition of basal ice from Byrd Station and to discuss a possible application for other oxygen-isotopic profiles from polar glaciers. There are some limitations to the use of their approach that I would like to point out.

1. While the fractionation at the water-ice interface is always given by the equilibrium fractionation coefficient (1.003 for ¹⁸O), the amount of observed fractionation is dependent on the freezing rate. This is due to the fact that the water close to the ice interface is more or less depleted in heavy isotopes. The controlling factor is the ratio between the diffusion coefficient of H₂ ¹⁸O in water and the freezing rate. As pointed out by Reference Posey and SmithPosey and Smith (1957), the consequence is that the ice is more or less enriched in ¹⁸O. The observed separation will be different from the true value. The isotopic range between the value at a certain percentage of freezing — that is for the authors at a certain value of discharge reduction - and the value of the initial water is thus not fixed unless diffusion phenomena and freezing kinetics are considered. It would only be fixed if water is always completely homogenized during freezing. This limiting case would give a Rayleigh distribution for an infinitely low freezing rate. Of course, this is probably never realized in Nature. The Reference Jouzel and SouchezJouzel and Souchez (1982) approach was concerned with a co-isotopic study - both D/H and ¹⁸O/¹⁶O - and put forward the concept of the slope of the freezing process on a δD−δ¹⁸O diagram. This slope is not dependent on the freezing rate but the relative positions of the points representing a certain percentage of freezing. If we refer to figures 2 and 3 of the paper by Reference Souchez and JouzelSouchez and Jouzel (1984), the apparent fractionation coefficients can be very different from equilibrium values in freezing experiments but the slope (α − 1)/(β − 1) is the same. This phenomenon has to be taken into account for a precise interpretation of the isotopic composition of basal ice.

2. If an open system is considered, then another difficulty arises. Reference Souchez and GrooteSouchez and De Groote (1985) showed that the freezing slope increases if initial water is mixed with isotopically more negative water in the course of freezing. Thus, the range of values for a single isotopic ratio between the value at a certain percentage of freezing and the value of the initial water will be different if there is some mixing or if there is none. By a study of a single isotopic ratio, it is not possible to tell whether the isotopic profile in basal ice is only due to fractionation by freezing without any mixing. Obviously, a co-isotopic study of basal ice from Byrd Station would be of great value. I am aware that the δD values of basal ice from Byrd Station are probably not available.

I am not sure that, at the present day, a theory can predict that the difference between subglacial water and the basal part of normal ice at Byrd Station would be 5.4% 0 and that the difference between 180 values for subglacial water and the glacier sole with which it is in contact would be 3‰ for the two reasons given above. One cannot exclude the possibility that the slight difference obtained by the authors is purely coincidental, as other combinations of factors would allow one to reach the same result. I do not yet see a means of testing this further with a single isotopic approach.