89, 2H, AND °H IN NATURAL WATERS 685 previously when the vertical transfer was considered, the turbulent exchange processes tend to reduce the differences in 489 content from one place to another, and the same presumably holds for horizontal turbulent exchange. Dansgaard’s measurements also show that the mean '89 concentrations vary only about 10 per mille within the latitude belt 55°N to 55°S (excluding only some continental stations for which other processes might be of importance) in spite of the fact that the total water content of an air column may vary from an average of about 6.5 g/cm? in the tropics to about 1 g/com? close to the polar re- gions. A much more careful study of the variations of the isotopic composition of rainwater must be conducted in conjunction with measurements of the isotopic composition of atmospheric moisture to arrive at more conclusive results about the flux of water vapor through the atmosphere. The preceding discussion is, of course, very tentative in many respects. It shows, however, that we may expect considerable differences in the distribution of various isotopes as a result of the relative importance of different physical processes contributing to the transfer processes. Data are still inadequate for a more quantitative evaluation of these differences. Obviously systematic measurements of the isotopic composition of atmospheric water vapor as a function of elevation and latitude seem very desirable, and simultaneous measurements of particles and artificial radioactivity would provide additional im- portant information. Studies of the isotopic composition of precipitation as conducted at present are of course also of value, although they are insufficient since such measurements represent an integration over the condensation and evaporation processes, the result of which may be difficult to interpret by itself. It should be added that some simple studies of the tritium budget of the atmosphere as compared with that of *Sr give some further support to the general picture of the verticaltransfer affecting tritiated water vapor and particular matter differently as previously indicated. Machta? has shownthat the sum of the total deposition of Sr over land and by rain over the oceans added to the amount remaining in the atmo- sphere is about 80% of the total injected (summer 1963). A corresponding inventory of tritium? shows quite a different result. In this inventory it is estimated that the direct air—sea transfer of tritium must be about twice that due to precipitation to obtain an approximate balance. This probably is a lower limit since only bomb-produced tritium was considered in the computation. REFERENCES 1. V. Starr and J. P. Peixoto, The Hemispheric Eddy Flux of Water Vapor and Its Implications for the Mechanics of the General Circulation, Arch. Meteorol. Geophys. Bioklimatol., Ser. A, 14: 111-130 (1964).

Select target paragraph3