G8 



DATA OF GEOCHEMISTRY 



and doubtless for other elements as well. The river 

 forms a dynamic system in which the biological ele- 

 ments are continually exchanging, and at a particular 

 time the fraction of the total mobile phosphorus, nitro- 

 gen, or silicon that is dissolved in the water may vary 

 from an indetectable amount to almost the whole of it. 



Strakhov (1948) has investigated the state of iron in 

 natural waters and has come to the conclusion that for 

 this element, under most circumstances, the proportion 

 of the total amount that is transported in a dissolved 

 form in river waters is very small. The fraction of geo- 

 chemical significance is bound on the surfaces of the 

 fine mineral grains carried in suspension. Vernadskii 

 (1948, in the Discussion of Strakhov's paper) has further 

 pointed out that even Strakhov overestimated the im- 

 portance of the dissolved iron, for the solubility product 

 of ferric iron is such that an insignificant amount is in 

 simple solution under ordinary conditions in most river 

 waters. This need not concern us, however, for we are 

 more interested in using "dissolved iron" in the sense 

 of the analysts who report this entity in their analyses 

 than in a rigorous physico-chemical way. Shapiro 

 (1957) has shown that some iron may be dissolved in 

 natural waters, despite the low solubility of ferric ion. 

 Organic compounds of moderate molecular weight 

 stabilize the iron and keep it in solution. 



Iron is an extreme example, but it is not alone. The 

 behavior of manganese, cobalt, and nickel must be 

 rather similar. One must not lose sight of the solu- 

 bility of the solid silicates, including not only the dia- 

 toms, sponges, and other minute silica particles in the 

 river, but also the silica in the glass bottles that are still 

 commonly used as sample containers. The last source 

 of error is most likely to be serious for strongly alkaline 

 waters. Hutchinson (1937) found the following in- 

 creases in the silica content of some water samples from 

 Indian Tibet between the time of their collection in the 

 field and their analysis in the United States: 



Tso Moriri—- 



Tso Kav 



Khyagar Tso. 



Yaye Tso 



Mitpal Tso.. 

 Pangur Tso.. 

 Pangong Tso. 

 Ororotse Tso. 



Besides these sampling errors there is a certain am- 

 ount of error in the analytical procedures used to deter- 

 mine the composition of water samples. In earlier 

 editions of this work considerable effort was devoted to 



selecting trustworthy analyses. To a great extent this 

 was possible, because the analyst responsible for a partic- 

 ular analysis was usually known, and something of his 

 skill and experience were known as well. In the much 

 larger scientific community of today it is impossible to 

 have a critical familiarity with the competence of more 

 than a small fraction of the analysts concerned, and, 

 perhaps for this practical reason, the analyst is less 

 often specified in the published account of an analysis. 

 For papers having several authors, one might presume 

 that the analyses had been carried out by the authors 

 themselves, but frequently the analytical responsibility 

 is shared by an entire organization. Obviously un- 

 trustworthy or outmoded analyses have been avoided. 

 Another method of distinguishing good from bad 

 analyses that was used in the earlier editions was to 

 check the equivalence of anions and cations reported 

 in the analyses. Apart from the possibility of compen- 

 sating errors, which it cannot reveal, this method has 

 the objection that uncertainties arise about the position 

 of dubiously ionized components. 



In this edition the most suitable of the available 

 analyses were chosen, which means, for a large part of 

 the earth's surface, all available analyses. These 

 explanatory notes are intended as a general caveat 

 concerning the reliability of the results. 



Temporal variations in the chemical content of rivers, 

 as we have seen, are associated principally with varia- 

 tions in river discharge. To some extent lakes, par- 

 ticularly those of arid regions, become more concen- 

 trated as the level falls, but this is not the principal 

 cause of changes in lake-water chemistry. Lakes in 

 general are chemically more stable than streams and 

 they do not show such striking changes in the amount 

 and proportions of the principal dissolved substances. 

 Very concentrated lakes in cold arid lands may 

 display an annual chemical cycle, as the lowered temp- 

 erature of winter causes the water to fall below the 

 saturation temperature of one or more of the materials 

 that dissolved in the water during the heat of summer. 

 When this happens salts crystallize out of the water, 

 and their concentration falls. In addition, because the 

 least soluble salts come out first, the percentage 

 composition of the remaining mineral substance is 

 altered. 



Even less concentrated lakes may display wide 

 annual fluctuations in concentration if they are shallow 

 in relation to their ice cover. Figure 3, for example, 

 shows how the chloride and magnesium contents of a 

 shallow Arctic lake increased by freezing out of salts into 

 the water that remains under the whiter ice. Imikpuk, 

 the lake described in the figure, occasionally receives 

 some sea water, but a similar change is to be expected 

 even in completely fresh Arctic waters. 



