trons for the dark-adapted eye of acareful observer to detect this phenom-enon. It remains for future research tocharacterize the relationships of flashbrightness and shape with charge, mo-mentum, and trajectory.

THoMAs F. BUDINGERDonner Laboratory and LawrenceRadiation Laboratory, University ofCalifornia, Berkeley 94720

HANS BICHSELDepartment of Radiology, Universityof Washington, Seattle 98105

CORNELIUS A. TOBIASDonner Laboratory andLawrence Radiation Laboratory

References and Notes

1. C. A. Tobias, T. F. Budinger, J. T. Lyman,Doc. UCRL-19868 (Lawrence Radiation Labor-atory, Berkeley, Calif., 1970); Nature, in press.

Gibbs's (1) stated intention in writinghis report was "to elucidate the majornatural mechanisms controlling worldwater chemistry." The three naturalmechanisms that he cited are the fol-lowing: (i) atmospheric precipitation,(ii) rock dominance, and (iii) the evap-oration-crystallization process. The ar-

gument, in brief, was that some streams,especially those draining well-leached,tropical basins, have water compositionsdominated by Na+ and Cl-, an indi-cation that the composition is con-

trolled by airborne salt with strongoceanic affinities [figure 2 in (1)]. An-other end-member is dominated byCa2+ and HCO3- with generous con-

tributions from K+ and SiO2, an in-dication that the composition is con-

trolled by mineral components leachedfrom the rocks of the drainage basin.The third type of water, also domi-nated by Na+ and CI-, but with a highconcentration of dissolved solids, re-

flects "evaporation, which increasessalinity, and . . . precipitation ofCaCO3 from solution, which increasesthe relative proportion of Na to Caboth from the tributaries and in themainstream" (1, p. 1090). The PecosRiver and the Rio Grande were citedas typical examples of the evaporation-crystallization process.

Inasmuch as specific data pointswere not given, it is impossible to ex-

amine Gibbs's (1) conclusions rigor-ously. Examination of his figure 4,

870

2. G. G. Fazio, J. V. Jelley, W. N. Charman,Nature 228, 260 (1970).

3. A. Prince, in Californium-252, U.S. At. EnergyComm. Conf. Rep. 681032 (1968), p. 23.

4. J. H. Fremlin, New Sci. 47, 42 (1970); M. J.Chamberlain, J. H. Fremlin, D. K. Peters, H.Philip, Brit. Med. J. 2, 581 (1968).

5. W. B. Nelp, H. E. Palmer, R. Murano, K.Pailthorp, G. M. Hinn, C. Rich, J. L. WSlliams,T. G. Rudd, J. D. Denney, J. Lab. Clin. Med.76, 151 (1970).

6. E. Tochilin and G. D. Kohler, Health Phys. 1,332 (1958); J. N. Schultz and W. A. Glass,ibid. 13, 1237 (1967). The available spectrumis for 20 Mev, although the experiment wasdone with 22-Mev deuterons.

7. C. Gaffey, in International Symposium on theResponse of the Nervous System to IonizingRadiation, 2d, Los Angeles, T. J. Haley andR. S. Snider, Eds. (Little, Brown, Boston,1964), p. 243.

8. S. Hecht, S. Shlaer, M. Pirenne, J. Gen. Phys-iol. 25, 819 (1942).

9. We thank J. D. Denney and W. B. Nelp forhelp in the conduct of these experiments. Thisresearch is supported jointly by the U.S.Atomic Energy Commission, National CancerInstitute (PHS grant CA-12446), and Na-tional Aeronautics and Space Administration.

5 March 1971 M

however, suggests that the choices ofexamples of the evaporation-crystalli-zation process are questionable. Gibbs'sfigure 4 shows "total dissolved salts"in Pecos River water increasing fromslightly more than 1000 parts per mil-lion (ppm) to nearly 6000 ppm; theincrease in total dissolved salts in theRio Grande is from a little more than300 ppm to nearly 900 ppm.

There is no argument with the prop-

osition that concentration by evapo-transpiration and irrigation return wa-

ter contribute to the mineralization ofPecos River and Rio Grande waterprogressively downstream. But the ma-

jor increments of increased mineraliza-tion to the Pecos are from ground-water inflows, some of which are brinesnearly saturated with respect to sodiumchloride.

Hale et al. (2) reported that in a

reach of only about 3 miles (4.8km) in the Malaga Bend south ofCarlsbad, New Mexico, seepages andsprings added about 420 tons (380,000kg) of dissolved minerals to the Pecosdaily, of which about 370 tons. is com-mon salt (NaCI). Since the Pecos istributary to the Rio Grande, it is rea-

sonable to infer that the increase bothin the mineral concentration and inthe proportion of Na+ among thecations in the Rio Grande shown byGibbs [figure 1 in (1)] is accountedfor largely by a comparison of theanalyses of water from the Rio Grande

taken above and below the junctionof the Pecos.

Gibbs's (1) reliance on the deposi-tion of CaCO8 to explain increasedNa+ percentages in waters dominatedby the evaporation-crystallization proc-ess is unsupported by any evidencethat in those waters the solubility limitof CaCO3 is exceeded. Nor is evidencegiven that, if supersaturation does oc-cur, it will be succeeded by nucleationand precipitation of CaCO3.

Gibb's figure 4 (1) also shows theSacramento River to have a concen-tration of dissolved solids of between4000 and 5000 ppm, and a high per-centage of Cl- among the anions. Forthe years 1950-1957 Van Denburghand Feth (3) calculated an averageconcentration of dissolved solids of168 ppm for the Sacramento at KnightsLanding, California, the farthest down-stream sampling station on the riverthat is still above the intrusion of sea-water. The concentration indicated byGibbs can occur in the Sacramentoonly in the delta reaches where mix-ing with seawater increases both theoverall mineral concentration and thepercentage of Cl- among the anions.In light of the foregoing, one viewswith some hesitation the Jordan Riverexample as well. The sample site isnot indicated, but it is possible thatthe site is downstream from materialsaline inflows from springs in the vicin-ity of Lake Tiberias.

It appears, further, that Gibbs (1)used data on precipitation chemistryfrom the Mojave Desert to calculatethe percentage -of several cations inthe Rio Grande derived from rainfall.The difficulties associated with the der-ivation of mass balances for chemicalloads of streams have been discussedby Van Denburgh and Feth (3). Oneof the dangers they mentioned arisesfrom extrapolations of incompletedata across large regions. The massbalance for the Rio Grande must, inconsequence of its data input, be re-garded as a first approximation at best.The above comments imply that

Gibb's selection of data was not whollysound and that some data were inap-propriate to his argument. Unfortu-nately, the visible flaws suggest thatthe conclusions presented should havebeen supported with more rigorouslyscreened data.

J. H. FETHU.S. Geological Survey,345 Middlefield Road,Menlo Park, California 94025

SCIENCE, VOL. 172

Mechanisms Controlling World Water Chemistry:Evaporation-Crystallization Process

on August 7, 2018

http://science.sciencem

ag.org/D

ownloaded from

References and Notes

1. R. J. Gibbs, Science 170, 1088 (1970).2. W. E. Hale, L. S. Hughes, E. R. Cox, "Pos-

sible Improvement of Quality of Water of thePecos River by Diversion of Brine at MalagaBend, Eddy County, New Mexico" (PecosRiver Commission, Carlsbad, N.M., 1954).

3. A. S. Van Denburgh and J. H. Feth, WaterResouir. Res. 1, 4 (1965).

4. Publication authorized by the director, U.S.Geological Survey.

21 JaInuary 1971; revised 18 March 1971

Feth's questions (1 ) regarding theevaporation-crystallization process inarid regions provide an opportunityfor the elaboration of several pointswhich the brevity of my original re-port (2) did not allow. Contrary toFeth's initial observation, the sourcesof the specific data for all rivers namedwere clearly identified in the secondparagraph of my report (2) as areadily available U.S. Geological Sur-vey publication by Livingstone (3);the source for the Amazon data is apaper in preparation (4).

Examination of Livingstone's data(3) indicates that the choice of thePecos River, the Rio Grande, and theJordan River as examples of riverscontrolled by the evaporation-crystalli-zation process is far from question-able. Rather, Livingstone's data (3)strongly support the inclusion of thesethree rivers as arid-region rivers ex-emplifying this process. Unfortunately,a drafting error in figure 4 (2) is re-sponsible for the misplacement of thename of the Sacramento River withthe group of rivers controlled by theevaporation-crystallization process. Liv-ingstone's data (3) show the correctgrouping of the Sacramento River withthose rivers exemplifying control ofcomposition by rock dominance.

Feth's attempt to discredit thegrouping of the Pecos River and theRio Grande with those rivers con-trolled by the evaporation-crystalliza-tion process is based on two errone-ous assumptions: that groundwater wasnot considered as a part of the processand that the trend of the Rio Grandetoward increased salinity and a higherratio of Na to (Na + Ca) is basedsolely on the mixing of the Pecos withthe Rio Grande.

In my original discussion of theevaporation-crystallization process, con-sideration of groundwater was cer-tainly implied in my statement (2,p. 1090) that the change in composi-tion and concentration along thelength of the arid-region rivers is dueto evaporation "and to precipitation ofCaCO3 from solution,"-including, nat-21 MAY 1971

urally, both surface and ground solu- of the Itions-"which increases the relative contradi(proportion of Na to Ca both from the stone (3tributaries and in the mainstream"- ficationwith the phrase "from the tributaries points irand in the mainstream" again inferring Three oconsideration of the entire basin, of upper piwhich groundwater is an integral part. evidenceFigure 1, which is based on data of both creasedLivingstone (3) and Love (5), clearly Na to (shows that the shift in chemical com- fined priposition along the entire length of the mixingPecos is a steady one, with none of Grande.the spectacular variation that, accord- compilating to Feth's implication, occurs with Livingstcthe addition of groundwater at Malaga ports theBend. -entire]

Feth's further assumption that the contribulincreased-salinity trend of the Rio ample oGrande results solely from the mixing its chem

10,000 ,

0

Fort Sumner, N.MM

I- ~~~~~~~ElPaso, Tex.

Labatos,-..-Colo

*World river average100- .: : : - *:

0.1 0.2 0.3 0.4 0.5 0.6

Na/(Na+Ca)Fig. 1. Variations of the weight ratio Na/(Na + Ca)solved salts of arid-region surface waters.

Pecos with the Rio Grande iscted by the data of Living-'). These data give clear identi-of the four Rio Grande datan figure 1 of my report (2).f these points, located in theortion of the Rio Grande, give

that the trend toward in-salinity and a higher ratio ofNa + Ca) is already well de-ior to and independent of theof the Pecos with the RioFigure 1, based on Love's

tions (5) [which substantiateone's work (3) ] further sup-e inclusion of the Rio Grandely apart from the tributarylion of the Pecos-as an ex-A an arid-region river havingical composition controlled by

as a function of the total dis-

871

on August 7, 2018

http://science.sciencem

ag.org/D

ownloaded from

the evaporation-crystallization process.The basis for Feth's objections to

the inclusion of the Jordan River as anarid-region river in the evaporation-crystallization grouping-that "the sam-ple site is not indicated"-is also un-founded. Feth hypothesizes that theJordan sampling site is dominated bygroundwater seepage. The samplingpoint at Jericho (3) does, indeed, re-flect the "saline inflows from springsin the vicinity of Lake Tiberias" (1)resulting from the fact that most ofthese springs enter Lake l7iberias (6)which serves thus not only as a mixingbasin, but also as the source of theJordan. However, as shown in Fig. 1,the 507-ppm salinity of Lake Tiberias(6) [also given by Livingstone (3)],serves as the base point for the steadyincrease in the salinity of the Jordan to1310 ppm at Jericho (3) after thewater flows through a region that iswithout major springs (6). This in-creasing salinity and the changingratio of Na to (Na + Ca) are evi-dence in support of the inclusion of theJordan River as an arid-region river inthe evaporation-crystallization processgrouping.

Figure 1 shows the distinct trendsof the Rio Grande (upstream fromthe Pecos junction), the Pecos, theColorado, and the Jordan rivers to-ward steadily higher salinities andhigher Na values relative to Ca, basedon data compiled by Love (5) andBentor (6), which substantiate that ofLivingstone (3). For each of theserivers the site sampled farthest down-

stream is far from either oceans orseas and, therefore, is not at all in-fluenced by seawater.

Feth also desires elaboration on theprecipitation of CaCO3. It is commonknowledge that the soils in arid-regionriver basins have an abundance ofCaCO3 precipitate in the form ofcaliche which, in some instances, pro-duces a limestone-like layer in thesoils.With regard to the mass-balance

calculation shown in table 1 (2) forthe Rio Grande (as an evaporation-crystallization river type),, the rain-fall data used are of the best qualityavailable for the area. Although thiscalculation is, indeed, a first approxi-mation, it does vividly demonstrate thecontribution of Na, K, Mg, and Cacations from rocks (99.9 percent) tothe evaporation-crystallization rivertypes. Further calculations based onadditional data would not change thesignificance of the conclusion alreadydrawn.

R. J. GIBBSDepartment of Geological Sciences,Northwestern University,Evanston, Illinois 60201

References

1. J. H. Feth, Science, this issue.2. R. J. Gibbs, ibid. 170, 1088 (1970).3. D. A. Livingstone, U.S. Geol. Surv. Prof. Pap.

440-G (1963).4. R. J. Gibbs, in preparation.5. S. K. Love, U.S. Geol. Surv. Water Supply

Pap. No. 1950 (1966), pp. 448-588; U.S. Geol.Surv. Water Supply Pap. No. 1951 (1966), pp.33-222.

6. Y. K. Bentor, Geochim. Cosmochimn. Acta 25,239 (1961).

8 March 1971

Precautions with Alkyl Mercury

This technical comment has beenwritten in response to inquiries con-cerning recommended safety precau-tions in the handling of alkyl mercurycompounds. In 1865 Edwards first re-ported that two laboratory techniciansdied from intoxication caused bymethylmercury (CH3Hg) while study-ing valencies of metallic compounds(1). Although the lethal effects of thevolatile compounds in this case werecommunicated from one laboratorytechnician to another, additional casesof poisoning of laboratory personnelcaused by alkyl mercury have beenreported. In these cases, inhalation wasthought to be the mode of contactwith the organic mercurial (2). Hun-ter et al., in 1940, described intoxica-tion caused by CH3Hg in four individ-

872

uals in a seed-dressing factory (3). Allhad taken some precautions in handlingthe reagents (for example, masks andgloves). They developed subacute ill-ness with initial symptoms of cerebellarataxia, paresthesias, disturbances of thevisual field, dysarthrias, increased sali-vation, and dysphagia. The diseaseprogressed to more severe neurologicalsymnptoms, including deafness, blind-ness, spasticity, paralysis, urinary in-continence, and seizures.Some alkyl mercury compounds are

very volatile. The saturated vapor con-centration of methylmercury chlorideat 20°C is 94,000 jig per cubic meterof air as compared with 5 ,tug per cubicmeter of air for phenylmercury chlo-ride and 14,000 ,tg per cubic meter ofair for metallic mercury (4). The high

volatilities of alkyl mercury compoundsincrease the risks to those handlingthe compounds.The International Committee to

Determine Maximum Allowable Con-centrations of Mercury met in Swedenin 1968 and recommended that no oneshould be exposed to more than 0.01mg of alkyl mercury per cubic meterof air over an 8-hour period and thatthe mercury concentration in wholeblood should not exceed 10 ,ug per100 ml (as total mercury) (5). Forlaboratory work we suggest that thefollowing precautions be taken:

1) All handling of CH3Hg shouldbe done in a well-ventilated hood.Syringes should be filled in a hood; ifthe solution is spilled, cysteine or someother sulfhydryl-rich solution must bemixed with the spilled alkyl mercurialin order to decrease the volatility. Alltreatments of animals should be per-formed in a hood.

2) Methylmercury, when not in use,should be stored in disposable, closed,glass containers (for example, a peni-cillin ampoule) at refrigerator tem-perature. Feces, although a majormetabolic excretion pathway for alkylmercury in rats, contain CH3Hg innonvolatile, bound form, thus mini-mizing the danger from this source.

3) Concentrations of mercury in theblood of all those working with thesesubstances should be measured period-ically. If the concentration exceedsthe recommended maximum bloodconcentration, the individual shouldbe immediately removed from thearea of further exposure to alkylmercury, and his handling techniquesshould be evaluated for carelessness orother flaws.

4) Women of childbearing age, andespecially pregnant women, should notwork with alkyl mercury compoundsbecause of possible teratogenic effectson the central nervous systems of theunborn children (5).

RONALD KLEINSHELDON HERMAN

National Institute of EnvironmentalHealth Sciences, Research TrianglePark, North Carolina 27709

References

1. G. N. Edwards, St. Bartholomew's Hosp. Rep.i, 141 (1865).

2. L. T. Kurland, S. N. Fars, H. Siedler, WorldNeurol. 1, 370 (1960).

3. D. Hunter, R. R. Bomford, D. S. Russell,Quart. J. Med. 9, 193 (1940).

4. U. Ulfvarson, in Fungicides, An AdvancedTreatise, D. C. Torgeson, Ed. (AcademicPress, New York, 1969), vol. 2, pp. 303-329.

5. Report of an International Committee, Arch.Environ, Health 19, '1 (1969).

12 April 1971 LT c

SCIENCE, VOL. 172

on August 7, 2018

http://science.sciencem

ag.org/D

ownloaded from

Mechanisms Controlling World Water Chemistry: Evaporation-Crystallization ProcessJ. H. Feth and R. J. Gibbs

DOI: 10.1126/science.172.3985.870 (3985), 870-872.172Science 

ARTICLE TOOLS http://science.sciencemag.org/content/172/3985/870

CONTENTRELATED file:/contentpending:yes

REFERENCES

http://science.sciencemag.org/content/172/3985/870#BIBLThis article cites 2 articles, 1 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceGovernment Works. The title reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S.Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. 2017 © The Authors, some rights

(print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for theScience

on August 7, 2018

http://science.sciencem

ag.org/D

ownloaded from