Applied Geochemistry, Vol. 13, No. 8, pp.

941±952, 1998
# 1998 Elsevier Science Ltd
All rights reserved. Printed in Great Britain
PII: S0883-2927(98)00029-8 0883-2927/98 $ - see front matter

Chemistry of rainwater in the Massif Central (France): a strontium

isotope and major element study

Philippe NeÂgrel*
BRGM Service GeÂologique National, Avenue C. Guillemin, BP 6009, 45060 OrleÂans Cedex 2, France

and

SteÂphane Roy{
Laboratoire de GeÂochimie et Cosmochimie, Institut de Physique du Globe de Paris, U.F.R. des
Sciences de la Terre, Universite de Paris VII. Tour 14±24, 3 eÂt., 4 place Jussieu, 75252 Paris, France

(Received 11 June 1996; accepted in revised form 10 February 1998)

AbstractÐAtmospheric aerosols (sea salt, crustal dust, and biogenic aerosols) are the primary source of
dissolved species in rainwater as well as one of the sources of dissolved species in river water. Chemical
weathering studies require quanti®cation of this atmospheric input. The crustal component of atmos-
pheric input can have various origins, both distant and local. The proportions of the various inputs
(marine, distant or local) are determined in this study.
Strontium isotope ratios and Ca, Na, K, Mg, Al, Cl, SO4, NO3 and Sr concentrations were measured
in rainwater samples collected in the Massif Central (France) over a period of one year. Each sample,
collected automatically, represents a monthly series of rain events. Chemical composition of the rain-
water samples varied considerably and the 87Sr/86Sr ratios ranged between 0.709198 and 0.713143.
Using Na as an indicator of marine origin, and Al for the crustal input in rain samples, the pro-
portion of marine and crustal elements was estimated from elemental ratios. A marine origin of 4 to
100% of Cl, of 0.6 to 20% of the SO4, of <1 to 10% of Ca, <1 to 40% of K, 4 to 100% of Mg and
1 to 44% of Sr was determined.
Strontium isotopes were used to characterize the crustal sources. The 87Sr/86Sr ratios of the crustal
sources varied considerably from 0.7092 to 0.71625 and indicate the occurrence of multiple sources for
the crustal component in the analysed rainwaters. # 1998 Elsevier Science Ltd. All rights reserved

INTRODUCTION Terrestrial aerosols may originate from soil ero-

Aerosols are the main source of chemical elements
carried by rainwater (Junge, 1963). They are
removed from the atmosphere by rainout (conden-
sation processes within clouds), or by washout
(impaction with falling raindrops below clouds).
The atmosphere, by means of either dry (aerosols)
or wet (rainwater) deposition, has been recognised
as a major source of dissolved species in rivers
(Gibbs, 1970; Likens et al., 1977; Stallard and
Edmond, 1981; Meybeck, 1983, Sarin et al., 1989;
NeÂgrel et al., 1993). The dissolved components in
rainwater can be divided into 3 groups: (a) those
derived from sea salt aerosols; (b) those derived
from terrestrial aerosols (soil dust, biological emis-
sions); and (c) those derived from anthropogenic
sources (industry, agriculture, burning of vegetation
and fossil fuels and fertilizers).
*Corresponding author.
{Now at BRGM, Service minier National, Avenue C.
Guillemin, BP 6009, 45060 OrleÂans Cedex 2, France.
941
sion near the collector (local source), from distant
areas within the continent (external source) or even
from other continents (exogenic source). Likewise,
aerosols derived from anthropogenic sources may
originate in the vicinity of the collector (local
source) or far from it (distant source). It is dicult
to unravel the di€erent sources, unless isotopes are
used in the atmosphere (Anderson et al., 1990). In
this work the use of 87Sr/86Sr ratios are proposed to
®nd the sources of the terrestrial aerosols.
The radiogenic isotope 87Sr is produced by the
radioactive decay of the 87Rb isotope. The abun-
dance of 87Sr (expressed as the 87Sr/86Sr ratio) in a
given geological material is a function of its age
and Rb/Sr ratio. The Sr released by aerosols of
di€erent origins results in di€erent 87Sr/86Sr ratios
in rainwater, which may reveal the various types of
aerosols that are dissolved. Given the short time
scale of the processes studied, Sr isotope variations
are due mainly to the mixture of di€erent sources
with di€erent Sr isotope compositions.
942 P. Negrel and S. Roy
Strontium isotope ratios in precipitation have
been reported by several authors. Graustein and
Armstrong (1983) and Gosz and Moore (1989)
reported 87Sr/86Sr values between 0.7088 and 0.7200
for rain in New Mexico. Aberg et al. (1989) found
large variations in dry plus wet deposition (0.711±
0.717) in central Sweden. Anderson et al. (1990)
reported values ranging between 0.7098 and 0.7194
in snow from central Scandinavia. NeÂgrel (1992)
measured values between 0.70968 and 0.71277 in
rainwater over the Congo Basin. Dupre et al.
(1994) have shown that in a rain event collected in
the Congo Basin, the isotope composition of Sr var-
ies considerably (0.71035±0.71326). Herut et al.
(1993) have reported a small range in Sr isotopes
(0.70792±0.70923) in rainwater of Israel. Seimbille
et al. (1989) have reported a small range in Sr iso-
topes (0.70829±0.70895) in rainwater of the Parisian
Basin. These variations are explained by mixing
between several sources in the atmosphere and by
atmospheric circulations.
Data are for major (Cl, SO4, NO3, Al, Ca, Na,
Mg and K) and trace (Sr) element concentrations
and 87Sr/86Sr ratios in rainwaters (hereafter refered
to PSM which stands for the name of the sampling
site Sainte Marguerite) collected in the Massif
Central (France). The aim of this study was to use
Sr isotope ratios correlatively with major and trace
elements in order to determine the di€erent sources
(sea salt and crustal sources) of chemical elements
in rainwaters.
METHODOLOGY
Collection procedure, sampling site
An automatic precipitation sampler was designed and
constructed at BRGM for collecting rain for chemical and
isotope (87Sr/86Sr) analyses. The basic requirements for the
collector were automatic detection and collection of rain-
fall, elimination of dry fallout, collection of frozen precipi-
tation, and avoidance of sample contamination for the
elements studied (especially trace elements). Several fea-
tures of previously reported collectors (Likens et al., 1977;
Lacaux and Warburton, 1980; Galloway et al., 1982;
Church et al., 1984; Colin et al., 1987; NeÂgrel, 1992;
Dupre et al., 1994) have been incorporated into the design
of this sampler. The body of the sampler is a box made of
1 cm-thick PVC plates. A polypropylene funnel 45 cm in
diameter is used to collect the rainfall. A removable PVC
lid covers the funnel when no rain is falling. The cover
exposes the funnel when electrical contact is made in the
rain detector. Movements are controlled by a reversible
motor mounted inside the box. The movement of the
cover is guided by two nylon wires which prevent metal
contamination. The funnel ori®ce is about 1.5 m high. The
funnel support has extended vertical sides (15 cm) which
counter aerodynamic e€ects on sampling eciency
(Sculdark and Church, 1988).
The precipitation detector is made of two grids of 12
parallel bars placed 5 mm apart. Rain drops close the elec-
trical circuit between the grid which triggers the opening
of the sampler. A 12 L polypropylene bottle is mounted
below the funnel inside the box. From the base of the fun-
nel, water ¯ows through a 165 mm nylon mesh ®lter to
prevent fallout of large particulate matter (leaves, insects)
and contamination of the sample. Samples were normally
removed once a month, and more often during periods of
heavy rainfall.
The sampling site is in an open area (altitude 340 m),
far from dirt roads or other sources of contamination, in
the Massif Central (Fig. 1) between the towns Issoire and
Clermont-Ferrand, near Sainte Marguerite and was
selected to minimise local sources of contamination. There
are no trees or rock outcrops in the vicinity of the collec-
tor. Site accessibility was a major consideration for easy
maintenance during the winter months. The
Mediterranean is 300 km to the S and the Atlantic Ocean
is 380 km to the W. The nearest towns, Clermont-Ferrand
and Issoire, are located 15 km to the NW, and 14 km to
the S, respectively.
Analytical methods
Conductivity, temperature and pH were determined on
site on a 50 ml aliquot of monthly collections. Electrical
conductivity and water temperature were determined with
a microprocessor conductivity meter WTW LF96 standar-
dized to 208C for the measurement of electrical conduc-
tivity. pH was measured with an Ingold electrode and an
Orion 250 pH-meter calibrated with standard bu€ers (pH
4.01 and 6.87).
All samples were ®ltered through a pre-washed 0.2 mm
Millipore PVDF ®lter with a pre-cleaned Nalgene holder
before analysis. One 250 ml aliquot of the ®ltered rain
sample was placed in a Nalgene polypropylene bottle.
Polypropylene storage bottles were cleaned with pH 2
ultra-pure water (Milli-Q water system plus redistilled con-
Fig. 1. Location map of rainwater collector in the Massif
Central (France). The shaded area corresponds to the
Massif Central. The dotted lines represent the limits of the
air mass trajectories and arrows indicate the main direc-
tions of the air mass movements. The ®rst is W and
mainly of marine origin (Atlantic Ocean) but this air mass
crosses over a part of the Aquitanian deposits and the sili-
cate W part of the Massif Central. The second one is NE-
NW and is both marin (North Atlantic and North Sea)
and continental (Great Britain) and crosses over a more
or less polluted area. The third one is E and totally conti-
nental over polluted areas (Germany and more Eastward).
Lastly, the fourth one is SE-SW and of marine origin plus
a possible origin of natural aerosols from Africa and pol-
lution from Spain
 Chemistry of rainwater in the French Massif Central 943

centrated HNO3) and conditioned with ultra-pure water
for several weeks. Samples were acidi®ed to pH 2 with
redistilled concentrated HNO3 and analysed for 87Sr/86Sr,
and for major and trace elements. A 100 ml aliquot of
each ®ltered rain sample was placed in polyethylene bot-
tles for anion and HCO3 determinations.
Major element concentrations were measured in the lab-
oratory using atomic absorption spectrometry (Ca, Na,
K), ion chromatography (Cl, SO4), and ICP-MS (Al, Mg,
Sr). Accuracy was better than 210% (Ca, Na, K, Cl, SO4)
and 5% (Al, Mg, Sr). Bicarbonate was analysed by HCl ti-
tration and Gran's method. Low values close to 0 were
obtained as observed earlier by Wedraogo-Dumazet (1983)
on rain collected in southeastern France.
Chemical separation and mass spectrometry procedures
for Sr followed the standard method used at BRGM.
Strontium was separated by using a cation exchange col-
umn (DOWEX AG50X8) with 2N HCl as eluant. Total
blanks for Sr are less than 0.5 ng and were negligible with
regard to the total Sr analysed in rain waters. After chemi-
cal separation, 1/5 of each sample was placed on a single
W ®lament and analysed using a Finnigan MAT 262 mul-
tiple collector mass spectrometer. The 87Sr/86Sr ratios were
normalized to 86Sr/88Sr = 0.1194. An average internal pre-
cision of 210  10ÿ6 (2s) was obtained during this study.
The reproducibility of the 87Sr/86Sr ratios measurement
was tested through duplicate analyses of the NBS 987
standard and the mean value appears to be close to
0.710228220 10ÿ6 (2s; n = 110).
RESULTS
Meteorological data
Meteorological data were obtained from the local
oce of the National Meteorological Service
located 4 km from the collector for the period from
April 1994 to March 1995. In April 1994, total rain-
fall was 1.5 to 3 times higher than the average
monthly value for April. There were 23 rainy days.
In May, precipitation was above average in most of
the area, although in some places it was below aver-
age. The irregularity of the rainfall is due to the
stormy nature of the period. In June, rainfall was
for the most part normal, with values slightly below
average in some places. Most of the rain fell in 3 to
5 days. In July and August, rainfall was below aver-
age despite several storms. In September, rainfall
was 2 to 3 times higher than average. In October,
rainfall was above average, especially during the
last 10 days. In November, 60 to 80% of the aver-
age monthly rainfall fell during the ®rst 10 days,
and rainfall during the rest of the month was below
average. In December, the rainfall was far below
average. In January and February 1995, rainfall
was 1.2 to 2 times greater than the monthly aver-
age. The highest total monthly rainfall was
measured in February. In March, rainfall was
below average (by a factor of 1.1 to 1.5).
Average air mass trajectories in Corsica (Loye-
Pilot and Morelli, 1988; Losno et al., 1991), in
Paris (Ja€rezo, 1987) and in the Vosges mountains
(Colin, 1988), and weather patterns (MeÂteÂorologie
Nationale, Centre DeÂpartemental du Puy de DoÃme)
enabled sectorizing air mass origins. Four sectors
can be identi®ed as shown in Fig. 1. The ®rst is
westerly and mainly of marine origin (Atlantic
Ocean) but the air mass passes over the Aquitanian
Basin with sedimentary deposits and the western
part of the Massif Central. The second is NE±NW
and is of both marine (North Atlantic and North
Sea) and continental (Great Britain) origin and
passes over a more or less industrialized area. The
third is easterly and completely continental, passing
over polluted areas (Germany and eastern Europe).
The fourth is SE±SW and is of marine origin, poss-
ibly carrying natural aerosols from Africa and pol-
lution from Spain. Ja€rezo (1987) and Colin (1988),
studied rain events and weather patterns over
France and showed that the W sector is predomi-
nant (52% of rain events) and the E sector never
exceeds 5%. The NE±NW and SE±SW sectors rep-
resented 16 and 27% of the rain events, respect-
ively. It is assumed that there was similar
sectorisation during the sampling period, but due to

Table 1. Field data of rainwater samples collected during the one year survey in the Massif Central. Q represents the
amount of collected water expressed in liters. RFA represents the rainfall amount in mm for each collection period.
Temperature T (in 8C), pH, electrical conductivity (in mS/cm standardized to 208C) of rainwater samples were measured
in ®eld
Sample
reference PSM Sampling period Q RFA T Cond pH
1 31-3-94/22-4-94 12 120 26 40.9 4.29
2 22-4-94/3-5-94 4 40 14.9 9.26 5.1
3 3-5-94/25-5-94 11 110 12.2 9.15 5.55
4 25-5-94/17-6-94 3.7 37 12.3 14.6 6.2
5 17-6-94/28-7-94 9.5 95 24.3 51.6 6.05
6 28-7-94/18-8-94 11 110 19.6 17.5 4.53
7 19-9-94/5-10-94 12 120 8.6 7.06 6.04
8 5-10-94/27-10-94 11 110 9.7 5.36 4.95
9 27-10-94/17-11-94 11 110 11 4.95 5.22
10 17-11-94/16-12-94 0.5 5 0.8 25.4 4.54
11 16-12-94/17-1-95 2.8 28 5.5 24.6 4.56
12 17-1-95/30-1-95 2 20 16.5 5.05 5.05
13 30-1-95/6-3-95 6.5 65 7.1 7.85 5.76
944 P. Negrel and S. Roy

Table 2. Major (Ca, Na, Mg, K, Al, HCO3, Cl, SO4, NO3) and trace (Sr) concentrations expressed in mmole/l, 87Sr/86Sr
ratios in dissolved load of rainwater samples collected during the one year survey in the Massif Central. d.l refers to the
detection limits and n.d refers to non analysed samples
Sample
reference
87
PSM Sr/86Sr 2s HCO3 Cl SO4 NO3 Ca Na Mg K Al Sr
1 0.710025 1.00E-05 nd 49 51 75.8 12.2 39.1 4.4 4.1 d.l 0.033
2 0.711182 1.10E-05 nd 7.1 23.4 16.9 10.7 6.1 2.1 4.1 d.l 0.030
3 0.709481 1.20E-05 26 7.1 16.1 34.7 8 3.9 1.7 2.6 d.l 0.017
4 0.710418 8.00E-06 24 16.9 19.8 d.l 14.7 9.1 2.1 7.2 d.l 0.034
5 0.70927 1.00E-05 160 31 47.9 41 78.8 20 4.2 23.8 1.56 0.121
6 0.709435 1.70E-05 nd 16.9 33.3 30.6 16 4.8 0.8 3.3 0.19 0.021
7 0.709198 9.00E-06 29 8.5 9.4 12.9 16.7 4.3 1.9 2.3 d.l 0.027
8 0.713143 8.00E-06 nd 10.2 7.3 d.l 2 8.7 1.2 2.6 d.l 0.004
9 0.712257 1.40E-05 nd 13.5 4.2 d.l 1 0.4 1.2 13.6 0.04 0.008
10 0.711326 9.00E-06 nd 13 36.6 54.7 9.5 6.5 6.5 3.8 1.52 0.030
11 0.709513 9.00E-06 nd 44.3 21.6 39.4 12 32.2 8.2 1.8 0.44 0.019
12 0.710773 9.00E-06 nd 14.1 6.1 d.l 3.2 8.7 5.8 3.1 0.04 0.019
13 0.709245 1.10E-05 17 22.6 13.2 19.4 5.7 43.1 4.1 2.3 1.00 0.020

the sampling procedure (sampling periods of up to
one month), air mass trajectories for individual rain
events are not available.
Chemical and isotopic data
Thirteen samples, representing a total of 97 l of
water and 970 mm of rainfall, were collected
between March 31, 1994 and March 6, 1995. The
®eld data are given in Table 1, and the water chem-
istry and Sr isotope values are given in Table 2.
During the PSM5 period (17/6/94±28/7/94), con-
ductivity increased to 52 mS/cm, then decreased to 5
mS/cm during the periods PSM8 (5/10/94±27/10/94),
PSM9 (27/10/94±17/11/94 and PSM12 (17/1/95±30/
1/95). There was no correlation between conduc-
tivity and the amount of rainwater collected.
The pH of rain samples ranged between 4.29
(PSM1) and 6.2 (PSM4). The mean value was
5.22 2 0.65. Pristine rainwater is moderately acidic
with a theoretical value of 5.7 (Berner-Kay and
Berner, 1987). The higher values reported here
could be the result of dissolution of windblown
dust with a high CaCO3 content, but the high pH
value does not correspond with the high HCO3 con-
centration. Rainwater can also have a pH below 5.7
due to the presence of natural H2SO4, weak organic
acids, or anthropogenic emission of H2SO4 and/or
HNO3. In this study, 9 values were lower and 4
greater than 5.7.
Concentrations of several elements varied greatly
during the sampling period. Five ions were clearly
predominant. Ca and Na were the most abundant
cations with a mean weighted concentration of 17.7
and 13.99 mmoles/l, respectively. Ca content varied
80-fold (1 mmole/l in PSM9 to 78.8 mmole/l in
PSM5). Sodium content varied 100-fold (0.43
mmole/l in PSM9 to 43 mmole/l in PSM13).
Magnesium was the least abundant of the major
cations with a mean weighted content of 2.8 mmole/
l. Magnesium content varied 45-fold (0.18 mmole/l
in PSM4 to 8.2 mmole/l in PSM11). No anions were
strongly dominant. The mean weighted contents of
Cl (19 mmoles/l), SO4 (23 mmoles/l) and NO3 (26
mmoles/l) were similar. Chloride content varied 6-
fold (7 mmole/l in PSM2 to 44.3 mmole/l in
PSM11). Nitrate content was below detection limit
in four samples (PSM4, 8, 9 and 12). Sulphate con-
tent varied 12-fold (4 mmole/l in PSM9 to 51
mmole/l in PSM1). Strontium ranged between 0.004
mmole/l (PSM8) and 0.034 mmole/l (PSM4).
The ion imbalance between cations and anions in
rainwaters collected in France is well-documented
(Losno et al., 1991; Galloway et al., 1982;
Wedraogo-Dumazet, 1983, Ja€rezo, 1987; Colin,
1988). This imbalance can be large and might have
two possible causes. The ®rst is the lack of NH4
measurements, which can cause a signi®cant cation
de®cit. Ja€rezo (1987) and Colin (1988) have shown
for marine and inland rains collected in France that
this species can make up 40% of the total cations.
The second is that lack of measurements of both
H3PO4 and organic species (acetate, formate, humic
acid . . .) may induce an anion de®cit. The observed
imbalance for anions or cations is, therefore, not a
re¯ection on the quality of the data set but, de
facto, the charge balance cannot be calculated.
The 87Sr/86Sr ratios ranged between 0.709198
(PSM7) and 0.713143 (PSM 8). The ratio increased
and decreased several times between PSM1 (end of
winter) and PSM5 (summer) from 0.7112 to 0.7093.
The isotope ratio was then relatively constant
during the summer (PSM5, 6 and 7: 0.7094, 0.7093,
0.7092). The greatest increase was observed between
PSM7 and PSM8 (beginning of autumn) from
0.7092 to 0.7131. 87Sr/86Sr ratios then decreased
between PSM8 and PSM11 (winter) from 0.7131 to
0.7095, increased in PSM12 (0.71077) and ®nally
decreased in PSM 13 (winter, 0.70925)
 Chemistry of rainwater in the French Massif Central 945

DISCUSSION

Quanti®cation of marine and terrestrial contri-

butions using concentration data
To calculate sea salt (ss) and terrestrial (t) contri-
butions in rainwaters (rw) a Na reference was used
(Hofmann et al., 1974; Rahn, 1976; Bonsang et al.,
1980; Galloway and Gaudry, 1980; Keene et al.,
1986; Suzuki and Tsunogai, 1988), hereafter
referred to as Yref. The following equation is used
to calculate the contribution of the sea salt com-
ponent (ss) with seawater (sw) characteristics:
Xss ˆ Yref  …X=Y †sw …1†

The contribution of the terrestrial component (t)

is the di€erence between the total composition of
rainwater (rw) and the sea salt (ss) contribution:
Xt ˆ Xrw ÿ Xss …2†

To calculate the contribution for a given element

X, the main problem to resolve is the choice of the
reference species (marine or terrestrial). In order to
be able to use an element as a marine or terrestrial
reference, there must be no fractionation during
aerosol formation, transport, or deposition by rain
impaction. Furthermore, the element must be of
exclusively marine or terrestrial origin.
Sodium has normally been used as a reference el-
ement for the sea salt component. Magnesium is
used mainly in coastal or marine rains (Berner-Kay
and Berner, 1987). Chlorine has rarely been used, in
spite of its dominant marine origin, because inter-
action with gas phases can cause non conservative
behavior. It could be used as a reference, however,
if the Cl/Na ratios of rainwater do not di€er from
those of seawater (1.16 mols/mols, Berner-Kay and
Berner, 1987).
In the Na vs Cl plot (Fig. 2a), only 2 samples
(PSM2 and PSM8) lie near the seawater line (sw)
indicating that both Na and Cl in these samples
came from sea salt. Furthermore, the fact that all
of the other rain samples, except PSM13, lie to the
left of the seawater line indicates either a signi®cant
anthropogenic Cl source (automobiles, coal com-
bustion, incinerators, etc., Berner-Kay and Berner,
1987) or a fractionation process between Cl and Na
during rainfall. Galloway et al. (1982) and Stallard
and Edmond (1981) have shown, however, that
fractionation does not occur in many moving mar-
ine aerosol cycles. Chlorine cannot be used as a
marine reference if anthropogenic sources exist, as
illustrated in Fig. 2a. Several studies have shown
that both Na and Mg contents in rainwater can
also be in¯uenced by non marine inputs (Ichikuni,
1978; Keene et al., 1986; Suzuki and Tsunogai,
1988). Depending on the site location and atmos-
pheric characteristics at the time (convective or stra-
tiform situation, changes in the proportions of
Fig. 2. Plot of Cl versus Na (Figure 2a) and of Mg versus
Na contents (Figure 2b), expressed in mmole/l, in rain-
waters collected during the one year survey in the Massif
Central. SW represents the seawater line and most samples
plot above this line
di€erent dust types), Na or Mg might be the best
tracers of sea salt input (Keene et al., 1986). Figure
2b illustrates the relationship between Na and Mg.
Only 4 samples lie on the seawater line (SW). The
rest of the rain samples lie above this line, re¯ecting
the terrestrial in¯uence on the Mg content reported
by various authors (e.g., Berner-Kay and Berner,
1987).
Sodium was used, therefore, as a marine tracer in
rainwater. However, the choice of this element as a
reference must be validated, particularly with regard
to possible terrestrial in¯uence. Indeed, 6 rain
samples have Al contents above the 0.5 ppb detec-
tion limit which indicates the presence of material
of land origin, since Al comes almost exclusively
from terrestrial material (Colin et al., 1987;
Sculdark and Church, 1988; Losno et al., 1988,
1991). In these samples, therefore, the presence of
terrestrial Na is assumed. Colin (1988) and Losno
et al. (1991) used the Al content in rainwater to
determine Na from terrestrial input (t), according
to the following equation:
Nat ˆ Alrw …Na=Al †t …3†
This calculation assumes that fractionation
between Al and Na during the transportation of
aerosols or rainfall is absent, which appears to be
the case (Rahn, 1976; Galloway et al., 1982; Losno
et al., 1991; Ja€rezo, 1987; Colin et al., 1987). The
Na contribution of terrestrial dust has been based
on the elemental Na/Al abundance ratios in shales
946 P. Negrel and S. Roy

(Hofmann et al., 1977; Church et al., 1984) or in
the upper continental crust (Losno et al., 1991).
The Na/Al ratio of the upper continental crust is
approximately 0.42 (Taylor and McLennan, 1985;
Condie, 1993), and the ratio for shales is 0.11
(Condie, 1993). The di€erence between these values
is the result of continental weathering. As soils are
the main source of terrestrial material in aerosols,
the crustal source must be representative of soils.
Hofmann et al. (1977) and Church et al. (1984)
have shown that shales are more representative of
average soils than the rocks of the upper continen-
tal crust. Therefore, the correction of crustal Na
with Al has been applied using shales as a terrestrial
dust reference. Sodium in rainwater samples can be
considered to originate from sea salt when Al < 0.5
ppb. If Yt=0 in equation (1), the sea salt and ter-
restrial contributions can be obtained using
equation (2). When Al is di€erent from 0, part of
the Na is derived from a terrestrial source and ter-
restrial Na is calculated with equation (3), using a
Na/Al ratio of 0.11.
The correction of terrestrial Na with Al shows
that less than 1% of Na was of terrestrial origin,
except for PSM10 where terrestrial Na is close to
2.6%. If the upper continental crust is used as the
terrestrial dust reference, there is an input of from
0.2% to 3% (except for PSM10 where terrestrial
Na is close to 9.8%). This shows the small contri-
bution of the terrestrial component in the Na con-
tent. The proportions of sea salt and terrestrial end
members were therefore calculated using equation
(1) and equation (2) with Na as marine reference.
The X/Na ratios for X = Cl, SO4, Ca, Mg, K, Sr
were determined according to the composition of
sea water given by Berner-Kay and Berner (1987).
Results are given in Table 3 in percentage of the
marine source.
The proportion of Cl coming from sea salt ran-
ged from 3.6% in PSM 9 to 100% in PSM 2 and
PSM 8 (Table 3). Values between 33 and 60% are
commonly observed. PSM 13 has a low Cl/Na
ratio, which translates to a marine Cl value of
around 220%. The non marine input of Cl could
come from various sources of pollution including
automobiles, coal combustion, incinerators and fer-
tilizers.
Sulphate from sea salt (Table 3) ranges from less
than 1% (PSM6 and 9) to 19.4% (PSM13). Excess
SO4 in rain over unpolluted land in Amazonia has
been estimated at around 5 mmol/l (Stallard and
Edmond, 1981). Excess SO4 over land commonly
ranges from 10 to more than 100 mmol/l and high
values can be related to pollution (mainly due to
burning of fossil fuel, Berner-Kay and Berner,
1987). In the rain collected in the Massif Central,
4.1 mmol/l to 49 mmol/l of excess sulfate was
measured. Two samples have SO4 values similar to
those measured over unpolluted areas (PSM 9 and
12). PSM1 (31/3/94±22/4/94) had the highest SO4
content (48.7 mmol/l) and PSM5 (16/6/94±28/7/94)
also had a high value (46.8 mmol/l). In the spring,
excess SO4 was between 15 and 25 mmol/l. SO4
increased during the early summer and generally
decreased after October.
Nitrate in rain comes from both human activities
(like fuel combustion) and chemical reactions in the
atmosphere (Berner-Kay and Berner, 1987), the lar-
gest source of NO3 in rain being the production of
N oxides during fossil fuel combustion. In rainwater
collected in the Massif Central, the positive trend
between excess SO4 and total NO3 re¯ects the input
of pollutants from fossil fuel combustion over the
studied period (Fig. 3). Chlorine input from the pol-
lutant sources cannot be linked to that of SO4 and
NO3 as indicated by the lack of correlation between
SO4 and NO3, and Cl (not shown).
Marine Ca values were low (<1.4%) for almost
the whole data set (PSM2-7, 9, 10, 12; Table 3),
similar to results reported by Berner-Kay and
Berner (1987). In PSM 11 and 13, 4±6% of Ca was
of marine origin, and 7 and 9.5% of Ca was of

Table 3. Results expressed in % of Cl, SO4, Ca, Na, Mg, K and Sr of the marine contribution in rain samples. Samples
with an asterisk have Na corrected using Al for crustal inputs
Sample
reference
87
PSM Sr/86Sr t Cl m SO4 m Km Na m Ca m Mg m Sr m
1 0.71027 93.2% 4.6% 21.0% 100% 7.0% 100.0% 22.5%
2 0.71126 100.0% 1.6% 3.3% 100% 1.2% 33.5% 3.9%
3 0.70950 64.3% 1.5% 3.4% 100% 1.1% 26.2% 4.3%
4 0.71048 62.9% 2.8% 2.8% 100% 1.4% 50.2% 5.1%
5* 0.70927 72.7% 2.4% 1.8% 97% 0.5% 51.6% 3.0%
6* 0.70945 32.4% 0.8% 3.1% 98% 0.6% 64.7% 4.4%
7 0.70920 59.9% 2.8% 4.2% 100% 0.6% 26.6% 3.0%
8 0.71625 99.9% 7.2% 7.5% 100% 9.5% 79.7% 43.9%
9* 0.71229 3.6% 0.6% 0.1% 96% 0.1% 3.8% 1.0%
10* 0.71141 52.9% 1.0% 3.4% 90% 0.8% 10.2% 3.8%
11* 0.70967 84.3% 8.9% 39.3% 99% 4.2% 44.0% 31.3%
12* 0.71092 71.8% 8.5% 62% 100% 1.2% 17.0% 8.5%
13 0.70930 220.4% 19.3% 41.2% 99% 5.7% 100.0% 40.5%
 Chemistry of rainwater in the French Massif Central
Fig. 3. Plot of the excess SO4 versus total NO3 in rain-
waters (all expressed in mm/l)
marine origin in PSM 1 and 8, respectively. The Ca
in rainwater from the Massif Central is mainly of
terrestrial origin, coming from the dissolution of
CaCO3 dust (Duce et al., 1980; Prospero et al.,
1981; Maenhaut et al., 1983; Uematsu et al., 1983;
Ichikuni, 1978; Wagner and Steele, 1987; Hofmann
et al., 1977). Calcium can also be produced by pol-
lution (burning coal, cement factories), but the lack
of correlation (not shown) between crustal Ca and
excess SO4 indicates that Ca comes mainly from
soil dust while excess SO4 comes from a pollutant
source (Pearson and Fisher, 1971).
Four to 100% of Mg in the samples was of mar-
ine origin (Table 3). The values are highly variable.
While 100% of Mg was derived from sea salt in
PSM 1 and PSM 13, other values are very low,
3.8% in PSM 9, 11% in PSM10, etc. These low
values for marine Mg show that, like Ca, Mg can
come from soil dust (Berner-Kay and Berner, 1987).
Marine K (Table 3) ranged from less than 1%
(PSM9) to 41.2% (PSM13). Values of less than 7%
were commonly observed in the data set. There are
various non-marine origins of K in continental rain,
including soil dust from silicate and calcareous soils
(Berner-Kay and Berner, 1987). This dust can orig-
inate far from the collector (Saharan soil dust for
example, Losno et al., 1991; Ja€rezo, 1987; Colin,
1988). Another source is agricultural soil dust with
fertilizer (Berner-Kay and Berner, 1987). A third
source are biogenic aerosols, even in temperate cli-
mates (Graustein, 1981). Plant exudates on leaf sur-
faces contain soluble salt (K2SO4) that can
contribute K and SO4 to rain. However, the lack of
correlation (not shown) between K and excess SO4
indicates that this is not the primary source and K
from anthropogenic sources does not have the same
origin as SO4.
The marine end member accounted for between
1% (PSM9) and 43.9% (PSM8) of the Sr content
(Table 3). However, values close to 3 to 8% were
commonly observed for the sea salt contribution
(PSM2-7, 10) and the terrestrial source was always
dominant in these rains. Values of 25 to 45% were
measured in PSM 1, 8 and 11, similar to values
947
measured in Central Africa (NeÂgrel, 1992) but with
a smaller range than those measured by Herut et al.
(1993) in Israeli rains.
Based on the study of elemental composition,
some elements (e.g., Cl, SO4 and NO3) are de®nitely
linked to anthropogenic sources. Calcium comes
predominantly from a natural terrestrial source,
and a marine source may or may not be signi®cant
for Mg, K and Sr. However, major element chem-
istry cannot precisely characterize the various ter-
restrial end members. Therefore, Sr isotope
compositions were used to obtain additional infor-
mation concerning the di€erent possible terrestrial
sources and their respective proportions.
Strontium isotope systematics
When considering an ideal mixing model between
sea salt and terrestrial dust, Sr isotope systematics
enable us to calculate the terrestrial isotope ratio
(87Sr/86Sr)t in each rain water sample. The method
was developed by Herut et al. (1993) for Israeli rain
water. The binary mixing equation for these two
end members is (Faure, 1986):
…87 Sr=86 Sr†rw ˆ a…87 Sr=86 Sr†sw ‡ …1 ÿ a†…87 Sr=86 Sr†t
…4†
where (87Sr/86Sr)rw is the measured isotope ratio in
rainwater, (87Sr/86Sr)sw is the isotope ratio of sea-
water (0.70917 2 0.00001, Hoddel et al., 1990)
which is the same in all oceans because of the high
Sr residence time in seawater, and (87Sr/86Sr)t is the
crustal isotope ratio. d is Srmarine =Srterrestrial ‡
Srmarine (determined in equations 1±3).
We thus obtain:
…87 Sr=86 Sr†t ˆ f…87 Sr=86 Sr†rw ÿ a…87 Sr=86 Sr†sw g=…1 ÿ a†
…5†
The terrestrial ratios in the rain samples vary sig-
ni®cantly (Table 3), ranging from 0.70920 (PSM7)
to 0.71625 (PSM8), with most values falling
between 0.709 and 0.711. This ratio is an indication
of terrestrial aerosol heterogeneities because isoto-
pic variations in the crustal source mean that di€er-
ent terrestrial components with di€erent isotope
ratios are represented in the mixture.
Limestone has a high Sr content (600±1000 ppm)
and low 87Sr/86Sr ratio (0.706±0.709) (Faure, 1986).
The 87Sr/86Sr ratios of CaCO3 carbonate are similar
to those of the water where deposition took place
because of its very low Rb/Sr ratio (Faure, 1986).
On the other hand, silicate rocks (granite and gneiss
bedrock) have lower Sr contents than limestone and
higher 87Sr/86Sr ratios (Wadleigh and Veizer, 1985;
Faure, 1986; Palmer and Edmond, 1989; NeÂgrel et
al., 1993). Basalt has low isotope ratios (0.703±
0.706) but its Sr content is in the same range as sili-
948 P. Negrel and S. Roy

cate rocks (Faure, 1986). Thus, the 87Sr/86Sr ratio
of silicate dust depends on the age and Rb/Sr ratio
of the parent rocks.
The isotopic composition of Sr in dusts can be
estimated by studying 87Sr/86Sr ratios of river
water. On a carbonate watershed, because of the
congruent dissolution of carbonate rocks, the
87
Sr/86Sr ratio of river water is the same as that of
the rocks. Leaching experiments with double dis-
tilled water of soils from silicate watersheds have
shown that soluble Sr in soils and Sr in the dis-
solved load in rivers, also have similar isotope

Fig. 4. Schematic geological map of France and location map of the di€erent crustal components con-
sidered in this work (1-6)
 Chemistry of rainwater in the French Massif Central 949

ratios (NeÂgrel, 1992). Strontium can be given o€ by
vegetation in a manner similar to that described by
Crozat (1979), Clairac et al. (1988) and Artaxo and
Maenhaut (1990) for K. Vegetation extracts cations
from soil waters, and Graustein and Armstrong
(1983) and Graustein (1981) have shown that water
in soils and plants have identical 87Sr/86Sr ratios,
which are higher than that of Sr in the dissolved
load in rivers. Strontium can also be linked with
anthropogenic emissions and fertilizers (more than
1500 ppm Sr).
The terrestrial end member: nature and origin of the
possible terrestrial sources in rainwater, Sr isotopes
constraints
The sea-salt corrected chemical composition of
rainwater in the Massif Central indicates various
terrestrial components, but no single origin can be
considered. Numerous sources, both local and
remote, natural and anthropogenic, must be con-
sidered. One of the terrestrial components can be
related to calcic particles from soil or limestone
dusts. This input of carbonate particles to the at-
mosphere could originate from Tertiary deposits
(Miocene and Oligocene marls and limestones) near
the sampling site and is called the local carbonate
source (LCS) (Fig. 4). The 87Sr/86Sr ratio of springs
draining these rocks and soils is 0.71102 (Table 4,
and ®rst sampling site in Fig. 4).
The Massif Central is bordered to the N, SW and
SE by marine deposits of the Paris Basin (Jurassic
to Oligocene), the Aquitaine Basin, and the Rhone
River valley (BRGM, 1980), respectively. These
deposits might represent a remote source of carbon-
ate particles in the rain samples, here called remote
carbonate source (RCS), the two principal sources
being the Paris and Aquitaine basins (Fig. 4). The
study of rivers draining only one carbonate facies:
Lower Jurassic (LJ), Upper Jurassic (UJ), Lower
Cretaceous (LC), Upper Cretaceous (UC), Tertiary
(T) in the Paris Basin (Table 4, and numbers 2, 3, 4
in Fig. 4) provided 87Sr/86Sr ratios indicating the
possible sources of dusts (NeÂgrel et al., 1989). Two
®elds were identi®ed. The ®rst included the Lower
Jurassic, the Lower Cretaceous, and the Tertiary
with 87Sr/86Sr ratios ranging between 0.7086 and
0.7094. The second included the Upper Jurassic and
the Upper Cretaceous with 87Sr/86Sr ratios of
0.7081 to 0.7084.
The second end member was related to dust emis-
sions from silicate terrains (clay minerals, mica,
feldspar) from the Massif Central called the local
silicate source (LSS, Table 4, and sampling site 5 in
Fig. 4) and from other granitic areas, here called
remote silicate source (RSS, Table 4, and number 6
in the Fig. 4) bordering the Massif Central
(Margeride, Morvan, Lozere). The Sr isotope ratios
of this silicate end member can be approached
through the dissolved load of rivers draining silicate
basement (granite, gneiss and basalt bedrock) in the
Massif Central and in remote granitic watersheds.
For granite and gneiss basements, the 87Sr/86Sr
ratios range between 0.7135 and 0.717 (NeÂgrel et
al., 1988, 1989; Seimbille et al., 1991; NeÂgrel and
Deschamps, 1996; Table 4) and for basaltic base-
ment the 87Sr/86Sr ratios range between 0.703 and
0.705 (NeÂgrel and Deschamps, 1996).
The last end member is made up of anthropo-
genic input (Berner-Kay and Berner, 1987), such as
fertilizers with an 87Sr/86Sr ratio ranging between
0.7079 and 0.7087 (NeÂgrel and Deschamps, 1996).
No data exist on Sr concentrations or isotope com-
positions of other possible pollutants (automobile
exhaust, coal combustion, incinerators).
Figure 5 shows the distribution of the terrestrial
87
Sr/86Sr ratios in the 13 rain samples in the classic

Table 4. 87Sr/86Sr ratios for the possible terrestrial components in rainwaters collected in the Massif Central encompass-
ing the local carbonate source LCS, the remote carbonate source RCS with carbonate facies from Lower Jurassic (LJ),
Upper Jurassic (UJ), Lower Cretaceous (LC), Upper Cretaceous (UC), Tertiary (T) estimated through the dissolved load
of rivers draining this type of basement. The local silicate source (LSS, granitic, gneissic and basaltic bedrocks) and the
remote silicate source RSS bordering the Massif Central (Margeride, Morvan, Lozere) estimated through river draining
silicate basements and lastly by anthropogenic inputs (F) by fertilizers and carbonate amendments.
87
Identi®cation Type Key code Sr/86Sr
Sedimentary bedrocks LCS
Tertiary 0.71120
Sedimentary bedrocks RCS
Tertiary T 0.70895
Upper Cretaceous U.C 0.70810
Lower Cretaceous L.C 0.70940
Upper Jurassic U.J 0.70845
Lower Jurassic L.J 0.70860
Silicated Basements LSS/RSS
Basalts 0.703±0.705
Granites/gneiss 0.7135±0.717
Anthropogenic Inputs
Fertilizers (n = 3) F 0.7079±0.70832
Carbonate amendments F 0.7087
950 P. Negrel and S. Roy
Fig. 5. Plot of the Sr isotopic ratios for the terrestrial end-
members in rainwaters in the diagram 87Sr/86Sr vs. the
inverse of Sr content. Numbers refer to PSM samples.
RCS1 represents the ®rst remote carbonate sources includ-
ing the Lower Jurassic (LJ), the Lower Cretaceous (LC)
and the Tertiary (T) deposits, RCS2 represents the second
carbonate sources including the Upper Jurassic (UJ) and
the Upper Cretaceous (UC) deposits, the fertilizers (F) are
also represented. RSS represents the remote silicate
sources, LSS represents the local silicate source, LCS rep-
resents the local carbonate source
diagram 87Sr/86Sr vs. the inverse of Sr content
(Faure, 1986). The following possible end members
representing: the local carbonate source (LCS); the
local silicate source (LSS) and the remote granitic
areas (RSS); the part of the RCS represented by the
RCS1 (Lower Jurassic, the Lower Cretaceous and
Tertiary deposits); the rest of the RCS represented
by the RCS2 with the Upper Jurassic and Upper
Cretaceous deposits are also represented. All the
data are scattered between 3 extreme apexes, one
corresponds to the silicate source (LSS-RSS), one
to the local carbonate source (LCS) and the last
one encompasses the remote carbonate source
(RCS1 and 2) and the fertilizer source (F).
PSM3, 6, 7, 11 and 13 lie between the two
extreme end members LSS-RSS and RCS2-F, the
main contribution in this mixing is related to
RCS2-F components. If PSM8, with the most
radiogenic 87Sr/86Sr ratio, is assumed to represent
the LSS-RSS component, more than 85% of Sr in
PSM3, 6, 7, 11 and 13 is derived from RCS2-F
components. However, the similarity between these
components RCS2 and F makes it very dicult to
di€erentiate between these two end members. PSM5
is mainly derived from the RCS1 end member (up
to 90%). The samples PSM2, 9 and 10 lie between
the two extreme end members LSS-RSS and LCS.
Likewise, if PSM8 represents the LSS-RSS com-
ponent, more than 90% of Sr in PSM2 and 10 and
around 75% of Sr in PSM9 is derived from the
LCS component. A ternary mixture LSS-RSS; LCS
and RCS1-RCS2-F is required to explain the terres-
trial 87Sr/86Sr ratio in samples PSM1, 4 and 12. In
this mixing, Sr is also mainly derived from the two
preceeding components LCS and RCS1-RCS2-F.
Summary of the nature and origin of the di€erent
sources in rainwaters
The concentrations of the analysed elements in
the rain samples varied greatly during the sampling
period. The 87Sr/86Sr ratio also ¯uctuated signi®-
cantly. These variations are for the most part re-
lated to variations in the sources. The coupled use
of a major element (mainly Na as marine reference)
and Sr isotope systematics, enabled the calculation
of the 87Sr/86Sr ratio of the crustal end member.
These results show that the crustal source is a mix-
ture of several components. The 87Sr/86Sr ratios
show that carbonate sources predominate in many
samples, indicating the origin of the aerosols, given
the observed isotopic characteristics in the rain-
water from the Massif Central. However, because
this carbonate component is present in the N, SE
and SW of the Massif Central, the origin of the air
masses cannot be determined using Sr isotopes
alone.
AcknowledgementsÐThis work was ®nancially supported
by the BRGM (Bureau de Recherche GeÂologiques et
MinieÁres) research Program S20 (on isotopes in hydrosys-
tems). Thanks are due to Michel Brach for technical as-
sistance in physico-chemical analyses and sample
collection, Dominique Breeze provided chemical analyses,
FreÂdeÂric Daudu developed the automatic rain collector.
We thank the TIMS team managed by C. Guerrot which
provided rapid Sr isotopes analyses. Professor Annie
Michard, C. A. J. Appelo and G. Faure are acknowledged
for their helpful comments on the original manuscript.
Editorial handling:ÐA. Appelo
REFERENCES
Aberg G., Jacks G. and Hamilton P. J. (1989) Weathering
rates and 87Sr/86Sr ratios ratios: an isotopic approach.
Jour. of Hydrol. 109, 65±78.
Anderson P., Lofvendahl R. and Aberg G. (1990) Major
element chemistry, 2H, 18O and 87Sr/86Sr in a snow pro-
®le accross Central Scandinavia. Atmos. Envir. 24A,
2601±2608.
Artaxo P. and Maenhaut W. (1990) Trace element concen-
tration and size distribution of biogenic aerosols from
the Amazon basin during the wet season. Nuc. Ins. and
Meth. in Phys. Res. B49, 366±371.
Berner-Kay E. and Berner R. A. (1987) The global water
cycle. Geochemistry and Environment. Prentice-Hall, 394
p.
Bureau de Recherche Geologiques et Minieres (1980)
Carte geÂologique de la France. 1/1500000. Notice expli-
cative.
Bonsang B., Nguyen B. C., Gaudry A. and Lambert
G. (1980) Sulfate enrichment in marine aerosols owing
to biogenic gaseous sulfur compounds. J. Geophys. Res.
85, 7410±7416.
Church T. M., Tramontano J. M., Sculdark J. R., Jickells
T. D., Tokos J. J. and Knap A. H. (1984) The wet de-
position of trace metals to the western Atlantic ocean at
the mid-Atlantic coast and on Bermuda. Atmos. Envir.
18, 2657±2664.
 Chemistry of rainwater in the French Massif Central
Clairac B., Delmas R., Cros B., Cachier H., Buat-Menard
P. and Servant J. (1988) Formation and chemical com-
position of atmospheric aerosols in an equatorial forest
area. J. Atmos. Chem. 6, 301±322.
Colin J. L., Ja€rezo J. L., Pinart J. and Roulette-Cadene
S. (1987) Sequential sampling of snow in a rural area.
Experimentation and identi®cation of the acidifying
agents. Atmos. Envir. 21, 1147±1157.
Colin J. L. (1988) Variabilite des concentrations des
espeÁces mineÂrales dans les preÂcipitations humides en re-
lation avec l'aeÂrosols atmospheÂrique. Ph.D. thesis, Univ.
Paris VII.
Condie K. C. (1993) Chemical composition and evolution
of the upper continental crust, contrasting results from
surface samples and shales. Chem. Geol. 104, 1±37.
Crozat G. (1979) Sur l'eÂmission d'un aeÂrosol riche en pot-
assium par la foreÃt tropicale. Tellus. 31, 52±57.
Duce R. A., Unni C. K., Ray B. J., Prospero J. M. and
Merril J. T. (1980) Long range atmospheric transport of
soil dusts from Asia to the tropical North Paci®c.
Science 209, 1522±1524.
Dupre B., NeÂgrel Ph., Seimbille F. and Allegre C. J. (1994)
87
Sr/86Sr ratios variation during a rain event. Atmos.
Envir. 28, 617±620.
Faure, G. (1986) Principles of Isotope Geology. John
Wiley.
Galloway J. N. and Gaudry A. (1980) The composition of
precipitation on Amsterdam Island. Indian Ocean.
Atmos. Envir. 18, 2649±2656.
Galloway J. N., Likens G. E., Keene W. C. and Miller
J. N. (1982) The composition of precipitation in remote
areas of the world. J. Geophys. Res. 87, 8771±8786.
Gibbs R. J. (1970) Mechanisms controlling world water
chemistry. Science 170, 1088±1090.
Gosz J. R. and Moore D. I. (1989) Strontium isotope stu-
dies of atmospheric inputs to forested watersheds in
New Mexico. Biogeochem. 8, 115±134.
Graustein W. C. (1981) The e€ect of forest vegetation on
solute acquisition and chemical weathering, a study of
the Tesuque watershed near Sante Fe. New Mexico.
Ph.D Thesis. Yale Univ.
Graustein W. C. and Armstrong R. L. (1983) The use of
87
Sr/86Sr ratios to measure atmospheric transport into
forested watershed. Science 219, 298±392.
Herut B., Starinsky A. and Katz A. (1993) Strontium in
rainwater from Israel, Sources, isotopes and chemistry.
Earth Planet. Sci. Lett. 120, 77±84.
Hoddel D. A., Mueller P. A. and Garrido J. R. (1990)
Variation in the strontium isotopic composition of sea-
water (8 Ma to Present), Implications for chemical
weathering rates and dissolved ¯uxes to the oceans.
Chem. Geol. 80, 291±307.
Hofmann E. J., Hofmann G. L. and Duce R. A. (1974)
Chemical fractionation of alkali and alkaline earth
metals in atmospheric particulate matter over the North
Atlantic. J. Geophys. Res. 8, 675±688.
Hofmann E. J., Hofmann G. L., Fletcher I. S. and Duce
R. A. (1977) Further consideration of alkali and alka-
line earth geochemistry of marine aerosols, results of a
study of marine aerosols collected on Bermuda. Atmos.
Envir. 11, 373±377.
Ichikuni M. (1978) Calcite as a source of excess Ca in
rainwater. J. Geophys. Res. 83, 6249±6252.
Ja€rezo J. L. (1987) Etude du lessivage des aeÂrosols atmo-
spheÂriques par les preÂcipitations en milieu urbain. Ph.D.
Thesis, Univ. Paris VII.
Junge C. E. (1963) Air chemistry and radioactivity. New
York, Academic Press.
Keene W. C., Pszenny A. A. P., Galloway J. and Hawley
M. E. (1986) Sea-salt corrections and interpretation of
constituent ratios in marine precipitation. J. Geophys.
Res. 91, 6647±6658.
951
Lacaux J. P. and Warburton J. A. (1980) The disposition
of silver released from soviet Oblako rockets in precipi-
tations during the hail suppression experiment
Grossversuch IV. Part I, measurements of blackground
and a preliminary seeding test. J. Appl. Meteorol. 19,
771±778.
Likens E. G., Bormann F. H., Pierce R. S., Eaton J. S.
and Johnson N. M. (1977) Biogeochemistry of a forested
ecosystem. Springer-Verlag.
Losno R., Bergametti G. and Buat-Menard P. (1988) Zinc
partitioning in Mediterranean rainwater. Geophys. Res.
Lett. 15, 1389±1392.
Losno R., Bergametti G., Carlier P. and Mouvier
G. (1991) Major ions in marine rainwater with attention
to sources of alkaline and acidic species. Atmos. Envir.
25A, 763±770.
Loye-Pilot M. D. and Morelli J. (1988) Fluctuations of
ionic composition of precipitations collected in Corsica
related to changes in the origins on incoming aerosols.
J. Aerosol Sci. 19, 577±585.
Maenhaut W., Raemdonck H., Selen A., Van Grieken
R. and Winchester J. W. (1983) Characterisation of the
atmospheric aerosol over the Eastern Equatorial Paci®c.
J. Geophys. Res. 88, 5353±5364.
Meybeck M. (1983) Atmospheric inputs and river trans-
port of dissolved substances. Proc. Hamburg Symp.
I.A.H.S publ. 141, 173±190.
NeÂgrel Ph., Dupre B., Seimbille F. and Allegre C. J. (1988)
Quantitative modelisation of di€erential erosion between
cristalline and sedimentary area of a French basin by
isotopic analysis of strontium in river waters. Chem.
Geol. 70, 13.
NeÂgrel Ph., Dupre B., Seimbille F., Birck J. L. and Allegre
C. J. (1989) Erosion on the Seine-Yonne basin studied
with the strontium isotopes. Terra Nova 1, 100.
NeÂgrel Ph. (1992) Utilisation des isotopes du Strontium,
des alcalins et alcalino-terreux pour la deÂtermination des
bilans des eÂleÂments chimiques dans les ¯euves, Apports
atmospheÂriques, alteÂration des roches. Exemple du bas-
sin du Congo. Ph.D Thesis, Univ. Paris 7.
NeÂgrel Ph., Allegre C. J., Dupre B. and LeÂwin E. (1993)
Erosion sources determined from inversion of major,
trace element ratios and strontium isotopic ratio in riv-
erwater, the Congo Basin case. Earth Planet. Sci. Lett.
120, 59±76.
NeÂgrel Ph. and Deschamps P. (1996) Natural and anthro-
pogenic budgets of a small watershed in the Massif
Central (France), chemical and strontium isotopic
characterization in water and sediments. Aquat.
Geochem. 2, 1±27.
Palmer M. R. and Edmond J. M. (1989) The strontium
isotope budget of the modern ocean. Earth Planet. Sci.
Lett. 92, 11±26.
Pearson F. J., Jr. and Fisher D. W. (1971) Chemical com-
position of atmospheric precipitation in the
Northeastern United States. U.S.G.S Water Supply
Paper 1535.
Prospero J. M., Glaccum R. A. and Nees R. T. (1981)
Atmospheric transport of soil dust from Africa to South
America. Nature 289, 570±572.
Rahn K. A. (1976) Silicone and aluminium in atmospheric
aerosols, crust-air fractionation? Atmos. Envir. 10, 597±
601.
Sarin M. M., Krishnaswami S., DilliI K., Somayajulu B.
L. and Moore W. S. (1989) Major ion chemistry of the
Ganga-Brahmaputra river system, weathering processes
and ¯uxes to the Bay of Bengal. Geochim. et
Cosmochim. Acta 53, 997±1009.
Seimbille F., NeÂgrel Ph., Dupre B. and Allegre C. J. (1991)
Geochemical study of granitic watersheds, what kind of
information can the isotopic strontium ratio tool can
provide. Terra Nova 3, 11.
952 P. Negrel and S. Roy
Seimbille F., NeÂgrel Ph., Dupre B. and Allegre C. J. (1989)
Nature of the strontium atmospheric input in the
Parisian Basin. Terra Abstracts 1, 346.
Sculdark J. R. and Church T. M. (1988) The atmospheric
deposition of arsenic and association with acid precipi-
tation. Atmos. Envir. 22, 937±943.
Stallard R. F. and Edmond J. M. (1981) Geochemistry of
the Amazon. Precipitation chemistry and the marine
contribution to the dissolved load at the time of peak
discharge. J. Geophys Res. 86, 9844±9858.
Suzuki T. and Tsunogai S. (1988) Origin of calcium in
aerosols over the Western North Paci®c. J. Atmos.
Chem. 6, 363±374.
Taylor S. R. and McLennan S. M. (1985) The continental
crust, its composition and evolution. Publications ed.
Uematsu M., Duce R. A., Prospero J. M., Chen L.,
Merril J. T. and McDonald R. L. (1983) Transport of
mineral aerosol from Asia over the North Paci®c
Ocean. J. Geophys. Res. 88, 5343±5352.
Wadleigh M. A. and Veizer J. (1985) Strontium and its
isotopes in Canadian Rivers, ¯uxes and global impli-
cations. Geochim. et Cosmochim. Acta. 49, 1727±1736.
Wagner G. H. and Steele K. F. (1987) Rain chemistry at a
midcontinent site. U.S.A., 1980±1984. Atmos. Envir. 21,
1353±1362.
Wedraogo-Dumazet B. (1983) Modi®cation de la charge
chimique des eaux au cours du transit aÁ travers 3 eÂco-
systeÁmes distincts du Mont LozeÁre. Ph.D Thesis, Univ.
OrleÂans.