A natron source at Pikrolimni Lake in Greece? Geochemical evidence

Journal of Geochemical Exploration 103 (2009) 133–143
Contents lists available at ScienceDirect
Journal of Geochemical Exploration
journal homepage: www.elsevier.com/locate/jgeoexp
A natron source at Pikrolimni Lake in Greece Geochemical evidence
E. Dotsika a, ⁎, D. Poutoukis b , I. Tzavidopoulos a , Y. Maniatis a , D. Ignatiadou c , B. Raco d
a Lab. of Archaeometry, NCSR “Demokritos”, 153 10 Aghia Paraskevi, Attiki, Greece
b General Secretariat for Research and Technology, 10-14 Messogion, Athens, Greece
c Archaeological Museum of Thessaloniki, 6 M. Andronikou Street, Thessaloniki, Greece
d Institute of Geosciences and Earth Resources, Via G. Moruzzi 1, 56124 Pisa, Italy
article
info
abstract
Article history:
Received 2 February 2009
Accepted 6 August 2009
Available online 21 August 2009
Keywords:
Non-marine salt
Natron
Brine
Alkaline water
Lake Pikrolimni
The geochemical conditions that are responsible for the formation of “Chalastraion nitron” in the basin of
Lake Pikrolimni (Northern Greece) were investigated in this study. The goal of the study is to confirm Pliny's
description: “at Clitae in Macedonia it is found in abundance the best, called soda of Chalestra…” The ionspecific
interaction model based on Pitzer's equations showed that the lake brine samples are saturated
versus carbonate–bicarbonate sodium and sodium–chloride–sulfate minerals. The results of individual X-
Ray Diffraction (XRD) analyses on the evaporative deposits showed that the salts consist mostly of trona,
burkeite, thenardite and halite confirming Pliny's description.
© 2009 Elsevier B.V. All rights reserved.
1. Introduction
During the last decades, the origin of natron (Na 2 CO 3 ) has been a
major issue among the researchers of ancient glassmaking. Natron
was used during the Greek–Roman period as a source of soda, which is
one of the three basic components of glass. Its use was located in
Eastern Mediterranean, Egypt and Phoenicia, from the Bronze Age
until the medieval time. In Mesopotamian glassmaking, the main
source of soda was halophyte ash. The glasses made of plant ash have
different composition in relation to those made of natron, so it is
practically possible to distinguish the West Asian glass products from
those from Mediterranean areas. Glasses from Archaic, Classical and
Hellenistic times which have been found in Greece are all made of
natron (Ignatiadou, 2002).
Natural natron deposits, created by evaporation and desiccation
processes in isolated lakes, exist in Wadi Natrun (De Cosson, 1936;
Coulson and Leonard, 1979; Henderson, 1985), northwest of Cairo,
Egypt. This source was used in antiquity but also until today and it is
considered the main or even the only source of natron for glassmaking.
Natron deposits also existed in Greece, in the region of Macedonia.
Natron is referred in literature, with earliest the quotation of Plato,
who refers to it as Chalastraion, used as detergent (Testimonium 1).
Education, he says, must be like an indelible dye that cannot be
washed out by detergents. It must not be affected by pleasure, which
is stronger than the most powerful chalastraion. One of his
commentators added that “Chalastra is a city in Macedonia and a lake
⁎ Corresponding author. Tel.: +30 2106503305; fax: +30 2106519430.
E-mail address: edotsika@ims.demokritos.gr (E. Dotsika).
where the Chalastraion nitron is formed or dissolved over a period of
nine days” (Testimonium 2). Others mention Chalastra as a city or lake
in Macedonia where natron was being burned (Testimonium 3) or
natron as Chalastraion nitron, named after Chalastra, the lake in
Macedonia (Testimonium 4).
A relatively unknown quotation in international studies about
ancient glass is that of Pliny, the Elder, which includes the most
thorough and enlightening reference to Chalastraion nitron, referring
that it abounds is the Macedonian city of Clitae (Testimonium 5). In
this quotation, the following are mentioned:
“At Clitae in Macedonia it is found in abundance the best, called
Chalastraion, white and pure like salt. There is an alkaline lake
there, with a little spring of fresh water rising up in the center.
Nitron forms there when the Dog Star rises for nine days, ceases
for nine days, comes to the surface again and then ceases”.
Additional characteristics of the lake follow and the author continues
referring to the other famous nitron, the one from Egypt. “Here it is
natural, but in Egypt it is made artificially and in much greater
abundance but of inferior quality, because it is darker and stony”.
Based on the reference of Pliny and other newer literature sources,
Hatzopoulos and Loukopoulou (1989) suggested that the lake
Chalastra is the modern lake Pikrolimni.
The whole issue is of great importance in studies of glassmaking, as
glass products were abundant in Macedonia during the period of
Pliny. The possibility that Chalastraion nitron was used for their
manufacture can lead to the determination of the origin of the
products, both Roman and pre-Roman.
0375-6742/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.gexplo.2009.08.003

134 E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
Lake Pikrolimni is located 20 km to the NW of the town of
Thessaloniki, in the region of Macedonia, in Northern Greece, where a
spa and installations for mud baths exist.
In order to confirm Pliny's description, we investigated the
geochemical data of the area and the evaporative conditions, which
are responsible for the formation of “Chalastraion nitron”.
2. Natron use
Evaporitic rocks have been used since the dawn of human history.
Salt and brines have been used in all cultures, i.e. Knossos Palace is built
on Crete Island with big blocks, some of which were pure massive
gypsum. Also in more “modern” civilisations, the salt was used for
payments. The Romans paid a “salarium” in NaCl and therefore the
actual word “salary” came up for wages, originally meaning salt money.
Ancient use of sodium carbonate is known since 5500–4000 BC.
However, findings of the material are rare. In general, the main use of
natron was as a purifier in mummification. Herodotus provides the
best description of the use of natron in mummification.
Natron seems to have been used in glassmaking in Egypt since the
Badarian Period (early 4th millennium BC), indicated by the very low
potash contents of glasses (Shortland et al., 2006). The use of natron
becomes apparent since the 1st millennium BC. Reported glass from the
tomb of Nesikhons in Egypt with low content in potash, magnesia and
lime indicates that the soda was derived from natron (Schlick-Nolte and
Werthmann, 2003). Other evidence of natron use in the same period is
the blue glasses from 8–9th century BC Nimrud, Iraq (Reade et al., 2005).
The Roman glassmaking industry seems to have been based on natron,
indicated by the low-magnesia content. Until the 9th century AD, natron
continued to be the basic material in glassmaking in Levant, the
Mediterranean and Europe. The use of natron as the flux in glass
production declines in Near East and in Europe in the 9th century AD,
when soda-rich plant ash replaced it (Shortland et al., 2006). However,
at al-Barnuj in the Western Delta, Egypt, natron was collected during
periods that waters were dried up until the 18th century AD (Martin and
Sauneron, 1982). From the 1920s and onwards natron was collected at
the same area by the Egypt Salt and Soda Company (Evelyn-White and
Hauser, 1926–1933).
In ancient Greece natron was mentioned for the first time by Plato
in the 5th century in Republic (Testimonium 1).
Today, sodium carbonate is formed in the subtropical region of
Lake Natron in Tanzania and Lake Magadi in Kenya (Magadi means
bitter, exactly as the name of Lake Pikrolimni, which means bitterlake).
In Table 2, the mineral sequence that precipitates during marine
water evaporation and lacustrine water evaporation appears (Lake
Natron, Lake Magadi, Deep Springs Lake and Wadi Natrun).
3. Meteorological and geological setting
Lake Pikrolimni is located in the basin of Kilkis plain, near Thessaloniki
(23 km), in northern Greece. It is a small shallow lake which
usually dries out during summer. It has an average depth of about 0.5–
0.7 m and covers an area about 4.5 km 2 when it is flooded. The mean
annual precipitation in meteorological station of Thessaloniki (during
1959–1997, Hellenic National Meteorological service) is 448 mm, while
the temperature ranges from 1.3 to 32 °C for the same period. Mean
temperature during July and August is 26.3 °C and mean precipitation is
22 mm (1959–1997, Hellenic National Meteorological service). The
annual real evapotranspiration for Thessaloniki area (Thermi) in 2005
was estimated 482.01 mm, with precipitation 481.8 mm and mean
temperature 15.4 °C (Anatoliki AE, 2006). During summer months, July
and August (2005), mean temperature was 22.3 °C and 25.5 °C
respectively, mean precipitation was 95 and 48 mm respectively while
real evapotranspiration was 95 and 48 mm for the same months
(Anatoliki AE, 2006). Also Dalezios et al. (2002) report that the maximum
values of reference evapotranspiration rates are observed mainly over
the northern region and particularly over the plains surrounding the
stations of Thessaloniki and Kilkis in the majority of months. This could
be merely due to the fact that mainly in summer months strong winds
prevail over Greece, which along with the lack of precipitation cause a
great deal of aridity variability among different sites. In particular, the
observed maximum values of reference evapotranspiration rates in the
summer months over the Thessaloniki plain could be due to a strong and
dry N–NW wind pattern named Vartharis.
Sediments of Neogene and Quaternary age are prevalent in the
region of Lake Pikrolimni (Fig. 1). The Quaternary deposits are
Holocene river and lake sediments, alluvial weathering material
composed mainly of schists and carbonate sandy clays, sands and
gravels of lower terrace system (carbonates). Pleistocene deposits
contain gravels, red clays and sands with calcareous concretions and a
big amount of mica. The prevalent Neogene formation (Mountrakis,
1985) is the sandstone–marl series, including Hipparion mediterraneum,
Mastodon sp., Gazella cf. deperdita.
The bedrock in the region consists of Mesozoic rocks, mainly by
conglomerates from Upper Jurassic, including Nerineae, Cladocoropsis sp.
and Pseudocyclammina sp. (Mercier, 1966), with quartzite, greywacke
(quartz and feldspars), quartz and limestone pebbles, basic rocks
(gabbro) from Upper Jurassic with pyroxene, altered into biotite and
actinolite, hornblende, plagioclase feldspar and epidote and limestone
from Middle–Upper Triassic. In a distance less than 10 km from Lake
Pikrolimni, the Paleozoic basement is uncovered consisting of maficrocks
(amphibolite with amphibole as hornblende) and plagioclase feldspars.
4. Sampling and analysis
Lake Pikrolimni is not fed by any major river and the hydrography
of the area shows that no perennial streams enter the lake directly. In
the past, the major source of water to the lake was ground water
(Pliny; Many public testimonies vouch that there were a lot of springs
in the middle of the lake, but after the big earthquake of 1981 these
springs have disappeared). Specifically, there were springs very close
to the lake and within the lake itself. These percolating groundwaters
probably brought the necessary ionic charge into the lake. So during
the summer, according to Pliny, the precipitation of the salt was
taking place. Actually, the main input of fresh water to Pikrolimni is
the rain. Also, a number of wells and two springs are located in the
lake area (one spring is located in the south perimeter of the lake,
nr. 21 in Table 1 and one spring emerges in the Pikrolimni Village).
The spring water (nr. 21 in Table 1), ranges in temperature from 28 to
30 °C and does not vary in discharge and concentration. It emerges in
the ancient basin of the lake and feeds the lagoon which is perennially
wet marginal lake areas. During the winter, when temperature is low,
the inflow of water (mainly rain water) into Lake Pikrolimni exceeds
the evaporation and the lake fills with water. During the summer,
when the temperature and evaporation increase, the level of the lake
begins to fall, the concentration of the ions increases and evaporite
minerals begin to precipitate out of the solution. During some
summers, when the lake water is exposed to very intense evaporative
concentration, the brines cover all the bottom of the lake. This brine
body vanishes during the dry seasons exposing salt deposit, as shown
in Fig. 1. In the next year, during the rainy seasons, the lake refills with
meteoric water and this water dissolves the salts and the shallow
brine body covering the main lake surface. This sequence of filling/
depletion of the lake was observed during the five years of this study.
Therefore, evaporative concentrations dominate the chemical composition
of the lake water and it is the driving force for the evolution of
Pikrolimni brines, although only one segment is observed directly: the
lake brines. The spring is postulated as sub-surface flow.
Samples were collected, in different times, from the borehole in
the thermal spa of Pikrolimni (which is located at the banks of Lake
Pikrolimni), from spring water and samples of brine and salts from the
lake itself. We also sampled fresh water of the region. The depth of the

E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
135
Fig. 1. Geological map of the area of Lake Pikrolimni and sampling sites (•).
perforation of the borehole in the thermal spa is approximately 250 m.
This water is naturally sparkling, with a metallic aftertaste and a slight
organic smell. Stratigraphic data are absent for the area, however
reports of the borehole at the thermal spa mention that across the
totality of the borehole's depth, only lake sediments were drilled.
The samples were taken twice during the year: in summer and in
winter (2002, 2003, 2004, 2005, 2006 and 2007). The samples were taken
in different seasons. Samples with codes PL-8/2002 and PL-9/2006 regard
residual brines on the lake bottom. The lake was totally dried out once, on
2002, and never again until 2006. The samples from 2003 to 2005 concern
mixing between the residual brines and meteoric water. The analytical
scheme includes field measurements of temperature, conductivity and pH.
Samples were also subjected to chemical analysis. Filtered
(0.45 µm), acidified (with HNO 3 1:1) water samples were collected
for determination of cations (Ca 2+ ,Mg 2+ ,K + and Na + ). Untreated
samples were collected for analyses of anions (Cl − ,SO 4 2− ,CO 3 2− and
HCO 3 − ). The major chemical constituents were analyzed with
standard methods described in Apha (1989). The anions and cations
of the water samples were analyzed with ion chromatography and
atomic absorption.

Table 1
Chemical compositions (mmol/L) of waters from boreholes, springs, water tubes and lake.
a/a Sample name Sample description Date T (°C) pH Cl −
(mmol/L)
SO 4
2−
(mmol/L)
1 S/1/2002 Spring–Pikrolimni Vil. 1/2002 18 9.8 3.81 1.23 3.20 5.61 0.08 0.12 14.78 0.08 0.15 Na–HCO 3
2 b/2/2002 Private borehole–Pikrolimni Vil. 1/2008 19 8.7 1.69 0.18 0.47 3.43 0.13 0.70 4.35 0.13 0.00 Na–HCO 3
3 b/3/2002 Private borehole–Pikrolimni Vil. 1/2002 19 8.6 0.59 0.38 0.87 10.61 0.48 1.03 9.78 0.13 0.00 Na–HCO 3
4 b/4/2002 Private borehole–Pikrolimni Vil. 1/2002 18 8.4 4.01 0.30 0.87 22.26 0.55 10.66 5.52 0.23 0.00 Mg–HCO 3
5 b/A/2002 Borehole–Pikrolimni Vil. 2002 21 6.6 0.79 0.11 0.80 5.20 1.05 1.93 1.26 0.03 Mg–HCO 3
6 wt/B/2002 Water tube 2002 22.5 6.8 3.70 0.30 51.00 7.60 11.60 15.43 0.18 Mg–HCO 3
7 b/B/7/2004 Borehole 2004 15.5 6.3 2.60 0.21 48.52 8.28 14.44 15.30 0.46 Mg–HCO 3
8 b/B/8/2004 Borehole 8/2004 22 6.6 3.10 0.27 55.25 9.78 14.73 17.35 1.82 0.16 Mg–HCO 3
9 b/G/8/2002 Borehole 2002 21 6.6 3.19 0.23 53.00 3.85 16.17 15.87 0.20 Mg–HCO 3
10 b/G/9/2002 Borehole 2002 18 6.6 3.02 0.24 55.51 3.98 16.67 14.57 0.18 0.18 Mg–HCO 3
11 b/G/2003 Borehole 2003 18 6.8 3.67 0.28 48.11 3.80 16.05 14.78 0.18 0.29 Mg–HCO 3
12 b/G/2004 Borehole 2004 18 6.3 3.05 0.28 41.72 7.33 10.78 15.17 0.28 Mg–HCO 3
13 b/G/7/2004 Borehole 7/2004 19 7.7 1.44 0.92 6.20 1.63 2.18 2.70 0.33 1.48 Mg–HCO 3
14 b/P1/7/2004 Borehole 7/2004 18 6.9 3.07 0.26 46.56 11.03 15.19 14.39 0.26 0.16 Mg–HCO 3
15 b/SPA Borehole/SPA 3/2007 18.2 6.1 6.77 0.25 39.49 7.10 12.92 13.83 0.15 Mg–HCO 3
16 wt/SPA Water tube/SPA 7/2004 26.9 6.1 7.42 0.32 38.00 7.23 10.91 15.96 0.20 Mg–HCO 3
17 MW/SPA Municipality water 7/2004 22.1 7.7 8.86 1.30 5.89 3.55 2.18 7.78 0.10 Na–Cl
18 PB Private borehole/Philadelphia 7/2004 18.1 6.9 3.81 1.02 9.89 4.23 2.14 4.83 0.00 Ca–HCO 3
19 MWP-1 Municipality water Pikrolimni 7/2004 23 7.2 5.11 1.49 6.00 2.90 2.30 6.39 0.08 Na–HCO 3
20 MWPh-1 Municipality water Philadelphia 7/2004 27.5 6.9 6.46 1.29 6.49 3.35 1.93 5.17 0.10 Ca–HCO 3
21 S/IGME Spring/near lake 7/2004 28–30 6.5 13.79 1.18 34.89 3.98 5.97 30.74 5.81 Na–HCO 3
21a Keratea a Spring 1977 23 6.5 13.79 1.19 34.90 3.98 5.97 30.70 0.59 Na–HCO 3
21b Pikrolimni-mud a Borehole mud spa 2000/2000 6.1 4.46 0.30 46.10 8.05 10.08 14.48 0.19 Mg–HCO 3
22 PL-8/2002 Lake Pikrolimni 8/2002 27.4 8.5 4146.69 331.98 826.67 609.84 0.09 0.29 5143.43 10.26 Na–Cl
23 PL-1/2003 Lake Pikrolimni 1/2003 22 9.6 170.13 18.20 28.00 14.00 8.83 0.74 246.87 0.51 Na–Cl
24 PL-3/2003 Lake Pikrolimni 3/2003 23 9 72.50 8.13 6.67 20.16 3.88 0.29 104.57 0.26 Na–Cl
25 PL-5/2003 Lake Pikrolimni 5/2003 25 9.2 120.37 14.01 12.00 19.10 6.83 0.45 174.13 0.41 Na–Cl
26 PL-8/2003 Lake Pikrolimni 8/2003 30.6 8.7 247.95 25.52 28.83 26.07 10.73 0.78 340.00 0.59 Na–Cl
27 PL-4/2004 Lake Pikrolimni 4/2004 25 8 241.47 25.63 26.17 43.44 1.35 2.26 412.61 7.83 Na–Cl
28 PL-5/2004 Lake Pikrolimni 5/2004 25.4 9.1 261.78 27.98 27.33 45.57 2.08 2.92 460.87 10.03 Na–Cl
29 PL-7/2004 Lake Pikrolimni 7/2004 28 9.5 499.29 54.58 32.07 44.00 22.53 7.41 723.04 14.14 Na–Cl
30 PL-8/2004 Lake Pikrolimni 8/2004 27.1 8.9 1357.97 150.21 36.83 40.74 13.48 7.94 1745.65 18.72 Na–Cl
31 PL-1/2005 Lake Pikrolimni 1/2005 27.5 9.1 556.56 62.08 30.50 50.90 17.10 6.13 881.30 30.43 Na–Cl
32 PL-9/2006 Lake Pikrolimni 9/2006 27.5 9.1 4100.00 527.92 2795.83 0.03 0.12 6321.74 20.03 Na–Cl
33 PL-3/2007 Lake Pikrolimni 3/2007 23.3 9.2 442.00 47.15 111.00 0.43 1.60 590.96 1.56 Na–Cl
34 PL-6/2007 Lake Pikrolimni 6/2007 34.2 8.9 603.50 68.24 142.00 0.35 1.65 838.57 2.25 Na–Cl
a
Unpublished data, sources in text.
CO 3
2−
(mmol/L)
HCO 3
−
(mmol/L)
Ca 2+
(mmol/L)
Mg 2+
(mmol/L)
Na +
(mmol/L)
K +
(mmol/L)
NO 3
−
(mmol/L)
Type
136 E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143

E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
137
5. Chemical results
5.1. Solute acquisition
Four chemically different water types (Table 1) are recognized in the
Pikrolimni area for diluted water and one type for the lake water. The
four recognized groundwater types, based on both chemical and TDS,
are: Mg–HCO 3 ,Na–HCO 3 ,Ca–HCO 3 ,andNa–Cl type (only one sample).
The two first types represent fresh water that has flown from boreholes
and spring in the old bottom of Lake Pikrolimni. The municipality water
of Pikrolimni village is also Na–HCO 3 type although municipality water
which waters the spa is Na–Cl type. The Ca–HCO 3 represents the aquifer
of local groundwater system in Philadelphia area. The lake waters are
Na–Cl, however when the lake began to evaporate until the evaporating
conditions were such that lake water is nearly totally evaporated, then
the residual water is Na–Cl(CO 3 –SO 4 ) type. Fig. 2 presents the
Pikrolimni area water, groundwater and Pikrolimni brine, in terms of
the major ions. The groundwaters are predominantly HCO 3 , whereas the
cations are mixtures of Na, Ca and Mg. In contrast, the cations of the
brines are dominated by Na and the anions are a mixture. This data is in
agreement with the model proposed by Hardie and Eugster (1970).
The Pikrolimni saline alkaline lake is enriched in dissolved minerals
that have accumulated in the brines following evaporation. Such brines
indicate a very considerable range in ionic composition and concentration.
The total salinity ranges from saline water to brine: total dissolved
solid (TDS) concentrations ranging from 7.5 to 426 g/L. The pH ranges
from 8.8 to 10.5 and increases as salinity increases. The surface water
temperature is between 22 and 31 °C. All the saline lake waters are Na–
Cl type, with Na + and Cl − representing about 95% and 70% respectively
of the total cations and anions. Concentrations of Ca 2+ and Mg 2+ are
very low in the brines of 8/2002 and 9/2006.
The two types of groundwater of Pikrolimni area, Na–HCO 3 ,Mg–
HCO 3 types, are tepid (temperature ranging from 18 to 30 °C), nearneutral
to alkaline (pH ranging from 6.3 to 9.8) and have total
salinities that range from fresh to brackish water: total dissolved solid
(TDS) concentrations ranging from 0.50 to 4.5 g/L. This variability of
TDS is due to the HCO 3 − concentrations.
The dominant ionic species in the Na–HCO 3 (soda) groundwater
are Na + and HCO 3 − –CO 3 2− with these two ions comprising 70–80% of
all ions in solution. The concentration of other major ions is low, with
Mg 2+ and Ca 2+ contents less than 2 mmol/L. The Cl − ,K + and SO 4
2−
concentrations are also low.
In the Mg–HCO 3 type water the dominant ionic species are Mg 2+
and HCO 3 − –CO 3 2− with these two ions comprising 70–80% of all ions in
solution.
The Na–HCO 3 waters were found in the Pikrolimni/Keratea area,
approximately 2 km away from Lake Pikrolimni, and these waters end up
in the basin of Lake Pikrolimni. Sample Keratea (IGME, unpublished data)
belongs in this category, which was measured with SiO 2 =31 mg/L and
CO 2 =668 mg/L. The Mg–HCO 3 groundwater was found within the
Holocene lacustrine sediments of Pikrolimni. Sample Pikrolimni-mud of
2000 (Manasis Mitrakas, unpublished data) belongs in this water type,
which was measured SiO 2 =47 mg/L and CO 2 =1650 mg/L. These
waters, of meteoric origin, flow through the clay minerals, rich in mica,
alternating with carbonate rocks. The Mesozoic and Paleozoic basements
in the area consist of basic–mafic rocks. When the aquifer matrix
is composed of alluvial fill, silicate hydrolysis is the typically invoked
explanation for the formation of Mg–HCO 3 groundwater and Na–HCO 3
groundwater. So it is tempting to assume that this process is controlling
the water chemistry. However, many other weathering reactions
contribute to the solute load of inflow waters. Thus, reconsidering the
facts and data, it could be said that generally rain water and weathering
reactions are the principal solute sources. Rain water contributes to all
principal solutes of dilute waters, excepting silica, but the most
significant contribution will be to Na + , Cl − , SO 4 2− and HCO 3 − . The
amounts of these solutes vary with the distance from seawater and
pollution. Regarding the weathering reactions, one of the most
important is the congruent dissolution of soluble minerals, e.g. gypsum
or halite that can load the solution with very high concentration — this is
a very important mechanism for recycling evaporites — and the silicate
hydrolysis. The alteration of feldspar to clay minerals charges the waters
with Na + ,HCO 3 − and silica. Other silicates provide the additional cations
Ca 2+ , Mg 2+ and K + but such waters are always dominated by
bicarbonate derived by atmosphere or from soil processes. The increase
in Ca 2+ ,Na + ,Mg 2+ versus HCO 3 − from the Pikrolimni/Keratea area
(nr. 1,2,3,19) to boreholes (nr. 4 to 16) near Lake Pikrolimni seems to
support silicate hydrolysis as the weathering reactions that contribute
to the solute load of inflow water. However silicate hydrolysis cannot be
the only control on the water chemistry. Moreover, the critical control
on the precipitation of trona, in Lake Pikrolimni waters, is the relative
amount of Ca 2+ and CO 3
2−
TOT ([CO 3 2− ]+[HCO 3 − ]+[H 2 CO 3 ], where
bracketed symbols refer to concentration) in the water to be
evaporated. According to the conceptual model of Hardie–Eugster
of evaporites formation (Hardie and Eugster, 1970), the first mineral
Fig. 2. Ternary diagrams of water samples from Pikrolimni area. White diamonds: Mg–HCO 3 groundwaters; white circles: Na–HCO 3 groundwaters; white squares: Ca–HCO 3
groundwaters; black diamonds: Pikrolimni Lake brines.

138 E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
to form from evaporating water is calcite, and calcite precipitation
continues until either Ca 2+ or CO 3 2− is exhausted. Waters with a
low proportion of CO 3
2−
TOT relative to Ca 2+ will be carbonate depleted
after calcite formation, and will thus not yield Na-carbonate
minerals upon further evaporation. So the elevated alkalinity/2[Ca 2+ ]
ratio is a main requirement, according to Hardie and Eugster (1970),for
the formation of significant quantities of Na-carbonate minerals.
Also, according to Hardie and Eugster (1970) the appearance of sepiolite
and gypsum determines the evaporative sequence. So, after the sepiolite
precipitation the water can become carbonate-enriched and alkaline
earth-poor or vice versa. If alkalinity is higher than 2[Ca 2+ +Mg 2+ ]
then the sepiolite precipitation is not capable to modify the evolution of
the solution along the path of alkaline facies (Risacher, 1992). The
significantly high proportion of CO 3
2−
TOT relative to 2[Ca 2+ +Mg 2+ ]
in Pikrolimni area waters causes them to move along the path,
in the Hardie–Eugster model, of alkaline faces, forming Na–CO 3
minerals.
However, because the ratio of alkalinity to Ca 2+ and Mg 2+ controls
the types of evaporite minerals formed, and specially the deposition of
trona mineral, it is very important to determine the potential
processes that affected the HCO 3 − contents in these waters.
5.2. Potential processes affected the concentration of HCO 3
−
There are a number of processes that can affect the HCO − 3 contents in
groundwater [these waters are rich in HCO − 3 (about 1.50–3 g/L)].
Eugster (1980) described in detail processes that can affect the
concentration of alkalinity in groundwater: dissolution or precipitation
of carbonate minerals, chemical weathering of silicates, redox reactions,
especially the reduction of NO − 3 to NH + 4 and SO 2− 4 to HS − , microbial
respiration or anaerobic decay and the conversion of CO 2 of deep origin
to HCO − 3 in the aquifer. The dissolution/precipitation of carbonate
mineral is a common control of HCO − 3 contents in groundwater. These
waters show an important over-saturation with respect to calcite
(Parkhurst and Appelo, 1999). In Fig. 3a the relationship Ca 2+ versus
alkalinity (HCO − 3 +2CO 2− 3 ) is shown. For the Mg–HCO 3 -type waters it is
observed that, from Pikrolimni village (Na–HCO 3 ) to boreholes near
Pikrolimni Lake (Mg–HCO 3 ), increase of HCO 3 contents takes place: the
Mg–HCO 3 groundwater that was sampled within the Holocene
lacustrine sediments of Pikrolimni has HCO − 3 contents higher than
38 mmol/L. It is also observed that the HCO − 3 increases with relatively
slight change in Ca 2+ , suggesting that carbonate mineral dissolution/
precipitation reactions are not an important control on HCO − 3 .Alsothe
absence of chloride in the soda and Mg–HCO 3 type waters suggest that
the dissolution of marine carbonates is not the only source of the carbon.
Furthermore, these waters are plotted to the right of the 2(Ca 2+ )=
(HCO − 3 ) equilibrium line (Fig. 3a). If the dissolution of the calcite was the
dominant process producing CO 2 in the system then these waters would
plot along the 2(Ca 2+ )=(HCO − 3 ) equilibrium line.
Chemical weathering of silicates is another major influence on
groundwater. The Na/HCO 3 ratio of dilute Na–HCO 3 inflow is very close
to unity. When accompanied by the absence of Ca–HCO 3 groundwater,
this shows that these Na–HCO 3 waters do not evolve from Ca–HCO 3
groundwater that undergoes normal ion exchange. Therefore the ratio
Na/HCO 3 , very close to unity, suggests that the chemistry of soda water
is controlled by simple weathering reactions like Na-feldspars:
Na–AlSi 3 O 8 þ H 2 CO 3 þ 4; 5H 2 O→Na þ þ HCO − 3 þ 2H 4 SiO 4
þ 1=2Al 2 Si 2 O 5 ðOHÞ 4
:
The hydrolysis of silicate minerals takes place rapidly producing
bicarbonate rich water with high silica content (sample Keratea,
nr. 21a). Probably this fresh water is from the deep cell of the local
groundwater system and probably circulates in the bedrock. On the
contrary, the Na/HCO 3 ratio of dilute Mg–Na–HCO 3 inflow (Fig. 3b)
Fig. 3. Ca 2+ versus alkalinity of all groundwaters (a) and Na + versus HCO 3 − of Mg–HCO 3
groundwaters (b). Same symbols as Fig. 2.
indicates that not only silicate alteration but also more processes are
involved in the chemistry of these waters. The correlation between
HCO 3 − and Mg 2+ indicates that basic rocks become a major target of
weathering reactions in the Mg–HCO 3 − waters. In summary, as referred
by Eugster (1980), the composition of inflow water depends largely on
the minerals present in the watershed: basic and ultrabasic rocks are
probable to give Mg-HCO 3 waters.
Redox reactions are an additional possible control on HCO 3 − in
particular the reduction of NO 3 − to NH 4 + and SO 4 2− to HS − ,bothofwhich
produce CO 2 . So, concerning the reduction reactions as a possible control
of HCO 3 − , although Eh was not measured, the high SO 4 2− and NO 3 − contents
in these waters suggest that these reactions are not likely to be occurring.
Also the microbial respiration or anaerobic decay process (potential
sources of CO 2 that could be converted to HCO 3 − ) appears unlikely to
apply significant control on the ground water chemistry because the
material that would indicate large accumulation of organic matter in the
borehole logs is absent.
The final major reaction that could influence the HCO 3 − contents in
the deep aquifer is the addition in the system of CO 2 of deep origin. The
pCO 2 values of the most Mg–Na–HCO 3 type waters (10 − 0.5 –10 − 1 )
are higher than the atmospheric value (10 − 3.5 ) and the value of the
soil (10 − 2.5 –10 − 1.5 ), suggesting that input of atmospheric/soil is not a
major source and evidencing that gas is being added to the aquifer
from the deep. The highest values are observed for the Mg–HCO 3 type
waters emerging in the Holocene lacustrine sediments of Lake
Pikrolimni. These waters present also the highest Na + ,Mg 2+ and
HCO 3 − contents, from Pikrolimni area (Pikrolimni Village) to Lake
Pikrolimni (borehole near Pikrolimni), suggesting probably that basic
mineral hydrolysis driven by injection of deep (mantle or metamorphism
source) CO 2 is acting to control the concentration of HCO 3 − in
the lacustrine basin.

E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
139
5.3. Brine evolution
The plot of Na + ,SO 4 2− ,CO 3 2− ,HCO 3 − versus Cl − contents of the
borehole sample in the thermal spa of Lake Pikrolimni and the domain of
brine from Lake Pikrolimni are illustrated in Fig. 4. In these diagrams we
also put published data for Lake Natron in Tanzania (Gueddari, 1984)
and Lake Magadi in Kenya (Gueddari, 1984), where the natron actually
precipitates and for Wadi Natrun which was the natron source in
antiquity (Taher, 1999). According to all diagrams, it is concluded that
the chemical composition of Lake Pikrolimni's water is similar to that of
the lakes that precipitate natron today. Such solute concentrations
support that the evaporative concentration is the dominant process in
the chemical evolution of these lake waters. The Na–Cl relation (Fig. 4a)
illustrates in general the stability of Na/Cl ratio in the lake brine. The Na/
Cl ratio of Lake Natron and Lake Magadi, as well as of Wadi Natrun which
was natron source in antiquity, remains stable between diluted waters
and concentrated brines. Comparatively to the waters of Lake Natron
(Na/Cl ratio=3.3) and Magadi (Na/Cl ratio=2), the water of Lake
Pikrolimni presents a lower Na/Cl ratio (Na/Cl ratio from 1.2 to 1.7),
almost two times. The Na/Cl ratios of Lake Natron, Lake Magadi and
Wadi Natrun tend to unity, close to Pikrolimni brines, only in the most
concentrate brines (Gueddari, 1984) showing a very small deviation
from the evaporative trend (from diluted to brine waters), until within
the concentration range appropriate for trona saturation. Once trona
precipitates, the Na + extraction from the solution enriches the brine in
chloride and thus, Na/Cl ratio decreases. In fact in the diagram log Na/Cl
versus Cl (Fig. 4b) the values of diluted waters and brines from Lake
Magadi are given as well as the corresponding data from Lake Pikrolimni.
As precisely referred by Eugster and Jones (1979), “from groundwaters
to all but the most concentrates brines Na/Cl remains constant, because
no fractionation occur. The most concentrated brines are saturated
with respect to trona and precipitation clearly leads to a decrease in
Na/Cl”. In Lake Pikrolini the most concentrated brine (8/2002)
shows a decrease in Na/Cl in relation to the most groundwaters that
are related to trona precipitation. Between groundwater and lake
brines, a decrease of Ca/Cl, Mg/Cl and HCO 3 /Cl ratios is observed
(Table 1). Carbonate species are subject to removal from solution by
different mechanisms. In Fig. 4c, the greater than 1:1 slope of the
regression line for Cl − versus CO 3 2− +HCO 3 − contents illustrates the
gradual loss of carbonate species, accompanying evaporative concentration
of Lake Magadi (Jones et al., 1977). The same is observed for the
waters from Pikrolimni area. This loss is attributed to equilibration with
the atmosphere [2HCO 3 − (aq) → CO 3 2− (aq)+CO 2 (gas)+H 2 O], calcite
precipitation from diluted waters [Ca 2+ (aq)+HCO 3 − (aq) → CaCO 3
(sol)+H + ] and trona precipitation from saturated brine [3Na + (aq)+
HCO 3 − (aq)+CO 3 2− (aq)+2H 2 O → Na 2 CO 3 ,NaHCO 3 ,2H 2 O (sol)]. In
most diluted lake waters, the carbonate loss is probably related to both
degassing and carbonate precipitation. Loss to atmosphere is expected
for pCO 2 greater than 10 −3.5 atm (Jones et al., 1977). In fact, in these
diluted waters the pCO 2 is lower than 10 −3.2 . In the lake brines, the
Fig. 4. Na + versus Cl − (a), log (Na/Cl) versus log Cl (b), Cl − versus CO 3 2− +HCO 3 − (c), and SO 4 2− versus Cl − (d). Black diamonds: Pikrolimni brines; white diamonds: groundwaters;
black circles: Lake Natron springs (Gueddari, 1984); white circles: Lake Natron brines (Gueddari, 1984); crosses: Wadi Natrun (Shortland, 2004); white squares: Lake Magadi
(Gueddari, 1984; Jones et al., 1977); white triangles: Wadi Natrun (Taher, 1999).

140 E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
carbonate loss is probably related to calcium and sodium carbonate
precipitation. An additional mechanism is also associated with efflorescent
crust (Eugster, 1966). These crusts are the products of complete
desiccation of lake water and consist predominantly of soluble salts.
When rain or dilute runoff comes in contact with these crusts only the
most soluble salts, such chlorides, carbonates and sulfates are resolved
into solution and increasing the solute load, while the less soluble phases
like the alkaline earth carbonates and silica remain in the crust. The
dissolution of these crusts leads to high alkaline water with a specifically
lower HCO 3 /CO 3 ratio. The high pH (9.6) of some waters of Lake
Pikrolimni (like PL-1/2003) can be attributed to recycling of residual
brines (PL-8/2002) from the fresh water. The finding of abundant calcite
precipitation in lake beds that are covered by crusts, consisting of trona,
burkeite, thenardite and halite minerals, as shown by the XRD analysis
(Table 2) of Pikrolimni crusts, shows that the above play a competitive
role in the geochemistry processes of Pikrolimni brines. In fact, salt
analyses from different parts of the lake showed that they consist of
carbonates (calcite, dolomite), trona, burckeite and halite. Salts from
Wadi Natrun show similar mineral composition (Shortland et al., 2006).
In the graph of SO 4 2− versus Cl − (Fig. 4d), the molar SO 4 2− /Cl − ratio
indicates generally neither loss nor gain of SO 4 2− between diluted and
brine waters. The relations SO 4 2− versus Cl − suggest a possible
incipient precipitation of a sulphate mineral only for PL-8/2002. All
the other lake waters indicate an excellent conservative behaviour of
sulphate relative to chloride. This also suggests that sulfate reduction
is not a major process at Lake Pikrolimni.
6. Geochemical model
6.1. Thermodynamic calculation model for groundwaters and lake water
An ion-specific interaction model based on Pitzer's equations has been
used (EQL/EVP by Risacher and Clement, 2001) to calculate the saturation
index versus different solid phases, of the Pikrolimni waters and brines.
The lake brine samples are put in ionic strength diagram versus the
saturation index (Fig. 5a, b, c) and in Table 3 the minerals that are
supersaturated in the lakes Pikrolimni, Magadi, Natron and Deep
Springs are given. The calculation of saturation index versus different
carbonate and/or bicarbonate sodium minerals shows that the borehole
waters are saturated in calcium minerals. The lake waters are saturated
in Na/Ca carbonate minerals and only the lake brines are saturated also
in Na–carbonate and bicarbonate, Na–sulfate and Na–chloride minerals.
Indeed, the brines of PL-8/2002 and PL-9/2006 are saturated in
sodium carbonate–bicarbonate minerals like nahcolite, trona and natron.
Τhe calculation of ionic strengths versus different sodium–sulfate
and sodium–chloride minerals show that the brines with Ι>6 are
saturated in halite (NaCl), thenardite (Na 2 SO 4 ) and mirabilite (Na 2-
SO 4 ⁎10H 2 O) confirming that in such lakes, the Ca–Na–HCO 3 salts
precipitate first and the SO 4 –HCO 3 salts follow. The results of
measurement of Na + and Cl − show indeed that only the saturation
in halite, thenardite and mirabilite is achieved.
6.2. Simulation of evaporation of groundwaters
Using geochemical program (EQL/EVP by Risacher and Clement,
2001), it is possible to investigate the mineralogical composition of
the minerals that precipitate from the water, with different initial
chemical compositions, as they evaporate. All sampled waters from
boreholes and the spring were evaporated in order to check if sodium
carbonates could precipitate by evaporation of these waters theoretically.
As shown in Table 4, most waters can give nahcolite (NaHCO 3 )
and thenardite (Na 2 SO 4 ) precipitates in the last phase of evaporation.
Ιn Table 4, the evaporated samples that can give trona and natron
precipitates are shown. In the same table, a diluted water from Lake
Pikrolimni (PL-5/2003) is also shown. It is observed that three
samples, S/1/2002; b/2/2002; PL-5/2003, give Na–carbonate–bicarbonate
minerals [gaylussite Na 2 Ca⁎(CO 3 ) 2 ⁎5H 2 O; pirssonite Na 2-
Ca⁎ (CO 3 ) 2 ⁎ 2H 2 O; trona Na 2 CO 3 ⁎ NaHCO 3 ⁎ 2H 2 O and natron
Na 2 CO 3 ⁎10H 2 O], Na–sulfate minerals (burkeite Na 2 CO 3 ⁎2Na 2 SO 4 ;
thenardite Na 2 SO 4 ) and Na–chloride minerals [halite NaCl]. Water
samples S/1/2002, b/2/2002 and PL-5/2003 give similar minerals
with those found by XRD analysis in collected salt samples but also
similar minerals referred in literature (Shortland et al., 2006; Taher,
1999; Gueddari, 1984). However, only the water sample coded S/1/
2002 and b/2/2002, Na–ΗCO 3 type, could give natron and trona.
Fig. 6 shows the results of S/1/2002 sample when the initial
chemical composition of this water was modeled. As it can be seen, as
the water is concentrated about 250 times of the initial chemical
composition, the first Na–carbonate mineral, gaylussite, deigns to
precipitate. After a concentration of 395 times of the initial chemical
composition, natron, pirssonite, and mirabilite precipitate. The gaylussite
and the pirssonite are among the first salts that precipitateand their
formation is related to calcite. The gaylussite [Na 2 Ca⁎(CO 3 ) 2 ⁎5H 2 O]
generally forms by the reaction of primary calcite (CaCO 3 )andthesodic
carbonate solution, according to the reaction
CaCO 3 þ 2Na þ CO 3 þ 5H 2 O↔Na 2 Ca⁎ðCO 3 Þ 2
⁎5H 2 O:
The pirssonite is usually found in the presence of gaylussite or of
calcite crystals. The direct precipitation of pirssonite takes place
according to the reaction
Na 2 Ca⁎ðCO 3 Þ 2
⁎2H 2 O↔2Na þ Ca þ 2CO 3 þ 2H 2 O:
Secondarily, pirssonite forms either by the dehydration of
gaylussite or by the reaction of Ca–Na solutions with calcite according
to the following equations
Na 2 Ca⁎ðCO 3 Þ 2
⁎5H 2 O↔Na 2 Ca⁎ðCO 3 Þ 2
⁎2H 2 O þ 3H 2 O
Table 2
XRD results on evaporites from different lakes.
Lake Pikrolimni 2002 Wadi Natrun a Lake Natron, Tanzania b Lake Magadi, Kenya b The Deep Springs Lake, California b
Calcite (CaCO 3 )
Dolomite c Calcite (CaCO 3 ) Calcite (CaCO 3 )
Dolomite [Ca(Mg)CO 3 )
Pirsonnite Na 2 Ca⁎(CO 3 ) 2 ⁎2H 2 O Pirsonnite Na 2 Ca⁎(CO 3 ) 2 ⁎2H 2 O
Nahcolite (NaHCO 3 )
Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O) Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O) Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O) Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O) Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O)
Thermonatrite (Na 2 CO 3 ⁎H 2 O), Thermonatrite (Na 2 CO 3 ⁎H 2 O),
Burkeite (Na 2 CO 3 ⁎2Na 2 SO 4 ) Burkeite (Na 2 CO 3 ⁎2Na 2 SO 4 Burkeite (Na 2 CO 3 ⁎2Na 2 SO 4 )
Kogarkoite (Na 2 SO 4 ⁎NaF)
Thenardite (Na 2 SO 4 ) not detected Thenardite (Na 2 SO 4 ) Thenardite (Na 2 SO 4
in 2006 salt
Halite (NaCl) Halite (NaCl) Halite (NaCl) Halite (NaCl) Halite (NaCl)
a Shortland A.J. (2004).
b Gueddari M. (1984).
c Wenigswieser (1992).

E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
141
64% of all salts that precipitate, while halite only 10%. On the contrary,
Fig. 6 shows the results of PL-5/2003 sample when the initial chemical
composition of this water was modeled. It seems that when the model
stops halite is more abundant in the precipitates than trona: trona
represents only 0.5% of all salts that precipitate, while halite 60%.
These two models show the change which may have happened in the
water chemistry of the lake's springs and why trona is so hard to find
in Lake Pikrolimni when the models reaches their lowest point: trona
salts are hoarded by the deposition of large quantities of halite, as
shown by Shortland et al. (2006) in their study about Wadi Natrun.
The alimentation of the lake in the past had happened mainly from
deep springs with similar chemistry while today is arisen mainly from
meteoric waters and irrigation groundwaters (Pliny; Many public
testimonies vouch that there were a lot of springs in the middle of
the lake, but after the big earthquake of 1981 these springs have
disappeared).
The chemical composition of the residual water and of three samples
(S/1/2002; b/2/2002; PL-5/2003) after the simulation of evaporation
was similar with the chemical composition of the lake's water in
conditions of intense evaporation (Table 5).
Especially the chemical composition of the residual water, coded
S/1/2002, after the evaporation simulation, is similar to that of the
lake brine PL-9/2006 (Table 5). Furthermore, the model shows that
natron starts to precipitate when Cl − content is 1417 mmol/L [when
the concentration factor (CF) is about 395 times: 395⁎3.8 mmol/L
(initial Cl of s/1/2002)=1500 mmol/L] while trona starts to precipitate
when Cl − content is approximately 2500 mmol/L [when the
concentration factor (CF) is about 720 times: 700⁎3.8 mmol/L (initial Cl
of s/1/2002)=2651 mmol/L]. The lake water reaches similar concentrations
only during the summer, August (but not every summer), as
observed by sampling. According to Pliny, natron was collected in
Macedonia only in the rising of the Dog Star for nine days. This timing
places this event at the beginning of the hottest part of the summer,
when the lake is found at the peak of evaporation and trona starts to
precipitate. However, the lake does not reach such high concentrations
every year, a fact that is due to the weather (temperature,
humidity, precipitation). Therefore, the collection would have been
taking place only during particularly hot and dry summers.
7. Conceptual model
Fig. 5. Saturation Index (Log saturation ratio) versus ionic strength (logarithmic). White
triangles: nahcolite; x: gaylussite; –: pirssonite; white diamonds: natron; black
squares: trona; white circles: halite; crosses: mirabilite; white squares: thenardite;
crossed x: thermonatrite.
CaCO 3 þ 2Na þ CO 3 þ 2H 2 O↔Na 2 Ca⁎ðCO 3 Þ 2
⁎2H 2 O:
Also when the concentration factor (CF) is about 700 times of the
initial chemical composition, burkeite and trona precipitation initiates.
Much later, when the water is concentrated by about a factor of
1200, halite begins to precipitate. When the models stop, halite and
trona are the more abundant precipitates. Trona is more abundant in
the precipitates than halite: trona (50%) and natron (14%) represent
The Na/Cl ratio in Lake Pikrolimni supports evaporative concentrations
as the dominant process in the chemical evolution of these lake
waters that have high proportions of Na + and HCO 3 − . The derived high
Na + and HCO 3 − contents are due to silicate hydrolysis. The computing
program has shown that the dilute spring inflow was concentrated (by
the assumption of Cl − conservation in solution throughout evaporation)
about 12000 times to obtain the most saline residual actual lake water,
PL-8/2002; PL-9/2006). However, this evaporation budget calculation
(Table 5), if compared with the actual analysis of the most saline
Pikrolimni brine, shows that some constituents are not conservative,
due to different removal mechanisms. Therefore, it is believed that in the
past, the solute composition of such waters, which are controlled by
silicate hydrolysis and evaporative concentration, gave the natron salts
of Lake Pikrolimni. Namely, in the past, after acquisition of solutes from
atmospheric precipitation, hydrolysis of silicate and possibly CO 2 of
deep origin, the waters are subjected to evaporation (at the surface or by
capillarity). Initially, the concentrations of all constituents increase until
the mineral precipitation occurs. The precipitation may take the form of
efflorescent crusts on top of the surface or of intergranural caliche-type
films and cements. The efflorescent crusts are the product of complete
desiccation and they are subjected to complete or differential solution
byrain.ThisrecyclingofsaltsisobservedactuallyatLakePikrolimni.
Thus, during summer when the water level of the lake is low or there is
no water at all (total desiccation), the first rains dissolve the residual
brines (or salts) and get charged by solutes. This recycling of salts is also

142 E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
Table 3
Thermodynamic calculation model: sequence of minerals precipitation.
Lake Pikrolimni 8/2002 Lake Pikrolimni 9/2006 Lake Magadi, Kenya a Lake Natron, Tanzania a The Deep Springs Lake, California a
Calcite (CaCO 3 )
Dolomite [Ca(Mg)CO 3 )]
Calcite (CaCO 3 )
Dolomite [Ca(Mg)CO 3 )]
Calcite (CaCO 3 ) Calcite–Mg [Ca(Mg)CO 3 ]
Dolomite (MgCO 3 )
Gaylussite [Na 2 Ca⁎
(CO 3 ) 2 ⁎5H 2 O]
Gaylussite [Na 2 Ca⁎
(CO 3 ) 2 ⁎5H 2 O]
Gaylussite [Na 2 Ca⁎
(CO 3 ) 2 ⁎5H 2 O]
Pirssonite [Na 2 Ca⁎
(CO 3 ) 2 ⁎2H 2 O]
Pirssonite [Na 2 Ca⁎
(CO 3 ) 2 ⁎2H 2 O]
Pirssonite [Na 2 Ca⁎
(CO 3 ) 2 ⁎2H 2 O]
Nahcolite (NaHCO 3 ) Nacholite (NaHCO 3 ) Nahcolite (NaHCO 3 ) Nahcolite (NaHCO 3 ) Nahcolite (NaHCO 3 )
Natron (Na 2 CO 3 ⁎10H 2 O) Natron (Na 2 CO 3 ⁎10H 2 O)
Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O) Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O Trona (Na 2 CO 3 ⁎NaHCO 3 ⁎2H 2 O
Thermonatrite (Na 2 CO 3 ⁎H 2 O) Thermonatrite (Na 2 CO 3 ⁎H 2 O)
Gypsum (CaSO 4 ⁎2H 2 O)
Glauberite [Na 2 Ca(SO 4 )]
Burkeite (Na 2 CO 3 ⁎2Na 2 SO 4 )
Mirabilite (Na 2 SO 4 ⁎10H 2 O) Mirabilite (Na 2 SO 4 ⁎10H 2 O)
Thenardite (Na 2 SO 4 ) Thenardite (Na 2 SO 4 )) Thenardite (Na 2 SO 4 )
Halite (NaCl) Halite (NaCl) Halite (NaCl) Halite (NaCl) Halite (NaCl)
a Gueddari M. (1984).
Table 4
Simulation of evaporation of groundwater.
Label
S/1_/2002
b/2/2002
b/3/2002
b/4/2002
b/A/2002
wt/B/2002
b/G/2003
b/G/2004
b/P1/7/2004
b/G/7/2004
b/B/8/2004
b/SPA
wt/SPA
MWP-1
MWPh-1
Pl-5/2003
Minerals
BURKEITE_CALCITE_GAYLUSSITE_GLASERITE_HALITE_MAGNESITE_MIRABILITE_NATRON_PIRSSONITE_TRONA
CALCITE_GAYLUSSITE_GLASERITE_HALITE_MAGNESITE_PIRSSONITE_THENARDITE_TRONA
CALCITE_GAYLUSSITE_GLASERITE_HALITE_MAGNESITE_MIRABILITE_NAHCOLITE_PIRSSONITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_SYLVITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_GLAUBERITE_GYPSUM_HALITE_MAGNESITE_MIRABILITE_SYNGENITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
CALCITE_GLASERITE_HALITE_MAGNESITE_NAHCOLITE_THENARDITE
HALITE_PIRSSONITE_THENARDITE_TRONA
In each line, the list of minerals corresponds to the whole sequence of precipitated minerals during concentrative evaporation.
the way of increasing the solute load of the groundwater. Precipitation
and re-solution inevitably lead to strong segregation of the initial
solutes. Therefore, sorption ion exchange and reduction reaction on
mineral surface may also remove certain solute.
8. Conclusions
Lake Pikrolimni is a saline lake that is characterized by alkaline
brine, poor in Ca 2+ and Mg 2+ . The dilute HCO 3 − spring fresh water
(350 g/L dissolved solid). Such brines show a considerable
range in ionic composition and concentration. High solute concentration
due to solar evaporation of water, mineral precipitation,
fractional dissolution and solute recycling are the main processes
responsible for these brines formation. In particular, these evaporating
conditions were such that incurred a hydrogeochemical environment
that was responsible for the lake to provide, in dry periods,
“nitrum chalestricum” (trona). Besides, the progressive concentration
of brines in alkaline lakes leads to a preferential precipitation of
sodium carbonate followed by sulfates and chlorides. This conclusion
comes in agreement with the results. Also, the mineralogical analysis
(X-Ray Diffraction of salts), the evaporation simulation and the
thermodynamic model showed that salts of trona, burkeite and halite,
deposit from these brines. So, the conditions that are responsible for
the formation of soda seem to be present in the basin and confirm
Pliny's description.
Fig. 6. Minerals molarity versus concentration factor.

E. Dotsika et al. / Journal of Geochemical Exploration 103 (2009) 133–143
143
Table 5
Chemical composition (mmol/L) of the residual water, after evaporation, when the models stop.
Label CF alk Cl SO 4 Na K Ca Mg TDS
S/1/2002 12,300 1913 4300 610 6590 845 0 0.02 455,699
b/2/2002 6850 1057 4510 780 6320 820 0 0.02 448,892
PL-5/2003 2700 182 4830 840 5870 830 0.2 0.45 431,224
S/1/2002 chemical composition of residual water when natron starts to precipitate 395 3143 1417 459 5448 30 0.004 0.03 318,000
S/1/2002 chemical composition of residual water when trona starts to precipitate 720 3246 2500 790 7273 53 0.001 0.02 434,000
Testimonia
Testimonium 1. Plato, 5th century BC, Republic 430a (ed. I. Burnet,
1963).
Testimonium 2. Scolia in Platonem, 159 (ed. Ruhnken).
Testimonium 3. Etymologicum Magnum, Halastri.
Testimonium 4. Stephanus Byzantius, Halastra.
Testimonium 5. Pliny, Naturalis Historiae 31, 46 (ed. W.H.S. Jones,
Loeb 1963). Chapter on soda, esp. 106–109.
Acknowledgements
The authors would like to express their gratitude to Mr. Aggelakis
for his help during sampling sessions as well as to Dr. Risacher F. and
Dr. Fritz B. for their contribution of the geochemical program EQL/
EVP. Finally, special thanks go to the two unknown reviewers for their
really contributory remarks and suggestions of the manuscript and to
the editor of this journal for his helpful guidance.
References
Anatoliki, A.E., 2006. Public Participation in Water Resources' Management at a Greek
River Basin. Project Water Agenda, LIFE04/ENV/GR/000099.
Apha, 1989. Standard Methods for the Examination of Water and Wastewater, 19th
end. American Public Health Association, Washington, DC.
Coulson, W.D.E., Leonard Jr., A., 1979. A preliminary study of the Naukratis region in the
western Nile Delta. Journal of Field Archaeology 6.2, 151–168.
Dalezios, N., Loukas, A., Bampzelis, D., 2002. The role of agrometeorological and
agrohydrological indices in the phenology of wheat in central Greece. Physics and
Chemistry of the Earth 27, 1031–1038.
De Cosson, F.C., 1936. El Barnugi. Bulletin de Societe Royale d'Archeologie — Alexandrie
30 (N.S. IX-1), 113–116.
Eugster, H.P., 1966. Sodium carbonate–bicarbonate minerals as indicators of pCO 2 .
Journal of Geophysical Research 71, 3369–3377.
Eugster, H.P., 1980. Geochemistry of evaporitic lacustrine deposits. Annual Review of
Earth and Planetary Sciences 8, 35–63.
Eugster, H.P., Jones, B.F., 1979. Behavior of major solutes during closed-basin brine
evolution. American Journal of Science 279 (6), 609–631.
Evelyn-White, H.G., Hauser, W., 1926–1933. The Monasteries of the Wadi 'n Natrun, 3
volumes, Publications of the Metropolitan Museum of Art Egyptian Expedition 2, 4,
8, New York.
Gueddari, M., 1984. Geochimie et thermodynamique des evaporites continentales, etude
du lac Natron en Tanzanie et du chott El Jerid en Tunisie. Sciences Geologiques
Universite Louis Pasteur de Strasbourg, Institute de Geologie, Mémoire 76, 144.
Hardie, L., Eugster, H., 1970. The evolution of closed-basin brines. Mineral. Soc. Amer.
Spec. Pap. 3, 273–290.
Hatzopoulos, M.B., Loukopoulou, L.D., 1989. Morrylos. Athenes, Centre de Recherches
de l'Antiquite Grecque et Romaine.
Henderson, J., 1985. The raw materials of early glass production. Oxford Journal of
Archaeology 4 (3), 267–291.
Ignatiadou, D., 2002. Research on the Nitrum Chalestricum. Proceedings of Archaeological
Work in Macedonia and Thrace 16, Thessaloniki, Greece.
Jones, B.F., Eugster, H.P., Rettig, S.L., 1977. Hydrochemistry of the Lake Magadi basin,
Kenya. Geochimica et Cosmochimica Acta 41, 53–72.
Martin, M., Sauneron, S. (Eds.), 1982. Claude Sicard. In: Bibliotheque d'etudes, vols. 3. Le
Caire, pp. 83–85.
Mercier, J., 1966. Etude géologique des zones internes des Hellénides en Macédoine
centrale (Grèce). Contribution a l'étude du métamorphisme et de l'évolution
magmatique des zones internes des Hellénides, Thèses, Paris 1966, Ann. geol. Pays.
Hellen., 20, 1–792.
Mountrakis, D., 1985. Geology of Greece. InUniversity Studio Press, pp. 50–72.
Thessaloniki.
Parkhurst, D.L., Appelo, C.A.J., 1999. User's guide to PHREEQC—a computer program for
speciation, reaction-path, 1D-ransport, and inverse geochemical calculation: US
Geological Survey. Water-resources Investigations Report 99, 4259–4312.
Reade, W., Freestone, I.C., St Simpson, J., 2005. Innovation or continuity Early first
millennium BCE glass in the Near East: the cobalt blue glasses from Assyrian
Nimrud. In: Cool, H. (Ed.), Annales du 16e Congres de l'Association Interntionale
pour l'Histoire du Verre-London, 2003, pp. 23–27.
Risacher, F., 1992. Geochimie. Sciences Géologiques Bulletin 45 (3–4), 135–214.
Risacher, F., Clement, Α., 2001. A computer program for the simulation of evaporation of
natural waters to high concentration. Computers and Geosciences 27, 191–201.
Schlick-Nolte, L.A., Werthmann, R., 2003. Glass vessels from the Burial of Nesikhons.
Journal of Glass Studies 45, 11–34.
Shortland, A., 2004. Evaporites of the Wadi Natrun: seasonal and annual variation and
its implication for ancient exploitation. Archaeometry 46, 497–516.
Shortland, A., Schachner, L., Freestone, I., Tite, M., 2006. Natron as a flux in the early
vitreous materials industry: sources, beginnings and reasons for decline. Journal of
Archaeological Science 33, 521–530.
Taher, A.G., 1999. Inland saline lakes of Wadi el Natrun depression, Egypt. International
Journal of Salt Lake Research 8 (2), 149–169.
Wenigswieser, S., 1992. Mineralogische untersuchungen an den Evaporiten und Tonen
des Wadi El-Natrun, Fakulat fur Bio- und Geowissenschaften, Universitat Karlsruhe,
Unpublished thesis.