Journal of Archaeological Science 35 (2008) 1453e1462

http://www.elsevier.com/locate/jas

Methods for calculating brine evaporation rates during salt production

D. Glen Akridge*
Arkansas Archeological Society, 5411 W. Wheeler Road, Fayetteville, AR 72704, USA
Received 11 July 2007; received in revised form 17 October 2007; accepted 22 October 2007

Abstract

Salt is recognized by archaeologists as an important commodity due to the biological need for sodium and other cultural uses. Numerous
studies have described the various techniques used in converting brine to crystallized salt, but few, if any, have attempted to quantify the physical
processes of evaporation in pre-industrial societies. Apart from the few areas where salt mining is possible, nearly all forms of salt production
require evaporation of water to concentrate brine and ultimately produce salt crystals. This study quantifies three of the most common evapo-
ration techniques and provides insight into the production rates of salt and fuel requirements. Methods of calculation are provided for determin-
ing evaporation through (1) direct solar heating of brine, (2) applied external heat to a vessel, and (3) an immersed heated object (e.g., stone).
These results provide physical constraints on the evaporation process and provide investigators with techniques for estimating efficiency and
total production of prehistoric and historic saltworks.
Ó 2007 Elsevier Ltd. All rights reserved.

Keywords: Salt; Evaporation; Brine; Stone boiling; Numerical methods

1. Introduction efficiency and total salt production, except in the few cases
where historic descriptions of saltworks exist (e.g., Chiang,
Saltmaking from brine has been a common worldwide in- 1976). Few have attempted to quantify the process of evapora-
dustry for thousands of years, beginning by at least the fourth tion. Quantifying the process of salt production does more
millennium B.C. in Europe (Olivier and Kovacik, 2006) and than provide estimates for the total salt produced. It provides
by the first millennium B.C. in China (Flad et al., 2005) and insights into the required human labor expenditure and the de-
Central America (Andrews, 1983). Solar evaporation of brine gree to which the local economy depended on salt. The salt
to form salt continues to be a viable commercial process to this trade would have largely been determined by the amount of
day along coastal areas (Kostick, 2002). The procedures used excess salt that could be produced. In the case of brine boiling
in making salt varied by geographic region and resources lo- to obtain salt, the need for substantial quantities of fuel (e.g.,
cally available. The quantity desired by the local population wood) often resulted in serious environmental changes to the
may have also influenced the choice of salt production surrounding landscape (Early, 1993).
methods. Although the process often involved techniques Quantifying the production of salt can be accomplished
such as leaching, extraction, filtering, and burning of salt- through experimentation or by numerical simulation. Experi-
enriched plants (Adshead, 1992), the final step in salt produc- mentation is often advantageous because it can provide
tion invariably required evaporation of water from brine to insights into the subtleties of the process that might not be
precipitate salt crystals. Numerous studies have focused on realized otherwise, but suffers greatly from the need to control
the techniques and archaeological remnants of saltworks. variables. Salt production is a labor intensive and time
These studies have tended to leave open the question of consuming task, and in the case of solar evaporation may
take weeks to obtain salt. This makes experimentation diffi-
* Tel.: þ1 479 790 0261; fax: þ1 479 575 5453. cult, especially in light of the multitude of potential variables
E-mail address: [email protected] such as ambient temperature and humidity, brine volume and

0305-4403/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2007.10.013
1454 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462

concentration, fire temperature, and vessel heat transfer prop- environmental conditions. This technique works best at low
erties that need to be monitored and maintained. In addition, latitudes where sunlight duration and intensity are highest
experimentation results are generally only relevant for the and areas with low relative humidity and rainfall. Solar evap-
tested scenario and provide only limited insight into other un- oration also becomes the default method when fuel resources
tested evaporation conditions. The great utility of numerical are scarce and boiling of brine is unfeasible. Historically
simulations is the speed at which ‘‘experiments’’ can be per- this technique was common in coastal areas (e.g., LeConte,
formed. Simulations of processes taking hours or weeks can 1862) and continues to be a viable commercial process world-
be performed in seconds allowing for greater exploration of wide (Kostick, 2002). Solar evaporation could have been prac-
variable effects and deeper understanding of the physical ticed at many inland salines where brine concentrations tend to
mechanisms underlying evaporation. be high (e.g., Demir and Seyler, 1999) thus reducing total
This paper lays out in detail all the necessary steps in evaporation time.
simulating evaporation in three different scenarios: (1) solar
evaporation, (2) evaporation from an externally heated pan, 2.1. Calculation method for solar evaporation
and (3) evaporation from a hot immersed object. Each scenario
involves evaporation of a fixed volume of brine to dryness Calculating evaporation rates for brine solutions requires
resulting in the precipitation of salt crystals. For any batch a slight modification to the standard Penman (1948) equation
evaporation the amount of salt produced can be determined by used by hydrologists to determine evaporation from open
water sources. The Penman approach combines the effects
ms ¼ mw 1:52  10ÿ4 S2 þ 9:50  10ÿ3 S
ÿ

ð1Þ of radiation and aerodynamic forces controlling evaporation
and has been shown to adequately predict evaporation in
where ms is the mass (kg) of salt crystallized, mw is the mass a wide variety of environments (e.g., Finch, 2001). The Pen-
(kg) of the water evaporated, and S is the initial salt concentra- man equation is generally expressed as:
tion of the brine in wt% NaCl (Fig. 1). For ease of calculation,
the dissolved salt is assumed to be pure sodium chloride. So- D g
lE ¼ Rn þ f ðuÞ ðes ÿ eÞ ð2Þ
dium and chlorine make up 85% of the inorganic constituents Dþg Dþg
found in seawater (Lide, 1991) and natural inland brines are
often of even higher purity (Bowman, 1956). Consequently, where E is the evaporation rate expressed as mm/day, l is the
the presence of other minor components has little impact on latent heat of vaporization, D is the gradient of the vapor pres-
the numerically simulated results. Exceptions may be inland sureetemperature curve, g is the psychometric constant, Rn is
lakes containing high concentrations of sulfates. In these the net solar radiation, f(u) is a function of wind speed, and es
cases, adjustments may need to be made to account for the and e are the saturation vapor pressure of water and ambient
mass, density, and vapor pressure differences of a sulfate- water vapor pressure, respectively. In this paper the Penman
rich brine. equation has been modified to reflect the reduced vapor pres-
sure of a salt solution. A similar approach has been used pre-
2. Solar evaporation viously in determining evaporation from saline lakes (Calder
and Neal, 1984). Additional details of the equation and its var-
In solar evaporation a vessel or ponded area containing iables are discussed by Allen et al. (1998) and Shuttleworth
brine is allowed to evaporate under the prevailing (1993).

20 2.1.1. Calculating aerodynamic terms

18 The aerodynamic forces acting on evaporation are primarily
16
the result of environmental variables governing diffusion of
NaCl Crystallized (kg)

26.2 wt% water molecules away from the liquid surface of water or
14
brine. The modified Penman equation for predicting potential
12
evaporation rates from a free water surface requires knowledge
10 20 wt%
of several local climate variables. The method shown here is
8 for daily calculation steps and uses 24 h averages for temper-
15 wt%
6 ature, humidity, and wind speed. For modeling past saltmaking
4 10 wt% endeavors, historical weather data including average climate
2
conditions for each month can be obtained for a variety of lo-
5 wt%
cales from several sources, including the National Climatic
0
0 5 10 15 20 25 30 35 40 45 50 Data Center in the United States.
Water Evaporated (kg) The latent heat of evaporation l (MJ kgÿ1) varies with tem-
Fig. 1. Graphical representation of Eq. (1). The maximum sodium chloride
perature according to
brine concentration is 26.2 wt% (at 25  C). For comparison, seawater has
a salt content of about 3.5 wt%. l ¼ 2:501 ÿ 0:002361T ð3Þ
 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462 1455

where T is in degrees celsius. The temperature T for daily time emax þ emin
e¼ ð11Þ
steps is simply defined as the mean of the daily maximum and 2
minimum temperatures. The vapor pressure e (kPa) can be determined from the rel-
Tmax þ Tmin ative humidity
T¼ ð4Þ
2 Hr es
e¼ ð12Þ
The saturation vapor pressure es (kPa) is determined from the 100
average daily temperature and is an indication of the rate at
which water molecules can escape from the liquid surface. where Hr is the relative humidity (%). Wind speed is incorpo-
When dissolved salts are present the saturation vapor pressure rated using empirically determined coefficients for the atmo-
is lowered due to the decreased chemical potential of the liq- spheric resistance encountered in diffusion of the water
uid water. To adjust for this lowering effect the water activity vapor away from a liquid surface. For an open water surface
coefficient aw is inserted into the basic equation. the wind function is given by

17:27T f ðuÞ ¼ 6:43ð1 þ 0:536U2 Þ ð13Þ

es ¼ 0:6108aw exp ð5Þ
237:3 þ T
where the wind speed U2 (m sÿ1) is measured at 2 m above the
The activity coefficient of water aw is a function of the con- surface.
centration of dissolved salts. The following correlation was de-
rived from experimental vapor pressure data published for 2.1.2. Determining net radiation
sodium chloride (Lide, 1991). Solar radiation aids in promoting evaporation by impart-
ing energy into the absorbing material. The radiational
aw ¼ ÿ0:0011m2 ÿ 0:0319m þ 1 ð6Þ energy available at the ground surface is a combination of
both short and long-wavelength radiation and is the differ-
where m is the concentration of sodium chloride in moles per
ence between the upward and downward radiation fluxes.
liter of water. Density of NaCl solutions has been measured by
The amount of solar energy reaching the ground surface
Romanklw and Chou (1983). Based on their published data,
can be reduced by cloudiness and atmospheric interferences
the following equation was developed to calculate NaCl solu-
or increased with increasing altitude. This net radiation, Rn,
tion density over the temperature range of 25e45  C and con-
can be determined in three ways: (1) direct measurement
centration range of 0e26.2 wt%.
using a radiometer, (2) published tables based on latitude,
 0:012
25 or (3) calculations incorporating Earth’s orbital characteris-
2:754  10ÿ5 S2 þ 6:872  10ÿ3 S þ 0:99704
ÿ

D¼ tics. Radiometers measure the net radiation by monitoring
Tsol
the temperature difference across two parallel plates. They
ð7Þ require periodic calibration and each measurement locale
must be generally free of any obstructions blocking incom-
where D is the solution density (g cmÿ3), S is the NaCl con-
ing or outgoing radiation. Approximate values for the net
centration in wt%, and Tsol is the solution temperature ( C).
radiation reaching the ground surface can also be found in
The saturation vapor pressure is a function of temperature
published tables (e.g., Lide, 1991). These tables are gener-
and the gradient of this function (kPa  Cÿ1) is also required
ally organized by latitude and indicate the average net
and can be calculated by
radiation for a cloudless sky during each month of the
4098es year. These tables do not usually account for altitude
D¼ 2 ð8Þ differences but are generally accurate to within 10% during
ð237:3 þ TÞ
summer months and 15% during winter months (Lide, 1991).
The psychrometric constant (kPa  Cÿ1) is given by

g ¼ 0:000655P ð9Þ 2.1.3. Calculating net radiation

The following series of equations are required when deter-
where P is the atmospheric pressure (kPa). If the atmospheric mining the net radiation by accounting for Earth’s orbital char-
pressure is unknown, an approximate value can be calculated acteristics. The extraterrestrial radiation Ra (MJ mÿ2 dayÿ1)
based upon a site’s elevation can be calculated for any latitude and day of year by adjusting
 5:26 the solar constant Gsc for the solar declination
293 ÿ 0:0065z
P ¼ 101:3 ð10Þ 24ð60Þ
293 Ra ¼ Gsc dr ½us sinð4ÞsinðdÞ þ cosð4ÞcosðdÞsinðus ފ
p
where z is height above sea level (m). ð14Þ
In a similar fashion to the mean temperature, the daily
mean vapor pressure is the average of the maximum and min- where Gsc ¼ 0.0820 MJ mÿ2 minÿ1, dr is the inverse relative
imum values. EartheSun distance, us is the sunset hour angle (rad), 4 is
1456 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462

the latitude (rad), and d is the solar declination (rad). Angles in temperature during a 24 h period (K), e is the water vapor
radians can be obtained by converting decimal degrees. pressure (kPa), as and bs are regression terms mentioned ear-
lier, Rs is the solar radiation (MJ mÿ2 dayÿ1), and Ra is the ex-
p
Radians ¼ ½Decimal degreesŠ ð15Þ traterrestrial radiation (MJ mÿ2 dayÿ1). The net radiation Rn
180
(MJ mÿ2 dayÿ1) is simply the difference between the incoming
The inverse relative EartheSun distance dr is given by effective solar radiation and outgoing long-wave radiation.
 
2p Rn ¼ Rns ÿ Rnl ð23Þ
dr ¼ 1 þ 0:033 cos J ð16Þ
365

where J is the number day of the year (e.g., 1 for January 1 or 2.2. Results from numerical simulations
365 for December 31). The solar declination d is
  To illustrate the utility of predicting solar evaporation rates,
2p
d ¼ 0:409 sin J ÿ 1:39 ð17Þ two examples from the literature will be briefly discussed here.
365 These examples were selected because sufficient evaporation
The sunset hour angle us is given by variables are specified to allow for numerical modeling of
the described saltmaking process. The first example is an eth-
us ¼ arccos½ÿ tanð4Þtanðdފ ð18Þ nographic description of saltmaking from the western coast of
The number of daylight hours N can be determined by Mexico. The second simulation uses historic descriptions of
19th century Chinese coastal tradition using portable wooden
24
N¼ us ð19Þ pans to evaporate brine. Each case requires the input of numer-
p ous weather data and site specific information. Historical
The solar radiation is calculated by adjusting the extrater- weather data including long-term averages can be obtained
restrial radiation for the relative sunshine duration from a variety of sources. The data used here were derived
 n from the online databases of the National Climatic Data Cen-
Rs ¼ as þ bs Ra ð20Þ ter of the National Oceanic and Atmospheric Administration
N
(www.noaa.gov); and Weather Underground (www.wunder-
where Rs is the solar radiation (MJ mÿ2 dayÿ1), n is the actual ground.com). Model inputs are summarized in Table 1.
duration of sunshine (hours), N is the maximum possible Williams (2002) describes a traditional Mesoamerican salt-
amount of sunshine (hours), and as and bs are regression pa- works operating on the western coast of Mexico in the state of
rameters with recommended values of 0.25 and 0.50, respec- Michoacan. The La Placita saltworks operates during the dry
tively. The effective solar radiation reaching the ground season, roughly from April to mid-June. Specialists, called sal-
surface is reduced by solar reflection caused by albedo ineros, scrape the upper surface of soil from a nearby dry es-
tuary. The salt-encrusted sand called salitre is first filtered
Rns ¼ ð1 ÿ aÞRs ð21Þ through a tapeixtle using seawater and the resulting enriched
brine is transferred to specially prepared evaporation pans
where Rns is the net solar radiation (MJ mÿ2 dayÿ1), and a is
called eras. According to Williams, La Placita has 18 eras
the albedo of the surface. For open water Shuttleworth
with most measuring approximately 3 m  3 m in dimension.
(1993) recommends an albedo value of 0.08. However, for
Each era is lined with beach sand and lime and can hold about
shallow evaporation pans the underlying reflectivity of the
400 L of brine. Every day 2e3 buckets of brine must be added
pan must be considered. For example, dark earthenware or
to maintain a consistent level. It takes 5 days of evaporation to
wooden pans filled with water may have an albedo closer to
concentrate the brine sufficiently to begin collecting salt, about
0.05. When evaporating brine, salt crystals will begin to
25e30 kg are then collected every other day from each era.
form, thus increasing the reflectivity. Rife et al. (2002) found
An average of 7 tons of salt can be produced at each saltmak-
that albedo values of 0.3 were appropriate for modeling diur-
ing site during the dry season.
nal weather cycles over a salt-encrusted playa.
The flux of long-wave radiation reflected by the ground
back into space is given by the StefaneBoltzmann law minus Table 1
Monthly weather averages for selected saltmaking locales
that which is absorbed by clouds, water vapor, dust, and car-
bon dioxide. The net long-wave radiation can be determined Michoacan, Mexico Shanghai, China
by Latitude 16.83 N 31.17 N
 4 Month May 2000 July
4 pffiffiffi
Temperature ( C)
  
T þTmin Rs 27.2 28.4
Rnl ¼s max
ÿ
0:34ÿ0:14 e 1:35 ÿ0:35 Humidity (%RH) 77 83
2 ðas þbs ÞRa
Solar Radiation (MJ mÿ2 day-1) 29.1 30.8
ð22Þ Wind Speed (m sÿ1) 5 4
Weather data were obtained from the National Climatic Data Center of the
where s is the StefaneBoltzmann constant (4.903  U.S. National Oceanic and Atmospheric Administration (www.noaa.gov). So-
10ÿ9 MJ Kÿ4 mÿ2 dayÿ1), T is the maximum and minimum lar radiation for a cloudless day was calculated from Section 2.1.3.
 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462 1457

Based on the information that Williams (2002) supplies, an of salt in these portable pans took only 1 day under fine
estimate for the average evaporation rate can be derived. Dur- weather in summers.’’ Although we are not told of the brine
ing the first 5 days of evaporation, the brine concentrates up to strength or volume in these pans, some reasonable assump-
a maximum value of about 26.2 wt%. Once the maximum tions can be made. If the brine is near saturation (26.2 wt%)
value is reached, further evaporation results in the formation and averages 3 cm in depth, then numerical calculations
of salt crystals. Eq. (1) indicates that producing 12.5e15 kg indicate that the evaporation rate would be 4.8 mm dayÿ1
of salt per day requires the evaporation of 35e43 kg of water. (see Table 1 for model inputs). Upon complete evaporation
Converting the mass of the water evaporated to volume and di- the depth of the salt crystals would be about 5 mm indicating
viding by the surface area of an era (w9 m2) results in an the total evaporation time to be about 5 days. This is somewhat
evaporation rate of 3.9e4.8 mm dayÿ1. longer than the estimate given by Chiang (1976) unless he was
Using the steps outlined in Section 2.1 we can numerically referring to incipient crystallization which could begin on the
simulate the evaporation process during Williams’ field season first day.
at La Placita in 2000. Input values used for the model are listed The two examples provided here illustrate the value of
in Table 1. The albedo of the pan is expected to be relatively numerical modeling in addition to historic descriptions. Com-
high (w0.3) owing to the lime-lined pan and the constant pre- puter simulation of the well documented La Placita saltworks
cipitation of salt crystals at the bottom of the pan. Results for in Mexico provides remarkable agreement between the
the numerical simulation suggest that the evaporation rate author’s observations and numerical results. However, historic
should be in the range of 4.1e4.7 mm dayÿ1 following the descriptions of 18th century Chinese saltmaking appear to
initial 5-day concentration period, essentially identical to that differ from the numerically calculated evaporation rate.
obtained in practice. The range in evaporation rate given by Numerical modeling provides an independent means for vali-
the numerically calculated results reflects the potential varia- dating historical claims regarding salt production.
tion in cloud cover from 0 to 20%. Assuming an initial brine
concentration of about 15 wt% and an average cloud cover of
10%, Fig. 2 represents model results from the numerically sim- 3. Evaporation from an externally heated pan
ulated evaporation occurring from a single era at La Placita dur-
ing the first 20 days of May 2000. Interestingly, extrapolating This technique typically involves the suspension of a vessel
the production rate of the era in Fig. 2 to the entire field season over a fire or the emplacement of a vessel directly onto a bed
(w75 days) yields approximately 790 kg of salt. This correlates of hot coals. Heat is transferred through the base and walls of
to an average of about 9 eras in use at La Placita based on the the vessel and warms the interior fluid. The amount of heat
estimate of 7 tons of salt produced each season. This agrees transferred to the brine is governed by the energy output of
well with Williams’ (2002: p. 243) assertion that not all of the fire and efficiency of heat transfer in a particular brine-
the 18 eras are in use at one time. boiling setup. The efficiency is defined as the amount of
Similarly, solar evaporation along the coastal margin of heat transferred into the brine relative to the total heat pro-
China utilized enriched brine obtained by leaching salty earth duced by the fire. For an open fire efficiency is low owing
or in some cases from ashes that were spread onto the ground to the loss of heat by combustion gases escaping around the
(Chiang, 1976). The practice of using portable wooden pans vessel. Glanville (2005) calculated an efficiency of about
for evaporation began in the 18th century in the Hangchou 20% for the 19th century methods of brine boiling in iron ves-
Bay area. Although variations in size were common, the typ- sels described by LeConte (1862) along the east coast of the
ical wooden pan was about 2.5 m long, 1 m wide, and 3e United States. This value is somewhat better than the approx-
6 cm deep. Chiang (1976: p. 526) states that ‘‘crystallization imately 15% efficiency for boiling water in cooking vessels
over an open fire (McCracken and Smith, 1998) but less
than the more optimal 28e40% efficiency of well vented
8 200
wood stoves (Joshi et al., 1991). For earthenware vessels
Evaporation Rate, mm/day 180
Evaporation Rate (mm/day)

with substantially lower thermal conductivity than iron

7 Salt Produced, kg 160
(Table 2), efficiency is reduced even further and may be as
Salt Produced (kg)

140 low as 2e5% (Glanville, 2005).

6 120 Although efficiency is a poorly known variable and is
100 unique to each setup for brine boiling with externally applied
5 80 heat, the amount of salt produced can be determined by simply
60 estimating the temperature on the exterior surface of the ves-
4 40 sel. This method eliminates (or sidesteps) the issue of effi-
20 ciency and focuses directly on the heat being transferred
3 0
through the vessel and into the brine. Thus, the amount of
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 heat calculated would be the absolute minimum required to ac-
Day of Month (May 2000) complish evaporation and would represent a system of 100%
Fig. 2. Numerical simulation results for a 9 m2 era at La Placita saltworks in heat transfer efficiency. Heat loss by escaping combustion
Michoacan, Mexico. gases, evaporating water molecules, and conduction through
1458 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462

Table 2
Physical properties of various materials at 20  C
Material Density (g cmÿ3) Thermal conductivity (W mÿ1 Kÿ1) Specific Heat (J gÿ1 Kÿ1) Source
Clay (9.3e18.3 wt% H2O) 1.33e1.50 0.4e0.7 C
Brick, clay 1.60e1.82 0.41e0.63 0.92 A, B, F
Iron (100%Fe) 7.87 74.48 0.4473 D
Cast iron (3.16%C) 7.15 46.9 0.837 D, G
Copper 8.96 393.7 0.3846 D
Bronze (89%Cu þ 11%Sn) 8.77 70.6 0.3771 E
Lead 11.36 34.7 0.1287 D
Data from [A] Dondi et al. (2004); [B] Bhattacharjee and Krishnamoorthy (2004); [C] Abu-Hamdeh and Reeder (2000); [D] Davis (1998); [E] Copper Develop-
ment Association (2007); [F] Lange (1952); [G] MatWeb (2007).

q
vessel walls to surrounding air all contribute to lowering the E¼ ð26Þ
l
overall efficiency.
using Eq. (3) to determine the latent heat of vaporization.
3.1. Calculation method for evaporation by externally
applied heat 3.2. Numerical simulation of applied external heat

By concentrating solely on the heat transferred through Simulating the effects of brine boiling requires knowledge
a vessel’s wall, a one-dimensional steady-state heat transfer of both the physical characteristics of the pan and the amount
equation can be applied of applied heat (i.e. temperature on pan exterior). Often one or
the other of these criteria is unknown for a specific saltwork.
k The discovery of several late Roman age lead pans in Britain
q ¼ A ðTe ÿ Ti Þ ð24Þ
Dx provides constraints on these variables which can be used to
demonstrate the numerical model. Lead melts at 327  C and
where q is the heat transfer rate in joules per second (J/s), A is salt practitioners would likely have kept the external tempera-
the heated surface area, Dx is vessel thickness, k is the thermal ture well below this value to avoid accidental melting of the
conductivity in watts per meter-kelvin (W mÿ1 Kÿ1), and Te pan. At Shavington, Cheshire, a lead pan measuring 100 cm
and Ti are the temperature (K) of the vessel exterior and inte- by 90 cm by 14 cm deep was found cut into eight pieces, pre-
rior surfaces, respectively (Geankoplis, 2003). This equation sumably as a first step to recycling the material (Penney and
assumes that a constant flow of heat is applied to the vessel ex- Shotter, 1996). At 0.8 cm thickness, the original weight of
terior. Materials of high thermal conductivity (Table 2) allow the pan would have been approximately 118 kg. Although
for substantially more heat to be transferred through the vessel none of the surviving Roman era lead pans have been found
whereas increasing the vessel’s wall or basal thickness reduces in situ, they likely would have been placed across earthen
heat flow. For most practitioners of brine boiling, the primary flue trenches much like many ceramic salt pans elsewhere in
means of evaporation control comes from varying the amount Britain (e.g., Bradley, 1992). Fires were presumably kept rel-
of external heat applied to the vessel which in turn determines atively low, 200e250  C, to prevent damage to the pan.
the temperature of the vessel’s exterior. The temperature of the Numerically simulating the evaporation from the Shaving-
internal surface Ti is typically the same as the boiling point of ton salt pan requires only an estimation of the initial brine con-
the brine and will change as evaporation proceeds and the centration. Here an arbitrary 10 wt% is used, but this value has
brine concentrates to a maximum value of about 29.0 wt% only a minor impact on the duration of the evaporation pro-
at the boiling point. Although at very high external tempera- cess. Even if the temperature on the pan’s exterior is kept rel-
tures, the interior surface may exceed that of the boiling brine. atively low (w200  C), Fig. 3 demonstrates a fairly quick
The boiling point of sodium chloride brine can be estimated by evaporation for the Shavington pan. Neglecting substantial
the boiling point elevation equation heat loss, the brine is brought to a boil in only 3 min and con-
tinues until 14 min, whereby evaporation has resulted in the
Tb ¼ Kb mi þ 100  C ð25Þ maximum allowable brine concentration. Further evaporation
after 14 min results in salt crystals forming inside the pan. Ul-
where Tb is the boiling temperature ( C) of the brine, Kb is the timately, about 14 kg of salt can be recovered in less than
proportionality constant for water (0.512  C/m), m is the brine 20 min of boiling, assuming 10 wt% brine in a 126 L pan.
concentration (mol NaCl per L of water), and i is the number
of dissociated ions per formula unit (NaCl ¼ 2). At high alti- 4. Evaporation using hot immersed objects
tudes where pure water boils at temperatures below 100  C,
the appropriate boiling point should instead be inserted into A third evaporation scenario considered here is that of a hot
the equation. Finally, the evaporation rate (g/s) can be deter- object (e.g., stone) placed inside a pan of brine. This method is
mined by believed to have been utilized for salt production in eastern
 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462 1459

35 20 analytical solution to the conduction equation becomes con-

Brine Concentration (wt% NaCl)

Brine Concentration 18 siderably more complex.

30
16 Finite difference methods can be used to obtain a numerical

Crystallized Salt (kg)

25 14 solution to Eq. (27), which involves computing temperature
Crystallized Salt
12 changes after small positional (n) and time (t) steps are incre-
20
10 mented (Geankoplis, 2003: pp. 386e387; Carslaw and Jaeger,
15
8
1959). The positional steps effectively divide the object into
6
concentric rings that are Dx (m) thick. For a sphere, the tem-
10
perature can be determined by
4
5
2  
1 2n þ 1 2n ÿ 1
0 0 tþDt T n ¼ T nþ1 þ ðM ÿ 2Þt T n þ T nÿ1 ð29Þ
0 5 10 15 20 25 30 M 2n t 2n t
Time (min)

Fig. 3. Brine evaporation from a lead pan placed over fire. Pan conditions are:
where M ¼ Dx2/(aDt). At the center (n ¼ 0) the equation
pan exterior ¼ 200  C, pan volume ¼ 126 L, pan thickness ¼ 0.8 cm, and pan changes to
heated area ¼ 0.9 m2. Initial pan heat-up time is negligible.
4 Mÿ4
tþDt T 0 ¼ T1 þ T0 ð30Þ
Mt M t
North America from about A.D. 1000e1400 (Brown, 1980,
1981). Stone boiling as a cooking technique probably began where M  4 for both Eqs. (29) and (30). At the surface, equa-
with the introduction of pottery (ca. 2500 B.C.), if not earlier, tions accounting for convection must be utilized assuming that
and continued into the historic period (Sassaman and Rudol- the heat capacity of the outer half-slab can be neglected
phi, 2001). Thick-walled ceramic ‘‘pans,’’ most in association
with salines, have been found with capacities ranging from 40 nN ð2n ÿ 1Þ=2
to 400 L. The enormous size and weight of these vessels when tþDt T n ¼ tþDt T a þ 2n ÿ 1 tþDt T nÿ1 ð31Þ
2n ÿ 1
filled with brine would have made them practically immovable þ nN þ nN
2 2
and suspension over a fire seems equally unlikely. Brown
(1980, 1981) concluded that these salt pans were placed in ba- where Tn represents surface temperature and Tnÿ1 the temper-
sin-shaped ground depressions and heated stones from nearby ature at 1 positional step below the surface.
fires were dropped into the pan to facilitate evaporation. The
lack of exterior discoloration from fires on many pans and 4.2. Numerical simulation of stone boiling
the occasional find of stones inside pans (e.g., Bushnell,
1907) lend support to the conclusion that stone boiling was Solving these equations allows for the determination of the
at least one method utilized by Native Americans to evaporate average stone temperature with time after immersion. Initially,
brine. the heat transferred raises the temperature of the brine up to its
boiling point. Any additional heat released by the stone serves
4.1. Calculation method for stone boiling to evaporate water and concentrate brine. Eventually the brine
is concentrated to a maximum value of about 29.0 wt% at its
A hot object transfers its internal heat energy to the sur- boiling point with further evaporation resulting in the forma-
rounding fluid through convective heat transfer at the object’s tion of salt crystals.
surface. This transfer of heat is governed by the unsteady-state For the numerical model, the boiling stone is assumed to be
conduction equation chert. Chert was well known to Native Americans and was
commonly used to make stone tools. Although the physical
vT v2 T properties of chert are not well studied, there are numerous
¼a 2 ð27Þ
vt vx studies on the analogous material of amorphous or fused
quartz. The thermal conductivity of chert can be determined
where a is the thermal diffusivity (m2/s), t is time (s), and T is
from a polynomial fit of published data by Kanamori et al.
the temperature (K) of the object. Thermal diffusivity is
(1968) and Clauser and Huenges (1995)
simply

k k ¼ 4:67  10ÿ9 T 3 ÿ 8:53  10ÿ6 T 2 þ 6:29  10ÿ3 T

a¼ ð28Þ
rCp ÿ 9:85  10ÿ2 ð32Þ

where k is the thermal conductivity (W mÿ1 Kÿ1), r the den- where k is thermal conductivity (W mÿ1 Kÿ1), and T is the
sity (kg mÿ3), and Cp the heat capacity of the object stone temperature (K). Similarly, Robie et al. (1978) have de-
(J gÿ1 Kÿ1). For simplicity, here it is assumed that the object veloped expressions for the heat capacity of quartz
is spherical and that only the one-dimensional  direction
need be considered. For irregularly shaped objects, the Cp ¼ 44:603 þ 3:7754  10ÿ2 T ÿ 1:0018  106 T ÿ2 ð33Þ
1460 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462

where Cp is the heat capacity (J molÿ1 Kÿ1) and T is the tem- 25

Volume Ratio
perature (K) for the range of 298e844 K. At higher tempera- Brine/Stone
tures the equation changes to 20 1.0

Water Evaporated (kg)

2.0
3.0
Cp ¼ 58:928 þ 1:0031  10ÿ2 T ð34Þ 4.0
15
ÿ3
for the range 844e1800 K. The density of chert is 2.6 g cm .
This information together can be used to determine the ther- 10

mal diffusivity (a) properties of chert for the numerical model.

Fig. 4 graphically represents a simulated evaporation of 5
placing 700  C chert stone(s), representing 25% of the total
40 L volume, into a salt pan containing 10 wt% brine at 0
25  C. Boiling begins immediately for brine surrounding the 200 300 400 500 600 700 800 900 1000

stone and continues for about 4 min, by which time the stone Initial Stone Temperature (°C)
temperature now equals that of the brine and no further heat is Fig. 5. Evaporation curves for stone boiling with various brine:stone volume
transferred. During the first 2 min the evaporating brine con- ratios. Curves represent an initial brine temperature of 25  C and 10 wt%
centrates from 10 wt% up to 29.0 wt%. At 1.6 min salt crystals concentration.
begin forming and continue until the stone and brine reach
thermal equilibrium. For this scenario, 373 g of crystallized obtained with various brine/stone volume ratios. The 10 wt%
salt would be produced. Emplacement of additional hot stones brine is assumed to initially be at 25  C before emplacement
into the pan would continue the evaporation process. of a hot spherical stone of chert. Even for a Vb/Vs ¼ 1 and
From the short timescales involved to obtain salt, it seems an initial stone temperature of 1000  C, the mass of the stone
this method would clearly be effective in evaporating brine. would still exceed the mass of the evaporated water by a factor
Unlike suspension of a pan over a fire, stone boiling releases of 2.5. Lower stone temperatures or smaller volumes of stone
all of its internal heat directly into the brine. However, there would result in greater discrepancies between stone mass and
may be practical limitations for manipulating large volumes evaporated water mass.
of very hot stones. Stones heated to high temperatures often
shatter and large stones would be difficult to transport from 4.3. Hot objects with negligible internal resistance
a fire to the brine pan. Smaller stones would make handling
easier, but would require more repeated firings to achieve An alternative mathematical treatment can be applied when
the same evaporation as large stones. The smallest salt pan the hot object has negligible internal resistance to the flow of
found at the Kimmswick site near St. Louis (Bushnell, 1907) heat. These objects (e.g., copper and iron) have high thermal
had a volume of approximately 40 L. Assuming a scenario conductivities and will have an approximately uniform inter-
similar to that outlined above, emplacement of 25 vol% stone nal temperature profile at any given time following immersion
into the salt pan would translate to 26 kg of extremely hot into brine. To maintain an energy balance, heat loss through
stone(s) that would have been manipulated. This seems to sug- the vessel wall or to the atmosphere is assumed to be minimal
gest that stone boiling may actually require a tremendous relative to the rapid rise in brine temperature. The total amount
amount of human labor to achieve significant quantities of of heat transferred at any given time can be determined by
salt. Fig. 5 indicates the potential evaporation that could be
Q ¼ Cp rVðT0 ÿ TN Þ 1 ÿ eÿðhA=Cp rVÞt
 
ð35Þ
20 0.50
18 0.45
where Q (J) is the total heat transferred, V (m3) is the volume
of the object, h is a heat transfer coefficient for natural convec-
Heat Transferred (MJ)

16 0.40
tion (w5000 W mÿ2 K), A is the heated area (m2), and t is the
Salt Produced (kg)

14 0.35
elapsed time (s) after immersion (Geankoplis, 2003: pp. 277e
12 0.30
286, 357e359).
10 0.25
8 0.20
5. Conclusions
6 0.15
4 0.10
Heat transferred, MJ The goal of this paper is to develop a mathematical basis
2 Salt Produced, kg 0.05
for brine evaporation that can be used by investigators to quan-
0 0.00
0 1 2 3 4 5 6 7 8 9 titatively describe salt production activities. The methods de-
Time (minutes) scribed herein cover three distinct techniques for evaporating
brine: (1) solar evaporation, (2) boiling due to an externally
Fig. 4. Numerical simulation results for Native American stone boiling. Rep-
resented here are results for placing a 700  C stone with a volume of 10 L into applied heat source, and (3) boiling caused by a hot immersed
a ceramic pan containing 30 L of 10 wt% brine initially at 25  C. The stone is object. Historically these techniques were sometimes used in
assumed to be a spherical nodule of chert. combination to achieve the desired evaporation results. Input
 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462 1461

parameters for the three models range from common weather Andrews, A., 1983. Maya Salt Production and Trade. University of Arizona
variables for solar evaporation to pan and stone physical Press, Tucson.
Bhattacharjee, B., Krishnamoorthy, S., 2004. Permeable porosity and thermal
dimensions for brine boiling. No one evaporation method conductivity of construction materials. Journal of Materials in Civil Engi-
can be considered superior since other factors such as fuel neering 16, 322e330.
availability, sunshine intensity, brine concentration, and even Bowman, H.H.M., 1956. Salinity data on marine and inland waters and plant
the cultural value placed on human labor make the choice of distribution. The Ohio Journal of Science 56 (2), 101e106.
evaporation technique unique to each culture. However, the Bradley, R., 1992. Roman salt production in Chichester Harbour: rescue exca-
vations at Chidham, West Sussex. Britannia 23, 27e44.
calculation methods given here do allow for direct compari- Brown, I.W., 1980. Salt and the Eastern North American Indian: an archaeo-
sons to be made regarding evaporation efficiency. Brine boil- logical study. In: Lower Mississippi Survey Bulletin, No. 6. Peabody Mu-
ing offers the quickest means of evaporation, typically on seum, Harvard University.
the order of minutes for the examples given here, but this Brown, I.W., 1981. The Role of Salt in Eastern North American Prehistory.
does not account for the time spent acquiring fuel. Solar evap- Louisiana Archaeological Survey and Antiquities Commission Anthropo-
logical Study No. 3. Department of Culture, Recreation and Tourism, Ba-
oration does not require fuel but may take days or weeks to ton Rouge.
accomplish and is limited to geographic areas with high evap- Bushnell, D.I., 1907. Primitive salt-making in the Mississippi Valley, I. Man
oration and little precipitation. No. 13. London.
Although beyond the scope of the current paper, quantify- Calder, I.R., Neal, C., 1984. Evaporation from saline lakes: a combination
ing salt production opens up new areas of research that can equation approach. Hydrological Sciences Journal 29, 89e97.
Carslaw, H.S., Jaeger, J.C., 1959. Conduction of Heat in Solids, second ed.
be addressed. Changes in production methods through time Clarendon Press, Oxford.
can be the result of either cultural or technological factors, Chiang, T.C., 1976. The production of salt in China, 1644e1911. Annals of
or both. In a simple least-cost model, humans would be ex- the Association of American Geographers 66, 516e530.
pected to seek strategies to minimize labor and fuel require- Clauser, C., Huenges, E., 1995. Thermal conductivity of rocks and minerals.
ments in the salt production process. Salt production has In: Ahrens, T.J. (Ed.), Rock Physics and Phase Relations e a Handbook
of Physical Constants. AGU Reference Shelf, vol. 3. American Geophys-
often been described as a labor intensive endeavor requiring, ical Union, Washington, pp. 105e126.
in the case of brine boiling, extraction of substantial quantities Copper Development Association. <http://www.copper.org> (accessed
of fuel from the local environment. The calculations presented 03.07.07.).
herein offer a direct means to evaluate technological changes Davis, J.R. (Ed.), 1998. Metals Handbook. ASM International, Materials Park,
for improved evaporation efficiency. For example, Eq. (24) Ohio.
Demir, I., Seyler, B., 1999. Chemical composition and geologic history of sa-
describes the direct relationship between vessel thickness line waters in Aux Vases and Cypress Formations, Illinois Basin. Aquatic
and the rate of heat transfer. Thinner vessel walls require Geochemistry 5, 281e311.
proportionately less fuel to achieve evaporation. In addition, Dondi, M., Mazzanti, F., Principi, P., Raimondo, M., Zanarini, G., 2004. Ther-
determinations of minimum energy (fuel) requirements can mal conductivity of clay bricks. Journal of Materials in Civil Engineering
be made which provide insight into local deforestation or other 16, 8e14.
Early, A.M., 1993. Caddoan saltmakers in the Ouachita Valley: the Hardman
resource depletion. Early (1993) noted that the removal of fruit Site. In: Research Series, No. 43. Arkansas Archeological Survey,
bearing plants and nut bearing trees from the vicinity of a salt- Fayetteville.
works may have altered diet and even changed the local pop- Finch, J.W., 2001. A comparison between measured and modeled open water
ulation of wild animals that depend on these resources. These evaporation from a reservoir in south-east England. Hydrological Pro-
and other concerns can now be better addressed by quantifying cesses 15, 2771e2778.
Flad, R., Zhu, J., Wang, C., Chen, P., von Falkenhausen, L., Sun, Z., Li, S.,
the brine evaporation process. 2005. Archaeological and chemical evidence for early salt production in
China. Proceedings of the National Academy of Sciences of the United
States of America 102, 12618e12622.
Acknowledgments Geankoplis, C.J., 2003. Transport Processes and Separation Process Principles,
fourth ed,. Prentice Hall, New Jersey.
The author thanks Ann M. Early and Dan F. Morse for Glanville, J., 2005. Brine transport and woodland salt making in southwestern
helpful discussions on this subject. Ashley Dumas and Robert Virginia: an hypothesis. Paper presented at the Eastern States Archeological
C. Mainfort provided insightful comments on an early draft of Federation Meeting, Williamsburg, Virginia, USA, November 12, 2005.
Joshi, V., Venkataraman, C., Ahuja, D.R., 1991. Thermal performance and
this paper. Valuable comments were also received from two emission characteristics of biomass-burning heavy stoves with flues. Pa-
anonymous reviewers that greatly improved this manuscript. cific and Asian Journal of Energy 1, 1e19.
Kanamori, H., Fujii, N., Mizutani, H., 1968. Thermal diffusivity measurement
of rock-forming minerals from 300 to 1100 K. Journal of Geophysical
References Research 73 (2), 595e605.
Kostick, D.S., 2002. Salt. In: Metals and Minerals. U.S. Geological Survey
Abu-Hamdeh, N.H., Reeder, R.C., 2000. Soil thermal conductivity: effects of Minerals Yearbook, vol. 1. U.S. Government Printing Office, pp. 1e17
density, moisture, salt concentration, and organic matter. Soil Science So- (Chapter 64).
ciety of America Journal 64, 1285e1290. Lange, N.A., 1952. Handbook of Chemistry, eighth ed. Handbook Publishers,
Adshead, S.A.M., 1992. Salt and Civilization. St. Martin’s Press, New York. Sandusky, Ohio.
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration: LeConte, J., 1862. How to Make Salt from Sea-Water? Governor and Council
guidelines for computing crop water requirements. In: Irrigation and of South Carolina, Columbia.
Drainage Paper, 56. Food and Agriculture Organization of the United Lide, D.R. (Ed.), 1991. CRC Handbook of Chemistry and Physics, 72nd ed.
Nations, Rome. CRC Press, Boca Raton.
1462 D.G. Akridge / Journal of Archaeological Science 35 (2008) 1453e1462

MatWeb, 2007. Material Property Database. Available from: http://www.mat- Robie, R.A., Hemingway, B.S., Fisher, J.R., 1978. Thermodynamic properties
web.com (accessed 03.07.07.). of minerals and related substances at 298.15 K and 1 bar (105 pascals)
McCracken, J.P., Smith, K.R., 1998. Emissions and efficiency of improved pressure and at higher temperatures. In: USGS Survey Bulletin, 1452.
woodburning cookstoves in highland Guatemala. Environment Interna- U.S. Government Printing Office, Washington.
tional 24 (7), 739e747. Romanklw, L.A., Chou, I.-M., 1983. Densities of aqueous NaCl, KCl, MgCl2,
Olivier, L., Kovacik, J., 2006. The ‘Briquetage de la Seille’ (Lorraine, France): and CaCl2 binary solutions in the concentration range 0.5e6.1 m at 25, 30,
proto-industrial salt production in the European Iron Age. Antiquity 80, 35, 40, and 45  C. Journal of Chemical and Engineering Data 28, 300e
558e566. 305.
Penman, H.L., 1948. Natural evaporation from open water, bare soil, and Sassaman, K.E., Rudolphi, W., 2001. Communities of practice in the early pot-
grass. Proceedings of the Royal Society, Series A 193, 120e145. tery traditions of the American Southeast. Journal of Anthropological Re-
Penney, S., Shotter, D.C.A., 1996. An inscribed Roman salt-pan from Shaving- search 57, 407e425.
ton, Cheshire. Britannia 27, 360e365. Shuttleworth, W.J., 1993. Evaporation. In: Maidment, D.R. (Ed.), Handbook of
Rife, D.L., Warner, T.T., Chen, F., Astling, E.G., 2002. Mechanisms for diurnal Hydrology. McGraw-Hill, New York.
boundary layer circulations in the Great Basin Desert. Monthly Weather Williams, E., 2002. Salt production in the coastal area of Michoacan, Mexico.
Review 130, 921e938. Ancient Mesoamerica 13, 237e253.