Detoxification and mineral supplementation
as functions of geophagy 1"23
Timothy Johns and Martin Duquette
ABSTRACT Clays employed historically in the consump­
tion of astringent acorns plus seven edible clays from Africa
were examined in relation to the functional significance of hu­
man geophagy. On the basis of sorptive maxima for tannic acid
ranging from 5.6 to 23.7 mg/g, we conclude that adsorption of
tannic acid in traditional acorn preparation methods in Cali­
fornia and Sardinia helped make these nuts palatable. Calcium
available in solution at pH 2.0 and 0.1 mol NaCl/L was 2.10
and 0.71 mg/g for the Sardinian and Californian clays, respec­
tively. The African clays released calcium, copper, iron, mag­
nesium, manganese, or zinc in amounts of nutritional signifi­
cance from some clays but not from others. A clay recovered
from an archaeological site occupied by Homo erectus and early
H. sapiens was indistinguishable mineralogically, in detoxifi­
cation capacity and in available minerals, from clays used in
Africa today. We suggest that the physiological significance of
geophagy made it important in the evolution of human dietary
behavior. Am J Clin Nutr 1991 *,53:448-56.
KEY WORDS Geophagy, pica, clay, dietary behavior, di­
etary evolution, Homo erectus, detoxification, minerals
Introduction
Background
Geophagy, or the deliberate consumption of earth substances,
is a complex and perplexing human behavior. It has been gen­
erally classified as a form of pica ( 1); indeed, it is often considered
as the prototype of pica. As such, it is lumped with aberrant
behaviors like the consumption of starch, ice, paint, cigarette
butts, and burnt matches. Historically, and in many traditional
subsistence societies today, geophagy is almost synonymous with
clay consumption (2) and is a normal human activity. Geophagy
is widespread in the animal kingdom, particularly among gen­
eralist herbivores (3), and is common among primates including
chimpanzees (4). While nutritionists tend to look at geophagy
as maladaptive or as a symptom of metabolic dysfunction, an­
imal behaviorists look for adaptive functions of such phenomena.
Efforts to examine traditional forms of geophagy within their
own ecological and cultural contexts not only help elucidate the
adaptive aspects of traditional diets but offer insights into evo­
lution of human dietary and medical behaviors. From this per­
spective we may better be able to understand truly aberrant di­
etary behaviors of humans within industrialized urban societies.
448 Am J Clin Nutr 1991;53:448-56. Printed in USA. © 1991 American Society for Clinical Nutrition
Downloaded from https://academic.oup.eom/ajcn/article-abstract/53/2/448/4694247
by guest
on 15 March 2018
We studied the properties of clays collected from several geo­
graphical locations in relation to the traditional ways the days
are used in those locations. Two case studies are considered in
relation to the following two explanations of the function of
geophagy.
Nutrient explanation
The most prevalent explanation for human geophagy is that
it is a response to alleviate nutrient deficiency (1,5). However,
in clinical studies where geophagy has been associated with min­
eral deficiencies of a pathological degree, it has been implicated
both as a cause and as a result of particular deficiencies (6).
Following the pica model of pathological craving, deficiencies
of specific nutrients have been the focus of the search for the
cause of geophagy ( 1, 5). Experimentally induced deficiencies,
such as those for iron and potassium, are known to evoke changes
in dietary behavior in animals. Rats selectively increase their
calcium intake during lactation but not during pregnancy (7).
Consumption of substances such as soil and bones (osteophagia)
appears to provide animals nutrients such as calcium, sodium,
and phosphorus under certain circumstances (8). In humans,
deficiencies of sodium and iron have been linked with specific
appetite for these minerals (9, 10).
Although iron deficiency has been established in some forms
of pica (1,5), no clear evidence has been presented to link geo­
phagy and iron need. Neither has sodium been linked in this
way with clay consumption. Sodium is seldom limiting in other
than pathological situations.
If clay is seen as one of several substances (and foods), the
consumption of which offers positive nutritional or other health
benefits, the concepts of craving and specific appetite are of sec­
ondary or no relevance in the explanation of geophagy. From
this perspective the analysis of minerals in edible clays in relation
to normal dietary requirements is more important evidence than
are data on deficiencies in the human subjects. Analysis of clays
1From the School of Dietetics and Human Nutrition and the De­
partment of Renewable Resources, Macdonald College of McGill Uni­
versity, Quebec.
2 Supported by a grant from the Natural Sciences and Engineering
Research Council, Canada.
3Address reprint requests to T Johns, School of Dietetics and Human
Nutrition, Macdonald College, Ste Anne de Bellevue, Quebec, H9X ICO,
Canada.
Received January 5, 1990.
Accepted for publication June 13, 1990.
 FUNCTIONS OF GEOPHAGY 449
consumed by people in Africa and Central America indicated Most nutrient explanations for geophagy have been based on
that in certain instances these clays are important sources of studies of clays consumed in Africa (11, 13). The use of African
supplementary calcium, copper, iron, magnesium, or zinc (11- clays has seldom been linked to the detoxification of dietary
13). However, no single nutrient factor explains geophagy in a constituents (see Ref 22 for an exception). A group of seven
general sense. edible clays collected in six countries across sub-Saharan Africa
was used in comparison with the acorn clays, and the data were
Detoxification explanation considered in relation to the two alternative explanations for
geophagy.
An alternative explanation is that geophagy serves to adsorb
Another clay was obtained from an archaeological site in
dietary xenobiotics and thereby detoxify foods and/or make them
Zambia. Its presence as an anomaly in a lower horizon in the
more palatable. Ecologists have suggested adsorption of alkaloids,
excavation (23) indicates that it may have been consumed by
tannins, oxalates, and other natural plant constituents as an ex­
hominids (Homo erectus or early H. sapiens) who occupied the
planation for many cases where animals in natural settings are
site (24). Properties of this clay were compared with those of
observed to eat clay (14). Under laboratory conditions rats eat
clays consumed by humans today.
clay in response to gastrointestinal malaise induced by poisoning
and other stresses (15, 16).
Geophagy is common in nonhuman primates including Methods
chimpanzees (4, 17). The potential role of clay ingestion in neu­
tralizing plant toxins in primate diets has important implications Clay sources and handling
for the understanding of the evolution of human dietary behav­
Samples of the two clays used traditionally to detoxify acorns
ior, suggesting that plant foods may have been more important
were collected in Baunei, Sardinia, Italy (Lat 40°04' N, Long
to our ancestors than was previously thought (3, 18). However,
9°39' E) in March 1988 and in Hopland, CA (Lat 38°58' N,
the hypothesis that primate geophagy serves a detoxification
Long 123°08' W) in November 1986. Additional clays from the
function in relation to plant secondary compounds (19, 20) is
Porno Indians of California deposited in the Lowie Museum,
difficult to test experimentally. Because humans can explicitly
University of California, Berkeley, were obtained from that in­
communicate their purposes, certain instances of human geo­
stitution and are listed in Table 1. The limited amounts available
phagy offer better case studies for testing this hypothesis.
of the latter clays restricted their use to mineralogical and sorp­
We presented previously a detoxification explanation for geo­
tion studies only. A sample of Sardinian soil that was not con­
phagy in relation to one ethnographically documented case ( 18).
sumed was collected for comparative purposes from a site west
Clays consumed traditionally with certain bitter and toxic types
of Tortoli (Lat 39°54' N, Long 9°36' E).
of potatoes (Solanum spp) in North and South America bind in
vitro, under physiological conditions of pH and ionic strength,
up to 680 mg/g of the glycoalkaloids characteristic of these tubers.
TABLE 1
Case studies o f clay consumption Sorption of clays used to make acorn bread
In the present study we consider detoxification and nutrient Clay source: locality, year Sorption Affinity
availability in relation to two case studies, one where adsorption (accession number) maximum constant
of plant toxins is the explicit and overriding explanation and
the other where it is not. mg/g
The first case considers two clays that were used with acorns Sardinia, Baunei 13.5 0.15
(Quercus spp) with the explicit purpose of making these nuts, California, Hopland 6.8 0.04
which are high in tannins, more palatable. The Porno Indians Samples from Lowie Museum*
in California and natives of Sardinia used similar methods to Mendocino County, 1904
make acorn bread (2, 21 ). The Porno mixed clay with the ground (LM 1-4471) 19.1 0.29
meal of bitter acorns (principally Quercus lobata Née) in a ratio Yukiah, Mendocino County, 1935
(LM 1-29073) 13.7 0.19
of 1:10-1:20 by volume. Water was added to make a dough,
Talemage, Mendocino County, 1935
and small loaves were baked gently in an earth oven for 12 h. (LM 1-29074) 5.6 0.22
Breads were made without clay from nonbitter acorns or from Bonnie Brae Colony, Mendocino
meal from which tannic acid was removed through a leaching County, 1935 (LM 1-29075) 11.8 0.05
process. The purpose of adding clay is specifically to eliminate Yukiah, Mendocino County, 1935
bitterness. (LM 1-29076) 15.5 0.26
In Sardinia a large amount of clay was stirred with cold water Los Lomas, Sonoma County, 1950
and the suspended clay fraction was added to a pot containing (LM 1-79880) 18.8 0.08
dried acorn meats (Q. ilex L) (~~12 g suspended clay/100 g Stewart’s Point, Sonoma County,
1961 (LM 1-144935) 23.7 0.14
acorn). The mixture was boiled for 5 h, a small amount of ash
Standards
from grape vines was added, and the mixture was cooked a few
Sardinia, Tortoli (nonedible) 0.8
more minutes to a solid mass. Kaolin 2.5 0.09
In an analysis of the tannic acid available to digestion from Montmorillonite 21.2 0.12
breads made according to traditional production methods, as
much as 77% of the tannic acid activity was eliminated when * Clays used by the Pomo Indians of California and deposited at the
clays were added to breads (21 ). University of California, Berkeley.

Downloaded from https ://academic.oup.com/ajcn/article-abstract/53/2/448/4694247

by guest
on 15 March 2018
 450 JOHNS AND DUQUETTE
TABLE 2 for 1 h with 1.0 mol HC1/L in a boiling water bath to destroy
Sorption of African clays the chlorite before slide preparation. The samples were x-rayed
with a copper source with a Rigaku Miniflex (Tokyo), with a
Clay source: country, locality Sorption Affinity
(name) maximum constant scan speed of 0.5 or 2 0 degrees/min and a scan range from 70
to 2.5 0 degrees.
mg/g
Total chemical analysis
Cameroon, Douala (cholk) 3.9 0.05
Gabon, Libreville (calabar) 7.0 0.03 The total chemical analyses were performed on each sample
Kenya-1, Siaya District by x-ray fluorescence. The detection limit for the major elements
Nyanza Province 6.5 0.05 is contained in Table 3.
Kenya-2, Kisumu (okumba) 9.9 0.16
Nigeria, Calabar—purchased in Sorption and mineral release with tannic acid
Douala, Cameroon One gram of clay and 30 mL tannic acid (Anachemia reagent,
(calabar cholk) 6.6 0.07 lot #780626; Anachemia Canada Inc, Lachine, Quebec) solution
Togo, Lomé 9.0 0.04 were placed in test tubes. Six tannic acid concentrations varied
Zambia, Kalambo Falls* 8.6 0.03
from 0 to 40 mg/30 mL and were prepared in an indifferent
Zaire, Lubumbashi 9.6 0.08
electrolyte solution of 0.1 mol NaCl/L, adjusted to pH 2.0 with
* Recovered from archaeological site. HC1. Each experiment was performed in duplicate. The suspen­
sions were shaken for 2 h at room temperature and centrifuged
(600 X g for 10 min).
Tannic acid in the supernatant was measured by absorbance
African clays were collected in the field or obtained from vis­ at 275 nm and compared with a standard curve. Tannic acid
itors to the region (Table 2). The sample from Zambia was re­ and iron form a blue-violet complex that significantly interfered
covered during the excavation of the Kalambo Falls (sample with the adsorbance when the iron concentration was > 3 mg/
KF/8/63/B2, B site) archaeological site in 1959 (23). L. Therefore, when iron was above this amount, a correction
Clays were air-dried, ground in a mortar, and passed through was made by comparing tannic acid absorbance with a standard
a 0.5-mm (#35) sieve before use. curve of tannic acid-FeCl3 mixtures. Only the samples from
The smectite standard contained 85% montmorillonite and Zambia and Zaire required correction in this way.
15% crystobalite as an impurity. A commercial standard of ka- Supernatants from the sorption experiments were filtered
olinite (Georgia Kaolin Co, Elizabeth, NJ) was also used. (polycarbonate 0.4 pm) and analyzed for calcium, aluminum,
iron, and potassium by atomic-absorption spectroscopy (model
Mineral composition 2380, Perkin-Elmer, Norwalk, CT).
Identification of minerals in edible earth samples was carried Ionic strength and pH affect sorption of organic molecules
out by x-ray diffraction on the total sample and on the clay-size and the exchange of ions. The conditions chosen for each factor
fraction. The samples were pretreated with NaOCl and citrate- are intended to provide a representative degree of sorption and
bicarbonate-dithionite to destroy organic matter and oxide coat­ ion exchange for the clays relative to each other and to the bio­
ings (25). The clay fraction (particles < 2 pm) was then separated logical issues under consideration. A pH of 2 is representative
by centrifugation. Oriented-aggregate slides were prepared in of normal stomach pH. The use of 0.1 mol NaCl/L provides
different states [magnesium, magnesium-ethylene glycol, KC1 constant ionic strength for all the experiments, whereas adjusting
air-dried, and KC1 heated to 300 and 550 °C (26)]. When ka- the pH with HC1 avoids the use of buffers that could interfere
olinite and chlorite were both present, the samples were treated with the sorption-exchange process. The sodium and chloride

TABLE 3
Total chemical analysis of clays

Sample source Ca P Mg FeJ+ Zn Cu Mn Cr V Si Ni A1 Ti Ba

P g /g P g /g P g /g m g !g P g /g P g /g p g /g P g /g P g /g mg/g P g /g m g /g P g /g Pg/g
California, Hopland 4217 393 3980 39.3 25 40 3980 1247 98 349.4 76 61.0 3969 1117
Sardinia, Baunei 5718 436 8924 51.6 110 26 1007 86 112 274.0 47 120.8 4816 77
Sardinia, Tortoli 3788 131 5608 34.1 57 15 310 38 49 295.3 12 102.9 2752 34
Gabon 143 349 1628 15.0 28 24 155 155 113 231.6 23 192.4 9155 139
Kenya-1 4145 262 1688 58.3 46 47 775 85 86 254.7 44 137.2 4551 76
Kenya-2 286 175 1266 60.0 185 32 3408 <DL* <DL 212.2 18 171.9 2381 <DL
Nigeria 572 349 1206 15.2 26 40 155 116 102 226.1 30 198.7 9684 104
Togo 143 175 603 22.8 19 37 155 203 154 209.3 78 214.1 4816 182
Zambia 1429 175 603 11.6 <DL 68 155 28 77 363.1 16 77.3 3387 25
Zaire <DL 349 663 20.4 12 28 155 148 114 174.1 113 170.4 5874 132
Detectable limit 71 44 60 0.1 10 10 77 10 10 0.1 10 5.4 60 9

* <DL, less than detectable limit.

Downloaded from https://academic.oup.eom/ajcn/article-abstract/53/2/448/4694247

by guest
on 15 March 2018
 FUNCTIONS OF GEOPHAGY
ions are usually considered as making an indifferent electrolyte
solution, meaning that they do not interfere with the sorption
process.
The sorption of tannic acid by each clay was reported by use
of the Langmuir adsorption isotherm (18, 27).
pH-dependent ion release
In a second experiment, four samples (Kenya-1, Kenya-2,
Nigeria, and Zambia) were used to assess the pH dependence
of ion release in the absence of tannic acid. Two hundred mil­
ligrams of the clay was shaken for 2 h with 30 mL 0.1 mol NaCl/
L solution adjusted to pH 1, 2, 3, or 4 with a constant ionic
strength of 0.1. The suspension was centrifuged (600 X g for 10
min), filtered, and analyzed spectrophotometrically as above for
aluminum, iron, magnesium, calcium, potassium, manganese,
copper, and zinc. When the concentrations of the elements were
low, a graphite furnace was used. Each experiment was dupli­
cated.
Results
Mineral composition and total chemical analysis
The results of the mineralogical analysis of the clay samples
are presented in Table 4. All the samples were characterized by
a dominance or abundance of kaolinite in the clay fraction. The
clays used with acorns in California and Sardinia also contain
minerals such as smectite, mica, and quartz, usually with a small
amount of aluminum-hydroxy interlayered vermiculite, Ver­
miculite, amphiboles, and feldspars. Quartz and feldspars are
more abundant in the sand and silt fractions than in the clay
fractions.
The mineralogy of the eight samples from Africa is dominated
by kaolinite, with the exception of the Kenya-2 sample. Mica,
quartz, and smectite were found in small amounts. These clays
come from a geographical location where the weathering time
is long enough to produce soils with only kaolinite as the major
mineral. Thus, major kaolinite deposits are common in this re­
gion. In these samples quartz is usually dominant or abundant
in the coarse fraction (sand and silt).
The total chemical analysis of mineral elements is presented
in Table 3. In general, silicon and aluminum are the dominant
elements in these clays. They are the basic elements of any clay
phyllosilicate minerals such as kaolinite, smectite, mica, ver­
miculite, and chlorite. Where quartz is abundant, most of the
silicon is attributed to this mineral.
The silicon content is generally assumed to be part of the solid
phase of aluminum-silicate minerals and should not easily be
released into solution in a sorption-exchange experiment. Alu­
minum could occur in aluminosilicate minerals or in the form
of aluminum oxides. Iron is usually in the oxide form in these
clays and can be dissolved by increasing acidity. This is especially
true for three samples from Lowie Museum in which hematite
(a-Fe20 3) was detected.
Most of the magnesium, potassium, calcium, and sodium oc­
curs in exchangeable forms or as part of specific minerals. Mag­
nesium occurs in smectite, and potassium, calcium, and sodium
occur in feldspars.
Sorption o f tannic acid
The sorptive maxima and affinity constants for the two clays
consumed with acorns and standard samples of kaolin and
Downloaded from https ://academic.oup.com/ajcn/article-abstract/53/2/448/4694247
by guest
on 15 March 2018
451
montmorillonite are recorded in Table 1. Data for the African
clays are recorded in Table 2. Sorption maximum is a measure
of the maximum amount of an organic molecule that can be
sorbed by the clay sample. Values of sorption maxima from
both groups varied between 3.9 and 23.7 mg/g. Georgia kaolinite
had a value of 2.5 mg/g, and the montmorillonite standard had
a sorptive maximum of 21.2 mg/g. The clays used by the Porno
in California varied in sorptive maxima between 5.6 and 23.7
mg/g. The Porno clay used in the following mineral-release ex­
periments with a value of 6.8 mg/g was one of the poorest sorbers
of tannic acid in this group. The acorn clay from Sardinia had
a sorptive maximum of 13.5 mg/g.
Mineral-element release
The presence of tannic acid had no significant effect on the
release of calcium, potassium, and aluminum. Only the quan­
tities of these minerals released without tannic acid (at pH 2.0
with 0.1 N NaCl) are reported (Table 5). Included in this table
is the amount of iron released with 40 mg tannic acid/g clay
and in the absence of tannic acid. The relationship of iron re­
leased to the amount of tannic acid adsorbed is shown in Figure
1. Although the Pomo clay released 0.08 mg Fe in the presence
of tannic acid, the complex formed with tannic acid would make
this iron unavailable as a nutrient.
The release at pH 2.0 of magnesium, manganese, copper, and
zinc from four of the African clays is shown in Table 6. As pH
decreased from 4 to 1, the concentrations of most elements re­
leased into solution from these four clays increased, with iron
increasing the most dramatically. The effect of pH on the release
of calcium, magnesium, iron, and zinc for the clay from Zambia
and clay 2 from Kenya is shown in Figure 2. The availability of
cations from these clays is in part a function of cations that are
already exchangeable as well as a function of the dissolution of
the solid phase. Dissolution will increase as pH decreases be­
low 4.0.
Aluminum is released at low pH values. Although this ion is
a nonnutrient, it was shown to form aluminum-tannate com­
plexes with tannic acid in solution that can be adsorbed subse­
quently by the clay (28). In other words, aluminum may be
involved in the adsorption of tannic acid. Alternatively, con­
sumption of aluminum in the amounts present in these clays
may have adverse affects on human health (29).
Discussion
Detoxification
Tannins are one of the most ubiquitous toxins in the plant
kingdom. They are toxic to rats in amounts > 8% of the diet
and, as plant defenses, are strongly deterrent to most mammals
in amounts > 2% of fresh weight (30). In moderate amounts,
however, they may have a positive role in normal dietary ecology
(3). Acorns vary in their tannic acid content in the range of 0.5-
8% fresh weight (31).
Recipes used by the Pomo called for clay:acom ratios of 1:
10-1:20 by volume ( ~ 1:5—1:10 by weight) (21 ). California clays
have sorption maxima for tannic acid between 5.6 and 23.7 mg/
g. If clays and acorns are consumed simultaneously without ad­
ditional processing, in the optimum circumstance (ie, when the
solution concentration of tannic acid equals maximum adsorb-
ance) 1 g clay could adsorb 23.7 mg tannic acid from 10 g acorns.
 452 JOHNS AND DUQUETTE
TABLE 4
Mineralogy of clay samples*

Sample source and

fraction KAOL SMEC CHLO VERM HIV MICA QUAR FELD AMPH PYR HEM

C a lifo rn ia , H o p la n d
T o ta l XXX X —- xx xxxx XXX
<2 xxxx XX x x XX — — —
L o w ie M u s e u m
L M 1 -4 4 7 1
T o ta l XXX XXX XXX XX XX
<2 pm XXX XXX XX XX XX — —
L M 1 -2 9 0 7 3
T o ta l XXX xxxx — — xx XXX XX — — —

<2 pm XXX xxxx — — xx XX XX XXX — —

L M 1 -2 9 0 7 4
T o ta l XX — — — xx XXX XX xxxx — XX
<2 pm xxxx — XX XX XXX XX — XX — —
L M 1 -2 9 0 7 5
T o ta l XXX XX — — XX xxxx — — — XX
<2 pm xxxx XX — — XX XX X — — —
LM 1 -2 9 0 7 6
T o ta l XXX XXX — — XX XXX XX XX — —

< 2 Mm xxxx XXX XX XX XXX — —

LM 1 -7 9 8 8 0
< 2 Mm xxxx — XX XX XX X — — —
L M 1 -1 4 4 9 3 5
T o ta l XX — xxxx — — XX XX
< 2 Mm — xxxx XXX — — — — XX XX — —
S a rd in ia , B a u n e i
T o ta l XXX — XX — — XX xxxx XX — — —
< 2 Mm XXX X XX — — XXX XX X — — —
S a rd in ia , T o rto li
T o ta l XXX — — — — XX xxxx xxxx —
< 2 Mm XXX XXX — — — XXX X — —
C am ero o n
< 2 Mm xxxx X — — — XX
G abon
T o ta l xxxx — XXX — — — —
< 2 Mm xxxx —

K e n y a -1
T o ta l XXX — XXX — — — —
< 2 Mm xxxx — — — — XX
K en y a-2
_ _ _
T o ta l XXX — xxxxt
< 2 Mm XXX — — xxxxt — — —
N ig e r ia
T o ta l xxxx — XX XXX —
< 2 Mm xxxx — XX X — — —
Togo
T o ta l xxxx —

< 2 Mm xxxx — X
Z a m b ia
T o ta l XXX — XX xxxx — — — —
< 2 Mm xxxx — XX X
Z a ir e
T o ta l xxxx — XXX
< 2 Mm xxxx — X

* KAOL, kaolinite; SMEC, smectite; CHLO, chlorite; VERM, vermiculite; HIV, Al-hydroxy interlayered vermiculite; MICA, mica; QUAR,
quartz; FELD, feldspar; AMPH, amphiboles; PYR, pyroxene; HEM, hematite, xxxx, dominant; xxx, abundant; xx, present (2-15%); x, trace (0-
2%); and —, absent,
t Sanidine (KAlSi30 8).

Downloaded from https ://academic.oup.com/ajcn/article-abstract/53/2/448/4694247

by guest
on 15 March 2018
 FUNCTIONS OF GEOPHAGY 453
TABLE 5
Release of calcium, potassium, aluminum, and iron

Fe (4 g
Clay source Ca K A1 Fe tannic acid)

mg/g clay

00
California, Hopland 708 ± 5 223 ± 12 244 ± 3 3± 1 27

+1
Sardinia, Baunei 2099 ± 90 109 ± 1 120 ± 2 5± 5 11 ± 5
Cameroon 77 ± 1 45 ± 2 479 ± 0.4 9± 0.4 10 ± 0.2
Gabon 68 ± 2 87 ± 1 520 ± 0.4 4± 0.2 9± 0.4
Kenya-1 791 ± 16 432 ± 18 170 ± 9 7± 1 108 ± 2
Kenya-2 220 ± 2 793 ± 7 536 ± 9 12 ± 0 178 ± 10
Nigeria 19 ± 4 102 ± 2 487 ± 8 10 ± 2 10 ± 2
Togo 120 ± 4 177 ± 10 215 ± 8 5± 0.5 7± 0.4
Zambia 142 ± 3 93 ± 4 56 ± 1 74 ± 0 212 ± 1
Zaire 16 ± 1 84 ± 12 231 ± 8 497 ± 30 905 ± 17

If all the tannic add from acorns with tannic add at the deterrent
level of 2% was in solution, ^ 12% of this would be adsorbed.
The Sardinian recipe uses clay and acorns in the ratio of —1:8
by weight (21). Based on a sorptive maximum of 13.5 mg/g,
^8% of the total tannic acid in unpalatable acorns (with 2%
tannic acid) would be adsorbed. The Sardinian bread, which is
made from the suspension of a soil sample stirred in an excess
of water after the removal of the sand particles, contains only
the clay and silt fractions and in practice it would have higher
sorption than the whole sample as we used it.
Sorption of organic compounds by clays is affected by pH,
ionic strength, and the presence of competing adsorbents. Under
the complex physical and chemical conditions encountered dur­
ing cooking, mastication, and digestion, the amount of sorption
of tannic acid by either clay is likely less than that predicted by
the sorption maxima.
The data presented here do not strongly support the hypothesis
that clays consumed directly with plant foods without prior pro­
cessing adsorb tannic acid sufficiently to reduce this compound
from a toxic to a safe level. Increased palatability may be a more
appropriate appraisal of what can occur than detoxification in
a narrow sense. It is, in fact, unusual for humans to consume a
diet with toxic levels of tannic acid. These compounds have,
Iron release <ug/g of clay)
however, a strong astringent effect on proteins of saliva and oral
membranes and their presence is detected by the trigeminal nerve
(32). Realistically, only a small portion of the tannic acid in a
food should pass into solution during the short mastication pe­
riod; palatability should, therefore, be greater than one might
assume because a relative excess of clay in solution would adsorb
the small portion of soluble tannic acid that would be available
to interact with oral tissues.
Although the Californian, Sardinian, and African clays would
not adsorb tannic acid at the toxic levels that are released during
prolonged digestion in the gastrointestinal tract, the traditional
processes for acorn-bread preparation involving temperatures
^ 100 °C for several hours reduce tannic acid activity by several
times what can be accounted for by the adsorptive maxima alone.
In fact, the ancillary data from digestion of acorn breads suggest
that cooking acorns with clay reduces the toxicity of acorns
caused by protein precipitation by as much as 77% [48 mg/g
acorn (21 )]. Heating may enhance the adsorption but probably
contributes to a catalytic activity of clay in the decomposition
of tannic acid (33). Interactions of tannic acid with iron or alu­
minum may contribute to the considerable detoxification that
is observed empirically.
We chose a case study involving tannins, in part because of
the ecological importance of tannins as a deterrent to primates
(34) and other animals (35). In fact, the capacity of all of the
clays tested to adsorb tannic acid is small relative to chaiged
compounds such as the glycoalkaloids, which are adsorbed in
the potato case mentioned above ( 18). However, in the specific

TABLE 6
Release of magnesium, manganese, copper, and zinc from African
clays

Mineral release

Gay source Mg Mn Cu Zn

mg/g clay
Kenya-1 135 ± 2 63 ± 2 2 ±0.1 3 ±0.1
Kenya-2 112 ±0.5 349 ± 12 1 ±0.4 5 ±0.1
Nigeria 9± 1 0± 0 2± 1 3 ±0.1
Zambia 60 ± 1 19 ± 1 11 ± 1 2± 1
FIG 1. Effect of tannic acid on iron release.

Downloaded from https://academic.oup.eom/ajcn/article-abstract/53/2/448/4694247

by guest
on 15 March 2018
 454 JOHNS AND DUQUETTE
case of acorn bread, the properties of the clays make them ef­
fective for what they were intended. Both the acorn bread and
the potato case studies, which compare parallel behaviors of
humans in distinct locations, support the detoxification hypoth­
esis in a general sense. However, other data suggest that a nu­
trition function for the clays should not be dismissed.
Nutrient release
Acorn clays. Data on the release of minerals from the acorn
clays suggest that these clays contribute calcium to the diet. The
Sardinian clay released 2.10 mg/g clay, by far the highest value
from any material studied. The total calcium content of the
sample is 5.72 mg/g (total chemical analysis). This means that
under the experimental conditions, 37% of the total calcium can
be released into solution and be potentially available as nutrients.
The nutrient release values recorded in Tables 5 and 6 are
theoretical maxima at pH 2.0 and are representative of gastric
conditions. Biological availability of minerals may be affected
by various factors. Except for copper, the minerals discussed
here are absorbed in the small intestine, where pH is higher and
mineral release undoubtedly less than in the stomach.
The California clay from Hopland released 0.71 mg Ca/g,
more than all but one of the African samples. The fact that
acorn breads were the primary staple [“the staff of life” (36)] of
the indigenous people of California suggests that they could pro­
vide significant amounts of calcium. To meet 10% of an 800-
Downloaded from https ://academic.oup.com/ajcn/article-abstract/53/2/448/4694247
by guest
on 15 March 2018
mg recommended dietary allowance (RDA) one would need
0.5-1.0 kg acorns (depending on the portion of clay used) in the
form of bread (37). Daily consumption of > 1 kg of a staple
cooked food is not unreasonable in subsistence societies. Acorns
are reported to have made up 25% of the diet of poor people in
Spain and Italy in the past (38).
Sardinian bread made from 1 kg acorns would contain 145 g
clay and 305 mg Ca, 38% of the adult RDA. The Sardinian
bread was made by boiling acorns to dryness with the supernatant
of a large amount of clay stirred in an excess of water. Thus, the
actual amounts of calcium and other minerals present in the
product were probably considerably higher than those originally
present in the actual weight of clay found in the final product.
In summary, geophagy in the case of acorn breads would ap­
pear to offer a nutritional benefit.
African clays. The African clays, likewise, offer specific mineral
nutrients in sufficient quantities to have an impact on dietary
status in some cases. Vermeer and Ferrell (39) estimated that
Ewe women in Ghana consume an average of 30 g clay/d but
that 300 g and, commonly, > 150 g are eaten by some persons.
Pregnant women in Guatemala consume between 39 and 232
g/d ( 12). The clay from Zaire would supply > 100% of the RDA
of iron for anyone eating ^ 20 g/d. Other clays, on the contrary,
offered little available iron; 7 g of Kenya-2, 40 g of Kenya-1,
and 133 g of the Zambian sample would meet 100% of man­
ganese needs; 30 g of the Zambian clay and 150 g of the clays
from Nigeria and Kenya (ie, Kenya-1) would provide 15% of
the RDA for copper; 150 g of Kenya-1 would provide 15% of
the RDA for calcium; 150 g of each of the African clays would
provide 5-7.5% of the RDA for zinc; and 150 g of the Kenyan
clays would provide 8-10% of the RDA for magnesium. The
pattern we see of certain clays (but not others) contributing im­
portant levels of some minerals is consistent with the work of
Hunter (11) and Hunter and deKleine (12).
The samples from Gabon and Nigeria are probably both ex­
amples of the clays from Calabar, Nigeria, that are widely dis­
tributed in West Africa. These clays have aluminum as their
major exchangeable cation and are comparatively poor in min­
erals of nutrient importance. Their wide distribution in Africa
is better explained on other than nutrition grounds (40).
The etiology o f geophagy
Nothing supports a notion of a specific appetite for any nu­
trient as an acceptable explanation for geophagy. Although cer­
tain clays undoubtedly make important dietary contributions,
if nutrient gain is other than a random event, the consumption
of clays in specific contexts is best seen as part of an empirical
adaptation of a group of people to a particular ecological setting
over many generations.
It is difficult to imagine how such empirical knowledge is ob­
tained. Specific appetite is an unsatisfactory explanation, and
conditioned preferences, which are generally weak, have not been
demonstrated for mineral nutrients. The cravings associated with
geophagy during pregnancy suggest, however, that unrecognized
nutrient appetites may exist. However, it is significant that nu­
trient benefit is not explicitly important in the rationales given
for the consumption of clays in traditional contexts. Even in
cases where geophagy appears to make a dietary contribution,
such as during pregnancy, its etiology is associated with treatment
of nausea (12). The clay from Zaire was sold as a stomach med­
icine.
 FUNCTIONS OF GEOPHAGY
In both the acorn and the African case studies, detoxification
broadly defined is a more satisfactory explanation for human
behavior because of its more overt purpose. In the first situation,
the purpose of improving palatability, at least, is explicitly stated.
In the African case, possible mechanisms of detoxification cor­
responding to the stated uses of the clays are not easily tested.
Nonetheless, the common uses of clays to combat vomiting and
nausea in pregnancy and to counter diarrhea (40) suggest a de­
toxification function for geophagy in this case.
The neutralization of toxins by clays in scenarios besides those
involving dietary constituents form the basis for several related
explanations of geophagy in humans and animals. Kaopectate
(The Upjohn Co of Canada, Don Mills, Ontario), which is pre­
dominately kaolin, serves an antidiarrheal purpose in indus­
trialized societies. The role of clay in these situations is in part
to adsorb toxins and bacteria (41). Bentonite is a common ad­
dition to cattle feed and has been shown to prevent fungal toxicity
in rats (42). Geophagy as well serves to alleviate upset of the
gastrointestinal tract by countering acidosis (14).
Geophagy and the evolution o f diet and medicine
The clay recovered from a site occupied by ancestors of H.
sapiens has mineralogical, nutritional, and pharmacological
characteristics compatible with and essentially indistinguishable
from those of clays consumed by humans in Africa today. This
circumstantial evidence, in conjunction with the use of clays by
higher primates, suggests that geophagy could have been part of
the subsistence activities of early hominids and has been a con­
stant behavior throughout the evolutionary period that deter­
mined human dietary habits.
Although nutritionists and other observers have tended to view
geophagy (and pica in general) as compulsive or driven by crav­
ing, as a medicine to alleviate discomfort, clay is not different
from any drug that humans in industrial and preindustrial so­
cieties take in attempting to modify their physiological state.
Clays could adsorb dietary toxins, bacterial toxins associated
with gastrointestinal disturbance, hydrogen ions in acidosis, or
metabolic toxins such as steroidal metabolites associated with
pregnancy (43). All of these sources of disturbance result in
common symptoms of nausea, vomiting, and diarrhea. Gas­
trointestinal upset may form part of the etiology of geophagy,
which may be linked in a nonspecific way with the role of nausea
in preventing toxicity (44). Geophagy has been posed as a sen­
sitive index of gastrointestinal malaise in rats (16) and it may
be closely tied with the conditioned aversion response that is a
classical and powerful means by which omnivores avoid toxicity
(15). An involvement of conditioned learning in human geo­
phagy could explain why some humans engage in geophagy and
others do not. Conditioned learning takes place at a subconscious
level and it is thus understandable that geophagy may seem
compulsive and inexplicable.
Geophagy is a behavioral trait humans share with many
members of the animal kingdom. In a medical model it can be
understood as a component of the pharmacophagous (45) be­
havior of animals that is the basis for the phenomenon of med­
icine that is essential to all human cultures (3). Rudimentary
forms of self-medication or pharmacophagy are identifiable in
the ingestion by chimpanzees (46) and other mammals (47) of
plant allelochemicals with the apparent purpose of treating gas­
trointestinal parasites. Ingestion of plant products that result in
Downloaded from https://academic.oup.eom/ajcn/article-abstract/53/2/448/4694247
by guest
on 15 March 2018
455
behavioral and mood changes is another type of pharmacophagy
with roots in animal behavior (47).
From an evolutionary perspective, the dietary and medical
models find a commonality that is manifest in the case of geo­
phagy. Nutrients and toxins are encountered together in the diets
of animals, and they interact to affect nutritional status (48). A
pharmacophagous behavior uses nonnutrient components that
are inevitably present in the diet in a beneficial way; in certain
circumstances, obtaining these benefits becomes the primary
reason for ingesting a substance. Geophagy facilitates a favorable
balance between nutrients and toxins. Geophagy can be a source
of nutrients. Its primary way of enhancing nutritional status
appears to be, however, to counter dietary toxins and, second­
arily, the effects of gastrointestinal parasites.
We believe geophagy plays a useful role in its proper context
and should be appreciated as a normal human behavior. In tra­
ditional societies it is usually a nonpathologic activity that in
appropriate circumstances can contribute to human health. Like
any behavior, it can be inappropriate when engaged in out of
context or to excess. No doubt pathological cases of nutrient
deficiency (cf Ref 5), ingestion of environmental pollutants, and
parasitic transmission can result from ingesting soils. In the in­
dustrial world many behaviors must be modified in accordance
with the conditions of modem lifestyles. But even in this context,
geophagy is not inappropriate with a commercial antidiarrheal
preparation such as Kaopectate. The distaste with which many
of us regard geophagy is culturally based. Its sound biological
basis, however, has made geophagy important in the evolution
of human dietary behavior. E3
We thank J Desmond Dark for the clay from Kalambo Falls, Lawrence
Dawson and the Lowie Museum for the samples of Porno clays, Dan
Harder for the clay from Zaire, and Antoinette Lopez for the clay from
Togo. We gratefully acknowledge the assistance of Francis Jack in Hop-
land, Pasquale Zucca in Sardinia, Elizabeth Achola in Kenya; the advice
of Philip Salzman and Harriet V Kuhnlein; and the technical assistance
of Aghdas Zamani and Farhad Saboohi.
References
1. Danford DE. Pica and nutrition. Annu Rev Nutr 1982;2:303-22.
2. Läufer B. Geophagy. Field Museum Natl Hist Anthropolog Ser
1930;18:99-198.
3. Johns T. With bitter herbs they shall eat it: chemical ecology and
the origins of human diet and medicine. Tucson, AZ: University of
Arizona Press, 1990.
4. Hladik CM, Gueguen L. Geophagy and mineral nutrition in wild
primates. CR Acad Sei [D] (Paris). 1974;279:1393-6 (in French).
5. Danford DE, Smith JC, Huber AM. Pica and mineral status in the
mentally retarded. Am J Clin Nutr 1982;35:958-67.
6. Halstead JA. Geophagia in man: its nature and nutritional effects.
Am J Clin Nutr 1968;21:1384-93.
7. Woodside B, Millelire L. Self-selection of calcium during pregnancy
and lactation in rats. Physiol Behav 1987;39:291-5.
8. Kreulen DA, Jager T. The significance of soil ingestion in the uti­
lization of arid rangelands by large herbivores, with special reference
to natural licks on the Kalahari pans. In: Gilchrist FMC, Mackie
RI, eds. Herbivore nutrition in the subtropics and tropics. Jo-
hannesburg. South Africa: Science Press, 1984:204-21.
9. Richter CP. Increased salt appetite in adrenalectomized rats. Am J
Physiol 1936;115:155-61.
10. Rozin P. The selection of food by rats, humans and other animals.
Adv Stud Behav 1976;6:21-76.
11. Hunter JM. Geophagy in Africa and in the United States: a culture—
nutrition hypothesis. Geogr Rev 1973;63:170-95.
 456 JOHNS AND DUQUETTE
12. Hunter JM, de Kleine R. Geophagy in Central America. Geogr Rev
1984;74:157-69.
13. Vermeer DE. Geophagy among the Tiv of Nigeria. Ann Assoc Am
Geogrs 1966;56:197-204.
14. Kreulen DA. Lick use by large herbivores—a review of benefits and
banes of soil consumption. Mammal Rev 1985;15:107-23.
15. Mitchell D, Winter W, Morisaki CM. Conditioned taste aversions
accompanied by geophagia: evidence for the occurrence of “psy­
chological” factors in the etiology of pica. Psychosom Med 1977;39:
402-12.
16. Watson PJ, Hawkins C, McKinney J, Beatey S, Bartles RR, Rhea
K. Inhibited drinking and pica in rats following 2-deoxy-D-glucose.
Physiol Behav 1987;39:745-52.
17. Uehara S. Seasonal changes in the techniques employed by wild
chimpanzees in the Mahale Mountains, Tanzania, to feed on termites
(Pseudacanthotermes spiniger). Folia Primatol (Basel) 1982;37:44-
76.
18. Johns T. Detoxification function of geophagy and the domestication
of the potato. J Chem Ecol 1986;12:635-46.
19. Davies AG, Baillie IC. Soil-eating in red leaf monkeys (Presbytis
rubicunda) in Sabah, Northern Borneo. Biotropica 1988;20:252-8.
20. Oates JF. The guerza and its food. In: Clutton-Brock TH, ed. Primate
ecology: studies of feeding and ranging behaviour in lemurs, monkeys
and apes. London: Academic Press, 1977,275-321.
21. Johns T, Duquette M. Traditional detoxification of acorn bread
with clay. Ecol Food Nutr (in press).
22. Irvine FR. Supplementary and emergency food plants of West Africa.
Econ Bot 1952;6:23-40.
23. Clark JD. Kalambo Falls prehistoric site. Vol 1. London: Cambridge
University Press, 1969.
24. Sampson CG. The stone age archaeology of South Africa. New York:
Academic Press, 1974.
25. Lavkuhlich LM, Weins JH. Comparison of organic matter destruc­
tion by hydrogen peroxide and sodium hypochlorite and its effect
on selected mineral constituents. Soil Sei Soc Am Proc 1970;34:
755-8.
26. Whittig LD. X-ray techniques for mineral identification and min-
eralogical composition. In: Black CA, ed. Methods of soil analysis.
Part I. Agronomy. Madison, WI: American Society of Agronomy,
1965:671-98.
27. Kinniburgh DG. General purpose adsorption isotherms. Environ
Sei Technol 1986;20:895-904.
28. Goh TB, Huang PM. Influence of citric and tannic acids on hydroxy-
A1 interlayering in montmorillonite. Clays Clay Min 1986;34:37-
44.
Downloaded from https ://academic.oup.com/ajcn/article-abstract/53/2/448/4694247
by guest
on 15 March 2018
29. McLachlan DRC, Famell BJ. Aluminum in human health. In: Havas
M, Jaworski JF, eds. Aluminum in the Canadian environment. Ot­
tawa: National Research Council Canada, 1986:153-73.
30. Freeland WJ, Janzen DH. Strategies in herbivory by mammals: the
role of plant secondary compounds. Am Nat 1974;108:269-89.
31. Ofcarcik RP, Bums EE. Chemical and physical properties of selected
acorns. J Food Sei 1971;36:575-8.
32. Silver WL. The common chemical sense. In: Finger TE, Silver WL,
eds. Neurobiology of taste and smell. New York: Wiley, 1987:65-
87.
33. Theng BKG. The chemistry of clay-organic reactions. New York:
Wiley, 1974.
34. McKey D, Waterman PG, Mbi CN, Gartlan JS, Struhsaker TT.
Phenolic content of vegetation in two African rain forest: ecological
implications. Science 1987;202:61-4.
35. Harborne JB. Introduction to ecological biochemistry. 3rd ed. Lon­
don: Academic Press, 1988.
36. Barrett SA. Material aspects of Porno culture. Part one. Bull Pub
Mus City Milwaukee. 1952;20:1-260.
37. National Research Council. Recommended dietary allowances. 10th
ed. Washington, DC: National Academy Press, 1989.
38. Hill AF. Economic botany. New York: McGraw-Hill, 1937:374-5.
39. Vermeer DE. Geophagy among the Ewe of Ghana. Ethnology
1971;10:56-72.
40. Vermeer DE, Ferrell RG. Nigerian geophagical clay: a traditional
antidiarrheal pharmaceutical. Science 1985;227:634-6.
41. Gilman AG, Goodman LS, Rail TW, Murad F, eds. The pharma­
cological basis of therapeutics. 7th ed. New York: Macmillan, 1985.
42. Carson MS, Smith TK. Role of bentonite in prevention of T-2 tox­
icosis in rats. J Anim Sei 1983;57:1498-1506.
43. Weigel MM, Weigel RM. The association of reproductive history,
demographic factors, and alcohol and tobacco consumption with
the risk of developing nausea and vomiting in early pregnancy. Am
J Epidemiol 1988;127:562-70.
44. Pelchat ML, Rozin P. The special role of nausea in the acquisition
of food dislikes by humans. Appetite 1982;3:341-51.
45. Boppré M. Redefining “pharmacophagy.” J Chem Ecol 1984; 10:
1151-4.
46. Wrangham RW, Nishida T. Aspilia spp. leaves: a puzzle in the feeding
behavior of wild chimpanzees. Primates 1983;24:276-82.
47. Janzen DH. Complications in interpreting the chemical defenses of
trees against tropical arboreal plant-eating vertebrates. In: Mont­
gomery GG, ed. The ecology of arboreal folivores. Washington, DC:
Smithsonian Institution Press, 1978:73-84.
48. Hathcock JN, ed. Nutritional toxicology. Vol II. Orlando, FL: Ac­
ademic Press, 1987.