untitled
144 ACTA BOT. CROAT. 75 (1), 2016
Acta Bot. Croat. 75 (1), 144–148, 2016 CODEN: ABCRA 25
DOI: 10.1515/botcro-2016-0019 ISSN 0365-0588
eISSN 1847-8476
A comparison of the infl uences of fl otation and
wet sieving on certain carbonized legume and
cereal remains
Sara Mareković*, Renata Šoštarić
Department of Biology, Faculty of Science, University of Zagreb, Marulićev trg 20/2, HR-10000 Zagreb, Croatia
Abstract – In order to determine the infl uence of recovery techniques with water (fl otation and wet sieving)
on carbonized plant remains, a certain amount of wheat, barley, millet, horsebean and lentil macrofossils from
archaeological sites was taken and treated with water. Moist recovery was also applied to in-laboratory, artifi -
cially, charred barley, millet and lentil samples. After the treatments, the investigated remains were re-count-
ed and the percentages of still recognizable remains for every plant species and for each method were re-
corded. Comparisons were made of the sensitivities of the investigated species and of the differences in the
degree of macrofossil breakup depending on the method of recovery. Our investigation proved that fl otation
is a less aggressive method than wet sieving and that barley, horsebean and wheat carbonized macrofossils
are resistant to moist treatments, while the breakup percentage of lentil and millet (from archaeological sites)
is higher than 30%, which should be taken into account when deciding on the (non)use of water recovery in
the investigations.
Key words: carbonized plant remains, cereals, fl otation, legumes, wet sieving
* Corresponding author, e-mail: sara.marekovic@biol.pmf.hr
Introduction
At archaeological investigation sites, botanical (as well
as zoological and archaeological) material is often collected
together with large amounts of soil and before analysis of
that material, it is necessary to conduct a recovery process
in order to remove the soil and to make sample analysis
easier and faster. The most common recovery methods use
water: fl otation or wet-sieving (water screening). In water
screening, soil is placed on a screen and washed with a wa-
ter jet. All the particles smaller than the screen mesh are
washed through the screen and the larger materials (like bo-
tanical fi ndings) are recovered and later analyzed. Flotation
is based on the difference in density between organic and
inorganic material. In the fl otation system, soil is poured
into a body of water. Agitation breaks up the soil, allowing
light materials (seeds, charcoal) to fl oat and heavier materi-
als (pebbles, potsherds) to sink. Archaeobotanists analyse
the organic material collected from sieves with different
mesh sizes, to which the water with fl oating material is di-
rected (Jacomet and Kreuz 1999, Pearsall 2000). Wagner
(1982) in his paper introduces the poppy seed recovery test
for determining the effectiveness and consistency of any
particular botanical fl otation system. This is important be-
cause it helps investigators to be aware of the recovery rate
of found remains in their fl otation devices.
The aim of this study is to examine whether the wet siev-
ing destroys carbonized plant remains more than the fl otation
method and besides that, to investigate whether there is a dif-
ference in sensitivity among the species horsebean, lentil,
wheat, barley and millet. In our previous studies, we acci-
dentally noticed that the lentil has greater sensitivity to water
treatments than wheat and barley grains and therefore our
goal was to verify whether this difference is constant and to
what extent the difference between the species exists. Tanno
and Willcox (2006) hypothesised that the rarity of faba bean
fi nds in their investigation might have been due to their fra-
gility (and devastation during fl otation and transportation).
Because of the limited access to real archaeological material
not previously treated with water, in this study we could not
include all the legumes and cereals which could theoretically
be found on archaeological sites. But the species that were
analyzed in this paper, are certainly some of the most com-
mon species found in samples, so we believe that this analy-
sis will give some important answers to archaeobotanists,
when the need to make a decision about the right water re-
covery method for their sample or want to discuss the ratio of
different carbonized fi ndings previously water recovered.
A COMPARISON OF THE INFLUENCES OF FLOTATION AND WET SIEVING
ACTA BOT. CROAT. 75 (1), 2016 145
Wright (2005) concluded that the results of archaeobo-
tanical analysis depend on fl otation sample size, how the
sample is measured and processed and how well the plant
material within the sample withstands the rigors of fl ota-
tion. Some other previous studies have also shown that wet
sieving affects plant residues of different species different-
ly, so Hosch and Zibulski (2003) demonstrated in their pa-
per that the number of fragile remains (e. g. cereal chaff re-
mains and Malus sylvestris Mill. pericarp fragments) were
reduced by increasing wet-sieving intensity, but more or
less round and robust remains were scarcely affected. They
suggested that a less destructive technique for fragile plant
macroremains should be used, although the researchers will
need more time for the recovery of the sample. Tolar et al
(2010) did an experiment with macroremains found in the
Neolithic pile dwelling site of Stare Gmajne, Ljubljansko
Barje (Slovenia). They compared three subsamples which
were differently rough-handled during the wet sieving and
their study showed that some sensitive species were almost
completely destroyed and no longer recognizable. This
study clearly shows that the proportion of species found in
the sample can largely depend on the method of recovery of
the sample and that this should be considered when select-
ing methods and in the interpretation of results. We are
aware that the above mentioned studies deal with water-
logged sediments, which are certainly not the same as min-
eral soils, but there are also some similarities that can help
us draw conclusions about recovery techniques.
It should be taken into account that the plant remains
can be damaged not only by the mechanical force of water
rinsing, but also because of the contact of the carbonized
material with the water and during the drying processes,
which affects the mechanical structure and cracking of the
fragile carbonized structures. Pearsall (2000) describes the
work of Jarman et al. (1972), which analyses the percentage
of destroyed carbonized macroremains after repetition of
the fl otation and drying. Their experiment showed that after
the fi rst fl otation and drying, only 4% of the sample was
destroyed. However after the second repetition of the pro-
cedure, an additional 56% of the sample was destroyed and
the third fl otation with drying destroyed the remaining car-
bonized macrofossils. Badham and Jones (1985) also con-
fi rmed that carbonized are far more sensitive than mineral-
ized remains and therefore water screening should be
avoided for the sake of fi nding such remains.
Materials and methods
Species that we have studied are: horsebean (Vicia faba
L.), lentil (Lens culinaris Medik.), wheat (mix: Triticum
turgidum ssp. dicoccon (Schrank) Thell.+ T. aestivum ssp.
aestivum L. + T. aestivum ssp. spelta (L.) Thell. + Triticum
turgidum ssp. dicoccon/T. aestivum ssp. spelta), barley
(Hordeum vulgare L.) and millet (Panicum miliaceum L.).
For all these species we took 300 well-preserved whole
grains from the Late Bronze Age sites Kalnik-Igrišče
(Mareković et al. 2015), for both pretreatments that we ex-
plored. Additionally, we have treated 180 carbonized lentil
seeds from the Roman site Poreč (which was explored in
2008) and 100 horsebean seeds from Bronze Age settle-
ments Nova Bukovica (Šoštarić 2001).
As we had just one additional (beside those from
Kalnik-Igriše site) archaeological carbonized sample, for
horsebean and lentil species only, in the laboratory we
made 6 extra samples with 30 grains for millet, barley and
lentil species. Millet and lentils were chosen because they
in previous research they showed high rates of destruction
during moist treatments, and barley was chosen as the con-
trol species, because it had shown great resistance to the
treatments. The material was charred in a muffl e furnace.
The barley and lentil species were heated up to 500 °C, for
3 or 10 minutes. Millet was heated up to 300 °C, for ap-
proximately 10 minutes. In our experiment we found the
best heating temperature and duration by consulting litera-
ture (Wright 2003, Babrauskas 2013) and by using the
method of trial and error for each plant species. For each
plant we took 3 samples for fl otation and another three for
wet sieving.
In the fl otation treatment we put the macroremains in a
container, which was supplied with water via a rubber tube
placed in lower part of the container. A steady supply of
water caused the carbonized plant remains to move around
the container and fi nally overfl ow the container on which
they were collected in a sieve. The entire procedure lasted
between 5 and 10 minutes for each sample.
In the second treatment we put the macroremains on
metal sieve, and rinsed it with the water jet from above. The
strength of the water stream was regulated to medium
strength, and the process lasted for ca 5 to 10 minutes.
Because different soils interact with macrofossils in dif-
ferent ways (some are much more viscous than others, for
example), we deliberately did fl otation and wet sieving
only with carbonized remains, in order to stay focused just
on the fragility of the plant remains of every species, re-
gardless of the infl uence of the soil type.
After rinsing, the macrofossils were left to dry in the
open air. Subsequently they were all examined under the
microscope, and we counted a number of carbonized re-
mains still recognizable after the wet treatments. We calcu-
lated the percentage of recognizable, slightly damaged
plant remains for every species and compared for any sig-
nifi cant difference between our two treatments and among
the different species investigated. Photographs of lentil and
barley demonstrate the differences in plant remains before
and after treatment, and also show grains or seeds that are
either recognisable or unrecognisable (Figs. 1, 2)
The experiment with the plant material carbonized in
the laboratory was, because of absence of decomposition
after the fi rst moistening, repeated 5 and 10 times to check
whether repeated water treatments and drying lead to great-
er destruction of the material. We wanted to see if the artifi -
cially charred remains can show us fragility differences
among species similar to those found in genuinely archaeo-
logical remains, however much they are more resistant due
to their date of origin as well as the absence of pedological
and elemental infl uences that cannot easily be simulated in
laboratory conditions.
MAREKOVIĆ S., ŠOŠTARIĆ R.
146 ACTA BOT. CROAT. 75 (1), 2016
Statistical analysis of the obtained data was performed
with the software package Statistica (ver 8), manufacturers
StatSoft Inc., USA. The aim was to check statistically
whether there is a difference in the proportion (percentage,
labelled p1) of recognizable seed after the fl otation and the
proportion (percentage, labelled p2) of recognizable seed
after the wet sieving treatment. Beside that we wanted to
see whether there are signifi cant differences among species
in the same pretreatment.
For this purpose, we set the null hypothesis H0: p1 = p2
and the alternative hypothesis H1: p1 ≠ p2. We tagged with
n1 the total number of seeds before the fl otation and with
n2 the total number of seeds before the wet sieving treat-
ment. The value of the test statistics is calculated with the
following formula:
where
Beside the value of the test statistic z we needed also the
corresponding p-value which is calculated as follows:
– p = P{Z ≥ z} if alternative hypothesis is in this form
H1 : p1 − p2 > 0,
– p = P{Z ≤ z} if alternative hypothesis is in this form:
p1 − p2 < 0,
where Z is a random variable with a standard normal distri-
bution. P-values were counted with the probability calcula-
tor in the software package Statistica. Calculated p-values
are compared with the level of signifi cance α = 0.05 = 5%
(α is the maximum probability that we will make mistakes,
if we reject the null hypothesis, i.e., maximum probability
that we will reject the null hypothesis if it is correct – so
called error of the fi rst kind) and one of the following two
conclusions is made:
– if p < α we reject the null hypothesis and at the level
of signifi cance α we accept the alternative hypothesis
H1,
– if p > α we conclude that we do not have enough ar-
guments to support the decision to reject the null hy-
pothesis.
Results
The results of treatment comparisons made on macro-
fossils collected from archaeological sites and afterwards
on those carbonized in the laboratory are shown in Figs.
3–5.
Horsebean collected from the Kalnik-Igrišče site showed
the least sensitivity to treatments and as many as 91% of
specimens were recognizable after both pretreatments.
Horsebean is followed by barley grains, preserved 85% af-
ter fl otation and 82.3% after wet sieving. Wheat also proved
to be a resistant macrofossil, because both treatments pre-
served almost three quarters of the grains in a recognizable
form (75.3% after fl otation and 74.7% after wet sieving)
(Fig. 3).
During fl otation as well as during wet sieving the re-
mains most destroyed were the carbonized remains of the
lentil, as only 68.7% of these remains (Kalnik-Igrišče) and
35% (Poreč) were recognizable after fl otation, and, that
percentage decreased after wet sieving to 52% (Kalnik-
Igrišče) and 22.2% (Poreč) (Figs. 4, 5).
p1 p2
1 1
1
n1 n2
z
p' p'
n1 p1 n2 p2
n1 n2
p'
0
10
20
30
40
50
60
70
80
90
100
Vicia faba Hordeum
vulgare
Triticum mix. Panicum
miliaceum
Lens culinaris
Fig. 1. The lentil seeds from the site Kalnik-Igrišče: A) before the
treatment, B) upon wet sieving: the recognizable lentil remains are
on the top and unrecognizable ones on the bottom of fi gure.
Fig. 2. The barley grains from the site Kalnik-Igrišče: A) before
the treatment, B) upon wet sieving: the recognizable barley remains
are on the top and unrecognizable ones on the bottom of fi gure.
Fig. 3. Comparison of percentage of recognizable macrofossils
after fl otation (black bars) and wet sieving (grey bars) on the site
Kalnik-Igrišče.
A COMPARISON OF THE INFLUENCES OF FLOTATION AND WET SIEVING
ACTA BOT. CROAT. 75 (1), 2016 147
The results show that in all cases, except for the horse-
bean sample from the site Kalnik-Igrišče, carbonized mate-
rial is destroyed more by the wet sieving method. The big-
gest difference in destruction beyond recognition, is evident
on the lentil (16.7% and 12.8%) and millet (8.3%) sample.
The results show that for the horsebean sample from
both sites, we cannot claim there is a signifi cant difference
(if we use the level of signifi cance α = 0.05) between the
fl otation and wet sieving treatments. For wheat and barley,
there are no very strong arguments for stating that the pro-
portions (percentages) of recognizable seeds differ accord-
ing to the treatment. In both lentil samples it has been prov-
en that (by the level of signifi cance α = 0.05) wet sieving
damages the carbonized remains more than fl otation. At the
level of signifi cance 0.05, we accepted the hypothesis that
the proportion of recognizable millet seeds is greater after
fl otation than after wet sieving. The samples of millet, lentil
and barley experimentally carbonised in a furnace are much
more resistant and less prone to destruction real archaeo-
logical material.
The experiments with the laboratory-carbonized, mate-
rial showed that the disintegration of the plant remains,
with both recovery methods and regardless of the number
of repetitions, was the biggest in millet samples. Barley
samples were the most resistant and barley grains remained
undamaged during all fl otation and wet sieving procedures
(Tab. 1).
The number of repetitions of the procedures done on the
millet and lentil samples, increased the decomposition of
the grains, and the samples were more damaged after wet
sieving, than after fl otation.
As for the fl otation process and the wet sieving of labo-
ratory-carbonised material (lentils, barley, millet), we can-
not (at the level of signifi cance α = 0.05) prove a statisti-
cally signifi cant difference in the destruction of the material
beyond recognition, regardless of the number of repetitions
of procedures (1, 5 and 10).
However, when the sensitivity of millet is compared
with that of barley, it can be proven (at the level of signifi -
cance α = 0.05) that sensitivity is signifi cantly bigger when
fl otation is repeated 10 times, and with wet sieving after the
fi rst repetition. Therefore it became apparent that wet siev-
ing is a more aggressive procedure and affects the carbon-
ized remains of millet more than fl otation.
When we compared the sensitivity of the lentil in rela-
tion to barley, we came to the conclusion that their sensi-
tivities are not statistically signifi cantly different (at the
level of signifi cance α = 0.05) after fl otation or wet sieving,
even after ten repetitions.
The sensitivity of millet compared to that of lentil is sta-
tistically signifi cantly different (at α = 0.05) only after the
tenth repetition of wet sieving.
Discussion
Laboratory samples of carbonized lentil, millet and bar-
ley clearly showed that recent material, carbonized in a fur-
nace, is more resilient and less prone to disintegration than
real archaeological material. In previous investigations it
has been already noted (Pearsall 1980, King 1987, Goette et
al. 1994, King 1994,) that it is not easy to get lab samples
0
10
20
30
40
50
60
70
80
90
100
Vicia faba Hordeum
vulgare
Triticum mix. Lens culinaris Panicum
miliaceum
Fig. 4. The percentage of recognizable carbonized remains after
fl otation. Black bars represent the results from the site Kalnik-
Igrišče, grey bar shows the results from Nova Bukovica and the
white one from Poreč site.
0
10
20
30
40
50
60
70
80
90
100
Vicia faba Hordeum
vulgare
Triticum mix. Panicum
miliaceum
Lens culinaris
Fig. 5. The percentage of recognizable carbonized remains after
wet sieving. Black bars represent results from the site Kalnik-
Igrišče, the white bar result from Nova Bukovica and the grey bar
the result of the lentil from Poreč site.
Tab. 1. The comparison of the preservation percentage of carbonized macrofossils obtained by methods of repeated fl otation and wet
sieving.
No. of
repetitions
Lentil Barley Millet
Flotation Wet sieving Flotation Wet sieving Flotation Wet sieving
1 x 100 % 100 % 100 % 100 % 97.77 % 96.67 %
5 x 100 % 99.44 % 100 % 100 % 96.67 % 93.33 %
10 x 98.89 % 98.33 % 100 % 100 % 94.47 % 86.67 %
MAREKOVIĆ S., ŠOŠTARIĆ R.
148 ACTA BOT. CROAT. 75 (1), 2016
(corn) the structure and properties of which identical or at
least satisfactorily similar to those found in real archaeo-
logical samples. Brady (1989) investigated the impact of
fl otation on the preservation of laboratory samples of car-
bonized wood and concluded that there is no signifi cant dif-
ference in the conservation of mass during fl otation, regard-
less of the differences in the density of the wood in different
plant species, which can also be result of greater resistance
of wood in the case of laboratory-carbonized material.
Therefore we presume that although it is impossible to get
accurate numerical data on the rate of destruction of some
plant species in wet recovery treatments on the basis of
studies done on laboratory carbonized material, it is possi-
ble to conclude which species are more sensitive and which
method is more aggressive.
This study showed that the carbonized grains of lentils
and millet are sensitive and therefore samples that contain
these grains should preferably be spared wet sieving. Car-
bonized remains of beans, wheat and barley after both treat-
ments keep a preservation percentage greater than 70% and
therefore we believe that they are suffi ciently robust and
their presence should not be a reason for avoiding wet siev-
ing. Therefore, we propose that, before the water recovery,
a preliminary insight into the diversity of plant remains
from each site, should be acquired. If the sample contains
carbonized lentils and millet, and it is possible to conduct
sieving without water, wet recovery techniques should defi -
nitely be avoided. However if the soil around the sample is
so compact that it demands the use of water, we recom-
mend avoiding wet sieving, because it will do more damage
to sensitive carbonized material than fl otation.
If wet sieving has already been conducted or must be
used to accelerate the recovery process, we certainly suggest
researchers should consider the fact that the decomposition
of lentils may be greater than 45% and the decomposition of
millet greater than 40%. That fact should be noted in order
to obtain a more accurate proportion ratio of found taxa.
In conclusion, our study is in agreement with previous
reports of Badham and Jones (1985) that the wet sieving is
a more aggressive method than fl otation and that it should
be used only when the sample does not contain many frag-
ile remains.
References
Babrauskas, V. 2013: Temperatures in fl ames and fi res. Retrieved
February 10, 2013 from http://www.doctorfi re.com/fl ametmp.
html
Badham, K., Jones, G., 1985: An experiment in manual process-
ing of soil samples for plant remains. Circaea 3, 15–26.
Brady, J. T., 1989: The infl uence of fl otation on the rate of recov-
ery of wood charcoal from archaeological sites. Journal of
Ethnobiology 9, 207–227.
Goette, S., Williams, M., Johannessen, S., Hastorf, C. A., 1994:
Toward reconstructing ancient maize: Experiments in process-
ing and charring. Journal of Ethnobiology 14, 1–21.
Hosch, S., Zibulski P., 2003: The infl uence of inconsistent wet-
sieving procedures on the macroremain concentration in wa-
terlogged sediments. Journal of Archaeological Science 30,
849–857.
Jacomet S., Kreuz S., 1999: Archäobotanik – Aufgaben, Metho-
den und Ergebnisse vegetations-und agrargeschichtlicher For-
schung. Eugen Ulmer, Stuttgart.
Jarman, H. N., Legge, A. J., Charles, J. A., 1972: Retrieval of plant
remains from archaeological sites by froth fl otation. In: Higgs,
E. (ed.) Papers in economic prehistory, 39–48. Cambridge
University Press.
King, F. B., 1987: Prehistoric maize in eastern North America: An
evolutionary evaluation. Dissertation, University of Illinois.
King, F. B., 1994: Variability in cob and kernel characteristics of
North American maize cultivars. In: Johannessen, S., Hastorf,
C. A. (eds.) Corn and culture in prehistoric New World, 35–
54. Westview Press, Boulder, Colorado.
Mareković, S., Karavanić, S., Kudelić, A., Šoštarić, R., 2015: The
botanical macroremains from the prehistoric settlement
Kalnik-Igrišče (NW Croatia) in the context of current knowl-
edge about cultivation and plant consumption in Croatia and
neighboring countries during the Bronze Age. Acta Societatis
Botanicorum Poloniae 84, 227–235.
Pearsall, D. M., 2000: Paleoethnobotany, a handbook of proce-
dures. Academic Press, San Diego.
Pearsall, D. M., 1980: Analysis of archaeological maize kernel
cache from Manabi Province, Ecuador. Economic Botany 34,
344–351.
Šoštarić, R., 2001: Carbonized plant remains of the prehistoric lo-
cality in Nova Bukovica on the site Sjenjak. Prilozi Instituta
za arheologiju u Zagrebu. 18, 79–82.
Tanno, K., Willcox, G., 2006: The origins of cultivation of Cicer
arietinum L. and Vicia faba L.: early fi nds form Tell el-Kerkh,
north-west Syria, late 10th millennium B.P. Vegetation History
and Archaeobotany 15, 197–204.
Tolar, T., Jacomet, S., Velušček, A., Čufar, K., 2010: Recovery
techniques for waterlogged archaeological sediments: a com-
parison of different treatment methods for samples from Neo-
lithic lake shore settlements. Vegetation History and Archaeo-
botany 19, 53–67.
Wagner, G., 1982: Testing fl otation recovery rates. American An-
tiquity 47, 127–132.
Wright, P. J., 2003: Preservation or destruction of plant remains by
car bonization?. Journal of Archaeological Science 30, 577–
583.
Wright, P. J., 2005: Flotation samples and some paleoethnobotani-
cal implications. Journal of Archaeological Science 32, 19–26.