The use of forensic botany and geology in war crimes

investigations in NE Bosnia

A.G. Brown *

Palaeoenvironmental Research Group, School of Geography, Archaeology and Earth Resources,

University of Exeter, Amory Building, Rennes Drive, Exeter EX4 4RJ, UK

Received 24 March 2005; received in revised form 17 February 2006; accepted 5 May 2006

Available online 30 June 2006

Abstract

From 1997 to 2002 the United Nations International Criminal Tribune for the former Yugoslavia (ICTY) undertook the exhumation of mass

graves in NE Bosnia as part of the war crimes investigations aimed at providing evidence for the prosecution of war criminals in The Hague. This

involved the location and exhumation of seven former mass graves (primary sites) dug following the fall of Srebrenica in July 1995. These primary

mass graves were secretly and hurriedly exhumed three months later and most of the bodies or body parts transported and reburied in a large

number of secondary sites many of which were subsequently exhumed by ICTY. The aim of the pollen and soil/sediment studies was to provide an

‘environmental profile’ of the original site of the samples and use this to match the relocated bodies to the original mass graves. This was part of

completing the chain of evidence, providing evidence of the scale and organization of the original atrocities and the subsequent attempts to conceal

the evidence related to them. All the primary sites were located in areas of contrasting geology, soils and vegetation, and this allowed matching of

the sediment transported in intimate contact with the bodies to the original burial sites, which in some cases were also the execution sites. In all,

over 24 sites were investigated, over 240 samples collected and analyzed under low power microscopy and 65 pollen sub-samples fully analyzed.

The pollen and sediment descriptions were used in conjunction with the mineralogy (using XRD) of primary and secondary sites in order to provide

matches. These matches were then compared with matching evidence from ballistic studies and clothing. The evidence has been used in court and

is now in the public domain. It is believed this is the first time ‘environmental profiling’ techniques have been used in a systematic manner in a war

crimes investigation.

# 2006 Published by Elsevier Ireland Ltd.

Keywords: Forensic palynology; Forensic geology; War crimes; Mass graves; Bosnia

www.elsevier.com/locate/forsciint

Forensic Science International 163 (2006) 204–210

1. Introduction

In recent years the use of forensic geology and botany in

criminal cases has increased in many parts of the world [1,2]

although it has a much longer history [3]. The combination of

geology and botany for provenancing is what has been termed

‘‘environmental profiling’’ [4] and it can be particularly

valuable in providing strong circumstantial evidence linking a

suspect or suspects to a scene of crime [4,5]. The basis of the

combination of techniques, such as mineralogy and pollen

analysis is to decrease the chances of a false match. Forensic

geology can often provide strong provenancing power at the

sub-regional or landscape scale and forensic botany can provide

* Tel.: +44 1392 26331; fax: +44 1392 263342.

E-mail address: a.g.brown@exeter.ac.uk.

0379-0738/$ – see front matter # 2006 Published by Elsevier Ireland Ltd.

doi:10.1016/j.forsciint.2006.05.025

evidence at the spatial scale of the crime scene. This can be

regarded as providing evidence of both the closeness of

association (match) and the uniqueness of that association as

formalized in a maximum likelihood ratio [6]. This paper

presents a summary of the results of environmental profiling

used as part of the forensic investigations of the war in the

former Yugoslavia and is believed to be the first attempts to use

such techniques in a war crimes context.

2. Exhumation sampling

In 1997 the United Nations International Criminal Tribune

for the Former Yugoslavia (UN ICTY) started exhumations of

mass graves in NE Bosnia associated with the massacre of

civilians in and around Srebrenica in July of 1995. It was known

from intelligence that 3 months after the initial executions of

civilians the original mass graves (primary sites) had been

A.G. Brown / Forensic Science International 163 (2006) 204–210 205

exhumed and the bodies transported over a 1–3 day period to a

number of unknown secondary grave sites. During exhumations

in 1997 a program of sampling soils and sediments associated

with both primary sites and secondary sites was begun. The aim

of this program was to link the secondary sites with primary

sites so providing evidence of the original location of the burials

as part of the war crimes indictments against individuals held in

The Hague or on the ICTY wanted list. The data would also add

evidential weight to the prosecution claims of the scale and

organization of both the original crimes and attempts to conceal

evidence. Mass graves were sampled during exhumation with

small bulk samples being taken of both the grave fills and the

country rock and soils surrounding the mass graves. In all five

primary sites were sampled and 19 secondary sites. This work

was done in conjunction with, but without any knowledge of,

the results of other forensic investigations of clothing,

documents and shell casings. All soils and sediments were

also described using standard geological and soil description

procedure [7]. Due to the high and variable alteration of

sediments in the graves caused by the variable state of

decomposition and saponification of bodies under different

water-table conditions it was decided to concentrate on robust

non-transient sediment and soil characteristics and specifically

‘included foreign items’, clasts, pollen and spores (palyno-

morphs) and sediment mineralogy. The combined use of these

parameters has been shown to have high provenancing power

with a low risk of error in criminal investigations under

appropriate circumstances.

Sampling took place at both primary and secondary sites. In

each case a series of samples was taken from the fill of the grave

(body part matrix) at locations both away from and close to

body parts (Fig. 1). In many cases clasts of soil or mud could be

Fig. 1. The sampling contexts of th

recognized which had been mixed into the fill and these were

sampled separately. Additional samples were taken from the

cranium and between clothing, skin, or bone of certain of the

bodies. Soil from these areas is unlikely to have been lost or

gained during the process of exhumation. Additional control

samples were taken from the walls of the grave once all the fill

had been removed. At least one control sample was taken for

every body-part matrix sample. These control samples

represented the background pollen and spore accumulation

profile for that location. This also allowed the determination of

a mineral profile of the local soils and sediments. The local

vegetation was also recorded to a distance of approximately

50 m where de-mining permitted. Vegetation was only recorded

as presence or absence with additional comments on abundance

for the major species present. The overall aim of the sampling

strategy was to trace as many stages as possible on the chain of

events that ended with the exhumation by ICTY and could have

allowed the addition of soil and pollen. Overall over 240

samples were taken and 65 sub-samples counted in the pollen

and spore analyses.

3. Methods

Samples for pollen and mineralogical analysis were

approximately 50 g and depending upon the context either

cut from the walls of the excavations, from sections, or taken

from clothing or direct body contact in the field or mortuary.

The samples were inspected under low power magnification

and then for pollen and spore analysis sub-samples were

subjected to standard chemical procedures [8]. This involved

the removal of carbonates using weak hydrochloric acid,

triple sieving (7 mm), removal of silicates using hydrofluoric

e environmental investigations.

A.G. Brown / Forensic Science International 163 (2006) 204–210206

acid, the removal of organic matter using an acetylation

mixture (sulphuric acid and acetic anhydride) and finally

mounting in silicon oil. In most cases relatively large sub-

samples were used (1–3 ml) due to the probable variable

pollen and spore concentrations of soils and sediments. Exotic

marker spores were not added due to pollen and spore

concentration values having little worth in such situations.

Silicon oil was preferred to glycerol due to the anticipated

difficulty of some identifications. Pollen and spore identifica-

tion took place at 600� magnification and 1000� for critical

identifications. An extensive reference collection held in the

Palaeoecology Laboratory at Exeter was used to aid

identification and standard pollen nomenclature is used [9].

Of particular importance was the differentiation of the cereal

type pollen which was achieved using specialist keys [10,11]

and which allowed the identification of Zea mays on the

basis of its grain size (>60 mm) and its pore and annulus

dimensions using [12]. Rosaceae pollen grains were differ-

entiated using a combination of the above keys, [13] and the

type collection at Exeter. Pollen and spore data were held on

spreadsheets and the results summarized in the reports

presented to ICTY.

Samples were prepared for XRD by crushing with a pestle

and mortar, homogenizing and smearing onto glass slides. XRD

analysis utilized a Philips PW1830 generator which employed

Philips APD software system to set parameters. All samples

were analyzed using 40 kV, 40 mA Copper Ka radiation from a

long fine focus tube with 18, 0.18 and 18 divergence, receiving

and scatter slits, with samples run from 48 to 708 2u. The count

rate was adjusted to 1.0 s with a step size of 0.028. Only major

mineral components were identified and used as this provides a

Fig. 2. The geology of the area around Srebrenica. Adapted from the Ge

more robust level of comparison and reduces the effect of local

and within profile variation.

North East Bosnia is a region of the central Balkan

Mountains and the local geology is dominated by a series of

thrusts extending E-W through the Zvornik area into Serbia

(Fig. 2). This thrust zone is predominately composed of

limestones and dolomites but there are also small outcrops of

amphibolites and serpentinites. The thrust zone has also been

intruded by igneous rock, such as lamprophyres, and there are

dyke swarms. There is also extensive regional metamorphism

throughout the area. To the south the geology is dominated by

folded and variably metamorphosed sedimentary rocks

including, sandstones, shales, conglomerates and schists.

An exception to this is the Srebrenica area where there is a

large mass of extrusive volcanics which include andesites,

dacites and pyroclastics. North of the river gorge at Zvornik,

the River Drina valley has a suite of terraces composed of

sand and gravels. The soils in the area are essentially

Mediterranean Brown Earths but there is a very strong

lithological control producing calcareous Brown Earths and

calcareous Pelosols on limestones, acid Brown Earths on the

terraces and podzolic profiles on the sandstones, shales and

schists. There is also a loessic input to many of the soils. The

regional vegetation would naturally be central European

montane beech and coniferous forest, and some woodland

survives composed of Fagus, Carpinus, Quercus, and Pinus

and Picea at higher elevations. However, the area has been

extensively cultivated and there exists a patchwork of small

unenclosed hay fields, small unenclosed arable fields, and

orchards in the valley floors and lower valley sides, with

woodland on the upper slopes. Only examples of the results

ology Map of Yugoslavia, Yugoslavian Geological Survey, Belgrade.

A.G. Brown / Forensic Science International 163 (2006) 204–210 207

Fig. 3. Photograph of striated serpentinite clast from the fill of HZ 3. Scale in

cm.

can be listed here in order to illustrate the range of

environmental evidence used.

4. Results

Initial investigations of the sedimentology, clasts and

included foreign bodies provided some initial linking evidence.

For example, the discovery of a striated clast of serpentinite in

secondary grave Hodzici road (HZ) 3 (Fig. 3) was found to

match the local geology of only one of the primary sites (Lazete

I) upslope of which was a serpentinite dyke. At secondary site

HZ 5 sections of stretched black plastic piping were recovered

which matched (including the extruded ends) with a water pipe

which had crossed the primary site Lazete II prior to excavation

of the mass grave.

Table 1

Summary of the environmental evidence linking primary and secondary sites

Primary site Soil type/lithology/

inclusions

Major minerals Veget

Kozluk River terrace gravels,

discoid fluvial gravel

(imbricated)

Ch, I/M, Qz, F, C Scrub

Branjevo Fm Tuff with deep

loessic soils

Ch, I/M, Ka, Qz, Fe Edge

(whea

rudera

Lazete I Thrust zone, limestone,

dolomites, sandstones,

serpentinite dyke

S, I/M, Ka, Qz, Fe Edge

forest

Lazete II Thrust zone, limestone,

dolomites, sandstones,

black water-piping

S, I/M, Ka, Qz, Fe Cleari

forest

Glogova 1E Sandstones and siltstones

with limestone rich

gravel in places

Qz, Ch, I/M Mixed

orcha

beech

Glogova 1F, 1H,

3, 5, 7, 8, 9

Sandstones and siltstones

with limestone rich gravel

in places, hay masses

(some with shell casings),

rubble, concrete, plaster

Qz, Ch, I/M Mixed

orcha

beech

S: swelling clays; Ch: chlorite; I/M: illite/mica; F: feldspar; Ka: kaolinite; Qz: qu

4.1. Primary sites

XRD analyses revealed that there was a major mineralogical

difference between the primary sites. In particular Lazete II and

I had swelling clay minerals present (Table 1) whilst Branjevo

Farm and Glogova did not. These differences were clearly

related to the soil and geology surrounding the primary sites

which was in all cases different from that of the secondary sites.

This provided a major differentiation between sediments

derived from the primary sites. Additionally, foreign plant

macrofossils could be recognized in several sites included

clumps of matted hay which were found in intimate contact

with body parts at both Glogova 1F, 1H, 8 and 9 (but not 1E) and

at secondary sites such as Zelanji Jadar 6. The location and

vegetation surrounding the Glogova sites was at variance with

this find and the discovery of shell casings within masses of hay

suggested it had originated at the location of execution, which

on the basis of witness statements was believed to be the

Krevice warehouse (Potocari).

The pollen and spore counts were used to provide lists of

types present with their relative percentages. The land use and

vegetation of the primary sites was both surveyed in the field

and known from aerial photographs. The major differences

between the primary sites allowed relatively straightforward

allocation based primarily on the dominant pollen and spore

types recorded. The summarized profiles of the primary sites

and linked secondary sites are given in Table 1 including the

dominant vegetation and pollen and spore types. It should be

noted that many other secondary sites exhumed were linked by

using other evidence directly to primary sites or to the

secondary sites listed below and thus to a primary site.

As can be seen from Fig. 4 the gross types present at each of

the primary sites varied significantly. At Branjevo Farm there

were low tree values, high herbs (mostly Poaceae) and very

ation/land use Dominant pollen/spores

(in descending value)

Linked

secondary sites

/grassland and arable – CR3

of arable field

t – communal farm)

ls

Cereals, Poaceae,

Pinus, Picea

CR12

of montane

(10 m)

Pinus, Cyperaceae, Poaceae,

Picea, Juglans

HZ3

ng in the montane

, wet meadow

Pinus, Cyperaceae, Poaceae,

Picea, Juglans

HZ2, HZ4,

HZ5

arable, hay meadow,

rds and forest (incl. pine,

, hornbeam and spruce)

Fagus, Picea, Pinus,

Carpinus, Corylus,

Poaceae

ZJ6

arable, hay meadow,

rds and forest (incl. pine,

, hornbeam and spruce)

Trees, Pinus, Picea, herbs

(high Poaceae) and high

cereals (Avena/Triticum)

occasional Z. mays,

Malus t. and Prunus t.

ZJ5

artz; Fe: iron; C: calcite.

A.G. Brown / Forensic Science International 163 (2006) 204–210208

Fig. 4. Summary pollen data from the primary grave sites.

Fig. 5. Summary pollen data from the secondary grave sites.

high cereals whilst at Lazete I trees were dominant with

moderate herbs and no cereals. This reflected the relative

locations of the sites with the Branjevo Farm site on the edge of

an arable field in an area of large fields (co-operative farm) with

little tree cover whereas the Lazete sites were located at the

edge of montane woodland. There were several primary sites at

Glogova but the natural soil and sediment surrounding all the

graves had pollen and spore contents too low to count (due to

ploughing) and so pollen from sediments within the grave

matrix had been brought in along with the topsoil and bodies.

The vegetation surrounding the Glogova sites included a mix of

woodland, orchards, hay meadows and arable fields. As can be

seen from Fig. 4 the fill matrix of Glogova 1F, 1H, 8 and 9 all

share similar pollen and spore spectra with relatively low trees,

high herbs (mostly Poaceae) and high cereal types. However,

Glogova 1E is entirely different with high trees (dominated by

Fagus and Picea), very low herbs and only a trace of cereals.

Table 2

Major pollen type comparisons (in percentage total land pollen) for secondary site

Selected pollen and spore types Hodzici road 25

Abies (fir) 3.6

Alnus (alder)

Carpinus (hornbeam) 16.8

Corylus (hazel)

Fagus (beech)

Juglans (walnut) 3.6

Picea (spruce) 2.4

Pinus (pine) 24.0

Quercus (oak) 1.1

Asteraceae (dandelion-like flowers)

Cereals (cereals)

Chenopodiaceae (goosefoot family)

Cyperaceae (sedges)

Hedera (ivy)

Lactuceae (daisy-like flowers) 25.4

Plantago lanceolata (ribwort plantain) 3.3

Poaceae (grasses) 13.2

Pteridium (bracken)

Ranunculus t. (buttercup family)

Reseda t. (mignonettes)

Vicia cracca (tufted vetch)

Filicales (unid. Ferns)

Others 6.6

The Glogova 3 matrix samples suggested they were derived

from an open environment, probably a meadow which has in the

past been under cereal cultivation probably maize (Z. mays).

The Glogova 5 matrix samples produced very similar pollen

assemblages all dominated by cereals and pine pollen with a

variety of meadow herbs and trees. The vast majority of the

cereal pollen is maize (Z. mays), a pollen type which does not

travel far. These samples are from a meadow which was under

maize the previous year. At least a part of the field from which

the soil comes was close to a walnut tree. The samples from

Glogova 7 had very low pollen counts, but the pollen present

was similar in types to Glogova 3 and 5. The most obvious

characteristic of all the Glogova body part matrix samples with

the exception of 1E was the high cereal pollen type, the

majority being Avena–Triticum type (probably wheat) with

occasional Z. mays (maize) pollen grains. Cereal pollen grains

are only found at such levels within wheat fields or where straw

or grain is stored. Also in grave 1F a straw and hay-rich mass

was observed and sampled (GL01/282A). The differences are

s along the Hodzici road

Hodzici road 23 Hodzici road 3 Hodzici road 4

1.2

2.4

1.3 2.3 3.9

1.1 1.4

2.0 1.1

18.2 8.7

56.3 23.0 14.6

1.2 2.9

2.2

1.6

2.0

1.3

1.0 30.1 5.7

6.7 7.8

6.7 9.1 13.6

2.8 1.6

2.0 1.0

1.9

2.3

4.8

6.5 18.3 29.4

A.G. Brown / Forensic Science International 163 (2006) 204–210 209

Table 3

Major pollen and spore type comparisons in percentage total land pollen and spores for secondary sites at Zelanji Jadar

Selected pollen

and spore types

Zelanji Jadar 5

AB13 control

Zelanji Jadar 5

AB4 BPM

Zelanji Jadar 5

AB1 BPM

Zelanji Jadar 6 19A

BPM (mixed)

Zelanji Jadar

6 43A BPM

Zelanji Jadar

6 266A BPM

Abies (fir) 4.0

Carpinus (hornbeam) 6.0

Corylus (hazel) 1.1

Fagus (beech) 5.0 13.5 2.3

Picea (spruce) 20.2 6.2 20.9 2.3

Pinus (pine) 52.0 9.4 1.5 45.3 4.6 6.6

Populus (poplar) 2.0

Apiaceae peucedanum t.

(Hog’s fennels)

27.6

Artemisia (mugwort) 2.0

Asteraceae

(dandelion-like flowers)

1.0

Caryophyllaceae

(pinks family)

1.2

Cereals (cereals) 6.2 2.2 64.4 1.9

Chenopodiaceae

(goosefoot family)

3.1

Cyperaceae (sedges) 1.4

Lactuceae

(daisy-like flowers)

6.1 5.4 1.4 2.3

Morus (mulberries) +

Polygonum persicaria t.

(redshank)

1.0

Plantago lanceolata

(ribwort plantain)

1.1 1.0

Poaceae (grasses) 4.7 16.9 95.0 4.6 22.6 84.2

Ranunculus t.

(buttercup family)

6.2

Serratula t. 2.5

Filicales (unid. Ferns) 2.7

Others 10.1 7.5 2.5 4.3 3.3 5.0

Fig. 6. The links made between primary and secondary sites using pollen, spore and mineralogical evidence. BF: Branjevo Farm; D: Dulici (Dam sites); GL:

Glogova; HR: Hodzici road; KBF: Kozluk bottle factory; KR: Kancari road; LZ: Lazete; NK: Nova Kasaba; P: Potocari (Krevice warehouse); ZJ: Zelanji Jadar.

A.G. Brown / Forensic Science International 163 (2006) 204–210210

believed to reflect a different original source with the bodies in

graves Glogova 1F-9 having been transported from the Krevice

warehouse (the execution site) whereas the bodies in Glogova

1E were the result of local killings. However, the general tree

content and particularly fruit trees (Malus t. and Prunus t.) in

the body matrix samples from one sample in grave 1H and the

similarity of the mineralogy of the body part matrix and natural/

control samples indicates that the sediment that forms the body

part matrix was of local origin. Whereas, several samples

contained masses of hay (cut and dried grass) and pollen spectra

dominated by Poaceae (including clumps from anthers). This

apparent conflict is the result of the mixture of bodies with hay

from the warehouse with the backfill of the sediments and soils

at the Glogova sites 1F-9.

4.2. Results from the secondary sites

The mineralogy and pollen/spores content of Kancari road

(KR) 3, KR12, Hodzici road (HZ) 3, HZ 4, and HZ 5 indicates

that in each case the grave matrix is foreign to the site and could

not have originated in situ. The sediment characteristics of KR3

were entirely compatible with its having been derived from the

gravel quarries near the Kozluk bottle factory and it could not

have come from Branjevo Farm, Lazete or Glogova primary

sites. For KR12 the pollen/spore content, the sediment type and

the stubble all point to the source being Branjevo Farm primary

site. The mineralogy, pollen/spores content, clast lithology and

inclusions (e.g. severed water pipe) all point to Lazete II as the

origin for HZ 3, HZ 4 and HZ 5, all of which have similar pollen

spectra (Table 2). The bifaces/tools and igneous clast lithology

further indicate that Lazete 1 was the source for HZ 3.

Fig. 5 shows the variation in gross pollen type spectra for

three secondary sites. It shows that the fill of Zelanji Jadar (ZJ)

5 contained both local sediment (similar to the surrounding

controls and dominated by tree pollen) and a very different

spectrum dominated by herbs (mostly Poaceae) and cereals

(Table 3). The site was located within the beach dominated

montane forest. The pollen from the body part matrix from ZJ 5

is similar to Glogova 3 and 5 and is consistent with it having

been derived from the same locality. A similar mixed fill was

found at ZJ 6 (ZJ6/028A). A rather different combination was

found at HZ 3, where the body part matrix component was

found to be a close match to Lazete I in both mineralogy and

pollen content. All the secondary sites, on the basis of clast

lithology, mineralogy and pollen and spore content, could be

associated with matching primary sites (Fig. 6).

5. Conclusions

Along with other forensic evidence (clothing, personal

effects, ballistics and documents), the environmental data

provided a high level of multiple circumstantial evidence which

linked the secondary sites to the primary sites and execution

sites. The robustness of the evidence was revealed by its

agreement with the other forensic evidence, particularly that

derived from shell casings as no discrepancies were found

between secondary and primary site allocation using the two

lines of evidence. Of particular importance was the recognition

that (a) the bodies may have entered primary mass graves with

an associated plant and pollen assemblage from an execution

site, and (b) that all grave sites are likely to contain both foreign

and local fill. It is therefore essential that in the field as much

effort as possible is made to differentiate between local matrix

and imported body part matrix. The techniques which are

increasingly being used in serious criminal investigations in the

UK, North America and New Zealand can provide reliable

evidence in appropriate circumstances. These are where bodies

have been exhumed and where the local geology, soils and/or

vegetation are spatially variable enough to provide easily

distinguishable non-transient characteristics, particularly pol-

len and spores and mineralogy. This is believed to be the first

time a systematic use of such environmental evidence has been

used in a major war crimes investigation.

Acknowledgements

The author owes a huge gratitude to Richard Wright for his

help and encouragement as well as to Jose Pablo Baryabar, Ian

Hanson, Dean Manning, Neil Ashcroft and all of the UN ICTY

Northeastern Bosnia Exhumation team. Additional thanks must

go to Sue Rouillard for the illustrations and to Art Ames for

sample preparation.

References

[1] V.M. Bryant Jr., J.G. Jones, D.C. Mildenhall, Forensic palynology in the

United States of America, Palynology 14 (1990) 193–208.

[2] A. Ruffell, J. McKinley, Forensic geoscience: applications of geology,

geomorphology and geophysics to criminal investigations, Earth Sci. Rev.

69 (2005) 235–247.

[3] D.C. Mildenhall, Forensic palynology in New Zealand, Palynol. Rev.

Palaeobot. 64 (1990) 227–234.

[4] A.G. Brown, A. Smith, O. Elmhurst, The combined use of pollen and soil

analyses in a search and subsequent murder investigation, J. Forensic Sci.

47 (2002) 614–618.

[5] R.C. Murray, Forensic geology: yesterday, today and tomorrow, in: K. Pye,

D. Croft (Eds.), Forensic Geoscience: Principles, Techniques and Appli-

cations, Conference Abstracts, Geological Society of London, 2004, pp.

7–10.

[6] M. Horrocks, K.A.J. Walsh, Forensic palynology: assessing the value of

the evidence, Rev. Palaeobot. Palynol. 103 (1998) 69–74.

[7] Soil Survey Field Handbook, Soil Survey Technical Monograph No. 5,

Harpenden, 1976.

[8] P.D. Moore, J.A. Webb, M.E. Collinson, Pollen Analysis, Blackwell,

Oxford, 1991.

[9] K.D. Bennett, Annotated catalogue of pollen and pteridophyte spore types

of the British Isles, Department of Plant Sciences, University of Cam-

bridge, 1994.

[10] S.T. Andersen, Identification of wild grass and cereal pollen, Danm. Geol.

Unders. Arbog (1978) 69–92.

[11] K. Faegri, P.E. Kaland, K. Krzywinski, Textbook of Pollen Analysis, 4th

ed., Wiley, Chichester, 1989.

[12] H.-J. Beug, Leitfaden der Pollenbestimmung fur Mitteleuropa und angren-

zende Gebiete, Verlag Dr. Friedrich Pfeil, Munich, 2004.

[13] R. Andrew, A Practical Pollen Guide to the British Flora, Quaternary

Research Association, Technical Guide No. 1, Cambridge, 1984.