Fractography Fracture Topography As A Tool in Fracture Mechanics and Stress Analysis PDF

Fractography: fracture topography
as a tool in fracture mechanics and stress analysis

Geological Society Special Publications
Series Editor A. J. FLEET
G E O L O G I C A L SOCIETY SPECIAL PUBLICATION NO. 92
Fractography: fracture topography

as a tool in fracture mechanics
and stress analysis
EDITED BY
M O H A M M E D S. A M E E N
Independent Consultant, UK
1995
Published by
The Geological Society
London
THE G E O L O G I C A L SOCIETY
The Society was founded in 1807 as The Geological Society of London and is the oldest geological society in the
world. It received its Royal Charter in 1825 for the purpose of 'investigating the mineral structure of the Earth'.
The Society is Britain's national society for geology with a membership of 7500. It has countrywide coverage and
approximately 1000 members reside overseas. The Society is responsible for all aspects of the geological sciences
including professional matters. The Society has its own publishing house, which produces the Society's
international journals, books and maps, and which acts as the European distributor for publications of the
American Association of Petroleum Geologists, SEPM and the Geological Society of America.
Fellowship is open to those holding a recognized honours degree in geology or cognate subject and who have at
least two years' relevant postgraduate experience, or who have not less than six years' relevant experience in
geology or a cognate subject. A Fellow who has not less than five years' relevant postgraduate experience in the
practice of geology may apply for validation and, subject to approval, may be able to use the designatory letters
C. Geol (Chartered Geologist).
Further information about the Society is available from the Membership Manager, The Geological Society,
Burlington House, Piccadilly, London W1V 0JU, UK. The Society is a Registered Charity, No. 210161.
Published by The Geological Society from: Distributors

The Geological Society Publishing House USA
Unit 7, Brassmill Enterprise Centre AAPG Bookstore
Brassmill Lane PO Box 979
Bath BA1 3JN Tulsa
UK Oklahoma 74101-0979
(Orders: Tel. 01225 445046 USA
Fax 01225 442836) (Orders: Tel. (918) 584-2555
Fax (918) 584-0469)
Australia
First published 1995 Australian Mineral Foundation
63 Conyngham Street
Glenside
The publisher makes no representation, express or South Australia 5065
implied, with regard to the accuracy of the information Australia
contained in this book and cannot accept any legal (Orders: Tel. (08) 379-0444
responsibility for any errors or omissions that may be Fax (08) 379-4634)
made.
India
9 The Geological Society 1995. All rights reserved. No Affiliated East-West Press PVT Ltd
reproduction, copy or transmission of this publication G-l/16 Ansari Road
may be made without prior written permission. No New Delhi 110 002
paragraph of this publication may be reproduced, India
copied or transmitted save with the provisions of the (Orders: Tel. (11) 327-9113
Copyright Licensing Agency, 90 Tottenham Court Fax (11) 326-0538)
Road, London W l P 9HE. Users registered with the
Copyright Clearance Center, 27 Congress Street, Japan
Salem, MA 01970, USA: the item-fee code for this Kanda Book Trading Co.
publication is 0305-8719/95 $07.00. Tanikawa Building
3-2 Kanda Surugadai
British Library Cataloguing in Publication Data Chiyoda-Ku
A catalogue record for this book is available from the Tokyo 101
British Library Japan
(Orders: Tel. (03) 3255-3497
ISBN 1-897799-32-2 Fax (03) 3255-3495)
Typeset by Bath Typesetting Ltd

Bath, England
Printed by The Alden Press,

Osney Mead, Oxford, UK.
Contents
AMEEN, M. S. Introduction 1
Part 1: Fractographic studies
Methods of observation, data collection and processing]quantitative fractography

AYDAN, O. & SHIMIZU,Y. Surface morphology characteristics of rock discontinuities
with particular reference to their genesis 11
JESSELL, M. W., Cox, S. J. D., SCHWARZE,P. & POWER, W. L. The anisotropy of surface
roughness measured using a digital photogrammetric technique 27
Application of fractography to fracture analysis from core

BANKWITZ, P. & BANKWITZ, E. Fractographic features on joints of KTB drill cores
(Bavaria, Germany) 39
Fractography applied to the analysis of experimentally produced fractures in rocks

and non-rocks
KULANDER, B. R. & DEAN, S. L. Observations on fractography with laboratory
experiments for geologists 59
BUTENUTH, C. & DE FREITAS, M. H. The character of rock surfaces formed in Mode I 83
Fractography applied to fracture analysis in field studies aimed at understanding

regional tectonics
AMEEN, M. S. Fractography and fracture characterization in the Permo-Triassic
sandstones and the Lower Palaeozoic Basement, West Cumbria, UK 97
AMEEN, M. S. Fracture characterization in the Chalk and the evolution of the Thanet
monocline, Kent, southern England 149
ROBERTS, J. C. Fracture surface markings in Liassic limestone at Lavernock Point,
South Wales 175
Part 2: Non-fractographic studies

Miscellaneous studies of fractures
COSGROVE, J. W. The expression of hydraulic fracturing in rocks and sediments 187
GOODWlN, A. M. Spatial change in joint geometry in the Chalk of eastern England 197
GROSS, M. R., FISCHER, M. P., ENGELDER, T. & GREENFIELD, R. J. Factors
controlling joint spacing in interbedded sedimentary rocks: integrating numerical
models with field observations from the Monterey Formation, USA 215
Index 235
From Ameen, M. S. (ed.), 1995, Fractography: fracture topography as a tool in fracture mechanics and stress analysis
Geological Society Special Publication No. 92, pp. 1-10
Fractography: fracture topography as a tool in fracture

mechanics and stress analysis. An introduction
M O H A M M E D S. AMEEN
Independent Consultant, 122 Tachbrook Street, Pimlico, London
S W1 V 2ND, UK
What is fraetography?
Fractography can be defined as the science which deals with the description, analysis and
interpretation of fracture surface morphologies (fracture topographies) and links them to
the causative stresses, mechanisms and subsequent evolution of the fractures. Fracto-
graphy can be classified according to the field (scale) of observation into:
(a) macrofractography, dealing with features which can be studied using the naked eye
or a hand lens, and
(b) microfractography, dealing with features which can be studied with the aid of
optical microscopes .or electron microscopes only.
The term fractography was first used by the metallurgist Carl A. Zapffe (Zapffe & Clogg
1944), and was introduced to geology by Byron Kulander in the 1970s (Kulander et al.
1979). The discipline of fractography (as a science) in geology is therefore new, even
though nearly a century ago Woodworth (1896) was the first to describe joint surface
morphology from field observations. Fractography is hardly mentioned in text books on
structural geology, consequently many geologists are unfamiliar with this term and its
implications and may confuse the term with fractals. The subject is largely neglected by
those who deal with fracture characterization of rocks (e.g. core laboratories) in spite of its
obvious potential. In addition, the interpretation of fracture surface features in rocks has
been almost exclusively qualitative, and frequently controversial. The state of infancy and
neglect of fractography in geology is further reflected by the fact that in this field there has
been no clear definition of fractography. In addition there is no universally accepted,
unified nomenclature for fractographic features. All these factors prompted this special
publication, the first thematic volume on fractography in geology.
The evolution of fraetography in geology

Fractography in geology has evolved to its present state through four stages:
(a) The embryonic stage started with Woodworth's studies (1896) and the classification
of surface structures on joint faces.
(b) Nearly half a century during which no studies on fracture surface features in rocks
were recorded.
2 M.S. AMEEN
(c) A phase of fracture surface studies started in the early 1940s (Parker, 1942) and
culminating in the publication of 2 key papers by Hodgson (1961) and Roberts
(1961). Hodgson reviews the definitions and implications of those features identified
by Woodworth (1896).
(d) The final phase was stimulated by Hodgson's and Roberts' publications (1961) and
included several studies (e.g. Bankwitz 1966; Syme-Gash 1971) and reached its
climax when Kulander et al. (1979) introduced the term fractography to the studies
of surface features in rocks. This initiated quantitative study of the topic both by
using the already established name from metallurgy (i.e. fractography of Zapffe
& Clogg 1944), and by leading to a thorough description and explanation of
macrofractographic features in both field and core studies. This phase continues to
the present and is characterized by relatively greater interest than the previous
stages, though is still limited (mostly field macrofractography). For a review of the
subject of fractography and comprehensive bibliography, the reader should refer to
Kulander et al. (1979) and a recent review by Bahat (1991).
Where does fractography stand within the wide field of fracture studies?
An understanding of fractures and their mechanics is a matter of great interest in both
material and geological sciences. Workers within these two fields are driven to determine
fracture mechanics for two different reasons. A metallurgist for example is concerned with
producing alloys which are less susceptible to failure. However, an earth scientist's or
engineer's interest is considerably more complicated in that it embraces mechanisms for
enhancing fracture development (e.g. in a hydrocarbon reservoir), methods of reducing the
chances of fracture development (e.g. in a nuclear waste repository), and the quantification
of fracture surface morphologies as they influence fluid flow and rock stability. In addition
to the considerably complex nature and wide spectrum of a geologist's task compared to
that of a metallurgist, the former is faced with dealing with a vastly heterogeneous natural
material (rocks) under circumstances of both spatial and temporal scale that fall well
beyond the well constrained time/space scales in the studies of metals and alloys. This
difference has undoubtedly affected the pace of progress within the disciplines of fracture
mechanics of rocks. Deprived of the chance of direct observation of natural fracture
development in rocks, geologists followed two approaches to infer fracture mechanics:
they study experimentally induced fractures in rocks, and observe fractures and their host
regional/local structures in the field. Both approaches involved analysis of fractures as a
'population' and as 'individuals'.
Fracture population studies generally separate a particular population of fractures in a rock

mass into sets according to their 3D orientation and spatial and temporal distribution, and
try to link the sets to their causative stresses. Individual fracture studies deal with the
evolution of an individual fracture and its causative stresses. An overlap of the two
approaches involves studying how a fracture interacts with neighbouring fractures during
its history of evolution and propagation, and how that affects the statistical observation of
the fractures population (e.g. their frequency, spacing, dimensions, etc.). Fractography
primarily falls within the last two of these areas (Fig. 1).
INTRODUCTION 3
Fig. 1.
Do all fractures develop fractographic features?

The answer to this is yes. All fractures possess fractographic features at the time of their
inception although they vary in the degree of development depending on the rock type and
the environment in which they have been initiated and propagated. However, consequent
history may obliterate or change these fractures, e.g. erosion, mineralization, stylolitiza-
tion, shear movement. Although the last three processes may introduce new surface
features (e.g. slickenlines) these are not usually considered as fractographic features by
geologists (Kulander et al. 1979). In addition some authors include some of the features
shown by the trace of fractures on a free surface, e.g. hooking and bifurcation, as part of
the fractographic features. The present author is of the opinion that all aspects of surface
morphology are fractographic features in that they are diagnostic characteristics of the
fracture history and its evolution from the time of inception of a fracture to the time of
observation. Therefore fractographers should incorporate all aspects of fracture surface
morphology in their observations. This practice is essential to the widening of the
fractographer's task from the limited sphere of the moment of inception of a fracture to
the universal sphere of the temporal evolution of a fracture/fractures, e.g. the development
of faults and fault rocks. This universal application of fractography will speed up and
enhance our understanding of fracture mechanics and related phenomena in rocks (e.g.
faulting and fault rocks development).
This volume
The eleven papers in this volume range from those directly linked to fractography (Part 1)
to those which, although not addressing the subject explicitly, have significant implications
for fracture mechanics, the main objective of fractography (Part 2).
Part 1: Fractographic studies

The eight papers in this part fall mainly within the subject of macrofractography and cover
the following areas:
4 M.S. AMEEN
Methods of observation, data collection & processing/quantitative fractography This area of

studies is important in two ways. Firstly it offers methods and techniques to quantify the
morphological characteristics of a fracture surface and their mechanical/physical
implications (e.g. modelling the strength and fluid flow characteristics of fractured rock
masses). Secondly it is of vital importance to develop fast and automated ways to record
and analyse fracture surface topography for a statistically representative number of
fractures. The latter is essential considering the fact that earth scientists deal with a
'natural laboratory' that is extremely difficult to access on the spur of the moment due to
financial, environmental and time constraints. Two papers by Aydan & Shimizu and
Jessell et al. discuss two different techniques of quantitative fractography. Aydan &
Shimizu present a profilometer technique for quantifying fracture surface morphology in
rock masses. Their approach incorporates three main components: classification of natural
discontinuities in rock masses, choice of functions or parameters which represent the
quantitative aspects of the topographical features, and use of a specially built device to
measure, process and digitize the data on site (in the field). They classify discontinuities
according to their origin into primary (eigen), secondary and tertiary. The first includes
discontinuities associated with the origin of the rock, e.g. bedding planes in sedimentary
rocks. The second includes discontinuities resulting from desiccation, cooling and freezing
of rocks. The third is of tectonic origin and includes shear and tension fractures. The
authors choose linear profile parameters (height of asperities, inclination of asperity walls,
length of asperity wall relative to base length, and periodicity of asperities) to express the
geometry of surface morphology in a group of functions. These parameters are measured
for two mutually perpendicular profile directions (the eigen directions) one parallel (X)
and one perpendicular (Y) to the axis of elongated protrusions (ridges) on the fracture
surface, which are considered as the eigen directions. A specially built, manually operated,
profilometer is used in this work in conjunction with a digitizer and a microcomputer for
digitizing and storing data which are analysed by a processing sub-system.
The technique is applied to different rock types in central Japan. It is concluded that the
fracture surface morphology can be represented using profiles along the two eigen
directions (i.e. X and Y). Asperity inclinations have their minimum values along the X-axis
and maximum values perpendicular to the ridge axis. The centre line average height
(CLAH), which is a function of the asperity heights for a specific type of discontinuity,
does not show significant variations with the variation of rock-types. However, in a given
rock type CLAH is generally higher in secondary and tertiary discontinuities than in the
primary discontinuities. The study shows a proportional relationship between the angle of
a weighted asperity inclination (WAI) and the ratio of profile length (RPL). However, the
correlation between WAI and CLAH is poor. Finally this study illustrates quantitatively
the anisotropy in the surface morphology functions measured in different directions on the
same fracture surface.
,Iessell et al. present a digital photogrammetric technique for the quantitative
determination of fracture surface morphology. They have developed a software package
that can automatically reconstruct surface morphology information from stereo pairs at
any scale, using edge correlation and point matching. After initial alignment of the images,
the software does not require user intervention, making it well suited to the production of
large datasets. To test the accuracy and reliability of their technique the authors made a
comparison between a traditional profilometer and the new digital photographs of a
quarrying-induced fracture face in granodiorite (Alexander Quarry, Harcourt, Victoria,
INTRODUCTION 5
Australia). They chose a test area of 256 by 116 mm of the fracture face. The results show a
strong degree of correlation between the sets of data obtained from the two techniques.
The digital technique is found superior to that of the profilometer in that it is applicable to
a wide range of horizontal scales from satellites to microscopes with a consistent
methodology.
The main drawback of the technique is that its vertical resolution is smaller than that
achievable using profilometers. This limitation can be reduced by either increasing the
magnification of the photographic techniques, or by scanning the resultant negative at a
higher resolution; both methods are easily achievable with the right equipment. However,
with the present technology the best vertical resolution achievable still falls short of the
0.1/~m achievable by the profilometers.
Application of fractography to fracture analysis from core This field of fractographic studies
is of great importance to analysing and understanding fractures observed in cores
(Kulander et al. 1990). Its significance lies in its application to distinguish natural fractures
from those induced by drilling, coring and handling of core. As a result the analyst is able
to discard induced fractures when considering the effects of the natural fractures and their
surface morphologies on the mechanical and physical properties of rock masses, e.g.
permeability, stability, strength. In addition, the topographical features of the drilling/
coring-induced fractures are very good indicators of the orientation of the maximum and
minimum horizontal in situ stresses (all and Crh respectively). This application of
fractography is largely neglected by core analysts and core laboratories who instead use
quick core scanning and classify fractures population according to their orientation into
'sets'. This can lead to unrealistic assumptions as to the origin of fractures.
Bankwitz & Bankwitz present an excellent illustration of the application of fractography
to the determination of the maximum horizontal in situ stresses (SH). Their study is
particularly interesting because it covers a great range of depths and two different core
diameters. The study is of coring-induced fractures in core sections recovered from the
superdeep (8000m) borehole in northern Bavaria (KTB: Continental Deep Drilling,
Windischeschenbach). The borehole cuts the crystalline basement of the Bohemian Massif,
which includes amphibolites, gabbro and layered paragneisses. In cores with a diameter of
10 cm the surface morphology of fractures is not easily recognizable due to the coarse-
grained texture of the basement rocks. However, the overall shape of the coring-induced
fractures is used to determine their origin and propagation direction. In cores of larger
diameter (23.5 cm) the fracture surface morphology is easier to recognize. Two main types
of drilling-induced disk fractures (according to their geometry) are recognized. The first
are referred to as core disking surfaces (CDS) which are characterized by saddle and
trough structures, and are restricted to fine-grained rocks. The second group lack the
saddle and trough structures, show clear fractographic features, occur more frequently
than the first and are referred to as the usual fractographic surface (UFS) structures.
The coring-induced fractures are initiated by the coupling of the stress states in turn
related to the combined effect of the in situ stresses within the rock mass and the stress
release due to the removal of overburden pressure. The direction of the maximum
horizontal stress axis Sn is determined from the pattern of fracture surface morphology on
the UFS. These data are mostly in agreement with those obtained from core disking. The
SH orientation is NW-SE (165~ This direction coincides with the results from other
techniques, e.g. borehole breakout analysis and this agreement further supports the
6 M.S. AMEEN
feasibility of fractography as a tool in stress analysis for both checking the reliability of
other techniques and for use on its own with a high level of confidence.
Fractography applied to the analysis of experimentally produced fractures in rocks and non-
rocks Unlike natural fractures, experimentally induced fractures in rocks and non-rocks
are easier to analyse, because they are formed within a better defined/constrained
environment. Under such circumstances a clear and direct link between the causative
factors (e.g. type and configuration of applied stresses) and the morphology of the
resulting fractures can be established. This technique is a double-edged sword in that it can
be used by both teachers and students of fractography and fracture mechanics to illustrate
and better understand these subjects. Furthermore, researchers can manipulate the science
of fractography to solve particular problems within the field of fracture mechanics in
rocks. In the present collection of papers these two applications of fractography are
illustrated in two papers by Kulander & Dean and Butenuth & de Freitas respectively. The
first paper is by two of the pioneers of fractography in geology, Kulander & Dean. The
paper highlights the principles of fractography through fascinating and easy to follow
experiments in which fractures are induced in cheap, readily available material. To achieve
their objectives the authors used microscopic slides, solid glass rods and capillary rods,
loaded by bending and torsion. Point loading and thermal stresses are also used to induce
fractures in slides. In discussing the results, Kulander & Dean use quantitative inference
based on more complicated, controlled 'quantitative' experiments of others in the field of
fracture studies in glass. The authors give a detailed anatomy of the fracture surfaces
produced in the laboratory and illustrate the different stages of evolution of a fracture
as can be recognized from its topographical features, e.g. origin, mirror, mist, velocity,
hackles and twist hackles. In doing so they draw attention to the common occurrence of
fractographic features which are linked to rapid fracturing, e.g. velocity hackles, Wallner
lines and forking. Their discussion of the results also focuses on the lack of these velocity-
related features in natural and drilling-induced fractures in rocks. They conclude that
such a contrast in brittle fracture morphology between glass and rock is a witness to the
different properties of the two materials and the manner in which they respond to stresses.
The authors also stress their belief that shear joints do not exist.
The second paper by Butenuth & de Freitas illustrates the uses of fractography to
investigate the effect of boundary conditions during hoop tests (to determine tensile
strength) on the mode of failure. The hoop test used in this study is relatively new, and its
operation is yet to be fully analysed and assessed. The tests involve the insertion of two
hemi-cylindrical loading platens into the central hole of a hoop of rock. Then the platens
are expanded until the hoop fails. The tensile strength is obtained 'indirectly' from plotting
the ultimate load line force and the surface area of the fracture at the time of failure. Their
plots for two rock types used in their experiments show deviation of some hoop tests from
the otherwise linear force v. area relationship. In addition there exists a constant intercept
on the area axis for each of the two rocks. The authors address these two features using
fractography. The deviation of some of the results is found to be related to the fact that the
samples were not fully separated by fractures at failure. Subtracting the area that has not
been broken in tension brings the deviated results on line with the majority of samples.
Fractographic studies show that the induced fracture terminal end (in the part of the hoop
that is not fully broken in halves by the fracture) is characterized by rougher and more
branching surface. This evidence was used to re-examine samples which are already
broken in halves (after the test) to recognize the boundaries of the fracture surface at
INTRODUCTION 7
failure and correct the area data on the force/area plots. The constant intercept is linked to
imperfections in the configuration of the hoop samples and the platens and/or the dynamic
branching of the fractures. The former is proved to exist and induces compressive stresses
in certain parts of the hoop. In spite of that, fractographic inspection on microscopic and
macroscopic scale indicates that the experimentally induced fractures lack evidence of
shear movement (e.g. slickenlines or debris). Therefore they are of Mode-I. However,
using fractographic evidence the authors could not decide on whether dynamic branching
and the bending due to imperfections in the samples and the testing equipment are acting
separately or in combination.
Fractography applied to fracture analysis in field studies aimed at understanding regional

tectonics Field studies utilizing fracture analysis can benefit greatly from systematic
observations of fractographic features. Field analysis of fractures (apart from rare studies
mostly in the USA) has been based largely upon fracture orientations and subsequent
grouping into sets of parallel fractures, interpreted as 'conjugate shear' or 'tension/
extension' fractures on geometrical assumptions. The studies presented in the current
volume (two by Ameen and one by Roberts) present contributions to the newly emerging
school incorporating fractographic observations and interpretations as an essential part of
fracture/joint observations and analysis. These studies also offer an important contribu-
tion to the universal aspects of fractography by adding relatively large amounts of data to
those collected in other provinces. Such a database is a vital requirement to test the
persistence of fractographic features and their implications for fracture mechanics and
regional tectonics. Furthermore these fractographic studies permit observation on the
evolutionary and tectonic history of the studied areas. The three studies discuss different
regions in the United Kingdom. Ameen's papers cover West Cumbria in the English Lake
District, and the Isle of Thanet, Kent, southern England. Roberts' paper covers part of
South Wales.
In the first study Ameen presents the results of a comprehensive fracture characterization
of both the sedimentary cover rocks (Permo-Triassic) and the Lower Palaeozoic basement
rocks in West Cumbria. This study is mainly based on work carried out on behalf of UK
Nirex Ltd as a part of their ongoing geological investigation into the suitability of the
Sellafield area for an underground nuclear waste repository. The author combines
fractographic analysis with the conventional analysis of fracture orientations to determine
the fracture characterization and related stresses in the area. Three main fracturing phases
are recognized in the Permo-Triassic part of the cover rocks and their causative stresses are
discussed. In contrast, the Lower Palaeozoic basement rocks reflect a complex fracturing
history. The results suggest that the geometry and degree of development of fractographic
features in the sedimentary cover are controlled by sedimentary layering, lithological
heterogeneity, the presence of open fractures, pore fluid pressure, orientation and
magnitudes of the principal stresses, and their temporal and spatial variations across the
fractured horizon. The work also investigates exposed macrofaults in both basement and
cover rocks and their relationship with the mesofractures. It is suggested that 3D
configuration of the well exposed mesofractures zones should be considered when
modelling the scarcely exposed macrofaults and their interpretation from subsurface data.
In his second paper Ameen uses systematic studies of fractures (including fractography)
in the Chalk and seismic data to establish the evolution of the regional structure and
related fractures on the Isle of Thanet, Kent, southern England. The area has received
little attention with regard to its fracture pattern and tectonic evolution. Burial, tectonic
8 M.S. AMEEN
and uplifting stresses, mostly related to basement block faulting, including extensional and
transtensional/transpressional movements, caused the evolution of the fractures and the
host regional forced fold. The Thanet monocline is divided into a minimum of four zones
(domains), according to fracture characterization. The study illustrates the effectiveness of
combining seismic data with field observations to determine the tectonic evolution of a
region. In addition the results are significant in that they touch on the fracturing of the
Chalk, an important subject to a number of engineering and hydrocarbon projects (e.g.
North Sea reservoirs, the Channel Tunnel, nuclear power stations).
Roberts describes the joint system in a 2.5 m thick slice of interbedded limestones and
mudstones located in the upper part of the Saint Mary's Well Bay Formation (Lower
Liassic) exposed locally between Lavernock Point and Saint Mary's Well Bay, South
Wales. Three systematic joint sets are recognized which strike 190~, 280 ~ and 335 ~
respectively. The three sets show fracture surface markings including hackle plumes, point
sources (origins), ribmarks and twist hackle plumes. The origins are consistently located in
the limestone layers from which the hackle plumes spread with bedding-parallel axes. The
latter indicate that the leading edge of the joints propagated parallel to the bedding. The
joints terminate in the overlying and underlying mud rocks with deep borders of twist
hackles. The persistent location of joint origins in the limestone layers and their
termination in the mudrocks are attributed to the different elastic moduli of these two rock
types. The limestone acts as a stress concentrator and provides sites for fracture initiation.
Although limestone lithifies earlier than the mud sediments, there is no evidence that joints
in the studied limestone developed early and later propagated to the surrounding mud
rocks. The three sets of joints show the same value of twist angles (10~176 The sense of
twist in each set is consistent, counter-clockwise for the 190~ joint set and clockwise for
each of the other sets. The relatively small twist angle is attributed to jointing under
relatively low confining pressure. The sense of twist is used to deduce the orientation of the
maximum and minimum principal stresses during the propagation of these joints. Joint
architecture suggests that the 190~ joint set is younger than the other two sets. No firm
conclusion is reached concerning the relationship of the joints to the local structures.
Part 2: Non-fractographic studies

The second part of the volume includes three studies. The first by Cosgrove addresses
tensile fractures in unconsolidated, unlithified sediments which tend to have their delicate
surface patterns (e.g. hackle plumes) obliterated by weathering, filling, compaction and
buckling of the fractures. In such circumstances the dynamic evolution of the fractures
cannot be documented by fractographic features. Therefore some explanatory mechanism
must be found. To illustrate this problem Cosgrove briefly considers sedimentary dykes.
In some regional field studies fractographic features are not clear due to the quality of
exposure, direction and degree of sunlight, etc. Such studies rely mainly on statistical
fracture population analysis. As an example, Goodwin presents the results of a study of
joints in the flat lying layers of the Upper Cretaceous Chalk of eastern England.
Orientation data were collected from 20 localities (no morphological features on joint
surfaces were identified by the author) and analysed using stereographic projection and
cluster analysis. The studies reveal the occurrence of two regional dominant sets
(according to their orientation) striking N W - S E and NE-SW respectively. The author
suggests that these sets are the same as those that occur in southern England, and
recognizes a change in the joint plane geometry of each set from steep conjugate 'hybrid'
INTRODUCTION 9
joints in Norfolk to sub-vertical extension joints in Humberside. Such a change is mainly

referred to northward increase in rock strength and hardness and a possible decrease in
differential stress towards the north. Stratigraphic and regional structural control on
jointing are considered to play a minor role in controlling joint geometry and their
variations.
Spacing is one result of the interaction of neighbouring and parallel fractures that do not
interact by cross-cutting/abutting. Instead they affect the spatial and temporal distribution
of their neighbouring penecontemporaneous fractures. In cases where the fracture traces
are the only source of information available, and there is a lack of exposed fracture
surfaces for fractographic observations, a theoretical explanation is needed for the factors
controlling the spacing between the fractures of one set. This is thoroughly discussed by
Gross et al. who give a detailed explanation of Hobbs' (1967) analysis of joint spacing and
attempt to clarify the assumption and implications of the one-dimensional model offered
by Hobbs. Gross et al. use a two-dimensional finite element numerical simulation of a
crack confined within a bed, to describe the stress and displacement fields around the
profile section of a crack in an interbedded sedimentary rock mass. They use field studies
in the Monterey Formation of California to illustrate their findings.
Suggested target areas for future research in the field of fraetography in

geology
Fractography in geology is a 'young science' and can therefore benefit from more studies
into any of its aspects. However, to achieve balance, key areas must be recognized and
given priority in future studies. In need of most urgent attention are:
1. A comprehensive review of the relatively vast nomenclature used in fractography to

establish a universally accergted 'fractographic code' for fractographers.
2. Microscopic studies of fracture surfaces (microfractography). This has been the most
neglected aspect of fractog~: aphy in spite of its great potential in advancing our knowledge
and understanding of fracture mechanics.
3. Systematic and detailed 3D mapping of fractographic features of fractures that are

induced experimentally under well constrained conditions, to determine reliable criteria
for quantitative interpretation of surface features.
4. Analysis of surface features of fractures as a part of core logging. The amount of work
and data collected on this subject is insignificant compared to that carried out on field
exposures.
This volume stems from a joint Petroleum Group-Tectonic Studies Group meeting held at the
Geological Society in September 1993. I would like to thank these Groups for sponsoring the
meeting. The efforts of the Conference Officer at the Geological Society during and before the
meeting are greatly appreciated. Thanks are due to the authors who contributed to the meeting and
this volume, and to the reviewers of the papers.
10 M.S. AMEEN
References
BAHAT, D. 1991. Tectonic-fractography. Springer-Verlag, Berlin.
BANKWXTZ,P. 1966. Uber Klufle, II. Geologie, 15, 896-941
HOBBS, D. W. 1967. The formation of tension joints in sedimentary rocks: an explanation. Geological
Magazine, 104, 550-556.
HODGSON, R. A. 1961. Classification of structures on joint surfaces. American Journal of Science, 259,
493-502.
KULANDER, B. R., BARTON, C. C. & DEAN, S. L. 1979. The application offractography to core and
outcrop fracture investigations. Report No. METC/SP-79/3, United States Department of
Energy, Morgantown, West Virginia, USA.
, DEAN, S. L. & WARD, B. J. 1990. Fractured core analysis: interpretation, logging, and use of
natural and induced fractures in core. American Association of Petroleum Geologists, Methods in
Exploration Series, 8, Tulsa, USA.
PARKER, J. M. 1942. Regional systematic jointing in slightly deformed sedimentary rocks. Bulletin of
the Geological Society of America, 53, 381-408.
ROBERTS, J. C. 1961. Featherstone-fracture and the mechanics of rock jointing. American Journal of
Science, 259, 481-492.
SYME-GASH P. J. 1971. A study of surface features relating to brittle and semibrittle fracture.
Tectonophysics, 12, 349-391.
WOODWORTH, J. B. 1896. On the fracture system of joints, with remarks on certain great fractures.
Proceedings of the Boston Society of Natural History, 27, 163-184.
ZAPFFE, C. A. & CLOGG,M. Jr. 1944. Fractography: a new tool for metallurgical research. Preprint 36,
American Society of Metals.
From Arneen, M. S. (ed.), 1995, Fractography: fracture topography as a tool in fracture mechanics and stress analysis
Surface morphology characteristics of rock discontinuities

with particular reference to their genesis
OMER AYDAN l & YASUHIRO SHIMIZU 2

1Department of Mar&e Civil Engineering, Tokai University, Orido 3-20-1,
Shimizu 424, Japan
2Department of Civil Engineering, Meijo University, Shiogamaguchi,
Tenpaku-ku, Nagoya 468, Japan
Abstract: A research program was undertaken to characterize the surface

morphology of natural discontinuities with particular reference to their genesis.
For this purpose a special profilometer, which could easily be transported and
operated in the field, and was capable of measuring 480 mm long profiles, was
designed. The recorded profiles were digitized and analysed to obtain surface
morphology characteristics of the discontinuities in different rock types found in
Central Japan. Natural discontinuities have an anisotropic surface topography
which may be characterized by measuring profiles along their eigen directions.
Consequently, surface morphology parameters are directional and differ
depending upon the discontinuity type.
Rock discontinuities in the form of cracks, joints, bedding planes, schistosity planes, etc.,
are commonly found in rock masses in the upper part of the Earth's crust (Pollard &
Aydin 1988). Discontinuities in rock masses influence the stability of rock engineering
structures as well as thermo-hydrological characteristics of rock masses, which have
become a field of interest in relation to radioactive waste disposal projects (Aydan &
Kawamoto 1990). There have been numerous experimental and numerical studies for the
characterization of the surface morphology of discontinuities and for relating the surface
morphology parameters to their thermo-hydromechanical properties in recent years
(Sezaki et al. 1993). When the authors undertook a project to investigate the
hydromechanical characteristics of rock discontinuities, it was realized that there were
two areas of study missing in most of the previous work: firstly, hardly any attention had
been paid to the genesis of discontinuities; and secondly most experiments were carried out
in laboratories by splitting rocks or rock-like m a t e r i a l s - i.e. experiments on natural
discontinuities were limited to those which could be transported and tested in the
laboratory.
The purpose of this study was to present a procedure for quantifying surface
morphological characteristics of discontinuities in rock masses with particular reference
to their genesis. Firstly, a classification for discontinuities in rock masses is proposed, then
a system developed for measuring and quantifying the surface morphology characteristics
of the discontinuities is described. Finally, in situ measurements carried out on various
kinds of discontinuities in different rocks, found in Central Japan are presented and
discussed.
12 6. A Y D A N & Y. S H I M I Z U
Classification of discontinuities in rock masses

Discontinuities in rock masses can be classified in various ways (Pollard & Aydin 1988;
Aydan & Kawamoto 1990). Proposed here is a classification based on the order of
occurrence as follows:
(1) Primary (eigen) discontinuities which are associated with the genesis of rocks. Such
discontinuities are termed flow planes in igneous rocks, bedding planes in
sedimentary rocks and schistosity planes in metamorphic rocks. These planes are
easily distinguished in sedimentary and metamorphic rocks while they may be
difficult to recognize in igneous rocks. Nevertheless, observations on platy minerals
(e.g. micas) and elongated minerals (e.g. turmaline and hornblend) generally help in
the recognition of such planes.
(2) Secondary discontinuities which develop in rocks as a result of desiccation, cooling
and freezing of rocks. Desiccation discontinuities are generally observed in
sedimentary rocks while cooling discontinuities occur in initially hot igneous or
metamorphic rocks. Freezing discontinuities can be observed in any kind of rock.
These planes are usually perpendicular to primary (eigen) discontinuities. These
planes can generally be classified as grain planes, hardway planes and sheeting
joints. The cause of these discontinuities is mainly tensile stresses.
(3) Tertiary discontinuities which develop in rocks due to tectonic forces. They occur
during folding and faulting processes, and are classified as shear and tension
planes.
Surface morphology characterization

The characterization of the surface morphology of rock discontinuities is merely a
geometrical procedure. In other words, it is an identification procedure of the topography
of the discontinuity surfaces, which specifically involves height, shape and periodicity of
protrusions, and the ratio of the surface area to the base area.
It will be appropriate if a function which can represent the topography of discontinuity
surfaces can be found. This is extremely hard not only because of the difficulty in finding
an appropriate function but also because of the enormous efforts required in measuring
and processing the data. As a result of this, most of available techniques are based on
linear profiles and various characterization parameters are proposed (Myers 1962; Sayles
& Thomas 1977; Tse & Cruden 1979; Thomas 1982; Tfirk et al. 1987). The parameters
associated with linear profiles are height of asperities, inclination of asperity walls, length
of asperity wall relative to base length, and periodicity of asperities. These parameters are
briefly explained in the following:
Centre-line average height (CLAH) is defined as;
CLAH = Z1 fx =o
x=L Ir (1)
where L is the measurement length, x the distance from the origin and r the height of the
profile from the reference base line.
SURFACEMORPHOLOGYCHARACTERISTICSOF ROCKDISCONTINUITIES 13
Mean standard variation of height (MSVH) is defined as:
MSVH -- ~1 fx =o
x=L r (2)
Root mean-square of height (RMSH) is defined
RMSH = r (3)
=0
Ratio of profile length (RPL) is defined as:
1 fxz=L 1 fxx=L ( de) 1/2

RPL = ~ =o ds = ~ =o 1 + dxx dx. (4)
Weighted asperity inclination (WAI*) is defined as:

1 fz x=L de
WAI* = ~ =o I ~ xx I dx, WAI = tan-l(WAI*). (5)
Mean standard variation of inclination (MSVI) is defined as:
MSVI = ~ o dxx dx. (6)
Root mean-square of inclination (RMSI) is defined as:
RMSI= dxx dx
]12 (7)
=0
Auto correlation function (ACF) is defined as:
1
ACF = ~
fx =o
x=L r162 + ~)dx. (s)
Structure function (SF) is defined as:
SF = ~1 fx --o
x=L [r - r + ~)]2d~, (9)
where T is a measure of the periodicity of asperities.
Fractal dimension is defined as:

N = CU D, (10)
where N is the number of steps, C is a constant, l is step length and D is a fractal

dimension. Since the following relation holds between the total length of the profile and
the step length:
14 O. AYDAN & Y. SHIMIZU
I. 530 _,
L Rec~ ~1 l~ = ~ tR~
81 I Pen-Recorder 'l ,I1
Drum-likeHandle FRONT
VIEW
Supports
SIDEVIEW
~
0 i'01111
I. olo ,W
PLANVIEW
Fig. 1. Views of the profilometer.
L = N1, (11)
then the above equation can be rewritten as:
L= Ct ~-~ (12)
The most crucial aspect of characterization using linear profiles is how to select the
appropriate direction of measurement. Although there are no guidelines for this, the usual
practice seems to be a line which is perpendicular to the axis of elongated protrusions by
taking into account the maximum shear resistance due to geometrical dilatancy of the
surface. Nevertheless, it is difficult to say that such profiles are representative of the
discontinuity surfaces.
It is well known in fracture mechanics of rocks and metals that the ridges of
discontinuity surfaces develop almost perpendicular to the crack propagation direction.
SURFACE MORPHOLOGYCHARACTERISTICSOF ROCK DISCONTINUITIES 15
136E 137E 138E 139"~E

| | t / 'l
~Tsuge (granite)
(~)Otake (andesite, basalt)
37N" JAPANS E A j ~ | (granite)
(~Kitamatado(shale)
@Okumino(rhyolite, sandstone,
HONSHUISLAND chert, limestone, basalt)
(~)Chita(mudstone)
36N' | (~Echizen(tuff)
|
35N
34N
PACIFICOCEAN
l I l
Fig. 2. Locations of the field measurements.
Furthermore, the ridges in sedimentary rocks are also perpendicular to the flow direction
of fluids, as is well known in sedimentology. By taking these facts into account it is
possible to propose that there are generally two mutually perpendicular eigen directions;
one perpendicular to the ridge axis and the other parallel to it. Measuring profiles along
these directions may be sufficient to characterize the surface morphology of disconti-
nuities. Consequently, one may introduce an elliptical coordinate system so that the
principal axes of such a coordinate system coincide with those eigen directions. Such a
coordinate system would be appropriate for many discontinuity types found in rock
masses.
Surface morphology characterization system

To measure a linear profile there are two principal measuring techniques. (1) Mechanical
devices of the contact type, which can be further subdivided into two groups - stylus of
needle type and stylus with roller tip. (2) Optical devices of the non-contact type which
utilize reflected light beams, i.e. laser beams.
Measurements and data acquisition are generally carried out manually, semi-
automatically or automatically. The automatic systems are desirable in that measuring
is a laborious procedure. However, such systems are unsuitable for measurements under
various in situ conditions as they require heavy equipment for their power supply, driving
16 O. AYDAN & Y. SHIMIZU
BEDDING PLAN ES
- - ~ ~ ~ P 1
BP3
B P 6
BP7
BP8
TENSION D ISCON TIN U ITIES TD1
TO2
TD3
TD4
TD5
TD6
TD7
SHEAR PLANES
-- SD1
SD2
= S.D4
SO5
"""'---'~"-~'~_... SD6
0 50 mm
Fig. 3. Measured profiles.

I I
systems for the measuring stylus and data acquisition. As a result of this, their use is
generally restricted to the laboratory. Since our main purpose was to make measurements
on actual discontinuities, lightweight equipment was needed. Therefore, a measuring
system consisting of a manually operated profilometer for site measurements, a digitizing
system for recording profiles, a microcomputer for data storage and a data processing sub-
system for the digitized data has been developed. Figure 1 shows several views and
SURFACE MORPHOLOGY CHARACTERISTICS OF ROCK DISCONTINUITIES 17
dimensions of the profilometer; it is ma n u a l l y operated, weighs c. 5 kg and is made mainly

of aluminium. The diameter of the roller-type stylus is 3 mm, which is the smallest one
available on the market. Profiles are recorded on a sheet using a pen-recorder. Profiles up
to 480 mm, with a m a x i m u m asperity height of 90 mm, can be measured. The apparatus
can be used on surfaces having an inclination varying between 0 and 90 ~ and it can be
carried to any site to which h u m a n s have access. All recorded profiles are digitized using
a digitizer and stored in files using a m i c r o c o m p u t e r at the authors' universities. The
digitized data are then used to obtain surface m o r p h o l o g y parameters as defined in the
previous section.
Table 1. Rock and discontinuity types at measurement sites
No. Location Rock type Discontinuity type

1 Tsuge Granite Flow plane, grain plane, hardway plane
2 Otake Andesite Flow plane, grain plane, hardway plane
Basalt Flow plane, grain plane
3 Nakatsukawa Granite Flow plane, grain plane, hardway plane, sheeting joint
4 Kitamatodo Shale Bedding plane, tension joint
5 Okumino Rhyolite Flow plane, grain plane, hardway plane, shear discontinuity
Sandstone Bedding plane, tension discontinuity
Chert Bedding plane
Limestone Bedding plane
Basalt Cooling joint
6 Chita Mudstone Bedding plane, shrinkage joint
7 Echizen Tuff Bedding plane, tension discontinuity and shear discontinuity
8 Chausuyama Hornfels Schistosity plane, grain plane, hardway plane
Measurements and discussion

Measurements have been ma d e on discontinuities in different rocks found in Central
Japan. Figure 2 shows the locations of measurements and Table 1 gives rock and
discontinuity types in the respective localities. Figure 3 shows several of the profiles
l
Fig. 4. Notations for measuring profiles of discontinuities.
18 O. A Y D A N & Y. S H I M I Z U
T 5 mm
0
BEDDING PLANE (BP6) ! 1
0 20 mm
TENSION DISCONTINUITY (TD4)
SHEAR DISCONTINUITY (SDE)
Fig. 5. Profiles of selected discontinuity types shown in Fig. 3.
measured. Measurements on discontinuity surfaces were carried out along the eigen
directions of surfaces; the x-axis was selected to be parallel to the ridge axis and the y-axis
perpendicular to the ridge axis (Fig. 4).
Table 2. Surface morphologyparameters
Discontinuity Height parameters Inclination parameters Periodicity parameters RPL Fractal

plane CLAH MSVH RMSH WAI WAID MSVI RMSI T for ACF r for SF dimension
(mm) (mm 2) (ram) (o) ( o) (mm) (mm)
BP6 2.301 6.312 2.512 1 0 . 7 2 3 1 . 5 6 4 0.052 0.229 11.50 12.13 1.025 1.009
TD4 2.174 6.696 2.586 11.45 1 0 . 0 2 2 0.063 0.251 14.54 13.85 1.030 1.010
SD5 2.212 6.097 2.469 8.92 0.547 0.036 0.189 23.62 19.65 1.017 1.006
Figure 5 shows digitized profiles of a bedding plane (BP6), tension discontinuity (TD4)
and shear discontinuity (SD5) whose original profiles are shown in Fig. 3. The sampling
interval for digitizing profiles was 2 mm in these particular examples. Table 2 gives the
values of surface morphology parameters defined in the preceding section. However, it
should be noted that computed results are generally influenced by the sampling interval
and profile length. Figures 6 & 7 show the effect of sampling interval and profile length,
respectively, on parameters CLAH, WAI and RPL for each discontinuity. The effect of the
sampling interval on parameter CLAH is almost negligible; on the other hand, parameters
WAI and RPL decrease in magnitude as the sampling interval increases (Fig. 6).
Therefore, it is desirable to make the sampling interval as small as possible. However, it
should be noted that the sampling interval in digitizing profiles manually is generally
restricted to >I 1 mm.
4 .1 04~- BEDDINGPLANE 1 0 ~ ,
9 [ (BP6)
8~~
~: 1.02
a
F o
5
o ,.o, F ~ ~ ~-~
11 ' I I ]-~i L , J / , I ~ I , I , I
0 10 20 30 40 "0 10 20 30 40 0 10 20 30 40
S A M P L I N G I N T E R V A L (ram) S A M P L I N G I N T E R V A L (mm) S A M P L I N G I N T E R V A L (mm)
TENSION DISCONTINUITY
4- 1.04 r (TD4) 10
_,3 1.03 ~- 6- 8
E
v
E
~-2 _ ~ - ~ ~: 1.02 -~64
~:
5
o
1 1.01 2
i I t I i I , I
0 10 20 30 40 1 , 40
4 0 10 20 30 40
S A M P L I N G I N T E R V A L (mm) S A M P L I N G I N T E R V A L (mm) S A M P L I N G I N T E R V A L (mm)
SHEAR DISCONTINUITY
IB
1.04 - (SD5) 10 F
3 1.03 o-~8 ~
E
v
E
z2 ~ 1.o2
5
O
1
I I i I I I , I - I , I , I , I
0 10 20 30 40 10 10 20 30 40 0 10 20 30 40
S A M P L I N G I N T E R V A L (mm) S A M P L I N G I N T E R V A L (mm) S A M P L I N G I N T E R V A L (mm)
Fig. 6. The effect of the sampling interval on surface morphology parameters CLAH, WAI and RPL.
20 0. AYDAN & Y. SHIMIZU
1.04- BEDDINGPLANE 12-

(BP6)
1.03 9
Q.
,.,- 1.02
o 1 1.01 - 3
I I , I 10, I , I , I , I , I i I , I , I
40 80 120 160 4O 80 120 1600 4O 80 120 160
PROFILE LENGTH (mm) PROFILE LENGTH (mm) PROFILE LENGTH (mm)
..3
o
4
,o4
1.06 - ,
1.02
TENSION DISCONTINUITY
(mD4) 16
2 !
4
!
0 40 80 120 160 10 40 80 120'1600 40 80 120 160

PROFILE LENGTH (mm) PROFILE LENGTH (ram) PROFILE LENGTH (mm)
E
-r-
5
4!i 1.04r-
SHEAR DISCONTINUITY
(SD5) 10
o
1.01 ~
, I , I I I I I 4/ ~ ! a. I , 1 9 9 I I I t I , I .I I
0 40 80 120 160 "0 40 80 120 1600 40 80 120 160
PROFILE LENGTH (mm) PROFILE LENGTH ( m m ) PROFILELENGTH (mm)
Fig. 7. The effect of the profile length on surface m o r p h o l o g y parameters C L A H , W A I and RPL.
30*
6o"
120 ~
180~._._ _
240 e
o 50 mm
Fig. 8. Profiles of a sheeting joint.
The length of profiles may also influence computed results. For this reason
computations were carried out varying the length of digitized profile for a sampling
interval of 2 mm. As seen from Fig. 7, if the profile length is greater than the wavelength of
the main asperity of a given discontinuity, then the influence of the profile length becomes
less pronounced.
As pointed out in the preceding sections, we discuss whether eigen directions for
discontinuity surfaces exist or not. For this purpose, the direction for measuring profiles
was varied by 30 ~ on several discontinuity types. As a specific example, the profiles of a
sheeting joint in Nakatsugawa granite for respective directions is shown in Fig. 8. The
profile length was 100 mm for all directions. The sampling interval was taken as 1 mm
during digitization and computed results for surface parameters CLAH, WAI and RPL
are shown as a function of measuring direction in Fig. 9. As seen from Fig. 9, peaks and
troughs were observed in computed results and they were mutually orthogonal to each
other. The peaks were observed when a profile was perpendicular to the ridge axis and
troughs were observed when a profile was perpendicular to the ridge axis and troughs were
observed when a profile was parallel to the ridge axis. This tendency was also observed in
22 6. A Y D A N & Y. S H I M I Z U
m m
SHEETING JOINT
E 4
E
"1"- 3
(3 2-
I I ,,I I I I
0 60 120 180 240 300 360
15-- MEASURING DIRECTION (o)
6"-
~ 10-
I I I I I I
0 60 120 180 240 300 360
1.2+ MEASURING DIRECTION (o)
rrl.1
1.0
0 60 120 180 240 300 360
MEASURING DIRECTION (o)
Fig. 9. Variation of surface morphology parameters CLAH, WAI and RPL as a function of direction
of measurement.
other types of discontinuities measured. The directions of peaks and troughs may be taken
as spectra of eigen directions. Furthermore, these measurements implied that surface
morphology parameters were strongly anisotropic. As a result of this, it is most likely that
the associated hydromechanical characteristics of discontinuities will be also anisotropic.
Next, interrelations between parameters CLAH, WAI and RPL for discontinuities
found in Nakatsukawa granite, Okumino rhyolite, Otake andesite and basalt were
investigated. Discontinuities measured were classified as either primary or secondary.
25 I i I i I
X Y
9 o PRIMARY DISCONTINUITY
9 [] SECONDARY DISCONTINUITY
20
eo SECONDARY DISCONTINUITY
111 - 9
W 15
0
111
0
O
10 <> o
i 00~0 (~0
~eO 9 9
I ! I
0 1 2 3 4 5 6
CLAH (turn)
Fig. 10. Relation between parameters CLAH and WAI.
Figure 10 shows the relation between parameters WAI and CLAH for all discontinuities
found in these igneous rocks. The values of WAI for primary discontinuities seem to be
smaller than those for secondary discontinuities. Nevertheless, the data plotted do not
support any correlation between WAI and CLAH. This may be a natural consequence,
since WAI depends on the ratio of the height to the wavelength of asperities.
Figure 11 shows the relation between RPL and CLAH. The data plotted are highly
scattered, which may be a result of the difference in grain sizes of the rocks and the
direction of measurement. The correlation between these two parameters is not so well
defined, but it seems that RPL increases as CLAH increases.
Finally, we checked the relation between WAI and RPL (Fig. 12). A well-established
relation between these two parameters was observed. As RPL increased so WAI
24 ,3. AYDAN & Y. S H I M I Z U
1.10 I I I I I
X Y
0 PRIMARY DISCONTINUITY
[] SECONDARY DISCONTINUITY
,O, 0 SECONDARY DISCONTINUITY
.O.
_1
n
rr
1.05 0
oO ~ [] O o
9 0O 0
P oop q) o
eO e
I I I
3 4 5 6
CLAH (ram)
Fig. 11. Relation between parameters CLAH and RPL.
proportionally increased. Furthermore, the width of scattering was small as compared

with the data plotted in Figs 10 & 11.
Conclusions
In this paper, the quantification of surface morphology of discontinuities in rock masses
with particular reference to their genesis has been considered and the following
conclusions are drawn.
(1) The discontinuity surfaces can be characterized using the profiles along the eigen
directions of discontinuity surfaces. These eigen directions are mutually perpendi-
25 ' ' ' ' "' I ' ' ' '
x Y
9 O PRIMARYDISCONTINUITY
[] [] SECONDARYDISCONTINUITY
20 -
~ SECONDARYDISCONTINUITY
UJ
W
15 4~ .
(.9
00 ~1 . 0 5 1.10
RPL
Fig. 12. Relation between parameters RPL and WAI.
cular to each other. One eigen direction is parallel to the ridge axis and the other
perpendicular to it.
(2) Asperity inclinations have a minimum along the ridge axis and a maximum
perpendicular to the ridge axis.
(3) There is no great difference between rock types with regard to the function CLAH,
which is a measure of asperity height. However, C L A H is generally higher in the
case of secondary and tertiary discontinuities as compared with that in the case of
primary discontinuities.
(4) The relation between W A I and C L A H for all measured data is poor and does not
support any correlation between these two parameters.
(5) The relation between WAI and RPL is well established - as RPL increased so WAI
increased proportionally.
26 <3. AYDAN & Y. SHIMIZU
The authors acknowledge partial funding for this research by the Research Foundation for Electrotechnology of
Chubu (REFEC), Nagoya, Japan.
References
AYDAN,O. • KAWAMOTO,T. 1990. Discontinuities and their effect on rock mass. Proceedings of the
International Conference on Rock Joints, Loen, 149-156.
MYERS, M. O. 1962. Characterization of surface roughness. Wear, 5, 182-189.
POLLARD, D. D. & AYDIN, A. 1988. Progress in understanding jointing over the past century.
Geological Society of America Bulletin, 100, 1181-1204.
SAYLES,R. S. ,~ THOMAS,T. R. 1977. The spatial representation of surface roughness by means of
the structure function: a practical alternative to correlation. Wear, 42, 263-276.
SEZAKI, M., AYDAN,6. & SHIMIZU,Y. 1993. Development of a data-base system for the surface
morphology characteristics and mechanical properties of rock discontinuities. Proceedings of the
25th Japan Rock Mechanics Symposium, Tokyo, 11-15.
SHIMIZU, Y., AYDAN, (~. & KAWAMOTO, Z. 1993. Surface morphology characteristics of rock
discontinuities. Proceedings of the 14th Western Japan Rock Engineering Symposium, Fukuoka,
65-70.
THOMAS, T. R. 1982. Rough Surfaces, Longman, London.
TSE, R. & CRUDEN, D. M. 1979. Estimating joint roughness coefficients. International Journal of
Rock Mechanics and Mining Science, 16, 303-307.
TORK, N., GERIG, M. J., DEARMAN, W. R. & AMIN, F. F. 1987. Characterization of rock joint
surfaces by fractal dimension. Proceedings of the 28th US Symposium on Rock Mechanics, 1223-
1236.
The anisotropy of surface roughness measured using a

digital photogrammetric technique
MARK W. JESSELL, 1 SIMON J. D . C O X , 1'2 P H I L L I P SCHWARZE 1
& WILLIAM L. P O W E R 2
1Victorian Institute of Earth and Planetary Sciences, Department of Earth
Sciences, Monash University, Clayton 3168, Victoria, Australia
2CSIRO Division of Exploration and Mining, 39 Fairway, Nedlands,
WA 6009, Australia
Abstract: The characterization of fracture surface morphology is important for the

understanding of shear and normal deformation behaviour of rock masses, and for their
transport properties. It has been suggested that natural fracture surfaces might be
fractal. If true, this would have dramatic implications for these rock properties and
would provide a sound basis for future modelling.
Evaluating this hypothesis requires measurement of fracture surface morphology over
a large range of scales. A software package has been developed that can automatically
reconstruct surface morphology information from stereo pairs at any scale, using edge
correlation and point matching.
The data derived using this technique compare favourably with results generated
from a traditional surface profilometer survey, and results are presented from the
analysis of the anisotropy of blasting-induced tensile fractures in granite from Mt
Alexander Quarry, Harcourt, Victoria, Australia.
The mechanical and physical properties of low porosity rocks at low pressures are strongly
influenced by the presence of fractures. Examination of normal and shear deformation
(e.g. Bandis et al. 1983), and fluid and electrical transport (e.g. Brown 1989) have
highlighted the effects of the shape or roughness of the fracture surface. This is most
important at shallow levels in the crust, which is within the realm of most engineering
studies, or in zones of low effective confining pressures.
Traditional engineering analyses have considered surface roughness within the scales of
0.01-1 m as affecting frictional strength (Thomas 1982). However, recent studies of rock
surfaces suggest that roughness of the surfaces occurs at a wide range of scales, and the
surfaces may be approximately self-affine or fractal (Power & Tullis 1991). In this context,
fractal means that surface irregularities are present at all scales, with longer wavelength
irregularities having a larger amplitude and that some form of power law describes the
change in roughness with scale.
There are some reasons why rock fractures might be fractal or chaotic, namely that they
develop under high energy conditions where the small-scale local fracture orientations
might depend on a wide variety of influences, including material heterogeneity, pre-
existing fractures and damage features, or other flaws. Alternatively, rock fractures might
be essentially Euclidean at large scale because they develop under the influence of a
continuous, well-behaved stress field. In the latter case, it might be expected that the large
scale (1-10 m) fractures will be essentially flat, while at scales approaching the rock grain
size the surface will be relatively rougher. Furthermore, once fracture surfaces are
28 M . W . JESSELL E T AL.
Fig. 1. Locality of the quarry where samples were collected for this study. Harcourt is located at
144~ 36~
activated as the locus for sliding, there is ample evidence to suggest that their roughness
characteristics will evolve and that the anisotropy of roughness will then become a major
consideration in the analyses of fracture properties.
Evaluating these relationships requires measurement of fracture surface morphology
over a large range of scales. This is very difficult to do using a consistent method,
particularly in 2D, rather than profile, form, since most current techniques have an
inherently narrow range of the power spectrum over which reliable data can be collected.
We have developed a software package capable of automatically reconstructing surface
morphology information from stereo pairs at any scale (Schwarze et al. 1992). After initial
alignment of the images the software does not require user intervention, and thus is well
suited to the production of large datasets.
Method
The aim of all stereo photogrammetric techniques is to solve the stereo correspondence
problem, which involves the matching of the same location in both images of a stereo pair.
Heights are calculated by comparing apparent displacements (known as disparities)
between matched points in the two images. These disparities are proportional to the
heights of individual points (relative to some datum). In a manual technique matching of
points is performed by a human operator, but in the current technique it is solved by the
computer using a combination of algorithms based on vision theories. Current theories of
the human visual system suggest that the first visual cues used in depth perception are
edges and areas of marked contrast changes (Marr & Poggio 1979), and that other parts of
the image are then used to fill in the finer detail of the depth map. We have implemented a
technique based on this hierarchy (Schwarze et al. 1992).
DIGITAL PHOTOGRAMMETRY OF ROUGH SURFACES 29
et0
,'-', "~0
~ ,,...~
e~o
2~r~
e q ~
30 M . W . JESSELL E T .4L.
Fig. 3. Scattergram comparison of surface topographic information collected by profilometer and by

digital photogrammetry; > 90~ of the data fell within 1.1 pixels of an exact match.
It has been suggested that a 'short cut' method for the analysis of surface roughness may
be achieved by directly analyzing the spectral content of single grey scale images of
surfaces (Russ 1993). Unfortunately, this technique is not suited to the study of coarse
grained rocks of the type measured in this paper, as the speckled appearance of the rocks
resulting from dark and light minerals would overwhelm any shading variations due to
surface topography. Conceivably, this could be avoided by coating the surface with a
neutral density paint; however, this would not really be practical for the analysis of large
rock surfaces.
DIGITAL P H O T O G R A M M E T R Y OF R O U G H SURFACES 31
Quantitative assessment of technique

As a means of assessing the accuracy and precision of the results produced by our
algorithm a direct comparison with a different technique was necessary. To achieve this a
comparison was made between a profilometer survey over a quarrying-induced fracture
and a digital photogrammetric analysis of the same sample. The sample was a flake of
granodiorite from Mt Alexander Quarry, Harcourt, Victoria, Australia (Fig. 1), and a test
area 256 • l l 6 m m was defined for comparison (Fig. 2a). The profilometer works by
tracking a stylus across the sample, with the surface elevations measured on a rectangular
grid at 1 mm spacing. The heights were measured with a precision of 4-0.1 #m using a
linear variable differential transformer displacement sensor (Fig. 2b); the total height
range measured was 13.7 ram.
In order to produce a comparable grid using digital photogrammetric techniques two
medium format (4 • 5 cm negative) photographs were taken at a distance c. 30 cm from the
surface of the flake; 8 • 10 cm prints were then made of the negatives and each print was
then scanned at a resolution of 75 pixels per inch (c. 30 pixels per cm) on a standard flat-
bed scanner, creating two 8 bit (256 grey level) digital images of the area of interest. These
images provided the raw input to the photogrammetry program, with additional
alignment information provided at this stage. The output from the program was a data
file of heights for individual pixel locations, with null values for those pixels for which the
program could not make a successful match. These null values were filled in using a thin-
plate spline interpolation scheme, producing a complete coverage of the test area (Fig. 2c).
High frequency noise spikes remain where incorrect matches were made.
The output from each of the data sets covered the same area of the sample and were of a
similar horizontal scale. In order to allow direct comparison of the two data sets we first
normalized the horizontal scales; however, there are some relative distortions in position
(resulting from parallax) which have not been corrected for (as this post-processing option
has not yet been developed).
Since the horizontal scales were now the same, the images could be overlaid and
compared pixel for pixel to assess the amount of correlation between the height values.
The results of this comparison are displayed as a contoured scattergram (Fig. 3). A
scattergram is compiled by taking the value of each pixel in one image and plotting it
against the value at the same pixel position in the second image, i.e. the pixel in row one
column one in the profilometer image v. the pixel in row one column one in the
photogrammetric image, and so on for each pair of locations. In a pair of identical images
normalized to a range of 0-255 pixels the plot would be a straight line. The contour level
of each of the plotted points in the scattergram corresponds to the per cent of pixels out of
the total which fall on that position.
The scattergram of the data from the profilometer and the stereo correspondence shows
a spread away from the ideal line. The points which deviate from the straight line have
different height values for the two techniques, the greater the difference the larger the
deviation. Nevertheless, there is a strong degree of correlation between the two data sets,
with 90% of the data falling within 1.1 pixels (0.059mm) of an ideal correlation, and
80% of the data corresponding to 0.88 pixels (0.047mm). This result may actually
underestimate the accuracy of the technique because errors introduced by parallax and the
interpolation scheme were ignored; however, until further work is carried out it will be
assumed that a vertical resolution equivalent to the distance represented by 1 pixel in the
raw image can be achieved.
Fig. 4. Quarrying-induced fracture in granodiorite plug from Mt Alexander Quarry, Harcourt,

Victoria. The rock face immediately above the stepladder was the test area shown in Fig. 5. The
radiating ridges and grooves on the fracture surface emanate from the rupture nucleation site at the
base of the drill hole, where the explosives were sited. Scale bar below and to right of step ladder
represents 2 m.
To improve the resolution, the horizontal scale that each pixel represents needs to be
decreased. This can be done either by increasing the magnification of the photographic
technique or by scanning the resultant negative at a higher resolution, both of which are
easily achievable with the correct equipment. With improved scanning equipment a
photographic negative can be scanned to 10 #m per pixel so that a 10 • 10 cm photograph
would produce an image with the dimensions of 10 000 • 10 000 pixels. On a 1 : 1 scale
photograph, for example, the pixel size would represent a true length of 10 #m on the
ground. The vertical precision of any elevation data can therefore be determined through
any combination of scale and scanning resolution for each application. It should be noted
that this vertical resolution still falls short of the 0.1 #m achievable using the profilometer;
however, the horizontal range of the profilometer is quite limited. The usefulness of the
digital photogrammetric technique lies in its application to a range of horizontal scales
(e.g. from satellites to microscopes) with a consistent methodology, rather than in its
ability to measure the finest vertical resolution at a given scale.
Application to Mt Alexander Quarry fracture

Photographs and calibration data from a quarrying-induced fracture in a granodiorite
plug from Mt Alexander Quarry, Harcourt, Victoria, Australia have been collected (Fig.
1). This single fracture covers an area over 10 • 10m, and the radiating ridges and grooves
on the fracture surface emanate from the rupture nucleation site at the base of the drill
-6
..=
.e
112
..i.a
xw...
9-= _~
~g
r~
e,)
t,~ .,.-,
ea)~
N e~0
34 M.W. JESSELL E T AL.
~Z
Q
o
~ "''~
eat)
"~
,.~ r~
Fig. 7. 2D fast Fourier transform images of the topographic data at two scales of measurement,
based on 256 • 256 pixel subsets of the images in Fig. 6. Brightness of individual pixels is
proportional to the log power spectral density for each pixel. The position of each pixel in this image
is plotted in polar coordinates, with the radial distance from the centre a function of frequency and
angular position a function of orientation. Lowest frequency pixels are in the centre of the image.
hole where the explosives were sited (Fig. 4). This fracture therefore represents a good test
of the measurement technique as it allows analysis of the observed anisotropy of
roughness.
For comparison purposes stereo photographs were collected at a number of scales;
however, in this paper the results from two scales only will be presented (with a scale
change of 9.4: 1) which will be referred to as 'near' and 'far' photos, covering a horizontal
field of view of 32 and 281 mm, respectively (Fig. 5). These two pairs of photographs were
scanned using a fiat-bed scanner at 75 dots per inch and then processed using the digital
photogrammetry program. Once the null matches had been filled by interpolation (which
(a) (b)
Fig. 8. Contours of log power spectral density, based on spatially averaged versions of the 2D fast
Fourier transform images shown in Fig. 7. These contours clearly show the change from anisotropic
roughness at lower frequencies to isotropic roughness at higher frequencies. This is apparent both
within the larger-scale image and in comparison with the smaller-scale image, and can be correlated
with the local orientation of radially dispersed ridges in the quarry face. Contours are drawn at levels
of 94, 91, 84 and 52% of peak power (at the central pixel).
represented < 7% of the total image in the near image and < 0.3% of the total image in
the far image) we had complete data for the two areas, which can be represented as two
grey scale images of the topography of the surfaces (Fig. 6), with higher areas shown as
lighter shades of grey.
Although these spatial representations of the data provide a useful dataset, charac-
terization of the scale-dependent and anisotropy nature of the roughness is better made in
the spectral domain, and so we have selected representative 256 x 256 pixel sub-areas of
each image for spectral analysis. First, a cosine mask was applied to the raw height image,
which minimizes the boundary effects of the image. These areas are transformed using a
2D fast Fourier transform (FFT) algorithm and are plotted as log power spectral density
(PSD) v. log spatial frequency (SF). Values are proportional to PSD for an orientation
and frequency defined by position in the 2D frequency domain (Fig. 7).
Of particular interest in this study is the information provided on the anisotropy of
roughness. By using a 2D transform a complete picture of the azimuthal dependence of the
roughness can be obtained, which is very laborious to produce using 1D sampling
techniques such as profiling. By applying a spatial averaging filter to the images of the
FFT spectra (which smooths out local fluctuations) contours of PSD close to the centre of
each F F T image, where the majority of the power is located, can be produced (Fig. 8).
Comparison of PSD contours for the two images reveals a striking contrast in the
anisotropy of roughness between the small- and large-scale images. In the small-scale
image the contours are close to circular for all the high power values, whereas the large-
scale image shows a marked axis of increased power at an angle of c. 48 ~ anti-clockwise
from the horizontal, indicating an important frequency component oriented at 90 ~ to this
axis, which reflects the local orientation of the surface ridges and grooves seen in Fig. 4.
The anisotropy gradually gets swamped by isotopic spectral behaviour at lower power
levels, which correlates with the lack of significant anisotropy in the small-scale image.
DIGITAL PHOTOGRAMMETRYOF ROUGH SURFACES 37
Conclusion
We have used a new solution for the stereo correspondence problem as applied to studies
of surface roughness. The method takes a digital stereo pair and automatically matches
corresponding points in the images in order to extract height information. The package
generates a dense dataset of heights, and comparison with profilometry data suggests that
the matching accuracy was e. 1 pixel for > 90% of the data. Application of this technique
to a quarrying-induced fracture provides a first insight into the mapping of roughness
anisotropies which may be achieved using this technique. Using current image acquisition
techniques, the strength of the digital photogrammetric technique lies in its applicability to
a broad range of horizontal scales with a consistent methodology, rather than in its ability
to measure the finest vertical resolution at a given scale.
References
BANDIS, C. C., LUMSDEN, A. C. & BARTON, N. R. 1983. Fundamentals of rock joint deformation.
International Journal of Rock Mechanics, Mineral Science and Geomechanics Abstracts, 20, 249-
268.
BROWN, S. R. 1989. Transport of fluid and electric current through a single fracture. Journal of
Geophysical Research, 94, 9429-9438.
MARR, D. & POGGIO, T. 1979. A computational theory of human stereo vision. Proceedings of the
Royal Society, London, B204, 301-328.
POWER, W. L. & TULLIS, T. E. 1991. Euclidian and fractal models for the description of rock surface
roughness. Journal of Geophysical Research, 96, 415424.
Russ, J. 1993. Surface Fractals. Plenum Press, New York.
SCHWARZE, P., JESSELL, M. & Cox, S. J. D. 1992. Surface reconstruction using digital photo-
grammetric techniques. 6th Australasian Remote Sensing Conference, New Zealand, Vol. 1,
298-307.
THOMAS, T. R. 1982. Rough Surfaces. Longman, New York.
Fractographic features on joints of KTB drill cores

(Bavaria, Germany)
P. B A N K W I T Z & E. B A N K W I T Z
GeoForschungsZentrum, Telegrafenberg A3, 14473 Potsdam, Germany
Abstract: The cores of the superdeep drilling in Bavaria (KTB: Continental Deep
Drilling, Windischeschenbach) were used to estimate the orientation of the con-
temporary maximum horizontal stress by several methods in the KTB-field laboratory,
including a study of core disking, borehole break-outs, cuttings and relaxation of the
cores. Core disking surfaces are one type of coring-induced joint occurring in quasi-
isotropic parts of the rock. Fractographic features on core disking surfaces aid in
understanding the core disking process. We determined the in situ stress orientation
from fractographic structures on common coring-induced fractures, which are present
in all sections of the cores. This analysis is based on the symmetry of joint characteristics
(3D shape and surface features of joint planes). The symmetry axes of these joint
features are related to the principal stress axes, which permits a determination of the
orientation of the recent maximum and minimum horizontal stress axis (SHr Sh).
The World Stress Map documents a contemporary tectonic stress field (Zoback 1992).
Near the southwestern border of the Bohemian Massif in the area of northern Bavaria,
Germany (Fig. l a), the continental superdeep borehole KTB is located in the crystalline
basement and penetrates the upper crust down to a depth of 9100 m. A pilot borehole was
drilled down to a depth of 4000 m, at a distance of 300 m from the main borehole. The
cores recovered from the boreholes consist of amphibolites, gabbros and layered
paragneisses with more or less intense, steeply dipping foliation. Between depths of
6900 and 7200m the borehole cuts the northeastward dipping fault system of the
Franconian Line between the upthrusted Bohemian Massif and the South German Block,
the latter covered in its uppermost part by Mesozoic and late Palaeozoic rocks (Fig. lb).
Coring-induced fractures in the KTB cores reflect the contemporary tectonic stress field.
The results of the joint analysis can be checked by different methods which were used at
the KTB-field laboratory by a special team to determine the maximum (Srt) and minimum
(Sh) horizontal stress axis (SH ~ Sh) by means of core disking (restricted to homogeneous
rocks of the drilling), borehole break-outs, cuttings and relaxation of the cores (Vernik
et al. 1992). Our studies deal only with joints in cores of the main borehole. Using
fractographic analysis additional data from the entire recovered core were collected. The
collected observations provide new results on the fracturing process of core disking
surfaces.
Method
During the past 30 years surface morphology of joint planes has received significant
attention. After initial studies of Woodworth (1896), Roberts (1961), Hodgson (1961),
Bankwitz (1966) and others, numerous contributions from geology and physics continue
40 P. BANKWITZ & E. BANKWITZ
q2
I
N
..:: ~.......~...o,~.. .
..s ,~,:.,,~... ~ .
.>-
f
7"
2" /
Y
",'(7 50
~-1,_
~'@/'r-O,
KTB / ' ~
drilling site
[ - - 1 platform k ~ ~
cover "~,4
~ latesand
phyllites
- ~ gneissesand
~
granulites
allochthonous
complexes %
~ Tepla-Barrandian
Complex %
~ Moldanubiangneisses
~ granulites 0 50 km
(a)
FRACTOGRAPHY KTB CORE 41
KTB MAIN B O R E H O L E
FORELAND BOHEMIAN MASSIF

I NE
o o o 4-
o o o o o o -I-
4- 4- _+0-
::- 4- 4-
4- 4-
.\§ 4-
.
-t-
- ooo -1000 -
. . . . .
4-
--2000 -2000
\
--3000 -3000
\
--4000 \ \ ' -4000-
Upper Cretaceous
\ j\
Triassic \
--5000 -5000-
Permocarboniferous ~ q
\
Granite ~. \ 5778 Fig. 7 and 14b
--6000 6359 Fig. 12 b -6000-
Orthogneisses ~ \ 6434
6435
Fig. 11
Fig. 3, 6 and 9
[ I Paragneisses 6668 Fig.
. . . . .12a,5,~:in
. =. 10a and b
--7000 6669 Hg. t~, I.~.
7012 Fig. 4 t-lg. 14c
b2
--8000 M b3
-8000
Metabasites
b4
--9000 -9000
b5
b6
Co)
Fig. 1. (a) Geological map of the western part of the Bohemian Massif (after Weber 1985). Broken
line: boundary between Saxothuringian and Moldanubian Zones. (b) Interpretation of the drilled
profile in a SW-NE section near the southwest boundary of the Bohemian Massif (after Hirschmann
1993). The drilling was finished at a depth of 9100m. SE1 and SE2 are strong reflectors in the
reflection seismic profile crossing the KTB. SE1 correlates with the Franconian Line which is the
main neotectonic boundary between the Bohemian Massif and the South German Block. O,
location of the core with joints shown in the figures listed.
to improve our understanding of brittle failure of rocks. The surface features on the
coring-induced joints are the traces of crack initiation, propagation and termination
(arrest lines of the fracture). In addition, the 3D shape of fractures is useful for the
determination of stress states during fracturing (Bankwitz & Bankwitz 1984).
Information on coring-induced fractures can also be gained by comparing them with
natural joints because the surface morphology is, in principle, the same on both types of
brittle-fracture surfaces. The most important components of surface morphology on
coring-induced fractures are diverging plumes with a centre of origin, hackle plumes, twist-
42 P. B A N K W I T Z & E. B A N K W I T Z
hackle fringes and arrest lines. Analysis of fracture surface morphology enables us to
determine: the area of origin; the direction and relative velocity of propagation of the
induced fractures; the relative timing of the development of induced fractures; and the
orientation of the principal stresses during fracture initiation and growth.
Surface morphology and the shape of coring-induced fractures in KTB cores formed
during Mode I fracture propagation, where the surfaces of the growing fracture were
opened in the direction of 0.3 (least compressional or greatest tensile stress). The direction
of 0.1 and 0"2 are apparent from the distribution of fracture surface structures and their
symmetry (Bankwitz & Bankwitz 1984, 1994). The correlation between stress field and
surface morphology arises from the symmetry between stress field and deformation field
during crack propagation, and the consequent symmetry of fractographic structures on
fracture surfaces. This relationship holds where cored intervals have abundant coring-
induced fractures.
Symmetry of fracture surface morphology

If the overall morphology of brittle fracture is observed then some regularities concerning
the shape of the fractures are found (Bankwitz & Bankwitz 1993). For example, the
growing fracture planes tilt or deviate from the initial fracture orientation, depending on
the local stress state. Deviation from the plane or origin generally culminates near to the
end of fracture propagation into ring-like zones, seen as either a tilt or as twist-hackle
fringes. Figure 2 shows schematically three different types of deviation from the initial
fracture plane (Fig. 2B-D) and the accompanying symmetry (zero, without deviation;
plus, upward deviation; minus, downward deviation), as it is found on both natural joints
and coring-induced fractures. It should be noted that the ring-like marginal zone can
deviate in the same sense from the initial plane of the fracture (Fig. 2B), or in an opposite
sense at two points (Fig. 2C), or the deviation sense can change at four points (Fig. 2D).
Depending on these deviations of the fringe itself, and depending on changes of hackle
steps of the twist-hackle fringe (sinistral or dextral steps) a higher or lower symmetry exists
and the symmetry axes are identified relative to 0.1 and 0.2. The latter corresponds to the
horizontal stresses S H r Sh in a core.
Basic types of symmetry (Fig. 2A-D) of the 3D shape of joints (first-order
characteristics) and of fractographic structures on joint surfaces (second-order character-
istics) occur as the response to development conditions. Besides the shape of joints, the
arrangement of twist hackles, including changes of their overlap, indicate the symmetry of
fractographic surface features. In many cases twist hackles are not restricted to the fringe
zone, or they can occur without a defined fringe zone but they always appear more dense
near the margin. In the types shown in Fig. 2A-D, fine lines indicate the position of a ring-
like marginal zone, sometimes without a well-defined boundary to the main plane. Twist
hackles would follow the direction of the radial fine lines. The scheme is derived from
natural joints, which always have circular or curved natural margins, independent of their
size, or parts of a curved margin. If not, then only a part of the whole joint is exposed.
In cases of plume axis on planar joints at least one symmetry axis forms. If a coarse-
grained core shows only roughly structured fracture surfaces, then the overall shape of the
ring-like fracture zones, described above, can be used to determine the symmetry axes
coinciding with the principal stress axes during origin and propagation of the joints.
=.
O'1.
o
%.%
~= .,-, ~ , . ~ ~ .
,,,
9, 0 , 0")
~
~ , ~~ 0~ i~
~.~..
,; b ~-
+
c'?'/
+b'- o . ~ ~ ~1[~ 0 II ..~
,,..--~. 0 ~-,~ ~
k,-~
..,-.
o
r
1=.,.~
~ ~
~ r
+ + +
~'~ o'= c~ o - '-~= . ~

U .~,~ ~~ ~
,
~ ~' ~.~
.=, ! vr= \k--
-%
+-~ ~. 0 .~ ~ 0
,
~-~'~ . ~ c~ 0 ~ ~
,~0 I:l ~
o~. ~ ~ o ~ ~
N , ~.~.~ ~.~ ~ ~ ~I
~'~ = = ~ ,~ ~,-~
~"~ , - ~ 0 ~ . ~ ~I ~.-~ . ~ "
.=~" \
N "'~ ~'~;-i ~ O
~'~ o. b ~ [ ~ ~ ~'~ o
~ 7N o~
l=:l ~ ~ .,- 0 ~ " ~ ,--~
~ ~-~ 0 ~-~ ~ ~ l::::l ~ l:::l
~.o~ ~ h~ ~_,.~
o c~
o o 0
o,-~ o o 0~-~
r
o
~D
~., . ~ ~ .,.~ ~ ~,
9 ~0 ~ ~ ~
I~I) r ,,~ ,,*~ @ r.~ rT I ~,~
44 P. BANKWlTZ & E. BANKWlTZ
(a)
Fig. 3. KTB cores (23.5 cm in diameter, depth 6435 m; amphibolite) with core disking spaced 2-5 cm
and forking structures, which are opening towards the high points (left side). (a) The wavy-like trace
of the core disking surfaces are visible. Low points near the right side; I, point of origin. Right lower
comer: two relatively younger core disking surfaces | terminated at an older fracture ~ t h a t is
oblique to the centre line. (b) A core adjacent to (a) which demonstrates the direction of core disking
propagation (arrows) by forking.
(b)
46 e. B A N K W I T Z & E. B A N K W I T Z
Conditions of K T B investigation
In the coarse-grained cores of the crystalline rocks of the KTB (Fig. l b) with a core
diameter of 10cm well-developed fracture planes are found, but without good
fractographic features, as are found in sedimentary rocks. Additionally, veins and veinlets
disrupt propagation and, hence, cause irregular fracture surfaces. In many cases the
coring-induced fracture surfaces contain only a rough pattern without detailed structures,
but their overall shape reflects the origin, propagation direction and the ring-like fracture
zones. At depths of 3000-8000 m, cores with a diameter of 23.5 cm were taken. Due to the
large diameter of these cores the fracture surface structures are better developed.
On cores at the KTB, orientation is marked by two colour lines rather than a scribe
knife; therefore, no scribe knife fractures occur in the KTB core.
Fracture types
There are several types of fractures in core from KTB, described below.
Drilling-induced centre-line fractures

These fractures (parallel and oblique to the core axis) are rarely developed. Their drilling-
induced origin is evident from fracture prolongation into the wall. These joints are not
terminated in the core. In some cases the centre-line fractures represent fracture
bifurcation, which implies a rapid fracture growth with early forking and subsequent
propagation in two different parallel planes 2 m m apart in the centre of the core. In a few
places, fractures seem to have started as petal lips, but their further propagation continued
as core disking surfaces, not as petal-centre-line fractures.
Coring-induced fractures
These fractures, initiated and terminated in the core, are dominant. Some of these
fractures are oriented perpendicular to the core axis and show effects of torsion. Cores of
23.5cm diameter from 3000 to 8000m depths are intensely fractured but are sometimes
without loss of cohesion (for example, Fig. 3). Therefore, not all fracture surfaces of such
cores could be studied directly, otherwise the disks would have to be separated. In some
cases, 0nly the traces of the fractures could be studied on the outside of the core. These
traces indicate the fracture type and reflect twist hackles by the forking trace (Fig. 3).
None the less, various disk fracture types are easy to distinguish because the borehole with
a depth of 9100 m gives the opportunity to observe enough separated disk surfaces.
Generally two main groups of disk fractures, which are dominant and occur nearly
perpendicular to the core axis, can be distinguished. One group includes the core disking
surfaces (CDS), characterized by saddle and trough structures, and is restricted to fine-
grained parts of the core. The second group contains disks with usual fractographic surface
structures (UFS) similar to those found on natural joints, without the particular saddle and
trough relief of the CDS. The second group is much more frequent than the first. Several
subgroups are used to divide the groups depending on the differing stress conditions.
The CDS on the KTB drill cores (Fig. 4) were investigated previously by R6ckel
& Natau (1990) and Wolter et al. (1990). These authors have observed additional
fractographic surface structures on CDS (Figs 3, 5 & 6). Sometimes there is a difference of
down-core
3 1
up-core SH j
4
MAIN DRILL HOLE ] 2, 4 low points

H019A4b / m 5778 ~r 1, 3 high points
SH "" 165"
Fig. 4. Core disking fracture with trough structure downcore (depth 7012 m, hornblende gneiss). The
axis between the low points is SH.
several centimetres between high and low points, effecting a wavy trace on the core (Figs
3a & 6). The CDS-type is equivalent to the four-part fractographic type of natural joints
(Fig. 2D). This type represents a special case of the more frequent UFS fractures, which
are equivalent to type C natural joints (Fig. 2C). The CDS (Figs 4 & 6) appear in fine-
grained and quasi-isotropic rock, whereas the UFS (Fig. 2A-C) appear in cores of all
kinds of rock.
Besides the CDS, all other coring-induced fractures perpendicular to the core axis, with
UFS, represent the second, and considerably more common, group (Fig. 2A-C). Their
rough fractographic pattern is arranged on a surface (Figs 7-9) without the trough
morphology of type D. They often initiate in the centre of the core and grow in regular
fashion with plumes, hackles and arrest lines, terminating in twist-hackle fringes (Fig. 2A
& B). Mostly, the fringe tilts from the opposite sides of the main plane in opposite
directions, as can be seen in Fig. 2C. The core in Figs 7 & 8 shows such fracture surface
structures as well, though roughly, developed. Barbed-type fractures similar to those
described by Kulander et al. (1990) are sometimes formed, with the point of origin always
located in the core, often near the centre and propagating downwards with the barb (Fig.
9).
Transitional forms between CDS and UFS types have been recognized. In the literature
opinion dominates that the CDS, with their saddle-trough structure, represent a special
case of disk development. The assumption that CDS were initiated at the high-point areas
of the core margin, and that the fracture propagated from these two points towards the
centre of the core, seem to be supported (Dyke 1989) by experimental results concerning
the strain distribution in the core, influenced by the drill equipment. From the physical
~t T
S H ... 165 ~ SH
Fig. 5. Core disking surface with overall triangular structure (depth 6668 m, amphibolite). The
typical trough structure, with two low (T) and two high (H) points, is superimposed by a central
ridge in the trough which broadens towards the propagation direction (top of the right-hand figure).
This shows three low points and is evidence for crack propagation in a preferred direction. It is a
transitional fracture type between core disking and common fractographic joints. Arrows, direction
of SH; + and - , positive and negative surface morphology.
I
a b D
V
S N ... 165 ~
sH
final
"-~~.... propagation
\ ~ ~ forking .--
C ~.
initial cracking
sym-axis 1 sym-axis 2
Fig. 6. Fractographic structures of core disking fractures from the KTB borehole (23.5cm in
diameter, depth 6435m, amphibolite). The fracture origin near the low points (L), small-scale
forking and the termination at the high points (H) is evident. Small arrows, direction of fracture
propagation; broad arrows, direction of S,; Sym, symmetry axis; i, initial point.
49
~
~..~
0
.~
n:D . , . ~
m~
~'~
~"~ ~
~~
,~ 0 ""
9~,~.~ ~
"~ w
~'~
9~ . ~
..~,~
~.-~
O
O
,.~ o
=
o
~o
o.~
~ ""~
0~.-~
"~"~ 0
u~
~.~
0
o,.~
~
e~0
.=8
",~ I
" ,.-.,
"6o
ao~ ~
r,o
.r .,-
point of view, if the CDS initiates at two points of the margin there should be some
evidence of overlap in the centre (T. Engelder, pers. comm.). Such CDS propagation could
not be confirmed on the KTB cores.
On CDS we have observed some hitherto unknown fractographic features and details of
the shape of the core disking fracture which give additional information on the disking
process. Such fractographic features, superimposed on the saddle structures of the CDS,
are often very smooth and not easy to detect. In nearly all cases our observations indicate
an initiation of the CDS in a low-point region and not at the marginal high-points. The
fracture grew from this single point through the entire core to the opposite margin. No
fractographic structures indicating a propagation from two marginal points towards the
centre could be observed.
From our observations it is apparent that the development of the CDS does not differ
from the development of the UFS joints. The fractographic features superimposed on core
disking saddle relief include fine hackle marks (rare) and small twist-hackle fringes
(common). The fringe zones (1-2 cm) become broader in the high-point region and may
disappear in the low-point area. Often core disking is characterized by a triangular surface
structure (Fig. 5), caused by a ridge within the trough of the downcore curved plane. In
this case three low points develop. Also of significance is a faint forking (Figs 3a, b & 6)
and larger forking (Fig. 10a & b), which is also known from common tension fractures.
The forking (bifurcation) marks the termination of CDS which are not developed through
the whole core (Fig. 10). Concerning their shape and surface structures, a range of
fractures was defined, differing from both CDS and UFS. These are both end members
with many transitional forms in between.
Fracture process (origin and propagation)

An origin of drilling-induced fractures at inhomogeneities (rock-internal boundaries or
crossing veins) is rarely observed. The disk fracture growth is sometimes influenced by
older shear planes parallel to the foliation, which dips c. 60 ~ (deviation of 30 ~ from the
core axis). More commonly, thicker veins parallel to the layering have interrupted fracture
growth. Sometimes the disks have started in the centre of the core and propagated towards
the margin, terminating in small fringe zones with rough twist hackles. Plumes are often
poorly developed. The propagation traces are marked by rough hackles.
Dominantly, disk origin is initiated near or at the margin, and the disk fractures grow
across the entire core, preferentially towards SH. This is counter-controlled by break-out
and core relaxation analysis. Consideration of fine fractographic features reveals that CDS
originate in the same way as the disks of the UFS. They initiate at the centre, or at one
point, of the core margin, spreading across the whole core. The larger sections of the core
provide good evidence for the relative timing of the recent joint succession (Fig. 11).
In some cases slickenlines (a few centimetres) were observed in the area of origin of the
brittle tensile fracture (Fig. 12). These suggest two stages of fracturing, separated by a brief
hesitation of fracture growth. Probably, during propagation of an initial small fracture,
sliding occurred on the initial fracture area which had developed before the fracture
hesitation. Alternatively, the disk fracture may have propagated across the core, followed
by a slight change in the stress field, which permitted slippage and slickenline development
on the smooth section of the fracture. We prefer the second assumption because no arrest
line can be observed.
(a)
Fig. 10. (a) Core disking surfaces (KTB core,

23.5cm in diameter, depth 6669m, metagab-
bro) strongly forking before termination; not
growing across the whole core. Top, zone Fig. 11. Relative timing of coring-induced
without traces of fractures; broken line, fractures in KTB cores (23.5cm in diameter,
boundary of the fractured zone. (b) Schematic depth 6434m, amphibolite). The oldest frac-
drawing from the core in (a). Broken lines, tures are parallel O and oblique (~) to the core
zone with veins parallel to the foliation. axis; the youngest fractures are the horizontal
Arrows indicate the direction of fracture disk joints (~). Arrow, downcore direction.
54 P. BANKWITZ & E, BANKWITZ
SYMMETRY AXIS
.S J / =SHmox-150 ~
- ' ~ ~ " ~ - " \ ist hackle steps
?;1% / , /:/
SHmi n SHmox ~ forking

SYM-AXIS 1
/
d ~ =SHmox=15@o
C ///"C////
slickenlines
Fig. 12. Two examples of coring-induced fractures with slickenlines in the area of origin, (a) Origin
near the margin and preferred direction of growth through the whole core (depth 6668m,
amphibolite). (b) Origin near the centre (~), followed by smooth areas (~) and forking (bifurcation) at
opposite sites in S~ direction (depth 6359 m, amphibolite). (e) & (d) Interpretation of the core in (a)
and in (b), schematic diagrams at the left: R, fringe; R1, inner fringe; R2, o u t e r fringe (both cores
23.5 cm in diameter); C, centre of primary initiation.
Recent stress field

All joints reflect the stress state during their initiation and growth, Coring-induced
fractures reflect the in situ stress field and the regularity of their fractographic geometry
along the whole KTB cores (over a height of several kilometres), suggesting a relationship
F R A C T O G R A P H Y KTB CORE 55
Orientation of drillhole and drillcore instabilities
Sh SH Sh SH
0 45 19o la5 118o 225 127o 31s 136o
3000 lllililtlitilllillililiittliittillillllllllllllliittllitillliililililli
V V
3050
V V
V V
3100
VV
3150
3200 V *
W vw w wv
3250 w w
3300
3350
v
E
c"
3400
V
a - t V
3450 V v V V
V V
3500
- V
3550 _ V V V V
-v e~.. vWvv ~ ~ vvvv

3600 ~ . ~ V VV
3650 -S V V, Y w v,
-v ~ w'vV ---eJlr .... i
3700 - o'1,-" ~ L'r_- vv I
g"" ,r
IIo
380O
Fig. 13. Core disking high points 9 and centre4ine fractures ~' in cores of the pilot hole of the KTB,
with SH at c. 1 6 5 ~ (Th. R6ckel, pers. comm.). X, dip of the centre-line fractures;., dip of the non-
vertical boring induced fractures; V, trend of the non-vertical boring-induced fractures.
to the in situ stress orientation determined by other techniques, e.g. borehole break-outs
(Brudy et al. 1993).
The above mentioned method to infer the principal stress axis on the basis of fracto-
graphic symmetry led to the determination of the recent SH. Fractures suitable for such
56 P. BANKWlTZ & E. BANKWITZ
I 1 l I I Symmetryaxis
M~nchberg
dBerneck (~H
"~ ~ s0*00'--
origin
'
z~ dorf
"45;-- 1
? 1p 2,o km b
11'30' 11~ ' 12~ ' 12115' Symmetrya x i s

I I (IH 165~
Fig. 14. (a) Maximum horizontal stress axes (small arrows) determined from disking surface
symmetry of KTB cores in relation to the Franconian Line, which is the platform cover boundary
between the Bohemian Massif and South German Block. Large arrows, supposed regional stress
distribution. (b) The dominant type of fracture surface structures with the origin near the margin and
a directed propagation through the core, mostly from south to north (depth 5778 m). The dominant
NW-SE compression (b) changes below a depth of 6000 m (e.g. depth 6669 m), sometimes to a 30~
NE-SW compressional direction (e). O, Point of origin; arrows, direction of SH correlating with the
symmetry axis.
investigations suggest that SH is oriented at c. 165 ~ This direction coincides with the
results from the core disking analysis of Wolter et al. (1990) and our break-out research
(Fig. 13). At a depth below 6000 m a partial change of the SH direction to 30 ~ (NE-SW) is
evident from our data. Such a change is possibly of local nature (Fig. 14). At greater
depths SH seems to again have a persistent N W - S E direction. The recent main horizontal
stress orientation is oblique to the fault boundary of the Bohemian Massif (130-145~ in
some cases the stress orientation rotates to a perpendicular direction (30~ The dominant
N W - S E compression coincides with the data in the World Stress Map.
Agreement between stress directions derived from our method of analysing surface
feature symmetry and other techniques is very close, so the former is suitable for both
checking the reliability of other techniques and for use in its own with high level of
confidence. Because it is applicable for all types of rocks, the method could become a
routine method in core analysis for finding stress orientation in oriented cores.
Main results
Most of the studied fractures from the core barrels are coring-induced fractures, restricted
to the core itself. They are initiated by the coupling of the stress states related to in situ
stresses within the undisturbed rock, as well as a stress imbalance caused by removal of
overburden pressure (Kulander et al. 1990). CDS, as well as UFS, are part of this group.
1. The direction of the maximum horizontal stress axis was determined by means of
UFS features known from natural joints (Bahat 1991). In most cases the results were
the same as from core disking data, indicating SH with a N W - S E direction (c. 165~
The Six orientation reflects compression oblique to the main boundary.
2. A sequence of joints is commonly developed (Fig. 11) and includes three groups (a)
joints quasi-parallel to the drill core axis, as the oldest fractures; (b)joints oblique to
the core axis (35 ~ deviation) formed next. Both (a) and (b) formed in response to
drilling-induced stresses coupled with in situ stresses within the undisturbed rock. (c)
Joints perpendicular to the core axis, these are the latest group formed primarily in
response to unloading stresses coupled with in situ stresses, most likely assisted by
torsional stresses (and vibrations accompanying drilling and coring). Frequently,
only this last group (c) is present in the core.
3. According to their frequency, CDS with a saddle and trough structure (two high
points and two low points) is the special case of UFS. Other UFS joints (Fig. 2A-C)
represent the most common cases. These UFS joints are widespread and occur in
different shapes (classes) in all rocks throughout the World, including highly
anisotropic rocks.
4. Joint categories can be differentiated. Using data from fracture physics it was
possible to differentiate between categories of fracture planes in KTB drill cores as:
tension fractures (Six c. Sh), relaxation fractures (SH > Sh), compression-tensional
fractures (Six > > Sh), compressional fractures (Srt > > > Sh, relatively large part of
differential stress), torsion fractures. Joints approximately parallel to the core axis
are rare and will not be discussed. UFS joints occur mainly in two classes: (1) as pure
tensional fractures, initiating at the centre of the joints, reflecting a low amount of
energy, proved by the equal dip direction of the twist-hackle steps on the entire
marginal fringe; (2) as fractures formed under higher energy conditions, indicating
differential stress. They originate near one border of the core and continue across the
whole core from one side to the other, following the direction of the main stress. A
N W directed polarity of the main horizontal compression Six seems probable.
Conclusions
1. CDS, and other joints with symmetric UFS structures, can occur together or
separately. The high and low points of the CDS represent symmetry axes. CDS may
also show fractographic surface elements, e.g. forking structures towards the margin,
which reflect symmetry and reveal the mechanics of their growth. In many cases the
CDS is the lower plane between the forking fractures, which are superimposed on the
trough morphology (wavy trace on the core).
2. In many cases core disking fractures are characterized by a triangular structure with
the point of origin at or near the margin of the fracture. The typical trough structure,
with two low and two high points, is superimposed by a central ridge in the trough
which broadens in the propagation direction. The shape of the CDS with three low
points reveals crack propagation in a preferred direction. The high-point areas of this
CDS type appear like fringes.
3. In a few cases slickenlines occur on small central parts of coring-induced KTB disk
joints. The lines are positioned at the area of origin of these tensional joints with the
central area being mostly tongue-shaped. It appears t h a t a small initial disk fracture
experienced brief hesitation in p r o p a g a t i o n , followed by continued crack g r o w t h
across the core. A slight change in the local compressional stress field caused m i n o r
sliding along the initial fracture segment.
This investigation was supported by the Deutsche Forschungsgemeinschaft (DFG) since 1992. It is
part of the programme 'Stresses in Rocks' with contributions from several teams, above all in the
KTB-field laboratory, in several universities, and in the GeoForschungsZentrum Potsdam. We
thank the D F G for financial support. We thank, for sharing their knowledge on various topics, H.-
G. Dietrich and Th. R6ckel, KTB-field laboratory. We are profoundly indebted to Terry Engelder
and Stuart L. Dean for stimulating discussions. We thank Mohammed Ameen for helpful comments
and correcting this manuscript. The paper has much benefited from the reviews.
References
BAHAT, D. 1991. Tectonofractography. Springer, Berlin.
BANKWlTZ, P. 1966. Uber Kliifte. II. Geologie, Berlin, 13, 896-941.
& BANKWITZ, E. 1984. Die Symmetrie von Kluftoberflfichen und ihre Nutzung ftir eine
Palfiospannungsanalyse. Zeitschrift fiir Geologische Wissenschaften, 12, 305-334.
- - &- - 1993. Stress analysis on KTB drill cores derived from fractographic features. KTB
Report 93-2, 213-218.
- - & - - 1994. Event related jointing in rocks on Bornholm (Denmark). Zeitschrift flit
Geologische Wissenschaften, 22, 97-114.
BRUDY, M., FUCHS, K. & ZOBACK, M. D. 1993. Stress orientation profile to 6 km depth in the KTB
main borehole. KTB Report 93-2, 195-197.
DYKE, C. G. 1989. Core disking: Its potential as an indicator of principal in situ stress directions. In:
MAURY, V. & FOURMAINTI~UX, O. (eds) Rock at Great Depth. Balkema, Rotterdam, 1057-
1064.
HIRSCHMANN, G. 1993. Seismischer Reflektor SE1 erbohrt - und was folgt darunter? 6.
Schwerpunktkolloquium 'KTB', Abstracts, 1.-2.4.1993, GieBen, 63-64.
HODGSON, R. A. 1961. Classification of structures on joint surfaces. American Journal of Science,
259, 493-502.
KULANDER, B. R., DEAN, S. L. & WARD, JR., B. J. 1990. Fractured core analysis interpretation,
logging, and use of natural and induced fractures in core. AAPG Methods in Exploration Series,
8.
ROBERTS, J. C. 1961. Feather-fracture, and the mechanics of rock-jointing. American Journal of
Science, 259, 481492.
R6CKEL, TH. & NATAU, O. 1990. Results from rock mechanical index tests of the pilot hole 'KTB
Oberpfalz VB'. KTB Report 90-8, H 1-H 13.
VERNIK, L., ZOBACK, M. D. & BRUDY, M. 1992. Methodology and application of the wellbore
breakout analysis in estimating the maximum horizontal stress magnitude in the KTB pilot hole.
Scientific Drilling, 3, 161-169.
WEBER, K. (Coord.) 1985. Excursion Guide Oberpfalz. IUCL Coordinating Committee "Continental
Drilling".
WOLTER, K. E., ROCKEL, TH., BIJCKER, CH., DIETRICH, H. G. & BERCKHEMER, H. 1990. Core
disking in KTB drill cores and the determination of the in situ stress orientation. KTB Report
90-8, G1-G13.
WOODWORTH, J. B. 1896. On the fracture systems of joints, with remarks on certain great fractures.
ZOBACK, M. L. 1992. First- and second-order patterns of stress in the lithosphere: the world stress
map project. Journal of Geophysical Research, 97, 11 703-11 728.
Observations on fractography with laboratory

experiments for geologists
B. R . K U L A N D E R 1 & STUART L. D E A N 2
1Department of Geological Sciences, Wright State University, Dayton,
OH 45435, USA
2Department of Geology, University of Toledo, Toledo, OH 43606, USA
Abstract: Brittle fracture growth proceeds through unique stages, each marked by
distinct fractographic features that can only develop during Mode I loading at the
propagating crack tip. Fractographic features in any substance can be interpreted in
terms of the location of the failure origin, as well as changes in propagation velocities,
stress directions and stress magnitudes at the crack tip during failure. Joints in granular
pervious rock, however, do not contain fractographic features, commonly formed in
glass, that develop at unstable propagation rates under the influence of appreciable
amounts of stored strain energy. Yet, features that develop at lower stable rates of
propagation are present on fractures in glass and rock implying that the absence of
certain fractographic features provides useful information.
Simple laboratory experiments, primarily on glass, are discussed to provide geologists
an opportunity to recognize and qualitatively interpret fractographic features diagnostic
of brittle failure. The exercises demonstrate the changing propagation dynamics that
control the morphological evolution of artificially-induced fractures and natural joints
in rock.
The term fractography was first used in its modern sense by Zapffe & Clogg (1944) who
described it as a new tool for studying fractures in metals. The current field of
fractography investigates fractographic features (fracture surface structures) that comprise
the topography of a fracture surface. The goal of these investigations is to interpret the
dynamic history of a fracture from its origin to final termination.
This paper compares fractographic features developed during simple laboratory
experiments, utilizing primarily glass, to surface structures that are common on joints.
The comparison provides a basis for the discussion section that focuses on selected aspects
governing the dynamics of joint propagation. All experimental observations are
qualitative in the sense that complex equipment and measurements are not necessary.
Nevertheless, the experiments provide geologists an opportunity to gain experience in
recognizing and interpreting fractographic features in all materials that undergo brittle
failure.
Primary fractographic features in any material can be classified as either transient or
tendential (Frechette 1972). Transient structures are those that comprise the surface
morphology of a fracture and commonly act together to form a complete plumose
structure. Plumes with well-developed plume axes are particularly evident in uniform
layers of contrasting lithology. Plumes in glass, however, can differ morphologically from
those in rock.
60 B. R. K U L A N D E R & S. L. DEAN
(a)
(b)
61
(c)
(d)
Fig. 1. Electron micrographs of a fracture surface on a glass rod subjected to three-point bending. (a)
Fracture origin surrounded by mirror surface disrupted locally by two twist hackle faces. Lower face
curves abruptly into upper face to form twist-hackle step. Twist hackle disappear as irregular stress
distribution about the origin becomes uniform and mirror surfaces become a single plane. ~) Same
fracture origin surrounded by mirror, then mist, region containing fine to coarse velocity hackle,
then recurring mirror surface broken by pronounced twist-hackle faces and steps. (c) Mirror region.
(d) Velocity hackle.
62 B . R . K U L A N D E R & S. L. D E A N
Fig. 2. Wallner lines on mirror surface in glass rod.
Tendential features are shown by the trace of a fracture plane on a free surface and may
reflect specific transient structures, as well as gradual or abrupt changes in the propagation
direction. Tendentially viewed features, including abutting patterns, may also result from
interaction of multiple cracks (Kulander et al. 1979; Pollard & Aydin 1988). This paper
focuses primarily on transient features. A brief description of terms commonly used in
ensuing sections follows.
A joint is a naturally occurring brittle fracture in rock that originates and propagates
due to Mode I loading at the crack tip (Pollard & Aydin 1988). Plumose structures
(Hodgson 1961) containing twist and inclusion hackle, and rib marks occur on joint faces
(Kulander et al. 1979, 1990). Plumes on glass fractures can also contain Wallner lines and
velocity hackle. An origin flaw that concentrates applied stress commonly serves as the
location from which a single fracture develops. Origins have also been termed focal points
or pits by Kies et al. (1950). An isolated inclusion that served as an origin in glass is shown
in Fig. 1A.
The mirror region is a flat featureless surface that surrounds the origin (Poncelet 1965;
Frechette 1972). Extent of the mirror region ranges from microscopic to the entire fracture
surface. A complete mirror developed in a glass rod is shown in Fig. lB. The featureless
nature of the mirror is suggested by Fig. 1C.
Development of velocity hackle terminates the mirror and gives the fracture surface a
foggy appearance that typifies the mist region (Poncelet 1958). Preston (1939) referred to
the mist region as a mat surface. The mirror to mist transition and the chaotic nature of
velocity hackle is shown in Fig. 1B & D.
Wallner lines (Wallner 1939; Quackenbush & Frechette 1978), or ripple marks (Poncelet
OBSERVATIONS ON FRACTOGRAPHY FOR GEOLOGISTS 63
2 CM
, /I.
I
. .
! '~ ~} " ~.~
Fig. 3. Sketch of joint in siltstone. All parts of the joint developed during the same fracture event and
propagation directions can be traced back to a single origin. Overall propagation direction from left
to right is shown by plume geometry about plume axis 'A'. A twist-hackle band is marked 'B', twist-
hackle faces and steps in twist-hackle fringe marked 'C' and 'D', respectively. One rib mark
surrounding the origin, and two constructed fracture front lines that are geometrically regular and
parabolic, decorate the main joint face. Arrows show local propagation directions that can be
determined at this scale.
1958), are common features in glass. They are generally subtle rounded features of low
relief that are convex in the overall direction of fracture propagation (Fig. 2).
Fracture forking occurs when a fracture, accelerated by increasing tip stresses in the mist
region suddenly bifurcates into two or more divergent planes. The point where forking
occurs is called the radient by Preston (1926).
Twist hackle (Frechette 1972) has been discussed under the general term hackle or
hackly fractures by Gash (1971), Murgatroyd (1942) and Preston (1926). Identical features
have been termed striations by Poncelet (1958, 1965) and Preston (1939), and river lines by
Pugh (1967). Geologists have discussed twist-hackle morphology as feather fracture
border planes, cross fractures (Woodworth 1896), plumose structure, F-joints and C-joints
(Hodgson 1961), fringe faces and steps (Bankwitz 1965), and dilatent echelon cracks
(Pollard et al. 1982). Twist hackle develops when a fracture breaks into en echelon twist-
hackle faces and steps that are generally more closely spaced than adjacent main fracture
planes (Fig. 3).
Inclusion hackle, a form of twist hackle, can be a common constituent of hackle plumes
in granular pervious rock (Kulander & Dean 1985; Kulander et al. 1990). It forms when a
fracture plane is warped in passing around, or through, an inclusion and consequently
does not maintain a single planar surface immediately on the inclusion's other side
(Kulander et al. 1990, fig. A8).
Rib marks (Preston 1926; Murgatroyd 1942) have been referred to as arrest lines
64 B. R . K U L A N D E R & S. L . D E A N
"r; gi~)~,._!
9' .~.-.'r ~:..
' L ~ ~
"w"'~':'-~":" : ' : 1 / , ',:.'~,'22
~ ; ~ . ~ 9 .:-:.\
Fig. 4. Common cross-sectional morphologies of rib marks. (A) In form of cuspate wave. (13)and (C)
In form of line separating tilted panels. (D) Rounded form that may not mark points of arrest.
(Frechette 1972; Kulander et al. 1979, 1990), annular structures (Bankwitz 1965, 1966) and
hesitation lines (Barton 1983). Rib marks represent a velocity discontinuity and commonly
develop when propagation hesitates before or after abrupt deviation from the mean
fracture plane (Fig. 4A-C). Rounded rib marks (Fig. 4D) may not mark points of
hesitation.
Fracture front lines that mimic rib marks can be constructed on fracture faces
adequately marked by accompanying fractographic features (Fig. 3).
Subsequent surface structures, caused by pressure solution or shear, include stylolites
possessing cone and socket structure, undulating pressure solution seams, and slickenlines
that may be marked by fibrous mineralization. These can be superposed on primary
fractographic features formed initially by brittle failure. This sequence of events is
geologically common in orogenic regions of lateral deformation. When these subsequent
features are developed on a joint surface, they should be included in a complete
interpretation of past dynamic events that led to final surface form. In this sense they are
viewed as fractographic features.
Simple laboratory experiments in fractography - experimental methods

The authors relied primarily upon the destruction of glass objects to provide surface
structures that were compared to those found in rocks. Glass objects included large
microscope slides, solid glass rods and capillary rods. One or both sides of several slides
,~cribe mark--~ ,
tapedfaceJ~ ~ A
slide face under

greatest tension
Wallner
lines
Fig. 5. (A) Glass lath prepared for cross-bending. (B) Broken lath folded back using tape as hinge.
were scoured, using a circular motion, with fine grit sandpaper and some slides were left
unscoured. A number of the solid and capillary rods were also scoured. Transparent tape
was then placed on one side of each slide, except those destined for thermal tests, to
prevent scattering of broken pieces (Fig. 5A). The tape also acted as a hinge permitting
exposure of fracture surfaces for examination (Fig. 5B). For slides broken by bending,
origin flaws of varying size were scribed on some untaped slide surfaces. All slides were
then broken by applying bending, torsion, point load and thermal stresses. Torsion and
point load stresses were applied as shown in Fig. 6A & B. Due to lack of precise controls,
several tests for a given experiment were necessary to produce results that most clearly
show contrasting fractographic features related to scribe flaw size and vigour of scouring.
The solid glass and capillary rods were induced to fail by stresses related to bending and
torsion. Torsion stresses were applied to glass rods with pliers after wrapping their ends
with layers of tape. Blackboard chalk is also an instructive medium. For example, failure
was induced by hand applied torsion. Torsion was applied to chalk sticks while simul-
taneously adding varying amounts of compression and tension parallel to the cylindrical
axis of the chalk.
Failure was also caused in slides by stresses related to thermal expansion and
contraction. For thermal fractures, a Bunsen burner was placed in a large open box to
facilitate retrieval of broken fragments. For expansion fractures, scoured and unscoured
slides, preferably square ones, were held over the flame so that the reducing flame tip was
in contact with the slide centre. After rupture, that can be explosive, the slides were
reassembled. For contraction fractures, a slide corner was heated until a yellow sodium
flame was observed. The slide was then removed from the flame. Failure occurred within
10 s. If failure did not occur the slide edge was scoured and reheated. Next, the long edge
of a slide was heated in similar fashion and cooled.
66 B. R. K U L A N D E R & S. L. D E A N
~Stress
~"'~ Tape ~-BB ~
B
Fig. 6. (A) Glass lath prepared for torsion test. (B) Glass lath prepared for point load test.
Fracture surfaces were exposed for examination on slides by folding along the tape
hinge. Slides and rods were then anchored in modelling clay for microscopic examination.
Oblique illumination was adjusted to highlight subtle transient features observed on
scoured and unscoured specimens, and those with scribed flaws were sketched. Local
propagation directions were shown with arrows. Sketches of tendential features for
fractures in glass slides, rods and chalk also proved instructive.
Fracture development and development of fractographic features in any material
selected for experimentation can be caused by diverse loading configurations applied
singly or in mixed mode (Barton 1983, chapter 7). Slow (stable) propagation rates in brittle
substances possessing lower tensile strength can be controlled by varying the intensity of
applied stress, in such materials, rib marks commonly develop at points where fracture
progress stopped. In addition, configuration of loading stress during slow fracture can be
altered during tests to observe fracture reaction.
Observed experimental results
Fracture surfaces in glass slides and solid glass rods

All methods of stress application led to fractures that were each initiated at a discrete
origin flaw. Furthermore, unscoured glass slides and rods with no scribed origin flaws,
excluding fractures developed during thermal contraction, commonly show a progression
of distinct fractographic features (Fig. 1B). These features, in order of development,
include the mirror region containing Wallner lines that are most pronounced on slides with
lightly scoured surfaces. The mirror is terminated by the mist region marked by
development of velocity hackle that may become increasingly rough and chaotic in the
direction of propagation. The mist region may revert to a mirror region that is commonly
disrupted by smooth twist-hackle faces and steps, as shown in Fig. 1B. The mist region is
also commonly terminated by forking of the initial fracture into diverging planar
velocity-
related no scribe flaw
forking
IAI
(-~ torsion both sides scoured
i
j Cx\\~ Y
~ top origin and overall
propagation direction
unscoured
bottom origin and overall

propagation direction
B
4- 4~-4rr~-.~_ _' , ~ - - - ~ ~ , ~ . ] [unscouredA)jLI

Slide ~ "(~
r~ Scoured I I ~Th_'~.._.
i~176 IC
side
Higher Heat
D at Fracture
Fig. 7. (A) Glass laths broken by bending that show typical tendential view of fracture patterns. Slide
bent concave towards the viewer and fractures led on reverse side. (B) Tendential fracture pattern in
laths that failed under torsion. (C) Tendential fracture pattern in laths that failed under point load.
View of slide face upon which pressure was applied. (D) Tendential fracture patterns on plates that
failed explosively.
segments. After the inception of forking velocity hackle disappears, or its development is
diminished.
Fractures originating from scribed origin flaws in rods and slides required less applied
stress to cause failure. This difference in failure stress was noticeable, even though stress
magnitudes were not precisely measured in these qualitative tests. In every case, fractures
from scribed flaws of sufficient dimension are all mirror surfaces. Furthermore, velocity
hackle and forking did not develop.
Fracture forking in slides

Fracture forking was produced by all stress applications, except thermal contraction, and
is most commonly developed in unscoured slides with no scribed origin flaws. In every
case, the point of bifurcation is immediately preceded by development of velocity hackle.
After inception of forking, velocity hackle disappears or its development is commonly
diminished. In this case, individual bifurcating segments may fork again. The initial angle
between forking segments is c. 45 and 15~ respectively, on unscribed slides broken by
bending and torsion stresses (Fig. 7A & B). On slides broken by point load and thermal
stresses during expansion this forking angle develops over a short distance and is e. 180~
(Fig. 7C & D). These angles, predicted by a relationship between maximum and minimum
failure stresses (Preston 1935), can be precise in controlled tests but are more variable in
these qualitative experiments.
Fracture hooking in slides and rods

Fracture hooking is most pronounced on slides and rods containing no scribed or scoured
flaws. The hook develops towards the face that is under the least tension at failure
and initially attempts to turn away from this surface (Fig. 5B). A twist-hackle fringe is
commonly developed within the zone of hooking. There is an inverse relationship between
stress to failure, related to origin size and distance from origin to initiation of the hook.
This observation is highlighted qualitatively by the fact that hooking initiated closest to
the point of fracture inception on unscoured rods with no scribed origin flaw (Fig. 8).
Pronounced hooking of this nature requires rapidly accelerating propagation.
Hooking of a different nature commonly develops when a fracture approaches a free
surface at an oblique angle. The fracture tip will turn towards the free surface in an
attempt to meet it orthogonally. Fracture hooking of this nature is particularly evident in
the tendential view of themal contraction fractures that propagate slowly (Fig. 9).
Torsion fractures in slides and rods

Unscoured slides, subjected solely to torsion, generally contain fractures that can be
grouped into two sets orthogonal to each other (Fig. 7B). This angular relationship can be
varied by a compression or tension added parallel to the slide faces during torsion. Slides
scoured on one face commonly contain only one fracture set that originates on the scoured
side. Slides with both faces lightly scoured may contain two fracture sets; however,
fracture frequency is reduced. Chronology of development for fracture sets is shown by an
abutting relationship that can be reversed on opposite ends of the slide (Fig. 7B).
Furthermore, fractures of one set are initiated and lead on one slide face, whereas fractures
within the other set originate and lead on the opposite side. This reversal occurs because
large
flaw
Fig. 8. Tendential hooking in glass rods. Distance from origin to inception of hooking varies with
origin size.
t__
E:
0
r
"(3 v v
(1)
(..-
t..-
(1)
13)
"13
(1)
"13
E:
t,- Thermal
Contraction Fractures
Fig. 9. Tendential fracture patterns on laths that failed during cooling.
torsion fractures remain perpendicular and parallel to resultant tensile and compressive
stress directions that are reversed on opposite ends of the slide (Fig. 7A).
Some fractures within each set induced by torsion appear to mutually transect each
other in a pattern commonly attributed to development of conjugate sets of shear joints.
However, close examination reveals that one fracture is throughgoing (Fig. 10). The
others, that can be seemingly continuous across the first-formed fracture, actually
originate at adjacent origins on opposite sides of the throughgoing fracture. These origins
are commonly twist-hackle faces and steps within a twist-hackle fringe.
Torsion stresses applied to a glass rod produce a fracture surface seen tendentially as a
70 B.R. KULANDER & S. L. DEAN
m S~
> " " x J--

slideface origin "
9P '
t
,l~j
/) toward ~,, t ,,
(h) viewer '77
~ "-"'"~,
Wallner j
lines
Velocity origin
hackle ~ slidefacesunder
greatesttension
Fig. 10. Transient morphology of abutting torsion fractures in glass laths.
helical trace. The trace ideally makes an angle of 45 ~ with the rod axis. As with slides, this
geometry is attributed to the reversed sense of simple shear induced on diametrically
opposite faces of the rod. Blackboard chalk, easily twisted by hand, demonstrates that this
angle is dependent upon the amount of compression or tension applied parallel to the
chalk axis during torsion. Axial tension and compression increase and decrease the angle,
respectively. Similar fractures commonly develop in rock cores when the core jams in the
barrel.
Point load fractures in glass slides

Fractures produced in slides by point load pressure also commonly form at right angles to
each other (Fig. 7C). Again, fractures that seemingly intersect are actually three different
fractures. In this case, however, each fracture is parallel to a slide boundary. Subsequent
fractures induced in previously broken segments also parallel previous fractures and slide
boundaries. In any case, unscoured slides and those scoured on the loaded face require
greater point load pressure to cause failure.
Circumferential
.__ tension
A 9r a d i a !
compression
Annealed"
Edge Corner
Fig. 11. (h) Expansion stresses in glass plate caused by heat application at centre. (B) Contraction
stresses in a glass lath subjected to corner and edge heating, then cooling.
Thermal fractures in glass slides

Thermal expansion fractures develop at explosive rates and produce abundant velocity
hackle and forking segments (Fig. 7D). Fracture is initiated immediately away from the
heated area at a point of maximum circumferential tension. Circumferential tensile and
radial compressive stresses causing fracture are shown in Fig. 11A. In contrast, thermal
contraction fractures can grow at visibly slow rates and follow a sinuous propagation path
(Fig. 9). Wallner lines are not developed. Failure is controlled by rapidly changing stress
fields of low intensity caused when the heated and annealed edge or corner attempts to
become circular upon cooling (Fig. 11B). Consequently, contraction fractures commonly
arrest and then reinitiate propagation. Rib marks are commonly developed at these
hesitation points. Thermal contraction fractures do not reach a propagation velocity
necessary to develop velocity hackle and are consequently all mirror surfaces.
Bending fractures in capillariy rods

Fracture surfaces transecting glass capillary rods contain two unique fractographic
features that are related to the capillary bore. Both features originate at the inclusion on
the side opposite that of initial fracture approach. Neither feature is developed on
fractures in solid rods (Fig. 12A & B). One such feature is a set of paired Wallner lines,
h o o k ~
~~ve!ocit](. hack!e
region
ook
mist with
velocity
hackle
...irror with
A origin Wallnerlines
Fig. 12. Transient features commonly develop on fracture surfaces in (A) glass rods and (B) glass
capillary rods during three-point bending.
commonly called gull wings because of their combined shape. Gull wings are subtle and
must be viewed with the microscope under proper oblique lighting. The second feature is
an inclusion hackle tail that is an isolated twist-hackle step.
Discussion
Each stage of fracture development in any brittle medium is marked by development of
unique fractographic features. The features are produced when the fracture or joint
deviates locally from its mean plane of propagation. Deviations are attributed to tensional
crack-tip stresses that are affected by the amount of strain energy locally released, the
passage of sonic waves, changes in propagation rate, changes in far-field stresses, and
material discontinuities related to varying elastic moduli and textures. Even though these
far-field stresses may be attributed to Mode II or III shear loading, resultant crack-tip
stresses in a brittle material are tensile.
Glass fractures, however, can develop velocity-related fractographic features that are
absent from joints in granular pervious rock. Conversely, fractographic structures,
common on joints, also form in glass. Appreciation of all features is vital for a wider
understanding of factors that affect propagation dynamics. For example, realizing why
certain fractographic features are absent in rock can benefit analysis.
The sequential development of fracturing stages varies with mechanical and textural
properties of the material and stress concentrated at the origin flaw that is required for
failure. For example, glass can contain few and small flaws, high tensile strength, and
a capability to store large amounts of elastic strain energy. In this case, fracture
development can proceed through stages containing fractographic features that are
dependent upon velocities that accelerate to higher levels than in rock. Such features
include the mirror region, the mist stage containing velocity hackle, velocity-related
forking and Wallner lines. Wallner lines are included in this category because propagation
velocity must be elevated to the level where sonic waves do not immediately overtake and
pass the fracture front.
Joints in permeable clastic rocks propagate at lower rates then fractures in glass and,
therefore, do not display fractographic stages related to propagation at high velocity.
Fig. 13. Complex hackle plume geometry in siltstone marked by irregular and bifurcating rib marks
and large irregular twist-hackle faces with distinct plume components. Development of subsequent
twist-hackle steps has altered original extent of twist-hackle faces. Two origins shown at 'A', third
origin off lower right. Irregular and bifurcating rib marks shown at 'B'. Arrows show local directions
of joint advance determined from plume components at megascopic scale.
Nevertheless, many joints do progress through distinct stages of development. For

example, a relatively smooth portion of a joint surface may contain only low relief twist-
hackle, inclusion hackle, rib marks and minor undulations attributed to vagaries of the
stress field. However, this relatively smooth area may break abruptly on a single joint into
twist-hackle and rib marks, both consisting of adjoining panels, that stand in high relief
(Bankwitz 1965).
Origins
Failure origins in glass and rock occur at material flaws that allow local stress
concentrations sufficient to cause failure (Fig. 1A). Flaws in glass, unless artificially
induced, are generally small and widely dispersed. Whereas those in granular pervious
rock are commonly larger and more numerous, including pore spaces, large grains, fossils,
previously developed microcracks and joints, and irregularities in the bedding surface
(Kulander et al. 1990, fig. A2). Externally applied stresses that cause brittle failure in both
glass and rock generate local failure stresses that are magnified by such flaws.
Glass fractures and individual joints in rock, developed during a single jointing event,
progress from a single origin. Closely spaced multiple origins can, however, occur in glass
when fracturing is violently induced. Hot glass quenched in water is an example. Here,
thermal shock produces widespread stresses well above the minimum necessary to cause
failure (Frechette 1972). In this case, velocity-induced hackle and forking, related to
unstable propagation at rapid rates, are common. An exposed surface caused by jointing
in rock may also contain multiple origins, with each marking a unique jointing event. In
this case, the individual joint surfaces that coalesce can produce an irregular pattern (Fig.
13). However, fractographic features developed at high unstable velocities are absent.
Mirror-m&t regions and velocity hackle

The featureless mirror region (Fig. 1B & C) surrounding the origin in glass fractures
develops perpendicular to the principal tensile stress at the crack tip (Preston 1926;
Frechette 1972, 1984). Poncelet (1958) states that this anatomy develops because fracture
acceleration, caused by persistent stress at the crack tip, has not reached critical levels
necessary to break a large number of bonds oblique to the mean fracture plane.
Fractures in glass, initiated by larger stresses, have smaller mirror radii. Here, greater
stress to failure and elevated stored strain energy accelerate the fracture tip. Consequently,
the mirror region is terminated by development of velocity hackle that marks the inception
of the mist region (Fig. 1B & D). At initial development, the roughened surface of the mist
region appears frosted due to development of minute velocity hackle during uncontrolled
(unstable) fracture propagation. A quantitative relationship between mist and mirror is
described by Shand (1959). He determined from tests on glass rods, containing no residual
stress that failure stress varies inversely with mirror radius (average distance from origin to
inception of mist), if this radius is small when compared to rod diameter.
Development of the mist region is generally attributed to sonic waves generated along
the fracture front by parting bonds that may be oblique to the main fracture plane. At
some critical propagation velocity, these waves begin to react with one another and
with the stress field at the crack tip to cause local deviations from the main fracture plane.
Individual velocity hackle so produced are irregular and chaotic in appearance.
Furthermore, velocity hackle size increases with increasing angles of deviation from the
main fracture plane at accelerating fracture velocities (Poncelet 1958).
Shand (1959) and Poncelet (1965) state that, in glass under given test parameters, the
propagation velocity critical to development of velocity hackle is approximately one-half
the speed of transverse wave velocity. Schardin (1959) states that this velocity, under
normal circumstances, is the approximate final fracture velocity. He found these velocities
ranged from 750 to 2155ms -1 in glasses containing 40.7% PbO and 100% SiO2.
The authors have not observed velocity hackle on joints in rocks, even when an electron
microscope is used to examine individual grains cut by joints (Kulander et al. 1979, fig.
41A). This implies that natural joints propagate at stable rates and can be stopped at any
point by decreasing the applied stress. In this case, the entire joint surface, including twist-
hackle and rib marks, would comprise the mirror. However, smooth mirrors, as in glass,
would not develop because of granularity, porosity and fractographic features developed
at low propagation velocities. Neither would rock mirrors, encompassing the entire joint,
be planar because stress distributions in large natural systems are commonly more
complex at all scales than in smaller glass samples that are more homogeneous and have
regular dimensions.
Fig. 14. (A) Diagram showing intersecting points of sonic wave, generated at a surface flaw, and
crack front at selected times along a Wallner line. (B) Diagram showing how Wallner lines shaped
like gull wings are developed. Displacement stresses of a sonic wave, generated on the lee side of an
interior inclusion, catch up and couple with crack-tip stresses. Selected intersection points of sonic
wave and crack front along the gull wings are shown. Two different fracture fronts (A and B) during
propagation are shown to demonstrate this effect on Wallner line geometry.
Wallner lines
Wallner lines are produced w h e n displacement stresses related to a passing sonic wave
couple with principal stresses at the propagating crack tip. The sonic waves can be
generated when a spreading fracture tip encounters a surface inhomogeneity. Successive
6~. ~ A twist hacklefringe ~.B

hiP, :ill flit I tl tlll :!116 III II :11"J h,'l le
overall fracture ,. [
J
propagationdirection-/
/ ,~ / ,.~.
I
_1
\\
f c
~ vel~ .. ~ X
hackle "1~ ~ ~ D
A - ~ 5~ Fra~:ture,e ment"N 1
s,6e,ace
D under greatest tension
Fig. 15. Transient features on a fracture that forked.
elements of the sonic wave overtake the fracture tip and cause it to oscillate briefly out of
the mean fracture plane in an attempt to remain perpendicular to the resultant tension
(Fig. 14A). In a similar sense, Frechette (1984) mentions Wallner lines caused by sonic
waves generated by vibration resulting from stress release during cracking.
Wallner lines are convex in the overall direction of fracture propagation and can be
used to infer the general shape of the fracture front. They do not show the exact shape
at a particular instant because a sonic wave does not overtake the entire fracture
instantaneously. Consequently, Wallner line distortion is most pronounced near the wave-
generating flaw where curvature of the wave front is most pronounced.
A unique pattern of paired Wallner lines, commonly called gull wings (Frechette 1972),
is generated by sonic waves as the fracture passes interior inhomogeneities (Fig. 14B). The
geometry of all Wallner lines is dependent upon the shape of the fracture front and
inclusion, as well as the velocity, of sonic waves and fracture propagation. It follows that
analysis of Wallner line-shape in the mirror region can be used to calculate fracture
velocity if thc sonic velocity is known (Poncelet 1965; Congleton & Petch 1967; Kulander
et al. 1979). The distance between sections of adjacent Wallner lines qualitatively suggests
variations of tensile stresses and propagation velocities along the crack tip. Away from the
inception point of the causative sonic wave these quantities are low where Wallner lines
are more closely spaced (Fig. 10).
The authors have not observed Wallner lines on joints at a microscopic scale in
individual grains or at larger intergranular scales. Their absence persists even though
granular pervious rocks contain a myriad of inhomogeneities at which sonic waves are
undoubtedly generated. Absence of Wallner lines in rock cannot be attributed, in every
case, solely to a masking effect imposed by granularity. For example, Wallner lines, with
maximum amplitude of several millimetres and maximum half-wavelengths e. 1 cm, have
been observed on fractures in ceramic insulators (Kulander et al. 1979, fig. 41a). The
authors, however, have never seen Wallner lines on rocks of similar fine-grained textures.
In this case, absence of Wallner lines is attributed to low joint propagation velocities that
permit sonic waves to overtake the entire joint tip over a short distance.
Fork&g
The propagation velocity at which fracture forking occurs is dependent on the material's
elastic parameters. The velocity critical to development of velocity hackle and forking is
similar, and is approximately one-half that of a transverse wave in a given material
(Schardin 1959). Furthermore, if development of velocity hackle and subsequent forking
are not artificially suppressed, the velocity at the inception of forking approximates the
final propagation velocity.
Fractographic details common to forking are depicted in a sketch of a fracture caused
by bending (Fig. 15). The point of initiation for the bifurcation is on the slide surface
under greatest tension. Forking is preceded by development of velocity hackle. After
forking, fracture propagation slowed and then accelerated as shown by velocity hackle
that disappears and then reappears. Such renewed acceleration may lead to additional
forking.
The number of individual bifurcations that are bounded by the outermost forking
segments is dependent upon the intensity of applied stresses and the resulting amount of
stored strain energy to be released. For example, Schardin (1959) concludes that an excess
of elastic energy can exist within the fracturing material after forking. Consequently,
surface energy, attributed to additional surface area of the bifurcating fractures, does not
adequately slow propagation velocities, and individual bifurcations may fork again. In a
similar fashion, excess tip stresses and elastic strain energy, that may exist immediately
after fracture initiation, decreases the distance between the origin and the point of initial
bifurcation.
Bieniawski (1967) produced fracture forking at critical propagation velocities in
crystalline rocks during controlled experiments. He concluded that forking and excess
grain damage mark the boundary between stable and unstable propagation. Bieniawski's
stable growth, at lower propagation rates, is controlled by the applied load and a
deficiency in related stored strain energy. During this stage, fracture propagation can be
stopped by decreasing the load stress. However, if unstable propagation is induced,
marked by velocity hackle and forking, elastic energy, even though reduced, is sufficient to
maintain propagation under a reduced load. Similarly, Irwin (1960) implies that the
transition from stable to unstable propagation in brittle metals occurs when the energy
released per unit of crack surface attains a critical value.
Velocity-related forking of joint surfaces in sedimentary rocks, including fracture
surfaces induced by drilling and coring, has not been observed by the authors. Joints
studied in these observations include those developed regionally in horizontal rocks of
diverse lithology within cratonic basins, and joints within foreland fold belts. In every case,
joints of all sets contain fractographic features that must be attributed to Mode I loading
and propagation. By adhering to this premise, the authors conclude that they have never
seen a joint in sedimentary rocks within the described geological settings that propagated
solely by Mode II or III shear stresses acting on the crack tip.
Twist and &clusion hackle

Twist and inclusion hackle, unlike velocity hackle, can form at any propagation velocity
and location on a crack surface. Twist hackle, common on joints, is initiated when a
fracture progresses into a region where the principal tension rotates from perpendicular to
I
I I
I I /
leading rib
segment
bifurcation
point
B
Fig. 16. (A) Hypothetical plumose structure confined within an individual layer. Rib marks are
shown by solid lines. Orthogonal plume components are shown by dashed lines. Straight and parallel
rib segments are shown at A, circular rib mark at B, a parabolic section of rib mark is shown at C.
See text for discussion of points D and E. (B) Plan view of bifurcating rib mark.
the crack tip line. Such a rotation implies a corresponding rotation of the crack plane.
However, the entire tip cannot rotate by an instantaneous and equal rotation.
Consequently, the fracture or joint breaks into en echelon, but disconnected, twist-hackle
faces, with the tip lines on each face turning to become perpendicular to the new and
rotated principal tension. Some, or all, twist-hackle faces may then curve abruptly, while
maintaining a continuous surface, into an adjacent neighbour to form a twist-hackle step,
thereby completing local separation (Kulander et al. 1990, fig. A-5). Alternatively,
adjacent twist-hackle faces may be joined by later-formed twist-hackle steps that are not
related to the initial fracture event. In any case, primary faces and steps can break into
secondary, then tertiary, and even smaller, faces and steps at increasingly diminished scales
(Kulander et al. 1990, fig. A-4). Development of such twist-hackle components attests to
the complex stress distributions, at decreasing scales, along a propagating crack tip.
Inclusion hackle, a form of twist hackle, can be a common constituent of hackle plumes
in granular pervious rock (Kulander & Dean 1985). It is generated when a fracture plane is
warped by local alterations of tip stresses near or within an inclusion. Inclusions are
commonly pore spaces or material possessing elastic moduli in contrast to the matrix, as
demonstrated in capillary rods.
The planar warping and separation, during passage of a fracture through or around an
inclusion, causes the initial plane to be at different levels on the inclusion's opposite side.
To complete material separation, one segment of the original plane will hook abruptly into
the other. This connecting segment is an isolated twist-hackle step that diminishes in relief
away from the inclusion and is elongate in the direction of local fracture advance,
However, on an even more local level, the direction of propagation within the step is
oblique to this direction (Kulander et al. 1990, fig. A-8). In any case, the diverse shape and
distribution of inclusions can cause related hackle to be irregularly shaped, producing a
surface that is rough overall.
Twist hackle commonly form distinct fringes near a stratum boundary where a joint
commonly terminates. Twist hackle can also develop as bands where a single joint
completely transects a layer of contrasting lithology and continues into adjacent layers
where the twist hackle may disappear (Fig. 3), Twist hackle can also be irregularly
distributed and form a complex collage of faces and steps (Fig. 13). This latter case is
common in massive rock units
In glass slides and layered rock, the most widespread twist-hackle fringe generally shows
greatest relief between adjacent hackle faces and is located toward the stratum boundary
furthest from the origin and plume axis. At the inception of this fringe, more closely
spaced Wallner lines in glass (Fig. 11), as well as rib marks or constructed fracture front
lines in rock (Fig. 3), show that fracturing velocities and related tip stresses were
diminished. In this case, the hackle plume generally has a well-developed and fairly linear
plume axis that reflects a regular distribution of crack-tip stresses. In a similar sense, the
irregular collage of twist hackle faces (Fig. 13) reflects a highly irregular distribution of low
differential stresses at the crack tip throughout the stratum. Consequently, no initial
planar segment of the joint develops that contains a fairly linear and extensive plume axis
(Fig. 3).
Rib marks
Rib marks are produced at a velocity discontinuity where fracture propagation can
hesitate in response to an abrupt change in the direction and magnitude of principal
tension at the crack tip. If propagation continues this tension direction is commonly
altered causing the new fracture surface to be abruptly inclined to that formed previously.
Figure 4 depicts common morphologies of rib marks caused by fracture arrest. However,
the rib mark in Fig. 4D is rounded in cross-section and may not reflect total arrest of
fracture propagation. For this reason, the authors suggest that the term arrest line be
replaced by rib mark.
A geometrically regular rib mark in plan view, with sharp crest and no inflection points,
or a constructed fracture front line drawn perpendicular to hackle components, traces
points of instantaneous hesitation all along the fracture front. Both naturally developed
rib marks and constructed fracture front lines with this geometry are convex everywhere in
the direction of crack propagation. They may be nearly circular near the origin but
approach a parabolic form as the fracture progresses in a layered medium (Fig. 16).
Geometrically regular rib marks and front lines contain no sections that are locally
concave in the direction of fracture propagation. Such regular features are useful for
inferring and comparing, to the smallest feasible scale, stress configurations, relative stress
magnitudes and relative propagation velocities during advance along a past fracture tip
(Kulander & Dean 1985; Kulander et aL 1990). For example, a straight segment of a rib or
front line marks points of instantaneous hesitation on a planar crack and indicates
constant propagation velocity and stresses unchanged in magnitude and direction along its
length (Fig. 16A). A circular rib or rib segment on a planar crack also marks
instantaneous hesitation, thereby indicating constant fracturing velocity and stresses that
remained unchanged in magnitude. In this case, however, the greatest and intermediate
principal stresses changed direction at a constant rate along its length (Fig. 16B). An entire
rib or isolated rib segment with parabolic form (Fig. 16C) on a planar crack indicate
that propagation velocities, stress magnitudes, and the directions of the greatest and
intermediate stresses were altered at a regularly changing rate along its length. Adjacent
rib marks and front lines that are geometrically repetitive with respect to a plume axis
show maximum separation along the axis (Fig. 16A, point D). Here, tensile stresses (~r3)
along a given rib or line acting perpendicular to the fracture plane at the tip, and related
and propagation velocities, were greatest. Conversely, these values were lowest where
adjacent rib marks were most closely spaced (Fig. 16A, point E). Maximum (Crl) and
intermediate (a2) tip stresses within the fracture plane at the tip were perpendicular and
parallel to rib mark tangents at any point. These stresses rotate and generally change
magnitude from D to E as a3 decreases. If these stress parameters do not hold, the regular
and repetitive geometries of propagating crack tips could not be maintained. Con-
sequently, the crack tip would become geometrically irregular.
A geometrically irregular rib mark and fracture front line possess sections that are
alternately convex, then locally concave, in the direction of local propagation. Such
features imply propagation rates that change irregularity and may not trace points of
instant fracture hesitation. This observation is strengthened by geometrically irregular rib
marks that bifurcate from a common point (Figs 13 & 16B). Here, the crack front
delineated by the lone rib mark remained stationary on one side of the bifurcation point.
Whereas, the crack front, on the other side of the bifurcation point, spreads to the position
of the leading rib segment. Irregular geometry of rib marks mandates an irregular
geometry for other fractographic features within a plume.
These combined geometries reflect irregularly changing fracturing velocities, stress
magnitudes and stress directions along the crack tip. Frechette (1972) states that, in glass,
irregular and very closely-spaced arrest lines develop at stable propagation velocities
below 10-3m s. Furthermore, rib marks are not developed on glass fractures that are
accelerating. They may develop, however, on thermal contraction fractures.
Conclusions
Simple fractography experiments utilizing easily obtained glass objects give geologists a
greater appreciation of joint propagation dynamics than would be gained by rock study
alone. For example, glass fractures are characterized by fractographic features, absent on
joints, that develop at high (unstable) propagation velocities. This acceleration is
attributed to mechanical and textural characteristics of glass that lead to levels of stored
strain energy much elevated over that stored in rocks during jointing. Furthermore,
fractographic features that develop at high propagation velocities are absent on joints in
granular pervious rock.
Fractographic knowledge provides a basis for conclusions centred upon something
other than joint-trend geometry and postulated geometrical relationships with surround-
ing structures. For example, primary fractographic features at any scale are formed by
brittle failure during Mode I loading at the crack tip. A shear origin for joints, caused
by Mode II or III loading at the crack tip during propagation, is precluded. However,
externally applied stresses sufficient to cause failure may contain a strong shear
component. Secondary structures such as slickenlines or solution features, are superposed

on joint surfaces by subsequent shear movement or compression.
Rib marks and the related geometry of other plume components, as well as constructed
fracture front lines, record the details of propagation dynamics. Geometrically regular rib
marks show the location of the fracture tip at some past instant of time. Such rib marks
show principal stress directions and relative magnitudes of tensile stress operating
perpendicular to the propagating crack tip as well as relative propagation velocities. These
observations can be compared throughout a hackle plume to a scale determined by surface
structure development and preservation. Conversely, geometrically irregular rib marks
may not mark points of instantaneous hesitation and are accompanied by an irregular
geometry of hackle patterns within a plume.
Geometrically irregular rib marks and those that are closely spaced, coupled with
irregular geometry of plume components, imply stable joint propagation at low velocities.
Furthermore, formation of velocity hackle and forking, commonly developed during
unstable propagation, have not been found by the authors in geometrically regular or
irregular plumes, on joints cutting granular pervious rock. Similarly, Wallner lines have
not been observed on joint surfaces in sedimentary rocks.
The authors appreciate reviews by M. Ameen, M. Cooke and an anonymous reviewer. M. Ameen's
time and effort in seeing this collection of papers through to publication is especially noted.
References
BANKWITZ,P. 1965. Uber Klufte, beobachtungen in Thurinischen schiefergebirge. Geologie, 14, 242-
253.
- - 1966. Uber Klufte II. Geologie, 15, 896-941.
BARTON, C. C. 1983. Systematic jointing in the Cardium Sandstone along the Bow River, Albena,
Canada. PhD Thesis, Yale University, New Haven, CT, USA.
BmNIAWSKI, Z. T. 1967. Mechanism of brittle fracture in rock, part II - experimental studies.
International Journal of Rock Mechanics, Mineral Science, 4, 407-423.
CONGLETON,J. & PETCH,N. J. 1967. Crack branching. Philosophical Magazine, 16, 749-760.
FRECHETTE, V. D. 1972. The fractology of glass. In: PYE, D. L. (ed.) Introduction to Glass Science.
Plenum Press, New York, 432-450.
- - 1984. Markings on crack surface in brittle materials: a suggested unified nomenclature. In:
MECHOLSKY,JR, J. J. & POWELL,JR, S. R. (eds) American Society for Testing and Materials, STP
827, 104--109.
GASH, P. J. 1971. A study of surface features relating to brittle and semibrittle fracture.
Tectonophysics, 12, 349-391.
HODCSON, R. A, 1961. Classification of structure on joint surfaces. American Journal of Science, 259,
293-502.
IRWIN, G. R. 1960. Fracture mechanics. In: GOODIER,J. N. & HOFF, N. J. (eds) Structural Mechanics:
Proceedings of the 1st Symposium~Naval Structural Mechanics, Pergamon, New York 557-591.
KIES, J. A., SULLIVAN,A. M. & IRWIN, G. R. 1950. Interpretation of fracture markings. Journal of
Applied Physics, 21, 716-720
KULANDER, B. R. & DEAN, S. L. 1985. Hackle plume geometry and joint propagation dynamics. In:
STEPHANSSON,O. (ed.) Fundamentals of Rock Joints Proceedings. Cenetek Press, Lulea, Sweden,
85-94.
, BARTON, C. C. & DEAN, S. L. 1979. The application offractography to core and outcrop
fracture investigations. Morgantown Energy Technology Center, United States Department of
Energy, METC/SP-79/3
, DEAN, S. L. & WARD, B. J, 1990. Fractured core analysis: interpretation, logging and use of
natural and inducedfractures in core. American Association of Petroleum Geologists, Methods in
Exploration Series, 8.
82 B. R. KULANDER & S. L. DEAN
MURGATROYD,J. B. 1942. The significance of surface marks on fractured glass. Journal of the Society
of Glass Technology, 26, 155-171.
POLLARD, D. D. ~I; AYDIN, A. 1988. Progress in understanding jointing over the past century.
Geological Society of America Bulletin, 100, 1181-1204.
--, SEGALL,P. • DELANEY,P. T. 1982. Formation and interpretation of dilatent echelon cracks.
Geological Sociey of America Bulletin, 93, 1291-1303.
PONCELET,E. F. 1958. The markings on fracture surfaces. Journal of the Society of Glass Technology,
42, 279-288
1965. Modern concepts of fracture and flow. Poulter Research Laboratories, Stanford
Research Institute, Menlo Park, Technical Report 002-65.
PRESTON,F. W. 1926. A study of the rupture of glass. Journal of the Sociey of Glass Technology, 10,
234-269.
- - 1935. Angle of forking of glass cracks as an indicator of the stress system. Journal American
Ceramic Sociey, 18, 175-176.
- - 1939. Bottle breakage - causes and types of fractures. Bulletin of the American Ceramics
Sociey, 18, 35~50.
PUGH, S. F. 1967. The fracture of brittle materials. British Journal of Applied Physics, 18, 129-162.
QUACKENBtJSH, C. L. & FRECHE~E, V. D, 1978. Crack-front curvature and glass slow fracture.
American Ceramic Society, 61,402-406.
SCHARDIN,H. 1959. Velocity effects in fracture. In: AVERBACK,B. L., FELBECK,D. K., HAHN, G. T. &
THOMAS, D. A. (eds) Proceedings, International Conference on Atomic Mechanisms of Fracture.
John Wiley, New York.
SI-IAND, E. B. 1959. Breaking strength of glass determined from dimensions of fracture mirrors.
Journal American Ceramic Society, 42, 474-477.
WALLNER, H. 1939. Linienstrukturen an Bruchflachen. Zeitschriftffir Physik, 114, 368-378.
Boston Society of Natural History Proceedings, 27, 163-184.
ZAPFFE, C. A. & CLOGG, JR, M. 1944. Fractography - a new tool for metallurgical research.
Transactions, American Society of Metallurgy, 34, 71-107.
From Ameen, M. S. (ed.), 1995, Fractography: fracture topography as a tool in fracture mechanics and stress analysis 83
The character of rock surfaces formed in Mode I
C. B U T E N U T H & M. H. de FREITAS
Centrefor Geological Engineering, Department of Geology, Imperial College
of Science, Technology and Medicine, London S14/'72BP, UK
Abstract: Samples failed in extension, i.e. in Mode I, are distinguishable from those
failed in shear and mixed modes, i.e. in Modes II and III, by the macroscopic and
microscopic character of their failure surface, i.e. by the fractography of their surfaces.
This paper describes the use of fractography to discover whether the results of a rock
test to failure have been influenced by both the properties of the rock tested and the test
method used. The test in question (the hoop test) is relatively new and its operation has
not yet been fully analysed; however, the use of fractography has confirmed the mode of
failure operating in the test is tensile (Mode I). Further, fractography enables the results
obtained from these tests on rock to be better interpreted and analysed with greater
confidence than was the case before.
Fractography has been used by metallurgists since the 1940s to study brittle failure but the
availability of increasingly sophisticated imaging techniques, and the associated processing
of data from them, has enabled the subject to be extended to cover both metallic and non-
metallic materials, including composite materials (Roulin-Moloney 1989). With rock it is
possible to obtain very useful data relevant to brittle fracture using field-scale, as well as
laboratory-scale, observations, as has been shown in a review of the subject by Bahat
(1991). This paper describes the application of fractographic observations made in the
laboratory to the analyses of a test of tensile strength (a hoop test): in doing this, the
observations so made also provided data on the characters of rock surfaces failed in
tension, i.e. in Mode I (Fig. 1).
Fig. 1. The basic modes of loading and their associated type of failure; (a) Mode I, opening; (b) Mode
II, sliding; (c) Mode III, tearing.
84 C. B U T E N U T H & M. H. de FREITAS
10
MQrbte hoops
I ! I I 10
Sandstone
I
hoops
I ! I
dry
---8
z
r~
o
x 6
- /.." _8
z
trt
o6
X
u
t_.
"2 ,,~
0 I I t I 0
| i
10 I I I I
12 saturated "/ saturated

8
10
xl,
GJ
u
~2
U-
] ,, | 0
0 2 l, 6 8 I0 0 2 4 6 8 10
(a) Area (x10-3 m2} (b) Area (xlO3m2)
Fig. 2. Results of direct tension tests (O) and hoop tests (e); Force = ultimate load line force at
failure; area = cross-section area of hoop, shown stippled (from Butenuth et al. 1993).
!
Z
D~~~c~i~ AF
QJ
t,,3
t,--
o
LL.
0 Area (m2)
Fig. 3. The ratios AF/AA and intercepts on the axis for area.
The hoop test and the results obtained with it have been described in some detail by Xu
et al. (1988), John et al. (1991) and Butenuth et al. (1993). In the hoop test an annular
hoop of rock is formed by coaxial drilling; two hemi-cylindrical loading platens are
inserted into its central hole and expanded until the hoop fails. Two parameters are
required to calculate the tensile strength of the material, namely the ultimate load line
force at failure and the area of failure at that time.
R O C K S U R F A C E S F O R M E D IN M O D E I 85
Examples of results obtained from wet and dry hoop samples of Penrith sandstone and
Italian marble are illustrated in Fig. 2, together with the strength of the material measured
in direct tension (Butenuth et al. 1993). Note, that in the hoop test, the slope of the force v.
area relationship describes a ratio AF/AA which is in very good agreement with the ratio
F/A defined by tests in direct tension (for details see Butenuth et al. 1993). This indicates
that there is a phenomenological correlation between the two tests which permits the ratio
AF/AA defined by the hoop test to be accepted as a measure of tensile strength (Fig. 3).
Hoop tensile strength, fiT, is thus defined as:
tYT = Aft'/mA. (1)
However, as can be seen from Fig. 2, the slope defining AF/AA for hoop tests sometimes
has a positive intercept on the area axis: thus, it holds that for tests in direct tension (suffix
D):
O'TD = (FD--O)/(AD--O) ----FD/AD, (2)
whilst for tests in hoop tension (suffix H):
OrTn = ( F H - O ) / ( A - AO)H = FH/(A--AO)H. (3)
Although the hoop test has proved to be a very convenient method for studying the
strength of rock in extension there are two aspects of the results obtained from it which
require clarification, namely (1) the deviation of some hoop test results from the line
defined by the majority of tests (see Fig. 2); and (2) the cause of the A0 intercept (see Figs 2
&3).
Spread of results
The results for saturated hoops of sandstone (see Fig. 2) show this spread of results most
clearly; samples which had a failure area of (4.2 to 4.6) • 10-3 m 2 (arrowed) did not fall on
the line defined by the results obtained for the same material, cut from the same block and
tested in the same way. There was obviously something different about them and the
samples were thus re-examined. Figure 4 illustrates one of these samples whose area, as
calculated from the external dimensions of the hoop (the stippled area shown in Fig. 3),
was 4.6 • 10-3 m 2. It can be seen that this is an overestimate of the actual area of failure
because only the right-hand side of the hoop has completely failed in tension; tensile
failure has not completely broken the left-hand side. Subtracting the area that has not
broken in tension from the originally calculated area reduces the area from 4.6 • 10-3 to
3.57 • 10-3 m 2 and brings the result of this hoop on line with those of the other samples.
That sample (Fig. 4) also showed the path of the fracture indicated by the heavy solid
line on the right and left sides of the hoop differs from that shown by dotted lines on the
left side of the hoop, the former having a more planar path than the latter. This difference
in macroscopic features permitted other samples in the saturated sandstone set of Fig. 2,
which had been completely broken in half during testing, to be re-examined on the basis
that essentially planar surfaces represented the areas which had failed by the time F had
reached its ultimate value: i.e. they represent the relevant A for the ultimate value of F. All
the samples whose values deviated from the straight line could be remeasured using this
principle: all values then fell on to the line.
86 C. BUTENUTH & M. H. de FREITAS
5 cm
9 I I I 1 I
Fig. 4. Main failure path in a hoop, shown by the continuous line, and branching of the main failure
on the left side of the hoop, shown by dotted lines (see text).
This simple application of the most visible and obvious features of fractography thus
resolved a problem with the results obtained that could not otherwise be answered.
Smaller scale observations revealed even more about the dynamics of the test used and the
manner of hoop failure.
Microscopic observations
The failure surfaces formed slowly by extension and prior to fast fracture, and ultimate
hoop failure, always started at the inner boundary of the hoop, opposite the area of platen
separation. It is here that unequivocal evidence of the fractographic character of failure in
tension can be found. Such evidence can be illustrated most clearly with reference to the
samples of sandstone.
The sandstone tested is of wind-blown origin having been accumulated as sand dunes: its
constituent particles are rounded grains of quartz. These particles have subsequently been
cemented by quartz which has formed an overgrowth to the original grains. The quartz
cement adopts the crystal structure of the grain it coats and shrouds the grain in quartz
that has a crystal form of rhombohedra and prisms when free to grow into pores. Under
tension these crystal overgrowths separate from each other revealing the mineral cement
bridges they have formed between grains and occasionally the faces of their crystal form
which were originally in contact with their neighbour. Sometimes the overcoat of quartz is
plucked cleanly away from the rounded grain it covers to reveal the grain itself. Figure 5a
illustrates these features from a location < 0.5mm from the inside edge of the hoop
opposite the area of platen separation.
At the other end of the fracture, i.e. near the hoop boundary furthest from the platens in
the area where the macroscopic planarity of the failure surface begins to be lost, or is lost
in a bifurcating system of cracks, the microscopic character of the surface is similar to that
ROCK SURFACES FORMED IN MODE I 87
,~ "0 ,,'0
~ ~.~
~ .-~ ~ !
~~
"~ "~ r~ -,-~
o ,,,,,I
0,~
=~ .
0 oO"
Fig. 6. Schematic illustration of failure surfaces with essentially planar surfaces shown stippled.
Fig. 7. Schematic illustration of types of failure surface with Aplanar shown stippled.
shown in Fig. 5a, with crystal faces of the overgrowth being visible and rounded grains
exposed from beneath their overgrowths. Thus, despite the macroscopic differences
referred to earlier (i.e. planar surfaces passing with distance from the initiation of failure to
bifurcating surfaces) the microscopic evidence suggests that the surfaces always separated
in tension.
Figure 5b illustrates the failure surface developed in a hoop of marble: this photograph
was also located within 0.5 mm of the inside edge of the hoop where failure commenced.
The crystalline nature of the marble is clearly seen with fracturing occurring between and
through the crystals of calcite. These microscopic characters are visible along the whole
length of the failure surface, from where it starts at the inside edge of the hoop to where it
ends at the outer boundary of the hoop. Here again the microscopic evidence suggests that
the entire failure surface was formed under tension.
The intercept A o
Reference to Fig. 4 and samples displaying similar incomplete failure across both sides
of the hoop indicated that the total cross-sectional area (A) defined by the external
dimensions of the hoop, as shown stippled in Fig. 3, did not correctly describe the area of
failure. In addition, many hoops had a failure surface which departed from the plane of
platen separation seen in plan and/or in elevation. This should not have occurred if they
had failed in pure tension (see the path of failure on the left side of the hoop shown in Fig.
4). Such failure is schematically illustrated in Fig. 6: the stippled areas represent failure
planes lying in the plane of platen separation and the unornamented areas represent failure
surfaces which depart from that plane. It should be noted that in many cases the failure
surfaces which depart from the plane of platen separation form part of a bifurcating
system of fractures. In such cases only one side of the bifurcating fracture pattern forms a
plane of sample separation, as can be seen in Fig. 4. Such bifurcations increase the total
area of failure.
The appearance of many hoop samples, once failed, is illustrated in Fig. 7, the stippled
areas being the area of failure in the plane of platen separation. Summing the total failure
area produces a larger area than that defined by the simple rectangular dimensions of the
width of the hoop and its depth, A. Thus, for any sample:
A=l+m+n+o+p, (4)
whereas:
A p l a n a r ~-~ m + n = A related to Fultimate , (5)
for A F / A A = o T.
Thus, the intercept is:

A o = A - Aplanar. (6)
It will be noticed (Figs 2 & 3) that Ao is a constant regardless of A.
The constancy of Ao
The areas of failure which have a consistent relationship with the load line force at failure
(Fig. 3) i.e. the stippled surfaces of the type m and n in Fig. 7, are bounded, and thus have
an area defined by the onset of bifurcation; the cause of this bifurcation is thus suspected
to be related to the constancy of Ao. There are two possible causes for these bifurcations
and they could be operating either independently or together. One cause is related to the
presence of bending in the sample and the other is related to crack branching associated
with dynamic crack growth.
Bending
The hoop test is not as straightforward a test as may first appear - its operation is the
combination of direct tension and pure bending, as shown in Fig. 8a (John et al. 1991).
This combination changes the magnitude of the hoop stress across the sample (Fig. 8b).
Photoelastic modelling and the strain gauging of elastic samples confirms the existence of
an axially oriented neutral axis (na in Fig. 8b) and an outer zone of compression (Gentier
90 c. BUTENUTH & M. H. de FREITAS
Pure bending
+
(a) (b) Nill

Dmect tension
Fig. 8. (a) Super position principle and its application to the analyses of a hoop test; (b) hoop stress
distribution in a curved beam subjected to pure bending: maximum negative hoop stress (i.e. tensile
stress) is at the inside edge of the hoop and decreases with an increase in radial distance becoming
positive (i.e. compressive) beyond the neutral axis (na) (from John et al. 1991).
et al. 1991; John et al. 1991), even though the microscopic characters of the surfaces
indicate that failure has always been in tension (Fig. 5). The position of this neutral axis on
either side of the hoop will only be symmetrical about the plane defined by the load line if
the platens and the sample are both coaxial and concentric. Unless extreme measures are
taken with sample preparation, some departure from such symmetry must be expected.
Platens can move in one or more of three ways;
1. They may separate as they are supposed to, so that each pair of points on opposite
sides of the plane of platen separation moves apart by the same amount as measured
in the direction of the force line. This assumes the platens expand uniformly in a
sample whose inner hole has perfectly parallel sides coaxial with the vertical axis
which passes through the centre of expansion of the platens.
2. They may tilt by opening about a vertical axis which is not concentric to the sample
so that one side of the platens, seen in plan, opens more than the other (Fig. 9a).
3. They may tilt about a horizontal axis so that one end of the platens, seen in vertical
section, opens more than the other (Fig. 9b).
As nothing other than ordinary care in good workshop practice is taken with machining
the platens, which are made from steel, and drilling the samples, which are cored with
good quality diamond impregnated bits, it is inevitable that condition (1) cannot always be
guaranteed. Conditions (2) and (3) commonly occur and at present are considered to be
the most likely reasons for failures of the type shown in Fig. 4.
The cause of tilting need not always be due to the geometrical mismatch between platens
and sample. A perfect match may be upset if one side of the sample begins to fail before
the other and this can be thought of as occurring either in plan or in section, or in both.
This has been observed using reflecting holographic interferometry (which is capable of
detecting displacements measured in nanometres); this has shown that although both sides
of a sample deform, and that the deformation can be almost symmetrical, one side
invariably reaches yield and ultimate failure before the other (A1-Samahiji 1992).
Whatever the cause, axes are created about which bending occurs (neutral axes).
(a)
onQ
"~149o9 '1
(b) ~176176
~176176
Fig. 9. (a) Asymmetric opening of the loading platens about a vertical neutral axis (na) at time
of maximum load line force, and appearance of failed sample; (b) asymmetric opening about a
horizontal neutral axis (na) and appearance of failed specimen.
Figure 10 illustrates the consequence of having such axes within a specimen; they divide
the sample into zones of tension and compression. One such axis, e.g. either vertically or
horizontally oriented (Fig. 10a or b) will divide the total potential failure area into two and
if both axes co-exist the potential failure area is quartered (Fig. 10c).
Figure l la is a photograph of the macroscopic character of a sample where
fractography suggests such axes existed, and Fig. 1 l b illustrates its basic features 9 The
sample is sandstone from the same batch as shown in Fig. 2 and has a failure similar to
that on the right side of the specimen shown in Fig. 4 (although it is not a photograph of
that specimen). Tensile failure commenced at the left-side margin of the sample, which was
the inside of the hoop adjacent to the platens, and progressed to the right. The surface
roughness increases towards the top-right quadrant where it is hilly and rough 9 A more
detailed inspection of the surface further revealed that escarpments shadow two areas of
origin (Fig. 11c), so that by using morphological evidence on the surface (see, for example,
Kulander & Dean 1985) the progression of failure can be retraced (Fig. 1ld). If bending is
accepted as the explanation for the surface topography shown in Fig. 11, then it can be
argued that this sample commenced its failure with one, vertically oriented, neutral
surface, as in Fig. 10a, but whose more rapid failure in its lower half (Fig. 1lc & d) resulted
in the later development of a horizontally oriented neutral axis, as in Fig. 10b, culminating
in a final region of compression in the top-right quadrant, shown stippled in Fig. 1lb.
Crack branching
The topography shown in Fig. 11 could also be attributed to crack branching during
dynamic crack growth. Fracture mechanics has demonstrated that branching occurs when
the release of stored strain energy at a propagating crack tip is so great that continued
) 8 i"i(2--
(a) 84 (b)
no
fzCx CzCx
I
t tx ICztx
I
(c)
Fig. 10. Schematic illustration of (a) vertical and (b) horizontal neutral axes (na), (c) quartering a
failure surface created by their coexistence: tz, Cz, tension and compression about a vertical neutral
axis; tx, Cx, tension and compression about a horizontal neutral axis, respectively.
growth of the existing crack produces insufficient new surface area to consume the energy
being released. Shock waves reflected back from the boundaries of the specimen can cause
new cracks to form close to the existing (or main) crack (Ewalds & Wanhill 1984). The
failure on the left side of the sample shown in Fig. 4 could be a result of this process.
The question thus arises: is Ao = constant defined by bending or branching or by both.
Fractography does not answer this directly but provides a powerful piece of evidence
which unites all the observations made. Of the many samples tested none have exhibited
slickensides or damage that can be clearly attributed to shear: all the surfaces appear to
have failed totally in tension. For example, the surface in the top-right quadrant of Fig.
1 lb is not slickensided but is rough, even though this area could well have been under
compression and susceptible to shear failure under compression. Such roughness as seen
here is a fractographic characteristic of branching. Branching is most likely to occur when
the tip of the main crack is 'blunt' as blunt ends to cracks tend to cause branching whereas
sharp ends do not (Ewalds & Wanhill 1984). The coincidence of the onset of branching
with possible positions of neutral axes suggests that crack branching and specimen
bending are in some way related. Although the nature of any such relationship has yet to
be defined the evidence implies the following.
As A = Ao + Aplanar, it follows that Aplanar (A is related to Fultimate for A F / A A = a.r:
equation 5) has to increase as A increases, given that Ao proves to be a constant. Thus, for
any ratio of inner and outer diameter of the hoop, the neutral axis or axes move with
progress of the test. However, the position of the axes is dictated by a balance between the
tensile strength of the material and the unfailed distance of hoop ahead of the major
propagating crack tip. As the hoop is progressively deformed prior to ultimate failure, and
as the neutral axes can be thought of as moving away from the boundary at which tensile
failure commenced, so the aspect ratio of the crack decreases, its tip bluntens and the
conditions conducive for branching commence. The failure path of the sample in Fig. 4
could be interpreted as showing how this tried to happen. Failure commences on the inside
of the hoop and developed a reasonably planar surface for c. 2.5 cm on the right-hand side
and 1.0cm on the left-hand side, i.e. the right side appears to have been failing in
preference to the left. The character of the surface on the right side then changes to a
rougher form (possibly reflecting incipient branching as the tip of the propagating crack is
progressively bluntened by the bending illustrated in Fig. 8a). This continues for the
remaining 2.0 cm of its length until it intersects the outer boundary of the hoop. The
failure surface on the left side also changes in character, it too having a rougher character
for its remaining 2.0 cm of propagation, possibly for the same reason as that suggested for
the roughness of the crack on the right. Branching had clearly begun to develop on the
left-hand side at c. 1.5cm from the inside edge of the hoop and was the sole form of
fracture after 3.0 cm. If branching is associated with blunting of the propagating crack tip
and linked to bending of the sample (Fig. 8a) then the onset of branching on the left side
could have occurred at the time when failure had completely broken the right side,
enabling the hoop to open like the letter C. Such movement would tend to blunten any
crack tip on the left side of the hoop.
Confirmation of such a model requires a linkage between the load-displacement plots
for such tests and the onset of bifurcation, which is most likely to occur when the load-
displacement curve has a negative gradient; future testing will attempt to reveal the nature
of this linkage.
Conclusion
Fractography has proved to be an extremely useful tool in studying the behaviour of rock
samples tested to ultimate failure. It provides data that cannot easily be generated from
numerical models founded on concepts of continuum mechanics, elasticity and statics. It
also enables the dynamic aspects of fracture propagation and development in rock to be
studied and offers considerable insight to the interpretation of results based on statics.
The surfaces formed in porous and crystalline rocks by Mode I type failure have
distinctive marks of tensile fractures. At microscopic level, at magnifications of as little as
x 30, clear evidence of grains being pulled apart can be seen in the porous sandstone. In
the marble, failure of complete crystals and crystal mosaics along cleavage and crystal
surfaces gives witness to the presence of Mode I failure. A striking feature in the specimens
of both sandstone and marble is the cleanness of the fractures - they are free from
slickensides and debris associated with shear displacements. At higher magnifications
94 c. B U T E N U T H & M. H. de FREITAS
0 o b b
ml~ i " "
d c
l I tension
0 5cm ~ compression
(b) L..., .... , i i , I
~escarpmenf
"*%.
~176176
(c) , (d)
Fig. 11. Macroscopic appearance of a failure surface formed in Mode I. (a) Photograph of failure; (b)
general appearance of the surface, tensile failure progressed from left; (e) escarpments on the surface
indicating the progress of failure in this plane; (d) reconstructed directions of fracture propagation
from the trace of the escarpments on the failure surface. Note this surface is not that shown in Fig. 4
but was formed in the same way.
(• 60 + ) the microscopic facets associated with brittle failure can be seen on grains and
crystals cut by failure surfaces.
N o t only does fractography enable the results of tests to failure to be analysed, it further
permits the relationship of the sample to its test method to be studied, thus enabling
properties which are a function of the material to be separated from those which are a
function of the test (e.g. the Ao v. A problem for the tests described in this paper).
Fractography also has the great advantage of enabling new ideas and observations on
the failure of samples to be applied to old results for force and displacement from samples
failed during earlier experiments, provided the fracture surfaces are carefully preserved
and are still available for study.
References
AL-SAMAHIJI,D. K. 1992. Experimental investigation of the tensile failure of rock in extension. PhD
Thesis, Imperial College of Science, Technology & Medicine, London University, UK.
BAHAT, D. 1991. Tectono-fractography. Springer-Verlag, Berlin.
BUTENUTH, C., DE FREITAS, M. H., AL-SAMAHIJI, D. K., PARK, H. D., COSGROVE, J. W. &
SCHETELIG, K. 1993. Observations on the measurement of tensile strength using the hoop test.
International Journal of Rock Mechanics and Mining Sciences and Geomechanics Abstracts, 130,
157-162.
EWALDS, H. L. & WANHILL, R. J. H. 1984. Fracture Mechanics. Edward Arnold, London.
GENTIER,S., POINCLOU,C. & BERTRAUD,L. 1991. Essai de traction sur anneux (Hoop Tensile Test),
application aux schistes ardoisiers. R 31682.BRGM. Orleans, France.
JOHN, S., AL-SAMAHIJI,D. K., DE FREITAS, M. H., COSGROVE, J. W., CLARKE, B., LOE,N. & TANG,
H. 1991. Stress analysis of a unidirectionally loaded hoop specimen. Proceedings of the 7th
International Congress on Rock Mechanics, Aachen, Germany, 513-518.
KULANDER,B. R. & DEAN, S. L. 1985. Hackle plume geometry and propagation dynamics. In:
STEPHANSSON,O. (ed.) Fundamentals of Rock Joints. Proceedings of Lulea University of

Technology, Sweden, 85-94.
ROULIN-MOLONEY, A. C. (ed.) 1989. Fractography and Failure Mechanisms of Polymers and
Composites. Elsevier Applied Science, London.
Xu, S., DE FREITAS, M. H. • CLARKE, B. A. 1988. The measurement of tensile strength of rock.
International Symposium, International Society of Rock Mechanics and Power Plants, Madrid,
September 1988, 1, 125-133.
From Ameen, M. S, (ed.), 1995, Fractography: fracture topography as a tool in fracture mechanics and stress analysis
Fractography and fracture characterization in

the Permo-Triassic sandstones and the Lower Palaeozoic
Basement, West Cumbria, UK
MOHAMMED S. A M E E N
GeoScience Ltd, Silwood Park, Buckhurst Road, Ascot SL5 7Q IV, UK
Present address: Independent Consultant, 122 Tachbrook Street, Pimlico,
London SW1 V 2ND, UK
Abstract: Three distinctive phases of fracturing (mesofractures and macrofaults) are

recognized in the Permo-Triassic part of the cover rocks of West Cumbria. The earliest
phase caused syn-depositional growth macrofaults with northerly strikes in the lower
part of the cover rocks. This phase is not evident in any of the macrofaults and
mesofractures which are exposed in the study area. The second phase led to the
occurrence of the easterly striking macrofaults and mesofractures and is clearly evident
from surface exposures. The latest phase caused the exposed northerly striking
macrofaulting and mesofracturing. Some of the latter macrofaults most probably
propagated upwards from the tips of the pre-existing (phase 1) northerly macrofaults.
Others resulted from the coalescence of mesofractures and propagated downwards.
The fractographic features on both the second and third phase mesofractures suggest
that they occurred as Mode I (tensile) fractures and were later modified by shear or
mixed-mode loading. The geometry and degree of development of fracture surface
morphology are controlled by a combination of factors.
The Lower Palaeozoic basement is characterized by complex meso- and macro-
fracture patterns, including a large number of local sets with a wide range of
orientations. Although 'statistically' prominent regional sets are recognized in the
basement rocks, these are purely geometrical and include various types of fractures that
developed under different stresses. Each local set of mesofractures shows varied
attitudes of fractographic features, indicating different dynamic implications of
fractures within the same set. Some macrofaults in the cover may have propagated
from the basement, or are related to movements on basement faults which controlled
the evolution of the sedimentary basin and its subsequent structures.
The study area is a part of the Lake District which is dominated by two distinctive
stratigraphic/tectonic divisions: the Lower Palaeozoic basement rocks and the Upper
Palaeozoic to Mesozoic sedimentary cover rocks. Each of these divisions includes several
lithostratigraphic units (Fig. 1). The area covered in this work includes parts of the Lake
District Block and the onshore parts of the northeastern edge of the East Irish Sea Basin,
and is flanked to the north by the Solway Basin (Fig. l a). The Lake District Block is a
structural high composed mainly of Palaeozoic igneous rocks (granitic batholiths and
volcanic rocks) which has received a thin sedimentary cover during the Carboniferous to
the Cenozoic. The detailed stratigraphic and tectonic evolution of the region (particularly
in the Palaeozoic) is complex, and has been the subject of ongoing debate which has been
described in a voluminous literature (see the bibliography given in Smith 1974; Jackson et
al. 1987; Mosely 1990; Chadwick et al. 1993 a, b). The post-Palaeozoic tectonic regime is
relatively better defined in terms of the regional structures (Fig. l a). However, the
98 M.S. AMEEN
Silioth ":.'.N.':.'.'.'.':.'.'.'.':.'.'.'.':.
,v-
9 - i~fldor
V~e !ii!i!iiiii!
! i!iii.!ii,,l
Basin
Ramsey- Whitehaven
Ridge / / / / ~ \ \/.~-,F,~.:-:-:
,/\ \7-/-/4/~
',, ~1 ~.\...::-Y "/~'r
%LakeDistrict.7 x" . .
.
.
.
.
.
.
.
.
. . . .
. o .
Outline of
/ Study Area
), (Fig Ib)
I-7 Permo-T~kYoug~rt
Cove
--480 1771 Car~ifems
[ ] Lo~rPaheo@ie }hsmmRocks
-9- - Er~io~ F~t lOhn
. a . RevereFault ' '
(a) 280 300
I 1
~ l h haven
0 E] Permo-Trias
Carboniferous
r-] Borrowdale Volcanics
1"~ Granite/Grano-Diorlto
0 ~ ~ ? ~ 9 ~,o,~o~a,,t~
SellafieldX~ll -- "~,, i~,/k [
_495 (b) -'~/ k..)~~

29s , 305 ,~
Fig. 1. (a) Location (insert) and general geology map of the English Lake District (modified after
Chadwick et al. 1993a). (b) A schematic map showing the distribution of the studied localities (stars).
FRACTOGRAPHY AND FRACTURE CHARACTERIZATION 99
temporal evolution of some of the tectonic events is poorly constrained and the overall
effect of these events is therefore equivocal. The summary of the post-Palaeozoic tectonic
events given below is mainly based on Chadwick et al. (1993a, b).
The East Irish Sea and the Solway Basin developed as extensional/transtensional basins
due to an E-W tension which dominated NW-Europe, in the Permian to Early Jurassic. In
the Late Jurassic to the Early Cretaceous these basins were affected by N-S tension. The
Mid to Late Cretaceous was characterized by regional post-extensional subsidence. In the
Palaeogene to Neogene a regional uplift was accompanied by some degree of reverse
movement on the pre-existing extensional faults in the East Irish Sea and the Solway
Basins. The latter movement is the result of N-S to N W - S E compression related to the
convergence of Africa and Europe which was culminated in the Oligocene to the Miocene.
The Lake District block was affected by a considerable uplift due to the crustal shortening
during this phase. In addition, both the Lake District Block and the surrounding basins
suffered a regional flexural uplift as a peripheral effect of doming associated with the onset
of seafioor spreading between Scotland and Greenland. The maximum depth of burial of
the Permo-Triassic St Bees and Calder Sandstones in the study area are estimated to vary
from 1400 m over the Lake District Block to c. 2000 m in the eastern flank of the East Irish
Sea Basin. The maximum horizontal in-situ stresses in the region are oriented N-S to N W -
SE.
Objectives of the present work

This work forms a part of the ongoing geological and hydrogeological investigation into
the suitability of the Sellafield area for a nuclear waste repository. The present study
focuses on the fracture patterns and their evolution in selected parts of the cover rocks and
the basement rocks. The study uses field observations of fractures (including joints, faults
and veins), particularly meso- and macrofractures, to achieve its objectives. It incorporates
fractography of mesofractures (see Definitions) as a tool to determine the dynamic
implications of the fractures and the tectonic evolution of the study area. Fracture data
were collected from coastal and inland exposures of the Ordovician basement (the
Borrowdale Volcanic Group and the Ennerdale and Eskdale Granites) and the Permo-
Triassic cover rocks (the Calder and St Bees Sandstones within the Sherwood Sandstone
Group). In the following sections of the paper these will be referred to as 'basement rocks'
and 'cover rocks' respectively. A total of 452 planar and 69 linear data were collected in
ten localities in the basement rocks, and a total of 500 planar and 31 linear data were
measured in eight localities in the cover rocks (Fig. 1). Individual localities cover an area
of 400 m wherever possible.
Definitions of terms used in the study
Fractures
The term 'fracture' is used here in the broad sense defined by Kulander et aL (1990) as a
break or physical discontinuity in a rock caused by stresses exceeding the rock's strength.
Therefore, it covers faults, joints, stylolites and veins. Faults are fractures which exhibit
appreciable shear displacements, i.e. displacement parallel to the fracture surface. Joints
are fractures which show little or no displacement parallel to the fracture surface. The limit
of displacement (parallel to the fracture plane) beyond which a fracture is classified as a
100 M.S. AMEEN
fault is arguable. The scale of observation should be considered, i.e. the presence of an
observable amount of offset along the fracture surface on the scale of observations should
be used as a criterion for that distinction. Stylolites are pre-existing fractures or bedding
plane surfaces along which pressure solution or dissolution has taken place leaving
insoluble residues (e.g. clay minerals) giving the surface a dark colour. They tend to show
various degrees of waveforms in profile section. Veins are fractures infilled with minerals
such as quartz and calcite.
Fractures develop in rocks on various scales and can accordingly be classified into four
scale groups (Turner & Weiss 1963). Microfractures occur on microscopic scale and
include those fractures which affect individual crystals or grains. Mesofractures can be
effectively observed in 3D with or without a hand lens in hand specimens or large
continuous exposures. Macrofractures are too large or too poorly exposed to be examined
directly in their entirety in one exposure. Megafractures are hundreds or thousands of
kilometres in length, and may occur on continental scale, e.g. the Dead Sea fault.
Fractures can be classified into two main groups according to their geometrical
consistency in terms of their orientation and shape (Ramsay & Huber 1987). Systematic
fractures are planar and develop into well defined sets, each include parallel or sub-parallel
fractures within the field of observation. Non-systematic fractures are non-planar and tend
to show no consistent orientation within the field of observation. The present paper deals
with systematic fractures.
Fractography and fractographic features

Fractography is defined here as the discipline of diagnosing, describing and interpreting
the fracture surface morphology or topography (fractographic features) which are related
to the initiation and propagation of the fractures. The fractographic features include two
main groups. The first are referred to here as 'surface structures' or 'surface patterns' and
cover all features which can be observed directly on the exposed fracture surface. They
occur as perturbations or undulations forming distinctive geometrical patterns. Such
features were first described by Woodworth (1896), and their anatomy is illustrated in
Fig. 2. Surface features related to shear movement (faulting) on fractures (slickenlines) are
not included in Woodworth's, or later, classifications (e.g. Kulander et al. 1979, 1990).
These are often related to later rejuvenation of fractures, and are therefore not included in
the present definition. The second group of fractographic features includes the forms of
the trace of the fractures as observed on a free surface (e.g. bedding surface or cliff face).
They are referred to here as fracture traces (Fig. 2).
The fractographic features are the result of the interaction between a complex
combination of factors including remote stresses, lithological and mechanical hetero-
geneity in the rocks, and pore-water pressure. The dynamic implications of fractographic
features are discussed in a separate section.
Mesofracture patterns in the cover rocks
Systematic fractures
The systematic fractures in the cover rocks of the study area include joints, veins and
faults. They are classified according to their orientation (and sense of slip of faults) into
sets. The fractures pattern is dominated by three distinctive sets, a southerly striking set
2a
2b
/i"
I
1. Main joint face 6. Twist-hackle face
2a. Abrupt twist-hackle fringe 7. Twist-hackle step
2b. Gradual twist-hackle fringe 8. Rib marks ( front lines of the fracture )
3. Origin 9. Hooking
4. Hackle plumes 10. En echelon fractures
5. Plume axis
Fig. 2. A block diagram illustrating the different types of surface structures (patterns) and fracture
trace architecture (partly based on Kulander et al. 1990).
(referred to here as set 1, which includes subsets la and lb), an easterly-striking set
(referred to here as set 2, which includes subsets 2a and 2b), and bedding-parallel fractures
(referred to here as set 3, or bedding fractures). A brief description of these fractures is
given in Table 1 and their particular features in each locality are summarized in Table 2
and plotted in Fig. 3.
Non-systematic fractures
These are a group of fractures which develop nearly at a right angle to the systematic
fractures, as irregular, often curved, relatively short and open fractures (Table 1). The
present study concentrates on the systematic fractures.
Surface features of the systematic mesofractures
Fracture traces
Most of the fracture traces are observed on gently dipping bedding surfaces and to a lesser
extent on coastal cliff faces. The observed characteristics of sets 1 and 2 are generally
similar and are summarized below.
On bedding surfaces the traces of the fractures form discrete discontinuities (Fig. 4a) or
fracture zones (Fig. 4b-d). The latter consist of smaller fractures parallel to the zone (Fig.
4b) or oblique to it (Fig. 4c), with or without a consistent en echelon pattern (cf. Fig. 4b
with 4c). In thick, cross-bedded sandstones some fractures show anastomosing traces (Fig.
4e). In poorly lithified sandstones the fractures occur as zones of anastomosing microfaults
(granulation seams or microgouges) which distinctly protrude above the wave-cut patchy
102 M.S. AMEEN
,..]
o~
z
00 9
b e d d i n g p a ra lle l f r a c t u r e s "
.. . -..:
9 o~149
'~
B~ O
9
Sets
,Ib
-. - ..
::i:
~, :" 9 & 9 J
oo 9
set 2 ".
~ " .
(b)
~leff ~, INI
i] [
r
f o r set 2
Fig. 3. (a) Pole diagrams of the fractures in the St Bees Sandstone (Loc. 1-8, Fig. lb). (b) Synoptic
pole diagram of all the fractures in the St Bees Sandstone shown in (a). (c) Synoptic great circles
diagram of all the fractures in the St Bees Sandstone shown in (a). The possible direction of the
causative maximum principal stress (0-] eft) is indicated for each set, the causative minimum
principal stress (0 3 eft) is horizontal and perpendicular to 0-1 eft. For details see text.
104 M. S. A M E E N
o
~ ,,..,~
~~
6~=8
r.
e~o~
o <
VI VI 2: e:m
~5
,.o
o
o
~ m~ o
.,-q
t'N r c~
<- o
r t-'q~
o
o ~ tr
~
Ot'N
o
o O0
0
d, - o
~ ,,...q
o,.
O
o
~
o . ,....~
i
_,~ ~ o
O
~'F::-~ -~'~ ~
~'~
9~ ~:~
~ O
.~ ~
~ 0.~
b I "" ~ N ' - d
c~ o
~
o
"" ~ N
c~ L
~ = g SN
.~. ~ . ~ =
o o
z~ o "~
~ .,.
;::1 . ~ ~
,-4
o" .... ,.....~
-~ W W "'-
m t~
A A A A o A A & d &A o~
",N
o o 7
V/ A A A r A\ A m'7' V / ~~ A \ o~
o
m ~ m. moo
~; ~o ~_ ~ ~? .~ o 9
O
NI
b~
o
mg ::^ A\ i A\ ~ W o ~ V/ WV/ A\ ~
0
t,4
i i -" i
.,.,~
0
o o o o o o o o o o
&
o
d
. . :dd,. . :d, a a 2 z a z r~
~ ,,,,4
d,
t~ O O O O ,...~ ~'~
a~,2a2 2 ~, 2 2 2 2 i ~'~-
o,,o
0 ~
z,~
~ z ~':~ Z ~ ~~ e~
0
~,~
ff
o e~
o
d ~g
[-, O JJ
106 M.S. AMEEN
9 ~,~l~.~ ~ :n =
~=eo ~ ~ ~ a o ~ . ~.
-,~ ~ ~ r "~ ~
9 O,.I~
,-.i
o
Z
o
d 7
r
~. ~. V/ V/ ~-- V/ m" ~- ~ ~ ~V/ i/h
.i,-. oo
~,.-,
A A ~ ~ A A A ~o o o A o ~ A .,,,~
r.~
A em A d V/ d A~ ~1~ ~[~I A Od d d A d A~ ,- b~
o
,q.
~
. o,,,-.,
o
N
o -. ~ ,= o
9-~ f~
~_. A\ I V/ A\ <:~ ~ . ~ . A i ~_. V/ ~V/
r
-- --- ,~ --- ,4' i~
0
2
o o o o o o o o
"-
IP-,- t",l ~ ~ '~
I'~01 ~"~ i P--.
~ ~ ~
: oo
I[""- ,,.-,,,~ ~ i" ~ I["~ ~
t",,i
~ n:::l
~0 I:I
,...d ,.u
b
r.~
~.;.r
0", ~1
o
FRACTOGRAPHY AND FRACTURE CHARACTERIZATION ] 07
0 ~ 0 ~
..~I o ,.~i o
o
0 0 0 0 ~.-~ O=
O
Z
o
o ~ ? ~
~
u~. r
a~ A A A A A A A A
o o o o o o o b~
A A A A A A\ A A
o
O
~
I o
V/ V/ o
t~
b~
o
o
o
I ~ V/ I I I I I o~
o
o
o o ~
N2 t I I I I I I I O
. ,,,.,
9 I::I
o~'~
o o
~Z
,~o
%
Z Z o
'tD
108 M.S. AMEEN
exposures of gently dipping beds (Fig. 4f). Similar features were recognized by Aydin &
Johnson (1978) who called them deformation bands.
Some fractures develop splay patterns and pinnate joints (Fig. 4g). Subsets la and lb,
and 2a and 2b do not occur together in each of the studied localities. However, when
subsets l a and l b and/or 2a and 2b occur together their traces appear to cross-cut each
other forming an 'X' shape or abut against each other forming a 'Y' shape (Fig. 4h & i). In
some places a ' l a ' or '2a' fracture zone is found to consist of smaller lb or 2b fractures,
respectively, and/or vice versa with an en echelon arrangement (Fig. 4j) similar to those
described by Beach (1975).
On cliff faces, the traces of set 1 and set 2 fractures occur as discrete discontinuities or
fracture zones which either cut across bedding discontinuities or abut against them (Fig.
5a & b). Some of the fractures are offset by bedding fractures (Fig. 5c), though this is rare
and is characterized by small displacement. In some cases a fracture is limited to one or a
few beds, depending on the bedding-joints spacing. Some fractures continue upsection as
vertical or steeply dipping discontinuities (Fig. 5b), others tend to become gentler, branch
off and may join bedding discontinuities (Fig. 5d).
The interaction of set 1 and 2 fractures is relatively simple. The filled or sealed set 2
fractures are consistently cut across, and in some places displaced by, set 1 fractures (Figs
4 & 6a, b). However, the latter abut the open set 2 fractures. The interaction of the traces
of the set 1 and set 2 fractures in each of the studied localities is schematically shown in
Fig. 7.
Surface patterns (structures)

No identifiable surface patterns occur on the bedding fractures, probably due to
weathering and the limited exposure. However, both set 1 and 2 fractures show surface
structures which are summarized in Table 3 and illustrated in Figs 8-11 (although these
are better developed and/or preserved on set 1). The fact that coastal cliff faces (which
mostly trend N-S) form the main source of data has undoubtedly limited the exposure and
thence observations of set 2 fractures. The interpretation of the surface patterns is
discussed in a later section.
Is the layering control on surface patterns universally pertinent?

The configuration and degree of development of the surface patterns on both sets 1 and 2
fractures are constrained by lithological heterogeneity, syn-sedimentary structures and the
occurrence of free surfaces (open fractures), such as bedding parallel joints. The shape of
the rib marks and associated plume axes are considered by some authors (e.g. Kulander et
al. 1979 ) to be constrained by the existence of layering, where fractures located far from
bedding discontinuities may develop circular rib marks, whereas those developed between
more closely spaced layer boundaries tend to show parabolic or elliptical rib marks and
layer-parallel plume axes. Such a constraint is unequivocal when a bedding discontinuity is
a free surface and is used here to date the fracture (chronology). However, the constraints
of bedding or layering, defined by lithological changes rather than free surfaces, on the
fracture surface patterns is equivocal, as is evident from this study. The systematic set 1
fractures commonly contain plume components that cross sedimentary layering, thereby
ruling out exclusive control of 'non-free surface' layering on the development and
geometry of surface features. Lithological heterogeneity and syn-sedimentary structures
FRACTOGRAPHY AND F R A C T U R E CHARACTERIZATION 109
110 M.S. AMEEN
Fig. 4. (a-k) Photographs illustrating the different patterns of fracture traces on bedding surfaces.
For details see text. Hammer shaft, compass and pencil are 0.3m, 10 and 14cm in length
respectively, rucksack is 0.5 m in height.
112 M.S. AMEEN
Fig. 5. (a--d) Photographs illustrating the architecture of the fracture traces on cliff faces, bedding
traces (B) are nearly horizontal. For details see text. Hammer shaft and measuring-tape box are
0.3 m and 5 cm in length, respectively.
(e.g. cross bedding) seem to affect the degree of development rather than the geometry of
surface features (Fig. 8). This may be related to the stress intensity at the propagating
fracture tip being high enough for the fracture to propagate across coherent sedimentary
layering or lithological heterogeneity. In addition, asymmetry in the development of the
fracture surface structures may reflect a corresponding asymmetry in the intensity of
fracturing stresses (Fig. 9a & b).
FRACTOGRAPHY AND FRACTURE CHARACTERIZATION 1 13
Fig. 6. An overall view (a) and a close-up view (b) showing an older fault of set 2 (F2) with two sets of
slickenlines (arrows); horizontal (older) and oblique (younger) and later calcite seal (c) which has
been cut and displaced by younger fracture of set 1 (F1).
114 M.S. AMEEN
od
I
d
o
,_.]
o ~ ~
o~'~ ~,I
..~ ~,',. ~~'~ ~

,..o
I ? .~ ,.o
z-.--- ~s
~ ~,o.. ~,o. o "a
,~,~ C~:~ ~'<

~
g
.,..9,
~ o ..,i "U
oo~,.~
& ~
"r~"/:
_X/ r..q ,%
<.,< ~ - / ~
C~
,.-I
o ~
o0 ~
I ~ I i i\~
Ib / ~i
~,~,, <.,,/ oI o
oL og Z ~ - ~ - d r-:
~ ~ fit ~\ ~ --
,m
Table 3. A summary of the surface patterns (surface features) on the systematic sets 1 and 2
Feature Set 1 Set 2
Origins Mostly located in the beds; in the middle Mostly located in the beds in the middle or
or at top or bottom third of the fractured at top or bottom third of the fractured bed or
bed of horizon (Figs 8 & 9). Voids, coarse horizon (Fig. 11). Voids, coarse grain or
grains or lithological heterogeneity form lithologlcal heterogeneity form the nuclei of
the nuclei of the fractures (O; Fig. 8a). the fractures.
Multi-origins are common on large
fractures (Fig. 8a).
Rib marks Well developed, mostly parabolic in shape Weakly developed, circular concentric around
with their axes mostly parallel or nearly the origin (Fig. 11) or parabolic in shape.
parallel to the bedding (Figs 8 & 9). The The axes of the latter are mostly parallel or
spacing of the rib marks, their geometrical nearly parallel to the bedding. The spacing
symmetry and degree of development vary and degree of development of rib marks vary
greatly (Figs 8 & 9). greatly.
Hackle plumes Well developed, chevron shaped with their Weakly developed, either radial associated
axes mostly parallel or nearly parallel to with concentric rib marks, or chevron shaped
the bedding trace on the fracture. Their with their axes gently inclined to the west at
geometrical symmetry and degree of ~< 15~ (Fig. 11).
development vary greatly (Figs 8 & 9).
Twist-hackle Both gradual and abrupt twist-hackle Mostly sharp with average twist angles of 11~
fringes fringes occur with average twist angles of for subset 2a and 13~ for subset 2b. The
30~ for subset la and 15~ for subset lb. sense of rotation of twist faces is clockwise
The sense of rotation of twist faces is in subset 2a and counter-clockwise in subset
clockwise in subset la and counter- 2b. The degree of development, geometry
clockwise in subset lb. The degree of and symmetry of the occurrence of the twists
development, geometry and symmetry of on the opposing fringes of a fracture vary
the occurrence of the twist hackles on the widely
opposing fringes of a fracture face vary
greatly (Figs 9b & 10).
Degree of The surface features of well-exposed Characterized by relatively faint surface features,
development, fractures continue across many bedding partly due to obliteration by later shear
symmetry and parallel fractures and sealed fractures of movements (faulting), and masking by calcite/
continuity of set 2 (Figs 9 & 10). However, the degree of haematite coating and filling. However, similar
surface development, symmetry and continuity of factors to those quoted for set 1 have most
features the surface features appear to be affected probably affected the evolution of the fracture
by lithological heterogeneity, syn- surface features for set 2.
sedimentary structures, and the magnitude
and orientation of local effective principal
stresses across the fractured horizon, in
addition to the existence of open
fractures (free surfaces).
C h r o n o l o g y o f the mesofractures
Cross-cutting and abutting relationships of the fracture sets described above suggest
that set 2 developed first, and were then sealed by calcite and/or iron oxide. Some
mineralization occurred during subsequent faulting, as evident from fibrous mineral filling
(Fig. 12a & b). This phase was followed by set 1 fractures which cut and displaced all
sealed set 2 fractures. However, where the seal never existed, or had been partly or fully
removed, set 1 fractures abut set 2 fractures. The continuation of surface structures on set
1 across the traces of sealed set 2 fractures support this conclusion (Figs 8b & 12c).
116 M . S. A M E E N
~O~o
~,~ ~,~ ,~ .0 0
e ' ~ O ~
cn
,- o ~
~o~
.$-~ ~ . . . " ~ .
:~.: o ~
~ ,~, C ~ , .,..~ ~ . ~
~0 N= o
, .~ 0 ;~.~ ~
oo2=~
,.~ ~,~ . = ~
~-q o
~g
. o~_~
The majority of bedding-parallel fractures occurred later than set 1 and 2 fractures.
However, some bedding fractures seem to have developed throughout the different phases
of deformation. Some set 1 and 2 fractures are terminated against bedding fractures or are
displaced by them. However, most of set 1 and 2 fractures cut across and displace the
bedding fractures (Fig. 5). In addition, the continuation of surface patterns on many
systematic fractures, across existing bedding fractures, indicates that the latter developed
after the former (Figs 8 & 12c). Alternatively, at least some bedding fractures could have
developed first and then healed before being cut by the systematic fractures. However, the
lack of any mineralization along bedding discontinuities makes the latter explanation less
likely.
The non-systematic fractures always end at the open systematic fractures or cut across
them, if they are healed (Fig. 13d), and are most probably the result of late-release stresses
which post-date the systematic fractures.
Dynamic implications of the mesofractures

The configuration of surface patterns (structures) and traces of fractures are considered to
be a direct result of the magnitudes and directions of the local principal stresses at the tips
of the propagating fractures, within a fractured stratum, as illustrated in Fig. 13a. The
dynamic implications of each of the systematic fractures (sets 1 and 2) are discussed
according to their chronology, starting with the older set (set 2). In this context, the author
uses local principal stresses to mean the net or effective maximum, intermediate and
minimum principal stresses (al eft, a2 eft, a3 eft, respectively).
Set 2 mesofractures
The majority of the set 2 fractures lack visible surface patterns, but show slickenlines
and/or are mineralized. Visible surface structures (Table 3) indicate that the fractures
originated within the beds. The fractures then propagated either radially and cyclically
(evident from rib marks, e.g. Fig. 1la), or steadily (evident from the lack of rib marks)
with their leading edge (defined by the hackle-plume axis) inclining at 15~ to the west.
These configurations suggest that the fractures originated as Mode I (tensile) fractures,
with the local minimum principal stress (cr3 eft) oriented at right angles to the main
fracture surface (nearly horizontal). The local maximum and intermediate principal
stresses (al eft and a2 eft) lie within the fracture surface. At the leading edge of individual
fractures al eff is inclined at 15~ to the west and a2 eft is at right angles to it (Fig. 11).
Fractures which are not mineralized, and show no visible surface patterns, can be
considered as Mode I (tensile) fractures which have propagated steadily and therefore lack
rib marks. However, fractures which show evidence of shear movements (e.g. slickenlines)
most probably originated as Mode I (tensile) fractures and were subsequently subjected to
shear loading. This explanation is evident from the occurrence of rib marks and hackle
plumes on some fractures which show indications of 'later' shear movement (faulting). In
addition, as shown in Fig. 7 (Loc. 5), some 2a faults are formed by the coalescence of small
en echelon 2b fractures. The latter, in turn, consist of smaller, right stepping en echelon
fractures (counter-clockwise twist faces). The occurrence of some 2a and 2b fracture zones
in which the constituent small en echelon 2a and 2b fractures enclose a small acute angle
(< 20~ as shown in Fig. 7 (Loc. 4), offers a good constraint on the directions of the
118 M.S. AMEEN
F R A C T O G R A P H Y AND FRACTURE CHARACTERIZATION 119
principal stresses during the mixed-mode loading. The sense of stepping and twist on these
zones indicate an overall N N W - N N E horizontal, 0-3 and E N E - W S W vertical 0-1/0-2plane.
The sense of slip on the fractures indicates that 0-1 was gently dipping to the west during
the final stages, which resulted in the sinistral normal and dextral normal faulting on the
2a and 2b fractures, respectively.
Therefore, subset 2a and 2b fractures indicate a consistent overall remote stress regime
characterized by a horizontal northerly 0-3 which varied in orientation temporarily and/or
spatially between a N N E and a N N W direction. The O'1/O"2 plane is perpendicular to 0"3,
with 0"1 > 0"2 and 0-1 gently inclined to the west during the latest phase of faulting (Fig. 13).
However, early (burial and/or tectonic) vertical or near vertical 0-1 is very likely. The
dynamic implications of the set 2 fractures are used to construct stress maps in the study
area associated with the evolution of these fractures in individual localities (Fig. 15a-c).
The occurrence of out-of-plane propagation of both subset 2a and 2b fractures is
manifested by the occurrence of both gradual and abrupt twist-hackle fringes. These are
the results of gradual or abrupt changes in the orientation and/or the magnitude of the
local principal stresses. Assuming a linear-elastic behaviour, the angle of rotation of a
twist-hackle face (twist angle) is a function of the rock's rheology and the ratio of shear to
normal effective stresses. Such a relationship was first analysed and quantified (in terms of
Poisson's ratio and the ratio of Mode I to Mode II stresses) by Lawn & Wilshaw (1975) for
brittle solids. Pollard et al. (1982) and Cruikshank et al. (1991) applied the latter analysis
to certain sandstones. This method for estimating stresses is useful, but the results have to
be considered cautiously, because it is based on the linear-elastic fracture mechanics, a
simplified theory of a complex phenomenon (Lawn & Wilshaw 1975). The average twist
angles on subset 2a and 2b fractures of 11 and 13~ respectively (Table 3), can be
accordingly used to estimate the ratio of shear to tensile stresses at the time of fracture
propagation.
Assuming a Poisson's ratio of 0.25 for the St Bees and the Calder Sandstones the tensile/
shear stress ratios for subset 2a and 2b fractures are 0.11 and 0.13, respectively. Due to the
lack of information about the pore-fluid pressure at the time of fracturing it is impossible
to estimate the regional stresses from these ratios. However, it can be concluded that the
net normal stresses were considerably higher than the shear stresses. The common
occurrence of calcite mineralization along the set 2 fractures suggests an important role of
pore-fluid pressure in the initiation and/or propagation of the fractures. Detailed fluid
inclusion analysis may help in estimating the pore pressure (Engelder & Lacazette 1990)
Fig. 9. Photographs (left) and maps (right) showing: (a) A set 1 fracture propagated with a maximum
rate along arrow 1 to rib mark R1 then downward (arrow 2), as evident from the larger spacing of the
rib marks in these directions. Note the convergence of several rib marks between R1 and R2 at I,
indicating (slower) propagation than along arrows 1 and 2. This drop in the propagation rate
coincides with the occurrence of a smaller, parallel joint which originated at O' and propagated along
arrow 4 (whose projection on the main fracture lies on the same line as I). Rib mark deflections at II
coincide with the intersection of the rib marks and a carbonate-rich lens and are accompanied by a
zone (E) of smaller, interfering rib marks and hackle plumes (small arrows indicate the local
direction of propagation). Note the continuation of the surface features across a sealed set 2 fracture
(F). (b) Asymmetrically developed surface patterns on a set 1 fracture. (c) A partly exposed set 1
fracture surface showing three main phases of propagation history: an early, cyclic, often
interrupted, phase evident from closely spaced rib marks (phase 1); a transitional, cyclic but less
interrupted phase, evident from wider spaced rib mark (phase 2); and a final non-interrupted phase
evident from the absence of rib marks (phase 3). Note the localized occurrence of twist hackles in
phase 2 near its border with phase 1.
120 M.S. AMEEN
Fig. 10. Set 1 fractures showing: gradual (a & b) and sharp (e) twist-hackle fringes, both with a
counter-clockwise sense of rotation of the hackle faces (1 and 2) relative to the main fractures (M).
Large and small arrows indicate the directions of propagation of the main fracture and twist faces,
respectively. The dashed line in (a) is a rib mark. The photograph in (b) shows the trace of the upper
fringe zone of the fracture in (a) on a bedding plane.
A,.o 0 ~
o o O - ~
~'~o ~.~ 0 ~
~ N ~
~ ' ~
~o~:~:
o ~ ~,-1 ~
~ . ~
~"CI 0 ,~ ",'~
I
.-I
I ~--~ . ~ ~ 0
~~ ~_._.
~.~ ~
,-~ ,.~ " ~ ,.~
~-.~ 0 ~ 0
,..~ ~ 0 ~
~ o =
0 o ,-~ r
. ~ N ~
~
.;..i -~,~
~<.~
122 M.S. AMEEN
Fig. 12. (a) A dextral-normal fault (F) of sub-set 2a with calcite filling exposed partly on a cliff face
(C) and partly on a bedding plane (B). (b) Fibrous calcite filling on the face of the fault in (a)
indicating an oblique normal-dextral movement (parallel to arrow). (c) A dextral fault (F2) of sub-
set 2a with incongruous step structures (S) and fibrous calcite coating which are cut by later set 1
fractures (F1) and bedding joints (B). (d) The traces (on bedding surface) of mineralized systematic
set 2 fractures (F2) cut by later non-systematic fractures (F).
124 M. S. AMEEN
9~ ~ ~,.~l ~-~,-~
0.~ ~ ~,.0 _n~n:~
cq
~l:~ ~ ~'t~ ~ ~
o u~
\l o ~=~" ~ ~:~ ~ ~ ~>~--~ ~.~
f\
Z ~ -- ,~ ~ ~ +~ +~ ~ 0
~.~.~ ~.: ~ ~:
~.~ ',~ ~ ~ ~ ~
~ l-~ " ~ ~ Q.)~ ~.;
~-.~ ~ .~..~ (1.) c~ ~.-~, q-~ ~ ,~
n9 ~ ~ oo o ~
+'~.~" 0,.0 ~[~ 9 0 "~
9~ > ~~7_ ~ , _ ~
,.~ ~ ~~ ~= . ~ "~ ~ ~",m
1 ~
z~ .g,"<.~ ~ ~ ~ -~.~
'~ ~0 0 " ~ O::l ~ e~
0
" ~ l= ~ . , . +.~ ~ ~.~ ~
, 0 . ~ "'~ 0 "~'~ +-'

,c~ "~ ~ 0 c~ 0 ~ ~.~
_'-i ~ ~ e.~ I:~ ~'- t~
~ ~ ~ c~'~ ,-., = < . , l o ,..n

~_ ~ o .=.'~,_._ ~
I::::l..,._, ~ ~ ~,-~, ;~ ~ 0
9-- ~ .~;-~ ~.~ c ~ ' ~ 0 ~ ~.~
"-- ~ 0 ''~ c~ ~ ' ~
c:l +.~ ~ , . ~ -~,--~ ~ ~_..~

~ ~ o ~ ~-~ ~ =
.~ c~ ~ , . ~ ~. ~
t~
and hence the regional stresses. Such studies are currently being carried out by UK Nirex
Ltd, as a part of their ongoing geological/hydrogeological investigation in the study area
(Andrew Bowden, pers. comm.).
Set 1 mesofractures
Visible surface structures on the fractures (e.g. twist-hackle plumes) indicate that they
originated within the beds as Mode I (tensile) fractures. The local minimum principal
stress (a3 eft) is nearly horizontal, at right angles to the main fracture surface. The local a~
eft and cr2 eft lie within the fracture plane and are horizontal and vertical, respectively,
along the leading edge of the majority of the fractures (Fig. 13). Some fractures lack rib
marks and therefore indicate a steady propagation. However, the majority of fractures
show rib marks, indicating incremental interrupted growth. Such interrupted growth is
linked to local fluctuation in the minimum stresses (tr3 eft), possibly due to changes in
pore-fluid pressure. The common hooking of lb fractures on to la fractures indicates that
the former developed later. Therefore, it can be concluded that la and lb fractures
developed due to NE-SW to E-W tensions (~r3), respectively. Both subsets were later
affected by a stress system characterized by an ENE-WSW trending a3. Therefore, they
suffered mixed mode loading which resulted in the breakdown of the fractures into twist
faces. The consistent clockwise and counter-clockwise twist of the l a and l b fractures,
respectively (Table 3 & Fig. 7, Loc. 2a & b), is indicative of dextral and sinistral sense of
shear, respectively. The average twist angles on la and lb fractures are 15 and 30 ~
respectively (Table 3). These indicate ratios of shear to tensile stresses of 0.15 and 0.36
for subsets la and lb, respectively. The former ratio is similar to those obtained for subset
2a and 2b fractures, and indicates that the net normal stresses were considerably higher
than the shear stresses during fracture development. However, the ratio for subset lb
fractures is considerably higher, indicating that they developed under considerably lower
normal stresses and higher shear stresses than those of set 2 and subset 1a fractures. This
may be the result of sub-set l b developing later, and therefore at shallower depths than
subset la. The coalescence of the la and lb mesofractures under such shear
loading resulted in the development of faults. Stress pattern maps in the study area based
on the temporal evolution and dynamic implications of the set 1 fractures are shown in
Fig. 15d-f.
The relationship of the mesofractures to the macrofaults in the cover rocks
Macrofaults pattern
Macrofaults in the St Bees and the Calder Sandstones crop out in limited coastal
exposures between Seascale in the south and Whitehaven in the north. The best exposures
of these macrofaults occur on the coastal cliff faces between South Head and North Head.
The observed faults dip steeply (> 60 ~ and strike at N-S to NW-SE and ENE-WSW to
E-W. The lower hemisphere, equal area projection of these faults is dominated by these
two sets which are referred to as set 1 (northerly set) and set 2 (easterly set), respectively
(Fig. 14a). The observed exposed width of the macrofaults ranges from a fraction of a
metre to several metres. They show various degrees of development of fault rocks
(cataclasites and gouges), as illustrated in Fig. 14b.
126 M.S. AMEEN
The 3D geometry and size of the macrofaults are difficult to determine from the small
field exposures. However, seismic and borehole data indicate the presence of faults up to
several kilometres in depth and several tens of kilometres in length. In addition, a wide
range of geometries are evident with linear, wavy and anastomosing traces of the faults on
horizontal plan, and linear or curvilinear (concave upward) profiles of individual faults
(Chadwick et al. 1993ab). The net slip on all the exposed macrofaults have a consistent
extensional component and vary from oblique-slip (dextral-normal or sinistral-normal)to
strike-slip. There lacks field evidence of syn-depositional movements on any of the exposed
macrofaults of sets 1 and 2. However, subsurface data indicate the occurrence of such
movements on some of the northerly faults at depths within the lower part of St Bees
Sandstones (Jon Gutmanis, pers. comm.). The implications of this are discussed in the
section on the regional stresses and the evolution of macrofaults.
Chronology of the exposed macrofaults

At the immediate vicinity of some of the macrofaults there occurs a localized group of
synthetic and antithetic fractures (Fig. 14c). Such a relationship suggests that the
mesofractures are related to the macrofaulting. However, all the exposed macrofaults are
consistently parallel or nearly parallel to the regionally diffused, systematic mesofractures,
i.e. set 1 and 2 macrofaults are parallel to set 1 and 2 mesofractures, respectively (cf. Fig.
14a with Figs 3c &13d). In addition, set 1 macrofaults consistently cut and displace the set
2 mesofractures, whereas set 1 mesofractures persistently abut against open segments of
the set 2 macrofaults and cut across healed segments of the same macrofaults. Therefore, it
can be concluded that the majority of the exposed set 1 and 2 macrofaults resulted from
the coalescence of the set 1 and 2 mesofractures, respectively, during the same phases of
deformation which caused these mesofractures. Hence, the exposed parts of the northerly
striking macrofaults (set 1) developed after the exposed parts of the easterly macrofaults
(set 2).
Mesofractures and the evolution of the macrofaults

The stress regimes which caused the exposed macrofaults are similar to those which caused
the mesofaults (cf. Figs 14a & 13b-d). The 3D geometry of the exposed macrofaults is not
clear due to their limited outcrops. However, the geometrically parallel, spatially and
genetically related mesofractures (faults and joints) form zones which are better exposed
and can therefore be used to shed some light on the macrofaults and the early stages in
their evolution.
Styles of mesofaults and mesofracture zones. There are four styles of smaller faults and
fracture zones which constitute part of the systematic fractures (Fig. 16): (1) Discrete
discontinuities which occur as discrete fracture planes (Fig. 16a); (2) Zones of granulation
seams (deformation bands), as shown in Fig 4f; (3) Zones of closely spaced fractures which
are curved in 3D, i.e. have curved vertical and horizontal profiles and occur as 'duplexes'
enclosing blocks of rocks up to a few metres in length (Fig. 16b). The shape of these
fractures will necessitate ploughing and friction large enough to produce substantial
amounts of fault rocks at relatively small displacements of a few metres; (4) Localized
zones of closely spaced fractures of a systematic set. These often include two subsets (e.g.
2a and 2b), one of which tends to end at the other dominant subset. The junction zone
Fig. 14. (a) Lower-hemisphere, equal-area projection of poles to macrofaults [dextral (O); sinistral
(A)] in the St Bees Sandstone and their slickenlines (corresponding solid symbols) showing two main
sets striking northerly (~ set 1) and easterly (, set 2) and are the ineffective principal stresses (cf. Figs
3b & c and 13b-d). (b) Cataclasite and gouge (G) developed along the macrofault shown in (c) & (d).
(e & tl) A photograph and its interpretation of a sinistral-normal macrofault dipping at 70-85 ~
towards 7~ with fibrous haematite growth (slickenlines) dipping at 38~ towards 285 ~ and vertical
offset of 6m exposed on a cliff face at South Head (NX9513 1195).
128 M.S. AMEEN
between the two subsets is highly susceptible to fault rocks development (Fig. 16c); (5)
Flower structures (Fig. 16d & e).
These different styles can be considered as small-scale models of the macrofaults.
Furthermore, they can be used to understand the embryonic stages of the macrofaults.
Therefore, any model for the evolution and mutual relationship of the macrofaults should
take into consideration the field evidence discussed here.
Regional stresses and the evolution of the mesofractures and the macrofaults
This study shows that there is a consistent mesofracturing, macrofracturing and related
stress regime in the exposed cover rock masses. An early N-S tension, evident from set 2
fractures (meso- and macrofractures) is most probably linked to the regional N-S tension
(Late Jurassic to the Early Cretaceous). Later E-W and NE-SW tensions which caused
the set 1 fractures (meso- and macrofractures) are in agreement with the Palaeogene to
Neogene N-S to NW-SE compression and uplifting which resulted from the convergence
of Africa and Europe. In addition there exists another phase of E-W and NE-SW tension
(Permian to Early Jurassic), evident only from subsurface data (Chadwick et al. 1993a, b;
J. Gutmanis, pers. comm.) in the lower part of the St Bees Sandstones and in older rocks
(e.g. growth macrofaults parallel to the set 1 of the exposed macrofaults). It is also evident
in exposed Carboniferous rocks where the northerly-striking mesofractures pre-date the
easterly-striking fractures (e.g. at Saltom Bay, NX 95801593). Therefore, it can be
concluded that the upper part of the St Bees Formation and the Calder Sandstone in the
study area were mostly not affected by and/or post-date this early phase of the E-W
tension. However, the lower part of the St Bees Sandstones was unlithified and thus
adapted itself by differential compaction and consolidation, and a limited degree of syn-
depositional macrofaulting.
Therefore, there are three main phases of fracture development (including macro-
faulting) in the cover rocks, which resulted from the three post-Palaeozoic regional stress
regimes (outlined above): (1) Phase 1 (Permian to Early Jurassic), is the earliest phase
during which syn-depositional growth macrofaults (parallel to the set 1 of the exposed
macrofaults) occurred. This was mostly restricted to the lower part of the cover rocks and
was accompanied by little or no mesofracturing. Therefore, this phase is not evident in any
of the exposed macrofaults which are covered in this study. (2) Phase 2 (Late Jurassic to
the Early Cretaceous) occurred after the deposition of the cover rocks and led to the
occurrence of set 2 macrofaults and mesofractures, and is clearly evident from surface
exposures. (3) Phase 3 (Palaeogene to Neogene) is the latest phase during which the
exposed set 1 macrofaulting and mesofracturing took place. Some of these macrofaults
most probably propagated upwards from the tips of the pre-existing (phase 1) macrofaults.
Others resulted from the coalescence of the set 1 mesofractures and propagated down-
wards. Some of the movements on the macrofaults during this phase were most probably
also accommodated by the pre-existing set 2 macrofaults. The occurrence of a few
slickenline sets on some of the set 2 faults may support this suggestion (Fig. 6).
It is difficult to put a depth constraint on the evolution of the set 1 and 2 fractures. Fluid
inclusion analysis of the calcite flling may be helpful. Such analyses are at an early stage
and their results are not available yet (Andrew Bowden, pers. comm.). However, an upper
limit of depth equivalent to the maximum estimated depth of burial of the rocks can be
applied. Such a maximum depth is estimated 1400-2000 m (Chadwick et al. 1993a) and
can be invoked for set 2 fractures. However, at least some of the set 1 fractures
49S plane (trace) \ ~.~ J

- --- tr~ ..~ -495 --f~ 1-495 "/,~
295 t 3~5. 295 , ~5 k ~ / 295 , 3~,5 \
,,, i l V I , , ,
F~
Fig. 15. Stress maps of the study area based on the temporal and spatial evolution of the set 2
fractures (a--e) and set 1 fractures (d-f) in the cover rocks. For details see text.
(particularly subset 1b) are likely to have occurred at shallower depths due to the uplifting
which accompanied their causative tectonic stresses.
Although bedding fractures (set 3) seem to have been active throughout the tectonic
history of the region most of them post-date set 1 and 2 fractures. Later stress release
associated with uplifting, and removal of overburden, was probably the main contributing
factor in the development of bedding-parallel fractures. This is evident from the
continuation of fracture surface structures, e.g. hackle plumes, on individual fractures of
sets 1 and 2 across bedding fractures (Figs 8a & 12c), and the occurrence of faults which
cross-cut bedding fractures (Fig. 5b ).
130 M.S. AMEEN
Table 4. Classification of the two statistically dominant sets of fractures in the Borrowdale Volcanics
Group (Fig. 18a & b) according to their types (joints, veins etc.), local set number, location and
direction of propagation
Regional Corresponding local fractures

set type/local set no. (Locality), direction of propagation
Joints Veins Faults

Strike-slip Dip-slip Oblique-slip Unknown
1 2(14), 6(9), 6(9), 4(11)~=~, 2(12)TR 2(14)D-R 1(15)

3(9)~, 3(11), 2(15), 2(15), 3(10)S-R 2(10)
3, 4(9), 2(10), 2, 3, 4(13) 4(12)yD 3(15)$D-N 3(14)
2(11)~=~, 3(15)/'/,," 5(12)
3(14)
2 2(9) B.J., 1(11) 1(13) 5(9)S 2(12)1R 3(9)S-R 5(12)
1(12) B.J. 4(14) 1(13)S 1(13) 1(14)
2(9), 5(9)~ 8(9)T l(13)D 1(11)
1(10), 3(10)~ 6(9)TD
1(13), 1(14)
4(14), 6(9)
3(11)
Same as those used in the observations given in the Appendices (e.g. 1(9): Local set no. 1 in Loc.
No. 9).
Arrows indicate direction of propagation; ~, top to bottom; T, bottom to top; r h o r i z a o n t a l ; / / ,
oblique and top to bottom or bottom to top (if the direction is not known none of these symbols is
shown). B.J., Bedding joint; D, dextral fault; S, sinistral fault; R, reverse fault; D-R, dextral-reverse
fault; S-R, sinistral-reverse fault; D-N, dextral-normal fault (if sense of slip is not known none of
these symbols is shown).
Mesofracture patterns in the basement rocks

Several sets of fractures occur in each of the studied localities. The summary of the
fracture observations in individual localities is reproduced in Figs 17 & 18 and in the
Appendices. Individual sets generally show a wide range of dip and strike, and evidence of
multiphase rejuvenation. Therefore, it is difficult to recognize clearly defined regionally
prevalent fracture sets according to orientation, type and relative chronology. However,
Fig. 16. Schematic block diagrams and photographs illustrating the different styles of smaller faults
and fracture zones in St Bees Sandstone. (a) Discrete fractures, illustrated by a set 1, dextral fault
dipping at 80-90 ~ towards 53-81 ~ although the fault's length and height are large (> 70, 20m,
respectively) its displacement is c. 0.1 m (St Bees Head, NX9445 1333). ~) Zones of closely spaced
fractures which are curved in 3D. The photographs illustrate profile and bottom views of dextral-
normal fault zones consisting of such fractures of set 1 at St Bees Head with width, height and length
of 1, > 20 and >30 m, respectively. The dip angle varies from 60 to 80~ and dip direction varies from
76 to 92 ~ The net slip on this zone is very small (c. 1 cm) and therefore there are no fault rocks. (e)
Localized zones of closely spaced systematic fractures. The photograph shows such a zone developed
between subsets 2a and 2b. Figure 5b shows the highly fractured profile section of the junction zone.
(d) Negative flower structures (transtensional faults), illustrated by a set 1 fault with a width ranging
from 2 m at the bottom of the cliff to a minimum of 12 m at the top of the cliff, and a minimum
length of 40 m. (e) Negative flower structure (transtensional fault), illustrated by a set 1 fault dipping
at 78~ towards 20 ~ with slickenlines dipping at 25~ towards 300~ associated with antithetic faults
dipping at 85~ towards the west (St Bees Bay).
132 M.S. AMEEN
~0,,,,,,
"0
0
c5
,d (D
0
m ~
.,,.,a 0.,)
.,,.a,
,d
~,,,,,,,
z rn
r.~
,,.2.
,L"
o~
t.:'-
"* 6
synoptic pole diagrams indicate the existence of two 'geometrical' sets in the Borrowdale
Volcanics. These are generally steep dipping (> 45~ striking NE-SW to ESE-WNW and
NW-SE to NNE-SSW, respectively (Fig. 18a & b). As shown in Table 4, each of the two
regional geometrical sets includes a wide spectrum of fractures and has no consistent type,
origin or fractographic features.
In the plutonic rocks, synoptic pole diagrams of the fractures are characterized by more
diffused data which may be vaguely divided into five geometrical sets. These include one
nearly horizontal and four steep dipping sets striking NW, NE, NNE and WNW,
respectively (Fig. 17c & d).
Surface features of the mesofractures in the basement

The surface features for the individual sets in each locality are summarized in the
Appendices. These include both the features described in Fig. 2 and features related to
shear movement (e.g. slickenlines). The latter vary from one set to another and generally
include fibrous mineral growths and frictional wear. No clear evidence of pressure solution
was noticed in the field on any of the fractures. Fractographic features are often
obliterated or masked by slickenlines, mineral filling and coating, and are described below.
Fracture traces
The exposed traces of the fractures are either discrete discontinuities (Fig. 19a) or zones of
closely spaced discontinuities. Fracture zones vary in width from a fraction of a centimetre
to a few tens of metres (Fig. 19b--d) and are made up of parallel fractures with or without
a consistent sense of overlapping. Some zones are defined by en echelon tension veins
(Fig. 20).
The different fractures in each locality may show consistent or varied cutting/abutting
relationships. This seems to depend on the presence or the absence of fillings or seals and
the number of phases of deformation (see Appendices). However, on a regional scale, it
is difficult to find a clearly defined, consistent cutting/abutting relationship between the
regional geometrical sets. Examples of relatively well established cutting/abutting
relationships between local sets of fractures are given in the next section.
Surface patterns (structures)

These mainly include hackle plumes and twist hackles. The pitch of the plume axes varies
from 0 to 90 ~ (Fig. 21 and Appendices). Some individual sets show both horizontal and
steeply dipping plume axes (Table 4).
Chronology and dynamic implications of the mesofractures in the basement

As mentioned above, it is difficult to define regionally consistent sets of fractures in the
basement rocks with clear relative chronology and origin. Instead, local sets are commonly
observed that show no clear evidence for origin and chronology. The fractures in the
plutonic rocks are the least defined due to their complex pattern, and/or the relatively
small number of data.
In the Borrowdale Volcanic Group each of the two statistically dominant geometrical
134 M.S. AMEEN
Fig. 18. Synoptic pole and great circle diagrams of all the fractures in: (a & b) the Borrowdale
Volcanics; (c & d) the Eskdale Granite, Granodiorite and the Ennerdale Granite (e & f) all the
studied basement rocks shown in (a-d)
FRACTOGRAPHY AND FRACTURE CHARACTERIZATION 13 5
Fig. 19. Typical examples of fracture traces in the Basement: (a) Borrowdale Volcanics; feldspar-
phyric andesite, at Wasdale (Water Crag, Loc. 13, NY1553 0610); (b & d) Eskdale Granite, at
Hooker Crag (Loc. 17, SD1120 9830); (e) Ennerdale Granophyre, at Wasdale (Loc. 18, NY 1510
0543). Hammer shaft is 0.3 m in length and the coin is 2 cm in diameter. See text for details.
sets (Fig. 18) consists of several overlapping local sets of diverse types of fractures (e.g.
dextral, sinistral, reverse and normal faults). These fractures do not show consistent local
or regional dynamic implications. For example, some fractures within set 1 (Table 4) have
propagated from top to bottom, or vice versa, others horizontally. Such a difference
reflects differences in the orientation and/or relative amount of the causative effective
principal stresses. In addition, an individual fracture zone may show evidence of
136 M.S. AMEEN
Fig. 20. A nearly-vertical fracture zone in feldspar-phyric andesite of the Borrowdale Volcanics at
Wasdale (Water Crag, Loc. 13, NY1553 0610) defined by en echelon tension veins and showing
evidence of multiphase deformation: (a) early phase dextral movement; (b) late phase sinistral
movement; (e) a close-up view of the bottom vein in (b). The coin is 2.8 cm in diameter. For details
see text.
multiphase movements. An excellent example is shown in Fig. 20 where a fracture zone

with a N - S trend consists of sigmoidal tension veins with an en echelon arrangement,
which indicates that the zone was originally developed as a dextral shear zone. However, a
later sinistral fault parallel to, and along, the fracture zone cuts and displaces the veins
(Fig. 20). Further movement phases are indicated by fibrous quartz filling along the fault.
A minimum of four phases of fibrous growth can be recognized in the field. Earlier stages
are all perpendicular to the fault walls indicating extension. The latest stage is oblique to
the fault walls indicating oblique (dip-slip and dilational movement) on the fault.
The fracturing history in the Borrowdale Volcanic Group is complex. Fracture initiation
in the the group has most likely occurred since the emplacement of the volcanic rocks, due
to cooling (e.g. columnar joints), volcano-tectonics (caldera-collapse; Branney & Soper
1988), and later tectonic movements. Thus, a purely tectonic origin for all the fractures
would be an oversimplification. The release of tectonic strain would have occurred at least
partly by rejuvenating the suitably oriented pre-existing discontinuities (including
mineralization zones) irrespective of their origin.
The relationship of the mesofractures and macrofaults in the basement rocks

No macrofaults were observed in the plutonic rocks. This is most probably related to the
limited exposure and small number of localities covered in the study. Many macrofaults
occur in the Borrowdale Volcanic Group (Fig. 22). These include two sets which strike
NNW and ENE, respectively, marked by gouges and cataclasites. In addition, most of the
macrofaults are parallel or sub-parallel to immediately adjacent mesofractures. However,
the frequency of the latter does not always seem to be affected by the presence of
macrofaults. Multiphase movement on individual macrofaults is evident from the
occurrence of a few sets of slickenlines, as shown in Fig. 22.
Conclusions
The present study shows a clear contrast between the fracture characterization of the
Lower Palaeozoic basement and the Permo-Triassic cover rocks. The latter is charac-
terized by clearly defined regional systematic sets with consistent relative chronology,
whereas the former shows a complex fracture pattern with no identifiable consistent
chronology and origin. Such a contrast is the result of the long and complicated history of
the basement rocks compared to that of the cover rocks.
The cover rocks were systematically fractured during burial, uplift and unloading, in
addition to tectonic deformation. Three distinctive phases of fracture development are
recognized: (1) Phase 1 (Permian to Early Jurassic) is the earliest phase during which
syn-depositional growth macrofaults with northerly strikes occurred. This was mostly
restricted to the lower part of the cover rocks and was accompanied by little or no
mesofracturing. Therefore, this phase is not evident in any of the exposed macrofaults
which are covered in this study. This phase resulted from an E-W tension which
dominated NW Europe and caused the development of the East Irish Sea and the Solway
Basins as extensional/transtensional basins. (2) Phase 2 (Late Jurassic to the Early
Cretaceous) occurred after the deposition of the cover rocks and led to the occurrence of
the easterly striking (set 2) macrofaults and mesofractures and is clearly evident from
surface exposures. They resulted from burial stresses (vertical al) and tectonic stresses
characterized by E-W, horizontal to moderately, westerly inclined O"1 and N-S horizontal
a3. (3) Phase 3 (Palaeogene to Neogene) is the latest phase during which the exposed
northerly striking (set 1) macrofaulting and mesofracturing took place. Some of these
macrofaults most probably propagated upwards from the tips of the pre-existing (phase 1)
northerly, macrofaults. Others resulted from the coalescence of the set 1 mesofractures and
propagated downwards. Some of the movements on the macrofaults during this phase
138 M.S. AMEEN
Fig. 21. Fracture surface patterns and their interpretation in the Borrowdale Volcanics. For details
see text. Long arrow indicates the direction of propagation of the main fracture, small arrows
indicate the direction of propagation of twist faces. Stress directions are for the leading edge of the
fracture near its origin. Hammer shaft and pencil are 0.3 m and 14cm in length, respectively.
were most probably also accommodated by the pre-existing set 2 macrofaults. The
occurrence of a few slickenline sets on some of the set 2 faults may support this suggestion.
This phase resulted from N - S to N W - S E compression related to the convergence of
Africa and Europe which was culminated in the Oligocene to the Miocene (NNW-SSE
horizontal to gently, northerly dipping O"1 and E N E - W S W horizontal era) and a regional
flexural uplifting (vertical crl) associated with the onset of seafloor spreading between
Scotland and Greenland.
The fractographic features on both sets of mesofractures (sets 1 and 2) suggest that they
occurred as Mode I (tensile) fractures and were later modified by shear or mixed-mode
(tensile-shear) loading. The geometry and degree of development of fracture surface
morphology are controlled by sedimentary layering, lithological heterogeneity, the
Fig. 22. Lower-hemisphere, equal-area projection of poles to macrofaults in the Borrowdale

Volcanics (O), and their slickenlines (O) showing two main sets striking northerly (set 1) and
easterly (set 2). Note that slickenlines 1-3 occur on the same fault (F).
presence of open fractures, the level of pore-fluid pressure, the orientation and relative
amounts of the principal stresses, and their temporal and spatial variations across the
fractured horizon. The conventionally accepted control of sedimentary layering on the size
and shape of fractures and their fractographic features is found to be equivocal, unless the
layering forms free surfaces at the time of fracturing. The normal stresses were
considerably higher than the shear stresses during the development of set 2 and subset
1a fractures. However, the ratio of the normal to the shear stresses was considerably lower
during the development of the subset lb fractures. This difference is probably linked to the
fact that the l b fractures were the youngest systematic fractures, which could have
developed at shallower depths than set 2 and l a fractures.
It is difficult to recognize any consistent regional phases of deformation from the
fractures in the basement. A combination of weakness zones [e.g. mineralization bands
in plutonic rocks, flow surfaces, cooling shrinkage cracks (columnar joints), bedding (in
tufts) and volcano-tectonic discontinuities (in volcanic rocks)] are most likely to have
contributed to the wide spectrum of fracture orientations. In addition, during each phase
of tectonic deformation some of the pre-existing fractures are likely to have adjusted to
dissipate any suitably orientated stresses. Furthermore, new fractures were most likely
initiated during ailphases of tectonism where pre-existing anisotropies were locally absent
or ineffective. The most important feature to fracture characterization of the basement and
the cover rocks is the abundance of dilational movements and vein development and the
lack of clear evidence of pressure solution from field evidence.
Although the exposed macrofaults in the cover rocks are parallel, or nearly parallel, to
those occurring in the basement, they are not necessarily related. Those in the basement do
not show a clear and consistent sense of slip like those in the cover rocks. The majority of
faults in the basement are most probably older than those in the cover rocks (e.g. of
volcano-tectonic origin), and therefore suffered a complex movement history. However,
140 M.S. AMEEN
the macrofaults exposed in the cover rocks are probably the result of direct physical
propagation from those in the basement and/or they are merely controlled by the strain
pattern induced into the cover rocks by movements on the basement faults (Ameen 1988,
1990, 1992). The existence of the evaporite units in the Permo-Triassic, though patchy
(Chadwick et al. 1993b), may limit the propagation of macrofaults from the basement into
the cover rocks and hinder the connectivity of faults in the two units in some parts of the
study area. The 3D geometry and evolution of the small-scale faults and fracture zones in
the cover rocks, described here, should be used to constrain any model for the evolution of
macro faults.
The author is grateful to Terry Engelder, Jon Gutmanis and Byron Kulander for their review and
comments which helped to improve an earlier version of this paper. This paper is partly based on
work carried out on behalf of UK Nirex Ltd as a part of their ongoing geological investigation into
the suitability of the Sellafield area for a nuclear waste repository. Thanks are due to UK Nirex Ltd
and GeoScience Ltd for their permission to publish this paper. All interpretations and conclusions
presented here are solely the author's; they do not necessarily reflect the ideas of UK Nirex Ltd or
Geoscience Ltd.
References
AMEEN, M. S. 1988. Folding of layered cover due to dip-slip basement faulting. PhD Thesis, Imperial
College, London, UK.
- - 1990. Macrofaulting in the Purbeck-Isle of Wight monocline. Proceedings of the Geologists'
Association, 101, 31-46.
- - 1992. Strain pattern in the Purbeck-Isle of Wight monocline: A case study of folding due to
dip-slip fault in the basement. In: BARTHOLOMEW,M. J., HYNDMAN,D. W., MOGK, D. W. &
MASON, R. (eds) Basement tectonics 8: characterization and comparison of ancient and Mesozoic
continental margins (Proceedings of the 8th International Conference on Basement Tectonics).
Kluwer Academic Publishers, Dordrecht, The Netherlands, 559-578.
AYDIN, A. & JOHNSON,A. M. 1978. Development of faults as zones of deformation bands and as slip
surfaces in sandstone. Pure and Applied Geophysics, 116, 931-942.
BEACH, A. 1975. The geometry of en echelon vein arrays. Tectonophysics, 28, 245-264.
BRANNEY, M. J. & lOPER, N. J. 1988. Ordovician volcanotectonics in the English Lake District.
Journal of the Geological Society, 145, 367-376.
CHADWICK, R. A., BAILY,H. E., KIRBY, G. A. & ROWLEY,W. J. 1993a. Jurassic to Neogene structural
evolution of the Sellafield area. UK Nirex Report No. 518, UK Nirex Ltd, UK.
--, EVANS,D. J. & HOLLIDAY,D. J. 1993b. The Maryport fault: the post-Caledonian tectonic
history of southern Britain in microcosm. Journal of the Geological Society, 150, 247-250.
CRUIKSHANK,K. M., ZHOA,G. t~ JOHNSON,A. M. 1991. Analysis of minor fractures associated with
joints and faulted joints. Journal of the Structural Geology, 13, 865-886.
ENGELDER, T. & LACAZETTE, A. 1990. Natural hydraulic fracturing. In: BARTON, N. &
STEPHANSSON,O. (eds) Rock Joints. A.A. Balkema, Rotterdam, 35-44.
JACKSON, D. I., MULHOLLAND,P., JONES, S. M. & WARRINGTON,G. 1987. The geological framework
of the East Irish Sea Basin. In : BROOKS,J. 8s GLENNIE,K. (eds) Petroleum Geology of North West
Europe. Graham & Trotman, London, 191-203.
JOHNSON, M. R. W., SANDERSON,D. J. & lOPER, N. J. 1979. Deformation of the Caledonides of
Ireland and Scotland. In: HARRIS,A. L., HOLLAND,C. H. & LEAKE,B. E. (eds) The Caledonides
of the British Isles. Geological Society, London, Special Publication, 8, 165-186.
KULA~ER, B. R., BARTON,C. C. & DEAN, S. L. 1979. The application offractography to core and
outcrop fracture investigation. US Department of Commerce Technical Report METC/SP-79/3,
National Technical Information Service, Springfield, Virginia, USA.
--, DEAN, S. L. & WARD, B. J. 1990. Fractured Core Analysis: Interpretation, Logging, and Use
of Natural and Induced Fractures in Core: Methods in Exploration Series, No. 8. American
Association of Petroleum Geologists, Tulsa, Oklahoma, USA.
LAWN, B. R. & WlLSHAW, T. R. 1975. Fracture of Brittle Solids. Cambridge University Press,
"6"6
"8"8 "~ ~
ZZ ;~ ZZ
~ ~~
9 "~
~oo~ ~
~~ .~
~0
~=~'" ~0 ~
9 0~ ~ ~.~ o
~
9 ZOa=o .~ o ~ ' ~ ~ ~.~ ~ ~ ~ e ~,~
I 9
t'q cq
AA~ Ac~ A ~A A c,~A
A A~ Ao A ~A A ~A
q,
"2. "2. o. m.
r~
~5
I I I I tr~ i ,.--, A~ I I
" '~ =o
r~. I I
7,
kid tt'~ tr ,,~
.a ,,...r
.,..,
9...-~ c q
8
Ig
,...,
142 M.S. AMEEN
z ~ ~ ~"~'~ ~" ~'
9, ' ~ o ~ ~ ' ~
"~ ~ ~ ~"~ ~" . ~ I~.~ "~
.~ O
A A r A r A A
A A A r A A
o
o
A~ A~ A~ A~
o o
8
r
~ o~ 8
z
.N
~.~~ ~ ~
~':_ ~, ~ , . o ~ ' ~ ~ ~ ~8.~ ~ ~ .~
~9 ~ ~.~ --
~ = o ~ ' ~
'-6 ~
9 ~"o ~, ~ ;= o ' ~ ; B ~ --o~ ~
[..) ~,~, . ~ ~ -,_~,.e " o " ~ e,i " 0 z
~ ~ ~ ~
a o ~
"0
~2 N~
~:~ 0 ~'~ ~-~ 0
~ ~
~ v/
A A ~ A A A
~,~ c~ c~ A cS A A A
~
-- . -- o.o o.
cS c} c~ oo o
~:, ~1 V/ AX I AX ~..
-....
~o
~z
{'B
144 M. S. A M E E N
t",l ~ t ~ . O O ~ ,,,-~
~o
N
,.,=
8~-,~ 9
~o~=~
~-~~ ~ ~
~ ~ ~...&~ o
r
~.~ .o o o
6
s
N-- N
~.~ ~
d
a ~ ~f~'-
r~
V/ V/
A A
I I
o o
..8
r.u~,
d~
~-~
~ o ~ o
.,,,, ~E
a s
"--" O
r -,..., I~
~t
&
FRACTOGRAPHY
ANDFRACTURECHARACTERIZATION 145
I= o l=
o ~0" o
o Zg
..~ .~o
o o o o ~ ~ ~ ~"o~ ~
o ~ ~ ~o ~ ~ o~ o ~ o'0 ~ . ~ .,-,
9 O ~ ~ o "6
9,-, i~ ~.-~ ~
~ ~ ~ ~ 9
io N~
v/ "-- ~ ~-
eel ~ r
~,_., A A r A A A A A
,.el
A o A A ,~ A A
~1 ~
~.~ v/ ~ v/ o VlVl o o c~
c.-. I A~ e~.e~. I I V~
C~
o o
O~ 0 g :o
o I ~J
oo
~2
8
b 8 Z,
~Z
:s O~ .~
0
146 M.S. AMEEN
~ - ~ ~ .~ ',~
~ ~ ~;~ ~.~ !
d
O
~ , 8 ~ .~ # ~ ~
o N
O
N1
~.~ ~
-8o8
~ o .~ ~ o o
.~ I~
~'~ ~
d 6
o ~
d
e.=l ~
"1=
t-q t-q
c~ A A A
,~ A A A
~. .o
~, 9
m "2.
V/ V/ c~
A~ A~
o ~
~9 o ~
%}
~8
Cambridge, UK.
MOSELEY, F. 1990. The Lake District. The Geologists' Association, London.
POLLARD, D. O., SEGALL, P. • DELANEY, P. T. 1982. Formation and interpretation of dilatant en
echelon cracks. Geological Society of America Bulletin, 93, 1291-1303.
RAMSAY,J. G. & HUBER, M. I. 1987. The Techniques of Modern Structural Geology. Academic Press,
London.
TURNER, F. J. d~ WEISS, L. E. 1963. Structural Analysis of Metamorphic Tectonites. McGraw-Hill,
New York.
SMITH, R. A. 1974. A bibliography of the geology and geomorphology of Cumbria. Cumberland
Geological Society, 1-32.
WOODWORTH, J. E. 1896. On the fracture system of joints, with remarks on certain great fractures.
Fracture characterization in the Chalk and the evolution

of the Thanet monocline, Kent, southern England
M O H A M M E D S. A M E E N
Independent Consultant, 122 Tachbrook Street, Pimlico,
London S W 1 V 2ND, UK
Abstract: Systematic studies of fractures in the exposed Upper Cretaceous Chalk and
from seismic data are used to establish the evolution of the regional structure and the
related fractures on the Isle of Thanet, Kent. The results show that the main structure
developed as an easterly trending, south facing, monoclinal forced fold with a WNW-
ESE trending cross forced fold along a basement grain (Pegwell Bay-Minnis Bay
transtension/transpression zone). The folding started during the deposition of the
Chalk in the Upper Cretaceous due to an extensional tectonic phase accompanied by
differential sedimentation and compaction [vertical (0.1) and horizontal (0-3)] which
induced vertical and steep macrofractures (mainly joints and some extensional faults) in
the Chalk. This was followed late in the Upper Cretaceous by a horizontal, N-S
compression (0-1)which continued in the Tertiary and caused inversion of movement on
the basement faults and the occurrence of the final stages in the forced folding. The
latter phase led to the rejuvenation of the pre-existing fractures in the cover (mainly by
faulting) and the development of new fractures (mostly faults and joints). An uplifting/
unloading phase started late in the deformation history and caused the occurrence of
near horizontal (bedding-parallel) fractures (joints) and bedding-perpendicular or steep
mesofractures parallel to, and perpendicular to, the burial/tectonic macrofractures.
These have particularly developed at the top several metres of the exposed Chalk where
a rapid increase in frequency of fractures is clear.
The fracture characterization of the Chalk varies significantly in different parts of the
monocline. Based on such variation a minimum of four domains (zones) have been
identified. The heterogeneity in the fracture development is most probably controlled by
the variation in orientation and type of the causative basement faults and the amount
and type of displacement on t~em.
The study area covers the Isle of Thanet in Kent, an E - W trending peninsula protruding
into the southern part of the N o r t h Sea. The peninsula constitutes the upthrown gentle
limb and the outer anticlinal hinge zone of a south-facing monoclinal structure in the
Upper Cretaceous to Tertiary sedimentary rocks, referred to here as the Thanet monocline
(Fig. 1). Although the presence of the monocline is evident from a structure contour map
published by the Geological Survey of Great Britain (1980), the fold was not described
until 1988 when a very brief account of the structure of the country around Ramsgate and
Dover was given by Shephard-Thorn (1988), who called the fold 'Thanet anticline'.
There have been no detailed systematic fracture studies in the monocline. Cawsey (1977)
gave a brief account of the fracture sets in Kent, including the Isle of Thanet (according
to their strikes only), and considered all fractures as joints. Shephard-Thorn (1988)
schematically indicated the traces of some of the macrofaults on coastal cliff sections of the
Upper Chalk, and called them tension faults associated with the 'Thanet anticline';
however, neither their orientation nor their chronology or origin was mentioned.
150 M.S. AMEEN
Boundary of
Contour values in the Isle of Thanet are based largely
transpression zone
upon the distribution of the M a r s u p i t e s Zone of the Upper
Fault (ticks indicate
Chalk (approx. 225m above the base of the Chalk). _ _ _ t _ _ - - ' d o w n t h r o w n side)
Contour dravm on
---200 b a s e o f c h a l k (at
a n interval o f 10m)
- 70
Seismic Section
locati
- 60
(b) 30 40
A Pegwell-Minnis Bay ~ ~ . At
Ebbsfleet (Pegwell Transpression Zone _ oane rumpmg .... ,~
tBorehole~ B ~.,~ I I Station Well (Margate) r~ortnaown
Sea~ ~; i ~ - ' ~ - ~ - ~ Sea[- ~u
Level -' . . . . . " -- ~ - Leve u
-
I ---~---
......\ % r
J": ~.. . ; . ~ : : - : : : : - : : ~ - - ~ : ~ ~
-1oo
-200
~K~(~9". '~a~ : P r e c a m b n a n - L . Palaeoz01c Basemet~t . . . . m
(C) Horizontal Scale

CHALK FRACTURE CHARACTERIZATION, THANET 151
Fig. 1. (a) Location (inset) and sub-Quaternary (solid) geology map; (b) reinterpreted structure
contour map on the base of the Chalk Group; (e) reinterpreted geological section along 'A-A' in (d)
for the Isle of Thanet (modified from Shephard-Thom 1988); (d) general stratigraphic column.
152 M.S. AMEEN
The present study discusses the evolution of the Thanet monocline and the associated
fracture characterization in the Upper Chalk. The term 'fracture' is used here in the broad
sense defined by Kulander et al. (1990) as a break or physical discontinuity in a rock
caused by stresses exceeding the rock's strength. Therefore, it covers faults, joints, veins
and stylolites. According to their scale the fractures covered in the present study include
mesofractures and macrofractures. Mesofractures can be effectively observed in 3D with
or without a hand lens in hand specimens or large continuous exposures. Macrofractures
are too large or too poorly exposed to be examined directly in their entirety in one
exposure. Fractography is defined here as the discipline of diagnosing, describing and
interpreting the fracture surface morphology or topography (fractographic features),
which are related to the initiation, propagation and subsequent history of the fractures.
Details concerning the definitions of fractures and fractographic features (e.g. origins,
hackle plumes, hackle faces, hackle steps and rib marks) and their interpretation and
methods used in this study are the same as those outlined in Ameen 1995.
A total of 431 fractures (faults and joints) was studied in 11 field localities, mostly on the
coastal parts of the Isle of Thanet. In addition, an offshore seismic section is interpreted
and used to study the profile of the regional structure and understand its origin and
development, in particular the basement/cover interaction.
Regional structure
The Thanet monocline
The Thanet monocline is a relatively gentle monoclinal fold, facing south to southwest.
The steepest exposed part of the monocline (the outer anticlinal hinge zone) occurs in
Pegwell Bay and dips at 10~ to the south-southwest. The fold has a relatively small
amplitude (c. 50 m, measured on the base of the Chalk) and its axial trace strikes at an
E-W direction along most of its length onshore; however, it swings to a N W - S E direction
at Pegwell Bay (Fig. 1a, b & c). The onshore length of the monocline is c. 20 km; however,
the structure is expected to continue offshore for a minimum of a few kilometres to the
southeast of Pegwell Bay. The fold affects the relatively thin sedimentary cover (a few
hundred metres to 1 km in thickness) which is dominated by Upper Cretaceous and
Tertiary rocks (Fig. l a, c & d). The underlying basement rocks are of Lower Palaeozoic
and Precambrian age, and form the southern part of the London-Brabant Massif (defined
below).
The monoclinal flexure evolved as a result of the differential compaction 'draping' of the
Upper Cretaceous to Tertiary sediments over the topographic high of the London-
Brabant Massif and the late Cretaceous to Pliocene basement-block faulting. The origin of
the monocline will be discussed further in a later section.
The London-Brabant Massif

This covers a wedge-shaped area with an E - W to N W - S E trend, and extends from east
England (where it has its maximum width of nearly 200 km) across the southernmost part
of the North Sea into Belgium, where it is considerably narrower. The massif has a length
of c. 550 km. It is bounded by faults which separate it from surrounding Mesozoic basins
(Fig. 2). Geophysical evidence suggests the presence of Precambrian cratonic rocks
beneath the London-Brabant Massif. The massif, which received some Lower Palaeozoic
Fig. 2. Regional tectonic framework of southern England (based on Hamblin et al. 1992 and
Cameron et al. 1992) showing the location of the London-Brabant Massif.
sediments, became established as a relatively stable upland area early in the Dinantian and
formed the southern boundary of the southern North Sea Basin and the northern
boundary of the Weald Basin until mid-Cretaceous times. The southern boundary of the
massif, which is close to the study area, is defined by a Variscan thrust and related
transpression zones (Variscan Front in Fig. 2). The massif received very little and patchy
sediments during the Upper Palaeozoic to mid-Cretaceous. This is reflected by the shallow
depth to the top of the Lower Palaeozoic (c. 1 km), and to the top of Carboniferous rocks
(Cameron et al. 1992, figs 11 & 13).
In the Cretaceous the London-Brabant Massif was submerged for the first time since the
early Palaeozoic. This led to the deposition of the Wealden, Greensand, Gault and the
Chalk Group (Lower, Middle and Upper Chalk). However, the thickness of these rocks
over the London-Brabant Massif is considerably thinner than that in the North Sea Basin
and the Weald Basin (Cameron et al. 1992, fig. 82). The submergence of the massif is
probably due to a combination of a gentle (thermal?) crustal downwarping and a
contemporary global rise in sea level, and was accompanied by basement-block faulting.
During the Tertiary both shallow marine and continental deposits accumulated over the
154 M.S. AMEEN
Fig. 3. (a) A map outlining the 4 different domains (zones) of the Thanet Monocline and showing the
distribution of the studied localities (a-i); ~) Lower-hemisphere, equal-area stereographic projection
of the poles to the fractures and the slickenlines in the different localities (a-i).
London-Brabant Massif. The contact between the Upper Cretaceous Chalk and the
Tertiary rocks is marked by a time gap and erosional surface (Fig. l d). It is arguable
whether the surface is the result of subaerial or shallow marine erosion.
Fracture characterization of the Upper Chalk

The fracture patterns in the exposed parts of the Thanet monocline can be divided
(according to their host structural habitat) into two main groups, those in the upthrown
nearly horizontal limb of the monocline and those in the outer anticlinal hinge zone of the
monocline.
Fractures in the upthrown limb of the monocline

Three different zones (domains) have been recognized in this part of the monocline,
according to their fracture patterns. They occur in the northeastern, northwestern and
western parts of the upthrown limb of the monocline, respectively, referred to here as
zones 1, 2 and 3 (Fig. 3), respectively, and are described below.
Fractures in the northeastern part of the limb (zone 1). These fractures occur in the area
east of Margate (Walpole Bay, Palm Bay, Botany Bay and Kingsgate Bay). They include
two sets, which are nearly orthogonal, and dip steeply (>~ 70 ~ and strike at ESE to E - W
and NNE to N-S respectively. The former set is the most dominant. The fractures are
mostly joints - faults are rare and include extensional faults only (Fig. 3).
The joints include two distinctive groups according to their dimensions, fractography
and their relationship to the bedding. These are referred to as the macro- and mesojoints.
The macrojoints cut across the coastal cliff faces from top to bottom and continue across
bedding parallel joints, except some hard ground and marl horizons (Fig. 4). They have a
,,1
Bay
. o1 - ~ o ***/
**
\; 9 *' 2/
0 Juints
Joints with Flint
O Fault with Flint
A 9 Dextral/Normal-Dextral Fault" & Slickenlines
D 9 S i n s t r a l / N o r m a I - S i n s t r a l Fault" & Slickenlines
"~ ~ Normal Fanlt'" & Slickenlines
"~ Fanlt with an Extensional-Element (no slickenlines)
"'(Faults with multiphase movement are shown with a
symbnl that indicates the oldest movement: examples of
such fmdts and their slickenlines are indicated with
numbers)
156 M.S. AMEEN
Fig. 4. Easterly striking macrojoints affecting the nearly-horizontal beds exposed in Walpole Bay (a
& b) and Kingsgate Bay (c & d) and a northerly striking macrojoint exposed in Palm Bay (e). For
location see Fig. 3.
CHALK FRACTURE CHARACTERIZATION,THANET 157
Fig. 5. Fractographic features on: (aid) easterly striking macrojoints, indicating dominantly
horizontal, overall propagation (a--e) and less common vertical propagation (d); and (e) on a
northerly striking macrojoint indicating vertical, top to bottom propagation direction (a, in Walpole
Bay, b-e, Botany Bay). The 3D interpretation of the fracture shown in (c) is illustrated in the line
drawing to the right. For location see Fig. 3. Long arrow indicates direction of propagation of the
leading edge of the joints. Small arrows are local direction of propagation. Stresses are for the
leading edge of the joint near its origin.
158 M.S. AMEEN
minimum height of several to a few tens of metres (about cliff height) and trace lengths, on
bedding surfaces, of tens of metres (up to 50 m). They are mostly tight or slightly open (up
to a few millimetres). Surface structures of these joints are characterized by origins, hackle
plumes, hackle faces, hackle steps and rib marks. Rib marks are rare and twist angles are
very small (_<5~ The surface features indicate an overall horizontal direction
of propagation of most of the easterly striking joints, some show evidence of vertical
propagation from top to bottom (Fig. 5).
The main differences between the easterly and northerly trending macrojoints are the
apparent greater frequency of the former (see Fig. 3 and cf. Fig. 4b & d with e), and the
tendency of the northerly set to abut against the easterly set and to propagate from top to
bottom (Fig. 5d).
The mesojoints of both N-S and E-W striking sets are constrained by adjacent bedding-
parallel joints and tend to originate at such joints. In addition, the N-S joints are also
constrained by the E-W set. Hackle plumes and rib marks on the mesojoints indicate
horizontal or vertical propagation from top to bottom or vice versa (Fig. 6).
Fractures & the northwestern part of the limb (zone 2). These occur in the areas extending
from Margate westwards to Minnis Bay (Westgate Bay and Epple Bay) and include two
systematic, regionally persistent, vertical or steeply dipping sets, striking NNW-SSE to
NW-SE and N-S, respectively, and a third non-systematic set of fractures which show
flint fillings (Fig. 7a). The northwesterly set is the most prominent in this zone (Fig. 3).
As for the eastern part of the limb, the systematic fractures are dominated by macro-
and mesojoints with relatively rare extensional faults. The macrojoints occur as discrete
discontinuities (Fig. 7b & c), or discontinuity zones (Fig 7d) which cut across the bedding-
parallel joints (except some hardgrounds and marl horizons) from top to bottom of the
cliff faces. Most of the fractures are tight or slightly open (_< 1 mm) with no filling;
however, iron oxide staining is common on the NNW-SSE to NW-SE set. (Fig. 7c). The
exact dimension of each joint is difficult to determine due to the limited exposure; however,
the exposed lengths of the macrofractures is up to 20 m and their heights are equal to or
greater than those of the cliff faces, i.e. several to tens of metres (Fig 7c). Fractographic
features on these surfaces indicate that they mostly originate within a bed in the middle or
upper part, at a void, fossil or flint nodule. Hackle plumes, twist hackles and gradual twist-
hackle fringes occur and indicate an overall horizontal propagation of most joints. The
faults are dip-slip, extensional, and belong to the northwesterly striking set. Early
diagenetic, compaction joints and faults are mostly flint-filled and non-systematic, i.e. vary
in attitude (e.g. Westgate Bay, East, Fig. 3).
The mesojoints are parallel to the macrofractures and show similar fractographic
features and fracture/bedding relationship to those for the mesofractures in the
northeastern part of the limb (zone 1, described above).
Fracture patterns in the southeastern part of the limb (zone 3). This zone covers the part
east of the Pegwell-Minnis Bay Transpression Zone and south of Joss Bay and is
characterized by steeply dipping fractures, mostly macrofaults, striking NNE-SSW to
NNW-SSE, E-W to ESE-WNW and NW-SE, respectively (Fig. 3). The northerly
fractures are very rare. The fractures in this zone are well exposed on the coastal cliffs
between Joss Bay and Ramsgate, particularly in the Broadstairs area (North Cliff and
Dumpton Gap to East Cliff). The dimensions of the macrofractures are similar to those in
other parts of the monocline (Fig. 8).
O0 9
0.0 r
~..~
~ ~ .~
.= o ~ ~
" ~ .=
"~ ~ 0 ~
~ ~ 0
i ~~
~.r~ .,-~
"~-~ ~ ~ o
0 o,-~
r~ ~ 0
,~ 0 ~.
o "
~ ~0 -
~ ,-~ "'-~ .~
o ~ E ~ .o
o ~ o 9 ~ o
9 ~ ;~1~ o
160 M.S. AMEEN
Fig. 7. Macrofractures exposed in zone 2. (a) Flint-filled, early diagenetic fault zone (F), Westgate
Bay. (b) Traces of N-S (f 1) and N W - S E (f2) fractures on bedding-parallel wave cut platform, Epple
Bay. (e) Steeply dipping N W - S E macrojoints (J) stained with iron oxide, Westgate Bay. (d) A N W -
SE fracture zone (f) dipping at 80 ~ towards 230 ~ defined by smaller fractures (sf) dipping at 60-70 ~
towards 32 ~ Epple Bay. For locations of (a)-(d) see Fig. 3.
The faults are mostly strike-slip, although some show multi-phase movements with
strike-slip, dip-slip and oblique-slip components as evident from the occurrence of a few
slickenline sets on some faults (Figs 8 & 9). The slip indicators are pressure solution and
frictional wear grooves. It is difficult to analyse the slip history, i.e. chronology and
amount of each movement phase. However, it is possible to determine the net-slip on the
faults which mostly have an extensional component. Fault rocks (cataclasites and gouges)
have evolved to various degrees on the faults and tend to vary in thickness along
individual faults (Fig. 10a & b). The cross-cutting relationship indicates that the N W - S E
set were the latest active faults which, together with the N - S fractures, terminate against
the older E - W fractures. Fractographic features in this zone are not clear, they have
mostly been obliterated by frictional wear due to shear movement on the majority of the
fractures (Fig. 10c).
Fractures in the outer anticlinal hinge of the monocline (zone 4). This part of the monocline
is exposed at Pegwell Bay and inland in Monkton Quarry. The former exhibits a
spectacular group of steeply dipping fractures dominated by joint zones and multi-phase
faults. These include four sets, striking N-S, N W - S E , E - W and N E - S W , respectively. The
former two dominate the fracture pattern. In M o n k t o n Quarry, which is relatively
overgrown, the N E - S W and E - W fractures do not occur (Fig. 3).
The two dominant sets occur as macro- and mesofractures, the former cut across
bedding-parallel fractures (except some hard grounds and marl horizons) and extend from
top to bottom of the cliff faces. The exposed height of most of the macrofractures is equal
to or greater than that of the cliff face (i.e. several to tens of metres) and their exposed
length is equal to or greater than a few to several tens of metres (Fig. 11 a-d). The northerly
striking fractures occur mostly as discrete faults or narrow fault zones up to 1 m in width.
The northwesterly striking fractures include both faults and joints, and occur mostly as
zones up to several metres in width (Fig. 1 l a & c). The northwesterly fractures tend to
hook on to the northerly striking fracture (Fig. lid). Macrofaults of both sets show
anastomosing, wavy, stylolitized traces on the bedding-parallel, wave-cut platform (Fig.
1 le).
Slickenlines are common on fault surfaces of both sets and are particularly well
preserved inside coastal caves associated with differential weathering and erosion along the
fracture zones. Up to four sets of slickenlines occur on some faults and it is not always
possible to draw a firm conclusion as to their overprinting relationship, and thence their
relative age (Fig. 12). In addition, it is impossible to separate the complex movement
history on individual faults (which show multiphase movements) into its increments.
However, it is possible to determine the net (resultant) slip on each fault from the
orientation of the fault and the offset of key horizons, particularly flint bands. These are
consistently oblique-slip with extensional component (i.e. normal--dextral or normal-
sinistral). Such an extensional component to the net-slip is spectacularly displayed on the
nearly E - W trending cliff face where the Whitaker's 3 inch flint band steps downward
progressively from the eastern end to the western end of the bay along the traces of the
macrofaults with a net vertical offset of c. 10 m over a distance of 870 m.
Various degrees of cataclasis occur along the faults; these are evident from w e d g e -
shaped bodies of cataclasites and/or gouges which change in thickness drastically along
the exposed faults length and are similar to those shown in Fig. 10a & b.
Fractographic features (particularly hackle plumes) on the macrofractures of both
major sets are obliterated by the shear movements on them. However, the profile sections
162 M.S. AMEEN
CHALK FRACTURECHARACTERIZATION,THANET 163
of some fractures show a breakup of the main fracture in a fashion characteristic of major
twist hackles, similar to that shown in Fig. 5c, indicating that the fracture originates at the
top of the fractured horizon with an overall horizontal propagation, individual segments
(twist faces) have propagated mainly vertically from top to bottom.
There is a wide variety of hackle plumes and rib marks on the mesojoints of all sets.
They indicate cyclic or steady growth of the joints, evident from the occurrence or the lack
of rib marks, respectively. Most of the joints originate at flint nodules, fossils or voids
within the beds or at their borders. The use of blasting prior to concrete injection into
some of the coastal caves (to stabilize the cliffs) might have inevitably induced some joints
with clear hackle plumes and rib marks. Therefore, careful consideration should be given
to the analysis of fractographic features on the joints. The presence or lack of iron oxide
staining (common on the tectonic joints in the area), the closeness of the joints to the blast
locations and the local variation in their frequency are used as guidance for recognizing
tectonic from blast-induced joints.
Origin of the Thanet monocline and related fractures
Origin of the Thanet monocline

In order to understand the fractures, it is essential to throw some light on their host
regional structure, i.e. the Thanet monocline, which was described in the section on the
regional structure above. It is possible to evoke several hypotheses for the origin of the
Thanet monocline, these are:
(1) Halokinesis (salt diaperism) is very unlikely due to the absence of the Upper Permian
Salts underneath the monocline; such salts caused halokinetic activity in the North Sea
Basin, to the north of the London-Brabant Massif; (2) buckle folding requires the
detachment of the mainly Upper Cretaceous sedimentary cover from the early Palaeozoic
and Precambrian basement - both stratigraphic and geophysical evidence discredit this
hypothesis (Fig. 13); (3) drape or forced folding, this includes non-tectonic and tectonic
forced folding.
The non-tectonicforcedfolcling can be the result of differential compaction of sediments
over rigid topographic highs, or the result of the growth of igneous intrusions, particularly
laccoliths and sills. The latter is ruled out due to the lack of evidence. However, the former
is apparent from stratigraphic and geophysical data. These reveal that the sedimentary
cover (mainly Upper Cretaceous Chalk Group) is thinner over the London-Brabant
Massif compared to the Weald Basin and the southern North Sea Basin, to the south and
the north respectively (Cameron et al. 1992, fig. 82). Locally, in the Thanet monocline, the
Fig. 8. Macrofractures affecting horizontal beds exposed on the cliff faces in zone 3. (a) An older
easterly striking fault zone (F) dipping at 73~ towards the north abutted by later NW-SE joints (J).
Note that both sets are stained by iron oxide (North Cliff). (b) A NW-SE striking fault (F) dipping
at 85~ towards the southwest with three sets of slickenlines which pitch at 20~ NW, 20~ SE and 90~
SW. The fault is stained by iron oxide (South Cliff). (c) A NW-SE striking fracture zone (F) dipping
at 77~ towards SE. The zone is partly defined by en echelon fractures (EF) dipping at 85~ to the NE
(South Cliff). Notebook is 21 cm in length. (d) Northerly striking faults (F1) dipping at 60~ to the
east with an extensional slip (South Cliff) evident also from frictional-wear slickenlines. Note that
the NW-SE fractures (F2) abut against F1 (South Cliff). (e) A NW-SE striking macrofault (F)
dipping at 52~ to the southeast with c. 2 m dip-slip component (South Cliff). For locations see Fig. 3.
164 M. S. AMEEN
rO
~o
o ~
.... ~ 0
~=.-~
~m,= o
~ ~.~
~ "t=l 0 ,-,
-~o~
o~ ~o
"0o El W O O
~ o ~
0 " ~ " 0 ,.,.~
~9 '~ ,.~
~ ~'~, ~ ~
9 ~ "~ .~ ~
~ ,., ,.-~ ~=~'~

~ o
-o ~ o .-o
9 c~ c~
~~ o. ~ ~,,~ 9,'-"~ ~ -
~D.~ .,n. ~ ' ~

0 ~ , ~ . , - , c.)
~ . ~ z ~
~.~
o ~ o ~
, ~ . c~ . _ o~ "'---
0 ,,.~
CO C~ . ,~
- , ~.-~
o o ~:~ ~ " ~
c~ ..:. ~ ._~ 0
~ ~: ~ . ~
O~ ~0
~-~ ~ a.~
~ ~ . o .
~,i .~= o
"~:~~"~~.=..
o ~
166 M.S. AMEEN
Fig. 11. (a) A NW-striking nearly-vertical fracture zone

(> 7 m in width) consisting of closely spaced fractures
(Pegwell Bay). (b) A N-striking fault at Pegwell Bay dipping
at 72 ~ to the east with dextral-normal and later sinistral-
normal slips. The estimated net dip-slip is 0.21 m. (e, d & e)
The traces of: (c) a northwesterly fracture zone; (d)
northwesterly fractures (F1) hooking on to the older
northerly fractures (F2). (d) A northerly striking fault,
shown in (b), the dark colour of the trace is due to pressure
solution seams on the fault, all exposed on a bedding-
parallel wave-cut platform, Pegwell Bay. See Fig. 3 for
location.
Fig. 12. Slickenlines (frictional wear/pressure solution grooves), parallel to arrows, on macrofault
surfaces in zone 4. (a) Horizontal slickenlines on a northerly striking sinistral fault dipping at 85 ~ to
the east, Pegwell Bay. (b) Four sets of slickenlines on a northerly striking fault dipping at 78 ~ to the
west, Pegwell Bay. (c) Gently dipping slickenlines on a dextral-normal, northwesterly striking fault
in Monkton Quarry. Note the iron oxide staining of the surface and the relatively feeble development
of the slickenlines, compared to those in (a) & (b). For location see Fig. 3.
168 M.S. AMEEN
cover rock (mainly Upper Cretaceous Chalk) is considerably thinner in the upthrown limb
compared with the external, anticlinal hinge zone and the downthrown limb, respectively.
Furthermore, the present study discovered local differences in thickness within the
upthrown limb. Both variations are controlled by extensional and transtensional fault
escarpments in the basement which can be seen on an offshore seismic section of the
monocline interpreted in this study (Fig. 13).
The tectonic forced folding is caused by the draping of sedimentary layers over a growing
fault escarpment in the basement. The fault can be dip-slip (normal or reverse), oblique-
slip, positive or negative flower structures (i.e. transpression or transtension zones). The
strain patterns and hence the fracture patterns in the cover rocks, are strongly affected by
the type, angle of dip and amount of displacement of the basement fault/fault zones
(Ameen 1988, 1990).
Such a type of folding has affected the evolution of the Thanet monocline, which seems
to be underlined and controlled by at least two groups of faults/fault zones striking E-W
and WNW-SSE, respectively. The former are mainly dip-slip and control the dominantly
it
E-W trend of the monocline, west of Pegwell Bay (Fig. 1a,b). The second group forms a
transpressional fault zone (Pegwell-Minnis Bay Zone) detected in the present study from
seismic sections (Fig. 13), and is expressed in the cover rocks as deflections of the otherwise
E-W trending structure contours to a WNW direction along the zone (Fig. l b), which
defines an upward bulge in these rocks to accommodate the flower structure in the
basement. This is also supported by the change in the fracture patterns from zone 1, which
is far from the transpression zone, to those in the immediate vicinity (zones 2 and 3) and
those within the transpression zone (zone 4), respectively (Fig. 3). Both groups of
basement faults were most probably active during the Cretaceous and the Tertiary N-S
compression. Such faults would have affected the evolution of the monocline in two
different, though related, ways; differential deposition and compaction of sediments on
existing fault escarpments (non-tectonic forced folding) and the draping of the
sedimentary layers over active faults (tectonic forced folding). The evolution of the
forced fold occurred in two main phases; an early extensional phase and a late
compressional phase. The extensional phase included non-tectonic and tectonic forced
folding and took place during the deposition of the Chalk where a transtensional
movement on the Pegwell-Minnis Bay Zone resulted in thicker accumulation of the Chalk
in the zone. Similarly, extensional movements on the easterly striking faults caused thicker
Chalk to be deposited on the downthrown side of the faults (Fig. 13). The second
(compressional) phase started late in the Cretaceous and caused inversion of the
movement on the basement faults and was accompanied by mainly tectonic forced folding
(including cross-forced folding above the transpressed Pegwell-Minnis Bay zone),
uplifting and erosion evident from the angular erosional unconformity between the
Chalk and the Tertiary rocks. Further compression took place during the Tertiary when
the monocline evolved to its present shape.
Origin o f the fractures

The fractures in the Upper Chalk of the Thanet monocline resulted from a few different
phases, at least some of which are related to the evolution of the fold.
Phase 1 - burial compaction. Unlike limestone, chalk does not usually lithify during early
diagenesis at, or very close to the surface, therefore early jointing occurs mainly after a
o
O
4...a
_o
.3
~J
..=
O
O
c~
[..,
~D
e~
o
O
o,'~-
170 M.S. AMEEN
considerable burial has taken place. Burial stress regime is characterized by vertical
maximum principal stress (or1) and horizontal minimum principal stress (or3). Under such
circumstances the chalk will tend to consolidate, by expulsion of pore water, and lithify by
pressure solution and cementation (Jones et al. 1984). However, the intact chalk has an
exceptionally low permeability (13.9-37) • 10-9m s-1 for a rock with high porosity; 29.6-
45.6 (Bell 1977; Bell et al. 1990). Therefore, a buildup of pore-water pressure with burial-
induced consolidation and lithification is expected to facilitate the initiation and
propagation of fractures. The orientation of the effective least principal stress is affected
by any shallow, buried topographic features underneath the Chalk. On the Isle of Thanet
effective least principal stress is expected to be in a direction perpendicular to the local
trend (and parallel to the maximum slope) of the local topographic highs in the basement
and the associated boundary faults. Therefore, this resulted in the development of vertical
or steep macrofractures (joints and normal faults) striking easterly (zone 1), northwesterly
(zone 2), northerly and easterly (zone 3) and N-S (zone 4). Some fractures propagated
rhythmically others propagated smoothly with little hesitation. This is evident from the
occurrence of rib marks and rhythmic hackle plumes on the former and their lack on the
latter. Stress release after the development of these fractures led to a change in the relative
amounts of the principal stresses and the development of vertical or steep macrofractures
(mostly joints with some normal faults) orthogonal to the first set. These late occurring
burial fractures are generally rare, form a few to several per cent of the total number of
fractures, except in zone 3 (Fig. 3), and are constrained between the older burial fractures.
The majority of the macrofractures show fractographic features indicative of a
horizontal, bedding-parallel propagation. This resulted from local reorientation of the
vertical maximum effective remote stresses to horizontal due to the existence of horizontal
bedding-parallel lithological discontinuities. The Upper Chalk was deposited as a
relatively homogenous lithological unit which lacks real bedding discontinuities, except
for some widely spaced hardgrounds, thin marl zones and flint bands, which acted as
'stress deflectors'. Before the onset of uplifting there were no closely spaced horizontal
(bedding) joints (see below) to constrain the burial phase fractures to smaller dimensions.
This is evident from the fact that the majority of the macrofractures and their
fractographic features continue across existing bedding discontinuities (e.g. Fig. 5) but
tend to abut against hardgrounds and marl horizons.
The joints which occurred during this phase and the consequent phases (described
below) originated as tensile (Mode I) fractures with the local (net or effective) minimum
principal stress (cr3 eft) oriented at right angles to the main fracture surface. The local (net
or effective) maximum and intermediate principal stresses (crl eft and ~r2 eft) lie within the
fracture surface. At the leading edge of each individual fracture ~rl eft and or2 eft are
parallel and perpendicular to the direction of propagation of the fracture, respectively
(Figs 5, 6 & 10c). However, these and consequent faults resulted from the coalescence and
shear or mixed-mode (tensile-shear) loading of microscopic and/or mesoscopic tensile
fractures and/or shear or mixed-mode loading of macrojoints.
P h a s e 2 - tectonic p h a s e . This includes an early extensional (Upper Cretaceous) and a late

compressional stage (late Upper Cretaceous to Pliocene) associated with the development
of the monocline. The extensional period was caused by a vertical maximum principal
stress (Crl) and horizontal minimum principal stress (or3). This is similar to, and difficult to
separate (in terms of the resulting fractures) from, the burial phase. The two were acting
penecontemporaneously as discussed in the section on the origin of the Thanet monocline
(above). Therefore, many of the macrofractures developed in the burial phase are also
most probably linked to the extensional tectonic stage.
The second (compressional) stage was associated with the Alpine lateral N-S
compression (Crl) which affected the Weald Basin to the immediate south of the Thanet
monocline. Such a N-S compression is expected to have reactivated the suitably oriented
fractures (both in the basement and the sedimentary cover) and initiated new sets of
fractures, where necessary, to take up the strain. The reactivation of fractures is evident
from extensive faulting in zones 3 and 4, which was accompanied by the development of
new northwesterly striking fractures (mostly faults). These fractures displaced or abut the
older burial/extensional phase fractures. The majority of the northwesterly striking faults
are strike-slip or oblique-slip faults, with an extensional movement component (Fig. 3).
The existence of such faults is also evident from seismic sections a few kilometres offshore.
The section, interpreted in the present work, indicates the existence of a major fault zone
in the basement along the Pegwell-Minnis Bay grain shown on Fig. lb. The sense of slip
and geometry of the zone are characteristic of a convergent strike slip-fault (positive flower
structure) indicative of transpression (Fig. 13). Such a major fault zone is expected to
cause the development of a suite of fractures and the rejuvenation of pre-existing fractures
to accommodate the associated strain pattern. Transpression faults (Harland 1971) have a
component of horizontal shortening across the strike-slip fault zones which is
compensated by cataclastic deformation and uplift of the rock wedge enclosed in the
fault zone (Lowell 1972; Wilcox et al. 1973). The sedimentary cover above such a bulging
basement wedge is expected to act passively to accommodate itself to the uplift of the
wedge and the strike-slip movement of the main bounding blocks (Fig. 14). Such an
accommodation in the Thanet monocline has taken place by gentle, cross-forced folding
along the WNW-SSE basement grain, outlined on Fig. lb and also evident from the
pattern of exposure of the Tertiary and Cretaceous rocks and the structure contour map of
the base of the Chalk (Fig. l b). The amplitude of this flexure decreases dramatically from
a few tens of metres near Pegwell Bay to several metres near Minnis Bay. Such a decrease
is noticeably accompanied by an equally significant decrease in the intensity of faulting in
the exposed Upper Chalk (Fig. 3). Therefore, the causative faulting in the basement seems
to diminish significantly northwestwards.
The fracture patterns associated with this phase is expected to include a combination of
those typical of a strike-slip zone and those associated with mild drape folding (Fig. 14),
i.e. a combination of strike-slip, oblique-slip and normal faults (and joints parallel to
them). The complete suite of fractures shown in Fig. 14 do not always occur, especially
when pre-existing fractures exist which are orientated suitably to take up the strain, as is
the case on the Isle of Thanet.
Phase 3 - uplifting~unloading phase. This phase started late in the deformation history and
was accommodated by the occurrence of near-horizontal (bedding-parallel) fractures
(joints) and bedding-perpendicular or steep mesofractures parallel to, and perpendicular
to, the burial/tectonic macrofractures described above. These have developed particularly
at the top several metres of the exposed Chalk where a rapid increase in frequency of
fractures is clear. It is difficult to assess the effect of the Quaternary, relatively brief, glacial
burial which took place during the final stages of the tectonic uplifting. Such a burial
is expected to be taken up by movements on pre-existing fractures. Detailed studies,
including fault rocks dating, may help in recognizing neotectonic/glaciation related
faulting.
172 M.S. AMEEN
Fig. 14. A schematic block diagram illustrating the potential fracture patterns in a drape fold
associated with a transpression fault zone (positive flower structure) in the basement. For details see
text.
Conclusions and discussion

The Thanet monocline and related fractures resulted from a complex combination of non-
tectonic and tectonic (extensional and compressional) forced folding largely controlled by
dip-slip and oblique-slip faulting in the Precambrian to Lower Palaeozoic basement. The
non-tectonic and extensional-tectonic forced folding occurred throughout most of the
Upper Cretaceous and was characterized by a stress regime with a vertical maximum
principal stress (trl) and a horizontal minimum principal stress (a3). The largely Chalk
cover rocks responded to these stresses by vertical and steep macrofractures (mostly joints
and some extensional faults). The orientation of the fractures vary in the different parts of
the monocline depending on variation in the direction of or3, which vary with the type and
orientation, of the underlying fault escarpment in the basement. Accordingly, a minimum
of four fracture domains (zones) have been recognized in the monocline. The tectonic
forced folding and fracturing of the cover rocks continued during the compressional phase
which started late in the Upper Cretaceous and continued during the Tertiary. This phase
was characterized by a N-S compression (trl) which inverted the movement on the
basement faults and rejuvenated the macrofractures in the cover rocks (mostly by faulting)
and initiated new fractures (faults and joints).
The joints which occurred during the different phases of folding and fracturing
originated as tensile (Mode I) fractures with the local (net or effective) minimum principal
stress (cr3 eft) perpendicular to the main fracture surface. The local (net or effective)
maximum and intermediate principal stresses (Ol eft and cr2 eft) lie within the fracture
surface. At the leading edge of each individual fracture crl eft and cr2 eft are parallel and
perpendicular to the direction of propagation of the fracture, respectively. However, the
faults resulted from the coalescence and shear or mixed-mode (tensile-shear) loading of
microscopic and/or mesoscopic tensile fractures and/or shear or mixed-mode loading of
macro joints.
An uplifting/unloading phase started late in the deformation history and was
accommodated by the occurrence of near-horizontal (bedding-parallel) fractures (joints)
and bedding-perpendicular or steep mesofractures parallel to, and perpendicular to, the
burial/tectonic macrofractures described above. These have developed particularly at the
top several metres of the exposed Chalk where a rapid increase in frequency of fractures is
clear. It is difficult to assess the effect of the Quaternary, relatively brief, glacial burial
which took place during the final stages of the tectonic uplifting. Such a burial is expected
to be taken up by movements on pre-existing fractures. Detailed studies including fault
rocks dating may help in recognizing neotectonic/glaciation related faulting. In addition,
some joints are induced by blasting during remedial measures to protect the coastal cliff
from sea erosion. Other joints are probably related to, or opened by, episodes of sudden
subaerial erosion and sudden cliff retreats, common during the winter, and tend to
coincide with and follow periods of maxima of air frost and water surplus (rain).
The present results can be used as a guidance to review faults and folds in the immediate
vicinity of the Thanet monocline. The dominantly N W - S E to W N W - S S E striking faults
in the Kent Coalfield of the Richborough syncline, to the south of the study area, and in
the English Channel (described by Shephard-Thorn et al. 1972) are most probably related
to flower structures similar to the Pegwell-Minnis Bay transtension/transpression zone in
the Precambrian-Lower Palaeozoic basement rather than strike-slip or dip-slip faults. In
addition, the assumption that joints in the cover rocks, including the Chalk, in other parts
of east Kent are propagated from, or related to, basement faults of the same strike
(Middlemiss 1983) is an oversimplification. As the current study shows, the fractures are
caused by movements on basement faults and/or differential burial compaction over
basement fault escarpments. The latter and dip-slip faulting in the basement result in
fractures (faults and joints) in the cover rocks striking parallel to the causative fault in the
basement. However, strike-slip and oblique-slip (transpression and transtension) faults
result in fractures in the cover rocks striking at a range of angles to the causative fault/
fault zone in the basement.
The present studies show clearly the occurrence of extensional/transtensional move-
ments on the basement faults during the deposition of the Upper Cretaceous rocks,
including the Chalk. This raises serious doubts about the commonly held view that the
extensional movement in southern England ceased before the onset of the deposition of
the Chalk (e.g. Stoneley 1982) and hence the assumption that the Chalk thickness does not
vary drastically across basement faults. Recent surface and subsurface observations
support the occurrence of extensional movement on basement faults as evident from
significant and abrupt difference in the Chalk thickness across the basement faults in the
Dorset and the Isle of Wight area (Andrew Gale, pers. comm.).
The author is grateful to Dr J. Cosgrove for reviewing the paper, Dr E. R. Shephard-Thorn for
reviewing the manuscript and his helpful comments, Dr Brian D'olier for supplying the raw seismic
data, and Dr Andrew Gale for the helpful discussions during the course of this work.
174 M.S. AMEEN
References
AMEEN, M. S. 1988. Folding of layered cover due to dip-slip basement faulting. PhD Thesis, Imperial
College, London, UK.
- - 1990. Macrofaulting in the Purbeck-Isle of Wight Monocline. Proceedings of the Geologists'
Association, 101 (1), 31-46.
- - 1995. Fractography and fracture characterization in the Permo-Triassic sandstones and the
Lower Palaeozoic Basement, West Cumbria, England. This volume.
BELL, F. G. 1977. A note on the physical properties of the Chalk. Engineering Geology, 11, 217-115
, CRIPPS, J. C., EDMONDS, C. N. & CULSHAW, M. G. 1990. Chalk fabrics and its relation to
certain geotechnical properties. In: Chalk: Proceedings of the International Chalk Symposium.
Thomas Telford, London, 187-194.
CAMERON,T. n . J., CROSBY,A., BALSON,P. S., JEFFERY,D. J., LoTr, J. K., BULAT D. J. & HARRISON,
D. J. 1992. The geology of the southern North Sea, British Geological Survey, UK Offshore
Regional Report. HMSO, London.
CAWSEY, D. C. 1977. The measurement of fracture patterns in the Chalk of southern England.
Engineering Geology, 11, 201-215.
GEOLOGICAL SURVEYOF GREAT BRITAIN. 1980. Ramsgate Sheet, 1:50,000. HMSO, London.
HAMBLIN,R. J. E., CROSBY,A., BALSON,P. S., JONES,S. M., CHADWICK,R. A., PENN,I. E. & ARTHUR,
M. J. 1992. The geology of the English Channel, British Geological Survey, UK Offshore Regional
Report. HMSO, London.
HANCOCK, J. M. 1976. The petrology of the Chalk. Proceedings of the Geologists" Association, 86,
499-535.
HARLAND,W. I . 1971. Tectonic transpression in Caledonian Spitsbergen. Geological Magazine, 108,
27-42.
JONES, M. F., BEDFORD, J. & CLAYTON, C. 1984. On natural deformation mechanisms in the Chalk.
Journal of the Geological Society, London, 141, 675-676.
LOWELL,J. D. 1972. Spitsbergen Tertiary orogenic belt and Spitsbergen fracture zone. Geological
Society of America Bulletin, 83, 3091-3102.
MIDDLESMISS,F. A. 1983. Instability of Chalk cliffs between the South Foreland and Kingsdown,
Kent, in relation to geological structure. Proceedings of the Geologists" Association, 94, 115-122.
SHEPHARD-THORN,E. R. 1988. Geology of the country around Ramsgate and Dover, British Geological
Survey Memoir for 1:50 000 geological sheets 274 and 290, England and Wales. HMSO, London.
, LAKE, R. D. & ATITULLAH,E. A. 1972. Basement control of structures in the Mesozoic rocks
in the Strait of Dover region and its reflexion in certain features of the present land and
submarine topography. Philosophical Transactions of the Royal Society, London, A272, 99-113.
STONELEY, R. 1982. The structural development of the Wessex Basin. Journal of the Geological
Society, London, 131, 543-554.
WILCOX, R. E., HARDING, T. P. • SEELY, D. R. 1973. Basic wrench tectonic. American Association of
Petroleum Geologists, Bulletin, 57, 74-96.
Fracture surface markings in Liassic limestone at

Lavernock Point, South Wales
JOHN C. R O B E R T S
University of Ulster, Coleraine Campus, Coleraine, Northern Ireland,
BT52 1SA, UK
Abstract: A study of fracture surface markings (FSM) on three joint sets, striking at 190,
335 and 280~ cutting a 2.5 m thick slice of interbedded limestones and mudstones of
lower Liassic age, yields evidence which enables determination of the orientation of the
palaeostress field at the time of joint formation. A specific element of the FSM, namely
the border planes, or the twist hackle, has been used to determine the orientation of a~,
in that these twist-hackle structures are considered to be developed normal to the
remote tensile stress ~r3.In the case of the joint sets striking at 335 and 190~ the sense of
rotation of the twist plane would suggest that for both sets Crl was aligned N-S and
almost parallel to bedding. Other FSM features suggest that the 190~ set propagated
rapidly with S-type plumose markings being the most common. The small angle of
border plane rotation (10-15 ~) suggests that all of the joint sets propagated under
relatively low confining pressures. The sense of border plane rotation on the 335 and
190~ sets suggests that there is a possible conjugate relationship between them.
However, the other main joint set which strikes at 280~ cannot be accounted for by
simultaneous development in a N-S aligned ~rl which might have caused the
development of the 335-190 ~ joint sets.
In recent years there has been renewed interest in various aspects of rock jointing. These
studies have ranged from the analyses of interacting joint sets (Reches 1976; Hancock
1985; Bahat 1987), joint interpretation investigated regarding joint frequency (Rives et al.
1992), and the concept of fracture saturation and the sequential development of joint
patterns (Gross 1993). There has been a movement away from the investigation of regional
joint systems and attempted interpretations of regional joint patterns in terms of
palaeostress fields. The emphasis has moved towards an understanding of the factors that
produce fracture and propagation mechanics of individual joint planes. The nature of this
change is documented clearly in the synoptic paper by Pollard & Aydin (1988).
The relevance of the fracture surface markings (FSM) and what they mean in terms of
the mechanics of joint initiation and propagation has been made possible by using
fractography as a geological tool. Kerkof (in Bahat 1991) states, 'while in photography the
light fixes the picture of an event, in fractography it is the fracture that plays the
corresponding part'. FSM have been observed for almost a century in many different
kinds of rocks and in a variety of tectonic settings. Woodworth (1896) produced the
fundamental article on the subject in his study of the Mystic River Slates. However, the
adoption of fractographic principles by Kulander et al. (1979) to a theoretical and applied
approach to rock fracturing laid the path to future interpretative studies. The synoptic
study by Bahat (1991) provided an updated synthesis of the subject.
As a result of fractographic studies what has to be called into question are the former
interpretations of joints as kinematic indicators of rock deformation. The classification of
176 J.C. ROBERTS
Fig. 1. The components of fracture surface markings; descriptive classification from Hodgson
(1961b) and genetic classification from Kulander et al. (1979).
joints as being either 'shear' or 'tension' in origin is no longer tenable. There would be
broad general agreement that the bulk of joints result from Mode I tensile fracturing, but
where the total surface area of the joint is displayed (and in suitable lithologies) the joint
can be seen to be the result of mixed-mode failure. The result of mixed-mode failure, i.e.
Modes I and III, can be seen clearly in Fig. 1, where in broad terms the hackle plume area
represents Mode I failure (opening) and the fringe (border planes-twist hackle) the out-of-
plane propagation of the main joint plane in response to a twist imposed by the influence
of the remote stress field (a3). However, even in the absence of twist hackle most workers
would regard a3 to be normal to the joint plane.
The study area

The basal succession of the lower Liassic crops out west of Penarth, South Wales, between
Lavernock Point and Saint Mary's Well Bay (Fig. 2). The succession has been described as
the Saint Mary's Well Bay Formation (Waters & Lawrence 1987), consisting of alternating
blue-black shales, mudstones and impure muddy limestones. This formation comprises
the lowest 16m of the Liassic succession and passes upwards into the Lavernock
Formation, a predominantly shaley sequence c. 15 m thick. The studied horizon is located
in the upper part of the Saint Mary's Well Bay Formation (Fig. 3). The succession is well
FRACTURE SURFACE MARKINGS 177
Fig. 2. Location and general geology of the study area.
exposed in a low cliff line and also in a series of gently dipping limestone pavements. The
pavements provide information regarding the geometry of the joint patterns and their
interactive elements, whilst the FSM were studied in the cliff outcrops.
The structural geology of the area is that of a gently SSE plunging syncline (Fig. 2) and
one fault subparallel to the synclinal axis. The fault strikes at c. 330 ~ dips 65~ with a
dip-slip displacement of c. 2 m. Where the fault trace intersects the pavement exposures
there is a 50 cm zone of calcite mineralization, usually in the form of parallel-sided veins,
3-5mm in width and filled with sparry calcite. This is the only manifestation of
mineralization in the study area.
The fractured horizon

The best developed examples of FSM are stratigraphically located towards the top of the
Saint Mary's Well Bay Formation. Three limestone beds, i.e. 32-34, display abundant
FSM (Fig. 3). However, for descriptive purposes, it is from Bed 22 that most of the
observations are derived; the FSM geometries are similar in Beds 33 and 34, and with
particular reference to the sense of twist-hackle rotation, identical. Some FSM geometries
are recorded in Fig. 4a-d, all relating to Bed 32, whilst one example of joint pattern and
interactions is given in Fig. 4e. The distinctive lithology of this horizon had already been
identified by Waters & Lawrence (1987), who named it the laminated horizon. The most
178 J.c. ROBERTS
L.S. Lavernoc
34
I 33
t-
o I 1 32
0
E 34 } ES.M.
33
32 Horizon I
I
la_ I lOcm
1
m
m
m
31
1 m D m
Fig. 3. Simplified stratigraphic column with the location of the studied horizon.
Fig. 4. (a) A point source in Bed 32 with an elliptical termination. (b) Point source of an S-type
chevron in Bed 32. Note the influence of the distinct parting on the location of the plumes. (e) A
double point source in Bed 32. (d) An S-type chevron in Bed 32 with a deep fringe and twist-hackle
planes. (e) The joint trace pattern in Bed 32, looking towards 10~ Chalk marks are 1 m apart and are
parallel to 190~ joint traces and perpendicular to 280 ~ joint traces.
highly developed l a m i n a t i o n occurs in Bed 32, a 10cm thick limestone with at least 25
laminations, some of these can be seen clearly in Figs 4 a - c & 5a & b. Therefore, the study
is concentrated within a 2.5 m slice of interbedded limestones and shales, with the F S M
relating m a i n l y to observations within the limestone layers. However, their influence u p o n
the interbedded shales can be seen clearly in Fig. 5 b - d (deep twist-hackle structures).
180 J.C. ROBERTS
Fig. 5. (a) Parallel 190~ joint faces displaying localization of the chevron axis along a well-defined
lamination. (b) Parallel 190~ joint faces displaying asymmetric barbs. Note that the chevron at the
left of the figure displays within the limestone layer and exhibits breakdown into broader planes. (e)
An asymmetrically positioned point source in Bed 32. Note the strongly developed border planes
above the joint plane. These are rotated counter-clockwise in relation to the joint surface. (d) An
asymmetric chevron breaking down into deep fringe with en echelon border planes.
The joint system

The joint system within the study area has been described by Roberts (1974). The
systematic joint system is represented in the rose diagram of Fig. 2. However, the synoptic
nature of the diagram does not convey accurately the joint pattern, Beds 32-34 are cut by
three main joint sets (better referred to as spectra as there is a spread of azimuth), i.e. 335,
280 and 190 ~ (Fig. 4e, sets 280 and 190~ As is evident in Fig. 4e, the 190 ~ set displays a
high frequency and a short joint trace length, but is restricted to Beds 32-34.
Method of study
During the previous study of the systematic joint patterns (Roberts 1974) FSM were noted
at Lavernock, and many other localities along the coast. However, the best examples occur
within the described location. The attributes of FSM were recorded subsequent to the
investigation of the joint system, when ground truth data were recorded for the joints.
Details of FSM have been derived using oriented photographs, however, ease of obtaining
detailed surface morphologies was reduced by the general E - W alignment of the cliff face
which precluded extensive development of the 190~ joint plane surfaces. In addition, the
low relief of FSM induced a further constraint in that incident light (sun) angles could only
be optimized over short time intervals.
Fracture surface markings (FSM)

The morphological elements of FSM are illustrated in Fig. 1. The current descriptive
terminology is due to Hodgson (1961b), but this has been amplified by Kulander et al.
(1979) into a genetic scheme. The main departures in the respective terminologies relate to
structures in the fringe, and the recognition by Kulander et al. (1979) of this structure as
representing a particular mode of fracturing.
FSM on t h e 1 9 0 ~ j o i n t s e t
As stated above, excellent FSM occur in Beds 32-34; however, the bulk of the descriptions
relate to observations on Bed 32.
Hackle plume markings. These are ubiquitous features on the surfaces of the 190~ joint set,
and the bulk of the observations indicate that the common direction of propagation of the
joint plane was layer parallel and controlled by lamination within the bed. This control
exerted by lamination is easily seen in Figs 4a-c & 5a & b. The plumes, in a geometric
sense, can be described as chevron type, in that the plume barbs have a straight linear
relationship to the plume axis, in contrast to some of the wavy plumes which are
frequently associated with fracture initiation points on the base of a bed (Bahat &
Engelder 1984). Details of the variations in the location of the plume axis and its
asymmetry with respect to the bedding surfaces are given in Figs 4c & d & 5d. In Fig. 5d
the plume axis is inclined at an angle of c. 25 ~ to the left of the photograph, but it becomes
layer parallel within 10 cm of the cross joint striking at 280 ~ A parallel joint plane at the
same exposure displays identical plume axis orientation, indicating that the principal
effective stresses were uniform over this volume of rock (see also Fig. 5a). Divergence in
propagation direction is seen clearly in Fig. 5c, where there is a fan arrangement of the
plume structure indicating that in the initial stage of propagation the joint surface
propagated vertically as well as parallel to bedding. However, the majority of the hackle
plume marks on the 190~ joint set were of the S-type (Bahat & Engelder 1984).
Point sources. The reality that joints initiate at points (small discrete areas of failure) can
be demonstrated in many places (Roberts 1961). Point sources are seen clearly in Figs 4a-c
& 5b, where the hackle plumes diverge from a central point. Figure 4a shows a point
source that lies mid bed, a lamination-parallel plume axis and a rib mark 20 cm to the left
of the initiation point. A most uncommon situation is seen in Fig. 5b where two in-plane
point sources have given rise to a rhomboid-shaped interference pattern.
Rib marks. These structures are not common on the 190~ joint set, but where they appear
they indicate a deceleration of the fracture front (Fig. 4a), and in some cases actual
termination of the joint plane (Kulander et al. 1979, fig. 77).
182 J . c . ROBERTS
~'~'~~ 010~
335~ Twist
//! j hackle
plane
l
ccw Twist
c% cw Twist
oo
0 Twist angle
E Greatest principle driving stress
~ ....,,
Remote stress field
I \
155~
1900
Fig. 6. Relationships of joint traces and twist-hackle planes to the stress field.
Twist-hackle planes of the fringe. The recognition of twist-hackle planes is important

because according to Kulander et al. (1979) they form when 'a propagating fracture
abruptly enters a region of different stress orientation. The fracture breaks (twists) into a
series of en echelon individual lance-like twist-hackle faces, each perpendicular to a
resultant tension'. In this investigation the salient feature, as evidenced by Beds 32 and 33,
is that the twist-hackle component is not found in the limestone layer but in the
interbedded shale layers, i.e. above and below the limestone; this relationship is seen
clearly in Figs 4d & 5c & d. What should also be noted is the development of twist hackle
from the hackle plume (Figs 4d & 5d); in Fig. 5d the twist-hackle zone is seen to have
developed to a depth of 15 cm below the hackle-plume axis. In Fig. 5d the discrete planar
elements of the twist hackle are evident, also the distinct out-of-plane orientation of the
twist-hackle planes, this is recorded as the twist angle (Fig. 6), the values of which have
implications for the intensity of confining pressures at the time of fracture propagation
(see below). In addition, the recognition of a zone of twist hackle is important in that if the
twist hackle were to develop on a larger scale, then it could be mistaken for a zone of close
jointing (Kulander et al. 1979). Indeed, such a process of large-scale twist-hackle
formation leading to the production of close joint zones has been noticed in more quasi-
isotropic rocks, e.g. granites and volcanic rocks (P. Bankwitz, pers. comm.), and indicates
high velocity propagation of the joint planes in the close joint zone.
F S M on other j o i n t sets
FSM are not confined to Beds 32-34, it is that they attain their acme at this horizon, and
in particular on the 190~ joint set. The 335 ~ joint set frequently displays S-type plumes
mainly of a chevron geometry. In contrast, the 280 ~ joint sets are characterized by FSM of
a more delicate plumose pattern with many elliptical rib markings.
Discussion
FSM were studied in three limestone beds on both flanks of an open syncline- dips
< 10~ - which is crossed by a single dip-slip fault with a downthrow of 2 m to the west
(Fig. 2). Both limbs of the syncline are cut by three sets of joints (Rose diagram, Fig. 2).
This diagram, a cumulative frequency plot, gives an impression of the geometry of the
joint system which pervades the Saint Mary's Well Bay Formation of some 36 limestone
beds. There is nothing in the morphology of the joint sets to suggest that they can be
differentiated in terms of their origin, e.g. uplift as opposed to burial (Bahat 1991),
however, it was suggested by Roberts (1974) that the joint system resulted from a weak
tectonic deformation, with O"1 aligned approximately N-S. The somewhat tenuous
kinematic relationship of the joint sets to the local synclinal structure requires
substantiation, some of which might be provided by the evidence of FSM. However,
whatever conclusions are arrived at they must be considered solely with respect to the
nature of the fracture pattern and the associated FSM.
The pertinent observations are that all of the FSM in Beds 32-34 which are developed
on the 190 ~ joint set are of the S-type and show a chevron hackle plume, and that the joint
set is of a multilayer nature, i.e. they cut more than one bed. The jointing architecture of
Beds 32-34 suggests that the 190~ set formed last in the jointing sequence, as evidenced
by their high frequencies, large number of intersections upon other joint sets and self-
interaction. In addition, the high frequency of cross-cutting (intersections) indicates that
the 335 and 280 ~ joint sets were closed at the time the 190 ~ set propagated.
The development of clear FSM in this rock succession is not unexpected since the role of
lithology as a controlling factor has been demonstrated by many workers, and is
summarized by Syme-Gash (1971), who states that FSM are 'commonly developed on
joints with inter-laminated strata such as occurs in the Lias or flysch and on fractures in
metal plates', he also recognized the fundamental control of grain size on inhibiting FSM,
184 J.C. ROBERTS
their development being favoured by fine-grain sizes. In contrast, it must be noted that
large-scale FSM have been observed in coarse-grained igneous rocks, and therefore grain
size alone cannot be regarded as an inhibiting factor for their development.
The confinement of the hackle plume to the limestone layers, and its initiation within the
limestone layer, must result from differences in the elastic module of the interbeds. This is
borne out by the development of the twist hackle in the overlying and underlying shale
beds (Fig. 5c & d) where there is a reduction in the crack-tip stresses due to local plastic
flow (Kulander et al. 1979). The most important observation with respect to the twist
hackle is that the sense of twist remains constant along the length of the joint plane, and
that the angle of twist is usually in the order of 10-15 ~. This study has shown that the sense
of twist on the 190~ joint set is always counter-clockwise. Recent work by Cruikshank et
al. (1991) has shown that a relationship exists between the size of the twist angle and ratio
of Mode III to Mode I driving stresses. Although their calculations were based upon a
sandstone with a Poisson's ratio of 0.25, their values would indicate that the twist angles in
the shales suggest a stress ratio in the region of 0.2-0.3. The significance of the value of the
angle of twist has been discussed by Bahat (1987) who has shown that it can also be related
to the angle of internal friction of the rock. In the case of Bahat's study of Eocene chalks
the characteristic twist angle was found to be 17~
The 335 and 280 ~ joint sets display clockwise twists of 10-15 ~. These observations
suggest that all the joint sets formed under similar stress conditions, but not necessarily at
the same time. The salient point is that certain conditions are conducive to the production
of the twist-hackle component of the joint plane. The presence of the twist hackle suggests
low confining pressures, Bahat (1991) stated, 'It appears that en echelon cracking is
inhibited beyond certain confining pressures'. In terms of the orientation of the
palaeostress field, the twist-hackle plane is an indicator of the direction of maximum
tension immediately ahead of the crack front (Cruikshank et al. 1991). Bahat (1991) states
that the size of the twist angle is influenced by the magnitude of the remote stress field.
These conclusions were arrived at by Kulander et al. (1979) who stated that, 'Twist hackle
steps increase in relief towards stratum boundaries. In this direction principal effective
stresses decrease, superposed stresses have greater effect, and the resultant stress
perpendicular to the twist-hackle face may alleviate more from the principal stress'.
The observations on the magnitude of the twist angle may have a bearing upon
speculations regarding possible stratigraphic thickness of the Jurassic succession in the
study area. The remnant thickness of the Liassic is in the order of 150 m (Wilson et al.
1990), but Wobber (1966) and Owen (1967) have suggested that Mesozoic strata covered
the South Wales Coalfield, a depth of burial that would have required a much greater
thickness of post-Triassic strata. Evidence of a thicker Jurassic succession is found in fault-
bounded synclines in the Bristol Channel (Wilson et al. 1990). Therefore, with regard to
the small value of the twist angle, either the joint set in question evolved early during the
post-lithification stage of sedimentation or much later under conditions of low overburden
(confining pressure), which would suggest an attenuated Jurassic succession.
Finally, some of the lithological controls exerted by the succession should be considered.
The fractured succession consists of alternations of mud and lime rocks in almost equal
proportions, and has responded in two different ways with respect to the development of
the joint sets (Hodgson 1961a). The limestone layers respond to the stress as relatively
rigid members, and it would appear that limestones 32-34 acted either as stress con-
centrators or occupied a zone of high stress concentration within the succession. The role
of the limestone beds as stress concentrators might have been influenced by the relative
rates of diagenesis of the lime and mud sediments. The lime sediments lithify earlier and
would therefore be capable of taking stress earlier than the mud sediments, and so would
act as accumulators of early stresses. There is no evidence to suggest that if joints formed
early in the limestone layers that they propagated upwards or downwards at a later stage
into the mud rocks, as stated above where hackle plumes are exposed to the principal
direction of joint propagation which was layer parallel with joint limits terminating in the
mud rocks. What is evident is that the limestone layers acted as the stress concentrators
and provided sites for fracture initiation.
Conclusions
Within the basal part of the Liassic succession a 2.5m thick horizon comprising three
limestone layers with shale interbeds displays abundant, well-developed FSM. These
represent an episode of rapid, brittle fracturing during one phase of the evolution of the
joint system. The fracture pattern includes three joint sets striking at 335, 280 and 190 ~.
The F S M are best developed on the 190 ~ joint set which is characterized by having a high
joint frequency and short joint trace length. The other two sets also display F S M but not
as frequently as the 190 ~ set. The most common geometry of F S M on the 190 ~ set is S-
type, chevron hackle plumes. The joints terminate in the overlying and underlying mud
rocks with deep borders of twist hackle. The sense of rotation of the twist-hackle planes is
consistent for a given joint set, albeit the degree of twist suggests low confining pressures at
the time of joint propagation. The clockwise rotation of the twist hackle on the 335 ~ joint
and the counter-clockwise rotation on the 190 ~ set suggests that or1 was oriented
approximately N - S at the time of propagation of both joints. This, however, does not
imply that the joint sets are of a conjugate nature, but rather that if conjugate relationships
were being sought then this pair set could be so considered, whereas the clockwise twist on
the 280 ~ set excludes any conjugate relationship with either the 335 ~ or the 190 ~ set.
Many thanks are due to Nigel McDowell for the photographic work, Killian McDaid for the
graphics, and to Mary McCamphill and Anne Hayes for typing the manuscript. Also, in the earlier
years of the investigation, to the Research fund of the University. The manuscript was improved
greatly by the constructive comments of the referees - M. Cooke, A. Aydin and P. Bankwitz. I thank
them all.
References
BAHAT, D. 1987. Correlation between surface morphology and orientation of cross-fold joints in
Eocene chalks around Beer Sheva, Israel. Tectonophysics, 136, 323-333.
- - 1991. Tectono-fractography. Springer-Verlag, Berlin.
- - 8s ENGELDER, T. 1984. Surface morphology on cross-fold joints of the Appalachian Plateau,
New York and Pennsylvania. Techonophysics, 104, 299-313.
BANKWITZ,P. 1965. Uber Klufte, beobachtungen in Thurinischen Schiefergebirge. Geologie, 14, 242-
253.
- - 1966. Uber Klufte II. Geologie, 15, 896-941.
CRUIKSHANK, K.M., ZHOA, G. 8s JOHNSON, A. M. 1991. Analysis of minor fractures associated with
joints and faulted joints. Journal of Structural Geology, 13, 865-886.
GROSS, M. R. 1993. The origin and spacing of cross joints: examples from the Monterey formation,
Santa Barbara Coastline, California. Journal of Structural Geology, 15, 737-752.
HODGSON, R. A. 1961a. Regional study ofjointing in Comb Ridge-Navajo Mountain Area, Arizona
and Utah. Bulletin of the American Association of Petroleum Geologists, 45, 1-38.
186 J.C. ROBERTS
- 1961b. Classification of structures on joint surfaces. American Journal of Science, 259, 493-
-
502.
HANCOCK, P. L. 1985. Brittle microtectonics: Principles and practice. Journal of Structural Geology,
7, 437-457.
KULANDER, B. R., BARTON, C. C. & DEAN, S. L. 1979. The application offractography to core and
outcrop fracture investigations. Morgantown Energy Technology Centre, METC/SP-79/3.
OWEN, T. R. 1967. 'From the South': a discussion of two recent papers in the proceedings.
Proceedings of the Geologists" Association, 78, 595-599.
POLLARD, D. D. & AYDIN, A. 1988. Progress in understanding jointing over the past century.
Bulletin of the Geological Society of America, 100, 1181-1204.
RECHES, Z. 1976. Analysis of joints in two monoclines in Israel. Bulletin of the Geological Society of
America, 87, 1654-1662.
RIVES, T., RAZACK, M., PETIT, J. P. & RAWNSLEY, K. D. 1992. Joint spacing: analogue and
numerical simulations. Journal of Structural Geology, 14, 925-938.
ROBERTS, J. C. 1961. Feather-fracture, and the mechanics of rock-jointing. American Journal of
Science, 259, 481-492.
- 1974. Jointing and minor tectonics of the vale of Glamorgan between Ogmore-by-Sea and
-
Lavernock Point, South Wales. Geological Journal, 9, 97-114.

SYME-GASH, P. J. 1971. Surface features relating to brittle fracture. Tectonophysics, 12, 34-391.
WATERS, R. A. & LAWRENCE, D. D. 1987. Geology of South Wales Coalfield, Part III, the country
around Cardiff, 3rd Edition. Memoirs of the British Geological Survey, sheet 263 (England and
Wales).
WILSON, D., DAVIES, J. R., FETCHER, C. J. N. & SMITH, M. 1990. Geology of the country around
Bridgend. Explanation of the 1:50 000 sheet. British Geological Survey, sheet 262, England and
Wales.
WOBBER, F. J. 1966. A Study of the depositional area of the Glamorgan Lias. Proceedings of the
Geologists' Association, 77, 127-137.
Boston Society of Natural Historical Proceedings, 27, 163-183.
The expression of hydraulic fracturing in rocks and

sediments
J. W. C O S G R O V E
Department of Geology, Imperial College of Science, Technology &
Medicine, London SW7 2BP, UK
Abstract: The orientation and spatial organization of hydraulic fractures in rocks and
sediments is determined by the magnitude of the differential stress, the orientation of the
principal stress axes and the intrinsic properties of the rock, particularly its cohesive
strength. Depending on the values of these various parameters the expression of
hydraulic fracturing in a rock or sediment can be extremely varied ranging from
conjugate shear fractures through parallel tension fractures to randomly oriented
tension fractures and fluidized sediments.
The influence of a rock's properties on its response to hydraulic fracturing is
illustrated by considering the mechanics of formation of sedimentary dykes.
The theory of brittle failure and hydraulic fracturing are discussed in most texts on
structural geology (e.g. Price 1966, Phillips 1972; Price & Cosgrove 1990) and the reader is
referred to them. Only a brief review of aspects of the topic which bear directly on the
ideas discussed in this article is given here.
The stress conditions necessary for the formation of shear and tensile failures are shown
in Fig. 1. In this figure the Navier-Coulomb criteria of shear failure and the Griffith
criteria of tensile failure are expressed graphically, and the stress states are represented by
M o h r circles.
In order for tensile failure to occur the Mohr circle must touch the brittle failure
envelope at the point 7- = 0, 0"3 = T, where T is the tensile strength of the rock. It follows
from the geometry of the failure envelope (Fig. 1) that this can only occur if the diameter
0"1 0"1
a) 0 ~
x2+4Tcrn - 4T =
Fig. 1. The graphical expression of the two brittle failure criteria and Mohr circles representing stress
states capable of causing tensile (a) and shear (b) failures.
188 J.w. COSGROVE
"cj i I
D,
0-1 O'1
0"3 ~ ~ 0"3 0"3 ~ / ~ / ~-0"3
(3"1 0"1
0"1 0"1
t t
0"3 ~ ~ 0"3 0"3 ~ ~ 0"3
b
F?, 1
0"1 0"1
Fig. 2. (a) Mohr circles (i-iv) representing a range of stress states which will lead to tensile failure.
The Mohr 'circle' iv that represents hydrostatic stress is a point. (b) (i-iv) Patterns of tensile fractures
generated, respectively, by the four stress states shown in (a).
HYDRAULIC FRACTURING 189
Fig. 3. (a) A regular array of tensile fractures exposed on a bedding plane in Carboniferous
sandstone at MiUook, North Cornwall. The alignment of the veins indicates that a significant
differential stress existed during fracturing. The diameter of the coin is 3 cm. (lo) Devonian sandstone
cut by randomly orientated tension fractures which indicate a hydrostatic stress state during
fracturing, at Combe Martin, North Devon. The diameter of the coin is 3 era.
190 J.W. COSGROVE
of the Mohr circle (the differential stress 0 - 1 - 0"3) is small, i.e. < 4T. Mohr circles
representing four stress states that satisfy these conditions are shown in Fig. 2, and in the
following section the orientations of the fractures that form in response to these stresses
are discussed.
The orientation of tensile fractures

Consider the orientation of fractures that form in response to the range of differential
stress states represented by the Mohr circles shown in Fig. 2, all of which satisfy the
conditions for tensile failure. The differential stresses vary from just < 4T (circle i, Fig. 2)
to zero ('circle' iv, Fig. 2). Note that when the differential stress is zero, the stress state is
hydrostatic and the Mohr circle collapses to a point.
Tensile fractures form parallel to the maximum principal compressive stress, 0-1, i.e. they
open in the direction of the minimum principal stress, 0-3 (Fig. 1a). Thus, in the stress state
represented by Mohr circle i in Fig. 2, which has a relatively large differential stress, there
is a definite direction of easy opening for the tensile fractures, i.e. parallel to 0-3. The
fractures would therefore exhibit a marked alignment normal to this direction (Fig. 2b, i).
However, for the stress states represented by the Mohr circles ii-iv, the differential stress
becomes progressively smaller until, for the hydrostatic stress represented by Mohr 'circle'
iv, the differential stress is zero. In a hydrostatic stress field the normal stress across all
planes is the same and there is therefore no direction of relatively easy opening for the
tensile fractures. Thus, the fractures will show no preferred orientation and, if they are
sufficiently closely spaced and well developed, will produce a brecciation of the rock (Fig.
2b, iv). It is to be expected, therefore, that as the differential stress becomes progressively
lower (stress states i-iv, Fig. 2), so the tendency for the resulting tensile fractures to form a
regular array normal to a3 decreases. Tensile fracture systems ranging from well aligned
fractures to randomly oriented fracture arrays are to be expected in rocks, and field
observations (Fig. 3a & b), support this idea.
The role of fluid pressure in tensile failure

It is generally acknowledged that the state of stress in the Earth's crust tends to be
compressional and that true tensile stresses are uncommon. For example, in a tectonically
relaxed region of the crust, i.e. a region where the overburden is the main source of stress,
both the vertical and horizontal stresses are compressional. The relationship between the
two depends on the boundary conditions and the rock type.
If, for example, the boundary conditions are such that horizontal strains are prevented
by the constraints of the rock mass surrounding the area of interest, then it can be shown
(see Price 1966, p. 69) that the relationship between the vertical and horizontal stresses is
given by:
0-H = 0-v/(m- 1), (1)
where m is Poisson's number - the reciprocal of Poisson's ratio. Under these boundary
conditions the vertical stress will be 0-1 and the principal stresses will be compressive at all
depths in the crust. It can be seen from equation 1 that the differential stress (0-v - o-n) will
depend on the depth of burial and the material properties of the rock.
HYDRAULICFRACTURING 191
"s SJ'~ J"
s
t S
tI
. _ ...t -
', 2
s S a"
/ I~3 (l~l_PH2-~O) 0"1 O'v

(0"3-PH20)
Fig. 4. Mohr circle representation of four stress states (solid line circles 1-4) which will not cause
failure. The effect of fluid pressure is to move the stress state to the left by an amount PH20. If the
new circle (dashed) touches the failure envelope hydraulic fracturing occurs. Whether this failure is
by shear or tension depends on the differential stress.
The stress states in four different rocks at a particular depth are represented in Fig. 4 by
four Mohr circles. They have the same overburden stress, el, and their horizontal stresses
are determined by their material properties. Because the principal stresses are
compressional none of these stress fields can cause tensile failure. However, field and
borehole core observations show that tensile failure is common in sediments subjected to
the 'zero horizontal strain' boundary conditions mentioned above and this dilemma has
been resolved by arguing that the stress states shown in Fig. 4 (and represented by the
solid-line Mohr circles 1 4 ) can be modified by fluid pressure. The build up of fluid
pressure during burial acts so as to oppose the compressive stresses generated by the
overburden and causes them to be reduced to the effective stresses (0-1 - - P n 2 o ) and (0-3 -
PH20). Thus, the Mohr circles are moved to the left by an amount equivalent to PH20 and
by this means, if the differential stress is < 4T, can be brought into contact with the tensile
failure envelope at the point 0-3 = - T , r = 0.
Failure caused in this way is known as hydraulic fracturing and the type of fracture that
develops depends on the differential stress. For example, if the fluid pressures in rocks with
the stress states represented by the solid line Mohr circles in Fig. 4 were sufficiently large to
cause failure, rock 1 would fail by shear failure and rocks 2-4 by tensile failure, as
indicated by the dashed Mohr circles. As mentioned earlier, as the differential stress is
reduced, so the tendency for the tensile fractures to form normal to 0-3 decreases until in
the limit when 0-1 -- 03, the fractures are randomly oriented (Fig. 2b, iv).
It is a common misconception that the result of hydraulic fracturing in rock is the
formation of randomly oriented tensile fractures and the generation of breccia textures
(Fig. 3b). The above discusion shows that, depending on the differential stress, the
expression of hydraulic fracturing can range from randomly oriented tensile fractures
through aligned tensile fractures to shear fractures.
192 J.W. COSGROVE
Fig. 5. Diagrammatic representation of fluidization. The build up of fluid pressure in the pores of an
uncemented granular sediment (a) reduces the normal stress acting across the grain contacts, thus
facilitating flow (b). The normal stress may be reduced to zero and the grains may move apart (e).
Fluidization
Now consider the effect of hydraulic fracturing on uncemented sediments with no intrinsic
cohesion, such as an uncemented sandstone lens in a shale. A material such as this has no
strength and therefore cannot support a differential stress. Thus, the stress state within it
will be close to hydrostatic. If the fluid pressure is such that hydraulic 'fracturing' can
occur then the grains of the sediment will simply move apart slightly and the sediment will
fluidize. Thus, an additional diagram should be added to Fig. 2b showing the effect of
hydraulic fracturing under conditions of hydrostatic stress in a material with zero cohesion
such as an uncemented sediment (Fig. 5).
Evidence for two different modes of tensile failure occurring at the same depth and at
the same time in different 'rocks' is provided by sandstone dykes frequently found cross
cutting shales (Fig. 6a). Because of the electrostatic charges between clay particles making
up a shale there is an intrinsic cohesion in the sediments from the moment they are
deposited. This cohesion will increase with compaction and cementation. However, sand
(a) ]I.2 ' Sandstonel I

I
i I
I ,
I
m m
I
I
I
I
I
(b) (c) I
T
Fig. 6. (a) Diagrammatic representation of a folded sandstone dyke cutting the Mercia mudstone at
Watchet, North Somerset. (b) Unfolding of the dyke to determine the thickness ll, of the bed at the
time of dyke formation. (e) Original thickness of the sediment, 10.
lenses interbedded with the shale will remain cohesionless until the processes of
cementation are initiated. Thus, during the early stage of burial the shales will possess a
cohesion and the sandstones will not. If, during this stage, the fluid pressures become high
enough to cause hydraulic fracturing in both sediments, the shales, because of their
intrinsic cohesion, will be able to sustain a differential stress. If the differential stress is
< 4T of the shale, tensile failure will occur and the resulting fractures will be vertical. At
the same time the effect of hydraulic fracturing on the sand lenses which have no cohesion
will be to cause fluidization. Fluidized sand will then be injected into the vertical fractures
in the shales wherever these fractures intersect the sandbodies.
Subsequent compaction of the shales will cause these originally vertical sand dykes to
become buckled (Fig. 6a). The thickness, ll, of the shale when the dykes were intruded can
be calculated by unfolding the dykes (Fig. 6b). If the present state of compaction of the
shale (Fig. 6a) can be determined then its original thickness, 10, can be calculated together
with the stage in burial when hydraulic fracturing occurred.
Tensile failure in anisotropic rocks

The general condition for tensile failure by hydraulic fracturing along any plane is that the
fluid pressure, PH20 must be equal to or greater than T, the tensile strength of the rock
194 J . W . COSGROVE
(a)
(b)
(c)
Fig. 7. Bedding-parallel veins indicating hydraulic fracturing along bedding with fluid pressures in
excess of the lithostatic load, (a) calcite (beef) vein from Lulworth Cove (Cosgrove 1993, fig. 2a) and
(b) satin spar veins from the Mercia mudstone at Watchet Harbour, North Somerset. The bedding is
weakly developed and forms anastomosing, sub-horizontal planes of weakness which are exploited
by the veins. The diameter of the coin is 2 cm. (e) Vertical tension veins with small offshoots along
bedding cutting a Carboniferous sandstone in the Rusey Fault zone, Boscastle, North Cornwall.
normal to that plane and the normal stress acting across it, i.e.
PH~O > T + 0". (2)
In an isotropic rock this condition is first met along the plane normal to the least
compressive principal stress, 0"3 (Fig. la). However, in an anisotropic rock, i.e. one in which
the tensile strength varies in different directions, this is not necessarily so.
Consider a well-laminated shale with a marked planar anisotropy parallel to bedding.
The tensile strength in the direction normal to bedding, Tn, will be considerably less than
that parallel to it, Tp. From equation 2, the conditions for tensile failure parallel and
normal to bedding by hydraulic fracturing are:
PH20 > Tp + O"H, (3)
PH20 > T. + o"v, (4)
respectively, recalling that 0-v = 0-1 and 0-H = 0"3.Which of these two conditions is satisfied
first depends upon the relative values of the differential stress (0"1 - 0"3) and the difference
between the two tensile strengths (Tp - 7",). If (0"2 - 0"3) > ( T p - T n ) then, as the fluid
196 J.W. COSGROVE
pressure increases, equation 3 is satisfied first and the resulting hydraulic tensile fractures
form normal to a3, i.e. are vertical. If the differential stress is less than the difference
between the two tensile strengths then the condition represented by equation 4 will be
satisfied first and hydraulic tensile fractures will form parallel to bedding, i.e. normal to al.
Field evidence for these fractures can be found in the form of bedding parallel, horizontal
veins of fibrous calcite (beef) and gypsum (satin spar) (Fig. 7a & b). The fibres are vertical
and indicate the direction of opening of the veins. These veins show that fluid pressures
equal to or in excess of lithostatic pressure can develop (Cosgrove 1993).
If the condition (a3 + Tp) = (al + Tn) occurs then the fluid pressure required to form
hydraulic fractures parallel and normal to the bedding will be the same and there would be
an equal likelihood that both sets of fractures will form. An example of this is illustrated in
Fig. 7c, which shows vertical viens of quartz with small horizontal offshoots forming along
the bedding.
Conclusions
Because the state of stress in the Earth's crust is generally compressive it is usually
necessary for high fluid pressures to be present in the rock in order for tensile failure to
occur. The orientation of the tensile fractures can vary between parallel arrays of fractures
which form normal to the least principal stress, a3, when the differential stress is relatively
large (but still < 4T of the rock) to randomly oriented fractures which are generated when
the state of stress in the rock is approximately hydrostatic, i.e. when the differential stress
approaches zero.
Hydraulic failure of a rock with a very low cohesive strength such as an uncemented
sandstone, may lead to fluidization. Once fluidized the sediment may be injected along
hydraulic fractures in the surrounding, more coherent rocks, resulting in the formation of
vertical sedimentary dykes which buckle during subsequent compaction and dewatering.
The stage in the burial history at which the dyke formed (i.e. when hydraulic fracturing
occurred) can be determined by 'unfolding' the dykes to obtain the thickness of the
injected bed at the time of dyke formation.
Although tensile failure generally occurs along a plane normal to the least principal
stress, a3, it is possible for such failure to occur normal to al if the rock is mechanically
anisotropic. For example, in a sedimentary succession in which the main source of stress is
the overburden, failure will occur normal to O"1 when the differential stress is less than the
difference between the tensile strength of the rocks parallel and normal to bedding.
References
COSGROVE,J. W. 1993. The interplay between fluids, folds and thrusts during the deformation of a
sedimentary succession. Journal of Structural Geology, 15, 3-5, 491-500.
PHILLIPS, W. J. 1972. Hydraulic fracturing and mineralisation. Journal of the Geological Society of
London, 128, 337-359.
PRICE, N. J. 1966. Fault and Joint development in Brittle and Semi-Brittle Rock. Pergamon Press,
Oxford.
- - & COSGROVE,J. W. 1990. Analysis of Geological Structures. Cambridge University Press,
Cambridge, UK.
Spatial change in joint geometry in the Chalk of eastern

England
A. M . G O O D W I N
Department of Earth Sciences, Cambridge University, Downing Street,
Cambridge, CB2 3EQ, UK
Abstract: Prominent NE and NW striking joint sets previously recognized in southern

England have been traced northwards in the Chalk of eastern England. Both sets change
their geometry northwards, defining two joint continua, from steep conjugate hybrid
joints in Norfolk to a sub-vertical single joint set in Humberside. This spatial change is
consistent both with the observed northward increase in rock strength of the Chalk and
a possible northward decrease in differential stress across eastern England.
Fractures, particularly joints, form one of the most common rock structures. Although
studied for over a century (Pollard & Aydin 1988) it is only recently that their usefulness
within the field of brittle microtectonics has been recognized (Hancock 1985). Despite their
sometimes kinematically ambiguous characteristics many studies have highlighted the
valuable nature of fractures in microtectonic investigations, particularly for inferring
regionally significant stresses and their relative magnitudes. Such studies show the
contribution that systematic joints can make in establishing past and contemporary stress
regimes, especially in stable platform regions of mild deformation (Engelder & Geiser
1980; Engelder 1982; Eyal & Reches 1983; Hancock et al. 1984; Bevan 1985; Caputo 1991;
Zoback 1992). Despite the increase in use of brittle-microtectonic features the term 'joint'
has been defined in a variety of ways (e.g. Dunne & Hancock 1994) and the genetic
classification of joints and fractures is still somewhat disputed. For the purposes of this
paper, the following definition of the term 'joint' will be used - a planar discontinuity
which at the scale of observation possible in the field shows no discernible evidence for
offset related to shear, dilation, or shortening (altered from the definition given by
Hancock 1985).
Previous work has demonstrated the predominance of two joint strikes, NW-SE and
NE-SW, within the Mesozoic and Cenozoic lithologies of eastern and southern England
(Norfolk, Essex, Kent and Dorset), northwest France, and Belgium (Toynton 1983; Bevan
1985; Balson & Humphreys 1986; Bevan & Hancock 1986; Vandyke et al. 1991). Joints
have also been documented in the Chalk of Yorkshire (Patsoules & Cripps 1990) in a
similar NE and NW pattern, but no such study has been undertaken in Lincolnshire.
Bevan (1985) regarded the two identified joint sets as coeval (questionably the NW set was
the older) and due to a Miocene-Pliocene deformation phase associated with Alpine
compression. Balson & Humphreys (1986) assigned a Pleistocene age to their joint systems
whilst Vandycke (pers. comm.) believes the joints to be of a late Cretaceous to early
Tertiary age. Toynton (1983) makes no comment on the age and tectonic relationships of
the joints observed in Norfolk. There are several possible explanations for the formation
of the two joints sets within the tectonic and intra-plate stress regimes of northwest Europe
since the end of the Cretaceous: Palaeocene-Eocene uplift of NW Britain, Tertiary
198 A.M. GOODWIN
(a)
~ Location of
the study area
C Cambridgeshire
H Humberside
L Lincolnshire
N Norfolk
" i []~ Upper Cretaceous Chalk
700 O00m 600 O00m
. - ~ ' (b)
Permo- ~
Triassic I
<: l
~,tt~i~ Jurassic .~
Fig. 1. (a) Location map of the study area within the counties of eastern England. (b) Simplified
geological map of the study area showing the outcrop of the upper Cretaceous Chalk.
CHANGING JOINT GEOMETRIES IN CHALK 199
tectonic activity in the North Sea, a 'neotectonic' origin related to the present day stress
field (Hancock & Engelder 1989). Structural and stratigraphic controls are known to occur
in the chalk in southern England (Ameen 1990; Clayton 1990) affecting the joint patterns
formed. Toynton did not address this point in his study of Norfolk. Recent re-analyses of
existing gravity and magnetic data by the British Geological Survey have also shown the
existence of deep seated lineaments (NW-SE, NE-SW, ENE-WSW, E-W) within the
concealed Caledonide basement of eastern England (Pharaoh et aI. 1987; Lee et al. 1990,
1991). No manifestation of these structures is evident, superficially, in the Mesozoic and
Cenozoic units.
The fracture study reported here covers the Upper Cretaceous Chalk of eastern
England, the counties of Norfolk, Lincolnshire, north Cambridgeshire and south
Humberside (Fig. 1). Prolific joints are visible in an essentially undeformed Mesozoic
sedimentary sequence that is fiat lying or gently dipping (< 10~ to the NE. The study
reviews the occurrence of joint patterns in eastern England extending the existing studies
from the adjacent areas. The collection of outcrop data has allowed a classification of
identified systematic joints into individual joint sets (Hancock 1985). The analysis of these
data have allowed the parallelism and continuity to be tested between the eastern England
joints and those further to the south and any previously unrecognized controls and
variations on fracturing to be defined. Insights can also be made into the formation
environment and age of the joint sets and their previous regional tectonic interpretations.
Methods and techniques

The field data collected from the study area comprise joint orientation, spacing, geometry,
architecture, dihedral angle, continuity, dilation and surface morphological features. The
orientation data of Toynton (1983) have augmented these new data particularly for
recently infilled quarries. Bias in orientation readings due to sampling procedure was
minimized by collecting readings from faces of varying orientation. Toynton has shown
that 50 readings are sufficient to identify the significant trends within the data set. All
orientation data have been analysed using the 'STEREOPLOT' software (Mancktelow
1989) and cluster analysis software (SPSS 1990).
Based on the relationships with other planes, and the similarity of the criteria stipulated
above, systematic joint sets were identified and individual joint planes were assigned to a
joint set whilst at a field locality. Equal area stereonet scatter plots (dip and dip azimuth)
of both a standard lower hemisphere projection and a rotated lower hemisphere projection
(rotated about a horizontal axis of minimum point density) confirmed the field-based
observations. Single-linkage cluster analysis producing a heirarchy of cluster solutions
based on the squared Euclidean distance between variables (data points) enabled the
significant cluster(s) to be identified from any background noise occurring in each set at
every locality. This process also identified a mean plane for each joint set. Once the set was
identified, statistical parameters such as the standard distribution could be calculated.
Distribution analyses related to the point density on the sphere, primarily the Bingham
analysis (Cheeney 1983; Fisher et al. 1987) confirmed the mean plane.
This paper concentrates on presenting the variations seen in the field criteria and
characteristics of the joint sets identified, not the secondary statistical differentiation of the
data and joint sets. Average planes and joint set data are presented for representative
localities.
200 A.M. GOODWIN
Grasby
Fox Coven i (65)
(83)
(44)
(77)
Tefford Hill 0 20
" ~ (73) I I
Candlesby Hill Km
(58)
Hunstanton BurnhamOvery
North Ormsby' _ r . ~ (88) Wells-Next-the-Sea
(68) (122)
Mill Hilt
(72)
Lower Hcllesdon
(41)
Snettisham
(108) #
Gayton
(64)
Whittingto 1
(40)
Non wo#:'" ") Soo Pieken

(41)
m Caistor
(40) .~/ (118)
Fig. 2. Lower hemisphere equal area projections of collected data. For each locality scatter plots of
poles to joint planes are shown with an accompanying plot of the mean plane for the dominant joint
sets. The NW joint set is highlighted in each case, the number represents the sample size (N).
Joint patterns
Joint sets, represented by their mean plane, are shown in Fig. 2. The existence of two
clusters on the stereonets is clear, correlating with the division of joint planes into two
similar sets in the field (Fig. 3). The summary plots mostly show steeply dipping N W and
NE striking sets of similar orientation to those in southern England. This relationship is
clearest on a horizontal surface perpendicular to the joint planes effectively eliminating the
varying geometric styles of set members. Both joint sets comprise steep conjugate and
vertical joints, everywhere consistent with bedding. The N W set is generally, but not
consistently, the more prominent of the two (Table 1). Fractures showing small amounts
of dip-slip displacement (< 15 cm) form a minor division of the N W set at only a few
localities. N o surface morphological features are seen on any joint planes. Cross-cutting
and abutting relationships between (and within) the two sets are ambiguous. Little
variation in jointing is seen between the different stratigraphic horizons of the Chalk
202 A.M. GOODWlN
CHANGING J O I ~ F GEOMETRIES IN CHALK 203
Fig. 3. Photographs showing jointing in vertical sections from Humberside, Lincolnshire and
Norfolk. Examples of structures are shown in each photograph where: j, joint plane; b, bedding
plane; c, conjugate joint plane (steeply inclined); e, extension joint plane (vertical). (A) Dominance of
vertical joints (dip > 85~ on a vertical SW-NE face in Burnham, south Humberside (hammer is
40 cm). (B) Vertical SW-NE cliff face showing an increase in the number of steeply inclined joints at
Mill Hill, south Lincolnshire (pen is 15 cm). (C) Vertical SW-NE cliff face showing the dominance of
steeply inclined joints at East Harling, Norfolk. (D) Close-up of the conjugate nature of steeply
inclined joints at East Harling, Norfolk (compass is 10cm). (E) Dominant joint traces on a
horizontal surface are the NW trending set, West Runton, North Norfolk coast. Scale bar graded in
1 and 0.5cm.
(Upper, Middle and Lower). Minor discrepancies in the density of jointing can be seen
between different localities, higher intensities of jointing being seen in 'harder' chalk
horizons (in the present study mainly in localities in the Middle Chalk). Likewise,
digressions in the nature of jointing - orientation, architecture and geometry - are also
apparent but again are only slight.
A more significant difference in joint set characteristics can be observed on a regional
scale. Although both the N W and NE striking joint sets are apparently continuous
through the eastern England Chalk, this continuity hides a subtle, but equally important,
change in their detailed geometry across the area. This change is illustrated with respect
to the more prevalent N W striking joint set (Fig. 4a). The N E striking set follows an
equivalent pattern.
1. The southern area. In Norfolk, joint traces on a vertical N E - S W section typically

show a conjugate, steeply inclined form with a dihedral angle of between 25 ~ and 45 ~
Vertical joints occur less commonly. Displacement on joint planes is not evident in
204 A . M . GOODWIN
9 .,~ ~ ....~
0 0 0 0 0 0
~L
.o
~ = ~ ~ ~ =
o
L~
[-i
o
~ ~.~
:t0
,~,=o
- ~.~
~ ~~
r~
O
" ~.E
~ o
0,-"
9','-.+ .-.,n __ m ~
O ~ r
r ..o
++',,+"r +
/, s i~ .,.~
LT~ o m
_.~
~ ~'~'~
N~N
r,r
= 0:l
= ~ ~,~
. Z.,..,i
~o=o "'"~
.o= o
. , ~ ~ o~ e~z~
"~ ~ ~ .~, -= ~--

o o"~ ~
~ ,.~ .~ ~
r
t.tj
z,~
. . . . . . =e,i ~.o
' r.+
II ~ 9 9 ,~ ..~ ,..., r~,+- I
9 " " "~ 0"~"~%4

~ I
206 A.M. GOODWIN
the field (excepting the minority of recognizable fractures) and microscope analysis
has not proved effective in detecting any shear displacement. Joint traces terminate at
horizontal joints, with an overall length of c. 20-40 cm. Ten per cent of the vertical
joints have a continuous trace across these horizontal joints of up to several metres.
Dilation across joint planes is small ( < 2 mm), except for the more continuous traces
where it can reach 5-6 cm.
. The northern area. By contrast, vertical N E - S W sections across the joint sets in south
Humberside/north Lincolnshire show mainly vertical joints ('H' and T architectural
traces of Hancock 1985) with few steeply inclined planes occurring. As in Norfolk,
the vertical joints are truncated at the horizontal joints, although c. 10% are more
continuous and these joint traces show the greatest dilation.
The origin and stress implications o f the joint patterns

The joints of Norfolk (the southern area) are interpreted as the conjugate hybrid joints of
Hancock (1985) showing typical 'x' and 'y' architectural traces. They would have formed
in a stress field with the maximum principal stress (0-1) vertical and the minimum principal
stress (0-3) horizontal and perpendicular to joint strike (principal stresses are assumed to be
compressive when positive). The effective stress component acting normal to the failure
plane (when the dihedral angle is < 45 ~ would be < 0 ( Price & Cosgrove 1990), allowing
for an element of shear along the fracture plane as well an element of extension across the
plane - hence the term 'hybrid'. The lack of detection of any component of shear may
simply be a factor of the absence of any suitable attribute to record any occurrence of
shear (fractured grains, displaced microscopic veins).
The Humberside joints are interpreted as extension joints (dihedral angle < 10 ~ formed
perpendicular to the minimum principal stress (0-3). The effective minimum stress would,
necessarily, have to be < 0, possibly achieved through the effect of pore-fluid pressure
counteracting a compressive 0-3. Any fluid present in a sediment will, on compression,
counteract the applied stress because of its incompressible nature. This will decrease both
0-1 and 0"3 but not the differential stress (0-z-o3), effectively weakening the Chalk,
instigating failure (Fyfe et aL 1978). If 0-3 is lOW or the hydrostatic pressure is large enough,
the effective 0-3 can become zero or negative giving rise to tensile brittle failure.
There appears to be a transition between the conjugate hybrid joints in the south and the
extension joints in the north, with localities in the intervening south Lincolnshire area
showing both types of trace (Fig. 4a). In both south Humberside and Norfolk, the joints
form a joint continuum rather than a singular joint set. Hancock (1986) suggests that
systematic joints can form either a well-defined fracture set, or they may form a coaxial
spectrum of joints (including both hybrid and extension joints) describing an angular
continuum with a maximum dihedral angle of c. 45 ~ The dominantly conjugate joint set of
Norfolk and the singular set of Humberside are therefore seen as end members of a
transition from south to north through a single joint continuum.
The minor variation observed in jointing within different stratigraphic horizons of the
Chalk reflects the similar N - S control of joint geometry - 'harder' horizons show the same
characteristics as the northern area. Similarly, a control on fracturing by local features and
the reactivation of basement features is not clearly seen. No major structural features,
folds and faults, occur in this area to act as a controlling influence. Other than a superficial
correlation of trend to identified basement lineaments (NW and SE), a lack of information
s l l e ~ stress
northern
[ fen,'elope
extension [/ southern
~ envelope
noflllal stress
9 /~ -"~_._...2oft
con.iuoat~ [ ~ ....
h~0ints [ ~.~...~ard
shear S|l'ffss
norlhern
stress field
,~'- ~ ~ southem
~,NNN,~ ~ stressfield
/ V / # ....... ....
extension I~ /
I ~. c,~
(a) I hybridjoints
high porosity chalk
I 2 3 4
normal stress
(MP;II
Io~ porosity chalk
iion1~11 SIFCSS
Co) (MP:~)
Fig. 5. (a) H y p o t h e s e s - (i) variable rock strength, (ii) variable stress field - for the N-S contrast in
joint geometry (see text for details). (b) Representative Mohr circles for low and high porosity
English Chalk are shown for comparison (from Clayton 1990).
208 A.M. GOODWlN
on the deep structures and a paucity of surface exposure coverage does not allow
conclusive suggestion of a link between the two.
Causes of the spatial change in joint geometry

It is suggested here that there are two possible mechanical explanations for the change
from conjugate to extension joint geometry from Norfolk to south Humberside. These are
explained with reference to Mohr stress diagrams (Fig. 5a). They are not mutually
exclusive and the natural explanation may involve components of both mechanisms.
Variable rock strength hypothesis

The first possibility is that the differential stress field (0-1- 03) remained constant across the
region but that the rock strength increased from south to north. The weaker rock would
have a Mohr failure envelope which lay within that of the stronger rock with a less steep
gradient (smaller angle of internal friction; Fig. 5a). For a range of differential effective
stresses, the resulting stress circle would touch the failure envelope of the weak rock at two
points, giving conjugate hybrid joints, yet touch the envelope for the strong rock at only
one point, giving extension joints. The minimum effective principal stress (0-3) would need
to be < 0.
The northward increase in rock strength and hardness in the Chalk is well known (e.g.
Scholle 1974; Bell et al. 1990). In the mechanically variable Chalk of England induration,
stylolitization and density all increase northwards, whilst permeability and porosity
decrease (Table 2). The Chalk of southern Britain is thus considered to be 'softer' than
that occurring further north in Lincolnshire and Yorkshire. A definition of strength terms
can be found in Mortimore & Fielding (1990). The observed northwards hardening is
thought to be primarily due to a variety of diagenetic factors (Clayton 1983, 1990; Clayton
& Matthews 1987), influenced locally by tectonically induced deformation (Mimran 1975).
Representative failure envelopes for high porosity (soft) and low porosity (hard) English
Chalks are shown, for comparison, in Fig. 5b. Scholle (1974) and Carter & Mallard (1974)
have suggested that the changes in geotechnical properties of the Chalk are due to the
controlling processes of either consolidation ('compression associated with over pressuring
or positive excess pore pressures') or drained compression ('compression controlled by the
rate of application of an overburden in the absence of excess pore pressures') associated
with an increasing burial depth (increasing northwards). Clayton & Matthews (1987),
however, suggest that these processes were insignificant contributors to the confined
deformation and densification of the Chalk and that excess pore fluids were unimportant
(Jones et al. 1984). If excess pore pressure was not a significant factor at burial and
consolidation of the Chalk the minimum compressive stress (0-3) would ;need to be low or
< 0, therefore requiring only a minimal effect of pore pressure to promote extensional
failure.
Variable stress field hypothesis

The second hypothesis to explain the observed joint variation is that the rock strength
remained constant across the region, but that the differential stress (0-1-0-3) decreased
from south to north, Fig. 5a. A large stress circle would touch any Mohr envelope at two
points, giving conjugate joints, whereas a small stress circle would touch the same envelope
CHANGING JOINT GEOMETRIESIN CHALK 209
Table 2. Density, porosity and strength values for the Chalk of eastern England showing the N - S
variation highlighted in Fig. 4b
Yorkshire Norfolk Kent

Dry density (mg m-a)
Bell et al. 1990 2.06-2.14 1.76-2.17 1.44-1.61
Clayton 1983 1.97-2.25 1.54-2.19 -
Absolute porosity (%)

Bell et al. 1990 24.4-26.8 26.6-35.1 40.6-47.0
Scholle 1974 1-10 - 43
Compressive strength (MPa)

Bell et al. 1990 25.6-30.7 13.0-29.2 5.5-9.5
Tensile strength (MPa)

Bell et al. 1990 1.7-2.2 0.8-2.1 0,5-0.7
at only one point in the tensile zone. A greater stress difference in the south would require
a larger effective tr~ than in the north. Because crl is vertical, inferred from the geometry
and architecture of the joint planes, this implies a greater overburden in the south. There is
no obvious geological evidence for this greater overburden. Indeed, the higher porosity
and lower density of the soft 'southern' Chalk has been taken to indicate a lower
overburden to the south. A high differential stress also implies that the rock mass has a
similarly high degree of internal strength, soft rocks flowing to dissipate stress evenly into
a hydrostatic state and to minimize the difference between crland a3 (Engelder 1993).
Increasing the differential stress to the south would then suggest an associated increase in
Chalk strength rather than the observed decrease.
Discussion
It follows that the most likely explanation for the spatial variation in the geometry of
jointing in eastern England is variable mechanical properties of the Chalk, namely an
increase in strength and hardness northwards (Table 2, Fig. 4b). The strength variation
seen is directly correlated to changes in density and porosity and accompanying textural
differences (Fig. 4b) (Clayton 1983; Jones et al. 1984). Variations in initial porosity and
matrix behaviour have been shown to affect the stress-strain path of a sample significantly
(Clayton & Matthews 1987; Clayton 1990). This sedimentological difference between the
northern and southern Chalks may be the underlying factor in the changing hardness/
strength and therefore the joint geometry. Increasing the tensile and cohesive strength of
the Chalk will increase the internal angle of friction and so delimit a broader Mohr failure
envelope (Bell et al. 1990; Clayton 1990) influencing the mechanisms of deformation of the
rock. Despite a paucity of evidence for a northward increase in overburden, however, the
diagenetic gradient within the Chalk from south to north alone infers a greater depth of
burial northwards (Scholle 1974). Variations noted in the lithological horizons of the chalk
reflect the broader scale N-S change and is similarly linked to the hardness of the
individual horizon.
Chalk is a marine deposit, the originally deposited sediment having a very high porosity,
210 A.M. GOODWIN
possibly c. 80%, and presumably very fluid-rich. Although the differential stress (0.1- 03)
might have remained constant across the area, the effective stresses would have been
reduced due to an increase in pore pressure with burial, however small a contributing
factor. Extension failure in the north implies an effective 0"3 that is negative or equal to
zero with the pore pressure exceeding the minimum compressive stress by the tensile
strength of the rock. Excess hydrostatic pressures would have been feasible if depositional
and diagenetic fluids were partially trapped by a low permeability layer. Two possible
explanations could account for the increase in pore-fluid pressure within the Chalk:
1. There existed above the Chalk an unknown amount of Upper Cretaceous and early
Tertiary overburden. The Chalk is unconformably overlain in the southern North
Sea by argillaceous Palaeocene sediments. Following early Palaeocene deposition, a
major regression and widespread uplift led to extensive erosion (possibly in the
region of 200m or more) of these and Upper Cretaceous sediments (Ziegler 1987;
Cameron et al. 1992; Cope et al. 1992). The precise amount of Cretaceous and early
Palaeocene sediment that was removed is uncertain. On this hypothesis, the joint sets
would be due to the stress regime before the early Palaeocene uplift of the North Sea,
and therefore be of latest Cretaceous or early Palaeocene age.
2. An effective overburden was provided by later Palaeocene and Eocene sediments
unconformably overlying the Chalk, equivalent, for example, to the late Palaeocene
Lista Formation (Cameron et al. 1992, figs 87 & 88). Post-early Palaeocene
subsidence deposited upper Palaeocene and Eocene basinal sediments up to 2000 m
thick, averaging c. 1000m (Ziegler 1987; Cameron et al. 1992; Cope et al. 1992).
Much of northern, central and western Britain was at a structural high during this
period, shedding sediment eastwards into the North Sea Basin.
However, the extent of onlap of late Palaeocene and Eocene sediments on to onshore
Britain is unclear (Cope et al. 1992). On the hypothesis of significant post-Palaeocene
overburden, the joint sets would then have formed prior to or during the uplift associated
with the Alpine orogeny.
An additional factor in either of the above hypotheses is the relatively late Cretaceous
impermeability of the Chalk, its present porosity and permeability being mostly fracture
related. This impermeability may have been sufficient to form an effective seal without the
necessity of any overlying impervious unit. However, flow within the Chalk could have
been appreciable due to the deposits high original porosity possibly allowing only small
hydrostatic pressures to arise in the Chalk. Even if increased pore pressure was not a
significant factor in the brittle failure of the Chalk a small 0.3, as postulated by Bevan
(1985), may have been low enough to fall into the tensile field of the Mohr diagram by only
a minimal amount of hydrostatic pressure.
The joint geometries in eastern England suggest that the Chalk there was overlain by an
overburden, probably of missing Upper Cretaceous and earliest Tertiary sediments, of
substantial thickness. The vertical application of an overburden would have had the effect
of increasing the vertical principle stress to become 0.1, as implied by the joint geometries.
Bevan (1985) postulated that 150 m of unconsolidated sediment would have been sufficient
to achieve this. Clayton & Matthews (1987)estimated a possible past maximum vertical
effective stress of 2000 k N m -2, equivalent to 200 m of overburden (based on the average
thickness of post-Cretaceous sediments plus 150 m of ice). They note that this estimate
would be greater than all past in situ vertical effective stresses, tectonic and diagenetic
processes hardening even the softest Chalk by some degree. Clayton & Matthews (1987)
suggest that the most likely possibility for the maximum past overburden would be the
deposition of c. 240m of early Tertiary sediments (Thanet sands, Reading Beds and
London Clay). Although it is possible that enough overburden existed prior to the uplift
and erosion during the early Palaeocene, it is more likely that a greater thickness of
overburden of late Palaeocene-Eocene sediments prior to uplift associated with Alpine
compression was responsible for the main episode of jointing in the Chalk.
The effect of the Pleistocene glaciations cannot be ignored, as Bevan (1985) noted, a
thickness of < 150 m of (denser) ice would have the same effect as a sediment pile of
150 m. As can be seen from Fig. 4b, the southern limit of the hard Chalk could correlate to
a glacial ice front(s). It would be unreasonable to assume, especially in eastern England,
that the sequence of glacial overburdens exacted no toll on the underlying solid rock.
However, distinguishing such an effect from that of an earlier sedimentary overburden
would prove difficult. The fact that both joint sets also occur in 'unglaciated' southern
England suggests that their formation probably pre-dates the Pleistocene glaciations, with
possibly no effect other than re-activation being due to any ice overburden.
Conclusions
1. The N E - S W and N W - S E striking joint sets previously recognized in southern
England have been traced northwards in the Chalk of eastern England as far as the
Humber Estuary.
2. Both sets exhibit a change in their joint plane geometry northwards from dominantly
steep conjugate hybrid joints in Norfolk to sub-vertical extension joints in
Humberside.
3. This change might be due to a decreasing differential stress towards the north, but
more likely is an association with the observed northwards increase in rock strength
and hardness in the Chalk.
4. The higher strength, lower porosity and higher density of the northern Chalk may
record a greater overburden here, which may have included impermeable clays to
seal the Chalk. This would generate the high pore-fluid pressures necessary for
extensional failure. Alternatively, if o3 was small only a minimal effect of hydrostatic
pressure would be necessary to cause the formation of the extension joints seen.
5. Stratigraphic and structural controls on jointing are acknowledged as possible
influences on joint geometry but considered to play a minor role superimposed on the
broader regional variation.
This work has been completed whilst in receipt of a three-year postgraduate Natural Environment
Research Council award. I would like to thank Nigel Woodcock and Ben Goodwin (amongst others)
for their thoughts and help towards completion of this manuscript.
References
AMEEN, M. S. 1990. Macro-folding in the Purbeck-Isle of Wight monocline. Proceedings of the
Geologists" Association, 101, 31-46.
BALSON, P. S. & HUMPHREYS, B. 1986. The nature and origin of fissures in the East Anglian
Coralline and Red Crags. Journal of Quaternary Science, 1, 13-19.
BELL, E. G., CRIPPS,J. C., EDMONDS,C. N. & CULSHAW,M. G. 1990. Chalk fabric and its relation to
212 A.M. GOODWIN
certain geotechnical properties. In: Chalk: Proceedings of the International Chalk Symposium.
Thomas Telford, London, 187-194.
BEVAN, T. G. 1985. A Cenozoic tectonic history of Southern England inferred from mesofractures.
PhD Thesis, University of Bristol, UK.
- & HANCOCK,P. L. 1986. A late Cenozoic regional mesofracture system in southern England
-
and northern France. Journal of the Geological Society, London, 143, 335-362.
CAMERON, T. D. J., CROCBY, A., BALSON, P. S., JEFFERY, D. H., LOTT, G. K., BULAT, J. &
HARRISON, D. J. 1992. The geology of the southern North Sea. United Kingdom Offshore
Regional Report, No. 7. British Geological Survey.
CAPUTO, R. 1991. A comparison between joints and faults as brittle structures used for evaluation of
the stress field. Annales Tectonicae, 5, 74-84.
CARTER, P. S. • MALLARD, D. J. 1974. A study of strength, compressibility and density trends
within the Chalk of south-eastern England. Quarterly Journal of Engineering Geology, 7, 43.
CHEENEY, R. F. 1983. Statistical Methods in Geology for Field and Laboratory Decisions. Allen &
Unwin, London.
CLAYTON, C. R. I. 1983. The influence of diagenesis on some index properties of Chalk in England.
G~otechnique, 33, 225--241.
1990. The mechanical properties of the Chalk. In: Chalk: Proceedings of the International
Chalk Symposium. Thomas Telford, London, 213-232.
- & MATTHEWS, M. C. 1987. Deformation, diagenesis and the mechanical behaviour of Chalk.
-
In: JONES, M. E. & PRESTON, R. M. F. (eds) Deformation of Sediments and Sedimentary Rocks.
Geological Society, London, Special Publication, 29, 55-62.
COPE, J. W., INGHAM, J. K. & RAWSON, P. F. 1992. Atlas of Palaeogeography and Lithofacies.
Memoir of the Geological Society, London, 13.
DUNNE, W. M. & HANCOCK, P. L. 1994. Palaeostress analysis of small-scale brittle structures. In:
HANCOCK, P. L. (ed.) Continental Deformation. Pergamon Press, Oxford, 101-120.
ENGELDER, T. 1982. Is there a genetic relationship between selected regional joints and
contemporary stress within the liothosphere of North America? Tectonics, 1, 161-177.
- - 1993. Stress Regimes in the Lithosphere. Princeton University Press, N J, USA.
- & GEISER, P. A. 1980. On the use of regional joint sets as trajectories of palaeostress fields
-
during development of the Appalachian Plateau, New York. Journal of Geophysical Research,
85, 6319-6341.
EYAL, Y. & RECHES, Z. 1983. Tectonic analysis of the Dead Sea Rift regions since the late-
Cretaceous based on meso-structures. Tectonics, 2, 167-185.
FISHER, N. I., LEWIS, T. L. & EMBELTON, B. J. J. 1987. Statistical Analysis of Spherical Data.
Cambridge University Press, Cambridge, UK.
FYFE, W. S., PRICE, N. J. & THOMPSON, A. B. 1978. Fluids in the Earth's Crust. Elsevier, Amsterdam.
HANCOCK, P. L. 1985. Brittle micro-tectonics: principles and practice. Journal of Structural Geology,
7, 437-457.
- - 1986. Joint spectra. In: NICHOL, I. & NESBITT, R. W. (eds) Geology in the Real World- The
Kingsley Dunham Volume. Institution of Mining and Metallurgy, London, 155-164.
- - d~ ENGELDER,T. 1989. Neotectonic joints. Bulletin of the Geological Society of America, 101,
1197-1208.
--, AL KADHI, A. & SHA'AT, N. A. 1984. Regional joint sets in the Arabian platform as
indicators of intra-plate processes. Tectonics, 3, 27-43.
JONES, M. E., BEDFORD,J. d~ CLAYTON,C. 1984. On natural deformation mechanisms in the Chalk.
Journal of the Geological Society, London, 141, 675-683.
LEE, M. K., PHARAOH, T. C. & GREEN, C. A. 1991. Structural trends in the concealed basement of
eastern England from images of regional potential fields. Annales Socidtd G~ologique Belgique,
114, 45-62.
, PHARAOH, Z. C. & lOPER, N. J. 1990. Structural trends in central Britain from images of
gravity and aeromagnetic fields. Journal of the Geological Society, London, 147, 241-258.
MANCKTELOW, N. 1989. STEREOPLOT, Version 1.2 Faceware Software, FaceIt REsources.
MIMRAN, Y. 1975. Fabric deformation induced in Cretaceous Chalks by tectonic stresses.
Tectonophysics, 26, 309.
MORTIMORE, R. N. & FIELDING, P. M. 1990. Relationships between texture, density, and strength of
Chalk. In: Chalk: Proceedings of the International Chalk Symposium. Thomas Telford, London,
109-132.
PATSOULES, M. G. & CRIPPS, J. C. 1990. Survey of macro- and micro-fracturing in Yorkshire chalk.
In: Chalk: Proceedings of the International Chalk Symposium. Thomas Telford, London, 87-94.
PHARAOH, T. C., MERRIMAN, R. J., WEBB, P. C. & BECKINSDALE, R. D. 1987. The concealed
Caledonides of eastern England: preliminary results of a multi-disciplinary study. Proceedings of
the Yorkshire Geologists Society, 46, 355-369.
POLLARD, D. n. & AYDIN, A. 1988. Progress in understartding jointing over the past century.
Bulletin of the Geological Society of America, 100, 1181-1204.
PRICE, N. J. & COSGROVE, J. W. 1990. Analysis of Geological Structures. Cambridge University
Press, Cambridge, UK.
SCHOLLE,P. A. 1974. Diagenesis of upper Cretaceous Chalks from England, Ireland and the North
Sea. In: Hsu, K. J. & JENKYNS,H. C. (eds) Pelagic Sediments: On Land and Under the Sea.
Special Publication of International Association of Sedimentologists, 1, 177-210.
SPSS 1990. SPSS for the Macintosh, Version 4.0 SPSS Inc., Language Systems Corporation,
Chicago.
TOYNTON, R. 1983. The relationship between fracture patterns and hydraulic anisotropy in the
Norfolk Chalk, England. Quarterly Journal of Engineering Geology, 16, 169-185.
VANDYCKE, S., BERGERAT, F. dk DUPUIS, Ch. 1991. Meso-Cenozoic faulting and inferred
palaeostresses in the Mons Basin, Belgium. Tectonophysics, 192, 261-271.
ZIEGLER, P. A. 1987. Late Cretaceous and Cenozoic intra-plate compressional deformtion in the
Alpine foreland - a geodynamic model. Tectonophysics, 137, 389-420.
ZOBACK,M. L. 1992. First- and second-order patterns of stress in the lithosphere; the world stress
map project. Journal of Geophysical Research, 97B, 11 703-11 728.
From Ameen, M. S. (ed.), 1995, Fractography." fracture topography as a tool in fracture mechanics and stress analysis
Factors controlling joint spacing in interbedded

sedimentary rocks: integrating numerical models with
field observations from the Monterey Formation, USA
MICHAEL R. GROSS, 1 M A R K P. F I S C H E R , 2 T E R R Y E N G E L D E R 2
& R O Y J. G R E E N F I E L D 2
1Department of Geology, Florida International University, Miami, FL 33199,
USA
2Department of Geosciences, Pennsylvania State University, University Park,
PA 16802, USA
Abstract: Local tensile stress normal to a joint is reduced in the vicinity of the joint
because such stresses are not transmitted across free surfaces. This stress reduction
prevents the formation of new joints in the vicinity of existing joints, and thus influences
joint spacing. Lateral extent of this stress reduction shadow increases with joint height,
which corresponds to bed thickness for many sedimentary rocks. The linear correlation
between joint spacing and bed thickness commonly observed in outcrop is a direct result
of this relationship. However, other factors in addition to bed thickness influence joint
spacing. We evaluate these factors through both a review of the Hobbs model for joint
spacing and a 2D finite element simulation of a crack confined to a lithology-controlled
mechanical unit. The stress reduction shadow increases in length with increasing
Young's modulus of the jointing bed, though fracture stress, flaw size, flaw distribution
and extensional strain all interact with bed thickness and elastic properties ultimately to
control joint spacing. One explanation for the observed decrease in joint spacing with
increasing Young's modulus in outcrops of the Monterey Formation is that beds with
higher Young's moduli fail at lower magnitudes of extensional strain.
In a 1967 issue of Geological Magazine, D. W. Hobbs published a paper entitled 'The

formation of tension joints in sedimentary rocks: an explanation', which is commonly
cited as a theoretical explanation for joint spacing in rock masses composed of layers, each
with a different set of elastic properties. Based on the original derivation of Cox (1952), the
Hobbs model provides one theoretical basis for the well-documented linear relationship
between joint spacing and bed thickness in sedimentary rocks (e.g. Price 1966; McQuillan
1973; Ladeira & Price 1981; Huang & Angelier 1989; Narr & Suppe 1991; Gross 1993a).
By relating joint spacing to the Young's modulus of the jointing bed and shear modulus of
the bounding non-jointing beds, Hobbs implies that joint spacing is influenced by the
elastic properties of each bed.
Though both ingenious and remarkably accurate in its prediction of stress distribution,
the Hobbs paper is often misunderstood because it is compact and lacks illustrations. We
present a brief explanation of Hobbs' analysis in an effort to clarify the assumptions and
implications of his model. In addition, we use a 2D finite element numerical simulation of
a crack confined within a bed to describe the stress and displacement fields around the
profile section of a crack in an interbedded sedimentary rock mass. Results of the finite
element modelling are compared with predictions of the Hobbs model. Finally, we discuss
216 M.R. GROSS E T AL.
Fig. 1. Sketch of a lithology-controlled mechanical unit, with systematic joints confined to discrete
beds, and lithologic contacts serving as mechanical layer boundaries. Note that joint height equals
mechanical layer thickness (MLT), and is proportional to joint spacing (from Gross 1993a).
Fig. 2. Photograph of steeply dipping limestone beds in the Monterey Formation exposed at Lion's
Head along the Santa Maria coastline of California, demonstrating the relationship between joint
spacing and bed thickness. Tall joints, in thicker beds, are more widely spaced than short joints, in
thinner beds. Note the consistent joint height for each individual mechanical layer.
FACTORS CONTROLLING JOINT SPACING 217
Fig. 3. Photograph of interbedded siliceous (jointing) and mudstone (non-jointing) layers in the
Monterey Formation exposed at Purisima Point along the Santa Maria coastline. The light coloured
siliceous layers contain statistically regularly-spaced joints that terminate at contacts with adjacent
mudstone layers.
the major factors controlling joint spacing and attempt to reconcile these factors with field
observations that some stiffer beds in the Monterey Formation of California have more
closely spaced joints than beds with lower Young's moduli.
Stress reduction near joints

Joint spacing models for rocks commonly focus on an analysis of stress reduction in the
vicinity of joints because the criterion for joint propagation is a threshold stress value,
derived either from the tensile strength test (e.g. Jaeger & Cook 1976) or a measure of
fracture toughness (e.g. Irwin 1957) of the rock bed. The mechanism for stress build up in
these models is layer-parallel stretching, which can occur locally in the extensional region
of a fold or regionally upon uplift and erosion.
Stress perturbations occur in the vicinity of an open joint because the joint is a free
surface, and hence acts as a barrier to the transmission of tensile stress. Consequently,
218 M.R. GROSS E T A L .
Fig. 4. Photomicrograph of a thin section from core taken from Point Arguello Oilfield, offshore
Santa Maria Basin, showing a tensile fracture confined to a quartz chert layer. See text for details.
local crack-normal tensile stress at the joint surface is zero, and increases to the level of
far-field tensile stress at an infinite distance from the joint. New joints form where the local
tensile stress exceeds the fracture stress of the rock. Thus, models for joint spacing
commonly predict the location, and hence spacing, of new joints by evaluating the ratio of
local tensile stress to remote tensile stress as a function of distance from the stress-free
joint (e.g. Lachenbruch 1961; Pollard & Segall 1987).
Bed thickness - joint spacing relationship and sequential infilling

Joints in sedimentary rocks generally fall into two categories, those that terminate
randomly within the rock mass and those that terminate at discrete mechanical layer
boundaries. Lithologic contacts, as well as pre-existing fractures, can serve as mechanical
layer boundaries, thereby dividing the rock mass into discrete mechanical units (Gross
1993a). The analysis presented in this study focuses on joints confined to individual
lithologic beds, which are frequently observed in thin to medium bedded sedimentary
Fig. 5. Location map of the southern Santa Maria coastline showing outcrops where fracture data
were collected along with the photographs from Figs 2-4 (adapted from Sylvester & Darrow 1979).
Sites include Lion's Head (LIO), Purisima Point (PUR) and Lompoc Landing (LOM). Towns are
Santa Maria (SM) and Lompoc (L). SYRF refers to the Santa Ynez River Fault.
sequences (i.e. beds ranging in thickness from c. 1 to 100 cm) composed of diverse rock
types. In such lithology-controlled mechanical units joints extend from the bottom of the
bed to its top so that joint height equals bed thickness, which in turn is proportional to
joint spacing, measured as the orthogonal distance between adjacent parallel joints of a
particular set (Fig. 1). These relationships are demonstrated in the photographs in Figs 2-
4, taken from the Monterey Formation along the central California coastline (Fig. 5). In
the interbedded limestone and shale of Fig. 2, joint height in each mechanical layer is
uniform, and tall joints are more widely spaced than short joints. The outcrop shown in
Fig. 3 consists of a series of well-defined jointing and non-jointing lithologies, in which
joints are confined solely to the light coloured siliceous layers and display a regular
spacing. In contrast, the thin recessive shaley layers act as mechanical layer boundaries
and remain unjointed. An example of a joint confined to a thin quartz chert layer is shown
220 M . R . GROSS E T AL.
in Fig. 4. Note the joint terminates abruptly at the contact with adjacent shale layers.
Although slip has occurred in the immediate vicinity of the crack, in general the widest
contact does not serve as a detachment.
Though numerous statistical functions, ranging from negative exponential to normal,
have been proposed to describe the general distribution of joint spacing (e.g. Priest &
Hudson 1976; Rouleau & Gale 1985; Rives et al. 1992), the specific case of lithology-
controlled joints leads to a typically skewed log normal or gamma distribution for joints
belonging to an individual systematic set (e.g. Huang & Angelier 1989; Narr & Suppe
1991). The regular spacing may be due to the stress reduction shadow created by pre-
existing joints (e.g. Hobbs 1967) and the sequential joint infilling process, whereby new
joints form in a sequential fashion between adjacent pre-existing joints (Gross 1993a). The
non-normal distribution may result from the random distribution of flaws, which serve as
initiation points for joints. While other studies focus on interactive joint growth (e.g.
Pollard et al. 1982; Olson & Pollard 1989), we are concerned with identifying where new
joints will form with respect to pre-existing joints along the profile section of a layer.
Therefore, throughout this paper we consider the actual joint propagation process as
instantaneous relative to the time between sequential infilling stages of a systematic joint
set.
Material science studies provide evidence in favour of a sequential infilling mechanism
for joints confined to lithology-controlled mechanical layers. Tension tests performed on
layered ceramics demonstrate that in addition to an increase in crack spacing with layer
thickness, average crack spacing decreases systematically with increasing applied stress
(Garrett & Bailey 1977; Parvizi & Bailey 1978). The applied stress in turn scales linearly
with extensional strain for the elastic laminates. The decrease in crack spacing with
increasing load occurs by a process of sequential infilling. Geological evidence supporting
sequential infilling includes fractured piedmontite grains (Masuda & Kuriyama 1988) and
curving cross joints (Engelder & Gross 1993).
The Hobbs model for joint spacing

The 1D Hobbs model consists of three discrete beds and predicts variations in the crack-
normal tensile stress as a function of differences in elastic properties between a middle
jointing bed and two non-jointing bounding beds. The beds behave elastically and are
bonded together to prevent slip along the interfaces. Unlike the non-layered models of
Lachenbruch (1961) and Pollard & Segall (1987), stress perturbations in a bedded medium
are influenced by both elastic properties and shear stresses that develop along bed
interfaces. According to our interpretation, the Hobbs model consists of a middle 'jointing
bed' (Bed A) sandwiched between 'non-jointing beds' (Beds B). Hobbs assumes both
bounding beds have the same elastic moduli, and that the thickness of bounding beds is
much greater than the thickness (d) of the jointed bed. After initial extension all of the
beds remain unjointed, however, as extensional strain increases, the tensile stress in the
middle bed eventually exceeds the tensile strength of that bed, and a joint propagates. At
this point Hobbs assumes that tensile stress cannot be transmitted across the joint surface.
Shear stresses are produced within the bounding beds as a direct result of differences in
elastic displacement within the jointed bed and bounding beds.
The key parameter controlling joint spacing is the distribution of longitudinal tensile
stress (~rxx) within Bed A after jointing, which increases from zero at the joint surface to
a value equivalent to the remote stress at an infnite distance away from the joint.
FACTORS CONTROLLINGJOINT SPACING 221
Substituting Hobbs' equation 13 into equation 6 and rearranging the terms produces an
equation that evaluates normalized crack-normal tensile stress after joint propagation (i.e.
local stress divided by remote tensile stress) as a function of distance from the joint;
P ff local (1)
- - - 1 + sinh ~ ~ - cosh--d-- -~A
E A de1 t7 r e m o t e
where P is the applied load, s is the strain at which the first joint forms, d is the jointing
bed thickness (i.e. joint height), and EA and GB are the Young's modulus and shear
modulus of the jointing and non-jointing beds, respectively. We prefer to rewrite equation
(1) in the form:
P
- - = f(x/6), (2)
E A del
where,
= ~ ~ , (3)
and f ( x / 5 ) gives the dependence of the local stress on x. The function f ( x / 6 ) is 0 for x = 0
and increases to 1 for large x. The quantity 6 has units of length and can be thought of as a
decay length, or the distance over which the stress shadow decays back to a fixed fraction
of the remote stress. The distance 5 is proportional to d and to the square roof of (EA/GB).
1.21 . . . . . , , , I I I I I I I I
..I
1.0 -! .........EA/ GB= 1 ..............
I.M .~
ml_ o 0.8
~ 0.6 EA/GB=20
N
i
"~
,,.J o
r162 o 0 . 4
rr" lEA/GB=(1, 2, 215, 5, 10, 20) I
O
=, 0.2
0.0
0 1 2 3 4 5
NORMALIZED DISTANCE (x/d)
Fig. 6. Predicted plots of stress reduction shadows based on the Hobbs model for various EA/Ga
ratios.
222 M . R . GROSS E T A L .
Fig. 7. Finite element mesh and boundary conditions. Models were subjected to uniform longitudinal
extension, c. Inset shows an enlarged portion of mesh around joint in Bed A. Bed B is shale in all
models whereas the material properties of Bed A are varied. Joint shown slightly open for
illustration. Cartesian coordinate system with (0, 0) at the centre of the joint is used for all related
graphs and figures (after Fischer 1994).
A plot of normalized stress v. normalized distance from a joint for a series of EA/GB
ratios is presented in Fig. 6. Because Hobbs' analysis predicts that for a fixed shear
modulus of the bounding beds (i.e. holding GB constant), 6 will increase as the square root
of EA, a high EA will produce a long stress reduction shadow, whereas a low EA creates a
short perturbed zone.
Table 1. Boundary conditions and elastic properties assigned to finite element models
Model # egg EB (GPa) EA (GPa) uB UA GB (GPa) GA (GPa)
1 5• 10-4 16 56 0.14 0.26 7 22

2 5• 10-4 16 17.5 0.14 0.25 7 7
3 5• 10--4 16 35 0.14 0.25 7 14
4 5• 10-4 16 70 0.14 0.25 7 28
Finite element modelling

We conducted a series of 2D finite element models to determine the distribution of stresses
and displacements in a jointed, interbedded rock body, and to investigate the effects of
contrasting elastic moduli on stress perturbations near joints. Finite element numerical
simulation was accomplished using the interactive fracture analysis program F R A N C
(Wawrzynek & Ingraffea 1987). F R A N C simulates the r - ~ singularity in the elastic crack-
tip stress field by surrounding each crack tip with a rosette of eight quadratic, triangular,
isoparametric quarter-point elements (Barsoum 1976); the remainder of the model mesh is
composed of quadratic, isoparametric elements. The boundary conditions and geometry
of the model are shown in Fig. 7. Model material properties (Table 1) are median values
compiled from several published sources (Blair 1955, 1956; Clark 1966; Senseny & Pfeifle
1984; Atkinson & Meridith 1987).
In the first model, the 2D distribution of crack-normal tensile stress (crxx) was analysed
in detail for a model consisting of a dolostone bed (Bed A; E = 56 GPa; G = 22 GPa;
Fig. 8. Contour diagram showing the 2D distribution of crack-normal tensile stress (trxx) in MPa for
a portion of the dolostone and shale model to the right of the joint. Model subjected to a strain of
5 • 10--4 (after Fischer 1994).
u = 0.26) bounded by shale beds (Bed B; E = 16GPa; G = 7GPa; u = 0.14). A uniform

longitudinal extensional strain of 5 x 10-4 was applied, resulting in theoretical pre-
cracking stresses of 28 MPa in the dolostone and 8 MPa in the shale beds. Because of edge
effects in the finite element model, however, Crxxvaried longitudinally within each bed prior
to cracking. This variation was minor ( < 2.0% mean error in each bed), and in the model
the mean axx is 28.5 MPa in dolostone and 7.8 MPa in shale.
L o c a l c r a c k - n o r m a l tensile stress
Local crack-normal tensile stress (trxx) in the dolostone is reduced near the joint whereas in
the shale there is an increase in Crxxnear the joint (Fig. 8). At the joint wall ~rxx is zero and
near each end of the joint the stress is elevated due to the stress concentrating effect of the
sharp crack tip (Lawn & Wilshaw 1975; Brock 1978). Because the joint is elliptical, the
greatest amount of elastic strain (and hence stress) is relieved by crack opening along the
middle of Bed A. Consequently, the stress reduction is greatest and extends farthest along
the centre of Bed A. This combination of elliptical crack shape, high stress near the joint
tip and zero stress along the length of the joint is responsible for the concave-inwards
shape of the stress reduction shadow in the dolostone bed. Because there is no crack
opening in the shale beds, large elastic strains developed in shale adjacent to the joint tips
result in a localized increase in stress. An integral part of linear elastic fracture mechanics
(Irwin 1957; Paris & Sih 1965; Broek 1978), this stress concentration is maximum at the
joint tips and decreases rapidly away from the joint. At horizontal distances > 0.3 m from
the joint the stress in the shale beds is the same as the pre-cracking stress.
-o
1.2 I . . . . . . . . . . . . . . . . . . . . I ,~, 0.5 ~ [ ~ ~ = - - ~ = ~ , 0.9 (a)
~g EA/ Ga
1 ............. m 0
0.8 ............................
i .~ -o.25 EA=17.5 GPa
#: ~, -o.5
g o.~ . . . . . . . . . . . . . . . . . . . . . . o.5
0.25
t~ 0.4 ............. (b)1
== 02 ~- -o25
EA=35GPal
o o ~ -0.5
0 1 2 3 4 5 ~ 0.25
NORMALIZED DISTANCE (x/d) -~ .75
(a) N o
:~ -0.25
~, -o.5
z 0 1 2 3
Normalized Horizontal Distance from Joint (x/d)
(b)
Fig.9 (a) Graph of normalized axx v. normalized horizontal distance from the joint for various
EA/G8 values calculated using FRANC. GB for all models is 7 GPa. Stress values were recorded
along centre of Bed A. ~) Contours of normalized axx in a middle bed of v = 0.25 and variable EA.
Bounding beds are shale in all three models and EA/GBratios are 2.5, 5 and 10 in (a), (b) and (c),
respectively. Contours are in percentages of the pre-cracking stress in each layer (i.e. a value of 0.5
indicates the stress at that position is 50% of the pre-cracking stress value) (after Fischer 1994).
Elastic moduli effects

In order to investigate the effects of varying elastic moduli (i.e. EA/Ga ratios) on joint
spacing we conducted a series of models in which a bed with v = 0.25 and variable EA is
sandwiched between shale beds and subjected to a uniform longitudinal extensional strain
of 5 • 10-5. The ratio of EA/GB controls the lateral extent of the stress reduction shadow
around a joint, as shown graphically in Fig. 9a. To illustrate this effect better, Fig. 9b
shows contours of normalized stress in the middle bed for three geologically reasonable
values of EA/GB. For values of normalized Crxx < 0.75, the stress reduction shadows for
various EA/GBratios are nearly identical. However, for values of normalized stress > 0.75
the size of the stress reduction shadow increases significantly with increasing EA.
Comparison with Hobbs' analysis

In general there is good agreement between Hobbs' analysis and our finite element model
results. Although Hobbs' prediction of the dependence of stress reduction shadow size on
contrasts in elastic moduli is in qualitative agreement with the finite element model results
(i.e. the stress shadow increases in extent with increasing Young's modulus), the
dependence of the stress shadow decay length, d, on EA exhibited by F R A N C is weaker
than the square root dependence predicted by the Hobbs model. To show this we consider
the x position of the finite element contours of normalized axx at the centre of Bed A,
denoted by x (m) (Fig. 9b). These positions are compared to those predicted by the Hobbs
model, denoted by x (p). The x (p) values for each contour are computed as:
x(p) = x(m) , / EA
(EA) (17.5)V 17.5 ' (4)
and are given in Table 2. This definition of x (p) gives Hobbs' prediction of the x position of
each contour for the case in which EA is 35 and 70GPa (i.e. xlP~), xlP~)) based on the
position of the corresponding contour for the finite element modelinwhich EA is 17.5 GPa
(i.e. "~(17.5V"
,.(m) a The x (p) values calculated for the models in which EA is 35 and 70 GPa are in
all cases greater than the corresponding x (m) values, showing that dependence of the stress
reduction shadow size on EA is weaker for the finite element model than the square root
dependence predicted by Hobbs.
Model geometry and method of stress determination may contribute in part to the
differences in stress reduction shadow size predicted by the Hobbs and finite element
models. In the finite element model the joint terminates at the shale boundaries and is
entirely confined to the middle layer, thus reflecting geometry commonly observed in
interbedded rocks with sharp lithologic contacts (Figs 3 & 4). In contrast, the joint in the
Hobbs model extends significant distances into the neighbouring beds due to requirements
imposed by shear stress assumptions. Furthermore, as opposed to Hobbs' derivation, the
numerical model utilizes linear elastic fracture mechanics to describe the stress and
displacement fields in the vicinity of the crack.
Table2. Comparison of finite element stress reduction contours to the (EA/GB) '/2 dependence predicted
by the Hobbs model
: Contour level
0.25 0.50 0.75 0.90
EA(GPa) x(m) x~) x(m) x~) x(m) x@) x(m) x~)
17.5 0.403 0.403 0.655 0.655 1.042 1.042 1.613 1.613

35 0.420 0.570 0.682 0.927 1.075 1.473 1.848 2.281
70 0.470 0.806 0.706 1.310 1.176 2.083 2.318 3.226
In all models GB is 7 GPa and distances are normalized to a bed thickness of 1. x(m) refers to the
contour position in the numerical model and x(p) is the position predicted by Hobbs.
Joint spacing predicted from numerical models

Joint spacing models combine a rock fracture criterion with an analysis of the stress
distribution around a pre-existing joint to predict the location where subsequent joints will
form (e.g. Hobbs 1967; Pollard & Segall 1987; Narr & Suppe 1991; Gross 1993a). Hobbs
employs a tensile strength fracture criterion wherein joints form when the remote tensile
stress (O'remote) in a bed exceeds the tensile strength. We use a fracture criterion based on
the principles of linear elastic fracture mechanics (Lawn & Wilshaw 1975).
226 M.R. GROSS E T AL.
3 0 , l , I , , I , I , ! , I ,
""
2
9, \ + ' '
", i ~ . li,,+ dolostone I
=E 25. , i " ~ . . . .~. . . ...... ~..............;..........,t - - - - " sandstone t .....
- - ~ - .... shale 1/
i?:i,
.............. ~,. ........ %.+.....i ............. i ....................... i .............. i ............. ~-'............ :. .........
m i'.i ~i i ~ i +
~ 1 5 .............. i..............
r - ! "4 . = "~ , i i
o 5- ...........
cf = K m c . ( n c ) "I!~...........
i.............................
i...........................................
0
0 0.1 0.2 0.3 0.4 0.5
FLAW HALF SIZE r (r
Fig. 10. Flaw size dependent fracture criterion. Fracture stress for dolostone, sandstone and shale
illustrated for various flaw sizes. Fracture toughness values for sandstone (Kic = 1.2 MPa mW), shale
(Kic= 0.9 MPa m v2) and dolostone (Kic= 1.7 MPa m v') are the mean of Kic values compiled in the
literature (Senseny & Pfeifle 1984; Atkinson & Meredith 1987).
Fracture criterion
Inglis (1913) and Griffith (1920) showed that brittle fracture is often induced by a tensile
stress concentration at flaws (e.g. microcracks) in a material. Linear elastic fracture
mechanics describes the stress concentration at the tip of a Mode I opening (i.e. tensile)
crack by means of the stress intensity factor (Broek 1978):
gI ~- Yo'remote v/-~, (5)
where ~rr~moteis the remote crack-normal tensile stress, c is half the crack length, and Y is a
geometric constant, commonly taken as 1.0 for an infinite fiat blade crack (Lawn &
Wilshaw 1975). The fracture criterion states that a crack will propagate when stress
intensity at the crack tip exceeds some critical value known as the fracture toughness (Kit)
of the material (Irwin 1957). Substituting Kic for KI in equation (5), setting Y = 1, and
solving for the fracture stress (af), the crack-normal tensile stress required to cause a flaw
to propagate, it is found that:
gxe
o: - v/ft.
Using published values of K~c for a variety of rock types, af can be calculated for a given
flaw size (Fig. 10). However, because the distribution of flaw sizes in rocks is complex, it is
unlikely that the fracture strength of a rock can be defined by a single value of af.
Flaw size distribution in two dimensions

The remote tensile stress required to drive a crack decreases with initial flaw size (Fig. 10).
As a consequence, for a constant applied extensional strain, the minimum allowable
distance between two adjacent joints decreases with increasing flaw size. The dependence
Fig. 11. Stress reduction shadow in the dolostone bed of the dolostone and shale model. Tensile
stress contours are in MPa, and applied remote stress is 28.5 MPa. The closest distance at which a
new joint will form is dependent on the size of the flaw from which it initiates. Flaws ranging in size
from 0.3 to 0.5 cm will activate in the shaded region, with 0.5 cm flaws initiating closer to the joint.
Joint spacing therefore decreases with increasing initial flaw size.
of joint spacing on flaw size is shown schematically in Fig. 11, which displays stress
contours around a joint in the model dolostone bed (extensional strain of 5 x 10-4; applied
remote stress of 28.5 MPa). As flaw size increases, the corresponding fracture stress at
which point the flaw grows into a joint is attained at distances closer to an existing joint,
resulting in a decrease in joint spacing. Joint propagation occurs in the shaded region of
Fig. 11 for flaws ranging in size from 0.3 to 0.5cm. In the case of joints initiating in the
middle of the bed (i.e. y = 0) the minimum allowable joint spacing ranges from (0.82)d for
0.5 cm flaws to (1.25)d for 0.3 cm flaws, where d is the bed thickness.
In addition to flaw size distribution along the x-axis, another factor that determines
joint spacing is the position relative to the bed boundary (i.e. along the y-axis) from where
the joint initiates. Contour shapes in Fig. 11 predict that for a given flaw size, joints
initiating near bed boundaries will be more closely spaced than joints propagating from
the central region,of beds. The curvature of stress contours is less pronounced for large
distances from the joint, so that for the 90% contour level the dependence of joint spacing
on y-axis-initiation position is somewhat reduced. Furthermore, a WeibuU approach to the
strength of materials (e.g. Weibull 1951; D avidge 1979) implies that joint initiation is more
likely to occur near the centre of the bed rather than along its edges. This is due to the
greater activation area in the central portion of the bed as indicated by the width of the
shaded zone in Fig. 11.
Joint spacing determination

Several approaches have been proposed for analysing joint spacing in 1D models,
including the maximum tensile stress criterion that predicts precise midpoint fracturing
regardless of initial joint spacing (Hobbs 1967), the introduction of random flaws into
stress reduction models (Narr & Suppe 1991; Rives et al. 1992), and the minimum
Fig. 12. (a) Stress shadows representing 90% of the remote tensile stress for beds with Young's
modulus of 17.5GPa (Bed #1) and 70GPa (Bed #2). ~) Expected joint spacing distributions for
Beds #1 and #2 after an equal number of infiUing events. Note the larger stress reduction shadow
leads to a larger median joint spacing.
allowable spacing criterion (Gross 1993a). Regardless of the approach, two boundary
conditions always apply to models of joint spacing in lighology-controlled mechanical
layers. First, a new joint will form between two pre-existing joints when the fracture stress
is attained. Second, the stress reduction shadow inhibits the formation of new joints in the
vicinity of pre-existing joints.
Our numerical results from 2D models indicate that the extent of the stress reduction
shadow increases with increasing Young's modulus. The effect of elastic properties on
joint spacing is illustrated in an example of two jointing layers, #1 and #2, with Young's
moduli of 17.5 and 70GPa, respectively (Fig. 12a). Consider the case whereby each bed
has undergone an initial fracturing episode. A subsequent infilling event will result in the
joint spacing distributions depicted in Fig. 12b. Random flaw distributions along with the
2D shape of the activation zone will lead to a wide range of joint spacing values and a
characteristic skewed distribution. For example, joints will initiate preferentially from the
largest flaws, and joints initiating near bed boundaries will be more closely spaced than
joints propagating from the central portion of the bed. Nevertheless, one would initially
predict that for a population of mechanically confined fractures, the bed with the higher
Young's modulus will exhibit a larger median joint spacing because of the larger stress
reduction shadow.
Table 3. Fracture spacing measurements at Lompoc Landing along with physical properties of siliceous
mechanical units in the Monterey Formation
Chert Diatomite
Number of beds 6 7
Number of joint spacings 247 315
Matrix porosity 0.5% 37%
Young's modulus 70 GPa 5 GPa
Shear modulus 33 GPa 2 GPa
Fracture spacing index (FSI) 2.41 1.09
Fracture spacing ratio (FSR) 2.29 1.11
Porosity and elastic moduli measurements are preliminary and were determined from similar units
exposed along the Santa Maria coastline. Elastic properties were measured dynamically using the
resonance technique. From Gross 1993b.
Why stiffer beds have smaller joint spacings; a field example from the
Monterey Formation, California
The concept that differences in elastic properties among jointing lithologies will result in
different joint spacings for a given bed thickness is intuitive. Our analysis thus far suggests
the stress shadow increases in lateral extent with increasing Young's modulus, implying
that joint spacing should be greater in beds with higher Young's moduli (i.e. stiffer beds)
(Fig. 9b). However, joint spacing is often closer in stiffer beds. For example, the siliceous
Monterey Formation exposed at Lompoc Landing along the Santa Maria coastline of
California (Fig. 5) consists of a series of interbedded cherts and dolomitic diatomites, both
jointing lithologies, along with non-jointing clay-rich mudstones. Typical physical
properties of the two siliceous rock types, exposed along the Santa Maria coastline, are
markedly different from each other, with a Young's modulus of 70 GPa for the quartz
chert and 5 GPa for the impure diatomite (Table 3). Several beds at Lompoc Landing are
welded together along their boundaries, and consequently were subjected to equal
amounts of extensional strain. The vertical systematic joint set investigated trends N N E -
SSW and post-dates silica diagenesis, and therefore developed after differences in physical
properties were established among the mechanical units.
Joint spacing data for the diatomite and chert beds at Lompoc Landing are presented as
plots of median joint spacing v. mechanical layer thickness (Fig. 13). The slope of the best-
fit line, calculated with median joint spacing as the dependent variable, is defined as the
fracture spacing index (FSI) of Narr & Suppe (1991). The fracture spacing ratio (FSR)
equals the mechanical layer thickness divided by the median joint spacing, and represents
the mean for all beds of a given lithology. Chert beds at Lompoc Landing display
significantly higher FSI and FSR, and thus more closely spaced joints relative to bed
thickness, than adjacent diatomites. This spacing relationship exists despite the much
larger chert Young's modulus. How does such a joint spacing distribution develop in light
of the numerical models and theoretical considerations discussed in this paper? One
explanation is that despite the larger stress reduction shadow, a stiffer bed will contain
more joints relative to other beds because jointing in the stiffer bed occurs at lower strain
levels.
Consider a rock mass composed of a sequence of dolostone and sandstone beds
interbedded with non-jointing mudstone beds. The dolostone has a Young's modulus of
230 M.R. GROSS ET AL.
,, ,, I . . . . I . . . . I . . . . I ,, ,, I, ,, ,I , ,, [I , I L I , I I ~ l , l , l , I
70
A E 39
E Dolomitic ,j18 ~ 8_
o
v 60- f Chert
w
w
Diatomite ,"" : eI
t"
ir 5 0 - I
,~r ," =6~ " 67 m,,

a
/- 40- 9 nS - t~
I-- s
, 1117
."
=,, 3 0 - 36
9 ,,
9
: ~
_
tl
>,4-
i
s
146
m
,_1 9 .j
.~ 4 2
20- 82 ,'" 9 53 L o
r 111.64 FSI = 1.09 : .~-r 2 - e

s
6
FSI = 2.41
10- 9 FSR = 1.11 " .C
'~ 138 FSR = 2.29
o m'm 3 6 o -
o
=E 26 "
0 .... I''''I .... "~ ''''I .... I''' 'I ' ' ~''r'' 0 " I ' I ' I ' I ' I ' i ' I ' 1
0 10 20 30 40 50 60 70 0 2 4 6 8
Median Joint Spacing (cm) Median Joint Spacing (cm)
Fig. 13. Median joint spacing v. mechanical layer thickness for diatomite and chert beds of the
Monterey Formation exposed at Lompoc Landing. Numbers next to data points refer to total
number of joint spacing measurements taken in that particular bed.
5 6 G P a and a Kit of 1 . 7 M P a m ~, whereas the sandstone has a Young's modulus of

12GPa and a Kic of 1 . 2 M P a m ~. Hooke's Law states that for a given value of strain,
stress will be higher in the bed with the higher Young's modulus (Fig. 14). Because our
fracture criterion is a threshold stress value (i.e. the fracture stress, crf), jointing will occur
in a given bed at the point where the stress-strain curve intersects cry for that bed.
Assuming a range in flaw size from 0.3 to 0.5 cm (i.e. crack half-lengths of 0.15-0.25 cm)
for both jointing lithologies, the fracture stress is 19.2-24.8 M P a for the dolostone and
13.5-17.5MPa for the sandstone (Fig. 14).
Fracture stress ranges derived from the range in flaw size correspond in t u m to a range
of strain values for the first jointing event in each lithology. The intersections of stress-
strain curves with horizontal fracture stress lines in Fig. 14 indicate that joints will prop-
agate in the dolostone bed for strain values within the range of 3.43 x 10-4 to 4.42 • 10-4.
Likewise, joints will initiate in the sandstone unit between strains of 1.12 x 10-3 and
1.46 x 10-3. In the case whereby all beds in the rock mass are welded together, and thus
remote longitudinal strain remains uniform throughout, the first series of joints will
propagate in the dolostone bed once a strain of c. 4 x 10-4 is reached. At this juncture the
sandstone bed is unjointed because stress within the sandstone is less than the minimum
fracture stress of 13.5MPa. As longitudinal extension continues, the sandstone bed
remains unjointed until strain values of 1.12 x 10-3-1.46 • -3 are exceeded, at which
point the first set of joints propagates in the sandstone.
As a result of the relationship between Hooke's Law and the fracture stress criterion,
jointing occurs in the dolostone at lower strains, and hence earlier, than jointing in the
sandstone. In fact, the situation may arise where a stiff bed undergoes two or more jointing
events prior to initial jointing in a low modulus bed. Therefore, in spite of the larger stress
reduction shadows predicted by both Hobbs and finite element modelling, beds with
relatively high Young's moduli can actually contain relatively closely-spaced joints by
virtue of the small strain increments required for infilling jointing events.
Fig. 14. Fracture stress ranges superimposed on stress-strain plots for typical dolostone and
sandstone beds. Ranges in fracture stress correspond to a range in flaw size from 0.3 to 0.5 cm. Note
the dolostone bed will fracture at lower strain levels than the sandstone. See text for details.
In comparing two jointing beds with different elastic moduli (for Beds 1 and 2, E1 > E2),
the factor that determines which bed will fracture at lower strains is the ratio of Young's
modulus to the ratio of fracture stress. Assuming the same range of flaw sizes for both
beds, the stiffer bed will fracture at lower strains when:
E1 > K~__.)_) (7)
The inequality in equation (7) holds true for a majority of sedimentary rocks, which may
explain the higher joint densities often observed in stiffer beds despite the larger stress
reduction shadows.
Conclusions
Joint spacing models focus on the distribution of tensile stress about a joint in an effort to
predict the location of subsequent joints. Stress distributions derived from finite element
simulations compare favourably with Hobbs' theoretical predictions. Both the Hobbs
model and finite element modelling predict longer stress reduction shadows, and hence
wider joint spacings, for beds with higher Young's moduli. Finite element simulations,
232 M.R. GROSS ET AL.
though, show the dependence of the stress reduction shadow on Young's modulus in the
jointing bed is weaker than the square root dependence predicted by Hobbs.
Though bed thickness is a primary factor, elastic moduli, extensional strain and flaw size
are all important parameters determining joint spacing in interbedded sedimentary rocks.
A high Young's modulus contributes to an increase in stress reduction shadow length, yet
tends to lower the strain level required for joint propagation due to its relationship with
the fracture stress criterion. Small flaws situated in the middle of the jointing bed produce
widely spaced joints, whereas those same flaws located near bed interfaces result in more
closely spaced joints. Despite the complex interaction among these factors, creating a wide
range of joint spacings for an individual mechanical bed, the linear correlation between
median joint spacing and bed thickness for a given lithology, and hence given mechanical
properties, is remarkably strong.
Funding for this project was provided by a grant from Texaco (TE, MRG) a Shell Doctoral
Fellowship (MPF) and GRI Contract 5088-260-1746 (TE). We thank T. Ingraffea for providing the
finite element program, T. Bittencourt and D. Swenson for technical assistance and Wendy Bartlett
for helpful field information. We appreciate the thoughtful reviews provided by Wayne Narr, Atilla
Aydin, Byron Kulander, Ernie Rutter and Mohammed Ameen.
References
ATKINSON,B. K. & MEREDITH,P. G. 1987. Experimental fracture mechanics data for rocks and
minerals, ln: ATKINSON, B. K. (ed.) Fracture Mechanics of Rock. Academic Press, London, 477-
525.
BARSOUM,R. S. 1976. On the use of isoparametric finite elements in linear fracture mechanics.
International Journal of Numerical Methods in Engineering, 10, 25-37.
BLAIR, B. E. 1955. Physical properties of mine rock, Part 3. US Bureau of Mines Report of
Investigations 5130.
- - 1956. Physicalproperties of mine rock, Part 4. US Bureau of Mines Report of Investigations
5244.
BROEK, D. 1978. Elementary Engineering Fracture Mechanics. Sijthoff & Noordhoff, Dordrecht, the
Netherlands.
CLARK, S. P. 1966. Handbook of Physical Constants. Geological Society of America Memoir, 97.
Cox, H. L. 1952. The elasticity and strength of paper and other fibrous materials. British Journal of
Applied Physics, 3, 72-79.
DAVIDGE,R. W. 1979. Mechanical Behavior of Ceramics. Cambridge University Press, Cambridge,
UK.
ENGELDER,T. & GROSS,M. R. 1993. Curving cross joints and the lithospheric stress field in eastern
North America. Geology, 21, 817-820.
FISCHER, M. P. 1994. Application of linear elastic fracture mechanics to some problems of fracture
propagation in rock and ice. PhD Thesis, Pennsylvania State University, USA.
GARRETr, K. W. & BAILEY, J. E. 1977. Multiple transverse fracture in 90~ cross-ply laminates of a
glass fibre-reinforced polyester. Journal of Materials Science, 12, 157-168.
GRIFFITH, A. A. 1920. The phenomena of rupture and flow in solids. Philosophical Transactions of
the Royal Society of London, A221, 63-197.
GROSS, M. R. 1993a. The origin and spacing of cross joints. Journal of Structural Geology, 15, 737-
751.
1993b. The effects of mechanical stratigraphy on failure mode and fracture spacing in the
Monterey Formation of coastal California. PhD Thesis, Pennsylvania State University, USA.
HOBBS, D. W. 1967. The formation of tension joints in sedimentary rocks: an explanation. Geological
Magazine, 104, 550-556.
HUAN6, Q. & ANGELIER, J. 1989. Fracture spacing and its relation to bed thickness. Geological
Magazine, 126, 355-362.
INGLIS, C. E. 1913. Stresses in a plate due to the presence of cracks and sharp corners. Transactions
of the Institute of Naval Architecture, 55, 219-230.
IRWIN, G. R. 1957. Analysis of stresses and strains near the end of a crack traversing a plate. Journal
of Applied Mechanics, 24, 361-364.
JAEGER, J. C. & COOK, N. G. W. 1976. Fundamentals of Rock Mechanics. Chapman & Hall, London.
LACHENBRtJCH, A. H. 1961. Depth and spacing of tension cracks. Journal of Geophysical Research,
66, 4273-4292.
LADEIRA, F. L. & PRICE, N. J. 1981. Relationship between fracture spacing and bed thickness.
Journal of Structural Geology, 3, 179-183.
LAWN, B. R. & WILSHAW, T. R. 1975. Fracture of Brittle Solids. Cambridge University Press,
Cambridge, UK.
MASUDA, T. & KURIYAMA, M. 1988. Successive 'mid-point' fracturing during microboudinage: an
estimae of the stress-strain relation during a natural deformation. Tectonophysics, 147, 171-177.
McQUILLAN, H. 1973. Small-scale fracture density in Asmari Formation of southwest Iran and its
relation to bed thickness and structural setting. American Association of Petroleum Geologists
Bulletin, 57, 2367-2385.
NARR, W. & SUPPE, J. 1991. Joint spacing in sedimentary rocks. Journal of Structural Geology, 13,
1037-1048.
OLSON, J. & POLLARD, D. D. 1989. Inferring paleostresses from natural fracture patterns: A new
method. Geology, 17, 345--348.
PARIS, P. C. & SIH, G. C. 1965. Stress Analysis of Cracks. American Society of Testing and
Materials, Special Publication, 381, 30-76.
PARVIZI, A. & BAILEY, J. E. 1978. On multiple transverse cracking in glass fibre epoxy cross-ply
laminates. Journal of Materials Science, 13, 2131-2136.
POLLARD, D. D. & SEGALL,P. 1987. Theoretical displacements and stresses near fracture in rock:
with applications to faults, joints, veins, dikes, and solution surfaces. In: ATKINSON, B. K. (ed.)
Fracture Mechanics of Rock. Academic Press, London, 277-349.
- - , SEGALL, P. & DELANEu P. T. 1982. Formation and interpretation of dilatant echelon
cracks. Geological Society of America Bulletin, 93, 1291-1303.
PRICE, N. J. 1966. Fault and Joint Development in Brittle and Semi-brittle Rocks. Pergamon Press,
Oxford.
PRIEST, D. S. & HUDSON, J. A. 1976. Discontinuity spacings in rock. International Journal of Rock
Mechanics, Mining Sciences, and Geomechanics Abstracts, 13, 135-148.
RIVES, T., RAZACK, M., PETIT, J.-P. & RAWNSLEY, K. D. 1992. Joint spacing: analogue and
numerical simulations. Journal of Structural Geology, 14, 925-937.
ROULEAU, A. & GALE, J. E. 1985. Statistical characterization of the fracture system in the Stripa
Granite, Sweden. International Journal of Rock Mechanics, Mining Sciences, and Geomechanics
Abstracts, 22, 353-367.
SENSENY,P. E. & PFEIFLE,Z. W. 1984. Fracture toughness of sandstones and shales. Proceedings of
the 25th US Symposium on Rock Mechanics. Society of Mining Engineers of the American
Institute of Mining, Metallurgical and Petroleum Engineers, Inc., New York, 390-397.
SYLVESTER,A. G. & DARROW,A. C. 1979. Structure and neotectonics of the western Santa Ynez
fault system in southern California. Tectonophysics, 52, 389-405.
WAWRZYNEK, P. A. & INGRAFFEA, A. R. 1987. Interactive finite element analysis of fracture
processes: an integrated approach. Theoretical and Applied Fracture Mechanics, 8, 137-150.
WEIBULL, W. 1951. A statistical distribution function of wide applicability. Journal of Applied
Mechanics, 18, 293-297.
Index
Page numbers in italics refer to Figures and Tables
Alpine Orogeny 171, 210 chert

andesite discontinuity measurements 15, 17
Japan 15, 17 jointing study 229, 230
West Cumbria 144 Chita mudstone discontinuity study 15, 17
annular structure 64 compaction/burial and associated jointing 168-
see also rib mark 70
arrest line 42, 63, 79, 176 compressional tectonics 171
see also rib mark compressive strength in Chalk 209
Australia granodiorite roughness study 31 Continental Superdeep Borehole see KTB
auto correlation function (ACF) defined 13 cooling discontinuity 12
core disking surface (CDS) 46-7, 48, 53
transitional forms 47
B-plane 176 crack branching, role in failure of 91-3
basalt discontinuity measurements 15, 17 cross fracture 63
Bavaria 40, 41 Cumbria see West Cumbria
bedding planes 12
discontinuity profile 16, 18, 19, 20
beef 196 density of Chalk 209
bending, role in failure of 89-91 desiccation discontinuity 12
bending fracture, experimental production of diatomite jointing study 229, 230
71-2 dilatent echelon crack 63
Bohemian Massif 40, 41 discontinuities
Borrowdale Volcanic Group 100, 141-4 classification 12
see also West Cumbria basement rock Japan case study
Botany Bay 154, 157 effect of profile length 20, 21
burial/compaction and associated jointing 168- effect of sampling interval 18-19
70 equipment 16-17
Burnham chalk study 201 location of samples 15
primary v. secondary 22-4
role of eigen direction 21, 22
C-fracture 176 linear profile measurement 15-17
C-joint 63 linear profile parameters 12-15
calcite summary of observations 24--5
mineralization 115, 117, 122 disparity, use in roughness measurement of 28
veins 196 diverging plumes 41
Calder Sandstone 99 drape folds 163, 168, 172
cataclasite 127, 161,165 Dumpton Gap 158
centre-line average height (CLAH) dyke, sandstone 192-3
defined 12
effect of profile length on 20
effect of sampling interval on 18, 19 East Cliff 158
relation to ratio of profile length 24 East Harling chalk study 202
relation to weighted asperity inclination 23 East Irish Sea Basin 98, 99
Chalk joint features 168-70 Echizen tuff discontinuity study 15, 17
method of analysis 199 eigen direction 22
results 200-6 effect on discontinuity study 21, 22
results discussed 209-11 eigen discontinuity 12
significance of spatial changes 208-9 elastic moduli, effect on joint spacing of 224
stress implications 206-8 en echelon fracture 101
setting 197-9 Ennerdale granite and granodiorite 99, 134, 146
Chausuyama hornfels discontinuity study 15, 17 Epple Bay 158, 160
236 INDEX
Eskdale granite 99, 134, 145 fracture plume 176

experimental production of fractographic fracture processes 52-4
features fracture propagation 52
methods 64-6 fracture spacing index (FSI) 229
results fracture spacing ratio (FSR) 229
bending fracture 71-2 fracture surface studies (markings, morphology
fracture forking 68 and patterns)
fracture hooking 68 classification 27, 41-2, 176
origin of fracture 66-8 interpretation of stress patterns 43
point load fracture 70 photogrammetric measurement
thermal fracture 71 principles 28
torsion fracture 68-70 quantitation 31-2
results discussed regional studies
forking 77 KTB analysis
inclusion hackle 77-9 core disking surfaces (CDS) 46-7, 48,
origins 73-4 53, 54
rib marks 79-80, 81 drilling-induced centre-line fractures 46
speed of propagation effect 72-3 transitional forms 47, 52
twist hackle 77-9 usual fractographic surface structures
Wallner lines 75-7 (UFS) 46-7, 49, 50, 51
extensional tectonics 170-1 Mt Alexandra Quarry 32-6
St Mary's Well Bay 176--7
methods of analysis 180--1
F-joint 63, 176 results 181-3
failure mode 83 results discussed 183-5
experimental testing 84-5 setting 177-9
failure area analysis 89 West Cumbria 108, 115, 133
bending 89-91 symmetry 42
crack branching 91-3 fracture trace
microscope analysis 86-8 defined 100
fast Fourier transform (FFT) images 35, 36 West Cumbria study 101, 108, 109, 110, 111,
faulting, role in discontinuity of 12 112, 133
faults 100 Franconian Line 39, 40
West Cumbria study 104-7, 125-6 freezing discontinuity 12
feather fracture border plane 63 fringe 176
flow plane 12 fringe face 63
fluidization 192-3 fringe step 63
focal point 62
see also origin
folding, role in discontinuity of 12 Germany 40, 41
forced folds 163, 168, 172 glaciation, overburden effects of 211
forking 48, 54 gouge 127, 161,165
see also fracture forking grain plane, role in discontinuities of 12
fractal defined 27 granite
fractal dimension defined 13 Japan 15, 17
fractographic features classified 100 West Cumbria 100, 133, 145, 146
fractography granodiorite
defined 1, 99-100, 152 Mt Alexandra 31
evolution of studies 1-2 West Cumbria 145
first use of term 59 Griffith criteria 187
relation to fracture analysis 2-3 gull wing 72, 75, 76
research gypsum veins 196
future 9
past 3-8
fracture defined 99-100 hackle face 158
fracture face 176 hackle (hackly) fractures 63
fracture forking 63, 66, 68, 74, 77 hackle plume 79, 101
fracture hooking 68 KTB study 41
INDEX 237
Lavernock Point 181, 183, 184 fracture types

Thanet monocline 158, 159, 163 core disking surfaces (CDS) 46-7, 48, 53,
West Cumbria study 118, 121 54
hackle step 158 drilling-induced centre-line fractures 46
haematite mineralization 127 transitional forms 47, 52
Harcourt roughness measurements 31 usual fractographic surface structures
hardway plane, role in discontinuity of 12 (UFS) 46-7, 49, 50, 51
hesitation line 64 stress field analysis
see also rib mark methods 54-6
hooking 3, 101, 125 results 56-7
hoop test for failure 84-5 location 39, 40, 41
hornblende, role in discontinuity of 12
hornfels discontinuity measurements 15, 17
hydraulic fracturing 191 Lake District Block 98, 99
Lavernock Point 176, 177
fracture surface marking study
igneous rocks, discontinuities in 12 methods of analysis 180-1
inclusion hackle 63, 77-9 results 181-3
inclusion hackle tail 72 results discussed 183-5
iron oxide mineralization 115, 117, 163 setting 177-9
Italian marble, failure testing of 85, 87, 88 limestone
Japan discontinuity measurement 15, 17
St Mary's Well Bay Formation 176-7
Japan discontinuity morphology study methods of analysis 180--1
effect of profile length 20, 21 results 181-3
effect of sampling interval 18, 19 results discussed 183-5
equipment 16-17 setting 177-9
location of samples 15 lithology, effect on joint spacing of 229
primary v. secondary features 22-4 London-Brabant Massif 152--4
role of eigen direction 21, 22
joint
defined 62, 99, 197 macrofracture and macrofault
regional studies defined 100
Eastern England 168-70, 197-9 West Cumbria study 125-8, 137
method of analysis 199 marble, failure testing of 85, 87, 88
results 200-6 mathematical modelling of joint spacing
results discussed 209-11 1-D Hobbs model 220-2
significance of spatial changes 208-9 finite element model 222-4
stress implications 206-8 models compared 224-5
West Cumbria 104-7 testing of models 225-8
joint face 101, 176 mean standard variation of height (MSVH)
joint spacing defined 13
bed thickness effects 218-20 mean standard variation of inclination (MSVI)
history of research 215-17 defined 13
mathematical modelling megafracture defined 100
1-D Hobbs model 220-2 Mercia Mudstone 193, 194
finite element model 222-4 mesofracture
models compared 224-5 defined 100
testing of models 225-8 West Cumbria study
Monterey Formation study 229-31 mesofracture study in basement rock 131-3,
stress field effects 217-18 141-6
dynamic significance 133-7
surface features 133
Kingsgate Bay 154, 156 fracture traces 133
Kitamatado shale discontinuity study 15, 17 surface patterns 133
KTB study mesofracture study in cover rock
drill core fractures chronology 115-7
fracture processes 52 layering control 117
238 INDEX
non-systematic fractures 108, 112 Okumino discontinuity studies 15, 17

systematic fractures 100-1 origin 62, 70, 72, 101
dynamic significance 117, 125 action in experiments of 66
fracture traces 101, 108, 109, 110, 111, role in chalk failure 158
112 Otake andesite and basalt discontinuity study
surface patterns 108, 112 15, 17
mesofracture/macrofault relations in base- overburden, stress effects of 196, 210-11
ment rock 137
mesofracture/macrofault relations in cover
rocks 126-8 Palm Bay 154, 156, 159
regional stress patterns 128 Pegwell Bay 166, 167
metamorphic rocks, discontinuities in 12 Pegwell Bay-Minnis Bay Transpression Zone
micas, role in discontinuity of 12 158, 168, 171
microfracture defined 100 Penrith Sandstone failure testing 85, 86, 87
Mill Hill chalk study 201 photogrammetry
mineralization 3 Mt Alexandra Quarry tests 32-6
West Cumbria 115-7, 122 use in roughness measurement
mineralogy, effect on discontinuities of 12 methods 31-2
Minnis Bay 158 principles 28
mirror region pit 62
defined 62 see also origin
experimental production of 74, 66, 71, 72 plume axis 101
mist region plumose structure 63, 176
characterized 62 plumose-coarse twist hackle boundary 176
experimental production of 66, 72, 74 point load fracture 70
mode I (tensile) fractures 188, 190 point sources 181
Thanet monocline 170, 172 porosity in Chalk 209
West Cumbria 117-9 power spectral density (PSD) plots 36
modes of failure 83 pressure solution features 64
experimental testing 84-5 profile length, effect on discontinuity study of 20,
failure area analysis 89 21
bending 89-91 profilometer 14, 17, 31
crack branching 91-3
microscope analysis 86--8
mode I (tensile) failure 83, 176 radient 63
effect of tensile strength 193-4 ratio of profile length (RPL)
relation to stress 187 defined 13
role of fluid pressure 190-1 effect of profile length on 20
mode II (sliding) failure 83 effect of sampling interval on 18, 19
mode III (tearing) failure 83, 176 relation to centre-line average height 24
Mohr circles 187, 188 relation to weighted asperity inclination 25
chalk stress field 207 rhyolite discontinuity measurements 15, 17
Monterey Formation joint spacing study 216, rib mark 63, 71, 78, 79-80, 81,101
217, 219-20, 229--31 Lavernock Point 181
Mt Alexandra Quarry roughness measurements Thanet monocline 158, 159, 163
31 West Cumbria study 118, 120, 121
mudstone discontinuity measurements 15, 17 Richborough syncline 173
ripple mark 62-3
see also Wallner line
river line 63
Nakatsukawa granite discontinuity study 15, 17 root mean-square of height (RMSH) defined 13
Navier-Coulomb criteria 187 root mean-square of inclination (RMSI) defined
non-systematic fractures 13
defined 100 roughness measurement 27-8
West Cumbria study occurrences 101, 117, method 28-30
123 quantitation 31-2
North Cliff 158, 165 regional study 32-6
INDEX 239
St Bees Sandstone 99 surface patterns 108, 115

fracture pole diagrams 102, 103
macro faults 127, 130
St Mary's Well Bay Formation 176-7 tearing (mode III) failure 83, 176
fracture surface marking study tectonic discontinuity 12
methods of analysis 180-1 tectonic features of Thanet monocline 170-1
results 181-3 tectonic stress field study
results discussed 183-5 drill core fracture evidence 46
setting 177-9 fracture types
sampling intervals, effect on discontinuity study core disking surfaces (CDS) 46-7, 48, 53,
of 18, 19 54
sandstone drilling-induced centre-line fractures 46
discontinuity measurements 15, 17 transitional forms 47, 52
failure testing 85, 86, 87 usual fractographic structures (UFS) 46-
see also West Cumbria study cover rock 7, 49, 50, 51
satin spar 196 location 39, 40, 41
schistosity planes 12 tendential fractographic features 62, 67, 69
sedimentary rocks, discontinuities in 12 tendential penetration 176
shale, discontinuity measurements in 15, 17 tensile (mode I) failure 83, 176
shear failure and relation to stress 187 effect of tensile strength 193--4
shear planes failure area analysis 89
discontinuity profile 16, 18, 19, 20 bending 89-91
role in discontinuity of 12 crack branching 91-3
shearing 3 hoop test 84-5
sheeting planes microscope analysis 86-8
discontinuity profile 21, 22 relation to stress 187
role in discontinuities of 12 role of fluid pressure 190-1
Sherwood Sandstone Group 100 tensile (mode I) fracture 188, 190
see also West Cumbria study cover rock Thanet monocline 170, 172
shoulder 176 West Cumbria 117-9
slickenlines 3, 52, 54 tensile strength 209
KTB study 52, 54 tensile stress analysis
Thanet monocline 155, 161, 164, 167 application to joints
West Cumbria study 113, 127 finite element model 222-4
sliding (mode II) failure 83 Hobbs model 220--2
Solway Basin 98, 99 tension planes
South Cliff 165 discontinuity profile 16, 18, 19, 20
spatial frequency (SF) plots 36 role in discontinuity 12
stereographic projection 155 texture, effect on fracture of 72, 73
stress 187 Thanet monocline
KTB stress field analysis 39, 40, 41 fracture characterization
drill core fracture evidence 46 hinge area 161-3
core disking surfaces (CDS) 46--7, 48, 53, NE area 154-8
54 NW area 158
drilling-induced centre-line fractures 46 SE area 158-61
transitional forms 47, 52 fracture origin 168-71
usual fractographic structures (UFS) 46- origin 163-8
7, 49, 50, 51 seismic section 169
West Cumbria study 112, 124, 125, 128-9 setting 149, 150
striations 63 stratigraphy 151
structure function defined 13 structure 152
stylolites defined 100 synthesis 172-3
stylolitization 3 thermal fractures, experimental production of
surface features see fracture surface studies 71, 74
systematic fractures 99-100 torsion fractures 68-70
West Cumbria study experimental production of 68-70
dynamic significance 117, 125 tourmaline, role in discontinuity of 12
fracture traces 101, 108, 109, 110, 111, 112 trace 176
240 INDEX
transient fractographic features 59, 70, 72 weighted asperity inclination

transpression in Thanet monocline 171 defined 13
Tsuge granite discontinuity study 15, 17 effect of profile length on 20
tuff discontinuity measurements 15, 17 effect of sampling interval on 18, 19
twist angle 125 relation to centre-line average height 23
twist hackle 63, 66, 77-9, 158 relation to ratio of profile length 25
twist-hackle face 78 West Cumbria
defined 101, 176 basement and cover rocks compare d 137-~9
twist-hackle fringe 68, 69, 70, 79 macrofault study in cover rock 125-6
defined 101, 176 mesofracture study in basement rock 131-3,
KTB study 41-2 141-6
Thanet monocline 158 dynamic significance 133-7
West Cumbria study 118, 120, 125 surface features 133
twist-hackle plane 182-3 fracture traces 133
twist-hackle step 54, 72, 78 surface patterns 133
defined 101, 176 mesofracture study in cover rock
chronology 115-7
layering control 108-12
uplift/unloading, associated jointing 171 non-systematic fractures 101, 117
usual fractographic surface (UFS) structures systematic fractures 100-7
46--7 dynamic significance 117, 125
transitional forms 47 fracture traces 101, 108, 109, 110, 111,
112
surface patterns 108, 115
veins mesofracture/macrofault relations in base-
defined 100 ment rock 137
West Cumbria study 104-7 mesofracture/macrofault relations in cover
velocity hackle 62, 70, 72 rocks 126--8
experimental production of 66, 68, 71, 74 regional stress patterns 128
setting 97, 98
West Runton chalk study 203
Wallner line 62, 62-3, 70, 72 Westgate Bay 158, 160
experimental production of 66, 71, 71-2, 75-7
Walpole Bay 154, 156, 157
Watchet 193, 194 Young's modulus 229
Fractography: fracture topography as a tool
in fracture mechanics and stress analysis
editedby
M.S. Ameen
(Independent Consultant, London, UK)
This volume introduces the principles and procedures involved in the

description and analysis of fracture surface topographies. It contaim a
wide range of case studies from core and field samples, as well as from
experimentally produced fractures, which illustrate the application of
fractography to fracture studies.
Fractography is a powerful tool that has been largely neglected by those
who deal with fracture characterization. A greater understanding of the
subject is essential to the development of oil and gas fields, particularly
deep and marginal fields. In eddWon, incorporating fractographic data
into models of rod( masses, such as the permeability of hydrocarbon
reservoirs and the stability of tunnels or underground nuclear waste
repositories, will undoubtedly enhance their feasibility.
The book also pmsents new mathods of quantitative meesummem of
fracture topography, and outlines areas needing further research and
development.
The volume will be of interest to a broad range of researche~ and
professionals in frKture mechanics,as well as ,bdctural, engineering and
petroleum geology, material sciencesand mining engineedng.
• 248 pages
• over 200 Illustrations
• 12 papelrs
• index
ISBN 1-897799-32-2
Cover///usuJf~: ~ m m U t a d l y ~ surge
onamt I f r ~ m , WMt Cumbr~ UK him p.llm.

Fractography Fracture Topography As A Tool in Fracture Mechanics and Stress Analysis PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fractography Fracture Topography As A Tool in Fracture Mechanics and Stress Analysis PDF

Uploaded by

Copyright:

Available Formats

Fractography: fracture topography

as a tool in fracture mechanics and stress analysis

Fractography: fracture topography

Published by The Geological Society from: Distributors

Typeset by Bath Typesetting Ltd

Printed by The Alden Press,

Part 1: Fractographic studies

Methods of observation, data collection and processing]quantitative fractography

Application of fractography to fracture analysis from core

Fractography applied to the analysis of experimentally produced fractures in rocks

Fractography applied to fracture analysis in field studies aimed at understanding

Part 2: Non-fractographic studies

Fractography: fracture topography as a tool in fracture

The evolution of fraetography in geology

Fracture population studies generally separate a particular population of fractures in a rock

Do all fractures develop fractographic features?

Part 1: Fractographic studies

Methods of observation, data collection & processing/quantitative fractography This area of

Fractography applied to fracture analysis in field studies aimed at understanding regional

Part 2: Non-fractographic studies

joints in Norfolk to sub-vertical extension joints in Humberside. Such a change is mainly

Suggested target areas for future research in the field of fraetography in

1. A comprehensive review of the relatively vast nomenclature used in fractography to

3. Systematic and detailed 3D mapping of fractographic features of fractures that are

Surface morphology characteristics of rock discontinuities

OMER AYDAN l & YASUHIRO SHIMIZU 2

Abstract: A research program was undertaken to characterize the surface

Classification of discontinuities in rock masses

Surface morphology characterization

Centre-line average height (CLAH) is defined as;

Mean standard variation of height (MSVH) is defined as:

Root mean-square of height (RMSH) is defined

1 fxz=L 1 fxx=L ( de) 1/2

Weighted asperity inclination (WAI*) is defined as:

Mean standard variation of inclination (MSVI) is defined as:

MSVI = ~ o dxx dx. (6)

Root mean-square of inclination (RMSI) is defined as:

Structure function (SF) is defined as:

where T is a measure of the periodicity of asperities.

Fractal dimension is defined as:

where N is the number of steps, C is a constant, l is step length and D is a fractal

then the above equation can be rewritten as:

136E 137E 138E 139"~E

Fig. 2. Locations of the field measurements.

Surface morphology characterization system

TENSION D ISCON TIN U ITIES TD1

Fig. 3. Measured profiles.

dimensions of the profilometer; it is ma n u a l l y operated, weighs c. 5 kg and is made mainly

Table 1. Rock and discontinuity types at measurement sites

No. Location Rock type Discontinuity type

Measurements and discussion

TENSION DISCONTINUITY (TD4)

SHEAR DISCONTINUITY (SDE)

Fig. 5. Profiles of selected discontinuity types shown in Fig. 3.

Table 2. Surface morphologyparameters

Discontinuity Height parameters Inclination parameters Periodicity parameters RPL Fractal

1.04- BEDDINGPLANE 12-

0 40 80 120 160 10 40 80 120'1600 40 80 120 160

Fig. 8. Profiles of a sheeting joint.

Fig. 11. Relation between parameters CLAH and RPL.

proportionally increased. Furthermore, the width of scattering was small as compared

25 ' ' ' ' "' I ' ' ' '

The anisotropy of surface roughness measured using a