Advances in Complex Societal

Environmental and Engineered

Systems 1st Edition Mohamed Nemiche
Visit to download the full and correct content document:
https://textbookfull.com/product/advances-in-complex-societal-environmental-and-eng
ineered-systems-1st-edition-mohamed-nemiche/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Advances in Complex Societal Environmental and

Engineered Systems 1st Edition Mohamed Nemiche

https://textbookfull.com/product/advances-in-complex-societal-
environmental-and-engineered-systems-1st-edition-mohamed-
nemiche-2/

Mathematical Advances Towards Sustainable Environmental

Systems 1st Edition James N. Furze

https://textbookfull.com/product/mathematical-advances-towards-
sustainable-environmental-systems-1st-edition-james-n-furze/

An Introduction to Agent Based Modeling Modeling

Natural Social and Engineered Complex Systems with
NetLogo The MIT Press Uri Wilensky William Rand

https://textbookfull.com/product/an-introduction-to-agent-based-
modeling-modeling-natural-social-and-engineered-complex-systems-
with-netlogo-the-mit-press-uri-wilensky-william-rand/

Bioethanol and Natural Resources: Substrates, Chemistry

and Engineered Systems 1st Edition Ruben Michael
Ceballos

https://textbookfull.com/product/bioethanol-and-natural-
resources-substrates-chemistry-and-engineered-systems-1st-
edition-ruben-michael-ceballos/
Deep Learning for Vision Systems 1st Edition Mohamed
Elgendy

https://textbookfull.com/product/deep-learning-for-vision-
systems-1st-edition-mohamed-elgendy/

Deep Learning for Vision Systems 1st Edition Mohamed

Elgendy

https://textbookfull.com/product/deep-learning-for-vision-
systems-1st-edition-mohamed-elgendy-2/

Deep Learning for Vision Systems 1st Edition Mohamed

Elgendy

https://textbookfull.com/product/deep-learning-for-vision-
systems-1st-edition-mohamed-elgendy-3/

Advances in Energy and Environmental Materials Yafang

Han

https://textbookfull.com/product/advances-in-energy-and-
environmental-materials-yafang-han/

The Crisis of the 14th Century Teleconnections between

Environmental and Societal Change Bauch

https://textbookfull.com/product/the-crisis-of-the-14th-century-
teleconnections-between-environmental-and-societal-change-bauch/
Nonlinear Systems and Complexity
Series Editor: Albert C. J. Luo

Mohamed Nemiche
Mohammad Essaaidi Editors

Advances in
Complex Societal,
Environmental
and Engineered
Systems
Nonlinear Systems and Complexity

Volume 18

Series Editor
Albert C. J. Luo
Southern Illinois University
Edwardsville, IL, USA

More information about this series at http://www.springer.com/series/11433

Mohamed Nemiche • Mohammad Essaaidi
Editors

Advances in Complex
Societal, Environmental
and Engineered Systems

123
Editors
Mohamed Nemiche Mohammad Essaaidi
Ibn Zohr University ENSIAS College of Engineering
Agadir, Morocco Mohamed V University of Rabat
Rabat, Morocco

ISSN 2195-9994 ISSN 2196-0003 (electronic)

Nonlinear Systems and Complexity
ISBN 978-3-319-46163-2 ISBN 978-3-319-46164-9 (eBook)
DOI 10.1007/978-3-319-46164-9

Library of Congress Control Number: 2016962275

© Springer International Publishing Switzerland 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

This book addresses recent technological progress that has led to an increased
complexity in societal, ecological, and engineered systems. This complexity is
characterized by the emergence of new proprieties and structures resulting from
nonlinear interactions among system elements and between system and its envi-
ronment. This volume provides researchers and managers with qualitative and
quantitative methods, bottom-up and holistic approaches for handling many features
of the complex contemporary reality. This book is composed of three parts with
a total of 13 chapters: Part I (Chaps. 1–6) focuses on societal and ecological
systems, Part II (Chaps. 7–12) deals with approaches for understanding, modeling,
forecasting, and mastering complex systems, and Part III (Chap. 13) includes real-
life examples. Each chapter of this book has its own special features; it is a
self-contained contribution of researchers working in different fields of science
and technology relevant to the study of complex systems including Agent-Based
Modeling, General Systems Theory, and Mathematical Modeling.
In the chapter “ProtestLab: A Computational Laboratory for Studying Street
Protests,” Lemos, Coelho, and Lopes present an Agent-Based model for the
simulation of street protests, with multiple types of agents (protesters, police, and
“media”) and scenario features (attraction points, obstacles, and entrances/exits). In
this model agents can have multiple “personalities,” goals, and possible states. The
model includes quantitative measures of emergent crowd patterns, protest intensity,
police effectiveness, and potential “news impact,” which can be used to compare
simulation outputs with estimates from videos of real protests for parameterization
and validation. ProtestLab was applied to a scenario of policemen defending a
government building from protesters and reproduced many features observed in real
events, such as clustering of “active” and “violent” protesters, formation of moving
confrontation lines, occasional fights and arrests, “media” agents wiggling around
“hot spots,” and policemen with defensive or offensive behavior.
In the chapter “A Generic Agent Based Model of Historical Social Behavior
Change,” Ahmed M’hamdi et al. describe and discuss how human societies change
over time. The main objective of this work is to build a generic agent-based model

v
vi Preface

of artificial social evolution in order to explore the East–West and North–South

divides, the migration process, and some possible evolution ways.
In the chapter “Understanding Social Systems Research,” Klein embarks on the
challenge of social complexity. He illustrates how decomposing our notion of the
social into political and cultural aspects improves our possibilities to understand
social systems. The praxeological departure point of the chapter is the reception
of the volatile, uncertain, complex, and ambiguous (VUCA) world in project
management. Referring to the works of Niklas Luhmann the importance of meaning
creation and sensemaking for social systems comes into sight. Leading from
Systemic Inquiry to Systems Analytics, Klein describes the state of the art of social
systems research and the implications for problem structuring and research design.
Klein ends on a reflection on the necessity of systems literacy and social design
impact evaluation as a way forward for the twenty-first century to become systems
savvy.
In the chapter “ForestSim: An Agent-Based Simulation for Bioenergy Sustain-
ability Assessment,” Mark Rouleau proposes the use of Agent-Based Modeling
(ABM) for bioenergy sustainability assessment. His study focuses on the assessment
of second generation bioenergy from woody biomass as a potential renewable
energy alternative to fossil fuels. He uses ForestSim, an ABM simulation to assess
the sustainability of bioenergy development in the heavily forested region of the
Upper Peninsula of Michigan, USA. He explains how ABM makes it possible to
investigate bioenergy sustainability as a complex phenomenon that emerges within
a coupled human and natural system. He compares the ABM simulation approach
to conventional Life Cycle Assessment and explains why ABM is necessary to
understand the complex dynamics of a bioenergy system. Finally, he compares his
approach to an existing sustainability study in the UP with the goal of estimating
biomass availability within the context of economic viability, ecological balance,
and the social acceptance of bioenergy. He believes that his approach offers
valuable insight into bioenergy sustainability and provides an important alternative
assessment method.
In the chapter “Toward a Complex Concept of Sustainability,” Humberto Mariotti
examines the relationships between complexity theory and some situations not
usually approached in the pertinent literature. One of the studied aspects is the
need to insert human nature among the variables of the sustainability/complexity
relationships as a way to “complexify” them, as well as to provide a kind of realism
that is badly needed to work on complex issues. The complexity of human nature
is studied from several perspectives. The chapter is closed with the presentation of
what the author calls a complex concept of complexity.
In the chapter “Effects of Policy Decision-Making on Riparian Corridors in a
Semi-Arid Desert: A Modeling Approach,” Pope and Gimblett project the impact
of an urban community on vegetation along the Upper San Pedro River. The
riparian corridor is highly valued as important habitat in the American Southwest;
nearly 57,000 acres of public land in the watershed are protected by Congress as a
Riparian National Conservation Area. However, the corridor is at risk from extensive
groundwater pumping to satiate growing urban demand. This work was designed
Preface vii

to produce the potential effects of policy decision-making on riparian vegetation,

including how much, where, and the likelihood of change occurring. A model was
developed that combined both agent-based and Bayesian modeling techniques with
three submodels: social, hydrological, and ecological. Using a variety of policy
scenarios, the agent-based model predicted the percent change in vegetation area
as well as the area the change is likely to occur while the Bayesian model produced
likelihoods of change. The models were developed to be stakeholder-friendly, so
that local decision-makers could use them to help make better decisions.
In the chapter “Dialectical Systems Theory as a Way to Handle Complex Sys-
tems,” Matjaz Mulej et al. explain that complexity in natural, social, organizational,
and technological phenomenon is hardly to be handled by the specialist vision
unless one applies one or more of the versions of systems theory, which support
interdisciplinary creative cooperation, such as the Dialectical Systems Theory. They
also demonstrate that the related applied methodologies and methods are also very
useful, because they can be used without too much complexity of theories, such as
“Corporate Social Responsibility” (ISO 26000), “USOMID,” “Six Thinking Hats”
and their synergetic applications.
In the chapter “Reducing Complexity of Nonlinear Dynamic Systems,” Nagy-
Kiss et al. present a systematic procedure to transform a nonlinear system into
a polytopic one without causing any information loss, contrarily to most existing
studies in the field. They also present a robust observer synthesis with respect
to internal/external perturbations, modeling parameterization errors, and unknown
inputs for the estimation of the state variables. The above-mentioned points are
applied to an activated sludge wastewater treatment plant, which is a complex
chemical and biological process.
In his chapter “A Few Reflections on the Quality of Emergence in Complex
Collective Systems,” Vincenzo De Florio discusses the major factors that play a role
in the persistence of emergence in collective systems. By means of those factors,
a “general systems theory” of emergent systems is proposed. Several classes of
emergent systems are then exemplified, ranging from simple aggregations of simple
parts up to complex organizations of complex collective systems. The chapter also
discusses the relationship between quality of emergence and resilience, namely the
persistence of system identity. As suggested already by G. W. Leibniz, De Florio
argues here that emergence and its quality may be better assessed by considering
the interplay between intrinsic, extrinsic, and “social” aspects.
In the chapter “Link Structure Analysis of Urban Street Networks for Delineating
Traffic Impact Areas,” Wen et al. propose an innovative analytical procedure of
ranking algorithm, the Flow-based PageRank (FBPR), for investigating the traffic
flow concentration, complexity of street network structure, and traffic impact areas.
A network modularity algorithm is used for delineating the traffic impact areas
that will be affected by traffic congestion. The results indicate, by overlapping
the topological structure of street network and flow concentration, street segments
prone to traffic congestion are identified, including the Central Business Districts
(CBD), and the areas proximate to the stations of the combination of MRT and
viii Preface

train railway systems. Meanwhile, the delineation of traffic impact areas could be
spatially targeted at priorities of traffic improvement for city planners.
In the chapter “Logic, Mathematics and Consistency in Literature: Searching for
Don Quixote’s Place,” Montero et al. combine fuzzy logic with other techniques
to analyze the consistency of the linguistic discourse about the village Miguel de
Cervantes (1547–1616) decided not to reveal in his classical Don Quixote’s novel.
In particular, the authors consider Cervantes linguistic description of Don Quixote’s
trips from and to that place in order to check if such information is consistent with
the map of La Mancha, and allowing a more or less constant walking speed march
per day. From this complex system of information, it is then concluded that there
is in fact a small region in the center of Campo de Montiel that meets all estimated
walking times per day, showing that Cervantes linguistic description of the trips
involving the hidden place is in this sense consistent.
In the chapter “Energy-Efficient Buildings as Socio-Technical Complex Systems:
Approaches and Challenges,” Lachhab et al. present important metrics that assess
performance and occupants’ comfort in energy-efficient buildings and study their
relationships with building physical properties, equipment control, outdoor envi-
ronment, and occupants’ behavior and activities. The authors analyze occupants’
actions and behaviors in context taking into account the complex interlinked entities,
situations, processes, and their dynamics. Lachhab et al. then review existing control
approaches and solutions for energy efficiency in complex buildings. They highlight
simulation tools that aim to study and analyze approaches for energy consumption,
occupants’ comfort, and CO2 emissions. They also introduce their ongoing work
related to modeling and control of these complex systems by highlighting the
necessity of the development of intelligent building management systems that could
include run-time processing techniques of large amount of data for deploying
context-aware event-triggered control techniques.
In the chapter “Modeling Space-Time-Action Modularity and Evolution of
Living Systems,” Pierre Bricage develops a new paradigm of “the gauge invariance
of living systems” which allows him to define functionally and dynamically the
hierarchical fractal organization of all living systems, from the quantum of Planck to
the whole universe. The ontogeny of interactions and the interaction of ontogenies
are topologically related through a power law of exponent 3/2, which is basically
related to the Brownian motion. His work was motivated by the need to understand
why all living systems are emerging through juxtaposition and embedment of
previous ones in a new blueprint that is always an Association for the Reciprocal
and Mutual Sharing of Advantages and Disadvantages: “interaction is construction
and construction is interaction” both for the spaces and times, which are juxtaposed
and embedded simultaneously into limited spatial and temporal networks, within
an “independent of mass-entropy relationship inter-active optimal surface flow of
flows” organizing control.
Preface ix

The editors would like to thank all the authors who contributed to this book.
Special thanks also go to Mr. Christopher Coughlin and Mr. Abdeljalil Haib for
their collaboration in this book project.

Agadir, Morocco Mohamed Nemiche

Rabat, Morocco Mohammad Essaaidi
Contents

Part I Societal and Ecological Systems

ProtestLab: A Computational Laboratory for Studying Street Protests . . 3
Carlos M. Lemos, Helder Coelho, and Rui J. Lopes
A Generic Agent-Based Model of Historical Social Behaviors Change . . . . 31
Ahmed M’hamdi, Mohamed Nemiche, Rafael Pla Lopez,
Fatima Ezzahra SFA, Khabid Sidati, and Omar Baz
Understanding Social Systems Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Louis Klein
ForestSim: An Agent-Based Simulation for Bioenergy
Sustainability Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Mark Rouleau
Toward a Complex Concept of Sustainability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Humberto Mariotti
Effects of Policy Decision-Making on Riparian Corridors
in a Semi-arid Desert: A Modeling Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Aloah Pope and Randy Gimblett

Part II Approaches for Understanding, Modeling,

Forecasting and Mastering Complex Systems
Dialectical Systems Theory as a Way to Handle Complex Systems . . . . . . . . 145
Matjaz Mulej, Zdenka Ženko, and Nastja Mulej
Reducing Complexity of Nonlinear Dynamical Systems . . . . . . . . . . . . . . . . . . . . . 159
Anca Maria Nagy-Kiss, Georges Schutz, Benoît Marx,
Gilles Mourot, and José Ragot

xi
xii Contents

A Few Reflections on the Quality of Emergence in Complex

Collective Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Vincenzo De Florio
Link Structure Analysis of Urban Street Networks for
Delineating Traffic Impact Areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Tzai-Hung Wen, Wei-Chien-Benny Chin, and Pei-Chun Lai
Logic, Mathematics and Consistency in Literature: Searching
for Don Quixote’s Place . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Javier Montero, Gracia Rodríguez-Caderot, and Pilar Romero
Energy-Efficient Buildings as Complex Socio-technical
Systems: Approaches and Challenges. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
F. Lachhab, M. Bakhouya, R. Ouladsine, and M. Essaaidi

Part III Real-Life Examples

Modelling Space-Time-Action Modularity and Evolution
of Living Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Pierre Bricage
List of Editors and Contributors

Editors

Mohamed Nemiche is an Assistant Professor at Ibn

Zohr University, Morocco. He received his Ph.D. from
the University of Valencia, Spain, in 2002 and his Habil-
itation in Computer Science from Ibn Zohr University,
Morocco, in 2012. His research interests are related to
the study of complex systems with agent-based social
simulation and mathematical modeling. Dr. Nemiche is
the founder and the current president of the Moroccan
Society of Complex Systems and the founder and cochair
of the World Conference on Complex Systems. He is the
winner of the Kybernetes Research Award (Pittsburg, USA) in 2002. Dr. Nemiche
is the guest Editor of the special issue Advances in Complex Intelligent Systems
in International Journal of Intelligent Systems and of the special issue Conceptual
Models of Complex Systems in Systems Research and Behavioral Science.

Mohamed Essaaidi is an IEEE Senior Member

and the current Director of National College of IT
(ENSIAS) of Mohammed VI Souissi University, Rabat,
Morocco. He was a Professor of Electrical and Com-
puter Engineering at Abdelmalek Essaadi University,
Morocco, from 1993 till 2011. He is the founder and
Chairman of the IEEE Morocco Section, the founder
of IEEE Computer and Communication Societies Joint
Morocco Chapter, the founder and Chair of IEEE
Antennas and Propagation Society and Microwave Theory and Techniques Society
Morocco Joint Chapter, and the founder of IEEE Education Society Morocco
Chapter. Prof. Essaaidi has authored and coauthored 5 books and more than 120
papers in international refereed journals and conferences in the field of Electrical,

xiii
xiv List of Editors and Contributors

Information and Communication Technologies. He has been the Editor-in-Chief

of International Journal on Information and Communication Technologies, Serial
Publications, India, since 2007. Prof. Essaaidi also founded and has been the
General Chair of the Mediterranean Microwave Symposium since the year 2000,
Information and Communication Technologies International Symposium in 2005
and 2007, the International Conference on Multimedia Systems and Computing in
2009, 2011, and 2012, and the International Conference on Complex Systems in
2012. Prof. Essaaidi holds four patents on antennas for very high data rate UWB and
multiband wireless communication systems and high resolution medical imaging
systems.

Contributors

M. Bakhouya International University of Rabat, Sala el Jadida, Morocco

Omar Baz Ibn Zohr University, Agadir, Morocco
Pierre Bricage AFSCET The French Association for Systems and Cybernetic
Sciences, Paris, France
IASCYS The International Academy for Systems and Cybernetic Sciences, Vienna,
Austria
Helder Coelho BioISI and Faculdade de Ciências da Universidade de Lisboa,
Lisbon, Portugal
Wei-Chien-Benny Chin Department of Geography, National Taiwan University,
Taipei City, Taiwan
Vincenzo De Florio Vlaamse Instelling voor Technologisch Onderzoek, Remote
Sensing Department, Boeretang, Mol, Belgium
Vrije Universiteit Brussel, Global Brain Institute & Evolution, Complexity and
COgnition group, Pleinlaan, Brussels, Belgium
M. Essaaidi Ecole Nationale Supérieure d’Informatique et d’Analyse des Sys-
tèmes, Rabat, Morocco
Randy Gimblett School of Natural Resources and the Environment,
The University of Arizona, Tucson, AZ, USA
Louis Klein The Systemic Excellence Group, Berlin, Germany
Pei-Chun Lai Department of Geography, National Taiwan University, Taipei City,
Taiwan
List of Editors and Contributors xv

F. Lachhab International University of Rabat, Sala el Jadida, Morocco

Ecole Nationale Supérieure d’Informatique et d’Analyse des Systèmes, Rabat,
Morocco
Carlos M. Lemos CISDI and Instituto de Estudos Superiores Militares, Lisbon,
Portugal
BioISI and Faculdade de Ciências da Universidade de Lisboa, Lisbon, Portugal
ISCTE-Instituto Universitário de Lisboa, Lisbon, Portugal
Rui J. Lopes ISCTE-Instituto Universitário de Lisboa, Lisbon, Portugal
Instituto de telecomunicações, Lisbon, Portugal
Humberto Mariotti Center for Workplace Diversity Research (CWDR), School
of Advanced Studies, University of Phoenix, Phoenix, AZ, USA
Benoît Marx Université de Lorraine, CRAN, UMR 7039, Cedex, Vandoeuvre-les-
Nancy, France
CNRS, CRAN, UMR 7039, Vandoeuvre-les-Nancy, France
Ahmed M’hamdi Ibn Zohr University, Agadir, Morocco
Javier Montero Faculty of Mathematics, Complutense University, Madrid, Spain
Gilles Mourot Université de Lorraine, CRAN, UMR 7039, Cedex, Vandoeuvre-
les-Nancy, France
CNRS, CRAN, UMR 7039, Vandoeuvre-les-Nancy, France
Matjaz Mulej Faculty of Economics and Business, University of Maribor,
Maribor, Slovenia
Nastja Mulej Nastja Mulej s.p., Ljubljana, Slovenia
Anca Maria Nagy-Kiss Environmental Research & Innovation Department, Lux-
embourg Institute of Science and Technology, Esch/Alzette, Luxembourg
Mohamed Nemiche Ibn Zohr University, Agadir, Morocco
R. Ouladsine International University of Rabat, Sala el Jadida, Morocco
Rafael Pla Lopez University of Valencia, Valencia, Spain
Aloah Pope School of Natural Resources and the Environment, The University of
Arizona, Tucson, AZ, USA
José Ragot Université de Lorraine, CRAN, UMR 7039, Cedex, Vandoeuvre-les-
Nancy, France
CNRS, CRAN, UMR 7039, Vandoeuvre-les-Nancy, France
xvi List of Editors and Contributors

Gracia Rodríguez-Caderot Faculty of Mathematics, Complutense University,

Madrid, Spain
Pilar Romero Faculty of Mathematics, Complutense University, Madrid, Spain
Mark Rouleau Social Sciences Department, Environmental and Energy Policy
Program, Michigan Technological University, Houghton, MI, USA
Georges Schutz Environmental Research & Innovation Department, Luxembourg
Institute of Science and Technology, Esch/Alzette, Luxembourg
Fatima Ezzahra SFA Ibn Zohr University, Agadir, Morocco
Khabid Sidati Ibn Zohr University, Agadir, Morocco
Tzai-Hung Wen Department of Geography, National Taiwan University, Taipei
City, Taiwan
Zdenka Ženko Faculty of Economics and Business, University of Maribor,
Maribor, Slovenia
 Part I
Societal and Ecological Systems
ProtestLab: A Computational Laboratory
for Studying Street Protests

Carlos M. Lemos, Helder Coelho, and Rui J. Lopes

Abstract We present an Agent-Based model called ProtestLab for the simulation

of street protests, with multiple types of agents (protesters, police and ‘media’) and
scenario features (attraction points, obstacles and entrances/exits). In ProtestLab
agents can have multiple “personalities” (implemented via agent subtypes), goals
and possible states, including violent confrontation. The model includes quantitative
measures of emergent crowd patterns, protest intensity, police effectiveness and
potential ‘news impact’, which can be used to compare simulation outputs with
estimates from videos of real protests for parametrization and validation. ProtestLab
was applied to a scenario of policemen defending a government building from
protesters (typical of anti-austerity protests in front of the Parliament in Lisbon,
Portugal) and reproduced many features observed in real events, such as clustering
of ‘active’ and ‘violent’ protesters, formation of moving confrontation lines, occa-
sional fights and arrests, ‘media’ agents wiggling around ‘hot spots’ and policemen
with defensive or offensive behaviour.

1 Introduction

Street protests and riots are manifestations of social conflict, which results from
feelings of frustration due to relative deprivation (RD) and cleavages within the
society (Gurr 1968). They may be large or small, peaceful of violent, organized or

C.M. Lemos ()

CISDI and Instituto de Estudos Superiores Militares, Lisbon, Portugal
BioISI and Faculdade de Ciências da Universidade de Lisboa, Lisbon, Portugal
ISCTE - Instituto Universitário de Lisboa, Lisbon, Portugal
e-mail: cmrso@iscte-iul.pt
H. Coelho
BioISI and Faculdade de Ciências da Universidade de Lisboa, Lisbon, Portugal
R.J. Lopes
ISCTE - Instituto Universitário de Lisboa, Lisbon, Portugal
Instituto de Telecomunicações, Lisbon, Portugal

© Springer International Publishing Switzerland 2017 3

M. Nemiche, M. Essaaidi (eds.), Advances in Complex Societal, Environmental
and Engineered Systems, Nonlinear Systems and Complexity 18,
DOI 10.1007/978-3-319-46164-9_1
4 C.M. Lemos et al.

spontaneous, isolated or recurrent, and are an instrument by which citizens try to

challenge authorities and influence the public opinion (Klandermans et al. 2014).
History provides examples of large protests playing a decisive role in overthrowing
regimes, as in the October Revolution in Russia in 1917 and the Iran Revolution
of 1979 (Kuran 1989). Recently, widespread access to Social Networks (SN) and
Information and Communication Technologies (ICT) allowed activists to convoke,
coordinate and show protest events to a worldwide audience (Comninos 2011).
The relationship between protests and the political and social context can be
viewed as a complex and path-dependent process. It involves multiple scales
(ranging from social context to microscopic motivations and behaviours), factors
(social, situational and individual), actors and interactions. Macroscopic context
factors (political, social and economic) lead to conflict manifestations such as
boycotts, strikes or protests, which in turn change the social context (e.g. the
perceived grievance or the legitimacy of the government or central authority). These
direct and feedback links may lead to stabilization or growth of the social unrest.
Understanding, predicting and if possible controlling this process is an important
topic in sociology, social psychology, political science, military and security studies,
and more recently social simulation (Lemos et al. 2013).
Social conflict phenomena have been studied using Agent-Based models (ABM)
(Gilbert and Troitzsch 2005; Lemos et al. 2013). Existing ABM of social conflict
and violence are focused on specific scales and types of phenomena, and can
be broadly classified as ‘abstract’, ‘middle-range’ and ‘facsimile’ (Gilbert 2007).
‘Abstract’ ABM have been proposed for explaining the dynamics and characteristic
patterns of large-scale civil and ethnic violence (Epstein 2002; Epstein et al. 2001),
revolution (Makowsky and Rubin 2011), worker protest (Goh et al. 2006), or even
urban crime (Fonoberova et al. 2012), using a minimum number of agent types with
very simple behaviour and interactions. ‘Middle range’ models are used to describe
phenomena at definite space and time scales with context-specific information
(Davies et al. 2013; Parunak et al. 2014). ‘Facsimile’ models attempt to describe the
system’s dynamics with as much realism as possible, by considering the multiplicity
of relevant actors, behaviours, interactions and scenario features. Models of this
latter type have been proposed for studying crowd dynamics such as pedestrian flow
in open or closed spaces (Pelechano et al. 2007; Still 2000), panic evacuation and
crowd control (see TSO 2010a,b for a review), and for modelling riotous crowds
which involve (possibly violent) interactions between protesters and police forces
(Durupınar 2010; Lacko et al. 2013; Torrens and McDaniel 2013).
In this paper we describe a new ABM called ProtestLab of ‘facsimile’ type
for modelling street protests, which is part of an ongoing work on social conflict
simulation (Lemos et al. 2013). The purpose of ProtestLab is to simulate the emer-
gent patterns in street protests, such as clustering of violent and active protesters,
the formation of confrontation lines between violent protesters and police cordons,
fights and arrests, and ‘media’ agents wiggling near fights, using a relatively simple
reactive agent architecture common to all agent types. The research questions for
the present work are:
A Computational Laboratory for Studying Street Protests 5

– How do the features of the protest space and the crowd size, composition and
initial placement affect the crowd formation patterns (wandering, clustering and
fighting) and protest intensity?
– How does violent confrontation arise? Once initiated, does it spread to the bulk
of the crowd or remains confined to specific types and clusters of protesters?
– How can different police actions influence the intensity of violent confrontation
and the ability for defending a perimeter?
The novel features of ProtestLab with respect to other ABM are: (i) multiple
types of agents (‘protesters’, ‘cops’ and ‘media’), with multiple behaviours repre-
sented by agent subtypes; (ii) several types of scenario features (attraction points,
obstacles, entrances and exits); (iii) a simple but efficient reactive agent architecture
which allows all agents to have multiple goals and spatially oriented interactions;
and (iv) quantitative measures of emergent crowd patterns, protest intensity, police
vs protesters effectiveness and potential “media” coverage.
Three subtypes of ‘protester’ agents were considered, ‘hardcore’, ‘hanger-
on’ and ‘bystander’, which broadly correspond to typical protester behaviours
well characterized in micro-situational theories of violence (Collins 2008, 2009;
Wikström and Treiber 2009), ABM of clustering and fighting (Jager et al. 2001), and
also easily identified in many videos of real protests. Four subtypes of ‘cop’ agents
were included, namely ‘command’, ‘offensive’, ‘defensive’ and ‘multi role’, which
correspond to different mission profiles and assigned tasks in a police force. All
agents can be one of four states, ‘quiet’, ‘active’ (waving, shouting, etc.), ‘violent’
(ripe for confrontation) or ‘fighting’. Fights only occur when ‘violent’ protesters
and cops are in contact, such as in body fighting (fights with weapons such as stones
and tear gas are not considered in the present version of the model).
The advantage of ProtestLab with respect to existing ABM of crowd dynamics
and violent confrontation is the capability of running realistic simulations of street
protests with several hundred agents using small computers. The present version of
ProtestLab is not intended for modelling fights between two groups (e.g. supporters
of rival parties or football teams) mediated by a police force or events of collective
violence that are not politically motivated, although the model can be extended to
treat these cases. Consequently, the theories and models related to pedestrian flow,
panic evacuation, ethnic violence or violence in two-party crowds (e.g. in football
stadiums) will be mentioned only in the aspects relevant for the issues considered in
ProtestLab.
The remainder of this paper is organized as follows. In Sect. 2 we present the
theoretical background for ProtestLab. In Sect. 3 we describe the model entities,
design concepts, quantitative measures of the emergent properties and issues
related to parametrization and validation. In Sect. 4 we present the results of a
set of simulations for a scenario of a police force defending the entrance of a
government building from protesters, for different crowd densities and compositions
(proportions of subtypes of ‘protester’ agents), average aggressiveness, and police
‘personality’ or mission profile and the action of a police “command” agent.
Section 5 contains the discussion of the model results with emphasis on the model’s
6 C.M. Lemos et al.

capabilities, limitations and potential for a realistic description of the dynamics of

street protests. Finally, Sect. 6 contains a summary of conclusions and a reference
to intended further developments.

2 Theoretical Background and Previous Models

Several ABM have been proposed for the simulation of conflict phenomena of
different types and scales. Epstein et al. (2001) introduced an ‘abstract’ ABM of
large-scale rebellion against a central authority (Model I) or ethnic violence between
two rival groups (Model II) with two types of agents, population and cops in a
torus space environment. Population agents can be in three states (‘quiet’, ‘active’
or ‘jailed’). Both types of agents have one movement rule M (move to a random
empty cell within the agent’s vision radius) and one action rule. Population agents
that are not ‘jailed’ become ‘active’ according to the following threshold (action)
rule:

GN >T (1)

where G D H  .1  L/, G  U .0; 1/ is the (heterogeneous) grievance, L 2 Œ0; 1

is the (homogeneous) perceived legitimacy, N D R  P is the net risk perception,
R  U .0; 1/ is the (heterogeneous) risk aversion,

P D 1  exp.k.C=A/v / (2)

is the estimated arrest probability, and T is a threshold. In Eq. (2), k is a constant

with adopted value 2.3 (see Epstein et al. 2001 or Epstein 2002) and Cv and Av
are the numbers of ‘cops’ and ‘active’ citizens within the citizen’s vision radius,
respectively. Cops arrest a random ‘active’ agent within their vision radius (action
rule) and arrested agents are removed from the model space for J  U .0; Jmax /
cycles (jail term). This model successfully explained many important features
of large scale civil violence or rebellion, such as intermittent bursts of violence
(punctuated equilibrium), or the effects of gradual or sudden variations of the
legitimacy and security apparatus (number of cops). The ABM of Epstein et al.
(2001) has been extended by other authors and applied to other conflict phenomena
such as worker protest (Kim and Hanneman 2011) or even urban crime (Fonoberova
et al. 2012). However, ‘abstract’ ABM are unsuitable for modelling smaller scale
phenomena like street protests, because they do not describe multiple actors,
interactions and environmental (scenario) features.
Davies et al. (2013) presented a model of the London riots of 6th–10th August
2011, with rioters and policemen, which can be classified as ‘middle-range’. This
model considers definite space and time scales (streets and sites of London affected
by the riots) and considers three processes: the decision to riot, the choice of the site
for rioting and the interaction with the police forces. The decision to riot is modelled
A Computational Laboratory for Studying Street Protests 7

as a function of deprivation and the attractiveness of the sites for rioting. The choice
of the rioting site is modelled via an utility function which takes into account
the distance from the rioter’s present position, the attractiveness and the level of
deterrence expected at each site. As rioters interact with police they may be arrested
with probability described by a function similar to that in Eq. (2). Although models
of this type (e.g. Davies et al. 2013, Parunak et al. 2014) provide a representation
of rioting events and include context-specific information for their parametrization
and validation, they do not describe the micro-scale features of the interactions
between protesters and the police forces (such as clustering of violent, active and
quiet protesters, the formation of confrontation lines with fights and arrests, etc.).
Modelling micro-scale processes in crowd dynamics is a problem of great
theoretical and practical importance. There are two types of crowd models relevant
for the present work: models of pedestrian flow and crowd density, and models
of riotous crowds. The study of pedestrian flow and the formation of spots with
dangerous crowd density in panic evacuation has been intensely studied using meth-
ods from fluid dynamics (Helbing and Molnár 1995), cellular automata (Klüpfel
2003) and particle dynamics (Leggett 2004). Several well developed models are
available for such applications (see Still 2000, and TSO 2010a,b for a review of
existing models), which use a continuous space and time description of the motion
of pedestrians as self-propelled particles subject to a force with three components
(Social Force Model, Helbing and Molnár 1995) that express individual tendencies
to (i) maintain a desired speed towards a wanted destination point; (ii) keep
clear from other pedestrians or obstacles (repulsion force, decaying exponentially);
and (iii) approach attractive features (such as other persons or displays) (Helbing
and Molnár 1995; Helbing et al. 2001). The main difficulties are the need for
algorithms for collision avoidance, description of the contact interactions between
pedestrians and obstacles, and the need for integrating the equations of motion for
each pedestrian (Pelechano et al. 2007). Discrete space and time models, in which
pedestrians move in discrete steps and only occupy empty cells within their move
range are simpler, because integration of the equations of motion for each pedestrian
is not necessary (Klüpfel 2003). If the grid cell size is large enough for the repulsion
forces to be negligible and only one agent per grid cell is allowed, then the agents’
movement can be modelled in a very simple way. This latter approach was used in
ProtestLab due to its simplicity and advantage for handling large numbers of agents.
Models for describing processes specific to riotous crowds, such as the formation
of confrontation lines between protesters and police cordons, patterns of clustering,
simple provocation or physical fights and arrests, are comparatively less developed
than models of pedestrian flow and panic evacuation. Jager et al. (2001) presented
an ABM of clustering and fighting between two parties (e.g. hooligans supporting
two different football teams) with three kinds of agents, ‘hardcore’, ‘hangers-on’
and ‘bystanders’. These subtypes are consistent with typical behaviours identified
in micro-situational theories of violence (e.g. Collins 2008, 2009, Wikström and
Treiber 2009) and can be confirmed in many videos of real protests. The agents’
internal state is characterized by an ‘aggression motivation’ that increases/decreases
depending on the difference between the number of visible agents of own and
8 C.M. Lemos et al.

opposing groups. Agents approach members of own group with preference for their
‘acquaintances’ and, depending on their ‘aggression motivation’, may approach and
start fighting with agents of the other group. The simulated results showed that
fighting occurs when there are large asymmetric groups with significant proportions
of ‘hardcore’, and that ‘bystanders’ are victimised. This model has the advantages
of including aggression as a state variable, memory effects in the ‘aggression
motivation’, and purposeful movement to simulate the collective behaviour of get-
ting support before engaging in violence. However, it does not include connection
between aggressiveness and social context variables and deterrence effects (there
are no ‘cop’ agents).
Durupınar (2010) presented a sophisticated ABM for different types of crowds
with psychological factors (personality, mood and emotions), based on a more
sophisticated agent architecture which allows representation of emotion contagion.
This model includes several types of agent personalities (protesters, agitators, and
police agents) and events (such as fires and explosions), and has important features
such as emotion contagion, multiple actors and multiple actions (waving, kicking
and punching). Although the model is computationally demanding, Durupınar
presents simulations with several hundreds of agents, including protesters marching
in a street watched by policemen at the sides. However, it is not clear how the
model would represent clustering, moving confrontation lines, fights and arrests in
a realistic way over many time steps, and also lacks quantitative measures of these
emergent features.
Torrens and McDaniel (2013) proposed an ABM for simulation of riotous crowds
with four types of agents, ‘Civil’, ‘Rebel’, ‘Jailed’ and ‘Police’1 and the model space
is a 2D torus in which sites can be blocked to represent obstacles. The attributes
of the first three types of agents are the same as in Epstein’s model (H, G, Pa , N
and T) and the transition from ‘Civil’ to ‘Rebel’ is done using Eqs. (1)–(2). The
geographic behaviour is implemented with three modules: spatial cognition, loco-
motion and steering. The spatial strategy (goal-driven approach/avoid movement)
is implemented by defining weights for ‘Rebel’ and ‘Police’ agents to approach or
avoid the other ‘Police’ and ‘Rebel’ agents. Although this model has sophisticated
locomotion and steering modules and is able to represent both low and high density
crowds, it falls short in other aspects critical for simulation of street protests.
For instance, the set of agent types and their states is the same as in Epstein’s
‘abstract’ ABM. Also, the approach/avoid weights are constant, whereas in reality
they depend on local superiority or inferiority of protesters and police agents. The
results reported in Torrens and McDaniel (2013) do not show the crowd patterns
typical of street protests with violent confrontation, and for a simulation scenario
of mass protest with 5000 citizens 42 % of the crowd ends up arrested, which is
obviously unrealistic.

1
In fact there are only two types of agents. ‘Civil’, ‘Rebel’ and ‘Jailed’ are different states of a
single agent type, as in Epstein’s ABM (Epstein 2002; Epstein et al. 2001).
A Computational Laboratory for Studying Street Protests 9

Lacko et al. (2013) proposed an ABM for the simulation of urban riots, with
police, protesters and a remotely operated vehicle (called “BOŽENA RIOT”)
designed for riot control, which considers cognitive modelling of agents, collision
avoidance and crowd pressure calculations. The authors claim to have implemented
the PECS (Physical Conditions, Emotional State, Cognitive Capabilities and Social
Status) agent architecture (Schmidt 2002), and model fear in terms of the number of
other persons within an agent’s “intimate zone”. Pressure calculations are performed
by dividing the crowd in groups, then defining a group direction which is used to
compute the pressure for individual agents in a virtual grid. However, these authors
do not give details of their PECS implementation and do not report any results
in Lacko et al. (2013). Thus, it is not possible to know whether or not the model
reproduces realistic patterns of confrontation, how it handles variable density and
pressure within the crowd, or how many agents can be included in one simulation.
In summary, existing ABM of riotous crowds have limited ability for modelling
the variety of actors, states and micro-interactions in typical street protests. It would
be desirable to integrate in one model different types of actors (protesters and
policemen with differentiated behaviour), interactions (fights and arrests), scenario
features, and quantitative measures of crowd patterns and protest intensity, for more
realistic simulations. Thus, the purpose of ProtestLab is to overcome some of these
drawbacks.

3 Model Description

In this section we present a summary description of the ProtestLab ABM, includ-

ing the model entities and interactions, design concepts, process overview and
scheduling, quantitative measures of crowd patterns and protest intensity, and
parametrization and validation issues.

3.1 Agents, Scenario and Interactions

ProtestLab was implemented in NetLogo and has the following entities: observer
(world, global variables), turtles (agents) and patches (space cells in the
model space). Three types of agents were implemented, ‘cops’, ‘protesters’ and
‘media’. The first two types correspond to the main interacting players and the latter
is necessary for describing potential ‘news impact’ of protest events. The presence
of “media” agents also changes the microscopic context by attracting curious
protesters. Three subtypes of ‘protesters’ were considered, ‘hardcore’, ‘hanger-on’
and ‘bystanders’ which broadly correspond to characteristic behaviours observed
in protests and reported in the literature (Collins 2008; Jager et al. 2001), and four
subtypes of ‘cop’ agents, ‘defensive’, ‘offensive’, ‘multi-role’ and ‘command’, to
reflect mission profile differentiation and allow a combination of command-oriented
10 C.M. Lemos et al.

Fig. 1 Simplified class diagram for the agents and scenario features considered in ProtestLab

vs self-organized police behaviour. In the present version of ProtestLab, only one

police commander agent is considered. Figure 1 shows the simplified class diagram
for the model entities in ProtestLab.
In what follows Ai is a generic agent i and Ej a generic grid cell (patch)
in the environment space. Cells Ej are described by their position and type, i.e.
SEj 2 {‘empty’,‘flag’,‘obstacle’,‘entrance’,‘exit’} and the
scenario consists of a 2D grid (to represent a protest space with physical boundaries)
of max-pxcor per max-pycor patches. It is assumed that the patch size has
dimensions 1 m  1 m and the time scale (tick duration) is 1 s which gives a
constant speed of 1 m s1 .2

3.2 Design Concepts and Agent Architecture

ProtestLab is based on the following design concepts: (i) agents are reactive and
rule-based, move in discrete time and space increments (at most one agent per grid
cell) and change their behavioural state and position depending on the local context;
(ii) the agents’ goal-directed movement is implemented by minimizing a penalty
function that involves weights for avoiding or approaching other agents and scenario
features (as in Ilachinsky 2004); and (iii) agents are activated in random order with a

2
However, if the global variable move-radius is increased from its default value of 1 a velocity
varying in steps of 1/move-radius can be introduced.
A Computational Laboratory for Studying Street Protests 11

frequency that depends on their type (as explained in Sect. 3.3) and behavioural state
and perform a scan-plan-behave sequence. Upon activation agents scan the
environment to form their percept PrAi (other agents and scenario features in sight),
plan actions by updating their internal state, and behave by moving and doing an
action that depends on their type and state, such as “taking pictures” of fights in the
case of ‘media’ agents, or starting fights in the case of ‘cops’ or ‘protester’ agents.
The agent attributes common to all agents are their vision radius and angle (which
define the vision cone for detecting nearby agents and scenario features) and move
radius (distance an agent can travel in one time step measured in units of patch
length), which are defined as input variables. The internal state IAi of Ai consists
of the agent’s behavioural state SAi 2 {quiet,active,violent,fighting},
personality vector !Ai (described below), current position x D .x; y/, intended
(planned) position x , and if SAi D fighting the number of time steps to the
end of fight, stored in one of two variables, attacking or defending (to
differentiate fights initiated by ‘cops’ from those initiated by ‘protesters’). ‘Cops’
also have a waypoint variable for storing an eventual patch that the ‘command
cop’ ordered to defend. Figure 2 shows the agents’ architecture used in ProtestLab
and helps understanding the workings of the scan-plan-behave sequence of the
agent cycle. In this figure A is the set of all agents, E is the set of environment
features, SAi is the agent’s behavioural state, !0;Ai .IAi ; PrAi / and !Ai .IAi ; PrAi / are
the default and context-adjusted personality vectors of agent Ai respectively (as
explained below), Ac is the agent’s action and f .x; SAi ; !Ai .IAi ; PrAi // is the agent’s
function (see e.g. Wooldridge 2009 for the terminology).

Fig. 2 Agent architecture diagram

12 C.M. Lemos et al.

3.2.1 Agents’ Movement

Agents move to the centre of the empty patch .Ox; yO / within their move range that
minimizes the penalty function

!Ai .I; PrAi /

VAi .x; y; I; PrAi / D  .SD.x; y/  SD.x0 ; y0 // (3)
k!Ai .I; PrAi /k1

where
0 1
NFk
X X
SD.x; y/ D @ D.x;y/ Fk;j A (4)
Fk 2PrAi jD1

is a double summation, in which the outer sum is taken over the feature sets Fk
and the inner sum involves the distances from entity k (say one fighting ‘cop’)
belonging to a feature set Fk (in this case the agent set of visible fighting ‘cops’)
to point .x; y/, and .x0 ; y0 / is the current position of agent Ai .3 Table 1 shows the
components of the default personality vector !0;Ai for all agent types and subtypes.
Weights have absolute value between 5 (very high) and 0 (neutral, meaning that
the agent is indifferent to entities of the corresponding feature). Positive weights
result in attraction and negative weights in repulsion relative to the corresponding
feature. The method of minimizing a penalty function allows the implementation
of multi-goal driven movement with a single rule, without the complications of
the “combinatorial explosion problem” that would arise with a rule-based IF
<context> THEN <action> formulation (Ilachinsky 2004).
The penalty-minimizing method of inducing goal-driven movement weights
perceived entities with context-dependent motivations or tendencies, but the for-
mulation in Eqs. (3)–(4) is not suitable for targeted movement towards a particular
position or agent. For example, a ‘cop’ Ai may have a motivation to go to a way-
point .xwpi ; ywpi / to defend that position, support other ‘cops’ or engage a protester.
‘Cops’ may also be ordered by a command agent to go to a specific way-point
.xwpc ; ywpc / (e.g. site to be protected or violent protester to be engaged). Denoting
by .Oxwp ; yO wp / the position of the patch centre within the agent’s move range nearest
to the way-point, the next position will be .Ox; yO /

.OxVAi ; yO VAi / C ˛i .Oxwpi ; yO wpi / C ˛c .Oxwpc ; yO wpc /

.Ox; yO / D (5)
1 C ˛i C ˛c

where ˛i is the weight for moving towards an individually selected target and ˛c is
the command weight (or “obedience level”) to go to .xwpc ; ywpc /. Depending on the

P
N
3
The L1 norm kxk1 D jxi j in Eq. (3) was introduced as a normalization factor, so that
iD1
personality vectors of different agents can be inspected and compared if required.
Table 1 Components of the default personality vector for all agent types and subtypes
Fighting Violent Active Quiet Fighting Violent Active Quiet Fighting Quiet
protester protester protester protester cop cop cop cop media media Flag Obstacle Entrance/exit
!0 !1 !2 !3 !4 !5 !6 !7 !8 !9 !10 !11 !12
‘Hardcore’ protester 2 2 1 0 5 4 3 2 0 0 5 1 0
‘Hanger-on’ protester 2 1 0 0 2 1 1 1 2 3 1 1 0
‘Bystander’ protester 5 3 1 0 5 3 1 0 5 1 0 1 1
‘Command’ cop 0 0 0 0 0 0 0 0 0 0 0 0 0
‘Defensive’ cop 0 0 0 0 5 5 5 5 0 0 5 5 0
A Computational Laboratory for Studying Street Protests

‘Offensive’ cop 5 5 1 0 5 5 3 1 0 0 1 1 0
‘Multi-role’ cop 3 3 1 0 3 3 1 1 0 0 3 3 0
‘Media’ 3 1 0 1 3 1 1 0 1 1 1 1 0
A positive sign means attraction and a negative sign means repulsion
13
14 C.M. Lemos et al.

internal state IAi and percept PAi the components of the personality vector of agent
Ai may change by the application of context rules that provide adaptation to the local
context (as in Ilachinsky 2004). These context rules are different for each agent type
and subtype.
Table 2 shows the names of the context rules and the components of the
personality vector that change by their application for all agent types and subtypes
in the model. The ‘hardcore’ cluster rule decreases attraction to flags when less then
ten ‘violent’ or ‘fighting’ protesters are visible by setting !10 D 1; the ‘pursue
cops’ rule increases attraction towards cops when protesters have a superiority of
four to one, by setting !5;6;7 D 5; and the ‘avoid cops’ rule reverses the signs of
!47 when protesters are outnumbered in a proportion equal or greater then two
to one. Notice that the conditions that activate the avoid/approach context rules are
mutually exclusive. The context rules for all other agent types and subtypes follow
the simple principles illustrated for the ‘hardcore’ context rules.

3.2.2 Agents’ State Transitions

Transitions of behavioural state (‘quiet’ , ‘active’ , ‘violent’) are modelled

differently for each type of agent. For ‘protester’ agents, these transitions are
described using a variant of the Epstein et al. (2001) threshold rule (Eq. (1))
which, although introduced in the context of ‘abstract’ ABM, offers a simple
alternative for modelling aggression and fear in politically-motivated protests,
and is consistent with the micro-sociological theories of violence (Collins 2008,
2009): predisposition can be represented by the grievance and risk aversion terms,
situational influence and deterrence by the form of the arrest probability function,
and the “barrier” of tension-fear by the threshold. More specifically, behavioural
state transitions of ‘protester’ agents are modelled using the following rules:

if G  .R  Pa C T/ > 0; turn ‘active’ (6)

if G  2  .R  Pv C T/ > 0; turn ‘violent’ (7)

where Pa D 1  exp.kbC=Acv /, Pv D 1  exp.kb2  C=.V C .1=2/  A/cv /.

In these equations G is the grievance, R is the risk aversion, T is the threshold,
v is the number of visible ‘violent’ protesters, A is the number of visible ‘active’
protesters, C is the number of visible ‘cops’ and Pa and Pv are the estimated arrest
probabilities for ‘active’ or ‘violent’ protesters respectively. The factor two in Eq. (7)
and in the expression of Pv accounts for the increased risk and local superiority
required for protesters to engage ‘cops’. The factor 1=2 that multiplies A in the
denominator of the floor function accounts for the effect that one ‘energizing active’
is less influential than a ‘violent’ protester.
‘Cops’ change their behavioural state depending on ‘approaching’ protesters
(defined as those at a distance smaller than 5 m), ‘closing protesters’ (defined as
those at a distance smaller than 2.5 m), protesters at a distance from the nearest
Table 2 Context-rule names and personality vector components changed by them, for all agent types and subtypes in ProtestLab
Weights Weights Weights Weights
Rule name changed Rule name changed Rule name changed Rule name changed
‘Hardcore’ protester Cluster !10 Pursue cops !57 !10 Avoid cops !47
‘Hanger-on’ protester Keep-clear !07 Cluster !2 Turn ‘hardcore’ !012
‘Bystander’ protester Keep-clear !07 Turn ‘hanger-on’ !012 Turn ‘hardcore’ !012
‘Defensive’ cop On-station !012 Engage trespassing Variable
A Computational Laboratory for Studying Street Protests

‘Offensive’ cop On-station !012 Pursue protesters !2 !6;7;10;11 Retreat !1 !6;7;10;11 Support !012
‘Multi-role’ cop On-station !1011
‘Media’ Keep-distance !07
15
Another random document with
no related content on Scribd:
 THROUGH KIKUYU TO
GALLA-LAND
INTRODUCTION.
My friend, George Henry West, and myself left Cairo in the latter part
of the year 1899, with the intention of proceeding to Uganda viâ
Zanzibar and Mombasa. George was an engineer in the service of
the Irrigation Department of the Egyptian Government, and had
gained a large and varied experience on the new works on the
Barrage below Cairo, then being concluded, and in building, running,
and repairing both locomotives and launches. As a profession I had
followed the sea for three years, leaving it in 1896 in order to join the
British South African Police, then engaged in subduing the native
rebellion in Mashonaland. At the conclusion of hostilities I wandered
over South Africa, and finally found my way to Egypt, where I met
George West. A year later, accompanied by George, I was on my
way southwards again, en route for British East Africa.
When George and I left Cairo, our idea was to go up-country as far
as the Lake Victoria Nyanza, as we considered it extremely probable
that there would be something for us to do in the engineering line,
either in building launches or in the construction of small harbour
works.
We reached Mombasa in due course, and from there proceeded to
Nairobi by the railway then in course of construction to Uganda.
Nairobi is 327 miles from the coast, and is an important centre, being
the headquarters of both the Civil Administration of the Protectorate
and the Uganda railway. On our arrival, George received an offer,
which he accepted, to go up to the lake with a steamer, which was
then on the way out from England in sections, and on his arrival at
the lake with it to rebuild it. I remained in Nairobi.
 In course of time I met the personage referred to in these pages
as “El Hakim,”[1] whom I had known previously by repute. He was
said to be one of the most daring and resolute, and at the same time
one of the most unassuming Englishmen in the Protectorate; a dead
shot, and a charming companion. He had been shooting in
Somaliland and the neighbourhood of Lake Rudolph for the previous
four years, and many were the stories told of his prowess among
elephant and other big game.
It was with sincere pleasure, therefore, that I found I was able to
do him sundry small services, and we soon became fast friends. In
appearance he was nothing out of the common. He was by no
means a big man—rather the reverse, in fact—and it was only on
closer acquaintance that his striking personality impressed one.
He had dark hair and eyes, and an aquiline nose. He was a man
of many and varied attainments. Primarily a member of the medical
profession, his opinions on most other subjects were listened to with
respect. A very precise speaker, he had a clear and impartial manner
of reviewing anything under discussion which never failed to impress
his hearers.
He was a leader one would have willingly followed to the end of
the earth. When, therefore, he proposed that I should accompany
him on an ivory trading expedition to Galla-land, that vast stretch of
country lying between Mount Kenia on the south and Southern
Somaliland on the north, which is nominally under the sphere of
influence of the British East African Protectorate, I jumped at the
chance; and it was so arranged. He had been over much of the
ground we intended covering, and knew the country, so that it
promised to be a most interesting trip.
About this time I heard from George that he was coming down
country, as the steamer parts had not all arrived from England, and
consequently it would probably be months before it would be ready
for building. He had also had a bad attack of malarial fever in the
unhealthy district immediately surrounding the lake at Ugowe Bay,
and altogether he was not very fit. I suggested to El Hakim that
George should join us in our proposed expedition, to which he
readily agreed; so I wrote to George to that effect.
To render the prospect still more inviting, there existed a certain
element of mystery with regard to the river Waso Nyiro (pronounced
Wasso Nēro). It has always been supposed to rise in the Aberdare
Range, but, as I shall show, I have very good reason to believe that it
rises in the western slopes of Kenia Mountain itself. The Waso Nyiro
does not empty itself into the sea, but ends in a swamp called
Lorian, the position of which was supposed to have been fixed by an
exploring party in 1893. But, as I shall also show in the course of this
narrative, the position of Lorian varies.
The upper reaches of the Waso Nyiro were visited by the explorer
Joseph Thompson, F.R.G.S., on his way to Lake Baringo during his
memorable journey through Masai Land in 1885.
In 1887-1888 a Hungarian nobleman, Count Samuel Teleki von
Czeck, accompanied by Lieutenant Ludwig von Hohnel, of the
Imperial Austrian Navy, undertook the stupendous journey which
resulted in the discovery of Lakes Rudolph and Stephanie. Count
Teleki, on his journey north, crossed the Waso Nyiro at a point in
North-West Kenia near its source, while Lieutenant von Hohnel went
two or three days’ march still further down-stream.
A few years later, in 1892-1893, Professor J. W. Gregory, D.Sc., of
the Natural History Museum, South Kensington, made, single-
handed, a remarkable journey to Lake Baringo and Mount Kenia,
and in the teeth of almost insuperable difficulties, ascended the
western face of that mountain and climbed the peak.
At the same time, in the latter part of 1892, an American, Mr.
William Astor Chanler, accompanied by Count Teleki’s companion
and chronicler, Lieutenant von Hohnel, started from a point in
Formosa Bay on the East Coast, and made his way along the course
of Tana River to North-East Kenia, intending later to go on to Lake
Rudolph, and thence northward. He and his companion, deceived by
the reports of the natives, which led them to believe that the Waso
Nyiro emptied itself into an extensive lake, and fired by the idea of
the possible discovery of another great African lake, made their way
down to the Waso Nyiro, and after a fearful march, enduring the
greatest hardships, eventually reached Lorian. To their great
disappointment, it proved to be nothing more than a swamp, and
they turned back without examining it. A few weeks later, Lieutenant
von Hohnel, having been seriously injured by a rhinoceros, was sent
down to the coast, his life being despaired of. Shortly afterwards Mr.
Chanler’s men deserted him in a body, and returned to the coast
also, thus bringing his journey to a premature conclusion; a much-to-
be-regretted ending to a well-planned and well-equipped expedition.
As Mr. Chanler was returning to the coast he met Mr. A. H.
Neumann coming up. Mr. Neumann spent the greater part of 1893 in
shooting elephants in the Loroghi Mountains, after going north to
Lake Rudolph. He also crossed the Waso Nyiro at a point north-east
of Mount Kenia.
During the time Mr. Neumann was shooting in the Loroghi
Mountains he was obliged to make periodical visits to M’thara, in
North-East Kenia, in order to buy food from the natives, and on one
such excursion he met Dr. Kolb, a German scientist, who was
exploring North Kenia.
Dr. Kolb ascended Mount Kenia from the north, and then returned
to Europe. An interesting account of his ascent of the mountain is
published in Dr. Petermann’s “Mitteilungen” (42 Band, 1896). Dr.
Kolb then returned to Kenia in order to continue his observations, but
he was unfortunately killed by a rhinoceros a couple of marches
north of M’thara.
Lorian, therefore, with the exception of Mr. Chanler’s hurried visit,
was practically unexplored. At the commencement of our trip, El
Hakim proposed that, if an opportunity occurred of visiting Lorian, we
should take advantage of it, and endeavour to supplement Mr.
Chanler’s information. As will be seen, an opportunity did present
itself, with what result a perusal of this account of our expedition will
disclose.
 FOOTNOTES:
[1] Anglice, “The Doctor.”
 CHAPTER I.
PREPARATIONS AND START.

Engaging porters—Characteristics of Swahili, Wa’Nyamwezi, and

Wa’Kamba porters—Selecting trade goods—Provisions—Arms
and ammunition—The Munipara—Sketch of some principal
porters—Personal servants—List of trade goods taken—
Distributing the loads—Refusal of the Government to register our
porters—Reported hostility of the natives—Finley and Gibbons’
disaster—Start of the Somali safaris—We move to Kriger’s Farm
—I fall into a game-pit—Camp near Kriger’s Farm—Visitors—The
start.
One of the most important items in the organization of a “safari”
(caravan) is the judicious selection of the men. Choosing ours was a
task that gave us much trouble and vexation of spirit. El Hakim said
that for all-round usefulness the Wa’kamba were hard to beat, and
thought that we had better form the bulk of the safari from them, and
stiffen it with a backbone of Swahilis and Wa’Nyamwezi, as, though
the Wa’kamba were very good men when well handled, in the
unlikely event of hostilities with the natives it would be advisable to
strengthen them with an addition from the lustier tribes. To that end
we proposed to engage a dozen Swahili and half that number of
Wa’Nyamwezi. Porters at that time were very scarce; but having
secured one or two good men as a nucleus, we sent them into the
bazaar at Nairobi to bring us any other men they could find who
wanted employment.
The Swahilis are natives of Zanzibar and the adjacent coasts.
They are of mixed—very mixed—descent, being mainly the offspring
of various native slaves and their Arab masters. They were originally
a race of slaves, but since the abolition of slavery they have become
more and more independent, and they now consider themselves a
very superior race indeed. They call themselves “Wangwana”
(freemen), and allude to all other natives as “Washenzi” (savages).
They are incorrigibly conceited, and at times very vicious, lazy,
disobedient, and insolent. But once you have, by a judicious display
of firmness, gained their respect, they, with of course some
exceptions, prove to be a hardy, cheerful, and intelligent people,
capable of enduring great hardships without a too ostentatious
display of ill-feeling, and will even go so far as to make bad puns in
the vernacular upon their empty stomachs, the latter an occurrence
not at all infrequent in safari work away from the main roads.
The Wa’kamba, on the whole, are a very cheerful tribe, and though
of small physique, possess wonderful powers of endurance, the
women equally with the men. We calculated that some of our men, in
addition to their 60-lb. load, carried another 30 lbs. weight in
personal effects, rifle, and ammunition; so that altogether they
carried 90 lbs. dead weight during one or sometimes two marches a
day for weeks at a stretch, often on insufficient food, and sometimes
on no food at all.
The Wa’Nyamwezi are, in my opinion, really more reliable than
either the Swahili or Wa’kamba. They come from U’Nyamwezi, the
country south and east of Lake Victoria Nyanza. We had six of them
with us, and we always found them steady and willing, good porters,
and less trouble than any other men in the safari. They were very
clannish, keeping very much to themselves, but were quiet and
orderly, and seldom complained; and if at any time they imagined
they had some cause for complaint, they formed a deputation and
quietly stated their case, and on receiving a reply as quietly returned
to their fire—very different from the noisy, argumentative Swahili.
They appear to me to possess the virtues of both the Swahilis and
Wa’kamba without their vices. The Wa’kamba’s great weakness
when on the march was a penchant for stealing from the native
villages whatever they could lay their hands on, being encouraged
thereto by the brave and noble Swahilis, who, while not wishing to
risk our displeasure by openly doing likewise, urged on the simple
Wa’kamba, afterwards appropriating the lion’s share of the spoil: that
is, if we did not hear of the occurrence and confiscate the spoil
ourselves.
 We had pitched our tent just outside the town of Nairobi, and
proceeded to get together our loads of camp equipment, trade
goods, and provisions: no easy task on an expedition such as ours,
where the number of carriers was to be strictly limited.
In the first place, we required cloth, brass wire, iron wire, and
various beads, in sufficient quantities to buy food for the safari for at
least six months. Provisions were also a troublesome item, as,
although we expected to live a great deal upon native food, we
required such things as tea, coffee, sugar, jam, condiments, and also
medicines. The question was not what to take, but what not to take.
However, after a great amount of discussion, lasting over several
days, we settled the food question more or less satisfactorily.
During this time our recruiting officers were bringing into camp
numbers of men who, they said, wanted to take service with us as
porters. Judging from the specimens submitted for our approval, they
seemed to have raked out the halt, the lame, and the blind. After
much trouble we selected those whom we thought likely to be
suitable, and gave them an advance of a few rupees as a retaining
fee, with which, after the manner of their kind, they immediately
repaired to the bazaar for a last long orgie.
There was also the important question of arms and ammunition to
be considered, as, although we did not expect any fighting, it would
have been foolish in the extreme to have entered such districts as
we intended visiting without adequate means of self-defence. We
concluded the twenty-five Snider rifles used by El Hakim on a
previous trip would suffice. Unfortunately, we could get very little
ammunition for them, as at that time Snider ammunition was very
scarce in Nairobi, one reason being that it had been bought very
largely by a big Somali caravan under Jamah Mahomet and Ismail
Robli, which set out just before us, bound for the same districts.
We, however, eventually procured five or six hundred rounds: a
ridiculously inadequate amount considering the distance we were to
travel and the time we expected to be away.
With regard to our armament, El Hakim possessed by far the best
battery. His weapons consisted of an 8-bore Paradox, a ·577
Express, and a single-barrelled ·450 Express, all by Holland and
Holland. The 8-bore we never used, as the ·577 Express did all that
was required perfectly satisfactorily. The 8-bore would have been a
magnificent weapon for camp defence when loaded with slugs, but
fortunately our camp was never directly attacked, and consequently
the necessity for using it never arose. The ·557 was the best all-
round weapon for big game such as elephant, rhinoceros, and
buffalo, and never failed to do its work cleanly and perfectly. Its only
disadvantage was that it burnt black powder, and consequently I
should be inclined, if I ever made another expedition, to give the
preference to one of the new ·450 or ·500 Expresses burning
smokeless powder, though, as I have not handled one of the latter, I
cannot speak with certainty. El Hakim’s ·450 Express was really a
wonderful weapon, though open to the same objection as the ·557—
that of burning black powder. It was certainly one of the best all-
round weapons I ever saw for bringing down soft-skinned game. It
was a single-barrelled, top-lever, hammer-gun, with flat top rib. The
sights were set very low down on the rib, to my mind a great
advantage, as it seems to me to minimize the chances of accidental
canting. Its penetrative power, with hardened lead bullets, was
surprising. I have seen it drop a rhinoceros with a bullet through the
brain, and yet the same projectile would kill small antelope like
Grant’s or Waller’s gazelles without mangling them or going right
through and tearing a great hole in its egress, thereby spoiling the
skin, which is the great cause of complaint against the ·303 when
expanding bullets are used.
I myself carried a ·303 built by Rigby, a really magnificent weapon.
I took with me a quantity of every make of ·303 expanding bullets,
from copper-tubed to Jeffry’s splits. After repeated trials I found that
the Dum-Dum gave the most satisfactory results, “since when I have
used no other.”
I also carried a supply of ·303 solid bullets, both for elephants and
for possible defensive operations. For rhinoceros, buffalo, or giraffe, I
carried an ordinary Martini-Henry military rifle, which answered the
purpose admirably. A 20-bore shot-gun, which proved useful in
securing guinea-fowl, etc., for the pot, completed my battery. George
carried a ·303 military rifle and a Martini-Henry carbine.
It was essential that we should have a good “Munipara” (head-
man), and the individual we engaged to fill that important position
was highly recommended to us as a man of energy and resource.
His name was Jumbi ben Aloukeri. Jumbi was of medium height,
with an honest, good-natured face. He possessed an unlimited
capacity for work, but we discovered, too late, that he possessed no
real control over the men, which fact afterwards caused us endless
trouble and annoyance. He was too easy with them, and made the
great mistake—for a head-man—of himself doing anything we
wanted, instead of compelling his subordinates to do it, with the
result that he was often openly defied, necessitating vigorous
intervention on our part to uphold his authority. We usually alluded to
him as “the Nobleman,” that being the literal translation of his name.
Next on the list of our Swahili porters was Sadi ben Heri, who had
been up to North Kenia before with the late Dr. Kolb, who was killed
by a rhinoceros a couple of marches north of M’thara, Sadi was a
short, stoutly built, pugnacious little man, with a great deal to say
upon most things, especially those which did not concern him. He
was a good worker, but never seemed happy unless he was
grumbling; and as he had a certain amount of influence among the
men, they would grumble with him, to their great mutual satisfaction
but ultimate disadvantage. His pugnacious disposition and lax
morals soon got him into trouble, and he, together with some of his
especial cronies, was killed by natives, as will be related in its proper
sequence.
Hamisi ben Abdullah was a man of no marked peculiarities, except
a disposition to back up Sadi in any mischief. The same description
applies to Abdullah ben Asmani and Asmani ben Selim.
Coja ben Sowah was a short, thick-set man, so short as to be
almost a dwarf. He was one of the most cheery and willing of our
men, so much so that it was quite a pleasure to order him to do
anything—a pleasure, I fear, we appreciated more than he did. On
receiving an order he would run to execute it with a cheery “Ay
wallah, bwana” (“Please God, master”), that did one good to hear.
Resarse ben Shokar was our “Kiongozi,” i.e. the leading porter,
who sets the step on the march and carries the flag of the safari. He,
also, always ran on receiving an order—ran out of sight, in fact; then,
when beyond our ken, compelled a weaker man than himself to do
what was wanted. I could never cure him of the habit of sleeping on
sentry duty, though many a time I have chased him with a stirrup-
strap, or a camp-stool, or anything handy when, while making
surprise inspections of the sentries, I had found him fast asleep. He
was valuable, however, in that he was the wit of the safari. He was a
perfect gas-bag, and often during and after a long and probably
waterless march we blessed him for causing the men to laugh by
some harmless waggish remark at our expense.
Sulieman was a big, hulking, sulky brute, who gave us a great deal
of trouble, and finally deserted near Lorian, forgetting to return his
rifle, and also absent-mindedly cutting open my bag and abstracting
a few small but necessary articles. Docere ben Ali, his chum, was
also of a slow and sullen disposition, though he was careful not to
exhibit it to us. When anything disturbed him he went forthwith and
took it out of the unfortunate Wa’kamba.
Of the Wa’kamba I do not remember the names except of two or
three who particularly impressed themselves on my memory. The
head M’kamba was known as Malwa. He was a cheerful, stupid idiot
who worked like a horse, though he never seemed to get any
“for’arder.” Another M’kamba, named Macow, afterwards succeeded
him in the headmanship of the Wa’kamba when Malwa was deposed
for some offence. We nicknamed Macow “Sherlock Holmes,” as he
seemed to spend most of his leisure hours prowling round the camp,
peering round corners with the true melodrama-detective-Hawkshaw
expression in his deep-set, thickly browed eyes. He would often
creep silently and mysteriously to our tent, and in a subdued whisper
communicate some trifling incident which had occurred on the
march; then, without waiting for a reply, steal as silently and
mysteriously away.
 I must not conclude this chapter without some mention of our
personal servants. First and foremost was Ramathani, our head
cook and factotum. Ramathani had already been some three months
in my service as cook and personal servant, and a most capable
man I had found him. My acquaintance with him began one morning
when I had sent my cook, before breakfast, to the sokoni (native
bazaar) to buy bread, vegetables, etc. As he did not return I went
outside to the cook-house in some anxiety as to whether I should get
any breakfast. Several native servants were there, and they informed
me my cook was still in the bazaar, very drunk, and most likely would
not be back till noon. Of course, I was angry, and proceeded to show
it, when a soothing voice, speaking in very fair English, fell upon my
ear. Turning sharply, I was confronted by a stranger, a good-looking
native, neatly dressed in khaki.
“Shall I cook breakfast for master?” he inquired softly.
“Are you able?” said I.
“Yes, master.”
“Then do so,” I said; and went back to my quarters and waited with
as much patience as I could command under the circumstances.
In a quarter of an hour or so Ramathani—for it was indeed he—
brought in a temptingly well-cooked breakfast, such as I was almost
a stranger to, and at the same time hinted that he had permanently
attached me as his employer. My own cook turned up an hour or so
later, very drunk and very abusive, and he was incontinently fired
out, Ramathani being established in his stead.
Ramathani had two boys as assistants, Juma and Bilali. Juma was
an M’kamba. His upper teeth were filed to sharp points, forming most
useful weapons of offence, as we afterwards had occasion to notice.
Bilali was an M’Kikuyu, and a very willing boy. He was always very
nervous when in our presence, and used to tremble excessively
when laying the table for meals. When gently reproved for putting
dirty knives or cups on the table, he would grow quite ludicrous in his
hurried efforts to clean the articles mentioned, and would spit on
them and rub them with the hem of his dirty robe with a pathetic
eagerness to please that disarmed indignation and turned away
wrath.
Having finally secured our men, it only remained to pack up and
distribute the loads of equipment, provisions, trade goods, etc. We
did not take such a large quantity of trade goods as we should have
done in the ordinary course, as El Hakim already had a large
quantity in charge of a chief in North Kenia. The following is a list,
compiled from memory, of what we took with us:—
Unguo (Cloth).
2 loads Merikani (American sheeting).
2 ” kisuto (red and blue check cloths).
2 ” blanketi (blankets, coloured).
1 load various, including—
gumti (a coarse white cloth).
laissoes (coloured cloths worn by women).
kekois (coloured cloth worn by men).

Uzi Wa Madini (Wire).

seninge (iron wire, No. 6).
2 or 3 loads of masango (copper wire, No. 6).
masango n’eupe (brass wire, No. 6).

Ushanga (Beads).
sem Sem (small red Masai beads).
2 or 3 loads of sembaj (white Masai beads).
ukuta (large white opaque beads).
2 loads of mixed Venetian beads.

When all the loads were packed, they were placed in a line on the
ground; and falling the men in, we told off each to the load we
thought best suited to him. To the Swahilis, being good marching
men and not apt to straggle on the road, we apportioned our
personal equipment, tents, blankets, and table utensils. To the
Wa’Nyamwezi we entrusted the ammunition and provisions, and to
the Wa’kamba we gave the loads of wire, beads, cloth, etc. Having
settled this to our own satisfaction, we considered the matter settled,
and ordered each man to take up his load.
Then the trouble began. First one man would come to us and ask
if his load might be changed for “that other one,” while the man to
whom “that other one” had been given would object with much
excited gesticulation and forcible language to any alteration being
made, and would come to us to decide the case. We would then
arbitrate, though nine times out of ten they did not abide by our
decision. Other men’s loads were bulky, or awkward, or heavy, or
had something or other the matter with them which they wanted
rectified, so that in a short time we had forty men with forty
grievances clamouring for adjustment. We simplified matters by
referring every one to Jumbi, and having beaten an inglorious retreat
to our tents, solaced ourselves with something eatable till everything
was more or less amicably settled.
Nothing is more characteristic of the difference in the races than
the way in which they carry their loads. The Swahilis and
Wa’Nyamwezi, being used to the open main roads, carry their loads
boldly on their heads, or, in some cases, on their shoulders. The
Wa’kamba, on the other hand, in the narrow jungle paths of their
own district find it impossible, by reason of the overhanging
vegetation, to carry a load that way. They tie it up instead with a long
broad strip of hide, leaving a large loop, which is passed round the
forehead from behind, thus supporting the load, which rests in the
small of the back. When the strain on the neck becomes tiring they
lean forward, which affords considerable relief, by allowing the load
to rest still more upon the back. There were also six donkeys, the
property of El Hakim, and these were loaded up as well. A donkey
will carry 120 lbs., a weight equal to two men’s loads.
Finally, we had to register our porters at the Sub-Commissioner’s
office, as no safaris are allowed to proceed until that important
ceremony has been concluded, and the Government has pouched
the attendant fees. In our case, however, there appeared to be a
certain amount of difficulty. On delivering my application I was told to
wait for an answer, which I should receive in the course of the day. I
waited. In the afternoon a most important-looking official document
was brought to me by a Nubian orderly. In fear and trembling I
opened the envelope, and breathed a heartfelt sigh of relief when I
found that the Government had refused to register our porters, giving
as their reason that the districts we intended visiting were unsettled
and, in their opinion, unsafe, and therefore we should proceed only
at our own risk. We did not mind that, and we saved the registration
fee anyhow. The Government had already refused to register the
Somali’s porters, and they intimated, very rightly, that they could not
make any difference in our case.
Jamah Mahomet, who was in command of the Somali safari,
started off that day. He had with him Ismail Robli as second in
command. A smaller safari, under Noor Adam, had started a week
previously. Both these safaris intended visiting the same districts as
ourselves. We were fated to hear a great deal more of them before
the end of our trip.
In the evening I received a private note from one of the
Government officers, informing me that we were likely to have a
certain amount of trouble in getting across the river Thika-Thika
without fighting, as the natives of that district were very turbulent,
and advising us to go another way. My informant cited the case of
Messrs. Finlay and Gibbons by way of a cheerful moral.
Finlay and Gibbons were two Englishmen who had been trading
somewhere to the north of the Tana River. They had forty men or so,
and were trading for ivory with the A’kikuyu, when they were
suddenly and treacherously attacked and driven into their “boma”
(thorn stockade), and there besieged by quite six thousand natives.
From what I saw later, I can quite believe that their numbers were by
no means exaggerated. During a night attack, Finlay was speared
through the hand and again in the back, the wound in the back,
however, not proving dangerous. They managed to get a message
through to Nairobi, and some Nubian troops were sent to their relief,
which task they successfully accomplished, though only with the
greatest difficulty. It was not till six weeks after he received the
wound that Finlay was able to obtain medical assistance, and by that
time the tendons of his hand had united wrongly, so that it was
rendered permanently useless. This was a nice enlivening story,
calculated to encourage men who were setting out for the same
districts.
The following day I received a telegram from George to say that
he had arrived from Uganda at the Kedong Camp, at the foot of the
Kikuyu Escarpment, so I went up by rail to meet him. He looked very
thin and worn after his severe attack of fever. We returned to Nairobi
the same evening, and proceeded to our camp. El Hakim, who was
away when we arrived, turned up an hour later, and completed our
party. He had been to Kriger’s Farm about seven miles out. Messrs.
Kriger and Knapp were two American missionaries who had
established a mission station that distance out of Nairobi, towards
Doenyo Sabuk, or Chianjaw, as it is called by the Wa’kamba.
El Hakim, being anxious to get our men away from the pernicious
influence of the native bazaar, arranged that he would go on to
Kriger’s early on the following morning, and that George and I should
follow later in the day with the safari, and camp for the night near
Kriger’s place. Accordingly he started early in the forenoon on the
following day.
George and I proceeded to finish the packing and make final
arrangements—a much longer task than we anticipated. There were
so many things that must be done, which we found only at the last
minute, that at 3 p.m., as there was no prospect of getting away until
an hour or so later, I sent George on with the six loaded donkeys,
about thirty of El Hakim’s cattle, and a dozen men, telling him that I
would follow. George rode a mule (of which we had two), which El
Hakim had bought in Abyssinia two years before. They were
splendid animals, and, beyond an inconvenient habit, of which we
never cured them, of shying occasionally and then bolting, they had
no bad points. They generally managed to pick up a living and get fat
in a country where a horse would starve, and, taking them
altogether, they answered admirably in every way. I would not have
exchanged them for half a dozen of the best horses in the
Protectorate. One mule was larger than the other, and lighter in
colour, and was consequently known as n’yumbu m’kubwa, i.e. “the
big mule.” It was used by George and myself as occasion required.
The other, a smaller, darker animal, was known as n’yumbu m’dogo,
i.e. “the little mule.” It was ridden exclusively by El Hakim.
After George’s departure I hurried the remaining men as much as
possible, but it was already dusk when I finally started on my seven-
mile tramp. Some of the men had to be hunted out of the bazaar,
where they had lingered, with their loved ones, in a last long farewell.
There is no twilight in those latitudes (within two degrees of the
equator), so that very soon after our start we were tramping along in
the black darkness. I had no knowledge of the road; only a rough
idea of the general direction. I steered by the aid of a pocket-
compass and a box of matches. After the first hour I noticed that the
men commenced to stagger and lag behind with their lately
unaccustomed burdens, and I had to be continually on the alert to
prevent desertions. I numbered them at intervals, to make sure that
none of them had given me the slip, but an hour and a half after
starting I missed three men with their loads, in spite of all my
precautions. I shouted back into the darkness, and the men
accompanying me did the same, and, after a slight interval we were
relieved to hear an answering shout from the missing men. After
waiting a few moments, we shouted again, and were amazed to find
that the answering shout was much fainter than before. We
continued shouting, but the answers grew gradually fainter and more
faint till they died away altogether. I could not understand it at first,
but the solution gradually dawned upon me. We were on a large
plain, and a few hundred yards to the left of us was a huge belt of
forest, which echoed our shouts to such an extent that the men who
were looking for us were deceived as to our real position, and in their
search were following a path at right angles to our own. I could not
light a fire to guide them, as the grass was very long and dry, and I
should probably have started a bush fire, the consequences of which
would have been terrible. I therefore fired a gun, and was answered
by another shot, seemingly far away over the plain to the right.
Telling the men to sit down and rest themselves on the path, I
ordered Jumbi to follow me, and, after carefully taking my bearings
by compass, started to walk quickly across the plain to intercept
them.
It was by no means a pleasant experience, trotting across those
plains in the pitchy blackness, with the grass up to my waist, and
huge boulders scattered about ready and willing to trip me up. I got
very heated and quite unreasonably angry, and expressed my
feelings to Jumbi very freely. I was in the midst of a violent diatribe
against all natives generally, and Swahili porters in particular, which I
must admit he bore with commendable patience, when the earth
gave way beneath me, and I was precipitated down some apparently
frightful abyss, landing in a heap at the bottom, with all the breath
knocked out of my body. I laid there for a little while, and
endeavoured to collect my scattered faculties. Soon I stood up, and
struck a match, and discovered that I had fallen into an old game-pit,
about 8 feet deep. It was shaped like a cone, with a small opening at
the top, similar to the old-fashioned oubliette. I looked at the floor,
and shuddered when I realized what a narrow squeak I might have
had; for on the centre of the floor were the mouldering remains of a
pointed stake, which had been originally fixed upright in the earth
floor on the place where I had fallen.
“Is Bwana (master) hurt?” said the voice of Jumbi from somewhere
in the black darkness above.
I replied that I was not hurt, but that I could not get out without
assistance; whereupon Jumbi lowered his rifle, and, to the
accompaniment of a vast amount of scrambling and kicking, hauled
me bodily out.
We were by this time very near to the men for whom we were
searching, as we could hear their voices raised in argument about
the path. We stopped and called to them, and presently they joined
us, and we all set off together to join my main party. We reached it
without further mishap, and resumed our interrupted march.
It was very dark indeed. I could not see my hand when I held it a
couple of feet from my face. One of the men happening to remark
that he had been over the path some years before, I immediately
placed him in the van as guide, threatening him with all sorts of pains
and penalties if he did not land us at our destination some time
before midnight.
I was particularly anxious to rejoin George, as I had the tents,
blankets, and food, and he would have a very uninteresting time
without me. We marched, therefore, with renewed vigour, as our
impromptu guide stated that he thought one more hour’s march
would do the business. It didn’t, though. For two solid hours we
groped blindly through belts of forest, across open spaces, and up
and down wooded ravines, until somewhere about eleven p.m.,
when we reached a very large and terribly steep ravine, thickly
clothed with trees, creepers, and dense undergrowth. We could hear
the rushing noise of a considerable volume of water at the bottom,
and in the darkness it sounded very, very far down.
I halted at the top to consider whether to go on or not, but the
thought of George waiting patiently for my appearance with supper
and blankets made me so uncomfortable that I decided to push on if
it took me all night. We thereupon commenced the difficult descent,
but halfway down my doubts as to the advisability of the proceeding
were completely set at rest by one of the men falling down in some
kind of a fit from over-fatigue. The others were little better, so I
reluctantly decided to wait for daylight before proceeding further. I
tried to find something to eat among the multifarious loads, and
fortunately discovered a piece of dry bread that had been thrown in
with the cooking utensils at the last moment. I greedily devoured it,
and, wrapping myself in my blankets, endeavoured to sleep as well
as I was able on a slope of forty-five degrees. A thought concerning
George struck me just before I dropped off to sleep, which comforted
me greatly. “George knows enough to go in when it rains,” I thought.
“He will leave the men with the cattle, and go over to Kriger’s place
and have a hot supper and a soft bed, and all kinds of good things
like that,” and I drew my blankets more closely round me and
shivered, and felt quite annoyed with him when I thought of it.
At daylight we were up and off again, and, descending the ravine,
crossed the river at the bottom, and continued the march. On the
way I shot a guinea-fowl, called by the Swahilis “kanga,” and after an
hour and a half of quick walking I came up with George.
 He had passed a miserable night, without food, blankets, or fire,
and, to make matters worse, it had drizzled all night, while he sat on
a stone and kept watch and ward over the cattle. The men who had
accompanied him were so tired that they had refused to build a
boma to keep the cattle in. He seemed very glad to see me. We at
once got the tent put up, a fire made, and the boma built, and soon
made things much more comfortable. In fact, we got quite gay and
festive on the bread and marmalade, washed down with tea, which
formed our breakfast.
El Hakim was at Kriger’s place, about a mile distant. We had to
wait two or three days till he was ready to start, as he had a lot of
private business to transact. We left all the cattle except nine behind,
under Kriger’s charge; we sent the nine back subsequently, as we
found they were more trouble than they were worth.
In the evening I went out to shoot guinea-fowl; at least, I intended
to shoot guinea-fowl, but unfortunately I saw none. I lost myself in
the darkness, and could not find my way back to camp. After
wandering about for some time, I at last spied the flare of the camp
fires, halfway up a slope a mile away, opposite to that on which I
stood. I made towards them, entirely forgetting the small river that
flowed at the foot of the slope. It was most unpleasantly recalled to
my memory as I suddenly stepped off the bank and plunged, with a
splash, waist deep into the icy water. Ugh!
I scrambled up the opposite bank, and reached the camp safely,
though feeling very sorry for myself. El Hakim and George thought it
a good joke. I thought they had a very low sense of humour.
On the following morning George and I sallied forth on sport intent.
George carried the shot-gun, and I the ·303. We saw no birds; but
after an arduous stalk, creeping on all fours through long, wet grass,
I secured a congoni. Congoni is the local name for the hartebeeste
(Bubalis Cokei). The meat was excellent, and much appreciated. El
Hakim joined us in the afternoon, accompanied by Mr. Kriger and Mr.
and Mrs. Knapp, who wished to inspect our camp. We did the
honours with the greatest zest, knowing it would be the last time for
many months that we should see any of our own race.