100% found this document useful (1 vote)
23 views

Modelling Simulation and Control of Non linear Dynamical Systems An Intelligent Approach Using Soft Computing and Fractal Theory Numerical Insights Patricia Melin All Chapters Instant Download

Theory

Uploaded by

mossenrizqy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
23 views

Modelling Simulation and Control of Non linear Dynamical Systems An Intelligent Approach Using Soft Computing and Fractal Theory Numerical Insights Patricia Melin All Chapters Instant Download

Theory

Uploaded by

mossenrizqy
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 61

Download Full ebookname - Read Now at ebookname.

com

Modelling Simulation and Control of Non linear


Dynamical Systems An Intelligent Approach Using
Soft Computing and Fractal Theory Numerical
Insights Patricia Melin
https://ebookname.com/product/modelling-simulation-and-
control-of-non-linear-dynamical-systems-an-intelligent-
approach-using-soft-computing-and-fractal-theory-numerical-
insights-patricia-melin/

OR CLICK BUTTON

DOWLOAD EBOOK

Discover More Ebook - Explore Now at ebookname.com


Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Dynamical systems and control 1st Edition Firdaus E.


Udwadia

https://ebookname.com/product/dynamical-systems-and-control-1st-
edition-firdaus-e-udwadia/

ebookname.com

Modelling Simulation and Control of the Dyeing Process 1st


Edition R. Shamey

https://ebookname.com/product/modelling-simulation-and-control-of-the-
dyeing-process-1st-edition-r-shamey/

ebookname.com

Modelling Simulation and Control of Two Wheeled Vehicles


1st Edition Tanelli Mara

https://ebookname.com/product/modelling-simulation-and-control-of-two-
wheeled-vehicles-1st-edition-tanelli-mara/

ebookname.com

The College Blue Book 36th edition 2009 Volume 2 Tabular


Data Macmillan Reference Usa Gale

https://ebookname.com/product/the-college-blue-book-36th-
edition-2009-volume-2-tabular-data-macmillan-reference-usa-gale/

ebookname.com
The Topography of Thebes from the Bronze Age to Modern
Times Sarantis Symeonoglou

https://ebookname.com/product/the-topography-of-thebes-from-the-
bronze-age-to-modern-times-sarantis-symeonoglou/

ebookname.com

The Handbook of Juvenile Delinquency and Juvenile Justice


1st Edition Marvin D. Krohn

https://ebookname.com/product/the-handbook-of-juvenile-delinquency-
and-juvenile-justice-1st-edition-marvin-d-krohn/

ebookname.com

Yen Jaffe s Reproductive Endocrinology Expert Consult 6th


Edition Jerome F. Strauss Iii Md Phd

https://ebookname.com/product/yen-jaffe-s-reproductive-endocrinology-
expert-consult-6th-edition-jerome-f-strauss-iii-md-phd/

ebookname.com

I Am Not Myself These Days A Memoir 1st Edition Josh


Kilmer-Purcell

https://ebookname.com/product/i-am-not-myself-these-days-a-memoir-1st-
edition-josh-kilmer-purcell/

ebookname.com

Communication Language and Literacy from Birth to Five Mrs


Avril Brock

https://ebookname.com/product/communication-language-and-literacy-
from-birth-to-five-mrs-avril-brock/

ebookname.com
When Your Teen Has an Eating Disorder Practical Strategies
to Help Your Teen Recover from Anorexia Bulimia and Binge
Eating 1st Edition Lauren Muhlheim
https://ebookname.com/product/when-your-teen-has-an-eating-disorder-
practical-strategies-to-help-your-teen-recover-from-anorexia-bulimia-
and-binge-eating-1st-edition-lauren-muhlheim/
ebookname.com
Modelling, Simulation and Control
of Non-Linear Dynamical Systems

© 2002 Taylor & Francis


Numerical Insights
Series Editor
A. Sydow, GMD-FIRST, Berlin, Germany
Editorial Board
P. Borne, ~ c o l ede Lille, France G. Carmichael, University of Iowa, USA
L. Dekker, Delft University of Technology, The Netherlands A. Iserles, University
of Cambridge, UK A. Jakeman, Australian National University, Australia
G. Korn, Industrial Consultants (Tucson), USA G.P. Rao, Indian Institute of
Technology, India R. Rice, Purdue University, USA A.A. Samarskii, Russian
Academy of Science, Russia Y. Takahara, Tokyo Institute of Technology, Japan

The Numerical Insights series aims to show how numerical simulations provide valuable
insights into the mechanisms and processes involved in a wide range of disciplines. Such
simulations provide a way of assessing theories by comparing simulations with observa-
tions. These models are also powerful tools which serve to indicate where both theory and
experiment can be improved.
In most cases the books will be accompanied by software on disk demonstrating working
examples of the simulations described in the text.
The editors will welcome proposals using modelling, simulation and systems analysis
techniques in the following disciplines: physical sciences; engineering; environment; ecol-
ogy; biosciences; economics.

Volume 1
Numerical Insights into Dynamic Systems: Interactive Dynamic System Simulation with
Microsofto, Windows 95TMand NTTM
Granino A. Korn

Volume 2
Modelling, Simulation and Control of Non-Linear Dynamical Systems: An Intelligent
Approach using Soft Computing and Fractal Theory
Patricia Melin and Oscar Castillo

This book is part of a series. The publisher will accept continuation orders which may be cancelled
at any time and which provide for automatic billing and shipping of each title in the series upon
publication. Please write for details.

© 2002 Taylor & Francis


Modelling, Simulation and Control
of Non-Linear Dynamical Systems

An Intelligent Approach Using Soft Computing


and Fractal Theory

Patricia Melin and Oscar Castillo


Tijuana Institute of Technology, Tijuana, Mexico

Taylor & Francis


Taylor&Francis Group

Boca Raton London NewYork Singapore

A CRC title, part of the Taylor & Francis imprint, a member of the
Taylor & Francis Croup, the academic division of T&F lnforrna plc.

© 2002 Taylor & Francis


First published 2002 by Taylor & Francis
11 New Fetter Lane, London EC4P 4EE
Simultaneously published in the USA and Canada
by Taylor & Francis Inc,
29 West 35th Street, New York, NY 10001
Taylor & Francis is an inzprint of the Taylor & Francis Group
O 2002 Taylor & Francis
This book has been produced from camera-ready copy supplied by the authors
Printed and bound in Great Britain by
TJ International Ltd, Padstow, Cornwall
All rights reserved. No part of this book may be reprinted or reproduced or utilised in
any form or by any electronic, mechanical, or other means, now known or hereafter
invented, including photocopying and recording, or in any information storage or
retrieval system, without permission in writing from the publishers.
Every effort has been made to ensure that the advice and information in this book is true
and accurate at the time of going to press. However, neither the publisher nor the authors
can accept any legal responsibility or liability for any errors or omissions that may be
made. In the case of drug administration, any medical procedure or the use of technical
equipment mentioned within this book, you are strongly advised to consult the
manufacturer's guidelines.

British Library Cataloguing in Publication Data


A catalogue record for this book is available from the British Library
Library of Congress Cataloging in Publication Data
A catalog record for this book has been requested

ISBN 0-415-27236-X

© 2002 Taylor & Francis


CONTENTS

PREFACE ix

1 INTRODUCTION TO MODELLING, SIMULATION AND CONTROL


OF NON-LINEAR DYNAMICAL SYSTEMS 1
1.1 Modelling and Simulation of Non-Linear Dynamical Systems 2
1.2 Control of Non-Linear Dynamical Systems 5

2 FUZZY LOGIC FOR MODELLING


2.1 Fuzzy Set Theory
2.2 Fuzzy Reasoning
2.3 Fuzzy Inference Systems
2.4 Fuzzy Modelling
2.5 Summary

3 NEURAL NETWORKS FOR CONTROL


3.1 Backpropagation for Feedforward Networks
3.1.1 The backpropagation learning algorithm
3.1.2 Backpropagation multilayer perceptrons
3.2 Adaptive Neuro-Fuzzy Inference Systems
3.2.1 ANFIS architecture
3.2.2 Learning algorithm
3.3 Neuro-Fuzzy Control
3.3.1 Inverse learning
3.3.2 Specialized learning
3.4 Adaptive Model-Based Neuro-Control
3.4.1 Indirect neuro-control
3.4.2 Direct neuro-control
3.4.3 Parameterized neuro-control
3.5 Summary

4 GENETIC ALGORITHMS AND FRACTAL THEORY FOR


MODELLING AND SIMULATION 65
4.1 Genetic Algorithms 67
4.2 Simulated Annealing 72
4.3 Basic Concepts of Fractal Theory 75
4.4 Summary 80

© 2002 Taylor & Francis


vi CONTENTS

5 FUZZY-FRACTAL APPROACH FOR AUTOMATED


MATHEMATICAL MODELLING
5.1 The Problem of Automated Mathematical Modelling
5.2 A Fuzzy-Fractal Method for Automated Modelling
5.3 Implementation of the Method for Automated Modelling
5.3.1 Description of the time series analysis module
5.3.2 Description of the expert selection module
5.3.3 Description of the best model selection module
5.4 Comparison with Related Work
5.5 Summary

6 FUZZY-GENETIC APPROACH FOR AUTOMATED SIMULATION


6.1 The Problem of Automated Simulation
6.1.1 Numerical simulation of dynamical systems
6.1.2 Behavior identification for dynamical systems
6.1.3 Automated simulation of dynamical systems
6.2 Method for Automated Parameter Selection using Genetic Algorithms
6.3 Method for Dynamic Behavior Identification using Fuzzy Logic
6.3.1 Behavior identification based on the analytical properties of
the model
6.3.2 Behavior identification based on the fractal dimension and the
Lyapunov exponents
6.4 Summary

7 NEURO-FUZZY APPROACH FOR ADAPTIVE MODEL-BASED


CONTROL
7.1 Modelling the Process of the Plant
7.2 Neural Networks for Control
7.3 Fuzzy Logic for Model Selection
7.4 Neuro-Fuzzy Adaptive Model-Based Control
7.5 Summary

8. ADVANCED APPLICATIONS OF AUTOMATED MATHEMATICAL


MODELLING AND SIMULATION
8.1 Modelling and Simulation of Robotic Dynamic Systems
8.1.1 Mathematical modelling of robotic systems
8.1.2 Automated mathematical modelling of robotic dynamic systems
8.1.3 Automated simulation of robotic dynamic systems
8.2 Modelling and Simulation of Biochemical Reactors
8.2.1 Modelling biochemical reactors in the food industry
8.2.2 Automated mathematical modelling of biochemical reactors
8.2.3 Simulation results for biochemical reactors
8.3 Modelling and Simulation of International Trade Dynamics
8.3.1 Mathematical modelling of international trade
8.3.2 Simulation results of international trade

© 2002 Taylor & Francis


CONTENTS vii

8.4 Modelling and Simulation of Aircraft Dynamic Systems 165


8.4.1 Mathematical modelling of aircraft systems 165
8.4.2 Simulation results of aircraft systems 167
8.5 Concluding Remarks and Future Directions 174

9 ADVANCED APPLICATIONS OF ADAPTIVE MODEL-BASED


CONTROL 175
9.1 Intelligent Control of Robotic Dynamic Systems 175
9.1.1 Traditional model-based adaptive control of robotic systems 177
9.1.2 Adaptive model-based control of robotic systems with a
neuro-fuzzy approach 177
9.2 Intelligent Control of Biochemical Reactors 184
9.2.1 Fuzzy rule base for model 'selection 184
9.2.2 Neural networks for identification and control 190
9.2.3 Intelligent adaptive model-based control for biochemical reactors 192
9.3 Intelligent Control of International Trade 202
9.3.1 Adaptive model-based control of international trade 202
9.3.2 Simulation results for control of international trade 204
9.4 Intelligent Control of Aircraft Dynamic Systems 208
9.4.1 Adaptive model-based control of aircraft systems 208
9.4.2 Simulation results for control of aircraft systems 210
9.5 Concluding Remarks and Future Directions 213

References 215

APPENDIX A PROTOTYPE INTELLIGENT SYSTEMS FOR


AUTOMATED MATHEMATICAL MODELLING 225
A.l Automated Mathematical Modelling of Dynamical Systems 225
A.2 Automated Mathematical Modelling of Robotic Dynamic
Systems 229

APPENDIX B PROTOTYPE INTELLIGENT SYSTEMS FOR


AUTOMATED SIMULATION 235
B.l Automated Simulation of Non-Linear Dynamical Systems 235
B.2 Numerical Simulation of Non-Linear Dynamical Systems 239

APPENDIX C PROTOTYPE INTELLIGENT SYSTEMS FOR


ADAPTIVE MODEL-BASED CONTROL 242
C. 1 Fuzzy Logic Model Selection 242
C.2 Neural Networks for Identification and Control 245

© 2002 Taylor & Francis


PREFACE

This book presents a unified view of mathematical modelling, simulation and control for
complex non-linear dynamical systems using soft computing techniques and fractal theory.
Our particular point of view is that modelling, simulation and control are problems that can
not be considered apart because they are intrinsically related in real-world applications.
Control of non-linear dynamical systems can not be achieved if we don't have proper
mathematical models for the systems. Also, useful simulations of a model, that can give us
numerical insights into the behavior of a dynamical system, can not be obtained if we don't
have the appropriate mathematical model. On the other hand, we have to recognize that
complex non-linear dynamical systems can exhibit a wide range of dynamic behaviors
(ranging from simple periodic orbits to chaotic strange attractors), so the problem of behavior
identification is a very diffcult one. Also, we want to automate each of these tasks (mod-
elling, simulation and control) because in this way it is easier to solve a particular problem.
We then have three difficult tasks at hand: automated mathematical modelling of a dynami-
cal system, automated simulation of the model, and model-based control of the system. A
real world problem may require that we use modelling, simulation and control, to achieve
the desired level of performance needed for the particular application.
Soft computing consists of several computing paradigms, including fuzzy logic, neural
networks and genetic algorithms, which can be used to produce powerful hybrid intelligent
systems. We believe that solving the difficult problems of modelling, simulation and control
of non-linear dynamical systems require the use of several soft computing techniques to
achieve the level of intelligence needed to automate the processes of modelling and simula-
tion, and also to achieve adaptive control. On the other hand, fractal theory provides us with
powerful mathematical tools that can be used to understand the geometrical complexity of
natural or computational objects. We believe that, in many cases, it is necessary to use fractal
tools to understand the geometry of the problem at hand. For example, the fractal dimension
is a useful tool in measuring the geometrical complexity of a time series and for this reason
can be used to formulate the corresponding mathematical model for the particular problem.
This book is intended to be a major reference for scientists and engineers interested in
applying new computational and mathematical tools for solving the complicated problems
of mathematical modelling, simulation and control of non-linear dynamical systems. The
book can also be used at the graduate or advanced undergraduate level, as a textbook or
major reference, for courses like: mathematical modelling, numerical simulation, non-
linear control of dynamical systems, applied artificial intelligence and many others. We
consider that this book can also be used to get new ideas for new lines of research or to
continue the lines of future research proposed by the authors of the book. The software
accompanying this book provides a good basis for developing more advanced 'intelligent'
software tools for modelling, simulation and control of non-linear dynamical systems.

© 2002 Taylor & Francis


In Chapter 1, we begin by giving a brief introduction to the problems of modelling,
simulation and control of non-linear dynamical systems. We motivate the importance of
solving these problems, in an automated fashion, for real-world applications. We also
outline the importance of using soft computing techniques and fractal theory to really
achieve automated mathematical modelling and simulation, and model-based adaptive control
of non-linear dynamical systems.
We present in Chapter 2 the main ideas underlying fuzzy logic and the application of this
powerful computational theory to the problem of modelling. We discuss in some detail
fuzzy set theory, fuzzy reasoning and fuzzy inference systems. At the end, we also give
some remarks about fuzzy modelling. The importance of fuzzy logic as a basis for devel-
oping intelligent systems (sometimes in conjunction with other soft computing techniques)
for control has been recognized in many areas of application. For this reason, we consider
reading this chapter essential to understand the new methods for modelling, simulation and
control presented in later chapters.
We present in Chapter 3 the basic concepts, notation and basic learning algorithms for
neural networks. We discuss in some detail feedforward networks, adaptive neuro-fuzzy
inference systems, neuro-fuzzy control and adaptive neuro-control. First, we give a brief
review of the basic concepts of neural networks and the backpropagation learning algo-
rithm. We then give a brief description of adaptive neuro-fuzzy systems. Finally, we end the
chapter with a brief review on the current methods for neuro-fuzzy control and some
remarks about adaptive control and model-based control. We can not emphasize enough the
importance of neural networks as a computational tool to achieve 'intelligence' for software
systems. For this reason, neural networks have been applied for solving complex problems
of modelling, control and identification.
We present in Chapter 4 the basic concepts and notation of genetic algorithms, simulated
annealing and fractal theory. Both genetic algorithms and simulated annealing are basic
search methodologies that can be used for modelling and simulation of complex non-linear
dynamical systems. Since both techniques can be considered as general purpose optimiza-
tion methodologies, we can use them to find the mathematical model which minimizes the
fitting errors for a specific problem. We also present in this chapter the basic concepts of
dynamical systems and fractal theory, which are two powerful mathematical theories that
enable the understanding of complex non-linear phenomena. Dynamical systems theory
gives us the general framework for treating non-linear systems and enables the identifica-
tion of the different dynamical behaviors that can occur for a particular dynamical system.
On the other hand, fractal theory gives us powerful concepts and techniques that can be
used to measure the complexity of geometrical objects.
We present in Chapter 5 our new method for automated mathematical modelling of non-
linear dynamical systems. This method is based on a hybrid fuzzy-fractal approach to
achieve, in an efficient way, automated modelling for a particular problem using a time
series as a data set. The use of the fractal dimension is to perform time series analysis of
the data, so as to obtain a qualitative characterization of the time series. The use of fuzzy
logic techniques is to simulate the process of expert model selection using the qualitative
information obtained from the time series analysis module. At the end, the 'best' math-
ematical model is obtained by comparing the measures of goodness for the selected math-
ematical models. In Chapter 8, we show some advanced applications of this method for
automated mathematical modelling.

© 2002 Taylor & Francis


PREFACE xi

In Chapter 6, we describe the problem of numerical simulation for non-linear dynamical


systems and its solution by using intelligent methodologies. The numerical simulation of
a particular dynamical system consists in the successive application of a map and the
subsequent identification of the corresponding dynamic behaviors. Automated simulation
of a given dynamical system consists in selecting the appropriate parameter values for the
model and then applying the corresponding iterative method (map) to find the limiting
behavior. In this chapter, a new method for automated parameter selection, based on genetic
algorithms, is introduced. Also, a new method for dynamic behavior identification, based
on fuzzy logic, is introduced. The fuzzy-genetic approach for automated simulation con-
sists in the integration of the method for automated parameter selection and the method for
behavior identification.
We describe in Chapter 7 our new method for adaptive model-based control of non-
linear dynamical systems. This method is based on a hybrid neuro-fuzzy approach to achieve,
in an efficient way, adaptive robust control of non-linear dynamical systems using a set of
different mathematical models. We use fuzzy logic to select the appropriate mathematical
model for the dynamical system according to the changing conditions of the system. Adap-
tive control is achieved by using a neural network for control and a neural network for
identification. Combining this method for control with the procedure for fuzzy model
selection, gives us a new method for adaptive model-based control using a hybrid neuro-
fuzzy approach. This method for adaptive control can be used for general dynamical
systems or non-linear plants, since its architecture is domain independent. In Chapter 9, we
show some advanced applications of this new method for adaptive model-based control.
In Chapter 8, we present several advanced applications of the new methods for auto-
mated mathematical modelling and simulation. First, we describe the application of the
new methods for automated modelling and simulation to robotic dynamic systems, which
is a very important application in the control of real-world robot arms and general robotic
systems. Second, we apply our new methods for modelling and simulation to the problem
of understanding the dynamic behavior of biochemical reactors in the food industry, which
is also very important for the control of this type of dynamical system. Third, we consider
the problem of modelling and simulation of international trade dynamics, which is an
interesting problem in economics and finance. Finally, we also consider the problem of
modelling and simulation of aircraft, as this is important for the real-world problem of
automatic aircraft control.
In Chapter 9, we present several advanced applications of the new method for adaptive
model-based control. First, we describe the application of the new method for adaptive
model-based control to the case of robotic dynamic systems, which is very important for
solving the problem of controlling real-world manipulators in real-time. Second, we
describe the application of the method for adaptive model-based control to the case of
biochemical reactors in the food industry, which is a very interesting case due to the
complexity of this non-linear problem. Third, we consider briefly the problem of control-
ling international trade between three or more countries, with our new method for adaptive
model-based control. Finally, we also consider briefly the problem of controlling aircraft
with our new method for adaptive model-based control.
Finally, we would like to thank all the people who helped make this book possible. In
particular, we would like to acknowledge our families for their love and support during the
realization of this project; without them this book would never have been possible.

© 2002 Taylor & Francis


Chapter 1

Introduction to Modelling, Simulation and Control of


Non-Linear Dynamical Systems

We describe in this book new methods for automated modelling and simulation of
non-linear dynamical systems using Soft Computing techniques and Fractal
Theory. We also describe a new method for adaptive model-based control of non-
linear dynamical systems using a hybrid neuro-hzzy-fractal approach. Soft
Computing (SC) consists of several computing paradigms, including fuzzy logic,
neural networks and genetic algorithms, which can be used to produce powerful
hybrid intelligent systems. Fractal Theory (FT) provides us with the mathematical
tools (like the fractal dimension) to understand the geometrical complexity of
natural objects and can be used for identification and modelling purposes.
Combining SC techniques with FT tools we can take advantage of the
"intelligence" provided by the computer methods (like neural networks) and also
take advantage of the descriptive power of fractal mathematical tools. Non-linear
dynamical systems can exhibit extremely complex dynamic behavior and for this
reason it is of great importance to develop intelligent computational tools that will
enable the identification of the best model for a particular dynamical system, then
obtaining the best simulations for the system and also achieving the goal of
controlling the dynamical system in a desired manner. We also describe in this

© 2002 Taylor & Francis


2 INTRODUCTION

book the basic methodology to develop prototype intelligent systems that are able
to find the best model for a particular dynamical system, then perform the
numerical simulations necessary to identify all of the possible dynamical
behaviors of the system, and finally achieve the goal of adaptive control using the
mathematical models of the system and SC techniques.
As a prelude, we shall provide a brief overview of the existing
methodologies for modelling, simulation and control of non-linear dynamical
systems and also of our own approach in dealing with these problems.

1.1 Modelling and Simulation of Non-Linear Dynamical


Systems

Traditionally, mathematical modelling of dynarnical systems has been performed


by human experts in the following manner (Jamshidi, 1997): 1) The expert
according to his knowledge selects a set of models consider to be appropriate for a
specific given problem, 2) Parameter estimation of the models is performed with
methods similar to least-squares (using the relevant data available), and 3) The
"best" model is selected using the measures of goodness for each of the models.
Also, we can say that linear statistical models have been traditionally used as an
approximation of real dynamic systems, which is not the best thing to do since
many of the mechanical, electrical, biological and chemical systems are
intrinsically non-linear in nature. In this work, we achieved automated
mathematical modelling by using different Soft Computing techniques (Jang, Sun
& Mizutani, 1997). The whole process of modelling starts with a time series (data
set), which is used to perform a "Time Series Analysis" to extract the components
of the time series (Weigend & Gershenfeld, 1994). Time series analysis can be
achieved by traditional statistical methods or by efficient classification methods
based on SC techniques, like neural networks or fuzzy logic (Kosko, 1997). In our
case, we used fuzzy logic for classification of the time series components. After
this time series analysis is performed, the qualitative values of the time series
components are used to obtain a set of admissible models for a specific problem,
this part of the problem was solved by using a set of fuzzy rules (knowledge base)

© 2002 Taylor & Francis


MODELLING, SIMULATION AND CONTROL ... 3

that simulates the human experts in the domain of application. Finally, the "best"
model is selected by comparing the measures of goodness for each of the
admissible models considered in the previous step.
The simulation of mathematical models traditionally has been performed
by exploring the possible dynamic behaviors, for a specific system, for different
parameter values of the model (Rasband, 1990). More recently, it has been
proposed to use Artificial Intelligence (Russell & Norvig, 1995) techniques for the
simulation of mathematical models (for example, by using expert systems
(Badiru, 1992)). In this work, we used SC techniques to automate the simulation
of dynamical systems. In particular, we make use of genetic algorithms to generate
the "best" set of parameter values for a specific model with respect to the goal of
obtaining the most efficient simulation possible. Genetic Algorithms (GA)
essentially consist of methods for the optimization of a general function based on
the concept of "evolution" (Goldberg, 1989). In our particular case, the problem
consisted in specifying the appropriate function to be optimized, with the goal of
achieving the most efficient simulation possible, i.e., a simulation that enables the
identification of all the possible dynamic behaviors for a specific dynamical
system. For the identification of dynamic behaviors we make use of a fuzzy rule
base that will identifl a particular behavior according to the results of the
numerical simulations.
In general, the study of non-linear dynamical systems is very important
because most of the physical, electrical, mechanical and biochemical systems can
be mathematically represented by models (differential or difference equations) in
the time domain. Also, it is well known in Dynamical Systems Theory (Devaney,
1989) that the dynamic behavior of a particular system can range from very simple
periodic orbits to the very complicated "chaotic" orbits. Non-linear models may
exhibit the chaotic behavior for systems of at least three coupled differential
equations or at least one difference equation (Ruelle, 1990). In particular, for the
case of real-world dynamical systems the mathematical models needed are of very
high dimensionality and in general there is a high probability of chaotic behavior,
along with all sorts of different periodic and quasi-periodic behaviors (Castillo &
Melin, 1998b). For this reason, it becomes very important to be able to obtain the
appropriate mathematical models for the dynamical systems and then to be able to

© 2002 Taylor & Francis


4 INTRODUCTION

perform numerical simulations of these models (Castillo & Melin, 1997b), since
this enables forecasting system's performance in future time. In this way,
automated mathematical modelling and simulation of dynamical systems can
contribute to real-time control of these systems, and this is critical in real-world
applications (Melin & Castillo, 1998b). Also, an intelligent system for modelling
and simulation can be useful in the design of real dynamical systems with certain
constraints, since the information obtained by the numerical simulations can be
used as a feedback in the process of design. The main contribution of the research
work presented in this book is to combine several Soft Computing techniques to
achieve automated mathematical modelling and simulation of non-linear
dynamical systems using the advantages that each specific technique offers. For
example, fuzzy logic (Von Altrock, 1995) was used to simulate the reasoning
process of human experts in the process of mathematical modelling and genetic
algorithms was used to select the best set of parameter values for the simulation of
the best model.
The importance of the results presented in this book can be measured from
the scientific point of view and also from the practical (or applications) point of
view. First, from the scientific point of view, we consider that this research work
is very important because the computer methods for automated mathematical
modelling and simulation of dynamic systems that were developed contribute, in
general, to the advancement of Computer Science, and, in particular, to the
advancement of Soft Computing and Artificial Intelligence because the new
algorithms that were developed can be considered "intelligent" in the sense that
they simulate human experts in modelling and simulation. From the practical
point of view, we consider the results of this research work very important for the
areas of Control and Design of dynamical systems. Controlling dynamical systems
can be made more easy if we are able to analyze and predict the dynamic
evolution of these systems and this goal can be achieved with an intelligent
system for automated mathematical modelling and simulation. The design of
dynamical systems can be made more easy if we can use mathematical models and
their simulations for planning the performance of these systems under different set
of design constraints. This last two points are of great importance for the
industrial applications, since the control of dynamical systems in real-world plants

© 2002 Taylor & Francis


MODELLPJG, SIMULATION AND CONTROL ...

has to be very precise and also the design of this type of systems for specific tasks
can be very useful for industry.

1.2 Control of Non-Linear Dynamical Systems

Traditional control of non-linear dynamical systems has been done by using


Classical Linear Control Theory and assuming simple linear mathematical models
for the systems. However, it is now well known that non-linear dynamical systems
can exhibit complex behavior (and as a consequence are difficult to control) and
the most appropriate mathematical models for them are the non-linear ones. Since
the complexity of mathematical models for real dynarnical systems is very high it
becomes necessary to use more advanced control techniques. This is precisely the
fact that motivated researchers in the area of Artificial Intelligence (AI) to apply
techniques that mimic human experts in the domain of dynarnical systems control.
More recently, techniques like neural networks and fuzzy logic have been applied
with some success to the control of non-linear dynamical systems for several
domains of application. However, there also has been some limitations and
problems with these approaches when applied to real systems. For this reason, we
proposed in this book the application of a hybrid approach for the problem of
control, combining neural and fuzzy technologies with the knowledge of the
mathematical models for the adaptive control of dynamical systems. The basic
idea of this hybrid approach is to combine the advantages of the computer
methods with the advantages of using mathematical models for the dynamical
systems. In this work, new methods were developed for adaptive control of non-
linear systems using a combination of neural networks, fuzzy logic and
mathematical models. Neural networks were used for the identification and
control of the dynamical system and fuzzy logic was used to enable the change of
mathematical models according to the dynamic state of the system. Also, the
information and knowledge contained in the mathematical models was used for
the control of the system by using their numerical results as input of the neural
networks.

© 2002 Taylor & Francis


6 INTRODUCTION

Traditionally, the control of dynamical systems has been performed using


the classical methods of Linear Control Theory and also using linear models for
the systems (Albertos, Strietzel & Mart, 1997). However, real-world problems can
be viewed, in general, as non-linear dynamical systems with complex behavior
and because of this, the most appropriate mathematical models for these systems
are the non-linear ones. Unfortunately, to the moment, it hasn't been possible to
generalize the results of Linear Control Theory to the case of Non-linear Control
due to the complexity of the mathematics that will be required (Omidvar & Elliot,
1997). Of course, this mathematical generalization could still take several years of
theoretical and empirical research to be developed. On the other hand, it is
possible to use non-linear universal approximators that have resulted from the
research in the area of SC to the problem of system identification and control. In
particular, SC methodologies like neural networks and fuzzy logic have been
applied with some success to problems of control and identification of dynamical
systems (Korn, 1995). However, there are also problems where one or both
methodologies have failed to achieved the level of accuracy desired in the
applications (Omidvar & Elliot, 1997). For this reason, we have proposed in this
work the use of a hybrid approach for the problem of non-linear adaptive control,
i.e., we proposed to combine the use of neural networks and fuzzy logic with the
use of non-linear mathematical models to achieve the goal of adaptive control. In
the following lines we give the general idea of this new approach as well as the
reasons why such an approach is a good alternative for non-linear control of
dynamical systems.
Neural networks are computational systems with learning (or adaptive)
characteristics that model the human brain (Kosko, 1992). Generally speaking,
biological neural networks consist of neurons and connections between them and
this is modeled by a graph with nodes and arcs to form the computational neural
network. This graph along with a computational algorithm to specify the learning
capabilities of the system is what makes the neural network a powerful
methodology to simulate intelligent or expert behavior (Miller, Sutton & Werbos,
1995). It has been shown, that neural networks are universal approximators, in the
sense that they can model any general function to a specified accuracy (Kosko,
1992) and for this reason neural networks have been applied to problems of

© 2002 Taylor & Francis


MODELLING, SIMULATION AND CONTROL ... 7

system identification (Pham & Xing, 1995). Also, because of their adaptive
capabilities neural networks have been used to control real-world dynamical
systems (Ng, 1997).
Fuzzy Logic is an area of SC that enables a computer system to reason
with uncertainty. Fuzzy inference systems consist of a set of "if-then" rules
defined over fuzzy sets. Fuzzy sets are relations that can be used to model the
linguistic variables that human experts use in their domain of expertise (Kosko,
1992). The main difference between fuzzy sets and traditional (crisp) sets is that
the membership function for elements of a hzzy set can take any value between 0
and 1, and not only 0 or 1. This corresponds, in the real world, to many situations
where it is difficult to decide in an unambiguous manner if something belongs or
not to a specific class. Fuzzy expert systems, for example, have been applied with
some success to problems of control, diagnosis and classification just because
they can manage the difficult expert reasoning involved in these areas of
application (Korn, 1995). The main disadvantage with fuzzy systems is that they
can't adapt to changing situations. For this reason, it is a good idea to combine
both methodologies to have the advantages of neural networks (learning and
adaptive capabilities) along with the advantages of fuzzy logic (contain expert
knowledge) in solving complex real world problems where this flexibility is
needed (Yen, Langar & Zadeh, 1995).
In this work, we have proposed a new architecture for developing
intelligent control systems based on the use of neural networks, fuzzy logic and
mathematical models, to achieve the goal of adaptive control of non-linear
dynamical systems. The mathematical model of a non-linear dynamical system
consist of a set of simultaneous non-linear differential (or difference) equations
describing the dynamics of the system. The knowledge contained in the model is
very important in the process of controlling the system, because it relates the
different physical variables and their dependencies (Sueda & Iwamasa, 1995). For
this reason, our approach is to combine mathematical models with neural
networks and fuzzy logic, to achieve adaptive control of non-linear dynamical
systems.
The study of non-linear dynamical systems is very interesting because of
the complexity of the dynamics involved in the underlying processes (for

© 2002 Taylor & Francis


example, biological, chemical or electrical) and also because of the implications,
in the real world, of controlling industrial processes to maximize production. Real
non-linear dynarnical systems can have a wide range of possible dynamic
behaviors, going from simple periodic orbits (stable) to the very complicated
chaotic behavior (Kapitaniak, 1996). Controlling a non-linear dynamical system,
avoiding chaotic behavior, is only possible using the mathematical models of the
system (Sueda & Iwamasa, 1995). For this reason, model-based control is having
great success in the control of complex real-world dynamical systems. In our
approach, the neural networks were used for identification and control of the
system, fuzzy logic was used to choose between different mathematical models of
the system, and the knowledge given by the models was used to avoid specific and
dangerous dynamic behaviors.
We consider the work on non-linear control presented in this book very
important, from the point of view of Computer Science, because it contributed
with new methods to develop intelligent control systems using a new hybrid
model-based neuro-fuzzy approach for controlling non-linear dynarnical systems.
Also, from the point of view of the applications, this work is very important
because it contributed with new methods for adaptive non-linear control that
could eventually be used in the control of real industrial plants or general
dynamical systems, which in turn will result in increased productivity and
efficiency for these systems.

© 2002 Taylor & Francis


Chapter 2

Fuzzy Logic for Modelling

This chapter introduces the basic concepts, notation, and basic operations for
fuzzy sets that will be needed in the following chapters. Since research on Fuzzy
Set Theory has been underway for over 30 years now, it is practically impossible
to cover all aspects of current developments in this area. Therefore, the main goal
of this chapter is to provide an introduction to and a summary of the basic
concepts and operations that are relevant to the study of fuzzy sets. We also
introduce in this chapter the definition of linguistic variables and linguistic values
and explain how to use them in fuzzy rules, which are an efficient tool for
quantitative modelling of words or sentences in a natural or artificial language. By
interpreting fuzzy rules as fuzzy relations, we describe different schemes of fuzzy
reasoning, where inference procedures based on the concept of the compositional
rule of inference are used to derive conclusions from a set of fuzzy rules and
known facts. Fuzzy rules and fuzzy reasoning are the basic components of fuzzy
inference systems, which are the most important modelling tool based on fuzzy set
theory.
The "fuzzy inference system" is a popular computing framework based on
the concepts of fuzzy set theory, fuzzy if-then rules, and fuzzy reasoning (Jang,
Sun & Mizutani, 1997). It has found successful applications in a wide variety of

© 2002 Taylor & Francis


10 FUZZY LOGIC FOR MODELLING

fields, such as automatic control, data classification, decision analysis, expert


systems, time series prediction, robotics, and pattern recognition (Jarnshidi, 1997).
Because of its multidisciplinary nature, the fuzzy inference system is known by
numerous other names, such as "fuzzy expert system" (Kandel, 1992), "fuzzy
model" (Sugeno & Kang, 1988), " k z y associative memory" (Kosko, 1992), and
simply "fuzzy system".
The basic structure of a fuzzy inference system consists of three
conceptual components: a "rule base", which contains a selection of fuzzy rules; a
"data base" (or "dictionary"), which defines the membership functions used in the
fuzzy rules; and a "reasoning mechanism", which performs the inference
procedure upon the rules and given facts to derive a reasonable output or
conclusion. In general, we can say that a fuzzy inference system implements a
non-linear mapping from its input space to output space. This mapping is
accomplished by a number of fuzzy if-then rules, each of which describes the
local behavior of the mapping. In particular, the antecedent of a rule defines a
fuzzy region in the input space, while the consequent specifies the output in the
fuzzy region.
In what follows, we shall first introduce the basic concepts of fuzzy sets
and fuzzy reasoning. Then we will introduce and compare the three types of fuzzy
inference systems that have been employed in various applications. Finally, we
will address briefly the features and problems of fuzzy modelling, which is
concerned with the construction of fuzzy inference systems for modelling a given
target system.

2.1 Fuzzy Set Theory

Let X be a space of objects and x be a generic element of X. A classical set A,


AcX, is defined by a collection of elements or objects x E X, such that each x can
either belong or not belong to the set A. By defining a "characteristic function" for
each element x E X, we can represent a classical set A by a set of order pairs (x,O)
or (x,l), which indicates x P A or x E A, respectively.

© 2002 Taylor & Francis


MODELLING, SIMULATION AND CONTROL ... 11

Unlike the aforementioned conventional set, a fuzzy set (Zadeh, 1965)


expresses the degree to which an element belong to a set. Hence the characteristic
function of a fuzzy set is allowed to have values between 0 and 1, which denotes
the degree of membership of an element in a given set.

Definition 2.1 Fuzzy sets and membership functions


If X is a collection of objects denoted generically by x, then a "fuzzy set" A in X
is defined as a set of ordered pairs:
A = { (x, PA(X))I (2.1)
where pA(x)is called "membership function" (or MF for short) for the fuzzy set A.
The MF maps each element of X to a membership grade (or membership value)
between 0 and 1.
Obviously, the definition of a fuzzy set is a simple extension of the
definition of a classical set in which the characteristic function is permitted to
have any values between 0 and 1. If the values of the membership function pA(x)
is restricted to either 0 or 1, then A is reduced to a classical set and pA(x) is the
characteristic function of A.
A fuzzy set is uniquely specified by its membership function. To describe
membership functions more specifically, we shall define the nomenclature used in
the literature (Jang, Sun & Mizutani, 1997).

Definition 2.2 Support


The "support" of a fuzzy set A is the set of all points x in X such that pA(x)> 0:
'
support (A) = { x l CLAW0 1. (2.2)

Definition 2.3 Core


The "core" of a fuzzy set is the set of all points x in X such that p,(x) = 1:
core (A) = { x 1 pA(x)= 1 ). (2.3)

Definition 2.4 Normality


A fuzzy set A is "normal" if its core is nonempty. In other words, we can always
find a point x E X such that pA(x)= 1.

© 2002 Taylor & Francis


12 FUZZY LOGIC FOR MODELLING

Definition 2.5 Crossover points


A "crossover point" of a fuzzy set A is a point x E X at which pA(x)= 0.5:
crossover (A) = { x I pA(x)= 0.5 ). (2.4)

Definition 2.6 Fuzzy singleton


A fuzzy set whose support is a single point in X with pA(x) = 1 is called a "fuzzy
singleton".
Corresponding to the ordinary set operations of union, intersection and
complement, fuzzy sets have similar operations, which were initially defined in
Zadeh's seminal paper (Zadeh, 1965). Before introducing these three fuzzy set
operations, first we shall define the notion of containment, which plays a central
role in both ordinary and fuzzy sets. This definition of containment is, of course, a
natural extension of the case for ordinary sets.

Definition 2.7 Containment


The fuzzy set A is "contained" in fuzzy set B (or, equivalently, A is a "subset" of
B) if and only if pA(x)I pB(x) for all x. Mathematically,
AE8 pA(x) 5 pB(x). (2.5)

Definition 2.8 Union


The "union" of two hzzy sets A and B is a fuzzy set C, written as C = A u B or C
= A OR B, whose MF is related to those of A and B by

PC(X)= max( PA('), PB(X)) = PA(X) CLB(X). (2.6)

Definition 2.9 Intersection


The "intersection" of two fuzzy sets A and B is a fuzzy set C, written as C = A n B
or C = A AND B, whose MF is related to those of A and B by
~lc(x)= min( PA(X),PB(X) = PAWA PB(X). (2.7)

Definition 2.10 Complement or Negation


The "complement" of a fuzzy set A, denoted by A ( 1A, NOT A), is defined as
p;i(x> = 1- PA(X). (2.8)

© 2002 Taylor & Francis


MODELLING, SIMULATION AND CONTROL ... 13

As mentioned earlier, a fuzzy set is completely characterized by its MF.


Since most fuzzy sets in use have a universe of discourse X consisting of the real
line R, it would be impractical to list all the pairs defining a membership function.
A more convenient and concise way to define an MF is to express it as a
mathematical formula. First we define several classes of parameterized MFs of
one dimension.

Definition 2.1 1 Triangular MFs


A "triangular MF" is specified by three parameters {a, b, c) as follows:

I 0 , xla.
y = triangle(x;a,b,c) = alxlb. (2.9)
blxlc.
clx.

The parameters {a,b,c) (with a < b< c ) determine the x coordinates of the three
corners of the underlying triangular MF.
Figure 2.1 (a) illustrates a triangular MF defined by triangle(x; 10,20,40).

Definition 2.12 Trapezoidal MFs


A "trapezoidal MF" is specified by four parameters {a, b, c, d) as follows:

I
0 , xla.
(x-a)/(b-a) , a lx lb . (2.10)
trapezoid (x;a,b,c,d) = 1 , bsxlc.
(d-X)/ (d-C) , csxld.
0 , dlx.

The parameters {a, b, c, d) (with a < b l c <d) determine the x coordinates of the
four corners of the underlying trapezoidal MF.
Figure 2.1 (b) illustrates a trapezoidal MF defined by trapezoid(x; 10, 20
40, 75).
Due to their simple formulas and computational efficiency, both triangular
MFs and trapezoidal MFs have been used extensively, especially in real-time
implementations. However, since the MFs are composed of straight line segments,

© 2002 Taylor & Francis


14 FUZZY LOGIC FOR MODELLING

they are not smooth at the comer points specified by the parameters. In the
following we introduce other types of MFs defined by smooth and nonlinear
functions.

Definition 2.13 Gaussian MFs


A "Gaussian MF" is specified by two parameters (c ,o )

A "Gaussian MF is determined completely by c and o ; c represents the MFs


center and o determines the MFs width. Figure 2.2 (a) plots a Gaussian MF
defined by gaussian (x; 50,20).

(a) Triangular MF (b) Trapezoidal MF

Figure 2.1 Examples of two types of parameterized MFs

Definition 2.14 Generalized bell MFs

A "generalized bell MF" is specified by three parameters {a, b, c):

bell(x; a, b, c) = 1
1 + 1 (x-c) / a 12b

© 2002 Taylor & Francis


MODELLING, SIMULATION AND CONTROL ... 15

where the parameter b is usually positive. We can note that this MF is a direct
generalization of the Cauchy distribution used in probability theory, so it is also
referred to as the "Cauchy MF".

Figure 2.2 (b) illustrates a generalized bell MF defined by bell(x; 20, 4,


50).

Although the Gaussian MFs and bell MFs achieve smoothness, they are
unable to specify asymmetric MFs, which are important in certain applications.
Next we define the sigmoidal MF, which is either open left or right.

(a) Gaussian MF (b) Generalized Bell MF

Figure 2.2 Examples of two classes of parameterized continuous MFs.

Definition 2.15 Sigmoidal MFs


A "Sigmoidal MF" is defined by the following equation:
sig(x; a, c) = 1 (2.13)
1 + exp [-a(x-c)]
where a controls the slope at the crossover point x = c.
Depending on the sign of the parameter "a", a sigmoidal MF is inherently
open right or left and thus is appropriate for representing concepts such as "very
large" or "very negative". Figure 2.3 shows two sigmoidal functions yl =sig(x; 1, -
5) and y2 =sig(x; -2,5).

© 2002 Taylor & Francis


Discovering Diverse Content Through
Random Scribd Documents
under abnormal conditions to the zigzag stem, or to a flat nose
instead of a crooked one, and so on.
This consideration should make it clear that plant-galls are not in the
remotest degree a stone of stumbling for the determinant theory, as
some have supposed. Of course there can be no 'gall-determinants,'
for galls are not transmissible adaptations of the plants on which
they occur; they arise solely through the larvæ of the gall-insect
which has laid its eggs within the tissues of the plant. But the
specific nature of the different kinds of plant-cells, predetermined by
their determinants, is such that, through the abnormal influences
exercised upon them by the larvæ, they are compelled to a special
reaction which results in the formation of galls. It is marvellous
enough that these abnormal stimuli should be so precisely graded
and adjusted that such a specifically definite structure should result,
and in this case there is obviously a very different state of matters
from that obtaining in most other processes of development, in
which the chief determining factor is rather implied in the nature of
the idioplasm, that is, of the determinants, than in the nature of the
external influences. Here, however, the specific structure of the gall
depends mainly on the quality, variety, and successive effects of the
external influences or stimuli. In discussing the influences of
surroundings I shall return once more to the galls.
My determinants have generally been regarded as if they were like
grains of seed, from which either nothing may arise, under
unfavourable conditions, or just the particular kind of plant from
which the seed itself originated.
This simile is, however, to be taken cum grano salis. The whole
ovum is certainly comparable to a grain of seed, but single
determinants or groups of determinants will always be able to adapt
themselves to different influences, and to remain active even under
slightly abnormal conditions, though in that case the resulting
structures may be somewhat divergent. This relative plasticity is
indispensable even in relation to the ceaseless mutual adaptations of
the growing parts of the organism. Not only do the cells which live
beside each other at the same time influence each other mutually,
but the influence extends to the whole cell-lineage. No cell or group
of cells develops independently of all the others in the body, but
each has its ancestral series of cells on whose determinants it is so
far dependent, since these have taken part in determining its own
nature, in, so to speak, supplying the soil in which ultimately its own
determinants will be sown from the nucleus, and whose influence
modifies these last according to its quality. We might therefore say
that every part is determined by all the determinants of its cell-
ancestors.
If there be urged against the doctrine of determinants the
undoubted fact of the dependence of individual development on
external conditions, or the capacity that organisms have of functional
adaptation, or especially the power that some parts of the organism
have of taking a different form in response to different stimuli, I can
only say that I see no reason why certain cells and masses of cells
should not be adapted from the first for responding differently to
different stimuli.
Therefore I see no contradiction of the determinant theory when, for
instance, among the higher vertebrates, the cells of the connective
tissue exhibit a great diversity of form, becoming a loose 'filling'
connective tissue in one place, a tense fascia, ligament, or tendon
tissue in another, according as they are subjected to slight pressure
on all sides or to stronger pressure on one side. I see no difficulty in
the fact that this connective tissue forms in one case bone-tissue
with the most accurate adaptation of its microscopic structure to the
conditions of stress and pressure which affect the relevant spot, or
in another case cartilaginous tissue, when the cells are exposed to
varying pressure (as on the surface of joints), or even that it gives
rise to blood-vessels when the pressure of the circulating blood and
the tension of the surrounding tissues supply the necessary stimulus.
It is easy to see how important, indeed how necessary, the many-
sidedness of these cells is for the organism, even leaving out of
account such violent interference as the breaking of a bone, the
irregular healing of broken ends of bones, new joint formation, and
so on, and thinking only of the normal phenomena of growth. While
the bone grows it is continually breaking up in the inside and
forming anew on the surface, and this occurs through the power of
the connective tissue-cells to form different tissues under different
influences or stimuli.
We must therefore assume that there are side by side in the
connective cells of higher vertebrates determinants of bone, of
cartilage, of connective tissue in the narrower sense, and of blood-
vessels, and that one or other of these is liberated to activity
according to the stimulus affecting it. Phenomena occur also in the
development of lower animals which lead us to the same
assumption.
Among these is the remarkable behaviour of the primary mesoderm-
cells in the young embryo (gastrula) of the Echinoderms (Fig. 92). At
the point where the primitive gut or archenteron invaginates into the
interior of the hitherto single-layered blastula (Fig. 92, A), some cells
are separated off (M), and move independently, constantly
multiplying the while, into the clear gelatinous fluid (G) which fills
the cavity of the larva, and there they fix themselves, some on the
outer ectodermic layer, others to the various regions and outgrowths
of the archenteron (Ms). According as these cells have established
themselves at one or another point, they become connective tissue,
muscle, or skeleton cells of the dermis, or contribute to the muscular
layer of the food-canal and water-vascular system, or, finally,
become skeleton-forming cells of the calcareous ring which
surrounds the gullet of the sea-cucumber. In all this there is nothing
to indicate a determination of the cells in one direction; on the
contrary it seems as if the fate of the individual cells depended on
the chance conditions which may lead them to one place or to
another.
Fig. 92. Echinoderm-larvæ. A, blastula-stage; the primary
mesoderm-cells (M) are being formed at the subsequent
invagination-area of the endoderm (Ent). Ekt, the
ectoderm. B, gastrula-stage; the archenteron (UD) has
been invaginated (Ent), and between it and the ectoderm
(Ekt) the mesoderm-cells (Ms) migrate into the gelatinous
fluid which fills this cavity. There they attach themselves
partly to the ectoderm, and partly to the endoderm. After
Selenka.

There are thus three possibilities of development, three kinds of


reaction, implied in these cells, which are all outwardly alike, and we
can only understand their rôle in the building up of this very
symmetrical animal if we assume that of these three only one is in
each case liberated, by the specific stimulus exerted by the
immediate surroundings of the cell, so that it may become,
according to the chance position it takes up after its migration,
either a skin-cell, a muscle-cell, or a skeleton-forming cell.
This case may be compared in some respects with the permanent
colour-adaptation of those caterpillars, in regard to which Poulton
demonstrated that they become almost black if they are reared on
blackish-brown bark, light brown on light bark, and green if they are
kept among leaves, and in all cases permanently so. In this case also
the implicated pigment-cells of the skin may develop in three ways,
according to whether this or that quality of the light releases this or
that determinant.
But in many cases we do not know the quality of the liberating
stimulus, and must content ourselves with imagining it. This is so in
the case of dimorphism of the sexes. It is clear that in the males of a
species the germ-cells develop quite otherwise than they do in the
females, that different determining elements attain to activity in
each sex, and since the primary constituents of both sexes must be
contained in most animals in the ovum and in the spermatozoon, we
must assume that in both there are at once 'ovogenic' and
'spermogenic' determinants, of which, however, only one kind
becomes active in a given individual. There are, however, both
among plants and animals hermaphrodite individuals, in which both
kinds of sexual products are developed simultaneously or
successively.
It is not only the primary sexual characters, however, that compel us
to the assumption of double determinants in the germ-plasm, the
secondary sexual characters do so too. We know very well in relation
to ourselves that 'the beautiful soprano voice of the mother may be
transmitted through the son to the grand-daughter, and that the
black beard of the father may pass through the daughter to the
grandson.' Thus both kinds of sexual characters must be present in
every sexually differentiated being, some visible, others latent. In
animals the determinants are sometimes handed on from germ-
plasm to germ-plasm through several generations in a latent state,
and only make their appearance again in a subsequent generation.
This is the case in the water-fleas (Daphnids) and the plant-lice
(Aphides), in which several exclusively female generations succeed
one another, and only in the last of them do males occur again side
by side with the females.
The germ-plasm of the ovum which is ripe for development must
thus contain not only the determinants of the specific ova and
sperms of the species, but also those of all the male and female
sexual characters, which we discussed at length in the section on
sexual selection. I then showed that these secondary sexual
characters differ greatly in range and in strength, that among lower
animals they are almost entirely absent, and that among higher
forms, such Crustaceans, Insects, and Birds, they attain to very
different grades of development even among the same species. Thus
the birds of Paradise are in most species brilliantly coloured and
adorned with decorative feathers only in the male sex, while the
females are simply blackish-grey, but there is a single species in
which the males are almost as soberly coloured as the females.
Conversely, too, we find that in parrots both sexes are usually
coloured alike, but a few species exhibit a totally different colouring
in the two sexes. In the same way the secondary sex differences
may affect only a few parts of the animal or many, while in a few
species the sexes are so divergent in structure that almost
everything about them may be called different. Examples of this are
the dwarf males of most Rotifers, and the males, more minute still in
proportion to the females, of the marine worm Bonellia viridis (p.
227).
We have now to inquire what theoretical explanation of these facts
we can arrive at in accordance with the germ-plasm theory. That
double determinants, male and female, for the differently formed
parts of the two sexes must be assumed to exist in the germ-plasm
has been already said, and we have to suppose that the same
stimulus—usually unknown to us—which incites the determinants of
the primary sexual characters to activity also liberates those of the
secondary characters. But we may safely go a step further and
conclude that there are male and female ids, that is, that the male
and female determinants belong to different ids. I infer this from the
fact that in some groups, such as the Rotifers and certain plant-lice,
the ova are sexually differentiated even at the time of their origin.
Males and females of these animals arise from different kinds of
eggs, which are even externally recognizable. Both develop
parthenogenetically, so that fertilization has nothing to do with it;
from the first, therefore, they must contain ids which consist of
determinants of one sex alone.
If this conclusion be correct, then the sexual equipment of the
determinants of the sexual characters must have taken place in the
course of phylogeny in such a way that each id was affected in one
direction only, and we should thus have to assume male and female
ids, even before the separation of the sexes as males and females,
and the same conclusion must be extended to the primary sexual
characters. Only in this way can we understand the fact that
differences between the sexes, at first small, have increased in the
course of phylogeny to such complete divergence of structure as is
now exhibited in the forms we have named, Bonellia, the Rotifers,
and some parasitic worms.
But there is not only sexual dimorphism, there is also dimorphism of
larvæ, e.g. green and brown caterpillars in certain species of hawk-
moth (Sphinx), and there are sometimes not only two but three or
more forms of a species; and in all these cases determinants of the
differential parts must be represented twice, thrice, or several times
in each germ-plasm, in each fertilized ovum, at least in all cases in
which the different forms live together on the same area. In
discussing mimicry we spoke of species of butterfly which were
everywhere alike or nearly so in the male sex, while the females
were not only quite different from the males, but differed greatly in
many respects among themselves. Three different forms of females
of Papilio merope occur in the same region of Cape Colony, each of
these resembling a protected model. All three forms have been
obtained from the eggs of one female. In this case the female ids of
the germ-plasm must be represented by three different sets, one of
which, when it is in the majority in the fertilized ovum, gives rise to
the Danais-form, the second to the Niavius-form, and the third to
the Echeria-form of the species. Phylogenetically considered, it is
probable that each of these three kinds of ids originated by itself, on
a more limited area on which the protected model lived in
abundance; but with a wider distribution the different female ids
mingled together, were united through the males into a single germ-
plasm, and now occasionally exhibit all three forms on the same
area. I doubt whether there is any other theory that can offer an
interpretation of these facts, and I regard them, therefore, as
affording further evidence of the real existence of ids.
The polymorphism of social insects must be thought of as similarly
based in the germ-plasm.
In bees there are in addition to the males and females the so-called
workers, and this can only depend on the existence of special kinds
of ids. Those of the workers were originally truly female, but as
many of their determinants underwent variations advantageous for
the maintenance of the species, they were modified into special
'worker-ids.' I postpone for the present any inquiry into the causes
by which these ids come to dominate the ontogeny; obviously it
cannot be by the mere fact of being in a majority over the rest of
the ids, as I indicated in the case of the butterflies with polymorphic
females.
In many ants the division of labour goes further still; there are two
kinds of workers in the colony, the ordinary workers and the so-
called 'soldiers,' and in this case the worker-id must have developed
in two different directions in the course of phylogeny, and have
separated into two kinds of ids, so that the germ-plasm of these
species must contain four kinds of ids.
I might cite many more cases in regard to which the assumption of
two or more kinds of determinants seems imperative, but I believe
that what has been said is enough to enable any one to think out
other cases for himself.
LECTURE XIX
THE GERM-PLASM THEORY (continued)
Co-operation of the determinants to form an organ: insect appendages
—Venation of the insect-wing—Deformities in Man—Apex of the fly's
leg—Proofs of the existence of determinants—Claws and adhesive
lobes—Difference between a theory of development and a theory of
heredity—Metamorphosis of the food-canal in insects—Delage's theory
—Reinke's theory of the organism-machine—Fechner's views—
Apparent contradiction by the facts of developmental mechanics—
Formation of the germ-cells—Displacement of the germinal areas in
the hydro-medusoid polyps, a proof of the existence of germ-tracks.

It would be futile to attempt to guess at the arrangement of the


determinants in the germ-plasm, but so much at least we may say,
that the determinants do not lie beside each other in the same
disposition as their determinates exhibit in the fully-formed
organism. This may be inferred from the complex formative
processes of embryogenesis in which many groups of cells, which in
their origin were far apart, combine together to form an organ. Thus
the arrangement of the determinants in the germ-plasm does not
correspond to the subsequent arrangement of the whole animal, nor
are primary constituents of the complete organs contained within
the germ-plasm. The organ is undoubtedly predetermined in the
germ-plasm, but it is not preformed as such.
Here, again, the history of development gives us a certain basis of
fact from which to work. Let us consider, for instance, the origin of
the appendages in those insects which in the larval state possess
neither legs nor wings, but exhibit a gradual emergence of these
structures from concealment underneath the integumentary
skeleton. In these cases, as I have already shown in regard to the
wings, the development of the limbs arises from definite groups of
cells in the skin. These must therefore be regarded as the formative,
and therefore as the most important and indispensable, parts of the
rudiments, and may be designated the imaginal disks, as I many
years ago proposed[22] (Fig. 89, ui and oi).
[22] Die Entwicklung der Dipteren, Leipzig, 1864.

But these disks of cells do not contain the whole leg, but only the
skin-layer of it, the 'hypodermis,' which, however, in this case
undoubtedly determines the form. But the internal parts of the leg,
especially the nerves, tracheæ, and probably also the muscles, are
formed from other cell-groups and grow into the imaginal disk from
outside. Something similar probably takes place in the case of all
organs which are made up of many parts; they are, so to speak,
shot together from several points of origin, from various primordia;
and determinants are brought into co-operation whose relative value
in determining the form and function of the organ may be very
diverse.
For it is undoubtedly a very different matter whether a cell bears
within it the elements which compel it in the course of growth to
develop an organ, for instance a leg, of quite definite size, sculpture,
number of joints, and so on, or whether it only bears the somewhat
vague power of determining that connective tissue or fatty tissue is
to be produced. In the first case it controls the whole formation of
the part, in the second it only fills up gaps or lays down fat or other
substances within itself if these be presented to it. Between these
two extremes of determining power there are many intermediate
stages. Cells which contain the determinants of blood-vessels,
tracheæ, or nerves need not be so definitely determined that they
always give rise to precisely the same blood-vessels, the same
branching of the tracheæ, or the same bifurcation of nerves; they
may probably possess no more than the general tendency to the
formation of such parts, and the special form taken by the nerves,
tracheæ, or blood-vessels may be essentially determined by their
environment. Thus in the morbid tumours of Man, nerves, and
especially blood-vessels, may develop in a quite characteristic
manner, which was certainly not determined in advance, but has
been called forth by the
stimulus, the pressure,
and other influences of
the cellular basis of the
tumour. In short, the cells
were only determined to
this extent, that they
contained the tendency
to give rise to blood-
vessels under particular Fig. 89. Anterior region of the
influences. larva
of a Midge (Corethra plumicornis).
It would be a mistake, K, head.
however, to think of the Th, thorax. ui, inferior imaginal
primary constituents of all disks.
cell-groups as so oi, superior imaginal disks. ui1,
indefinite. Let us call to ui2, and
mind, for instance, the ui3, the primordia of the limbs. oi2
venation of the insect and oi3, the primordia of the
wing. It is well known wings and
that this is not only quite 'balancers.' g, brain. bg, chain of
ventral
different in beetles, bugs, ganglia with nerves which enter
and Diptera from that in the
the Hymenoptera, and imaginal disks. trb, tracheal
different again in the vesicle.
butterflies, but that it is Enlarged about 15 times.
quite characteristic in
every individual family of
butterflies, and indeed in every genus. We cannot conceive of the
absolute certainty of development of these very characteristic and
constant branchings as having its roots elsewhere than in the
determinants of the germ-plasm, which, lying within certain series of
cells, ultimately cause particular cell-series of the wing-rudiment to
become the wing-veins. If this were not so, how would it be possible
to understand the fact that every minute deviation in the course of
these veins is repeated in exactly the same way in all the individuals
of a genus, while in all the individuals of an allied genus the venation
turns out slightly different with equal constancy.
But it is quite certain that all determinations are in some degree
susceptible to modifying influences, that they are in very different
degrees capable of variation.
Many deformities of particular parts in Man and the higher animals
may be referred to imperfect or inhibited nutrition of the part in
question during embryonic development; the determinants alone
cannot make the part, they must have a supply of formative
material, and according as this material is afforded more abundantly
or more scantily the part will turn out larger or smaller. In the same
way the pressure conditions of the surrounding parts must in many
cases have a furthering or inhibiting influence, or may even
determine the shape. But it is quite possible, indeed even probable,
that other specific influences are exerted by the cells or cell-
aggregates surrounding an organ which is in process of being
formed, just as the stake on which a twining plant is growing may
prompt it to coil. If the stake be absent, the predetermined twining
of the plant cannot attain to more than very imperfect expression, if
indeed it finds any. The spirally coiled sheath of muscle-cells which
occurs so often around blood-vessels in worms, Echinoderms, and
Vertebrates is probably due to similar processes, that is, on the one
hand, to a specific mode of reaction characteristic of these cells, and
predetermined from the germ; on the other hand, to the external
influence of the cell-surroundings without which the determination of
the muscle-cell is not liberated, that is, is not excited to activity.
Fig. 93. The development of a limb in the pupa of a Fly
(Sarcophaga carnaria). A, apex of the limb from a pupa
four days old; the jointing is hinted at; hy, hypodermis;
ps, pupal sheath; ph, phagocytes; tr, tracheal branch. B,
the same on the fifth day; the lumen of the limb is quite
filled with phagocytes (ph); the last tarsal joint (t5) is
beginning to show a bifid apex. C, the same on the
seventh day; the claws (Kr) and the adhesive lobes (hl)
are formed.

But even if every determinant requires a stimulus to liberate it,


whether this stimulus consists in currents of particular nutritive
fluids, in contact with other cells, or, conversely, on the removal of
some pressure previously exerted on the cell by its surroundings, the
material cause of a structure is to be sought for not in these
conditions of its appearance, but in the primary constituents which
have been handed on to the relevant cell or cell-group from the
germ, in other words, through its determinants. How, for instance,
could the blunt rounded knob of the rough and clumsily jointed sac
of cells which represents the insect's leg at the beginning of the
pupal period (Fig. 93, A) be incited to thicken, to constrict at the
root (B), and to form a joint-surface, to broaden out at the end, and
produce two sharply cut points (C), which become incurved and
form claws (kr), while beneath these a broad flat lobe (hl) grows
forward, and with its regularly disposed cells gradually forms the
characteristic adhesive organ of the fly—how could all this happen if
there were not contained within these cells special formative forces
which determine them not only in their form and the rest of their
constitution, but above all in their power of multiplication? No special
external stimulus affects the still unfinished knob of the fly's leg
unless it be the removal of pressure; but this operates regularly, and
cannot be the cause of the growth, at definite places, of claws and
adhesive lobes with all their characteristically placed hairs.
We require to assume that each of the cells composing the primary
rudiment of the limb possessed a determining power which made it
grow and multiply under the given conditions of nutrition and
pressure in a prescribed manner and at a prescribed rate; and we
must make the same assumption in regard to all the daughter and
grand-daughter-cells, and so on. The strictest regulation of the
power of multiplication of each of the implicated cells is a necessary
condition of the constant production of the same two claws and
adhesive lobes, the same form of tarsal joint, the same regular
covering of hair, and so on. This exact determination of the cells can
only take place through material vital particles, and it is these which
I call determinants.
I have already said so much about the assumed 'determinants' of
the germ-plasm that it might perhaps be supposed that we have
now exhausted the topic; but the assumption of such 'primary
constituents' is so fundamental, not only for my own germ-plasm
theory of to-day and to-morrow, but also—unless I am much
mistaken—for all future theories of development and inheritance. In
point of fact, the conception of determinants has as yet penetrated
so little into the consciousness of biologists, that I cannot remain
content with what I have already said, but must endeavour to test
and to corroborate my thesis by additional illustrations.
As far as I am aware, only a few zoologists have expressly and
unconditionally agreed with the assumption of determinants; on the
other hand, several biologists have rejected it as fanciful and
untenable, while others have set it aside as a useless playing with
ideas. The last, I am inclined to believe, have not taken the trouble
to think out what the idea is. It has even been objected that there
can be no determinants because we can see nothing of them, and
that they must therefore be pure figments of the imagination,
invented to explain facts which could be explained much more easily
and simply in some other way. From the very first I have stated
emphatically that they have not been, and never will be seen,
because they lie far below the limit of visibility, and thus can at best
only become visible when they are collected in large aggregates like
chromatin granules. Nor have I any objections to make if any one
chooses to describe all the details of their activity as mere
hypotheses, such, for instance, as their distribution during
development, their 'maturation,' their migration from the nucleus,
and the manner in which they control the cell. All this is really an
imaginative picture which may be correct to a certain degree, but
may also be erroneous; no formal proof of it can be obtained at
present; and I am content if it be simply admitted to be possible. On
the other hand, the existence of determinants seems to me to be, in
the sense indicated, indubitable and demonstrable.
Let us return for a moment to the claws and adhesive lobes which
are developed on the foot of the fly. It may perhaps be thought that
it is possible to do without the assumption of determinants for these
parts, by assuming that although 'external' influences in the ordinary
sense could not possibly have determined that certain cells of the
apex of the leg should form claws and others adhesive lobes, the
result might be due to the differences of intercellular pressure within
the apical knob; these may have been stronger in one direction,
weaker in another, thus prompting the cells to grow here into claws
and there into adhesive lobes. If we had merely to explain from the
constitution of the germ-plasm the ontogeny or development of
these parts in an individual fly there might perhaps be no radical
objection to this view, though it would hardly be possible to explain
the assumed differences in pressure otherwise than as due to a
different intensity of growth in the cells in the various regions of the
limb-apex, which again would have to be referred to differences in
the germ-plasm. But when we reflect that these parts vary
hereditarily and independently of other parts, and owe their present
form to their power of doing so, and that they are differently formed
in every genus and species, we see at once that they must be
represented in the germ-plasm by particular vital particles, which are
the roots of their transmissible variability, that is, which must have
previously undergone a corresponding variation if the relevant parts
themselves are to vary. Without previous variation of the
determinants of the germ no transmissible independent deviation on
the part of the claws or adhesive lobes of the animal is conceivable.
All the opponents of my theory have overlooked this fact; both Oscar
Hertwig and Kassowitz have forgotten that a theory of development
is not a theory of heredity; they only aim at the former, and they
therefore dispute the logical necessity for an assumption of
determinants.
But as this is the very foundation of the theory, let me further submit
the following considerations in its favour.
In insects which undergo metamorphosis, not only the external but
the internal parts of the caterpillar or larva go through a more or
less complete transformation. In the flies (Muscidæ), for instance,
the whole intestinal tract of the larva is reconstructed in the pupa; in
fact it breaks up into a loose, flocculent, dead, but still coherent
mass of tissue. Within this there arises a new intestine, as I have
shown in an early work (1864); and Kowalewsky and Van Rees have
since made us aware of the interesting details of this reconstruction,
showing that the new intestine arises from definite cells of the old
one, which are present in the larval gut at certain fairly wide
distances, and which do not share in the general destruction, but
remain alive, grow, and multiply, and form islands of cells in the
dead mass. These living islands, continually extending, ultimately
come into contact and again form a closed intestinal canal which
differs entirely from that of the larva in its form, in its various areas,
and in its differentiation. In this case those formative cells of the
imago-intestine must have contained the elements which determined
their descendants in number, power of multiplication, arrangement,
and histological differentiation. In other words, each of these cells
must contain the determinants of a particular limited section of the
intestine of the imago. The other cells of the intestinal epithelium
could not do this, even though they were under exactly the same
conditions, were included in the same intimate cell-aggregate, and
had the same nutritional opportunities. They break up when the
formative cells begin to be active, for till then the latter had
remained inactive, and had not multiplied, although they lay
regularly distributed among the other cells. Whence, then, could the
entire difference in the behaviour of these two sets of cells arise, if it
does not depend on the nature of the cells themselves, and how
could this difference of nature have developed during the racial
history of insect-metamorphosis if determinants did not reach the
cell from the germ-plasm—determinants which conditioned that
some cells should be hereditarily modified into the cells of the
imago-intestine and others into the larval intestine? Quite similar
processes have been recently demonstrated in regard to the
reconstruction of the larval intestine in other insect-groups.
Deegener has done this, for instance, for the water-beetle
(Hydrophilus piceus); and it is certain that all these reconstructions
start from particular cells, which lie indifferently between the active
cells during the larval period, and contain the primary constituents
for the formation of a section of the intestine, but which only
become active when their hitherto living neighbours die and break
up.
The whole of the reconstruction of the external form of the fly takes
place in a similar manner. Not only the limb, the head, the stigmata,
but the skin itself is formed anew from imaginal disks. In each of the
abdominal segments three pairs of little cell-islands are formed
during larval life, and these only enter on the stage of formative
activity after pupation, when they multiply rapidly and grow together
to form a segment, whose size, form, and external nature is
determined by them. But it is well known that the abdominal
segments of the fly differ from those of the larva very markedly and
in every respect, so that each cell-island must contain determinants
which are quite different from those in the skin-cells of the
corresponding larval segments. These last break up at the beginning
of pupahood, while the former begin to grow vigorously, and to
spread themselves out. The most remarkable fact about the whole
business, and it seems to me also the most instructive, is that these
imaginal disks frequently appear for the first time during larval life,
as I found in the case of a midge, Coretha plumicornis, in regard to
the disks of the thorax, and as Bruno Wahl[23] has recently
demonstrated in the case of the abdominal cell-islands. Since in the
young larva the position of the subsequent imaginal disks is
occupied by cells which apparently in no way differ from the rest of
the skin-cells, and are also exposed to precisely the same external
and internal influences, the origination of the imaginal cells from
these can only depend on differential cell-division; the primordial cell
of each imaginal disk must have separated at the beginning of disk-
formation into a larval and an imaginal skin-cell.
[23] Bruno Wahl, Ueber die Entwickelung der hypodermalen
Imaginalscheiben im Thorax und Abdomen der Larve von
'Eristalis' L., Zeitschr. f. wiss. Zool., Bd. lxx. 1901.
In insects in which the larva and the imago differ widely, the perfect
insect, as regards all its principal parts, is already represented in the
larva, namely, in particular cells which lie among those of the
corresponding larval parts, and do not visibly differ from these,
although they are equipped with quite different determinants, and
consequently enter on their formative activity much later, and give
rise to quite different structures. As the determinants of the whole
animal with all its parts are contained in the ovum, so those of the
parts of its imaginal phase are contained in these cells of the
imaginal disks.
In addition to all this, we have incontrovertible evidence in favour of
the theory of determinants in the independent phyletic variations of
the individual stages of development, on which depends the whole
phenomenon of 'metamorphosis' which we have just been
considering. How could the larval stage have become so different
from the imago-stage, if the one were not alterable by variation
arising in the germ without the other being affected? If this absolute
independence of the transmissible variability of the individual stages
were not an indispensable assumption in the explanation of
metamorphosis and other phenomena of development, I should
regard an attempt at a theory of development without determinants
as justifiable. But I am forced to see in this fact alone an invalidation
of all epigenetic theories of development, that is, of all theories
which assume a germ-substance without primary constituents, which
can produce the complicated body solely by varying step by step
under the influence of external influences, both extra- and intra-
somatic. It is possible to conceive of an ovum in which the living
substance is of such a kind that it must vary in a definite manner
under the influence of warmth, air, pressure, and so on, that it must
divide into similar, and subsequently also into dissimilar parts, which
then interact upon each other in diverse ways and give rise to
further variations, which in their turn result in differentiations and
variations, till ultimately we have the whole complicated organic
machine complete and 'finished' in every detail. Certainly no mortal
could make any pronouncement as to the constitution of such a
substance, but even if we assume it, for the nonce, as possible, how
can we account for the transmissible variation of the individual parts
and developmental stages, on which the whole phylogenetic
evolution depends?
As the development of the butterfly exhibits the three main stages of
caterpillar, pupa, and perfect insect, each of which is independently
and hereditarily variable, and therefore implies a something in the
germ, whose variation brings about a change in the one stage only,
so the ontogeny of every higher animal is made up of numerous
stages, which are all capable of independent and transmissible
variation. How else should we human beings, in our embryonic
phase, still possess the gill-arches of our fish-like ancestors,
although much modified and without the gills? Truly, he who would
seek to deny that the stages of individual development are capable
of independent and transmissible variation must know very little
about embryology. But if the facts are as stated, how can they be
reconciled with the conception of a germinal substance developing in
epigenetic fashion? Every variation in this substance would affect not
only the whole succession of stages, but the whole organism with all
its parts. In this way too, then, we are driven to the conclusion that
there must be something in the germ whose variation causes
variation only in a particular part of a particular stage. This
something we define in our conception of the 'primary constituents'
(Anlagen)—the determinants. These are not to be thought of either
as 'miniature models,' or even as the 'seeds' of the parts; they alone
cannot produce the part which they determine, but they effect
changes in the cell in which they become active, causing it to vary in
such a manner that the formation of the relevant part results. While
I conceive of development as a continuous process, I supplement
this with the idea that from within, namely, from the nuclear
substance, new, directive, 'determining' influences are continually
being exerted on the developing cells.
I can hardly think of a better proof of the necessity of this
assumption than that furnished by Delage, one of the most acute
biologists of France, who, in his comprehensive book on Heredity,
has striven to replace the theory of determinants by something
simpler. Delage rejects all 'primary constituents' (Anlagen) in the
germ, all 'particules représentatives,' as much too complicated an
assumption, and thinks it possible to work with the conception of a
germ-plasm which is about as simple as the cell-substance of a
Rhizopod, that is to say, a protoplasm of definite chemico-physical
constitution and composition. Leaving out of account the
consideration that the protoplasm of an amœba is scarcely of such
extreme simplicity, but is certainly made up of numerous
differentiated and definitely arranged biophors, how could such an
extremely simple ('éminemment simple') constitution of the ovum as
is here assumed give rise to such a complicated organism, the
individual parts of which are capable of independent and
transmissible variation? According to Delage it does so because the
ovum, though not containing 'all the factors requisite for its ultimate
resultant,' does contain 'un certain nombre des facteurs nécessaires
à la détermination de chaque partie et de chaque caractère de
l'organisme futur'! Determinants after all, it may be said, but that is
far from the truth! It is not primary constituents that the germ
contains, according to Delage, it is chemical substances, for instance
muscle substances, probably 'les substances caractéristiques des
principales catégories de cellules, c'est-à-dire, celles qui, dans ces
cellules, sont la condition principale de leur fonctionnement.' All
these must be contained in the ovum. How they are to reach their
proper place in the organism, how the 'characteristic chemical
substance' of a mole is to land just behind the right or left ear of the
fully formed man, is not stated. But apart from this, there is a much
deeper error in this assumption of specific chemical substances in
the ovum as an explanation of the phenomena of local hereditary
variation, and I have already touched upon it: chemical substances
are not vital units, which feed and reproduce, which assimilate and
which bear a charm against the assimilating power of the
surrounding protoplasm. They would necessarily be modified and
displaced in the course of ontogeny, and would therefore—no matter
where they had been placed at first—be incapable of performing all
that Delage ascribes to them. Either the germ contains 'living'
primary constituents, or it is, as Delage maintains, determined
chemico-physically; but in the latter case there is no scope for
hereditary local variation. Delage must either renounce the attempt
to explain this, or he must transform his 'substances chimiques' into
real and actually living determinants.
Thus from all sides we are forced to the conclusion that the germ-
substance on the whole owes its marvellous power of development
not only to its chemico-physical constitution, whether that be
eminently simple or marvellously complex, but to the fact that it
consists of many and different kinds of 'primary constituents'
(Anlagen), that is, of groups of vital units equipped with the forces
of life, and capable of interposing actively and in a specific manner,
but also capable of remaining latent in a passive state, until they are
affected by a liberating stimulus, and on this account able to
interpose successively in development. The germ-cell cannot be
merely a simple organism, it must be a fabric made up of many
different organisms or units, a microcosm.
Yet another train of thought leads us to the same idea, and this has
its roots in the extraordinary complexity of the machine which we
call the organism.
The botanist Reinke has recently called attention once again to the
fact that machines cannot be directly made up of primary physico-
chemical forces or energies, but that, as Lotze said, forces of a
superior order are indispensable, which so dispose the fundamental
chemico-physical forces that they must act in the way aimed at by
the purpose of the machine. To produce a watch it is not enough to
bring together brass, steel, gold, and stones; to produce a piano it is
not enough to lay wood, iron, leather, ivory, steel, &c., side by side,
but these stuffs must be brought together in a definite form and
combination. In the same way, the mere juxtaposition of carbon and
water does not result in a carbohydrate like sugar or illuminating
gas; the component elements only yield what is desired when they
are placed in a particular and absolutely definite relation to each
other, in which they so act upon and with one another that sugar or
illuminating gas results, and the same is true of the component
elements of a watch or of a piano. In the watch and in the piano this
relation is arranged by human intelligence, by the workmen who
form the different materials and put them together in the proper
manner. In this case, then, human intelligence is, as Reinke says, the
'superior force' which compels the energies to work together in a
particular way.
But organisms also are machines which perform a particular and
purposeful kind of work, and they are only capable of doing so
because the energies which perform the work are forced into
definite paths by superior forces; these superior forces are thus 'the
steersmen of the energies.' There is undoubtedly a kernel of truth in
this view, and I shall return to it. Reinke, however, uses it in a way
which I cannot follow; that is, he infers from it a 'cosmic intelligence'
which puts these superior forces into the organisms, and thus
controls these machines to purposeful work, as the watchmaker puts
'superior forces' into the watch by means of wheels, cylinders, and
levers. In one case it is human intelligence which controls the
'superior forces,' in the other 'cosmic' intelligence. I cannot regard
this reasoning from analogy as convincing, because, in the first
place, these 'superior forces' are not 'forces' at all. They are
constellations of energy, co-ordinations of matter and the energies
immanent therein under complex and precisely defined conditions,
and it is a matter of indifference whether chance or human
intelligence has brought them together. If we take Reinke's own
example of carbohydrates it is certain that our coal-gas is due to the
intelligence of man, which brings together the carbon and the water
in such a way that coal-gas must arise. The 'superior forces' must
here be looked for in the arrangements of the coke-stove, and, in
the second place, in the intelligence of man. But when decaying
plants in the marsh form another carbon-compound, marsh-gas,
where do the directing 'superior forces' come in? Surely only in the
fortuitous concomitance of the necessary materials and the
necessary conditions. Or may 'cosmic' intelligence have established
this laboratory in the marsh? If not, what can compel us to refer the
formation of dextrin or starch in the cells of the green leaves of
plants to 'superior forces' which are placed in them by 'cosmic'
intelligence? I am far from believing that the great and deep
problem here touched upon can be put aside in any off-hand
manner, but I feel sure that it will never be solved by word-play
about energies and 'superior forces.'
Let us return to the kernel of truth in Reinke's thesis; it lies in this,
that, while the working of a machine does really depend on the
forces or energies which are bound up with the stuffs of which it
consists, it also depends on a particular combination of these stuffs
and forces, on a particular 'constellation' of them, as Fechner
expressed it. In the watch these 'constellations' are the springs, the
wheels, &c., and their position in relation to each other; but in the
organism they are the organs, down to the cells and cell-parts; for
the cell too is a machine, indeed a very complex one, as its functions
prove. There are thousands of kinds of 'constellations' of elementary
substances and forces which condition the activity of the living
machine, and only when all these constellations are present in the
proper manner and in the proper relations to each other can the
functions of the organism be properly discharged.
But the living machine differs essentially from other machines in the
fact that it constructs itself; it arises by development from a cell, by
going through numerous 'stages of development.' But none of these
stages is a dead thing, each is itself a living organism whose chief
function is to give rise to the next stage. Thus each stage of the
development may be compared to a machine whose function
consists in producing a similar but more complex machine. Each
stage is thus composed, just like the complete organism, of a
number of such 'constellations' of elementary substances and
elementary forces, whose number in the beginning is relatively
small, but increases rapidly with each new stage.
But whence come these 'constellations' or, to keep to our metaphor,
the levers, wheels, and cranks of each successive stage in the
making of the organic machine? The epigenetic theory of a germ-
plasm without primary constituents answers by pointing to internal
and external influences which cause the germ-plasm, originally
homogeneous, to differentiate gradually more and more, bringing it
into the most diverse 'constellations.' But how can such influences
introduce new springs, levers, and wheels of a quite specific kind, as
must be the case if apparently similar germinal substances are to
give rise to two such different animals as a domestic duck and a
teal? The cause must lie in the invisible differences in the
protoplasm, opponents will answer, and we with them. But our
studies up to this point have shown us that the differences cannot
be merely elementary differences, cannot be merely of a physico-
chemical nature depending on the composition of the raw material
and the implicated energies; they must depend on the definite co-
ordination of substances and energies, in other words, on the
occurrence of 'constellations' of these. Thus the germ-plasm must be
composed of definite and very diverse combinations of living units,
which are themselves bound up in a higher 'constellation,' so that
they act as a living machine at the first stage of development, and
liberate into activity the already existing constellations of the second
stage. The second stage in the series of living machines which arise
successively from each other liberates the sleeping 'constellations'
for the third, and so on.
These 'constellations' of matter and energy are the biophors, the
determinants, and the 'groups of determinants' which we may think
of as disposed in a manifold overlapping series. That they do not
enter into activity all at once, but successively take their part in
development, seems to me a necessary consequence of their
successive origin in the phylogeny; and the ontogeny, as we shall
see later, arises through a modified condensation of the phylogeny.
Now since every new determinant that arises in the course of
phylogeny can only develop by division and subsequent variation
from the determinants which were previously active at the same
place in the organism, it is quite intelligible that later on, when the
phylogeny has been condensed in the ontogeny, they should not
enter upon their active stage at the same time as their phyletic
predecessors, but after them. The theory of Oscar Hertwig, who
starts from a germ-plasm without primary constituents, that all parts
of the germ-plasm become active at the same time, seems to me
quite untenable. How could the wheels, levers, and springs of the
complete vital machine, which arose so very slowly in the course of
phylogeny, arise to-day in the ontogeny in such rapid succession
unless they were already present in the germ-plasm and only
required to be incited to activity, that is, liberated by the stage
preceding them? Even Fechner supported this view when he
supposed that the interaction and mutual influences of the parts in
the organism, that is, of the 'constellations,' gave rise of themselves
to the succeeding stage, that is to say, to the new constellations
peculiar to the succeeding stage. To this Reinke reasonably objected
that it was like expecting the window frames of a house in process
of building to produce the panes of glass. The panes in the organism
only develop in the window frames if their determinants have been
present in the germ-plasm from the beginning, and are liberated by
the development of the frames, just as the activity of the glazier is
liberated by the sight of the completed frames. Neither new panes
nor new determinants could be produced rapidly; the former must
be manufactured in the glass factory, the latter in the developmental
workshop of the form of life in question, which workshop we call its
phylogeny. But just as it is unnecessary to erect a new glass factory
for each new house that is built, so the development of each
individual does not require the establishment each time of those
numberless life-factories—the constellations—whose business it is to
produce anew the wheels, levers, springs, and cylinders of the
developmental machinery at each stage, for they are all provided for
in the germ-plasm, and it is only on this account that they are
capable of hereditary variation.
I have already directed attention to some embryological facts which
seem to be contradictory, if not to the germ-plasm theory itself, at
least to the assumption it makes that the germ-plasm is analysed
out during the ontogeny; and something more must be said on this
head. I refer to the numerous facts brought to light through the
science of developmental mechanics founded by Wilhelm Roux, and
particularly to the investigations as to the prospective significance of
the segmentation-cells of the animal ovum.
Among these investigations we find experiments in compressing
certain eggs (sea-urchin's) in the early stages of segmentation. The
blastomeres are prevented by artificial pressure from grouping
themselves in the normal manner; they are compelled to spread out
side by side in the same plane. If the pressure is removed, they
change their grouping, and yield a normal embryo. I will not here
discuss whether these results can only be interpreted as showing
that each segmentation-cell has the same prospective significance,
and that it is only its relative position which decides what part of the
embryo is to be formed from it; this could not be done without going
into great detail; I therefore assume it to be true, and confine my
survey to the second group of experiments, those on isolated
segmentation-cells.
It has been shown that in the eggs of the most diverse animals, for
instance in the sea-urchin once more, each of the two first
blastomeres, if separated from one another, can develop into a
complete larva. Indeed, in the eggs of sea-urchin and some other
animals each of the first four, or any of the first eight, blastomeres,
and indeed any segmentation-cell during the earlier stages,
possesses the power of developing to a certain point, namely, as far
as the so-called 'blastula-larva.' This seems to contradict a theory
which assumes that the primary constituents become separated in
the successive stages of ontogeny. But in the first place the
blastomeres of all animals do not behave in this way, and, moreover,
the facts can be quite well explained without entirely renouncing the
assumption of the segregation of the determinant-complexes. It is
only necessary to assume that the segmentation-cells, which
develop in the isolated condition as if they were intact eggs, still
contain the complete germ-plasm, and that the differential
segregation into groups of determinants with dissimilar hereditary
tendencies takes place later. This would certainly load the theory
with further complications, and I shall not enter into the question
here, since the facts which we should have to consider are as yet by
no means undisputed.
But in any case the facts of developmental mechanics referred to,
which we owe to numerous excellent observers of the last decade,—
I need only name W. Roux, O. Hertwig, Chun, Driesch, Barfurth,
Morgan, Conklin, Wilson, Crampton, and Fischel—not only leave the
essential part of the germ-plasm theory untouched, but rather
strengthen than endanger its more subordinate points, such as the
assumption of a segregation of the components of the germ-plasm
in the course of ontogenesis.
As to the fundamental ideas expressed in the theory, I have already
shown that these remain unaltered, even if we do not assume a
disintegration or segregation of the germ-plasm, but think of all the
developing cells as equipped with the complete germ-plasm. In that
case the determinants would be liberated to activity solely by
specific stimuli. But in regard to the assumption of disintegration, it
must be noted that the facts cited relative to the sea-urchin's ova do
not by any means hold true of the eggs of all animals.
In various animal types each of the first two segmentation-cells,
when separated from its neighbour, produces only a half-embryo,
and any one of the first four cells a quarter-embryo. This 'fractional
embryo' is, however, in some cases able later to develop into a
whole embryo (to 'postgenerate' itself, as W. Roux says). The
isolated blastomere shows, to begin with, an activity of only a half of
the primary constituents of the animal, as was first established by W.
Roux and maintained conclusively, in spite of many attacks, until it
was established beyond doubt by the detailed corroboratory
investigations of Endres. The secondary completion of the embryo,
which, however, is still disputed, must be regarded as a
regeneration, and, to explain it, a co-operation of the complete but
not yet wholly active germ-plasm in both segmentation-cells must
therefore be assumed.
It would carry us too far if I were to deal in detail even with the
most important of the numerous facts that the last decade has
brought to light; I shall restrict myself to the most essential.
That isolated segmentation-cells have the capacity of developing into
embryos which are complete but correspondingly smaller in size has
been demonstrated in animals of various groups, though it does not
seem to go to the same length in all. In the Medusæ we find that
not only one of the first two, but one of the first four, eight, and
even sixteen segmentation-cells may develop a whole larva when
isolated (Zoja). In the sea-urchin at least any one of the first eight
blastomeres may do so. And Driesch's experiments in cutting up the
young larvæ at the blastula-stage (a single-layered ball of cells)
leads us to assume that each of these cells still possesses the
complete germ-plasm. Beyond that stage, however, the primary
constituents obviously divide into those of the ectoderm and those
of the endoderm, for the subsequent two-layered stage in the sea-
urchin's development, the gastrula, does not complete itself if it be
artificially divided into fragments which consist only of cells from the
outer, or only of cells from the inner layer. In corroboration of this
experiment made by Barfurth, Samassa was able to demonstrate in
regard to the egg of the frog that, even after the third division of the
ovum, the segmentation-cells are so different from each other in
respect of their primary constituents that they were not able to
replace each other mutually. When this investigator killed the
ectoderm-cells alone by means of an induction current, or the
endoderm-cells alone, the dead half could not be replaced by the
half which remained alive, and the whole ovum perished.
If these facts may be adduced in favour of a separation of the
primary constituents at an earlier or later stage, we find even
stronger proofs among the Ctenophores, Gastropods, Bivalves, and
Annelids. In the last-named group Wilson has shown it to be
probable that development is really a 'mosaic work,' as Roux and I
had assumed. The older observations made by Chun at an earlier
date on the Ctenophora, and the more recent experiments of Fischel
on the same animals, prove the same thing for this group. In this
case complete larvæ are easily distinguished from mere 'partial
developments' by the number of the characteristic 'ciliated
meridional rows' or ribs, which extend from one pole of the larva to
another. In the complete larva there are eight of these, but in larvæ
from one of the first two blastomeres (isolated) there are only four,
and in those which have arisen from one of the first four
blastomeres there are only two. If an ovum at the eight-cell stage
can be successfully divided into separate blastomeres, each of these
will form an 'eighth larva,' always with only one ciliated rib. Even in
the succeeding sixteen-cell stage it could still be demonstrated that
the substance responsible for the formation of the ribs only lies in
particular places and always suffices only for eight ribs. The sixteen-
cell stage consists of eight large cells and eight small ones, the
'macromeres' and the 'micromeres'; if an ovum at this stage be cut
so that one piece contains five macromeres and five micromeres, a
partial larva will develop which possesses only five ribs, while the
larva from the other portion will have only three. But the localizing of
the rib-determinants can be followed still further, for in larvæ in
which individual micromeres have been displaced from their normal
position there is a correlated displacement of the corresponding ribs,
and a dislocation of their ciliated comb-plates. The determinants of
the ribs must therefore lie in the micromeres, and we must conclude
that at the antecedent division they were only imparted to one
daughter-nucleus, while the other, that of the macromere, did not
receive this kind of determinant. Here then we have an example of
dissimilar or differential division. Those who oppose this theory of
qualitative division will hardly be likely to admit this, but will rather
seek to maintain that 'external influences,' such as relative position,
determine which cells are to give rise to the ciliated ribs and which
are not. But the fact that artificial displacement of the micromeres
leads to a disarrangement of the ciliated comb-plates, of which the
ribs are made up, invalidates this suggestion, and at the same time
overthrows the interpretation that it may be the cells which lie on
particular meridians that are determined by this position to the
production of ciliated plates. Obviously, the converse of this is true;
those cells which contain the rib-determinants come to lie in the
regular course of development in these eight meridians, and the
cells lying between them, though of the same descent (from
micromeres), contain no such determinants and therefore form no
ribs. But if those cells which are equipped with rib-determinants be
artificially displaced, then they give rise to swimming-plates
elsewhere than on the aforesaid meridians.
The experiments made by Crampton on a marine Gastropod,
Ilyanassa, likewise go to prove that a disintegration or segregation of
the primary constituents does occur in the course of development.
In this case, when the first two or first four segmentation-cells were
artificially separated from each other, they developed exactly as if
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and


personal growth!

ebookname.com

You might also like