Download Full Version ebook - Visit ebookmeta.

com

Deep Learning with Python 2nd Edition Nikhil

Ketkar

https://ebookmeta.com/product/deep-learning-with-python-2nd-
edition-nikhil-ketkar/

OR CLICK HERE

DOWLOAD NOW

Discover More Ebook - Explore Now at ebookmeta.com

 Instant digital products (PDF, ePub, MOBI) ready for you
Download now and discover formats that fit your needs...

Start reading on any device today!

Deep Learning with Python 1st Edition Nikhil Ketkar

https://ebookmeta.com/product/deep-learning-with-python-1st-edition-
nikhil-ketkar/

ebookmeta.com

Fundamentals of Deep Learning Nikhil Buduma

https://ebookmeta.com/product/fundamentals-of-deep-learning-nikhil-
buduma/

ebookmeta.com

Applied Deep Learning with TensorFlow 2: Learn to

Implement Advanced Deep Learning Techniques with Python,
2nd Edition Umberto Michelucci
https://ebookmeta.com/product/applied-deep-learning-with-
tensorflow-2-learn-to-implement-advanced-deep-learning-techniques-
with-python-2nd-edition-umberto-michelucci-2/
ebookmeta.com

Developing Self and Self Concepts in Early Childhood

Education and Beyond 1st Edition Bridie Raban

https://ebookmeta.com/product/developing-self-and-self-concepts-in-
early-childhood-education-and-beyond-1st-edition-bridie-raban/

ebookmeta.com
The Consumer Society Myths and Structures Jean Baudrillard

https://ebookmeta.com/product/the-consumer-society-myths-and-
structures-jean-baudrillard/

ebookmeta.com

Insight Guides France Travel Guide eBook 7th Edition

Insight Guides

https://ebookmeta.com/product/insight-guides-france-travel-guide-
ebook-7th-edition-insight-guides/

ebookmeta.com

World s Great Men of Color Volume II Rogers J A

https://ebookmeta.com/product/world-s-great-men-of-color-volume-ii-
rogers-j-a/

ebookmeta.com

Chicken Essentials: A Chicken Cookbook with Delicious

Chicken Recipes (2nd Edition) Booksumo Press

https://ebookmeta.com/product/chicken-essentials-a-chicken-cookbook-
with-delicious-chicken-recipes-2nd-edition-booksumo-press/

ebookmeta.com

Another Kind of Eden First Edition James Lee Burke

https://ebookmeta.com/product/another-kind-of-eden-first-edition-
james-lee-burke/

ebookmeta.com
Lippincott Illustrated Reviews Biochemistry 7th Edition
2017 Denise R Ferrier

https://ebookmeta.com/product/lippincott-illustrated-reviews-
biochemistry-7th-edition-2017-denise-r-ferrier/

ebookmeta.com
Nikhil Ketkar and Jojo Moolayil
Deep Learning with Python
Learn Best Practices of Deep Learning Models with
PyTorch
2nd ed.
Nikhil Ketkar
Bangalore, Karnataka, India

Jojo Moolayil
Vancouver, BC, Canada

Any source code or other supplementary material referenced by the

author in this book is available to readers on GitHub via the book’s
product page, located at www.​apress.​com/​978-1-4842-5363-2. For
more detailed information, please visit http://​www.​apress.​com/​
source-code.

ISBN 978-1-4842-5363-2 e-ISBN 978-1-4842-5364-9

https://doi.org/10.1007/978-1-4842-5364-9

© Nikhil Ketkar, Jojo Moolayil 2021

Apress Standard

Trademarked names, logos, and images may appear in this book. Rather
than use a trademark symbol with every occurrence of a trademarked
name, logo, or image we use the names, logos, and images only in an
editorial fashion and to the benefit of the trademark owner, with no
intention of infringement of the trademark. The use in this publication
of trade names, trademarks, service marks, and similar terms, even if
they are not identified as such, is not to be taken as an expression of
opinion as to whether or not they are subject to proprietary rights.

The publisher, the authors and the editors are safe to assume that the
advice and information in this book are believed to be true and accurate
at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, express or implied, with respect to the material
contained herein or for any errors or omissions that may have been
made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

Distributed to the book trade worldwide by Springer Science+Business

Media New York, 233 Spring Street, 6th Floor, New York, NY 10013.
Phone 1-800-SPRINGER, fax (201) 348-4505, e-mail orders-
ny@springer-sbm.com, or visit www.springeronline.com. Apress Media,
LLC is a California LLC and the sole member (owner) is Springer
Science + Business Media Finance Inc (SSBM Finance Inc). SSBM
Finance Inc is a Delaware corporation.
Introduction
This book has been drafted with a unique approach. The second edition
focuses on the practicality of the topics within deep learning that help
the reader to embrace modern tools with the right mathematical
foundations. The first edition focused on introducing a meaningful
foundation for the subject, while limiting the depth of the practical
implementations. While we explored a breadth of technical frameworks
for deep learning (Theano, TensorFlow, Keras, and PyTorch), we limited
the depth of the implementation details. The idea was to distill the
mathematical foundations while focusing briefly on the practical tools
used for implementation.
A lot has changed over the past three years. The deep learning
fraternity is now stronger than ever, and the frameworks have evolved
in size and adoption. Theano is now deprecated (ceased development);
TensorFlow saw huge adoption in the industry and academia; and
Keras became more popular among beginners and deep learning
enthusiasts. However, PyTorch has emerged recently as a widely
popular choice for academia as well as industry. The growing number
of research publications that recently have used PyTorch over
TensorFlow is a testament to its growth within deep learning.
On the same note, we felt the need to revise the book with a focus
on engaging readers with hands-on exercises to aid a more meaningful
understanding of the subject. In this book, we have struck the perfect
balance, with mathematical foundations as well as hands-on exercises,
to embrace practical implementation exclusively on PyTorch. Each
exercise is supplemented with the required explanations of PyTorch’s
functionalities and required abstractions for programming
complexities.
Part I serves as a brief introduction to machine learning, deep
learning, and PyTorch. We explore the evolution of the field, from early
rule-based systems to the present-day sophisticated algorithms, in an
accelerated fashion.
Part II explores the essential deep learning building blocks. Chapter
3 introduces a simple feed-forward neural network. Incrementally and
logically, we uncover the various building blocks that constitute a
neural network and which can be reused in building any other network.
Though foundational, Chapter 3 focuses on building a baby neural
network with the required framework that helps to construct and train
networks of all kinds and complexities. In Chapter 4, we explore the
core idea that enabled the possibility of training large networks
through backpropagation using automatic differentiation and chain
rule. We explore PyTorch’s Autograd module with a small example to
understand how the solution works programmatically. In Chapter 5, we
look at orchestrating all the building blocks discussed through so far,
along with the performance metrics of deep learning models and the
artifacts required to enable an improved means for training—i.e.,
regularization, hyperparameter tuning, overfitting, underfitting, and
model capacity. Finally, we leverage all this content to develop a deep
neural network for a real-life dataset using PyTorch. In this exercise, we
also explore additional PyTorch constructs that help in the
orchestration of various deep learning building blocks.
Part III covers three important topics within deep learning. Chapter
6 explores convolutional neural networks and introduces the field of
computer vision. We explore the core topics within convolutional
neural networks, including how they learn and how they are
distinguished from other networks. We also leverage a few hands-on
exercises—using a small MNIST dataset as well as the popular Cats and
Dogs dataset—to study the practical implementation of a convolutional
neural network. In Chapter 7, we study recurrent neural networks and
enter the field of natural language processing. Similar to Chapter 6, we
incrementally build an intuition around the fundamentals and later
explore practical exercises with real-life datasets. Chapter 8 concludes
the book by looking at some of the recent trends within deep learning.
This chapter is only a cursory introduction and does not include any
implementation details. The objective is to highlight some advances in
the research and the possible next steps for advanced topics.
Overall, we have put in great efforts to write a structured, concise,
exercise-rich book that balances the coverage between the
mathematical foundations and the practical implementation.
Acknowledgments
I would like to thank my colleagues at Flipkart and Indix, and the
technical reviewers, for their feedback and comments. I will also like to
thank Charu Mudholkar for proofreading the book in its final stages.

—Nikhil Ketkar

I would like to thank my beloved wife, Divya, for her constant

support.

—Jojo Moolayil
Table of Contents
Chapter 1:​Introduction to Machine Learning and Deep Learning
Defining Deep Learning
A Brief History
Advances in Related Fields
Prerequisites
The Approach Ahead
Installing the Required Libraries
The Concept of Machine Learning
Binary Classification
Regression
Generalization
Regularization
Summary
Chapter 2:​Introduction to PyTorch
Why Do We Need a Deep Learning Framework?​
What Is PyTorch?​
Why PyTorch?​
It All Starts with a Tensor
Creating Tensors
Tensor Munging Operations
Mathematical Operations
Element-Wise Mathematical Operations
Trigonometric Operations in Tensors
Comparison Operations for Tensors
Linear Algebraic Operations
 Summary
Chapter 3:​Feed-Forward Neural Networks
What Is a Neural Network?​
Unit
The Overall Structure of a Neural Network
Expressing a Neural Network in Vector Form
Evaluating the Output of a Neural Network
Training a Neural Network
Deriving Cost Functions Using Maximum Likelihood
Binary Cross-Entropy
Cross-Entropy
Squared Error
Summary of Loss Functions
Types of Activation Functions
Linear Unit
Sigmoid Activation
Softmax Activation
Rectified Linear Unit
Hyperbolic Tangent
Backpropagation
Gradient Descent Variants
Gradient-Based Optimization Techniques
Practical Implementation with PyTorch
Summary
Chapter 4:​Automatic Differentiation in Deep Learning
Numerical Differentiation
 Symbolic Differentiation
Automatic Differentiation Fundamentals
Implementing Automatic Differentiation
Summary
Chapter 5:​Training Deep Leaning Models
Performance Metrics
Classification Metrics
Regression Metrics
Data Procurement
Splitting Data for Training, Validation, and Testing
Establishing the Achievable Limit on the Error Rate
Establishing the Baseline with Standard Choices
Building an Automated, End-to-End Pipeline
Orchestration for Visibility
Analysis of Overfitting and Underfitting
Hyperparameter Tuning
Model Capacity
Regularizing the Model
Early Stopping
Norm Penalties
Dropout
A Practical Implementation in PyTorch
Interpreting the Business Outcomes for Deep Learning
Summary
Chapter 6:​Convolutional Neural Networks
Convolution Operation
 Pooling Operation
Convolution-Detector-Pooling Building Block
Stride
Padding
Batch Normalization
Filter
Filter Depth
Number of Filters
Summarizing key learnings from CNNs
Implementing a basic CNN using PyTorch
Implementing a larger CNN in PyTorch
CNN Thumb Rules
Summary
Chapter 7:​Recurrent Neural Networks
Introduction to RNNs
Training RNNs
Bidirectional RNNs
Vanishing and Exploding Gradients
Gradient Clipping
Long Short-Term Memory
Practical Implementation
Summary
Chapter 8:​Recent Advances in Deep Learning
Going Beyond Classification in Computer Vision
Object Detection
Image Segmentation
 Pose Estimation
Generative Computer Vision
Natural Language Processing with Deep Learning
Transformer Models
Bidirectional Encoder Representations from Transformers
GrokNet
Additional Noteworthy Research
Concluding Thoughts
Index
About the Authors
Nikhil Ketkar
currently leads the Machine Learning
Platform team at Flipkart, India’s largest
ecommerce company. He received his
PhD from Washington State University.
Following that, he conducted
postdoctoral research at University of
North Carolina at Charlotte, which was
followed by a brief stint in high-
frequency trading at TransMarket in
Chicago. More recently, he led the data
mining team at Guavus, a startup doing
big data analytics in the telecom domain,
and Indix, a startup doing data science in the ecommerce domain. His
research interests include machine learning and graph theory.

Jojo Moolayil
is an artificial intelligence professional
and published author of three books on
machine learning, deep learning, and IoT.
He is currently working with Amazon
Web Services as a Research Scientist –
A.I. in their Vancouver, BC office.
In his current role with AWS, Jojo
works on researching and developing
large-scale A.I. solutions for combating
fraud and enriching the customer’s
payment experience in the cloud. He is
also actively involved as a technical
reviewer and AI consultant with leading
publishers and has reviewed over a
dozen books on machine learning, deep
learning, and business analytics.
You can reach Jojo at:
https://www.jojomoolayil.com/
https://www.linkedin.com/in/jojo62000
https://twitter.com/jojo62000
About the Technical Reviewers
Judy T. Raj
is a Google Certified Professional Cloud
Architect. She has great experience with
the three leading cloud platforms—
Amazon Web Services, Azure, and Google
Cloud Platform—and has co-authored a
book on Google Cloud Platform with
Packt Publications. She has also worked
with a wide range of technologies in
machine learning, data science,
blockchains, IoT, robotics, and mobile
and web app development. She is
currently a technical content engineer in
Loonycorn. Judy holds a degree in
computer science and engineering from Cochin University of Science
and Technology. A driven engineer fascinated with technology, she is a
passionate coder, a machine language enthusiast, and a blockchain
aficionado.

Manohar Swamynathan
is a data science practitioner and an avid
programmer, with more than 14 years of
experience in various data science-
related areas, including data
warehousing, business intelligence (BI),
analytical tool development, ad-hoc
analysis, predictive modeling, data
science product development,
consulting, formulating strategy, and
executing analytics programs. His career
has covered the life cycle of data across
multiple domains, such as US mortgage
banking, retail/ecommerce, insurance,
and industrial IoT. Manohar has a bachelor’s degree with a
specialization in physics, mathematics, computers, and a master’s
degree in project management. He is currently living in Bengaluru, the
silicon valley of India.
© Nikhil Ketkar, Jojo Moolayil 2021
N. Ketkar, J. Moolayil, Deep Learning with Python
https://doi.org/10.1007/978-1-4842-5364-9_1
1. Introduction to Machine Learning and Deep Learning
Nikhil Ketkar1 and Jojo Moolayil2
(1) Bangalore, Karnataka, India
(2) Vancouver, BC, Canada

The subject of deep learning has gained immense popularity recently, and, in the process, has given rise to
several terminologies that make distinguishing them fairly complex. One might find the task of neatly
separating each field overwhelming, with the sheer volume of overlap between the topics.
This chapter introduces the subject of deep learning by discussing its historical context and how the field
evolved into its present-day form. Later, we will introduce machine learning by covering the foundational
topics in brief. To start with deep learning, we will leverage the constructs gained from machine learning
using basic Python. Chapter 2 begins the practical implementation using PyTorch.
Defining Deep Learning
Deep learning is a subfield within machine learning that deals with the algorithms that closely resemble an
over-simplified version of the human brain that solves a vast category of modern-day machine intelligence.
Many common examples can be found within the smartphone’s app ecosystem (iOS and Android): face
detection on the camera, auto-correct and predictive text on keyboards, AI-enhanced beautification apps,
smart assistants like Siri/Alexa/Google Assistant, Face-ID (face unlock on iPhones), video suggestions on
YouTube, friend suggestions on Facebook, cat filters on Snapchat are all products that were made the state-
of-the-art only for deep learning. Essentially, deep learning is ubiquitous in the today’s digital life.
Truth be told, it can be complicated to define deep learning without navigating some historical context.

A Brief History
The journey of artificial intelligence (AI) to its present day can be broadly divided into four parts: viz. rule-
based systems, knowledge-based systems, machine, and deep learning. Although the granular transitions in
the journey can be mapped into several important milestones, we will cover a more simplistic overview. The
entire evolution is encompassed into the larger idea of “artificial intelligence.” Let’s take a step-by-step
approach to tackle this broad term.

Figure 1-1 The AI landscape

The journey of Deep Learning starts with the field of artificial intelligence, the rightful parent of the field,
and has a rich history going back to the 1950s. The field of artificial intelligence can be defined in simple
terms as the ability of machines to think and learn. In more layman words, we would define it as the process
of aiding machines with intelligence in some form so that they can execute a task better than before. The
above Figure 1-1 showcases a simplified landscape of AI with the various aforementioned fields showcased
a subset. We will explore each of these subsets in more detail in the section below.

Rule-Based Systems
The intelligence we induce into a machine may not necessarily be a sophisticated process or ability;
something as simple as a set of rules can be defined as intelligence. The first-generation AI products were
simply rule-based systems, wherein a comprehensive set of rules were guided to the machine to map the
exhaustive possibilities. A machine that executes a task based on defined rules would result in a more
appealing outcome than a rigid machine (one without intelligence).
A more layman example for the modern-day equivalent would be an ATM that dispenses cash. Once
authenticated, users enter the amount they want and the machine, based on the existing combination of
notes in-store, dispenses the correct amount with the least number of bills. The logic (intelligence) for the
machine to solve the problem is explicitly coded (designed). The designer of the machine carefully thought
through the comprehensive list of possibilities and designed a system that can solve the task
programmatically with finite time and resources.
Most of the early day’s success in artificial intelligence was fairly simple. Such tasks can be easily
described formally, like the game of checkers or chess. This notion of being able to easily describe the task
formally is at the heart of what can or cannot be done easily by a computer program. For instance, consider
the game of chess. The formal description of the game of chess would be the representation of the board, a
description of how each of the pieces moves, the starting configuration, and a description of the
configuration wherein the game terminates. With these notions formalized, it is relatively easy to model a
chess-playing AI program as a search, and, given sufficient computational resources, it’s possible to produce
relatively good chess-playing AI.
The first era of AI focused on such tasks with a fair amount of success. At the heart of the methodology
were a symbolic representation of the domain and the manipulation of the symbols based on given rules
(with increasingly sophisticated algorithms for searching the solution space to arrive at a solution).
It must be noted that the formal definitions of such rules were done manually. However, such early AI
systems were fairly general-purpose task/problem solvers in the sense that any problem that could be
described formally could be solved with the generic approach.
The key limitation of such systems is that the game of chess is a relatively easy problem for AI simply
because the problem set is relatively simple and can be easily formalized. This is not the case with many of
the problems human beings solve on a day-to-day basis (natural intelligence). For instance, consider
diagnosing a disease or transcribing human speech to text. These tasks, which human beings can do but
which are hard to describe formally, presented as a challenge in the early days of AI.

Knowledge-Based Systems
The challenge of addressing natural intelligence to solve day-to-day problems evolved the landscape of AI
into an approach akin to human-beings—i.e., by leveraging a large amount of knowledge about the
task/problem domain. Given this observation, subsequent AI systems relied on large knowledge bases that
captured the knowledge about the problem/task domain. Note that the term used here is knowledge, not
information or data. By knowledge, we simply mean data/information that a program/algorithm can reason
about. An example could be a graph representation of a map with edges labeled with distances and about of
traffic (which is being constantly updated), allowing a program to reason about the shortest path between
points.
Such knowledge-based systems, wherein the knowledge was compiled by experts and represented in a
way that allowed algorithms/programs to reason about it, represented the second generation of AI. At the
heart of such approaches were increasingly sophisticated approaches for representing and reasoning about
knowledge to solve tasks/problems that required such knowledge. Examples of such sophistication include
the use of first-order logic to encode knowledge and probabilistic representations to capture and reason
where uncertainty is inherent to the domain.
One of the key challenges that such systems faced, and addressed to some extent, was the uncertainty
inherent in many domains. Human beings are relatively good at reasoning in environments with unknowns
and uncertainty. One key observation here is that even the knowledge we hold about a domain is not black
or white but grey. A lot of progress was made in this era on representing and reasoning about unknowns and
uncertainty. There were some limited successes in tasks like diagnosing a disease that relied on leveraging
and reasoning using a knowledge base in the presence of unknowns and uncertainty.
The key limitation of such systems was the need to hand-compile the knowledge about the domain from
experts. Collecting, compiling, and maintaining such knowledge bases rendered such systems impractical. In
certain domains, it was extremely hard to even collect and compile such knowledge—for example,
transcribing speech to text or translating documents from one language to another. While human beings can
easily learn to do such tasks, it’s extremely challenging to hand-compile and encode the knowledge related
to the tasks—for instance, the knowledge of the English language and grammar, accents, and subject matter.
To address these challenges, machine learning is the way forward.

Machine Learning
In formal terms, we define machine learning as the field within AI where intelligence is added without
explicit programming. Human beings acquire knowledge for any task through learning. Given this
observation, the focus of subsequent work in AI shifted over a decade or two to algorithms that improved
their performance based on data provided to them. The focus of this subfield was to develop algorithms that
acquired relevant knowledge for a task/problem domain given data. It is important to note that this
knowledge acquisition relied on labeled data and a suitable representation of labeled data as defined by a
human being.
 Consider, for example, the problem of diagnosing a disease. For such a task, a human expert would collect
a lot of cases where a patient had and did not have the disease in question. Then, the human expert would
identify a number of features that would aid in making the prediction—for example, the age and gender of
the patient, and the results from a number of diagnostic tests, such as blood pressure, blood sugar, etc. The
human expert would compile all this data and represent it in a suitable form—for example, by
scaling/normalizing the data, etc. Once this data were prepared, a machine learning algorithm could learn
how to infer whether the patient has the disease or not by generalizing from the labeled data. Note that the
labeled data consisted of patients that both have and do not have the disease. So, in essence, the underlying
machine language algorithm is essentially doing the job of finding a mathematical function that can produce
the right outcome (disease or no disease) given the inputs (features like age, gender, data from diagnostic
tests, and so forth). Finding the simplest mathematical function that predicts the outputs with the required
level of accuracy is at the heart of the field of machine learning. For example, questions related to the
number of examples required to learn a task or the time complexity of an algorithm are specific areas for
which the field of ML has provided answers with theoretical justification. The field has matured to a point
where, given enough data, compute resources, and human resources to engineer features, a large class of
problems are solvable.
The key limitation of mainstream machine language algorithms is that applying them to a new problem
domain requires a massive amount of feature engineering. For instance, consider the problem of recognizing
objects in images. Using traditional machine language techniques, such a problem would require a massive
feature-engineering effort wherein experts identify and generate features that would be used by the
machine language algorithm. In a sense, true intelligence is in the identification of features; the machine
language algorithm is simply learning how to combine these features to arrive at the correct answer. This
identification of features or the representation of data that domain experts do before machine language
algorithms are applied is both a conceptual and practical bottleneck in AI.
It’s a conceptual bottleneck because if features are being identified by domain experts and the machine
language algorithm is simply learning to combine and draw conclusions from this, is this really AI? It’s a
practical bottleneck because the process of building models via traditional machine language is
bottlenecked by the amount of feature engineering required. There are limits to how much human effort can
be thrown at the problem.

Deep Learning
The major bottleneck in machine learning systems was solved with deep learning. Here, we essentially took
the intelligence one step further, where the machine develops relevant features for the task in an automated
way instead of hand-crafting. Human beings learn concepts starting from raw data. For instance, a child
shown with a few examples of a particular animal (say, cats) will soon learn to identify the animal. The
learning process does not involve a parent identifying a cat’s features, such as its whiskers, fur, or tail.
Human learning goes from raw data to a conclusion without the explicit step where features are identified
and provided to the learner. In a sense, human beings learn the appropriate representation of data from the
data itself. Furthermore, they organize concepts as a hierarchy where complicated concepts are expressed
using primitive concepts.
The field of deep learning has its primary focus on learning appropriate representations of data such
that these could be used to conclude. The word “deep” in “deep learning” refers to the idea of learning the
hierarchy of concepts directly from raw data. A more technically appropriate term for deep learning would
be representation learning , and a more practical term for the same would be automated feature engineering .
Advances in Related Fields
It is important to note the advances in other fields like compute power, storage cost, etc. that have played a
key role in the recent interest and success of deep learning. Consider the following, for example:
The ability to collect, store and process large amounts of data has greatly advanced over the last decade
(for instance, the Apache Hadoop ecosystem).
The ability to generate supervised training data (data with labels—for example, pictures annotated with
the objects in the picture) has improved a lot with the availability of crowd-sourcing services (like
Amazon Mechanical Turk).
The massive improvements in computational horsepower brought about by graphical processing units
(GPUs) enabled parallel computing to new heights.
The advances in both the theory and software implementation of automatic differentiation (such as
PyTorch or Theano) accelerated the speed of development and research for deep learning.
Although these advancements are peripheral to deep learning, they have played a big role in enabling
advances in deep learning.
Prerequisites
The key prerequisites for reading this book include a working knowledge of Python and some coursework in
linear algebra, calculus, and probability. Readers should refer to the following in case they need to cover
these prerequisites.
Dive Into Python, by Mark Pilgrim - Apress Publications (2004)
Introduction to Linear Algebra (Fifth Edition), by Gilbert Strang - Wellesley-Cambridge Press
Calculus, by Gilbert Strang - Wellesley-Cambridge Press
All of Statistics (Section 1, chapters 1-5), by Larry Wasserman - Springer (2010)
The Approach Ahead
This book focuses on the key concepts of deep learning and its practical implementation using PyTorch. In
order to use PyTorch, you should possess a basic understanding of Python programming. Chapter 2
introduces PyTorch, and the subsequent chapters discuss additional important constructs within PyTorch.
Before delving into deep learning, we need to discuss the basic constructs of machine learning. In the
remainder of this chapter, we will explore the baby steps of machine learning with a dummy example. To
implement the constructs, we will use Python and again implement the same using PyTorch.
Installing the Required Libraries
You need to install a number of libraries in order to run the source code for the examples in this book. We
recommend installing the Anaconda Python distribution
(https://www.anaconda.com/products/individual), which simplifies the process of installing
the required packages (using either conda or pip). The list of packages you need include NumPy, matplotlib,
scikit-learn, and PyTorch.
PyTorch is not installed as a part of the Anaconda distribution. You should install PyTorch, torchtext, and
torchvision, along with the Anaconda environment.
Note that Python 3.6 (and above) is recommended for the exercises in this book. We highly recommend
creating a new Python environment after installing the Anaconda distribution.
Create a new environment with Python 3.6 (use Terminal in Linux/Mac or the Command Prompt in
Windows), and then install the additional necessary packages, as follows:

conda create -n testenvironment python=3.6

conda activate testenvironment

pip install pytorch torchvision torchtext

For additional help with PyTorch, please refer to the Get Started guide at
https://pytorch.org/get-started/locally/.
The Concept of Machine Learning
As human beings, we are intuitively aware of the concept of learning. It simply means to get better at a task
over time. The task could be physical, such as learning to drive a car, or intellectual, such as learning a new
language. The subject of machine learning focuses on the development of algorithms that can learn as
humans learn; that is, they get better at a task over a period over time and with experience—thus inducing
intelligence without explicit programming.
The first question to ask is why we would be interested in the development of algorithms that improve
their performance over time, with experience. After all, many algorithms are developed and implemented to
solve real-world problems that don’t improve over time; they simply are developed by humans,
implemented in software, and get the job done. From banking to ecommerce and from navigation systems in
our cars to landing a spacecraft on the moon, algorithms are everywhere, and, a majority of them do not
improve over time. These algorithms simply perform the task they are intended to perform, with some
maintenance required from time to time. Why do we need machine learning?
The answer to this question is that for certain tasks it is easier to develop an algorithm that
learns/improves its performance with experience than to develop an algorithm manually. Although this
might seem unintuitive to the reader at this point, we will build intuition for this during this chapter.
Machine learning can be broadly classified as supervised learning , where training data with labels is
provided for the model to learn, and unsupervised learning , where the training data lacks labels. We also
have semi-supervised learning and reinforcement learning , but for now, we would limit our scope to
supervised machine learning. Supervised learning can again be classified into two areas: classification, for
discrete outcomes, and regression, for continuous outcomes.
Binary Classification
In order to further discuss the matter at hand, we need to be precise about some of the terms we have been
intuitively using, such as task, learning, experience, and improvement. We will start with the task of binary
classification.
Consider an abstract problem domain where we have data of the form

where x ∈ ℝn and y = ± 1.
We do not have access to all such data but only a subset S ∈ D. Using S, our task is to generate a
computational procedure that implements the function f : x → y such that we can use f to make predictions
over unseen data (xi, yi) ∉ S that are correct, f(xi) = yi. Let’s denote U ∈ D as the set of unseen data—that is,
(xi, yi) ∉ S and (xi, yi) ∈ U.
We measure performance over this task as the error over unseen data

We now have a precise definition of the task, which is to categorize data into one of two categories
(y = ± 1) based on some seen data S by generating f. We measure performance (and improvement in
performance) using the error E(f, D, U) over unseen data U. The size of the seen data |S| is the conceptual
equivalent of experience. In this context, we want to develop algorithms that generate such functions f
(which are commonly referred to as a model). In general, the field of machine learning studies the
development of such algorithms that produce models that make predictions over unseen data for such, and,
other formal tasks. (We introduce multiple such tasks later in the chapter.) Note that the x is commonly
referred to as the input/input variable and y is referred to as the output/output variable .
As with any other discipline in computer science, the computational characteristics of such algorithms
are an important facet; however, in addition to that, we also would like to have a model f that achieves a
lower error E(f, D, U) with as small a ∣S∣ as possible.
Let’s now relate this abstract but precise definition to a real-world problem so that our abstractions are
grounded. Suppose that an ecommerce website wants to customize its landing page for registered users to
show the products they might be interested in buying. The website has historical data on users and would
like to implement this as a feature to increase sales. Let’s now see how this real-world problem maps on to
the abstract problem of binary classification we described earlier.
The first thing that one might notice is that given a particular user and a particular product, one would
want to predict whether the user will buy the product. Since this is the value to be predicted, it maps on to
y = ± 1, where we will let the value of y = + 1 denote the prediction that the user will buy the product and
the value of y = − 1 denote the prediction that the user will not buy the product. Note that there is no
particular reason for picking these values; we could have swapped this (let y = + 1 denote the does not buy
case and y = − 1 denote the buy case), and there would be no difference. We just use y = ± 1 to denote the
two classes of interest to categorize data. Next, let’s assume that we can represent the attributes of the
product and the users buying and browsing history as x ∈ ℝn. This step is referred to as feature engineering
in machine learning and we will cover it later in the chapter. For now, it suffices to say that we are able to
generate such a mapping. Thus, we have historical data of what the users browsed and bought, attributes of
a product, and whether the user bought the product or not mapped on to {(x1, y1), (x2, y2), …(xn, yn)}. Now,
based on this data, we would like to generate a function or a model f : x → y, which we can use to determine
which products a particular user will buy, and use this to populate the landing page for users. We can
measure how well the model is doing on unseen data by populating the landing page for users, seeing
whether they buy the products or not, and evaluating the error E(f, D, U).
Regression
This section introduces another task: regression. Here, we have data of the form D = {(x1, y1), (x2, y2), …(xn,
yn)}, where x ∈ ℝn and y ∈ ℝ, and our task is to generate a computational procedure that implements the
function f : x → y. Note that instead of the prediction being a binary class label y = ± 1, like in binary
classification, we have real valued prediction. We measure performance over this task as the root-mean-
square error (RMSE) over unseen data

Note that the RMSE is simply taking the difference between the predicted and actual value, squaring it
so as to account for both positive and negative differences, taking the mean so as to aggregate over all the
unseen data, and, finally, taking the square root so as to counterbalance the square operation.

A real-world problem that corresponds to the abstract task of regression is to predict the credit score for an
individual based on their financial history, which can be used by a credit card company to extend the line of
credit.
Generalization
Let’s now cover what is the single most important intuition in machine leaning, which is that we want to
develop/generate models that have good performance over unseen data. In order to do that, first will we
introduce a toy data set for a regression task. Later, we will develop three different models using the same
dataset with varying levels of complexity and study how the results differ to understand intuitively the
concept of generalization.
In Listing 1-1, we generate the toy dataset by generating 100 values equidistantly between -1 and 1 as
the input variable (x). We generate the output variable (y) based on y = 2 + x + 2x2 + ϵ, where
is noise (random variation) from a normal distribution, with 0 being the mean and 0.1 being the standard
deviation. The code for this is presented in Listing 1-1, and the data is plotted in Figure 1-2. In order to
simulate seen and unseen data, we use the first 80 data points as seen data and treat the rest as unseen data.
That is, we build the model using only the first 80 data points and use the rest for evaluating the model.

#import packages
import matplotlib.pyplot as plt
import numpy as np

#Generate a toy dataset

x = np.linspace(-1,1,100)
signal = 2 + x + 2 * x * x
noise = numpy.random.normal(0, 0.1, 100)
y = signal + noise
plt.plot(signal,'b');
plt.plot(y,'g')
plt.plot(noise, 'r')
plt.xlabel("x")
plt.ylabel("y")
plt.legend(["Without Noise", "With Noise", "Noise"], loc = 2)
plt.show()

#Extract training from the toy dataset

x_train = x[0:80]
y_train = y[0:80]
print("Shape of x_train:",x_train.shape)
print("Shape of y_train:",y_train.shape)

Output[]
Shape of x_train: (80,)
Shape of y_train: (80,)
Listing 1-1 Generalization vs. Rote Learning
 Another Random Document on
Scribd Without Any Related Topics
that his account is hidden in the Proceedings of the Bath Natural History
and Antiquarian Field Club, vol. ii. no. 3, 1872.
[107] See N. C. vol. iv. p. 310.
[108] Chron. Petrib. 1088. “And syððon foron út of þam castele and
hergodon Baðon, and eall þæt land þær abutan.” Florence adds the
burning; “Rotbertus … congregato exercitu invasit Bathoniam, civitatem
regiam, eamque igne succendit.”
[109] Flor. Wig. 1088. “Illa [Bathonia] deprædata, transivit in
Wiltusciram, villasque depopulans, multorumque hominum strage facta,
tandem adiit Givelceastram, obsedit, et expugnare disposuit.”
[110] Geveltone, now Yeovilton, was held by one Ralph under William of
Eu (Domesday, 96 b). Givele, now Yeovil, was held by Count Robert
(Domesday, 93). All these names come in various corruptions from the
river Givel or Ivel, also called Yeo. Only in Yeovil we may trace a bit of
false etymology, which has also set the pattern to Yeovilton.
[111] I took with me to Ilchester a book by the Rev. W. Buckler,
“Ilchester Almshouse Deeds” (Yeovil, 1866), which contains the accounts
of Ilchester from Leland, Camden, and Stukeley, together with Stukeley’s
map. The last-named writer may have drawn somewhat on his
imagination; but I could trace the line of the walls, represented in a great
part of their course by modern buildings. Under the circumstances of the
site, the usual carfax is not to be found at Ilchester, any more than at
Godmanchester.
[112] Domesday, 86 a. “In Givelcestre sunt 107 burgenses, reddentes
xx. solidos. Mercatum cum suis appendiciis reddit xi. libras.”
[113] Flor. Wig. 1088. “Pugnant exterius spe capti prædæ et amore
victoriæ, repugnant intrinsecus acriter pro se suorumque salute. Tandem
inter utrumque necessitatis vicit causa; repulsus et tristis recedit Rotbertus
privatus victoria.” The Chronicle and William of Malmesbury do not speak
of Ilchester. William thus sums up the campaign; “Gaufridus episcopus,
cum nepote, Bathoniam et Bercheleiam partemque pagi Wiltensis
depopulans, manubias apud Bristou collocabat.”
[114] See N. C. vol. ii. p. 144.
[115] Chron. Petrib. 1088. “And eall Beorclea hyrnesse hi awæston.”
Florence more fully; “Willelmus de Owe Glawornensem invadit comitatum,
regiam villam deprædatur Beorchelaum, per totam ferro et flamma grande
perpetrat malum.”
 [116] See N. C. vol. ii. p. 557.
[117] See Domesday, 164. But it had already given a name to Roger and
Ralph of Berkeley; Domesday, 168. From Roger’s descendants it passed by
marriage to Robert the son of Harding. See N. C. vol. iv. p. 758.
[118]Domesday, 163. “In Nesse [Sharpness] sunt v. hidæ pertinentes
ad Berchelai quos W. comes misit extra ad faciendum unum castellulum.”
[119] Since I wrote the fourth volume of the Norman Conquest, there
has been much controversy about the origin of Robert Fitz-Harding. (See
Notes and Queries, Jan. 3rd, 1880.) I am confirmed on the whole in my
old belief that he was the son of Harding the son of Eadnoth.
[120] See N. C. vol. iv. pp. 590, 855.
[121] See above, p. 33.
[122] Chron. Petrib. 1088. “Þa men þe yldest wæron of Hereforde, and
eall þeo scír forþmid, and þa men of Scrobscyre mid mycele folce of
Brytlande.”
[123] See above, p. 33.
[124] Flor. Wig. 1088. “Cum hominibus comitis Rogerii de
Scrobbesbyria.” Yet the Chronicler says distinctly, “And Rogere eorl wæs
eac æt þam unræde.” That is, he joined in the conspiracy, but did not take
a personal share in the war.
[125] See above, p. 35, note 3.
[126]Flor. Wig. 1088. “Congregato magno Anglorum, Normannorum, et
Walensium exercitu.”
[127] See N. C. vol. ii. p. 395.
[128] Ib. vol. i. p. 520.
[129] Chron. Petrib. 1088. “Þa men … comon and hergodon and
þærndon on Wiðreceastrescire forð, and hi comon to þam porte sylfan,
and woldon þa þæne port bærnen, and þæt mynster reafian, and þæs
cynges castel gewinnan heom to handa.” Florence adds, “grandem de
regis incolis fidelibus sumpturos vindictam.” On the deliverance of
Worcester, see Appendix D.
[130] Florence brings in his own Bishop with a panegyric; “Vir magnæ
pietatis et columbinæ simplicitatis, Deo populoque quern regebat in
omnibus amabilis, regi, ut terreno domino, per omnia fidelis, pater
reverendus Wlstanus.” In the Chronicle he is simply “se arwurða bisceop
Wlfstan.” He goes on to make his exhortation after the manner of Moses.
[131] See N. C. vol. iii. p. 61.
[132] Ib. vol. iv. p. 579.
[133] See N. C. vol. iv. p. 174.
[134] See N. C. vol. iv. p. 379.
[135] Ib.
[136] Flor. Wig. 1088. “Normanni interim, ineuntes consilium, rogant
ipsum episcopum ut ab ecclesia transiret in castellam, tutiores se
affirmantes de ejus præsentia, si majus incumberet periculum; diligebant
enim eum valde. Ipse enim, ut erat miræ mansuetudinis, et pro regis
fidelitate, et pro eorum dilectione, petitioni eorum adquievit.”
[137] See N. C. vol. iv. p. 174.
[138] Flor. Wig. u. s. “Interea audenter in arma se parat episcopalis
familia.” On the nature of this “familia,” see N. C. vol. v. p. 496.
[139]Ib. “Inter quos [hostes] magna belli jam fervebat insania;
contumaciter enim episcopi contemnentes mandata, in terram ipsius
posuerunt incendia.” On the order of events, see Appendix D.
[140] Ib. “Conveniunt castellani et omnis civium turma, occurrere se
affirmant hostibus ex altera parte Sabrinæ fluminis, si hoc eis pontificis
annueret licentia. Parati igitur et armis instructi, ipsum ad castellum
euntem habent obviam, quam optabant requirunt licentiam; quibus
libentur annuens, ‘Ite,’ inquit, ‘filii, ite in pace, ite securi, cum Dei et nostra
benedictione.’ Confidens ego in Domino, spondeo vobis, non hodie nocebit
vobis gladius, non quicquam infortunii, non quisquam adversarius. State in
regis fidelitate, viriliter agentes pro populi urbisque salute.”
[141]Ib. “Episcopus ingenti concutitur dolore, videns debilitari res
ecclesiæ, acceptoque inde consilio, gravi eos, ab omnibus qui
circumaderant coactus, percussit anathemate.” See Appendix D.
[142] Ib. “Alacres pontem reparatum transeunt, hostes de longinquo
accelerantes conspiciunt.”
[143] See Appendix D.
[144] Flor. Wig. u. s. “Cæduntur pedites, capiuntur milites, cum
Normannis tam Angli quam Walenses, cæteris vero vix debili elapsis fuga
[were the ‘milites’ spared for the sake of ransom?] regis fideles cum
pontificis familia, exultantes in gaudio, sine ulla diminutione suorum,
redeunt ad propria; gratias Deo referunt de rerum ecclesiæ incolumitate,
gratias episcopo referunt de consilii ejus salubritate.”
[145] See N. C. vol. iv. p. 386.
[146] Chron. Petrib, 1088. “Þe wæs ærur heafod to þam unræde.”
[147] See above, p. 29.
[148] Chron. Petrib. 1088. “Ðe bisceop Odo, þe þas cyng of awocan,
ferde into Cent to his earldome and fordyde hit swyðe, and þæs cynges
land and þæs arcebisceopes mid ealle aweston, and brohte eall þæt gód
into his castele on Hrofeceastre.” This follows at once on the accounts of
Roger the Bigod and Hugh of Grantmesnil. So William of Malmesbury, who
here brings in the story of Lanfranc’s share in Odo’s imprisonment in 1082,
in order to account for Odo’s special hatred towards the Archbishop.
[149] See N. C. vol. i. pp. 267, 296. On the early history of Rochester
generally, see Mr. Hartshorne’s paper in the Archæological Journal,
September, 1863.
[150] This is brought out by Orderic, 667 B; “Oppidum igitur Rovecestræ
sollicita elegerunt provisione, quoniam, si rex eos non obsedisset in urbe,
in medio positi laxis habenis Lundoniam et Cantuariam devastarent, et per
mare, quod proximum est, insulasque vicinas, pro auxiliis conducendis
nuntios cito dirigerent.” The islands must be Sheppey and Thanet.
[151] See the siege of Rochester in 1215 and his defence by William of
Albini in Roger of Wendover, iii. 333.
[152] For the siege of 1264 see W. Rishanger, Chron. p. 25 (Camd. Soc.).
On Simon’s military engines he remarks that the Earl “exemplum
relinquens Anglicis qualiter circa castrorum assultationes agendum sit, qui
penitus hujusmodi diebus illis fuerant ignari.” A forerunner of Kanarês, he
had a fire-ship in the river; he also used mines, as the Conqueror had
done at Exeter.
[153] Mr. Hartshorne showed distinctly that the present tower of
Rochester was not built by Gundulf, but by William of Corbeuil. See the
passages which he quotes from Gervase, X Scriptt. 1664, and the
continuator of Florence, 1126. But we have seen (see N. C. vol. iv. p. 366)
that Gundulf did build a stone castle at Rochester for William Rufus
(“castrum Hrofense lapidum”), and we should most naturally look for it on
the site of the later one. On the other hand, there is a tower, seemingly of
Gundulf’s building and of a military rather than an ecclesiastical look,
which is now almost swallowed up between the transepts of the cathedral.
But it would be strange if a tower built for the King stood in the middle of
the monastic precinct.
[154] The odd position of the cloister at Rochester suggests the notion
that Gundulf’s church occupied only the site of the present eastern limb,
and that the later Norman nave was an enlargement rather than a
rebuilding.
[155] Domesday, 2 b. “Episcopus de Rouecestre pro excambio terræ in
qua castellum sedet, tantum de hac terra tenet quod xvii.s. et iv.d. valet.”
This is said of land at Aylesford; but the castle spoken of must surely be
that of Rochester. The Domesday phrase “sedet” seems beautifully to
describe either the massive square donjon or the shell-keep on the
mound; yet it may be doubted whether Rochester had either in the
Conqueror’s day.
[156] This ditch is said to have been traced right across the middle of
the cathedral, with the twelfth-century nave to the west of it. I can say
nothing either way from my own observation; but such an extension of
the church to the west would exactly answer to the extension of the
churches of Le Mans and Lincoln to the east. In both those cases the
Roman wall had to give way.
[157] See N. C. vol. iv. p. 367.
[158] Ord. Vit. 667 A. “Tunc Odo Bajocensis cum quingentis militibus
intra Rofensem urbem se conclusit, ibique Robertum ducem cum suis
auxiliaribus secundum statuta quæ pepigerant præstolari proposuit.” The
last clause of course implies the supposed earlier agreement with Duke
Robert, on which see above, p. 25, and Appendix B.
[159] Flor. Wig. 1088. “Rumore autem percussus insolito, comes exultat,
amicis nunciat, quasi jam de victoria securus triumphat, plures ad prædam
incitat; Odoni episcopo, patruo suo, auxiliarios in Angliam legat, se
quantocius, congregato majori exercitu, secuturum affirmat.”
[160] Ib. “Prædictus episcopus Baiocensis, munita Roveceastra, misit
Normanniam, exhortans comitem Rotbertum cito venire in Angliam,
nuntians ei rem gestam, affirmans paratum sibi regnum, et si sibi non
desisteret, paratam et coronam.”
[161] Ib. “Missi a comite Rotberto venerunt in Angliam, ab Odone
episcopo ad custodiendum receperunt Roveceastram; et horum ut
primates Eustatius junior, comes Bononiæ, et Rotbertus de Beleasmo
gerebant curam.” Here we have (see Appendix B) the true moment of
their coming. From this point we may accept the account in Orderic (667
B); “Prædictum oppidum Odo præsul et Eustachius comes atque Robertus
Bellesmensis, cum multis nobilibus viris et mediocribus, tenebant,
auxiliumque Roberti ducis, qui desidia mollitieque detinebatur, frustra
exspectabant.” We meet them again in 765 B.
[162] “Eustatius junior,” “Eustatius þe iunga.” See N. C. vol. iv. p. 745.
[163] They are mentioned in the Chronicle along with the incidental
mention of Eustace; “Innan þam castele wæron swiðe gode cnihtas,
Eustatius þe iunga, and Rogeres eorles þreo sunan, and ealle þa
betstboren men þe wæron innan þisan lande oððe on Normandige.” This
is followed by William of Malmesbury (iv. 306); “Erat tunc apud
Roveceastram omnis pene juventutis ex Anglia et Normannia nobilitas;
tres filii Rogerii comitis, et Eustachius Bononiæ junior, multique alii quos
infra curam nostram existimo.”
[164] The three sons of Earl Roger can hardly fail to be his three eldest
sons (see Will. Gem. vii. 16; Ord. Vit. 708 D), Robert, Hugh, and Roger, all
of whom figure in our story. Arnulf does not appear in English history till
later, and Philip the clerk does not appear at all. Geoffrey Gaimar (Chron.
Ang. Norm. i. 35), after setting forth the possessions of Robert of Bellême,
mentions the other three; but one does not exactly see why he says,
“Le conte Ernulf ert le quarte frère, Par cors valeit un emperère.”

Cf. Ord. Vit. 708 D, 808 C.

[165] See N. C. vol. iv. p. 488.
[166] See above, p. 33.
[167]Flor. Wig. 1088. “Rogerus fautor Rotberti erat in castello suo
Arundello, comitis prædicti opperiens adventum.”
[168] See N. C. iv. 66, v. 808.
[169] See Tierney’s History of Arundel, i. 43.
[170] Domesday, 23 “Modo inter burgum et portum aquæ et
consuetudinem navium reddit xii. libras et tamen valet xiii. libras. De his
habet S Nicolaus xxiiii. solidos.” “Clerici sancti Nicolai” are mentioned again
in the next column. The church then was secular in 1086; but the clerks
must have soon given way to the priory of Saint Nicolas, founded by Earl
Roger himself as a cell to his abbey at Seez; in 1386 it gave way to the
college of Arundel.
[171] See N. C. iv. p. 501.
[172] Domesday, 23. “Modo est ipsa civitas in manu comitis Rogerii.”
Here he had one quarter of a Roman chester, while the Bishop had
another; yet there were sixty houses more than there had been T. R. E.
[173]See the customs of Lewes and the rights of William of Warren in
Domesday, 26. The toll on selling a man was threepence. The two
mounds of the castle, the smaller known as Brack Mount, are rare,
perhaps unique. The inner gateway seems to be of Earl William’s building.
[174] I suspect that the original title of the Earls of Arundel was Earl of
Sussex, and that the name of the castle came to be used, much as the
successors of William of Warren, strictly Earls of Surrey, are more
commonly called Earls Warren. See more in Tierney’s History of Arundel.
[175] Lucan, iv. 819.
[176] See N. C. vol. iii. p. 161.
[177]Ord. Vit. 666 D. “Rex Guillelmus, ut vidit suos in terra sua contra
se pessima cogitare, et per singula crebrescentibus malis ad pejora
procedere; non meditatus est ut timida vulpes ad tenebrosas cavernas
fugere, sed ut leo fortis et audax rebellium conatus terribiliter
comprimere.”
[178]Will. Malms. iv. 306. “Nec minori astutia Rogerium de Monte
Gomerico, secum dissimulata perfidia equitantem, circumvenit.”
[179] Ib. “Seorsum enim ducto magnam ingessit invidiam; dicens,
Libenter se imperio cessurum, si illi et aliis videatur quos pater tutores
reliquerat. Non se intelligere quid ita effrænes sint: si velint, pecunias
accipiant pro libito; si augmentum patrimoniorum, eodem modo; prorsus,
quæ velint, habeant. Tantum videant ne judicium genitoris periclitetur:
quod si de se putaverint aspernandum, de se ipsis caveant exemplum;
idem enim se regem, qui illos duces fecerit. His verbis comes et
pollicitationibus incensus, qui primus factionis post Odonem signifer fuit,
primus defecit.” Roger of Wendover (ii. 33) adds the words “pœnitentia
ductus.”
[180] Orderic a little later (667 B) says, “Rogerus Merciorum comes,
multique Normannorum, qui cum rege foris obsidebant, clam adminiculari
quantum poterant inclusis satagebant.”
 [181] Orderic (680 C) puts the creation of this earldom somewhat later,
at the Gemót held just before the invasion of Normandy in 1090. He adds
that the new earl died soon after (“quem paulo post mors nulli parcens e
medio rapuit”), and records his burial at Lewes, and adds his epitaph.
There is no better authority than that of the Hyde writer (298) for placing
the creation at this time or for placing the Earl’s death a little later (see
below, p. 76). But his narrative is so minute that one would think that he
must have had some kind of ground for it. His words are; “Rex Willelmus
… videns igitur principes regni nutantes et exercitum a se dilabi, sapienti
usus consilio, Willelmum de Warennia, virum bellicosum, animo ferum et
corpore strenuum famaque præclarum, in amicitia Asarum [what this may
mean I have no notion, but the editor vouches that such is the reading of
the MS.] comitis honore sublimat, multa impendit multaque promittit.”
[182] See N. C. vol. ii. p. 251.
[183] Ord. Vit. 667 C. “Omnes episcopi Angliæ cum Anglis sine dolo
regem juvabant, et pro serena patriæ pace, quæ bonis semper amabiles
est, laborabant.”
[184] The appeal to the English is strongly marked in the Chronicle; “Ða
þe cyng undergeat ealle þas þing and hwilcne swicdom hi dydon toweard
his, þa wearð he on his mode swiðe gedrefed. Sende þa æfter Englisce
mannan, and heom fore sæde his neode and gyrnde heora fultumes.”
Simeon of Durham gives a free translation quite independent of Florence;
“Hoc audito, rex fecit convocare Anglos, et ostendit eis traditionem
Normannorum, et rogavit ut sibi auxilio essent.” But the appeal comes out
no less strongly in Orderic (666 D); “Lanfrancum archiepiscopum cum
suffraganeis præsulibus, et comites, Anglosque naturales convocavit, et
conatus adversariorum, ac velle suum expugnandi eos indicavit.” The writ
comes from William of Malmesbury, iv. 306; “Ille, videns Normannos pene
omnes in una rabie conspiratos, Anglos probos et fortes viros, qui adhuc
residui erant, invitatoriis scriptis accersiit.” It is singular that Florence
mentions the English only in an incidental way a little later; “Congregato
quantum ad præsens poterat Normannorum, sed tamen maxime
Anglorum, equestri et pedestri, licet mediocri, exercitu.” Does the precious
document spoken of by William of Malmesbury still lurk in any manuscript
store?
[185] Chron. Petrib. “And behet heom þa betsta laga þe æfre ær wæs
on þisan lande, and ælc unriht geold he forbead, and geatte mannan
heora wudas and slǽtinge.” William of Malmesbury (iv. 306) translates,
“Bonas leges et tributorum levamen, liberasque venationes pollicens.”
Florence is less literal; “Statuens leges, promittens fautoribus omnia
bona.” Simeon gives another version; “Eo tenore, ut si in hac necessitate
sibi fideles existerent, meliorem legem quam vellent eligere eis
concederet, et omnem injustum scottum interdixit, et concessit omnibus
silvas suas et venationem. Sed quicquid promisit, parvo tempore
custodivit. Angli tamen fideliter eum juvabant.”
[186] Flor. Wig. 1088. “Jure regio, militari, ut impiger, fretus audacia,
mittit legatos, vocat quos sibi credit fidos, vadit Lundoniam, belli
tractaturus negotia, expeditionis provisum, necessaria.”
[187] See above, p. 29.
Chron. Petrib. 1088. “Ac Englisce men swa þeah fengon to þam
[188]
cynge heora hlaforde on fultume.” The numbers come from Orderic
(667A); “Anglorum triginta millia tunc ad servitium regis sponte sua
convenerunt.”
[189]Ord. Vit. 667 A. “Passim per totum Albionem impera, omnesque
rebelles deice regali justitia.”
[190] Ib. “Viriliter age, ut regis filius et legitime ad regnum assumptus;
securus in hoc regno dominare omnibus.”
Ord. Vit. 667 A. “Solerter Anglorum rimare historias, inveniesque
[191]
semper fidos principibus suis Angligenas.” Fancy William Rufus sitting
down to study the Chronicles, as his brother Henry may likely enough
have done.
[192] Chron. Petrib. 1088. “Ferdon þa toweard Hrofeceastre and woldon
þone bisceop Odan begytan, þohtan gif hi hæfdon hine, þe wæs ærur
heafod to þam unræde, þæt hi mihton þe bet begytan ealla þa oðre.”
[193] It is somewhat singular that, though Richard appears in Domesday
as “Ricardus de Tonebrige” as well as “Ricardus filius Gisleberti comitis”
(14 et al.), and though his “leva” or “lowy” (see Ellis, i. 212) is often
spoken of, yet Tunbridge castle itself is not entered. See on Richard of
Bienfaite, Clare, or Tunbridge, N. C. vol. ii. p. 196; iv. 579. A singular story
is told in the Continuation of William of Jumièges (viii. 15), how Tunbridge
was granted in exchange for Brionne, and measured by the rope. See
Appendix S.
[194] At Tunbridge the mound and the gateway stand side by side, as
indeed they do, though less conspicuously, at Arundel and Lewes. A wall is
built from the gateway to the keep on the mound, losing itself, as it were,
in the side of the mound. The mound thus stands half within and half
without the enclosure formed by the gateway.
 [195] Chron. Petrib. 1088. “Þa Englisce men ferdon and tobræcon þone
castel, and þa men þe þærinne wæron griðodon wið þone cyng,” So
Simeon of Durham; “Sed viriliter Angli insilientes in illud, destruxerunt
totum castrum, et qui intus erant in manus regi dederunt.” Florence gives
some further details; “Tunebrycgiam cui præerat Gilebertus filius Ricardi,
contrarium sibi invenit: obsedit, in biduo expugnavit, vulneratum
Gilebertum cum castello ad deditionem coegit.” Is it possible that,
according to Orderic’s second account of the rebellion (765 A, B), we are
still only in the Easter week?
[196] See N. C. vol. iv. p. 366. While I am revising my text, an account
of this tower by Mr. Clark has appeared in the Builder, November 27, 1880.
[197] Chron. Petrib. 1088. “Se cyng mid his here ferde toweard
Hrofeceastre, and wendon þæt se bisceop wære þærinne, ac hit wearð
þam cynge cuð þæt se bisceop wæs afaren to þam castele on Pefenesea.”
Florence helps us to an hexameter in the middle of his prose; “Relatum
erat ei ibi esse episcopum Odonem cum omnibus suis et cohortem
ultramarinam….
Fama volans dicti pervenit Odonis ad aures,
et cum sociis inito consilio, relinquens Roveceastram, cum paucis adiit
castrum fratris sui Roberti Moritanensis comitis quod Pevenessa dicitur.”
Are the “cohors ultramarina” those who had come with Eustace and
Robert of Bellême?
[198] Flor. Wig. 1088. “Fratrem reperiens, cum ut se teneat hortatur,
pollicens se securos ibi posse esse, et dum rex ad expugnandam
Roveceastram intenderet, comitem Normanniæ cum magno exercitu
venturum, seque suosque liberaturum et magna fautoribus suis dando
præmia regnum accepturum.”
Ord. Vit. 666 D. “Statuerat præcursores suos vere redeunte sequi
[199]
cum multis legionibus militum.”
[200] Cont. Will. Gem. viii. 2. “Quum sui fideles eum exhortarentur ut
regnum Angliæ sibi a fratre præreptum velocius armis sibimet restitueret,
simplicitate solita et, ut ita dicam, imprudentiæ proxima, respondisse
fertur, ‘Per angelos Dei [Gregory’s pun in another form], si essem in
Alexandria, exspectarent me Angli, nec ante adventum meum Regem sibi
facere auderent. Ipse etiam Willelmus frater meus, quod eum
præsumpsisse dicitur, pro capite suo sine mea permissione minime
attentaret.’”
 [201] Chron. Petrib. 1088. “Betwyx þissum se eorl of Normandige
Rodbeard, þes cynges broðer, gaderode swiðe mycel folc, and þohte to
gewinnane Englelande mid þæra manna fultume þe wæron innan þisan
lande ongean þone cyng, and he sende of his mannan to þisum lande,
and wolde cuman himsylf æfter.”
[202] Florence seems here to translate what the Chronicler had said a
little before (see above, p. 67); “Inito itaque salubri consilio, illum eo
usque cum exercitu persequitur, sperans se belli citius finem
assequuturum, si ante triumphare posset de principibus malorum
prædictorum.”
[203] So I find it called in several papers in the Sussex Archæological
Collections. But the local antiquaries seem hardly to have fully grasped the
fact that there is a town in Normandy called Laigle, and that the family
with which we are concerned took its name from it.
[204] Chron. Petrib. 1088. “And se cyng mid his here ferde æfter, and
besætt þone castel abutan mid swiðe mycele here fulle six wucan.” The
artillery comes from Florence; “Accelerat, machinas parat, patruum
utrumque obsidet; locus erat munitissimus; ad expugnationem indies
laborat.” William of Malmesbury cuts the siege of Pevensey short, and
Orderic leaves it out altogether.
[205] See Appendix E.
[206] See N. C. vol. iii. p. 395.
[207] Liber de Hyda, 299. “Willelmus de Warennia apud obsidionem
Peveneselli sagitta in crure valde vulneratus, Leuwias cum omnium
mœrore deportatus est.” The writer goes on to describe Earl William’s last
testament and death. It will be remembered (see above, p. 62) that
Orderic makes William of Warren die quietly at a later time; but, small as
is the authority of the Hyde writer, it is strange if he altogether invented or
dreamed this minute account.
[208] Chron. Petrib. 1088. “Syððan heom ateorede mete wiðinnan þam
castele, þa gyrndon hi griðas, and agefan hine þam cynge, and se bisceop
swór þæt he wolde út of Englelande faran, and ná mare cuman on þisan
lande butan se cyng him æfter sende, and þæt he wolde agyfan þone
castel on Hrofeceastre.” So William of Malmesbury (iv. 306); “Captum ad
quod libuit jusjurandum impulit, ut Anglia decederet et Rovecestram
traderet.”
[209] Chron. u. s. “Ealswa se bisceop ferde and sceolde agifan þone
castel and se cyng sende his men mid him.” So Will. Malms. “Ad quod
implendum eum cum fidelibus suis præmisit, lento pede præeuntes
subsecutus…. Regii cum episcopo pauci et inermes (quis enim eo
præsente insidias timeret?) circa muros desiliunt, clamantes oppidanis ut
portas aperiant; hoc episcopum præsentem velle, hoc regem absentem
jubere.”
[210] Will. Malms. u. s. “At illi, de muro conspicati quod vultus episcopi
cum verbis oratorum non conveniret, raptim apertis portis ruunt, equos
involant, omnesque cum episcopo vinctos abducunt.” This explains the
shorter account in the Chronicle; “þa arisan þa men þe wæron innan þam
castele, and namon þone bisceop and þes cynges men, and dydon hi on
hæftmenge.” It is now that both the Chronicle and William give the names
of the chief nobles who were in the castle. Henry of Huntingdon (1088, p.
215) strongly marks Odo’s treachery; “Eustachius consul et cæteri
proceres qui urbi inerant, fallacia ipsius, episcopum regisque ministros
ceperunt et in carcerem retruserunt.”
[211] See N. C. vol. ii. p. 104.
[212] Will. Malms, iv. 306. “Ille [rex]…. Anglos suos appellat; jubet ut
compatriotas advocent ad obsidionem venire, nisi si qui velint sub nomine
Niðing, quod nequam sonat, remanere. Angli, qui nihil miserius putarent
quam hujusce vocabuli dedecore aduri, catervatim ad regem confluunt, et
invincibilem exercitum faciunt.” This leaves out the fact that the
proclamation was addressed both to French and English. The words of the
Chronicle are express; “Ða se cyng undergeat þat þing, þa ferde he æfter
mid þam here þe he þær hæfde, and sende ofer eall Englalande, and
bead þæt ælc man þe wære unniðing sceolde cuman to him, Frencisce
and Englisce, of porte and of uppelande.” We can hardly doubt that we
have here the actual words of the proclamation. It must not be forgotten
that, by the law of the Conqueror, Frenchmen who had settled in King
Eadward’s day were counted as English. See N. C. vol. iv. p. 620.
[213] Ord. Vit. 667 B. “Animosus rex … oppidum Maio mense cum grandi
exercitu potenter obsedit, firmatisque duobus castellis omnem exeundi
facultatem hostibus abstulit.” It must have been late in May, as six weeks
had been spent before Pevensey. Indeed, if the siege did begin in the
Easter week, it must have been June.
[214] See Mr. Clark in the Archæological Journal, vol. xxxii. p. 205.
[215] This appears from the words of Florence; “Hrofenses
Cantwariensibus et Lundoniensibus cædes inferunt et incendia.
Landfrancus enim archiepiscopus et pene omnes optimates ejusdem
provinciæ erant cum rege.” Orderic too (u. s.) points out the
advantageous position of Rochester for such purposes; “In medio positi
laxis habenis Lundoniam et Cantuariam devastarent.”
[216] See N. C. vol. v. p. 748.
[217] Ord. Vit. 667 C. “In oppido Rofensi plaga similis Ægyptiorum
plagæ apparuit, qua Deus, qui semper res humanas curat et juste
disponit, antiqua miracula nostris etiam temporibus recentia ostendit.”
Nobody could eat, unless his neighbour drove away the flies; so they
wielded the flapper by turns.
[218] See above, p. 62.
[219] Will. Malms. iv. 306. “Nec diutius potuere pati oppidani quin se
traderent, experti quamlibet nobilem, quamlibet consertam manum, nihil
adversus regem Angliæ posse proficere.”
[220] Ord. Vit. 667 D. “Guillermum regem nuntiis petierunt ut pacem
cum eis faceret, ac oppidum ab eis reciperet, tali tenore ut terras, fundos,
et omnia quæ hactenus habuerant, ab ipso reciperent, et ipsi eidem ut
naturali domino [cynehlaford] fideliter amodo servirent.”
[221] Ord. Vit. 667 D. “His auditis rex iratus est, et valde rigidus
intumuit, et in nullo flexus legatorum postulationibus non acquievit; sed
perfidos traditores in oppido virtute potenti capiendos juravit, et mox
patibulis suspendendos, et aliis mortium diversis generibus de terra
delendos asseruit.”
[222] Ib. “Ecce turgidi juvenes et cupiditate cæcati senes jam satis
edocti sunt quod regiæ vires in hac insula nondum defecerunt. Nam qui
de Normannia, tamquam milvi ad prædam, super nos cum impetu
advolarunt, et in Anglia regiam stirpem defecisse arbitrati sunt, jam
Guillelmum juvenem Guillelmo sene non debiliorem, cohibente Deo,
experti sunt.”
[223] Ord. Vit. 668 B. “Quid sceleratis peccavi? quid illis nocui? quid
mortem meam totis nisibus procuraverunt, et omnes pro posse suo contra
me populos cum detrimento multorum erexerunt?”
[224] Ib. “Quisquis parcit perjuris et latronibus, plagiariis et execratis
proditoribus, aufert pacem et quietem innocentibus, innumerasque cædes
et damna serit bonis et inermibus.” We seem to be reading the cover of
the Edinburgh Review.
[225] Ord. Vit. 668 C. “Baiocensis Odo patruus tuus est et pontificali
sanctificatione præditus est.” “Cum patre tuo Anglos subjugavit”—​a merit
which would hardly be pleaded in the hearing of the King’s army. He is
“antistes Domini,” and so forth. “Omnes precamur ut illi benevolentiam
tuam concedas et illæsum in Normanniam ad diocesim suam abire
permittas.”
[226]Ib. “Comes Boloniensis patri tuo satis fuit fidelis, et in rebus arduis
strenuus adjutor et contubernalis.” There must be some confusion
between father and son.
[227] Ib. “Magnam Normanniæ partem possidet, fortissimisque castellis
corroboratus pene omnibus vicinis suis et Neustriæ proceribus
præeminet.”
[228] Here (ib. D) a hexameter peeps out;
“Idem qui lædit, fors post ut amicus obedit.”
It is the doctrine of Aias in Sophoklês (659);
ἐγὼ δ’ ἐπίσταμαι γὰρ ἀρτίως, ὅτι
ὅ τ’ ἐχθρὸς ἡμῖν ἐς τοσόνδ’ ἐχθαρτέος,
ὡς καὶ φιλήσων αὖθις.
The balancing clause was not called for.
[229] They were (ib.) “eximii tirones”—​“swiðe gode cnihtas”—​“quorum
servitutem, inclite rex, parvi pendere non debes.”
[230] Ib. “Igitur, quos jam superasti potestate, divitiis, et ingenti
probitate, subjuga tibi magnificentia et pietate.” On the sense of
“magnificentia,” cf. N. C. vol. i. p. 261.
[231] Ord. Vit. 668 D. “Omnem spem habendi hæreditates et terras in
regno ejus, quamdiu ipse regnaret, funditus abscidit.”
[232] Ord. Vit. 668 D. “Tunc Odo pontifex a rege Rufo impetrare
temptavit, ne tubicines in eorum egressu tubis canerent, sicut moris est
dum hostes vincuntur et parvum oppidum capitur.” Why “parvum”?
[233] Ib. “Nec se concessurum etiam propter mille auri marcos palam
asseruit.”
[234] Ib. “Oppidanis cum mœrore et verecundia egredientibus, et
regalibus tubis cum gratulatione clangentibus.”
[235] Ord. Vit. 669 A. “Multitudo Anglorum quæ regi adhærebat cunctis
audientibus, vociferabatur, et dicebat; Torques, torques afferte, traditorem
episcopum cum suis complicibus patibulis suspendite. Magne rex
Anglorum, cur sospitem pateris abire incentorem malorum? Non debet
vivere perjurus homicida, qui dolis et crudelitatibus peremit hominum
multa milia.”
[236] Ib. “Hæc et alia probra mœstus antistes cum suis audivit.”
[237] Chron. Petrib. 1088. “Se bisceop Odo mid þam mannum þe innan
þam castele wæron ofer sæ ferdon, and se bisceop swa forlet þone
wurðscipe þe he on þis land hæfde.” Orderic (669 A)—​in his character of
“Angligena”—​moralizes; “Sic irreligiosus præsul de Anglia expulsus est, et
amplissimis possessionibus spoliatus est. Tunc maximos quæstus, quos
cum facinore obtinuit, justo Dei judicio cum ingenti dedecore perdidit, et
confusus Baiocas rediit, nec in Angliam postmodum repedavit.”
[238] Ord. Vit. 669 A. “Anno primo Guillelmi Rufi regis, in initio æstatis,
Rofensis urbs ei redita est, omniumque qui contra pacem enses
acceperant, nequam commotio compressa est.” We shall see by the story
of Robert of Rhuddlan, to which we shall presently come, that some of the
King’s followers were at home again by the end of June.
[239] See above, p. 74.
[240] Chron. Petrib. 1088. “Eac manige Frencisce men forleton heora
land and ferdon ofer sæ, and se cyng geaf heora land þam mannum þe
him holde wæron.”
[241] Ord. Vit. 669 B. “Quorumdam factiones sævissimis legibus puniit,
aliquorum vero reatus ex industria dissimulavit. Antiquis baronibus, quos
ab ipso aliquantum desciverat nequitia, versute pepercit, pro amore patris
sui cui diu fideliter inhæserant, et pro senectutis reverentia, sciens
profecto quod non eos diu vigere sinerent morbi et mors propria. Porro
quidam, quanto gravius se errasse in regiam majestatem noverunt, tanto
ferventius omni tempore postmodum ei famulati sunt, et tam muneribus
quam servitiis ac adulationibus multis modis placere studuerunt.”
[242] See above, p. 32.
[243] See above, p. 28.
[244] See above, p. 88.
[245] See N. C. vol. iv. pp. 409, 825, and below, p. 139.
[246] Mon. Ang. i. 245. “Tandem misi sibi rex abbatem sancti Augustini,
mandans ei ut, sicut prius mandaverat sibi, ad curiam suam cum abbate
veniret. Episcopus autem, inimicorum suorum insidias cum regis ira
metuens, sine bono conductu se non posse venire respondet et legatos
suos per abbatis conductum cum subscriptis litteris regi misit.”
[247] Ib. “Homines meos et terras et pecuniam quam vicecomites vestri
ubicumque poterant, mihi abstulerunt, scilicet Offedene et Welletune quas
diviserunt Odoni et Alano comitibus, cum cæteris terris in Ewerwickschire.”
See above, p. 31. On Count Alan, see N. C. vol. iv. p. 294, and on Odo,
vol. iv. pp. 301, 805.
[248] Ib. “Quod breve cum mississem Radulfo Paganello non solum mihi
pacem negavit sed et de parte vestra me diffidavit.” On diffidatio see
Ducange in voce. In N. C. vol. v. p. 270 we have a case of the man
defying his lord. Here the lord defies his man. In either case there is the
withdrawal of one side of the mutual duty of lord and man.
[249] Ib. “Hominum vero quosdam vendidit, quosdam redimi permisit.”
[250] Mon. Ang. i. 245. “Hoc in veritate vobis mando quod libenter cum
hoc abbate venissem, nisi plus inimicos meos et indoctam populi
multitudinem timuissem quod de vestro brevi et baronum vestrorum
fiducia dubitassem.”
[251]Ib. “Rex visis his litteris misit conductum episcopo et bene affidavit
eum per litteras suas quod per eum vel per suos homines nullum ei
damnum eveniret usque quo de rege rediens Dunelmum intraret. Perrexit
ergo episcopus ad regem.”
[252] Mon. Ang. i. 245. “Episcopus … deprecatus est eum ut
rectitudinem sibi consentiret sicut episcopo suo. Rex autem respondit ei,
Quod si laicaliter placitare vellet, et extra pacem quam rex ei dederat se
mitteret, hoc modo rectitudinem sibi consentiret, et, si hoc modo placitare
recusaret, Dunelmum faceret eum reconduci.”
[253] Ib. “Dunelmum rediit episcopus, cui rex interim plus quam
septingentos homines cum multa præda abstulerat.”
[254] They were to have (Mon. Ang. i. 246) the “securitas et conductus
regis” till they had crossed—​“donec ultra mare ad terram siccam cum
rebus suis essent.” The catalogue of the “res suæ” is curious; “Et liceret
eos per conductum regis secum ducere et portare [ἄγειν καὶ φέρειν]
aurum et argentum, equos et pannos et arma et canes et accipitres, et
sua prorsus omnia quæ de terra portari debent.” The hawks and hounds
remind us of Harold setting sail from Bosham in the Tapestry. See N. C.
vol. iii. p. 222.
 [255] Mon. Ang. i. 246. “Episcopus dedit fidem suam Rogero Pictavensi,
quod si ipse per præscriptam condicionem castellum reduceretur, et major
fortitudo in castello missa vel facta esset in hominibus vel in munitione vel
in castelli fortitudine quam eadem die ibi erat, episcopus totum illud
destrui faceret, ita quod episcopus inde nullum proficuum haberet nec rex
damnum.”
[256] Mon. Angl. i. 246. “In quarto nonas Novembris … venit episcopus
Salisbiriam, quem cum Ursus de Habetot unus ex servientibus regis ad
regem intrare moneret.” On Urse of Abetot, see N. C. vol. iv. pp. 173, 383,
579, 820.
[257] Ib. “Episcopus reqnisivit ab archiepiscopis utrum revestitus ingredi
deberet, dixitque, ‘Nihil se prorsus acturum ibi nisi canonice et secundum
ordinem suum et sibi videbatur quod ecclesiastica consuetudo exigebat ut
ipse revestitus ante revestitos causam suam diceret et causantibus
canonice responderet,’ Cui Lanfrancus archiepiscopus respondens, ‘bene
possumus,’ inquit, ‘hoc modo vestiti de regalibus tuisque negotiis
disceptare, vestes enim non impediunt veritatem,’”
[258] See William FitzStephen, iii. 56, Robertson.
[259] Mon. Angl. u. s. “Episcopus surgens precatus est regem ut
episcopatum suum quem jamdiu sine judicio abstulerat sibi redderet.
Lanfrancus vero, rege tacente, dixit, ‘Rex de episcopatu tuo nihil tibi
abstulit vel aliquis per eam neque breve suum vidisti per quod te de
episcopatu tuo dissaisiret vel dissaisiri præciperet.’”
[260] The Bishop now tells his grievances at length. After other wrongs
the King “misit comites et barones cum exercitu suo, et per eos totum
episcopatum meum vastavit, terras quoque et homines et pecuniam Sancti
Cuthberti et meam mihi abstulit. Nostram etiam sedem me ad tempus
abjuvare coegit; ipsi etiam casati ecclesiæ qui mei homines ligii fuerant et
quidquid habebant de casamento ecclesiæ tenebat ex præcepto regis
guerram mihi fecerunt, et terras suas de rege tenentes pacifice hic eos
cum rege video adversum me convenisse.”
[261] “Rectitudinem facere” is the technical phrase. See Appendix C.
[262] “Tunc laici hujusmodi verbis Lanfranci totius Angliæ primatis
animati, adversus episcopum exclamantes dixerunt ‘injustum esse quod
rex episcopo responderet antequam regi fecisset justitiam.’ Laicis vero
hæc et alia multa declamantibus et iterantibus, facto silentio, dixit
episcopus.”
 [263] “Domini barones et laici, permittite me, quæso, quæ dicturus sum
regi dicere, archiepiscopis et episcopis respondere, quia nihil vobis habeo
dicere, et, sicut huc non veni judicium vestrum recepturus, ita illud
omnino recuso, et si domino nostri regi et archiepiscopis et episcopis
placuisset vos hic negotio interesse, nec me taliter obloqui decuisset.”
[264] See the complaints from the ecclesiastical side in N. C. vol. iv. p.
436.
[265] Mon. Angl. i. 247. “Tunc Rogerus Bygotus dixit regi, ‘Vos debetis
episcopo dicere unde eum appellare vultis, et postea, si ipse nobis voluerit
respondere de responsione sua facite eum judicari; sin autem, facite inde
quod barones vestri vobis consulerent.’”
[266] I cannot identify this Hugh. “Hugo cognomento pauper” (Ord. Vit.
806 A), son of Count Robert of Meulan, and afterwards Earl of Bedford
(Gest. Steph. 61), was not yet born.
[267] See above, p. 30.
[268] Mon. Angl. u. s. “Rex te appellat quod, cum ipse audivit quod
inimici sui super eum veniebant, et homines sui, episcopus scilicet
Baiocensis et Rogerus comes et alii plures regnum suum pariter sibi et
coronam auferre volebant, et ipse per consilium tuum contra illos
equitabat.” There is something odd in this calm mention of Earl Roger as
an open rebel.
[269] See above, p. 28.
[270] Macaulay, ii. 496–499, 510, 511.
[271] Mon. Angl. u. s. “Episcopus autem Hugoni respondit, ‘Hugo, dicas
quidquid volueris, non tibi tamen hodie respondebo.’”
[272] Mon. Angl. u. s. “Tum multum tumultuantes laici, quidam
rationibus, quidam vero contumeliis, adversus episcopum deiterarent.”
[273] Ib. “Domini archiepiscopi, nos non oporteret diutius hæc ita
considerare, sed deceret nos surgere et episcopos et abbates convocare,
quosdam etiam baronum et comitum istorum nobiscum habere, et cum eis
juste decernere si episcopus debeat prius investiri vel ante investituram de
querelis regis intrare in placitum.” The text has “S. Constantiensis
episcopus,” but Bishop Geoffrey must be meant.
[274] Mon. Angl. u. s. “Ad hæc Lanfrancus archiepiscopus, ‘Non est
necesse,’ inquit, ‘nos surgere, sed episcopus et homines sui egrediantur, et
nos remanentes, tam clerici quam laici, consideremus equaliter quid inde
juste facere debeamus.’”
[275] Ib. “Vade, nos enim juste faciemus quidquid fecerimus.”
[276] Ib. “Si ego hodie te et tuum ordinem judicare non potero, tu vel
tuus ordo nunquam me amplius judicabitis.”
[277] Ib. “Vide autem qui in domo ista remanent et me judicare
disponunt ut et canonicos judices habeant et canonice me judicent; si
enim aliter agerent, eorum judicia penitus recusarem.”
[278]Ib. “Rege, cum suis episcopis et consulibus et vicecomitibus et
præpositis et venatoribus aliisque quorumlibet officiorum, in judicio
remanente.”
[279] We have met with Osgeat the Reeve in Domesday. See N. C. vol.
v. p. 812. Croc the hunter, like others of his craft, appears in 49, 74 b. See
Ellis, i. 403. This odd mixture of great and small officials is not unusual. In
the “Constitutio Domus Regis” in Hearne’s Liber Niger, i. 341, the descent
from the Chancellor to the bakers and cooks—​the huntsmen come at the
end—​is more sudden than one would have looked for, though certain
chaplains and seneschals break the fall.
[280] See N. C. vol. v. pp. 423, 878.
[281] Mon. Angl. u. s. “Dominus noster archiepiscopus et regis curia
vobis judicat quod rectitudinem regi facere debetis antequam de vestro
feodo revestiat.”
[282] Ib. “Nullus mihi hodie vel ego alicui de feodo feci verbum,” says
Bishop William. To which Archbishop Thomas answers, “Vobis judicat curia
ista, quia de nulla re debet vos rex resaissire antequam sibi rectitudinem
faciatis.”
[283]Mon. Ang. u. s. “Episcopi sunt judices, et eos ad consilium tuum
habere non debes.”
[284] Ib. “Cum tuis ibi consule, quia de nostris in consilio tuo nullum
prorsus habebis.”
[285]Ib. “Parum consilii in his septem hominibus habeo contra virtutem
et scientiam totius hujus regni quod hic adversum me video
congregatum.”
[286] Mon. Angl. u. s. “In lege nostra prohibitum invenio, ne tale
judicium suspiciam.” This strange phrase, twice repeated, most likely
refers to the False Decretals, of which he seems to have had a copy with
him. See below, p. 109.
[287] Ib. “Apostolicam sedem Romanam, sanctam ecclesiam et beatum
Petrum ejusque vicarium appello, ut ipsius ordinatione negotii mei justam
sententiam suscipere merear, cujus dispositioni majores causas
ecclesiasticas et episcoporum judicia antiqua apostolorum eorumque
successorum atque canonum auctoritas reservavit.” Yet, according to the
doctrine held long after by Thomas Stubbs (see N. C. vol. iv. p. 260), the
Bishop of Durham need not have gone very far to find a Vicar of Saint
Peter.
[288] See N. C. vol. iv. p. 338.
[289] Mon. Angl. u. s. “Dispoliatus episcopio extra provinciam meam,
absentibus omnibus comprovincialibus meis, in laicali conventu causam
meam dicere compellor.”
[290] Mon. Ang. u. s. “Nos non de episcopio sed de tuo te feodo
judicamus, et hoc modo judicavimus Baiocensem episcopum ante patrem
hujus regis de feodo suo, nec rex vocabat eum episcopum in placito illo,
sed fratrem et comitem.”
[291] Ib. “Quia Dei gratia sapientissimus et nominatissimus estis, in hoc
sapere vestrum tam sublime intelligo, quod parvitas mea illud
comprehendere non potest; sed apostolicam sedem quam ex necessitate
appellavi per licentiam regis et vestram adire volo.”
[292] Mon. Ang. u. s. “In omni loco in quo non violentia sed justitia
dominetur, de scelere et perjurio me purgare paratus sum, et hoc quod hic
pro judicio recitasti in Romana ecclesia falsum et injuste dictum esse
monstrabo.”
[293]Ib. “In curia ista nullum ad præsens placitum subintrabo, quia nihil
ibi tam bene dicerem quin fautores regis depravando perverterent, qui
ipsam et non reverentes apostolicam auctoritatem post ejus appellationem
me judicio non legali gravant, sed Dei et Sancti Petri postulans auxilium
Romam vadam.”
[294] Ib. “Tunc rex ait, ‘Modo volo ut castellum tuum mihi reddas,
quoniam judicium meæ curiæ non sequeris.’”
[295]Mon. Ang. i. 248. “Per vultum de Luca nunquam exibis de manibus
meis donec castellum habeam.”
 [296] Ib. “Ego passus sum per tres servientes vestros aufferri mihi terras
et pecuniam ecclesiæ, præsentibus centum meis militibus, et in nullo
prorsus vobis restiti.”
[297]Durham is described as “Urbs ipsa in qua sedes est ecclesiæ.” The
Bishop adds; “Paratus sum bonos obsides et fiducias dare vobis, quod
homines mei quos ibi dum Romam vado volo dimittere in fidelitate vestra
eam custodient, et, si volueritis, libenter vobis servient.”
[298] “Tunc rex ait, ‘In veritate credas, episcope, quod nullo modo
Dunelmum reverteris et quod homines tui Dunelmi nullatenus
remanebunt, nec tu manus meas evades donec castellum tuum liberum
mihi reddas.’”
[299] Mon. Ang. u. s. “Si episcopus amplius castellum suum vobis
contradixerit, bene eum capere potestis, quia conductum quem hactenus
habuit nunc dimittit, cum prior conventionem frangit, et barones vestros
probare appetit quod fidem suam servarent non bene.”
[300] On Randolf Peverel and his alleged connexion with William, see N.
C. vol. iii. p. 662; iv. 200; v. 26.
[301] Mon. Angl. i. 248. “Tunc Radulfus Piperellus et omnes laici
unanimiter conclamantes dixerunt; ‘Capite eum, capite eum, bene enim
loquitur iste vetustus ligaminarius.’” One would like to have the original
French of this somewhat irreverent description of the Archbishop, but
gaoler seems to be the most likely meaning of the unusual word
ligaminarius.
[302] Ib. “Multum precor dominum meum regem ne fidem meam inde
faciat me mentiri, nullum enim proficuum in me haberet ulterius.”
[303] Ib. “Rex bene vos adquietavit; plenam namque rectitudinem
episcopo obtulit, et ipse eam vobis audientibus recusavit, regem quoque
Romam injuste invitavit; recognoscat igitur episcopus hoc justum fecisse
judicium, et si illud sequi nollet, et rex sibi naves inveniet et conductum.”
[304]“Christianam legem quam hic scriptam habeo, testem invoco.” See
above, p. 104.
[305] Mon. Ang. u. s. “Non est justum ut placitum vel judicium regis pro
aliqua contradictione longius procedat, sed quotiens in curia sua judicium
agitur, ibidem necesse est ut concedatur vel contradicatur, tu ergo
judicium nostrum vel hic concede, vel hic evidenti ratione contradicito.”
 [306] Ib. “Rex ait, ‘Dicas licet quidquid velis, non tamen effugies manus
meas nisi castellum prius mihi reddas.’” The Bishop has just before spoken
of “Roma, ubi debeo et ubi justitia magis quam violentia.”
[307] Ib. “Cum vos non solum episcopatum, verum et omnia mea,
injuste abstuleritis, et ipsam modo sedem violenter auferre velitis, pro
nulla re quam facere possim capi me patiar.”
[308]Ib. “Constituta est ergo dies qua episcopus urbem suis hominibus
vacuaret et rex ibi suos poneret.”
[309] Ib. “Tu pro regis damno et omnium nostrorum dedecore vadis
Romam, et ipse tibi terram dimitteret? Remane in terra sua, et ipse
episcopatum tuum præter urbem tibi reddet, ea conditione quod in curia
sua judicio baronum suorum rectitudinem sibi facias.”
[310] Mon. Ang. u. s. “Ego apostolicam sedem appellavi, quia in curia
ejus nullum justum judicio audio et nullo modo dimittam quin illuc vadam.”
[311] Ib. “Tunc rex ait, ‘Faciat mihi episcopus fiduciam quod damnum
meum citra mare non quærat vel recipiat, et quod naves meas quas sibi
inveniam non detinebit frater meus vel aliquis suorum ad damnum meum
contra nautarum voluntatem.’”
[312] Mon. Ang. u. s. “Reginaldus Paganellus ait, ‘Certe comites vestri
promiserunt hoc quod dicit episcopus et convenienter inde eos custodite.’”
“Reginaldus” must surely be a slip for “Radulfus.”
[313] Ib. “‘Tace,’ inquit rex, ‘quia pro nullius fiducia naves meas perdere
patiar, sed, si episcopus inde se fiduciam fecisse cognoverit, super illam
aliam non requiram.’”
[314] Ib. “Tunc rex iratus ait, ‘Per vultum de Luca, in hoc anno mare non
transibis, nisi fiduciam quam de navibus requiro prius modo feceris.’”
[315] Ib. “Faciam hanc et multo majorem, si necesse fuerit, fiduciam
antequam hic in captione detinear; sed bene omnes audiant quod ea
invitus faciam et captionis timore coactus.”
[316]Ib. “Rex ait, ‘Nullum conductum habebis, sed Wiltone moraberis
donec ego vere sciam quod castellum habeam in mea potestate, et tunc
demum naves recipies et conductum.’” Wilton seems an odd place for the
purpose; should it be “Wintonie?”
[317] Mon. Ang. u. s. “Cum quod vellem et deberem facere non valeam,
hoc ipsum quod dicitis injuste patiar et coactus.”
 [318] See N. C. vol. iv. p. 215. “Walterus de Haiencora,” or “Haiencorn,”
must be a corruption of his name.
[319]Mon. Angl. i. 249. “Precamur vos ut faciatis domino meo reddi
pecuniam.” The name of the speaker is given as “Willelmus de Merlao.”
[320]Ib. “Rex ait, ‘Videant barones isti si ego juste possum implacitare
episcopum.’”
[321] Ib. “Injustum esset si amplius implacitaretis eum, cum de vobis
mihi teneat et securum conductum habere debeat.”
[322] Mon. Ang. u. s. “Bene scias, episcope, quod nunquam transfretabis
donec castellum tuum habeam; episcopus enim Baiocensis inde me
castigavit.”
[323] Gilbert of Bretevile appears as a considerable landowner in
Hampshire (Domesday, 48) and Wiltshire (71). He may have been Sheriff
of either shire.
[324]See N. C. vol. iv. pp. 215, 800. Besides Erneis himself, we have
heard of a Ralph Fitz-Erneis at Senlac, vol. iii. p. 494.
[325] Mon. Ang. u. s. “Dissaisiverunt episcopum de ecclesia et de
castello et de omni terra sua xviii. Kal. Dec., et liberaverunt hominibus
episcopi Helponem balistarium regis.” The King’s writ follows. Helpo must
be Heppo. See N. C. vol. iv. p. 216. See Appendix C.
[326] Mon. Ang. u. s. “Accepit Ivo Taillesbosci duos milites episcopi, et
coegit eos placitare de animalibus Constantiensis episcopi de quibus
judicatum fuerat ante regem Dunelmensi episcopo non debere
respondere.” It is of course possible that there might be some ground for
impleading the knights, though not for impleading the Bishop.
[327] He had before asked; “dum in Anglia fuero, habetote mecum
unum bonum hominem, qui et hospitia mihi inveniat et ab impedimento
me defendat.” The “good man” assigned is “Robertus de Comitisvilla.” One
would think that he was a kinsman of the husband of Herleva, the King’s
step-grandfather.
[328] Roger in the text; but Robert must surely be meant.
[329] Mon. Ang. u. s. “Illi responderunt se nullam sibi navem liberaturos,
et dixerunt regem sibi præcepisse ut bene servarent episcopum, ne de
potestate regis exiret usque quo quid de eo fieri præciperet, illis per suas
sigillatas literas remandaret.”
 [330] Mon. Ang. u. s. “Venerunt ad eum Salesberiensis episcopus et
Robertus de Insula et Ricardus de Cultura, et summonuerunt eum de
parte regis, Kal. Decembr., ut in nativitate Domini esset Londoniæ ad
curiam regis, et faceret ei rectitudinem de Gaufrido monacho suo, qui,
postquam episcopus ad curiam venerat, de dominicatu episcopi quingenta
et triginta novem animalia acceperat, et munitionem castelli abstulerat de
quibusdam suis aliis hominibus, qui unum hominem regis occiderant.” The
Gemót was therefore to be at Westminster, not in its regular place at
Gloucester.
[331] Ib. “Quamvis juste facere potuissem, potui enim de meis facere
quidquid volui, usquequo de mea sede me dissaisivit.”
[332] Ib. “Ad curiam ejus amplius ire non possum, ipse enim omnia mea
mihi abstulit, et equos meos jam venditos manducavi.”
[333] He offers, “Solus, si liceat, transfretabo.”
[334] Mon. Angl. u. s. “Rex misit ei Wintoniensem episcopum et
Hugonem de Portu et Gaufridum de Traileio, et per illos sibi mandavit ut
Gaufridum monachum ad placitandum de prædictis forisfactis Dunelmum
mitteret, et ipse Londoniam iret, ut in nativitate Domini de hominibus suis
ibi rectitudinem regi faceret.”
[335] Ib. “Episcopus tristis misit ad comites Alanum et Rogerum et
Odonem, mandans eis impedimenta sua, et conjuravit eos per eam fidem
quam in baptismo susceperant et quam sibi promiserant.”
[336] Ib. “A Roberto fratre regis comite Normannorum honorifice
susceptus, totius Normanniæ curam suscepit.”
[337] See above, p. 91, where he is afraid of the “indocta multitudo.”
[338] See N. C. vol. iv. pp. 502, 675.
[339] Ann. Camb. 1087. “Resus filius Teudur a regno suo expulsus est a
filiis Bledint, scilicet Madauc, Cadugan, et Ririt. Resus vero ex Hibernia
classem duxit et revertitur in Britanniam.” The Brut is to the same effect.
[340] Ib. “Ingentem censum captivorum gentilibus et Scotis filius Teudur
tradidit.” The Brut for “gentiles et Scoti” has “Yscotteit ar Gúydyl,” marking
the Gwyddyl as heathen Ostmen. This is the most common use of the
word in the British writers; but we can hardly think that the Scots here
spoken of are Scots in the elder sense.
[341] In Ann. Camb. 1082, Trahaern (see N. C. iv. 675), with others, “a
Reso filio Teudur et a Grifino filio Conani occidisus est.” This Gruffydd must
be distinguished from Gruffydd son of Meredydd. He may be the “Grifin
puer” of Domesday, 180 b. “Griffin rex” in p. 269 is surely Gruffydd son of
Llywelyn.
[342] Ord. Vit. 669 B. “Grithfridus rex Guallorum cum exercitu suo fines
Angliæ invasit, et circa Rodelentum magnam stragem hominum et
incendia fecit, ingentem quoque prædam cepit, hominesque in
captivitatem duxit.”
[343] Orderic (u. s.) specially marks Gruffydd’s invasion as happening
“cum supradicta tempestate vehementer Anglia undique concuteretur et
mutuis vulneribus incolæ regni quotidie mactarentur.”
[344]See above, pp. 34, 47. Now is the time for the exploits of the
grandsons of Jestyn ap Gwrgan. See N. C. vol. v. p. 822, and Appendix
DD.
[345] We have seen him among the rebels. See above, p. 34.
[346] Ord. Vit. u. s. “Robertus Rodelenti princeps de obsidione Rofensi
rediens, et tam atroces damnososque sibi rumores comperiens,
vehementer dolens ingemuit, et terribilibus minis iram suam evidenter
aperuit.”
[347] Ib. 670 B. “Tertio die Julii Grithfridus rex Guallorum cum tribus
navibus sub montem qui dicitur Hormaheva littori appulsus est.” It needs a
moment’s thought to see that Hormaheva is Ormesheafod, the Orm’s
Head. Here the name bears the Scandinavian form given to it doubtless by
Northern rovers. The Worm’s Head in Gower, in its English form, marks
the presence of Low-Dutch settlers, whether Flemish or Saxon.
[348] Ord. Vit. 670 B. “Incolis Britonibus sævo Marte repulsis, fines suos
dilatavit, et in monte Dagaunoth, qui mari contiguus est, fortissimum
castellum condidit.” Orderic has clearly got hold of the right names and
the right incidents; but he has misconceived the topography.
Dwyganwy passes as the stronghold of that Maglocunus or Maelgwyn,
whom Gildas (Ep. 33) addresses as “insularis draco, multorum tyrannorum
depulsor, tam regno quam etiam vita” (cf. Nennius, c. 62, and Ann. Camb.
547, the year of his death). See Giraldus, It. Kamb. ii. 10; Descrip. Kamb.
i. 5 (where he calls it “nobile castellum”), vol. vi. pp. 136, 176.
[349] Ord. Vit. 670 C. “Interim mare fluctus suos retraxit, et in sicco
litore classis piratarum stetit. Grithfridus autem cum suis per maritima
discurrit, homines et armenta rapuit, et ad naves exsiccatas festine
remeavit.”
 [350] See N. C. vol. iii. p. 176.
[351] Ord. Vit. u. s. “Clamor vulgi Robertum meridie dormitantem
excitavit, eique hostilem discursum per terram suam nuntiavit. Ille vero, ut
jacebat, impiger surrexit, et mox præcones ad congregandum agmen
armatorum per totam regionem direxit. Porro ipse cum paucis bellatoribus
imparatus Guallos prosecutus est, et de vertice montis Hormohevæ, qui
nimis arduus est, captivos a piratis ligari, et in naves cum pecoribus
præcipitari speculatus est.”
Orderic must surely have confounded the Orm’s Head itself with the
lower hill of Dwyganwy. It is there, in or near his own castle, that we must
conceive Robert sleeping, not on the Orm’s Head itself, or on any casual
point of the flat ground between the two. To climb the higher of the two
peaks of Dwyganwy would be perfectly natural, and would give him a
wide enough view over the whole country. But to conceive him first
crossing the flat, and then climbing a huge mountain for no particular
object, seems quite out of the question.
[352] Ib. “Marchisus audax, ut leo nobilis, vehementer infremuit,
hominesque paucos qui secum inermes erant, ut, antequam æstus maris
rediret, super Guallos in sicco litore irruerent, admonuit.”
[353] Ord. Vit. 670 C. “Prætendunt suorum paucitatem, et per ardui
montis præcipitium descendendi difficultatem.”
[354] Ib. “Nimis doluit, impatiensque moræ per difficilem descensum
sine lorica cum uno milite nomine Osberno de Orgeriis, ad hostes
descendit.” I cannot identify this Osbern, unless he be “Osbernus filius
Tezonis,” who in Domesday (267 b, 268 b) holds a good deal of land in
Cheshire under Earl Hugh, but none seemingly under Robert himself. For
Orgères see Stapleton, ii. lxxxv.
[355] Ib. 670 D. “Quem cum viderent solo clypeo protectum et uno
tantum milite stipatum, omnes pariter in illum missilia destinant, et
scutum ejus jaculis intolerabiliter onerant, et egregium militem letaliter
vulnerant. Nullus tamen, quamdiu stetit et parmam tenuit, ad eum
comminus accedere, vel eum ense impetere ausus fuit.” Cf. the account of
the death of Siccius in Dion. Hal. xi. 26. He has an ὑπασπιστής to play the
part of Osbern of Orgères.
[356] Ib. “Bellicosus heros spiculis confossus genua flexit, et scutum
missilibus nimis onustum viribus effœtus dimisit.”
[357] Ib. “In conspectu suorum caput ejus abscindunt ac super malum
navis pro signo victoriæ suspendunt.”
 [358] Ord. Vit. 670 D. “Classe parata piratas per mare fugientes
persequebantur nimis tristes, dum caput principis sui super malum puppis
intuebantur.”
[359] Ib. 671 A. “Cum nimio luctu Anglorum et Normannorum.” This may
be well believed. Normans and English soon forgot their own differences
in warfare with the Welsh.
[360] But Orderic has forgotten his dates when he says, “Nuper illud
cœnobium Hugo Cestrensis consul construxerat, eique Ricardus Beccensis
monachus abbas præerat.” We shall see as we go on that the monks were
not planted at Saint Werburh’s till 1092 (see N. C. vol. iv. pp. 312, 491). It
is now that Orderic speaks of the “belluini cœtus”—​we are not told
whether they were Norman, English, or Welsh—​among whom Abbot
Richard had to labour.
[361] See N. C. vol. iv. p. 489.
[362] His gifts in lands, tithes, and villains, in Normandy and in England,
are reckoned up by Orderic, 669 C, D. Among them was “in civitate Cestra
ecclesiam sancti Petri de mercato et tres hospites.”
[363] Ord. Vit. 671 B. “Rainaldus pictor, cognomento Bartolomæus, variis
coloribus arcum tumulumque depinxit.”
[364] Ib. “Vitalis Angligena satis ab Ernaldo rogatus epitaphium elegiacis
versibus hoc modo edidit.”
[365] See N. C. vol. iv. p. 490.
[366] Ord. Vit. 672 A;
 “Eripe tartareis Robertum, Christe, camœnis [caminis];
Est nimis ipse reus; terge, precor, facinus;”
with four more lines to the same effect.
[367] Ord. Vit. 671 C, D.
“Montem Snaudunum fluviumque citum Colvenum,
Pluribus armatis transiliit vicibus.
Præcipuam pulcro Blideno rege fugato
Prædam cum paucis cepit in insidiis.
Duxit captivum lorisque ligavit Hoëllum
Qui tunc Wallensi rex præerat manui.
Cepit Grithfridum regem vicitque Trehellum;
Sic micuit crebris militiæ titulis.
Attamen incaute Wallenses ausus adire,
Occidit æstivi principio Julii.
Prodidit Owenius, rex est gavisus Hovellus;
Facta vindicta monte sub Hormaheva.
Ense caput secuit Grithfridus, et in mare jecit,
Soma quidem reliquum possidet hunc loculum.”
The exploits of Robert fully entitled him to Orderic’s pet Greek word.
“Colvenus” must be some corrupt form of Conwy.
[368]We have seen that, in describing the rebellion of 1088, the words
of the Chronicler are, “þa riceste Frencisce men þe weron innan þisan
lande wolden swican heora hlaforde þam cynge.” In 1101 we read simply,
“þa sona þæeræfter wurdon þa heafod men her on lande wiðerræden
togeanes þam cynge.”
[369] See N. C. vol. ii. p. 308.
[370] I refer to the passage which I have already quoted in N. C. vol. v.
p. 830, where William Rufus, just before his death (Ord. Vit. 782 B),
mocks at the English regard for omens; “Num prosequi me ritum autumat
Anglorum, qui pro sternutatione et somnio vetularum dimittunt iter suum
seu negotium?”
[371] See N. C. vol. iv. p. 393.
 [372] Stigand appears in the list of deaths which accompanied that of
William in the Chronicle, where one would think that the persons spoken
of died after him; but in the less rhetorical account of the same year in
Florence they seem to have died before him. The Life of Lanfranc at the
end of the Chronicles records the consecrations and benediction of all the
three prelates with whom we are concerned, Geoffrey, Guy, and John, in
1088; “Cantuariæ, in sede metropoli, examinavit atque sacravit.” Cf.
Gervase, X Scriptt. 1654.
[373] See Stephens’ Memorials of Chichester, p. 47.
[374] See N. C. vol. ii. p. 459.
[375] Will. Malms. Gest. Pont. 195 draws a curious picture of him; “Erat
medicus probatissimus, non scientia sed usu, ut fama, nescio an vera,
dispersit. Litteratorum contubernio gaudens, ut eorum societate aliquid
sibi laudis ascisceret; salsioris tamen in obloquentes dicacitatis quam
gradus ejus interesse deberet.” He had just before described him as
“natione Turonicus, professione medicus, qui non minimum quæstum illo
conflaverat artificio.” The local writer in the Historiola (21) calls him “vir
prudens et providus.”
[376] See N. C. vol. iv. p. 417.
[377] See N. C. vol. ii. p. 411.
[378] See Appendix F.
[379] See above, p. 41.
[380]Will. Malms. Gest. Pont. 196. “Cessit Andreas Simoni, frater fratri,
minor majori.” Yet before the west front of the church of Wells there can
be no doubt who was there looked on as the very chiefest apostle.
[381] See Appendix F.
[382] See Appendix F.
[383] Will. Malms. 195. “Sepultus est in ecclesia sancti Petri, quam a
fundamentis erexerat, magno et elaborato parietum ambitu.”
[384] The like usage is still more remarkable at Durham and Carlisle,
churches which never had an abbot distinct from the bishop. At Carlisle
the “abbey” seems to mean the monastic precinct rather than the church
itself.
[385]See N. C. vol. iv. p. 409. The story is told in the Winchester
Appendix to the Chronicles.
 [386] Chron. Wint. App. 1089. “Post ejus [Lanfranci] obitum, monachi
sancti Augustini, præfato abbati suo Widoni palam resistentes, cives
Cantuariæ contra eum concitaverunt, qui illum armata manu in sua domo
interimere temptaverunt. Cujus familia cum resisteret, pluribus utrimque
vulneratis et quibusdam interfectis, vix abbas inter manus illorum illæsus
evasit, et ad matrem ecclesiam, quærendo auxilium, Cantuariam, fugit.”
This last odd expression must be owing to the fact that Saint Augustine’s
stood outside the walls.
[387] Chron. Wint. App. “Coram populo subire disciplinam, quia palam
peccaverant, ii qui advenerant, decreverunt; sed prior et monachi ecclesiæ
Christi, pietate moti, restiterunt; ne, si palam punirentur, infames deinceps
fierent, sicque eorum vita ac servitus contemneretur. Igitur concessum est
ut in ecclesia fieret, ubi non populus, sed soli ad hoc electi admitterentur.”
Thierry, who of course colours the whole story after his fashion,
becomes (ii. 140) not a little amusing at this point. The flogging was done
by two monks of Christ Church, “Wido et Normannus.” If one stopped to
think of matters of nationality at such a moment, we might admire the
impartiality of the Norman bishops in entrusting the painful duty to a
monk of each nation, somewhat on the principle of a mixed jury. For no
one can doubt that Normannus, Northman, was as good an Englishman as
Northman the son of Earl Leofwine and other English bearers of that
name. Thierry, on the other hand, tells us that the whipping was done by
“deux religieux étrangers, appelés Guy et Le Normand.” He seemingly
mistook the Christian name “Normannus” for the modern surname
“Lenormand,” and he forgot that this last could be borne only by one
whose forefathers had moved from Normandy to some other French-
speaking land.
[388] Chron. Wint. App.
[389] Ib. See N. C. vol. iv. p. 410.
[390] See Lanfranc, Ep. 67 (i. 80, ed. Giles); N. C. vol. iv. p. 439.
[391] Chron. Petrib. 1089. “On þisum geare se arwurða muneca feder
and frouer Landfranc arcebisceop gewat of þissum life, ac we hopiað þæt
he ferde to þæt heofanlice rice.”
[392] The exact date comes from his Life, 52 (i. 312, ed. Giles); “anno
archiepiscopatus xix, v. calendas Junii diem clausit extremum.” The Latin
Chronicler gives us the exact measure of his primacy; “In sede pontificali
sedit annis decem et octo, mensibus ix. duobus diebus.” The Life gives us
his epitaph, which begins;