Download the Full Ebook and Access More Features - ebooknice.

com

(Ebook) Sparse Estimation with Math and Python:

100 Exercises for Building Logic by Joe Suzuki
ISBN 9789811614378, 9811614377

https://ebooknice.com/product/sparse-estimation-with-math-
and-python-100-exercises-for-building-logic-35997448

OR CLICK HERE

DOWLOAD EBOOK

Download more ebook instantly today at https://ebooknice.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!

(Ebook) Sparse Estimation with Math and Python: 100

Exercises for Building Logic by Joe Suzuki ISBN
9789811614378, 9811614377
https://ebooknice.com/product/sparse-estimation-with-math-and-
python-100-exercises-for-building-logic-35997450

ebooknice.com

(Ebook) Statistical Learning with Math and Python: 100

Exercises for Building Logic by Joe Suzuki ISBN
9789811578762, 9811578761
https://ebooknice.com/product/statistical-learning-with-math-and-
python-100-exercises-for-building-logic-33792268

ebooknice.com

(Ebook) Kernel Methods for Machine Learning with Math and

Python: 100 Exercises for Building Logic by Joe Suzuki
ISBN 9789811904004, 9811904006
https://ebooknice.com/product/kernel-methods-for-machine-learning-
with-math-and-python-100-exercises-for-building-logic-43837738

ebooknice.com

(Ebook) Sparse Estimation with Math and R by Joe Suzuki

ISBN 9789811614453, 9811614458

https://ebooknice.com/product/sparse-estimation-with-math-
and-r-33795076

ebooknice.com
(Ebook) doing math with python doing math with python by
kan

https://ebooknice.com/product/doing-math-with-python-doing-math-with-
python-50196050

ebooknice.com

(Ebook) The Definitive Guide to Masonite: Building Web

Applications with Python by Christopher Pitt, Joe Mancuso
ISBN 9781484256015, 1484256018
https://ebooknice.com/product/the-definitive-guide-to-masonite-
building-web-applications-with-python-50194580

ebooknice.com

(Ebook) Math for Programmers: 3D graphics, machine

learning, and simulations with Python MEAP V10 by Paul
Orland
https://ebooknice.com/product/math-for-programmers-3d-graphics-
machine-learning-and-simulations-with-python-meap-v10-11069540

ebooknice.com

(Ebook) Math Adventures with Python: An Illustrated Guide

to Exploring Math with Code by Peter Farrell ISBN
9781593278670, 1593278675
https://ebooknice.com/product/math-adventures-with-python-an-
illustrated-guide-to-exploring-math-with-code-36151058

ebooknice.com

(Ebook) Math Adventures with Python: An Illustrated Guide

to Exploring Math with Code by Peter Farrell ISBN
9781593278670, 1593278675
https://ebooknice.com/product/math-adventures-with-python-an-
illustrated-guide-to-exploring-math-with-code-10674096

ebooknice.com
Joe Suzuki

Sparse Estimation with Math and

Python
100 Exercises for Building Logic
1st ed. 2021
Joe Suzuki
Graduate School of Engineering Science, Osaka University, Toyonaka,
Osaka, Japan

ISBN 978-981-16-1437-8 e-ISBN 978-981-16-1438-5

https://doi.org/10.1007/978-981-16-1438-5

© The Editor(s) (if applicable) and The Author(s), under exclusive

license to Springer Nature Singapore Pte Ltd. 2021

This work is subject to copyright. All rights are solely and exclusively
licensed by the Publisher, whether the whole or part of the material is
concerned, specifically the rights of reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other
physical way, and transmission or information storage and retrieval,
electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.

The use of general descriptive names, registered names, trademarks,

service marks, etc. in this publication does not imply, even in the
absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general
use.

The publisher, the authors and the editors are safe to assume that the
advice and information in this book are believed to be true and accurate
at the date of publication. Neither the publisher nor the authors or the
editors give a warranty, expressed 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.
This Springer imprint is published by the registered company Springer
Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04
Gateway East, Singapore 189721, Singapore
Preface
I started considering the sparse estimation problems around 2017
when I moved from the mathematics department to statistics in Osaka
University, Japan. I have been studying information theory and
graphical models for over thirty years.
The first book I found is “Statistical Learning with Sparsity” by T.
Hastie, R. Tibshirani, and M. Wainwright. I thought it was a monograph
rather than a textbook and that it would be tough for a non-expert to
read it through. However, I downloaded more than fifty papers that
were cited in the book and read them all. In fact, the book does not
instruct anything but only suggests how to study sparsity. The contrast
between statistics and convex optimization gradually attracted me as I
understand the material.
On the other hand, it seems that the core results on sparsity have
come out around 2010–2015 for research. However, I still think further
possibilities and expansions are there. This book contains all the
mathematical derivations and source programs, so graduate students
can construct any procedure from scratch by getting help from this
book.
Recently, I published books “Statistical Learning with Math and R”
(SLMR), “Statistical Learning with Math and Python” (SLMP), and
“Sparse Estimation with Math and R” (SEMR). The common idea is
behind the books (XXMR/XXMP). They not only give knowledge on
statistical learning and sparse estimation but also help build logic in
your brain by following each step of the derivations and each line of the
source programs. I often meet data scientists engaged in machine
learning and statistical analyses for research collaborations and
introduce my students to them. I recently found out that almost all of
them think that (mathematical) logic rather than knowledge and
experience is the most crucial ability for grasping the essence in their
jobs. Our necessary knowledge is changing every day and can be
obtained when needed. However, logic allows us to examine whether
each item on the Internet is correct and follow any changes; we might
miss even chances without it.
What makes SEMP unique?
I have summarized the features of this book as follows.
1. Developing logic

To grasp the essence of the subject, we mathematically formulate

and solve each ML problem and build those programs. The SEMP
instills “logic” in the minds of the readers. The reader will acquire
both the knowledge and ideas of ML, so that even if new technology
emerges, they will be able to follow the changes smoothly. After
solving the 100 problems, most of the students would say “I learned
a lot”.

2. Not just a story

If programming codes are available, you can immediately take

action. It is unfortunate when an ML book does not offer the source
codes. Even if a package is available, if we cannot see the inner
workings of the programs, all we can do is input data into those
programs. In SEMP, the program codes are available for most of the
procedures. In cases where the reader does not understand the
math, the codes will help them understand what it means.

3. Not just a how to book: an academic book written by a university

professor.

This book explains how to use the package and provides examples
of executions for those who are not familiar with them. Still,
because only the inputs and outputs are visible, we can only see the
procedure as a black box. In this sense, the reader will have limited
satisfaction because they will not be able to obtain the essence of
the subject. SEMP intends to show the reader the heart of ML and is
more of a full-fledged academic book.

4. Solve 100 exercises: problems are improved with feedback from

university students
 The exercises in this book have been used in university lectures and
have been refined based on feedback from students. The best 100
problems were selected. Each chapter (except the exercises)
explains the solutions, and you can solve all of the exercises by
reading the book.
5. Self-contained

All of us have been discouraged by phrases such as “for the details,

please refer to the literature XX.” Unless you are an enthusiastic
reader or researcher, nobody will seek out those references. In this
book, we have presented the material in such a way that consulting
external references is not required. Additionally, the proofs are
simple derivations, and the complicated proofs are given in the
appendices at the end of each chapter. SEMP completes all
discussions, including the appendices.

6. Readers’ pages: questions, discussion, and program files

The reader can ask any question on the book via https://​bayesnet.​
org/​books.

Acknowledgments The author wishes to thank Tianle Yang, Ryosuke

Shinmura, Tomohiro Kamei and Daichi Kashiwara for checking the
manuscript in Japanese. The author thanks Professors Shu-ichi Kawano,
Hidetoshi Matsui, and Kei Hirose for their helpful advices for three
years before the current book has been published. This English book is
largely based on the Japanese book published by Kyoritsu Shuppan Co.,
Ltd. in 2020. The author would like to thank Kyoritsu Shuppan Co., Ltd.,
in particular its editorial members Mr. Tetsuya Ishii and Ms. Saki Otani.
The author also appreciates Ms. Mio Sugino, Springer, for preparing the
publication and providing advice on the manuscript.
Joe Suzuki
Osaka, Japan
September 2021
Contents
1 Linear Regression
1.​1 Linear Regression
1.​2 Subderivative
1.​3 Lasso
1.​4 Ridge
1.​5 A Comparison Between Lasso and Ridge
1.​6 Elastic Net
1.7 About How to Set the Value of

Exercises 1–20
2 Generalized Linear Regression
2.​1 Generalization of Lasso in Linear Regression
2.​2 Logistic Regression for Binary Values
2.​3 Logistic Regression for Multiple Values
2.​4 Poisson Regression
2.​5 Survival Analysis
Appendix Proof of Proposition
Exercises 21–33
3 Group Lasso
3.​1 When One Group Exists
3.​2 Proxy Gradient Method
3.​3 Group Lasso
3.​4 Sparse Group Lasso
3.​5 Overlap Lasso
3.​6 Group Lasso with Multiple Responses
 3.​7 Group Lasso Via Logistic Regression
3.​8 Group Lasso for the Generalized Additive Models
Appendix Proof of Proposition
Exercises 34–46
4 Fused Lasso
4.​1 Applications of Fused Lasso
4.​2 Solving Fused Lasso Via Dynamic Programming
4.​3 LARS
4.​4 Dual Lasso Problem and Generalized Lasso
4.​5 ADMM
Appendix Proof of Proposition
Exercises 47–61
5 Graphical Models
5.​1 Graphical Models
5.​2 Graphical Lasso
5.​3 Estimation of the Graphical Model Based on the Quasi-
Likelihood
5.​4 Joint Graphical Lasso
Appendix Proof of Propositions
Exercises 62–75
6 Matrix Decomposition
6.​1 Singular Decomposition
6.​2 Eckart-Young’s Theorem
6.​3 Norm
6.​4 Sparse Estimation for Low-Rank Estimations
Appendix Proof of Propositions
 Exercises 76–87
7 Multivariate Analysis
7.​1 Principal Component Analysis (1):​SCoTLASS
7.​2 Principle Component Analysis (2):​SPCA
7.3 -Means Clustering
7.​4 Convex Clustering
Appendix Proof of Proposition
Exercises 88–100
References
© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021
J. Suzuki, Sparse Estimation with Math and Python
https://doi.org/10.1007/978-981-16-1438-5_1

1. Linear Regression
Joe Suzuki1
(1) Graduate School of Engineering Science, Osaka University, Toyonaka, Osaka, Japan

Joe Suzuki
Email: [email protected]

In general statistics, we often assume that the number of samples N is greater than the
number of variables p. If this is not the case, it may not be possible to solve for the best-
fitting regression coefficients using the least squares method, or it is too
computationally costly to compare a total of models using some information
criterion.
When p is greater than N (also known as the sparse situation), even for linear
regression, it is more common to minimize, instead of the usual squared error, the
modified objective function to which a term is added to prevent the coefficients from
being too large (the so-called regularization term). If the regularization term is a
constant times the L1-norm (resp. L2-norm) of the coefficient vector, it is called Lasso
(resp. Ridge). In the case of Lasso, if the value of increases, there will be more
coefficients that go to 0, and when reaches a certain value, all the coefficients will
eventually become 0. In that sense, we can say that Lasso also plays a role in model
selection.
In this chapter, we examine the properties of Lasso in comparison to those of Ridge.
After that, we investigate the elastic net, a regularized regression method that combines
the advantages of both Ridge and Lasso. Finally, we consider how to select an
appropriate value of .

1.1 Linear Regression

Throughout this chapter, let and be integers, and let the (i, j) element of
the matrix and the kth element of the vector be denoted by and

, respectively. Using these X, y, we find the intercept and the slope

 that minimize . Here, the L2-norm of

is denoted by .

First, for the sake of simplicity, we assume that the jth column of X,
and y have already been centered. That is, for each , define

, and assume that has already been subtracted from each so that is
satisfied. Similarly, defining , we assume that was subtracted in advance

from each so that holds. Under this condition, one of the parameters

for which we need to solve, say, , is always 0. In particular,

holds. Thus, from now, without loss of generality, we may assume that the intercept
is zero and use this in our further calculations.
We begin by first observing the following equality:

(1.1)

In particular, the jth element of each side can be rewritten as follows:

Thus, when we set the right-hand side of (1.1) equal to 0, and if is invertible, then
becomes

(1.2)
For the case where , write each column of X as ; we see that
 (1.3)

If we had not performed the data centering, we would still obtain the same slope ,

though the intercept would be

(1.4)

Here, and are the means before data centering.

We can implement the above using the Python as follows:

In this book, we focus more on the sparse case, i.e., when p is larger than N. In this
case, a problem arises. When , the matrix does not have an inverse. In fact,
since

is singular. Moreover, when X has precisely the same two columns, ,

and the inverse matrix does not exist.
On the other hand, if p is rather large, we have p independent variables to choose
from for the predictors of the variable when carrying out model selection. Thus, the
combinations are

(that is, whether we choose each of the variables or not). This means we have to
compare a total of models. Then, to extract the proper model combination by using
an information criterion or cross-validation, the computational resources required will
grow exponentially with the number of variables p.
To deal with this kind of problem, let us consider the following. Let be a
constant. We add a term to that penalizes for being too large in size.

Specifically, we define

(1.5)

or

(1.6)

Our work now is to solve for the minimizer to one of the above quantities. Here,
for , is the L1-norm and is the L2-

norm of . To be more specific, we plug mean-centered and into

either (1.5) or (1.6), then minimize it with respect to the slope (this is called

Lasso or Ridge, respectively), and finally, we use (1.4) to compute .

1.2 Subderivative
To address the minimization problem for Lasso, we need a method for optimizing
functions that are not differentiable. When we want to find the points x of the maxima
or minima of a single-variable polynomial, say, , we can differentiate

it and find the solution to . However, what should we do when we encounter

functions such as , which contain an absolute value? To address

this, we need to extend our concept of differentiation to a more general one.

Throughout the following claims, let us assume that f is convex [4, 6]. In general, we
say that f is convex (downward)1 if, for any and ,

holds. For instance, is convex (Fig. 1.1, left) because

is satisfied. To check this, since both sides are nonnegative, the RHS squared minus the
LHS squared gives . As another example, consider

(1.7)

This function satisfies the following:

Therefore, it is not convex (Fig. 1.1, right). If functions f, g are convex, then for any
, the function has to be convex since the following holds:

Fig. 1.1 Left: is convex. However, at the origin, the derivatives from each side differ; thus, it

is not differentiable. Right: We cannot simply judge from its shape, but this function is not convex
Next, for any convex function , fix arbitrarily. For all , we
say that the set of all that satisfies

(1.8)
is a subderivative of f at .
If f is differentiable at , then the subderivative will be a set that contains only 1
element, say, .2 We prove this as follows.
First, the convex function f is differentiable at ; thus, it satisfies
. To see this, since f is convex,
This can be rewritten as

In fact, whether or , we have that

holds. Thus, the above inequation is true.

Next, when the convex function f is differentiable at , we can show that is
the one and only value of z that satisfies (1.8). In particular, when , for (1.8) to be
satisfied, we need . Similarly, when , for (1.8) to be satisfied, we

need . Thus, z needs to be larger than or equal to the derivative on the

left and, at the same time, be less than or equal to the derivative on the right at . Since
f is differentiable at , those 2 derivatives are equal; this completes the proof.
The main interest of this book is specifically the case where and .
Hence, by (1.8), its subderivative is the set of z such that for any , . These
values of z lie in the interval greater than or equal to and less than or equal to 1, and

is true. Let us confirm this. If for any x, holds, then for and ,
and , respectively, need to be true. Conversely, if , then
is true for any arbitrary .
Example 1 By dividing into 3 cases , , and , find the values x that
attain the minimum of and . Note that for , we can find

their usual derivatives, but for at , its subderivative is the interval

.

Therefore, has a minimum at (Fig. 1.2, left).

Therefore, has a minimum at (Fig. 1.2, right). We use the following

code to draw the figures.

Fig. 1.2 (left) has a minimum at , and (right) has a minimum at

. Neither is differentiable at . Despite not being differentiable, the point is a minimum for

the figure on the right

The subderivative of at is the interval [−1,1]. This fact
summarizes this chapter.

1.3 Lasso
As stated in Sect. 1.1, the method considered for the minimization of

is called Lasso [28].

From the formularization of (1.5) and (1.6), we can tell that Lasso and Ridge are the
same in the sense that they both try to control the size of regression coefficients .
However, Lasso also has the property of leaving significant coefficients as nonnegative,
which is particularly beneficial in variable selection. Let us consider its mechanism.
Note that in (1.5), the division of the first term by 2 is not essential: we would obtain
an equivalent formularization if we double the value of . For the sake of simplicity,
first let us assume that

(1.9)

holds, and let . With this assumption, the calculations are made much

simpler.
 Solving for the subderivative of L with respect to gives

(1.10)

which means that

Fig. 1.3 Shape of the function when

Thus, we have that

Here, the RHS can be rewritten using the following function:

(1.11)

and hence becomes . We plot the graph of when in Fig. 1.3

using the code provided below:
 Next, let us consider the case where (1.9) is not satisfied. We rewrite (1.10) by

Here, we denote by , and let be . Next, fix ( ),

and update . We do this repeatedly from to until it converges

(coordinate descent). For example, we can implement the algorithm as follows:

Note that here, after we obtain the value of , we use and to

calculate the value of . The following centralize function performs data centering
and returns a list of 5 results.
 Thus, we may standardize the data first, then perform Lasso, and finally restore the
data. The aim of doing this is to examine our data all at once. Because the algorithm sets
all less than or equal to to 0, we do not want it to be based on the indices

. Each jth column of X is divided by scale[j], and consequently, the

estimated will be larger to that extent. We then divide by scale[j] as well.

Example 2 Putting U.S. crime data

https://web.stanford.edu/ hastie/StatLearnSparsity/data.htm into
the text file crime.txt, we set the crime rate per 1 million residents as the target
variable and then select appropriate explanatory variables from the list below by
performing Lasso.

Column Cov./Res. Definition of variable

1 Response Crime rate per 1 million residents
2 (we currently do not use this)
3 Covariate Annual police funding
4 Covariate % of people 25 years+ with 4 yrs. of high school education
5 Covariate % of 16–19-year-old persons not in high school and not high school graduates
6 Covariate % of 18–24-year-old persons in college
7 Covariate % of people 25 years+ with at least 4 years of college education

We call the function linear_lasso and execute as described below:

Fig. 1.4 Result of Example 2. In the case of Lasso, we see that as increases, the coefficients decrease.

At a certain , all the coefficients will be 0. The at which each coefficient becomes 0 varies

As we can see in Fig. 1.4, as increases, the absolute value of each coefficient
decreases. When reaches a certain value, all coefficients go to 0. In other words, for
each value of , the set of nonzero coefficients differs. The larger becomes, the
smaller the set of selected variables.
When performing coordinate descent, it is quite common to begin with a large
enough that every coefficient is zero and then make smaller gradually. This method is
called a warm start, which utilizes the fact that when we want to calculate the
coefficient for all values of , we can improve the calculation performance by setting
the initial value of to the estimated from a previous . For example, we can write
the program as follows:

Example 3 We use the warm start method to reproduce the coefficient for each
in Example 2.

First, we make the value of large enough so that all are 0 and
then gradually decrease the size of while performing coordinate descent. Here, for
simplicity, we assume that for each , we have ; moreover, the
values of are all different. In this case, the smallest that makes

can be calculated by . Particularly, for any

larger than this formula, it will be satisfied that for all , . Then, we
have that and

hold. Thus, when we decrease the size of , one of the values of j will satisfy
. Again, since ( ), if we continue to make it

smaller, we still have that , and thus, the value of for

that j becomes smaller than .

Fig. 1.5 The execution result of Example 4. The numbers at the top represent the number of estimated
coefficients that are not zero
The glmnet package is often used [11].
Example 4 (Boston) Using the Boston dataset and setting the variable of the 14th
column as the target variable and the other 13 variables as predictors, we plot a graph
similar to that of the previous one (Fig. 1.5).

Column Variable Definition of variable

1 CRIM Per capita crime rate by town
2 ZN Proportion of residential land zoned for lots over 25,000 sq.ft.
3 INDUS Proportion of nonretail business acres per town
4 CHAS Charles River dummy variable (1 if tract bounds river; 0 otherwise)
5 NOX Nitric oxide concentration (parts per 10 million)
6 RM Average number of rooms per dwelling
7 AGE Proportion of owner-occupied units built prior to 1940
8 DIS Weighted distances to five Boston employment centers
9 RAD Index of accessibility to radial highways
10 TAX Full-value property-tax rate per $10,000
11 PTRATIO Pupil-teacher ratio by town
12 BLACK Proportion of blacks by town
13 LSTAT % lower status of the population
14 MEDV Median value of owner-occupied homes in multiples of $1,000

So far, we have seen how Lasso can be useful in the process of variable selection.
However, we have not explained the reason why we seek to minimize (1.5) for .
Why do we not instead consider minimizing the following usual information criterion
 (1.12)

for ? Here, represents the number of nonzero elements of that vector.

Lasso, as well as Ridge, which is to be discussed in the next chapter, has the
advantage that minimization is convex. For a globally convex function, because the
minimum and minimal points coincide, we can search the optimal solution effectively.
On the other hand, the minimization by (1.12) requires an exponential time for the
number of variables, p. In particular, since the function in (1.7) is not convex,
cannot be convex either. Because of this, we need to consider instead the minimization
of (1.5). An optimization problem becomes meaningful only after there exists an
effective search algorithm.

1.4 Ridge
In Sect. 1.1, we made an assumption about that the matrix is

invertible, and, based on this, we showed that the that minimizes the squared error
is given by .

First, when , the possibility that is singular is not that high, though we
may have another problem instead: the confidence interval becomes significant when
the determinant is small. To cope with this, we let be a constant and add to the
squared error the norm of times . That is, the method considering the minimization
of

($1.6$)

is commonly used. This method is called Ridge. Differentiating L with respect to gives

If is not singular, we obtain

Here, whenever , even for the case , we have that is

nonsingular. In particular, since the matrix is positive semidefinite, we have that

its eigenvalues are all nonnegative. Therefore, the eigenvalues of
can be calculated by

and thus, all of them are positive.

Again, when all the eigenvalues are positive, their product, , is also

positive, which is the same as saying that is nonsingular. Note that this

always holds regardless of the sizes of p, N. When , the rank of is

less than or equal to N, and hence, the matrix is singular. Therefore, for this case, the
following conditions are equivalent:

As an example of the Ridge case, we can write the following program:

Example 5 We use the same U.S. crime data as that of Example 2 and perform the
following analysis. To control the size of the coefficient of each predictor, we call the
function ridge and then execute.
 In Fig. 1.6, we plot how each coefficient changes with the value of .

Fig. 1.6 The execution result of Example 5. The changes in the coefficient with respect to based

on Ridge. As becomes larger, each coefficient decreases to 0

1.5 A Comparison Between Lasso and Ridge
Next, let us compare Fig. 1.4 of Lasso to Fig. 1.6 of Ridge. We can see that they are the
same in the sense that when becomes larger, the absolute value of each coefficient
approaches 0. However, in the case of Lasso, when reaches a certain value, one of the
coefficients becomes exactly zero, and the time at which that occurs varies for each
variable. Thus, Lasso can be used for variable selection.
So far, we have shown this fact analytically, but it is also good to have an intuition
geometrically. Figures similar to those in Fig. 1.7 are widely used when one wants to
compare Lasso and Ridge.
Fig. 1.7 Each ellipse is centered at , representing the contour line connecting all the points that

give the same value of (1.13). The rhombus in the left figure is the L1 regularization constraint
, while the circle in the right figure is the L2 regularization constraint

Let such that is composed of 2 columns . In

the least squares process, we solve for the values that minimize
. For now, let us denote them by , respectively. Here,

if we let , we have

However, since for any ,

holds, we can rewrite the quantity to be minimized, , as follows:

(1.13)
and, of course, if we let here, we obtain the minimum ( RSS).

Thus, we can view the problems of Lasso and Ridge in the following way: the
minimization of quantities (1.5) and (1.6) is equivalent to finding the values of
that satisfy the constraints and , respectively, that also

minimize the quantity of (1.13) (here, the case where is large is equivalent to the case
where are small).
The case of Lasso is the same as in the left panel of Fig. 1.7. The ellipses are centered
at and represent contours on which the values of (1.13) are the same. We

expand the size of the ellipse (the contour), and once we make contact with the
rhombus, the corresponding values of are the solution to Lasso. If the rhombus
is small ( is large), it is more likely to touch only one of the four rhombus vertices. In
this case, one of the values will become 0. However, in the Ridge case, as in the
right panel of Fig. 1.7, a circle replaces the Lasso rhombus; hence, it is less likely that
will occur.
In this case, if the least squares solution lies in the green zone of Fig. 1.8,

then we have either or as our solution. Moreover, when the rhombus is

small ( is large), even when remains the same, the green zone will become

larger, which is the reason why Lasso performs well in variable selection.

Fig. 1.8 The green zone represents the area in which the optimal solution would satisfy either

or if the center of the ellipses lies within it

 We should not overlook one of the advantages of Ridge: its performance when
dealing with the case of collinearity. That is, it can handle well even the case where the
matrix of explanatory variables contains columns that are highly related. Let us define
by

the VIF (variance inflation factor). The larger this value is, the better the jth column
variable is explained by the other variables. Here, denotes the coefficient of

determination squared in which is the target variable, and the other variables are
predictors.

Example 6 We compute the VIF for the Boston dataset. It shows that the 9th and 10th
variables (RAD and TAX) have strong collinearity.
In usual linear regression, if the VIF is large, the estimated coefficient will be

unstable. Particularly, if two columns are precisely the same, the coefficient is
unsolvable. Moreover, for Lasso, if two columns are highly related, then generally one of
them will be estimated as 0 and the other as nonzero. However, in the Ridge case, for
, even when the columns j, k of X are the same, the estimation is solvable, and both
of them will obtain the same value.
In particular, we find the partial derivative of

with respect to and make it equal to 0.

Then, plugging into each gives
Example 7 We perform Lasso for the case where the variables and
have strong correlations. We generate groups of data distributed in
the following way:

Then, we apply linear regression analysis with Lasso to . We plot how

the coefficients change relative to the value of in Fig. 1.9. Naturally, one might expect
that similar coefficient values should be given to each of the related variables, though it
turns out that Lasso does not behave in this way.
Fig. 1.9 The execution result of Example 7. The Lasso case is different from that of Ridge: related
variables are not given similar estimated coefficients. When we use the glmnet package, its default
horizontal axis is the L1-norm [11]. This is the value , which becomes smaller as increases.

Therefore, the figure here is left-/right-reversed compared to that where or is set as the

horizontal axis

1.6 Elastic Net

Up until now, we have discussed the pros and cons of Lasso and Ridge. This section
studies a method intended to combine the advantages of the two, i.e., the elastic net.
Specifically, the method considers the problem of finding that minimizes

(1.14)

The that minimizes (1.14) for the case of Lasso ( ) is ,

while for the case of Ridge ( ), it is . In general cases, it is

 (1.15)

This is called the elastic net. For each , if we find the subderivative of (1.14)
with respect to , we obtain

Here, let , , and . We have used the

fact that

where is a function that returns for each element.

Then, we can write the program for the elastic net based on (1.15), as shown in the
following. Here, we added a parameter to the line of #, and the only essential change
lies in the three lines of ##.
Example 8 If we add the additional parameter alpha = 0, 0.25, 0.5, 0.75
to the function elastic_net of Example 7, we obtain the graph of Fig. 1.10. As
approaches 0, we can observe how the coefficients of related variables become closer to
one another. This outcome reveals how Ridge responds to collinearity.
Fig. 1.10 The execution result of Example 8. The closer is to 0 (the closer the model is to Ridge), the

more it is able to handle collinearity, which is in contrast to the case of (Lasso), where the

related variable coefficients are not estimated equally

1.7 About How to Set the Value of

To set an appropriate value for , the method of cross-validation (CV) is often used.3
For example, the 10-fold CV for each divides the data into ten groups, with 9 of
them used to estimate and 1 used as the test data, and then evaluates the model.
Switching the test group, we can perform this evaluation ten times in total and then
calculate the mean of these evaluation values. Then, we choose the that has the
highest mean evaluation value. If we plug the sample data of the target and the
explanatory variables into the CV procedure, it evaluates each value of and returns
the of the highest evaluation value as an output.
Fig. 1.11 The execution result of Example 9. We observe that was the best value

Example 9 We apply the function cv.glmnet to the U.S. crime dataset from
Examples 2 and 5, obtain the optimal , use it for the usual Lasso, and then obtain the
coefficients of . For each , the function also provides the value of the least squares of
the test data and the confidence interval (Fig. 1.11). Each number above the figure
represents the number of nonzero coefficients for that .

For the elastic net, we have to perform double loop cross-validation for both .
The function cv_glmnet provides us the output cvm, which contains the evaluation
values from the cross-validation.

Example 10 We generate random numbers as our data and try conducting the double
loop cross-validation for .
 For the problem, we changed the values and compared the evaluation for the best
to find that the difference is within 1% through .
Exercises 1–20
In the following exercises, we estimate the intercept and the coefficients
from N groups of p explanatory variables and a target variable data
. We subtract from each the
and from each the such that each of

them has mean 0 and then estimate (the estimated value is denoted by ). Finally, we
let . Again, we let and

1. Prove that the following 2 equalities hold.

Moreover, when is invertible, prove that the value of that minimizes

is given by . Here, for

, we define . Then, write a program

for a function called linear with Python. This function accepts a matrix

and a vector as inputs and then calculates the estimated intercept

and the estimated slope as outputs. Please complete blanks (1), (2) below.

Here, the function centralize conducting mean-centering to X, y is used. In the

3rd and 4th lines below, X, y has already been centered.

2. For a function that is convex (downward), for , we have

 The set of (subderivative) is denoted by . Show that the following
function f is convex. In addition, at , find .

3.
(a)

(b)

Hint For any and , if holds,

then we say that f is convex. For (b), show that .

(a) If the functions g(x), h(x) are convex, show that for any , the function
is convex.

(b) Show that the function is not convex.

(c) Decide whether the following functions of are convex or not. In addition,
provide the proofs.
(i)

(ii)

(iii)

Here, denotes the number of nonzero elements, denotes the sum of

the absolute values of all elements, and denotes the square root of the sum
of the square of each element. Moreover, let .
Discovering Diverse Content Through
Random Scribd Documents
 other. Cowpox appears where there are no horses, or possible contact
with horses; and may affect a number of cows in a dairy while the horses
are entirely free from Horsepox. [159]

At this point comes the tug of war. If cows have pox, how do they
contract the malady? Speaking at the London Conference on Animal
Vaccination in December, 1879, Professor J. B. Simonds, Principal of
the Royal Veterinary College, said—
My contention is, that the existence of Cowpox has to be proved.
Jenner’s account of the disease was an illusion. In my experience among
animals for forty years, I have never seen a case of Cowpox, and I do not
believe that any form of variola belongs to the bovine race. Sheep are
afflicted with pox, but not cattle. We hear of Cowpox, but who ever heard
of Bullpox? And is it credible that a disease should be confined to cows
and never attack bulls and steers? Let any one point out an affection of
females that does not extend to the males of the same species.

Professor Simonds and others believe that Cowpox as described

by Jenner was a parasitic affection of Smallpox, probably
communicated by milkers; and that Ceely, Badcock, and others did
intentionally, what milkers had done inadvertently, when they
inoculated cattle with Smallpox in order to create virus for
vaccination. On the other hand, those who assert the independent
existence of Cowpox, hold no terms with this heresy. As Dr. Cameron
says, “We can no more make Smallpox into Cowpox than by stunting
an oak-tree we can make it a gooseberry bush.” Fortunately I have
no call to pronounce judgment on the controversy. The more it
rages, the better I like it, and if the combatants disposed of each
other as did the Kilkenny cats, I might not be very sorry.
A last word as to Horsepox. There seems to be little doubt that
when inoculated on man it gives rise to vesicles indistinguishable
from those raised by Cowpox. In 1863 Professor Bouley of Alfort
produced pox on a cow by inoculating it with pox from a horse, and
children were successfully vaccinated with the virus. In the
Transactions of the Clinical Society, Vol. X., Mr. John Langton,
describes the case of a groom who came to St. Bartholomew’s
Hospital, 20th March, 1877, with an eruption caught from a horse
exactly like that induced by vaccination; and there could be no
question, says Mr. Langton, that the disease was the same as that
described by Jenner as grease.
There is much virus in currency as vaccine that is equine, and
many of us are equinated who suppose ourselves vaccinated; and it
might be argued that we have been saved from Smallpox by reason
of our equination. Why with all the notorious failures of vaccination,
and of re-vaccination, some of the more audacious medical quacks
do not recommend Horsepox as an infallible alternative, is not easy
to understand. It would be a Napoleonic stroke; nor is it improbable
that before vaccination is surrendered the attempt will be made.
How easily it might be asserted that vaccination is a failure in so far
as it has lost the original virtue of equination, that the remedy is to
dismiss the cow and revert to the horse, from whose poxy heels, as
the immortal Jenner observed, there issues “the true and genuine
life-preserving fluid.” The oracle might be worked thus—
“Let us hear no more of pure lymph from the calf, too often, alas!
an illusion. Sure and certain salvation from Smallpox can only be
guaranteed to those inoculated with pure pox from the horse. Come
then to the horse, the horse with pox! Come quickly! Come
yourselves! Come with your wives! Come with your children! Come
and be saved by Horsepox from the loathsome pestilence that
decimates the human race and brings myriads to untimely graves!”
 FOOTNOTES:
[142] As tradesman, however, Jenner was not honest. He took to
market what was not his to sell. The introduction of Cowpox for
inoculation (stigmatised in The Inquiry as spurious, not being
derived from Horsegrease) was effected by Pearson, and Jenner’s
claim to it was an act of piracy.
[143] Study of Medicine, vol. iii. p. 59.
[144] Life of Jenner, vol. i. p. 248.
[145] Treatise on the Cowpox. By John Ring. P. 336.
[146] Baron’s Life of Jenner, vol. i. p. 428.
[147] Copeland’s Medical Dictionary.—Art. Vaccination.
[148] Histoire de la Vaccination. Par Jean de Carro. Vienne, 1804.
[149] Baron’s Life of Jenner, vol. ii. p. 388.
[150] Ib. p. 388.
[151] Baron’s Life of Jenner, vol. ii. p. 390.
[152] Ib. p. 226.
[153] Ib. p. 226.
[154] Account of Varioloid Epidemic in Edinburgh. By John
Thomson. London, 1830.
[155] Baron’s Life of Jenner, vol. ii. p. 227.
[156] Ib., p. 399.
[157] Baron’s Life of Jenner, vol. ii. p. 390.
[158] Baron’s Life of Jenner, vol. ii. p. 456.
[159] Lancet, 29th May, 1880, p. 834.
 CHAPTER XIX.

JOHN BIRCH.

It is part of the Jennerian legend that the introduction of vaccination

was resisted by prejudice, fury and fanaticism, and that the practice
made its way by sheer force of its proven efficacy. The statement is
widely at variance with facts. Vaccination was accepted with instant
acclamation by the medical profession, the royal family, and the
public as an infallible and harmless preventive of smallpox; and the
subsequent course of experience was to disprove alike its
harmlessness and infallibility. That in some cases vaccination was
encountered with absurdity and violence lay in the nature of things,
even as it was advocated with absurdity, violence and prevarication.
It is always easy to raise a laugh by the exhibition of the
extravagance of either side in a hot dispute, but to what purpose? It
would have been no cause for surprise if some had been moved to
scorn by the facile credulity with which Jenner’s magical prescription
was so rashly accepted, but the world to which he appealed had no
scientific acquaintance with the laws of health, and it was in nowise
marvellous that, convinced of the prophylaxy of inoculated smallpox,
they should have been overcome by the plausibility of inoculated
cowpox. Yet were not all overcome, nor were all who resisted the
popular craze furious. There was John Birch, for example, surgeon to
St. Thomas’s Hospital, who with calmness and cogency steadily
protested against the introduction of “the new disease styled
cowpox;” and we may read his letters and pamphlets and fail to note
a fiery epithet or unkindly imputation. People who talk as if all who
opposed Jenner were steeped in ignorance and perversity can know
nothing of John Birch.
 Although satisfied with variolous inoculation, he had no objection
to vaccination in itself. He thought it fair that experiments with
cowpox should be tried, and the verdict of experience submitted to;
but he complained that experience was anticipated and success
proclaimed ere it was possible for the truth to be known, whilst
every objector was overwhelmed with abuse. As an illustration of the
unwarrantable persuasion that prevailed in favour of the new
practice before there was time to justify it, Birch mentions that at
the anniversary dinner at Guy’s Hospital in 1802, he was surprised to
find the usual business set aside to secure signatures to Jenner’s
petition for a vote of money from Parliament, and that after dinner
toasts, songs, and compliments in honour of Vaccinia were the order
of the day. Booksellers, he relates, declined to publish anything
against vaccination, and editors of newspapers and magazines would
not suffer a word to appear to its disparagement. Even the Post
Office carried the cowpox and correspondence of the Royal
Jennerian Society gratis until the collapse of the concern in 1806.
Those who resorted to doctors and hospitals for inoculation with
smallpox got cowpox instead in spite of assertions to the contrary.
Church vied with chapel in recommending the new practice. The
Archbishop of Canterbury was called upon to issue a letter directing
the clergy to recommend vaccination from the pulpit, but, with the
wariness of office, sent his chaplain to Birch to hear the other side,
and the chaplain retired with the judicious observation, “His Grace
must not commit the Church.” Many clergymen, however, not only
preached vaccination, but practised it with restless assiduity.
Erasmus Darwin was not without hope that baptism and vaccination
might be associated. He wrote to Jenner from Derby, 24th February,
1802—
As by the testimony of innumerable instances, the Vaccine Disease is so
favourable to young children, in a little time it may occur that the
christening and vaccination of children may always be performed on the
same day.

The Vaccine Disease so favourable to young children! The

assertion affords a vivid glimpse of the prevalent enchantment. “The
idea of connecting religious services with vaccination,” says Baron,
“had occurred to several individuals in this country as well as on the
Continent.”[160]
I viewed with indignant scorn [wrote Birch], the ungenerous artifice
adopted by the Jennerian Society of sticking up in every Station House, in
the Vestries of fanatical Chapels, and in Sunday Schools, that false
Comparative View of the effects of Smallpox and Cowpox representing to
the gaping multitude a frightful picture of Inoculation with the supposed
misery attendant on it; and exhibiting representations equally false and
exaggerated of the blessings of Vaccination.

The women were not behind the clergy in diffusing vaccine

salvation. They were Jenner’s most devoted allies. He took pains to
teach ladies to operate with “a light hand” so as not to draw blood,
and boasted that one of his pupils had ten thousand patients to her
credit, rescued from the terror and peril of smallpox!
As Birch observed, it was not a question of medicine or of surgery
that he and others had to deal with, but an outburst of enthusiasm
in which the methods and arguments of science were swept
heedlessly away. Any testimony to the credit of vaccination was
accepted with alacrity, whilst the facts to its discredit were denied or
explained away. This recklessness of procedure was most painfully
manifest in the conduct of the Committee of the House of Commons
which sat on Jenner’s first petition for money in 1802—
The number of witnesses in support of the application [wrote Birch] was
40, but out of the forty 28 spoke from mere hearsay, and not from
knowledge acquired in practice; while the three who spoke against it were
heard impatiently, though they corroborated their evidence with proofs.

Birch wished to know what cowpox was. Jenner had said it was
derived from horsegrease, but “that origin is proved to be erroneous,
and is now given up, even by his best friends. On all hands it is
admitted,” Birch continued, “that it is not a disease of the cow, but
communicated to the cow by the milker. No cow that is allowed to
suckle her own calf ever has the complaint.” What, then, is the
disease in the milker? asked Birch. Is it smallpox? Is it lues venerea?
Is it itch? A man came to St. Thomas’s Hospital from an adjacent
dairy with a hand and arm covered with ulcerations. He said several
of the milkers and the cows’ teats were affected in the same way,
and he was told they had got cowpox. Birch called one of his country
pupils and asked him what was wrong with the man. “It is itch—rank
itch,” was his reply. A box of Jackson’s ointment for the itch was
given to him, and at the end of a week he reappeared at the
Hospital cured. If cowpox be itch, argued Birch—
Then if a patient be inoculated with the disorder, though it may suspend
the capacity for Smallpox for a season in the constitution, it will ultimately
prove no security—

Which was to say, that it was not probable that smallpox and itch
could occur together, and that a person inoculated with itch would
pass through the variolous test successfully. In this connection we
may recall the fact that Jenner found it impossible to vaccinate a
regiment at Colchester, the men with their women and children all
being afflicted with itch.
Still farther to complicate the mystery of cowpox, Jenner began to
describe it as genuine and spurious, but which was the one and
which was the other he left in bewildering uncertainty. Said Birch—
Though Dr. Jenner could not tell us what Cowpox was, he soon came
forward to inform us that it was of two sorts—the one genuine and
harmless, the other spurious and hurtful.
Spurious Cowpox is a term I do not admit of. I know of no such thing as
spurious Smallpox, spurious Measles, spurious Lues Venerea, spurious
Scrofula.

Birch’s objection to spurious cowpox was forcible, but what in the

innocence of his heart he took for a blunder was proved out of
Jenner’s own mouth to be a deliberate dodge in 1807. Pressed by
the Committee of the College of Physicians to explain what he meant
by Spurious Cowpox, he had to own that he knew nothing of such a
malady, and that he had only meant to describe irregular effects of
cowpox on the arms of the vaccinated! In other words, when
vaccination turned out badly, he had found it convenient to ascribe
the disaster to spurious vaccine! The policy revealed in this
shameless avowal was cynically justified by Dr. Maunsell, who, in a
well known volume, wrote—
The term imperfect or Spurious Vaccination is frequently to be met with
in books, and has been the cause of no small degree of confusion in
practice, although, at the same time, it has frequently afforded the
practitioner an excellent asylum against the storms now and then arising
out of failures in the protective power of the vaccine disease.[161]

From out the muddle as to the origin of cowpox and its genuine
and spurious varieties, Birch demanded, What had Jenner
discovered? It is not that cowpox prevents smallpox; for that has
been asserted by dairy-folk for generations, and has been
disregarded by physicians because proved to be untrue. What then
is it? Let him define his discovery that we may know how to respect
it. Let him explain why it is forbidden to inoculate direct from the
cow. Is genuine cowpox invisible and to be taken on trust? Or is the
disease so virulent on its first communication that it has to be
meliorated in the body of some victim ere it is fit for public use?
Birch asked these questions as we continue to ask What is cowpox?
Is it a disease of the cow? Or is it communicated to the cow by man
or by horse? However definite the answers, the contradictions are
equally definite, and the authorities equally trustworthy.
Practical men answered for Jenner, as they presume to answer at
this day, “Whatever may be the origin of cowpox, we know that
vaccination is harmless, and that it prevents smallpox; and more we
neither demand nor care to inquire.” The credulity and conceit of
such practical men is that stupidity against which, says Goethe, even
the gods are powerless. It was practical men who “on the mere
show of reason” accepted vaccination before it could be tested, and
on most superficial evidence, said Birch—
Recommended Dr. Jenner to the munificence of Parliament for a
discovery in practice which was never to prove fatal; which was to excite
no new humours or disorders in the constitution; which was to be, not
only a perfect security against Smallpox, but would, if universally adopted,
prevent its recurrence for ever.

The harmlessness of the practice was soon belied—

 It gave rise [wrote Birch] to new and painful disorders. It was
sometimes followed by itchy eruptions; sometimes by singular ulcerations;
and sometimes by glandular swellings of a nature wholly distinct from
scrofula, or any other known glandular disease. Eruptions of the skin are
most frequent, and may be heard of in every parish of London; and
whether Vaccination shall be called the Cow Evil, or the Jennerian Evil,
posterity will have to determine.

The non-fatality of the practice was also speedily confuted. The

disorders it excited caused numerous deaths—from erysipelas
especially. It was then said, as it continues to be said to this day,
“Yes, but it was not vaccination, but erysipelas the patient died of”—
a form of words that seems to satisfy many minds accounted
rational. Birch mentioned three or four cases of death resulting from
vaccination, and adds—
These cases were as favourably palliated and ingeniously excused as
they could be; but it is admitted that each patient was punctured by a
lancet infected with what is called Cowpox; each arm so punctured
became inflamed and ulcerated, and each patient died.

The Variolous Test, used so unscrupulously to win converts to

vaccination, was proved by the Inoculators to be untrustworthy.
They had no difficulty in variolating the vaccinated. When it was
discovered that vaccination was no guard against smallpox, many of
the vaccinated resorted to inoculation with smallpox, and they “took”
as readily as did their unvaccinated acquaintance. Five in one family,
the Hignells of Cheltenham, vaccinated by Jenner were variolated by
Mr. Freeman, and smallpox resulted in the ordinary course. Nothing
indeed became plainer than that the vaunted Variolous Test was a
mere conjuring trick, and the more judicious vaccinators ceased to
refer to it.
The promise that the vaccinated would remain for ever secure
from smallpox Birch had no difficulty with. Londoners vaccinated by
the most approved operators caught smallpox, and died precisely as
did the unvaccinated. “Every post,” said Birch in 1804, “brings me
accounts of the failures of vaccination.” As the failures multiplied, so
did the excuses. There was the prime excuse of genuine and
spurious cowpox. If vaccination failed, it could only be through the
inadvertent use of spurious vaccine. Jenner had taught that one
puncture was all-sufficient for protection, but as one was not found
effective, it was asserted that two or three were requisite for
absolute safety. Many, it was alleged, had been imperfectly operated
upon, and the practice of the women and clergy and other
busybodies was thrown into discredit, although at the outset their
services and testimonies had been blazoned abroad as indisputable;
but Birch made this conclusive answer—
It cannot be meant to class Mr. Wachsel, Apothecary to the Smallpox
Hospital, or Mr. Bing, the Accoucheur, among ignorant and equivocal
practitioners; and yet from the patients vaccinated by these two persons I
could bring instance of more failures, more deaths, and more diseases
than have occurred in the practice of any other two persons who have
come within my knowledge.

Many, moreover, who had been vaccinated by Jenner fell victims to

the disease, and he was so pestered with awkward questions, says
Birch, “that to avoid the perplexing appeals that were made to him
daily, and the messages that were perpetually sent requiring him to
visit untoward cases, he retired from London.” Subsequently he had
to forsake Cheltenham for the same reason. The convictions of
quackery were too numerous for his endurance.
Having proved that vaccination did not prevent smallpox, whilst it
was a frequent cause of illness and death, Birch held up to derision
the fine promises wherewith its advocates had beguiled the people—
Were an architect to undertake to build an edifice which should be firm
in its foundation; all its rooms wind and water tight; and such as might be
inhabited with perfect security; if, before the edifice were well finished,
the foundations were discovered to be rotten; and if in less than seven
years, several apartments had fallen in and killed those who occupied
them, while in a great number of rooms, the wind or rain was continually
beating in, could I be blamed for declaring that the architect had broken
his contract, and that the edifice ought no longer to be inhabited?
Certainly not. Why then am I to be told that I am acting perversely when I
remonstrate against the practice of Cowpox? for such an edifice as I have
described, so rotten in its foundations, so ill built, so ruinous, is
Vaccination.
 Those who take success as the test of truth may say that Birch
was unsuccessful in his contention; but he was not unsuccessful.
Vaccination in London was discredited, and the imposture abated, as
the report of the College of Surgeons in 1807 attests. Where
retained, it was not so much as a preventive as a mitigator of
smallpox, its advocates being content to occupy the safe position
that it made milder a disease the severity of which was unknown.
Birch died in 1815. His sister reprinted his papers against
vaccination (from which have come my citations[162]), and erected a
monument to his memory in St. Margaret’s, Rood Lane, Fenchurch
Street, the inscription on which is noteworthy.
SACRED
To the Memory of
JOHN BIRCH, Esquire,
Many years an eminent Surgeon of this Metropolis;
who died on the 3rd February, 1815,
Aged 69 Years,
and whose earthly remains lie deposited under the Pulpit and Desk.

In his professional Character,

As humane as he was skilful,
He permitted not the daily sight of wounds and sores,
Afflictions and wretchedness of every kind,
To blunt the edge of his natural feelings,
For the sufferings of his Fellow creatures:
But, contemning a too hasty reliance on vaunted Theories,
Sparing of the Knife—abhorring unnecessary Torture—
A foe to wanton, cruel, or dangerous experiment,
Averse from rash operation, and the destruction of parts,
Redeemable by patient and judicious care—
He erected for himself a high and distinguished reputation,
On the solid, and only secure Basis of
ENLIGHTENED EXPERIENCE:
Stimulated throughout Life by a wise and Christian-like
Ambition, to cure, not maim—preserve, and not destroy.

Mankind is indebted to him

For a more intimate acquaintance with the powers
Of MEDICAL ELECTRICITY;
By his own ingenious and improved application of which
 He performed many remarkable and almost unhoped-for cures.
But the Practice of COW-POXING,
Which first became general in his Day,
Undaunted by the overwhelming influence of
Power and Prejudice, and the voice of Nations,
He uniformly, and until Death, perseveringly opposed;
Conscientiously believing it to be a Public Infatuation,
Fraught with peril of the most mischievous consequences to Mankind.
Whether right or wrong, Time will most surely determine:—
MAN’ S MERE OPINIONS MUST EVER BE LIABLE TO ERROR;
BUT BY THE MOTIVES THAT SWAY HIS HEART
SHALL HE ALONE BE JUDGED.

To perpetuate the remembrance of Qualities so excellent,

PENELOPE BIRCH,
His affectionate and only surviving Sister,
Hath raised this Monument:
Not out of a worldly and vain-glorious
Pride of Affinity;
But in order to hand down an Example worthy of Imitation
To succeeding Ages.
Jenner recognised Birch as a dangerous antagonist, and behaved
toward him with his usual meanness. Writing from Berkeley, 11th
October, 1812, to Moore in London, where smallpox was prevalent,
he observed—
I have not heard lately whether the fury of the Smallpox is abated in
town. I trust it is. Had I power to exercise vaccination as I liked, in one
fortnight this dismal work of death should entirely cease. What a sad
wicked fellow is that Birch! Moseley I hear nothing of now, but Birch is still
employing his agents to spread the pestilence. [163]

Birch a sad wicked fellow employing agents to spread pestilence in

London, whilst the good Jenner, capable of arresting the dismal work
of death, sat impotent at Berkeley! Comment is superfluous. Quack,
malicious and impudent, is written at large.
 FOOTNOTES:
[160] Life of Jenner, vol. i. p. 541.
[161] A Practical Treatise on the Management and Diseases of
Children. By Richard T. Evanson, M.D., and Henry Maunsell, M.D.,
Professors in the Royal College of Surgeons, Ireland. Dublin,
1838.
[162] An Appeal to the Public on the Hazard and Peril of
Vaccination, otherwise Cowpox, by the late John Birch, Esq.,
together with his Serious Reasons for uniformly objecting to
Vaccination: and other Tracts by the same Author. 3rd Edition.
London, 1817.
[163] Baron’s Life of Jenner, vol. ii., p. 382.
 CHAPTER XX.

GOLDSON AND BROWN.

William Goldson, member of the London College of Surgeons,

practising at Portsea, published a pamphlet in 1804[164] wherein he
set forth a number of instances within his own experience of
smallpox following vaccination by infection or inoculation. He did not
turn against vaccination, but suggested that its prophylaxy might
neither be so certain nor so enduring as at first asserted.
Vaccination, he pointed out, had been carried into practice on a
wave of enthusiasm, and it was not unreasonable to expect that on
closer acquaintance some of the claims made for it should be subject
to modification. Indeed so much was already admitted; for failures
had led to the discrimination of spurious from genuine cowpox, and
to the issue of new instructions as to the period of taking vaccine,
“on which point, it was now said, depended the whole success of the
operation.” Thus what was originally set forth as an operation for
which any novice was competent, had developed into one of
considerable delicacy with serious liability to miscarriage. Goldson,
therefore, had fair reason to believe that his own observations and
suggestions would meet with candid consideration, and, if verified,
serve for general guidance in the practice of vaccination.
It is unnecessary to recite Goldson’s cases. Interesting at the time,
they are now commonplace. He found that inoculation with smallpox
was possible at an interval after vaccination, and that infection with
smallpox was equally possible under the like circumstances. One
case is noteworthy for its connection with Jenner. A seaman, named
Clarke, was successfully vaccinated on 4th November, 1800, and,
returning from a voyage to the West Indies was put to the variolous
test on 24th March, 1802, when he sickened with smallpox and was
sent to Haslar. To prove that his malady was really smallpox, several
persons were variolated from him. The Committee of the House of
Commons was sitting on Jenner’s first claim for public money, and
Goldson wrote to Jenner to come to Haslar and see Clarke for
himself; but Jenner was too astute to cumber himself with difficulties
at a time when so much cash was in question. The case was
mentioned to the Committee, but was treated as of no moment in
presence of what they were pleased to regard as overwhelming
evidence as to the perpetual virtue of vaccination.
Goldson’s was a modest pamphlet—conjectural rather than
demonstrative. He ventured to think it was possible that the efficacy
of vaccine might be weakened by transmission from arm to arm, and
that security might be restored by reversion to the cow—
The casual Cowpox is produced by virus immediately from the animal;
while the inoculated disease is the effect of new matter generated by the
action of the other on the human subject. Whether that new matter be
possessed of the power to produce the same permanent properties as the
parent virus, time alone can decide.
He likewise suggested that horsegrease might be inoculated on
the nipple of the milch mare, and the virus used for equination.
These and other points were advanced with a philosophic grace that
ought to have commanded respect; but, on the contrary, his
pamphlet was received with a howl of fury, and its author
denounced as an ill-conditioned fellow—ignorant, prejudiced, pig-
headed. It was safer to be pronounced anti-vaccinist than a
vaccinator and harbour doubt as to any article of the Jennerian faith.
Ring plied his bludgeon over the heretic, and Jenner wrote of him
with malicious insolence—“All his reasoning is erroneous;” “his
arrogance is increased by attention;” “he obstinately holds a veil
before his eyes, and will not behold the vaccine light;” “one might as
well contend with a blind man on the nature of a prism;” and so on.
Goldson’s offence was that he laid his finger on some of the weaker
points of vaccination; that his sight was too keen, and his reasoning
too cogent. At this day the questions between him and Jenner are
decided by vaccinators themselves in Goldson’s favour.[165]
Perhaps the most able attack on the practice of vaccination was
delivered by Thomas Brown, surgeon, Musselburgh; and it is much
to be regretted that his book, published in Edinburgh in 1809,[166] is
so little known at this day. Brown had accepted vaccination, carried
away, he admitted, by the common enthusiasm, and the unqualified
audacity with which its claims were asserted—
The practice was introduced and recommended to the public by its
Author as a perfect antidote and security against Smallpox without any
exception or reserve, and capable of banishing Variola from the catalogue
of human misery. I have no hesitation in confessing that I became an
early convert and advocate of the new practice; and it is now eight years
and a-half since I have uniformly advised and practised Vaccination, in
which period I may safely say, I have vaccinated upwards of twelve
hundred patients, and have only inoculated three at the positive request
of parents. This course I persevered in until the present time,
notwithstanding I met with several instances where it appeared to fail in
giving security; some about three years after the introduction of the
practice; a few more about two years ago; and those which make part of
the present volume within the last six months.

An epidemic, in which his own perfectly vaccinated patients fell

victims to smallpox, at last opened his eyes to the delusion in which
he was walking, and to the perversity with which he and others had
resisted the light of truth—
I am convinced from what has passed under my own observation for the
last three or four years, that we have been all guilty of rejecting evidence
that deserved more attention, in consequence of the strong
prepossessions which existed, from the very persuasive proof of
Vaccination resisting inoculation and exposure to infection, and from our
judgments being goaded and overpowered with the positive and arbitrary
opinions of its abettors. I am now perfectly satisfied, from my mind being
under the influence of prejudice and blind to the impression of the fairest
evidence, that the last time Smallpox was prevalent, I rejected and
explained away many cases which were entitled to the most serious
attention, and showed myself as violent and unreasonable a partisan as
any of my brethren in propagating a practice, which I have now little
doubt we must ere long surrender at discretion.
 When Brown first saw the vaccinated prostrate with smallpox, he
concluded that there must have been some mistake about their
vaccination; “for after Vaccination it was impossible to contract
Smallpox;” but the evidence of his senses gradually overcame the
phantasy imposed upon him, and like an honest man he proclaimed
his error, and verified the experiences whereby he had been
reluctantly corrected. He set forth with all particulars forty-eight
cases of smallpox following vaccination within his own immediate
cognizance, and though aware of many cases outside that
cognizance, he limited himself to what he could attest with personal
assurance. He knew he would be told that the vaccinations had been
imperfect, or that what he took for smallpox was some other
eruption—
It is strenuously contended [said Brown] by nearly every author, and by
almost every practitioner, that Vaccination is a perfect antidote against
Smallpox, if the disease be properly communicated; and Dr. Jenner and his
relative, Mr. G. Jenner, positively assert, that they have had not one
instance of failure in their own practice. They all therefore, and without
hesitation, refer the whole series of failures that have been brought
forward to the sweeping power of imperfect Vaccination, or to the
blindness and stupidity of the medical practitioner who cannot distinguish
between Smallpox and Chickenpox, a rash, or bug-bites.

Nor did Brown rest satisfied with proving that vaccination did not
prevent smallpox. He also showed the fallacy of the variolous test.
He adduced twelve cases in which vaccinated persons had been
variolated as if they had never been vaccinated. Also four cases in
which vaccination and variolation were effected simultaneously, the
diseases running their courses concurrently, proving there was no
antagonism between them; and since they could occur together,
what reason was left for supposing that one might not succeed the
other?
Having found liberty in the truth, he reverted to Jenner’s writings,
and reading them with opened eyes, he was not slow to detect and
to demonstrate the laxity of statement, the contradictions, and
absurdities with which they were pervaded. No reply was attempted:
no reply, indeed, was possible. The surgeons of the Edinburgh
Vaccine Institution issued An Examination of Mr. Brown’s Opinions
and Statements,[167] but they merely carped over non-essential
details, and left the main issues wholly unaffected. What they had to
show was that Brown’s patients were either unvaccinated, or had not
had smallpox; and unable to do this, they were unable to do
anything.
Brown remained victor. He did not overthrow vaccination, nor
restore variolation, but he did make an end in Scotland of confidence
in vaccination as an omnipotent safeguard against smallpox. The rite
continued to be practised on humbler terms: “it did no harm”: even
Mr. Brown allowed that it might keep off smallpox for a time: and
“there was reason to believe that it tended to make the disease
milder when it did occur.” Thirty years after his first publication, in
1842, Brown reaffirmed his position in a series of letters[168] to Dr.
George Gregory, a sympathetic friend, and advised a return to
variolation in view of “the acknowledged defects of the Jennerian
practice”—a dismal alternative. But it is in vain to expect any man to
be much in advance of his time: it suffices for honourable distinction
that he be in advance. When Brown commenced practice, smallpox
and other fevers were regarded as inevitable as storms and
earthquakes, and the knowledge with which we are now so familiar,
that they are engendered in foul habits and habitations, was for
practical purposes unknown. Our reproach is, that knowing so much
better, we surrender ourselves to a superstitious observance
conceived in days of darkness.
 FOOTNOTES:
[164] Cases of Smallpox subsequent to Vaccination, with Facts
and Observations read before the Medical Society at Portsmouth,
29th March, 1804: addressed to the Directors of the Vaccine
Institution. By William Goldson. Portsea, 1884.
[165] Baron’s Life of Jenner, vol. ii. pp. 338, 346, 348.
[166] An Inquiry into the Anti-Variolous Power of Vaccination; in
which from the state of the Phenomena and the Occurrence of a
great variety of Cases, the most Serious Doubts are suggested of
the Efficacy of the Whole Practice, and its Powers at best proved
to be only temporary. From which also will appear the Necessity
of and the proper period for again submitting to Inoculation with
Variolous Virus. By Thomas Brown, Surgeon, Musselburgh.
Edinburgh, 1809. Pp. 307.
[167] Report of the Surgeons of the Edinburgh Vaccine
Institution, containing an Examination of the Opinions and
Statements of Mr. Brown of Musselburgh on Vaccination.
Edinburgh, 1809.
[168] An Investigation of the Present Unsatisfactory and
Defective State of Vaccination, and the Several Expedients
proposed for removing the now Acknowledged Defects of the
Jennerian practice. In a Series of Letters addressed to Dr. George
Gregory, Physician to the Smallpox and Vaccination Hospital,
London; which also are intended as an Answer to the Queries of
the Academy of Science in Paris, proposed as the subject of a
Prize Essay. By Thomas Brown, formerly Medical Practitioner in
Musselburgh. Edinburgh, 1842, pp. 139.
 CHAPTER XXI.

MOSELEY, ROWLEY AND SQUIRREL.

It may be well to devote a chapter to those antagonists of

Vaccination who, though right in their contention against cowpox,
did more or less to discredit their cause by scurrility and
extravagance. The faults of these men are frequently adduced as
evidence of the absurd and brutal resistance with which Vaccination
was encountered, but it is forgotten how intense was their
provocation, and how the bad on one side was matched by the bad
on the other. It was a contest between Smallpoxers and Cowpoxers,
alike ignorant of the conditions of physical well-being. It is plain,
however, in the light of our later experience, that much that was
asserted by the Smallpoxers of the uselessness and harmfulness of
cowpox must have been exactly and painfully true, though
persistently and ferociously denied.
In the Edinburgh Review for October, 1806, appeared an article
entitled “Pamphlets on Vaccine Inoculation,” which may be taken as
a reflection of the state of the controversy at that date, and as an
index to the chief offenders against propriety. The article was written
by the editor, Francis Jeffrey, and was a product of that perspicuous
intelligence, which reduced to order whatever was subjected to its
action, in much the same way that a housemaid “sets to rights” a
library by ranging the books according to their sizes and bindings,
and assorting the papers so that they lie neatly disposed. As is the
habit of able editors, a view of the variolous controversy was evolved
that might be comfortably accepted and confidently repeated by his
readers—the evolution of such rational mirage being regarded for
the time as veracious matter-of-fact.
 First we may take the reviewer’s evidence as to the extent and
fury of the controversy—
The ample and public testimony offered in favour of Vaccination seemed
for a while to set the question at rest; and, except in a few obscure
pamphlets and communications to the medical journals, little was heard in
opposition to it, till 1804, when Mr. Goldson of Portsmouth published six
cases of Smallpox occurring after Vaccination, accompanied with
observations, calculated to shake the confidence which was now very
generally placed in the security of the Jennerian inoculation. These were
answered by Mr. Ring and others, who endeavoured to show that, in some
of his cases, Mr. Goldson’s patients had not had the genuine Cowpox in
the first instance, and that in others, they had not had the genuine
Smallpox thereafter. This part of the controversy was conducted with
temper, and with a reasonable degree of candour. In the end of the same
year however, Dr. Moseley published his treatise on the Cowpox, in which
the ravings of Bedlam seemed to be blended with the tropes of
Billingsgate. Dr. Rowley followed on the same side, and in the same
temper, with 500 cases of “the beastly new diseases produced from
Cowpox,” and attracted customers by two coloured engravings at the head
of his work of “the Cowpoxed, ox-faced boy,” and the “Cowpoxed, mangey
girl.” The battle now became general. The Reverend Rowland Hill
thundered in defence of vaccination—Dr. Squirrel leaped from his cage
upon the whole herd of vaccinators—Mr. Birch insisted upon stating his
serious reasons for objecting to Cowpox—Drs. Thornton and Lettsom
chanted pæans in its praise—Mr. Lipscomb strutted forward with a
ponderous, wordy dissertation on its failures and mischiefs; and Messrs.
Ring, Merriman, and Blair answered everybody; and exasperated all their
opponents by their intemperance and personality. Charges of murder and
falsehood were interchanged among the disputants without the smallest
ceremony; the medical journals foamed with the violence of their
contention; it raged in hospitals and sick-chambers; and polluted with its
malignity the sanctity of the pulpit and the harmony of convivial
philanthropy.
In the whole course of our censorial labours, we have never had
occasion to contemplate a scene so disgusting and humiliating as is
presented by the greater part of this controversy; nor do we believe that
the virulence of political animosity or personal rivalry or revenge ever gave
rise, among the lowest and most prostituted scribblers, to so much
coarseness, illiberality, violence and absurdity as is here exhibited by
gentlemen of sense and education discussing a point of professional
science with a view to the good of mankind. At one time, indeed, we were
 so overpowered and confounded by the clamour and vehement
contradictions of the combatants, that we were tempted to abandon the
task we had undertaken, and leave it to some more athletic critic to
collect the few facts and the little reasoning which could be discerned in
this tempest of the medical world.

Furious was the controversy, but why was it furious? There are
often great fights over little matters, but the reason is that the little
matters are vitally related to the self-love of the combatants; and
thus it was with the Cowpoxers and the Smallpoxers. The Cowpoxers
set out with the absolute assertion that whoever submitted to their
prescription would be secure from smallpox for life. Without proof, or
with powerful sham proof, the assertion was endorsed by the mass
of the medical profession, and there followed the conversion of the
community in that sort of faith-panic which is described by Carlyle as
Swarmery—
All the world assenting, and continually repeating and reverberating,
there soon comes that singular phenomenon called Swarmery, or the
gathering of men in swarms; and what prodigies they are in the habit of
doing and believing when thrown into that miraculous condition! Singular,
in the case of human swarms, with what perfection of unanimity and
quasi-religious conviction the stupidest absurdities can be received as
axioms of Euclid, nay, as articles of faith, which you are not only to
believe, unless malignantly insane, but are (if you have any honour or
morality) to push into practice, and, without delay see done , if your soul
would live. [169]

People thus enchanted do not like to be brought to their senses;

and medical men, who in 1800 attested the perpetual prophylaxy of
cowpox, were naturally very unwilling to be proved deceivers and
deceived. When cases of smallpox were reported as following
vaccination, they at first denied the possibility, saying that either
there had been no vaccination, or that the smallpox was not
smallpox. On the other hand, the Smallpoxers who had been snuffed
out by the Cowpoxers, revived in presence of the discovered
impotence of the new practice, and stoutly maintained, and cruelly
demonstrated that unquestionable vaccinations were followed by
unquestionable smallpox. It needs little acquaintance with human
nature to see unlimited elements of bitterness in these conditions. To
be convicted of imposture does not beget equanimity, nor
contradiction as to plain matter-of-fact; and thus convicted were the
Cowpoxers and thus contradicted were the Smallpoxers.
The Edinburgh reviewer described Dr. Moseley’s treatise on
cowpox as blending “the ravings of Bedlam with the tropes of
Billingsgate.” Some Billingsgate I concede, but not Bedlam at all.
Much however depends on the point of view. Vaccination if regarded
as a blessing in which the inspiration of heaven was consummated in
the salvation of the human race from smallpox,[170] resistance
thereto might appear, as Carlyle observes of creatures under
enchantment, as “malignantly insane.”
Dr. Moseley’s book,[171] it is to be allowed, was singularly
exasperating. He had spoken against cowpox from the outset, and
was charged with condemning that of which he knew nothing; to
which he cogently replied that he could scarcely know less than the
gang of medical men who attested its perpetual efficacy in the
newspapers in 1800 before they had any proper experience of it
whatever. If his scepticism was premature, what was their credulity?
Moseley had patience: no argument could be heard in the rage that
set in for the new salvation. “Cowpox, I admit, is not contagious,” he
said, “but cow-mania is.” When, however, in process of time it was
seen in hundreds of cases that cowpox conferred no immunity from
smallpox, he published in 1804 Lues Bovilla—a somewhat pompous
treatise, with frequent touches of superfluous learning, and
permeated with the irritating superiority of the true prophet—“You
see it has turned out just as I predicted.” Nor was he content to
make general assertions: he specified the names and addresses of
those who had been correctly vaccinated, or had taken cowpox from
the cow, and had subsequently suffered from smallpox with their
neighbours; also of cases of severe illness, injury, and death
resulting from vaccination. Bluster was idle in presence of such facts.
Even the Royal Jennerians had to eat humble pie, for in their Report,
dated 2nd January, 1806, we read—
 It is admitted by the Committee that a few cases have been brought
before them of persons having the Smallpox who had apparently passed
through the Cowpox in a regular way.

With so much admitted by such furious fanatics, what might not

be inferred!
Moseley was held in high esteem alike by the profession and the
public, and his judgment enforced by so much serious evidence
contributed heavily to the discredit of vaccination, and unfortunately
to the resumption of variolous inoculation. That the reaction was
extensive, especially in London, appears from numerous
contemporary testimonies, which Moseley confirms in saying—
The people at large are not to be reproached for putting their faith in
this splendid imposition on humanity; and to the credit of their
discernment and parental feelings, the middle and inferior classes have
taken precedence in renouncing the delusion. At this moment, unless
attacked by surprise, or with threats, or cajoled by artifice (all of which
have been practised on them) there are now none among them in London
and the adjacent villages who will expose their children to Cowpox
Inoculation.

Rowland Hill was a religious and philanthropic notable in those

days, and in common with many of his kind, was an enthusiastic
vaccinator. A leading spirit in the Royal Jennerian Society, he had the
school-room of Surrey Chapel constituted a vaccination station
whereat Dr. Walker officiated. Nor was he content to patronise the
practice, but was himself an energetic operator. Speaking at the
annual meeting of the Jennerian Society, 17th May, 1806, he said—
“With my own hands I have vaccinated upwards of 5000 persons,” and,
lifting up his eyes to heaven, exclaimed, “I solemnly declare before God, I
have not had a failure in a single instance. What then shall we say of the
false and daring publications of those who denounce the benign practice,
and how shall they answer for their conduct to their King, their Country,
and their God!”

Hill and Jenner were great friends. Hill visited Jenner at Berkeley,
and Jenner heard Hill when he preached at Cheltenham. Introducing
Jenner to a nobleman, Hill remarked—
 “Allow me to present to your Lordship my friend Dr. Jenner, who
has been the means of saving more lives than any other man.”
To which Jenner, being of a pious turn, sighed with meek effusion
—
“Ah! would I, like you, could say souls.”[172]
So committed and so possessed, Hill naturally resented the
growing distrust of vaccination. It cut him deeply to be supposed a
quack; and in 1806 he issued a pamphlet[173] relating his
experiences as a Jennerite, defending his practice, and denouncing
those who treated it despitefully. Moseley especially was subjected
to severe and contemptuous condemnation. Hill’s sanctimony and
virulence, his vigour and venom compose a piquant mixture, and if
we could tarry for amusement we might produce it abundantly from
a variety of elegant extracts. Consider, for instance, this his
adjuration, and its pitiful object—
Oh, the blessing of the Jennerian inoculation! Did ever man stand as
Jenner so much like an Angel of God, an instrument in the hands of Divine
Providence between the living and the dead till the plague was stayed!

Hill’s latent assumption throughout his discourse was—

First, that all must have smallpox; and
Second, that all the vaccinated who escaped smallpox, owed their
salvation to their Jennerisation.
It never apparently occurred to him that before Jenner was heard
of, many passed through life exempt from smallpox; nor,
consequently, did he inquire how they escaped; nor why, when
vaccination was introduced, escape should be placed to its credit.
The belief in the vicarious influence of vaccination comes out
strongly, too, in Hill’s pamphlet. Of Londoners there were then over
1,000,000, and of these, he says, at least, 100,000 had been
vaccinated, and with this effect—
Vaccination reduced the deaths from Smallpox in London to 10 per
week; but after the Inoculators had been making their clamours, the
 applicants for Vaccination diminished, and the deaths soon rose to 100 per
week.
Now can effrontery itself deny that the introduction of Vaccination was
the sole cause of reducing the fatality on the Smallpox list?

Thus one in ten being vaccinated, smallpox was reduced

throughout the unvaccinated 9-10ths; and as soon as the vicarious
operations dropped, up went the rate of mortality! Nor was Hill
singular in this persuasion. He cited his friend Dr. Lettsom as writing
to him, 25th March, 1806—
Vaccination was gradually lessening the mortality in 1804, when about
the middle of 1805 false reports against Vaccination gained very general
credit, and Vaccination was nearly suspended; the consequence was the
death of 1286 children in four months (September to December) or ten
every day, each of whom might now have been alive had the blessing of
Vaccination been accepted.

And again I find Lettsom wrote to Moseley, November, 1808—

The increase of births and decrease of deaths has added 3000 lives
annually to the population of London during the period that Vaccination
has been practised.

Talk evidently sincere, and widely repeated, but with how little
consideration for truth!
To return to Moseley. He was not the man to endure Hill’s
aggression submissively, and in a pamphlet entitled An Oliver for a
Rowland,[174] he made a terrific reprisal. The public were delighted
with it, and in the course of a few months it ran through ten
editions. Hill was generally regarded as a clerical mountebank, with
more impudence than piety, and to see him knocked over, kicked,
and rolled in the mire, was sport that carried many sympathisers.
Moseley’s opening address gives the key to the whole performance—
R owland ,—I bought your pamphlet, entitled Cowpox Inoculation
Vindicated, dated the 25th of March, 1806.
I paid a shilling for it. Rowland,—it is not dear. The same quantity of
folly, falsehood, and impudence, could not have been bought for twice the
money of any other Cowpoxer from the Ganges to the Mississippi.
 But let me ask you, Rowland, what could induce you to take up your
pen to attack me on the subject of Physic, who never attacked you on the
subject of Religion? Would it not have been more prudent in you to have
continued to expose yourself in your own trade in your own shop?
As to my learned friend Dr. Lettsom (who is never out of the way when
there is good to be done) being moved to instigate you, a Methodist
parson, to enter into a medical controversy—that can only be accounted
for by supposing he owes you a grudge, and put you into my hands for
payment.

Paid he was with interest—gross and Rabelaisian; and Hill, when

he had picked himself up and recovered his senses, discreetly retired
from the combat.
Spite of his pomposity and buffoonery, there was good sense and
humour in Moseley, and his resistance to the Jennerian mania was
not ineffective. As he wrote in 1808—
It is ten years since I began this Trojan war against Vaccinia; and if it be
not yet ended, I have at least the satisfaction to see that her original
troops are no longer able to defend her throne; and that the mobled
Queen with “a clout upon her head where late diadem stood,” has fallen to
a new dynasty of mercenaries. [175]

In Dr. Munk’s Roll of the Royal College of Physicians we read, that

Dr. Moseley was appointed physician to Chelsea Hospital in 1788,
“an office which he filled with the highest éclat for more than thirty
years”—until his death in 1819—
Though a shrewd practitioner, and undeniably a man of extensive
mental capacity and very considerable attainments, Dr. Moseley was a
violent opponent of Vaccination, on which his communications to the press
were incessant. They did little credit to his medical penetration, or his
qualifications as a dispassionate searcher after truth, and, happily for his
reputation, are now well nigh forgotten. [176]

Are they? For what else is Dr. Moseley remembered? So that a

man does his duty in the world, whether he be forgotten or
remembered is not worth a thought; but Moseley’s early and
steadfast resistance to the Cowpox Imposture will long constitute his
title to grateful recollection.
 Dr. William Rowley, Physician to the Marylebone Infirmary, also left
his mark in medical history as a determined opponent of vaccination.
He had seen the profession and the public go mad about so many
absurd novelties, that it did not surprise him that they should go
mad about cowpox: and after due experience and investigation he
delivered judgment on the craze and its pernicious effects in a
pamphlet entitled Cowpox Inoculation no Security against Smallpox,
[177] containing two coloured engravings representing the Cowpoxed
Ox-Faced Boy, and the Cowpoxed Mangey Girl. Much ridicule was
expended on these pictures, and to this day whoever wishes to be
funny and create giggle over the early resistance to vaccination tells
how one Dr. Rowley maintained that Jenner’s benign virus induced
the face of an ox on a boy; but like the majority of comic anecdotes,
it is untrue. The engraving represents a comely lad with a swelling
on the upper part of his left cheek, which was thought to give that
side of his face an ox-like expression. Many a medical practitioner
among the poor would at this day have little difficulty in presenting
living examples of affliction answering to Rowley’s pictures—and
worse. It was, moreover, the fear or fancy of many at the time that
inoculation with cowpox might beget bovine characteristics in the
human species, and the fear or fancy was turned to inevitable
account in jest and earnest. The jest is visible in some of
Rowlandson’s caricatures, and stories like this got into circulation—
A child at Peckham, after being inoculated with Cowpox, had its former
natural disposition absolutely changed to the brutal, so that it ran upon all
fours like a beast , bellowing like a cow , and butting like a bull .

In order to discredit Rowley, it is thought fair policy to connect him

with such nonsense, and to have it supposed that he rested his case
upon “the cowpoxed ox-faced boy:” it was far otherwise. He
diligently tracked the vaccinators, and accumulated 504 cases of
smallpox and injury after vaccination with 75 deaths, particulars
being accurately specified. Nor was he content merely to report what
he had ascertained. “Come and see,” was his forcible argument. “I
have lately had under my care,” he wrote, “some of the worst
species of malignant smallpox in the Marylebone Infirmary, which
many of the faculty have examined and know to have been
vaccinated.” His trust in “Come and see,” he still more powerfully
exemplified in an exhibition of the injuries inflicted by vaccination in
his Lecture Room in Savile Row in October, 1805. “Knowing,” he said,
“the caviling character of the Cowpoxers, I determined to leave them
no hole for retreat”; and therefore he brought together Joules, “the
ox-faced boy, who also had a terribly diseased elbow-joint”;
Marianne Lewis, the mangey girl, “who was covered with blotches
like a leopard”; “a load of children in a cart from the south of
London,” and others accompanied by their parents, and displaying
their various maladies, said, “Behold the effects of the new disease
that has been taken from the cow and implanted in humanity!” This
painful exposition was continued over two days, and as he records,
“the scene was truly affecting and distressing to all who witnessed
it.” An antagonist like Rowley is a serious factor in any controversy,
and we may estimate the havoc he wrought by the extreme anxiety
of the Jennerites to have him estimated by the supposed absurdity
of the ox-faced boy.
To a man of practical temper like Rowley, the enthusiasm with
which vaccination was at first advocated appeared akin to delirium—
I have been in some vaccination storms, and have had the buttons torn
off my coat, cloth and all, to convince me of the great and infallible
excellence of Cowpox. I have seen some of the vehement vaccinators
redden like a flame with fury, their lips quivering, their eyes starting out of
their heads, their mouths foaming, their tongues dropping hard words,
and their fists clenched like pugilists, ready to accompany their violent
wrath with other knock-down arguments. In such circumstances, mild,
investigating Philosophy quits the scene and leaves the field of battle to
the Bedlamites.

The fury had subsided in 1805, and Rowley held that many
medical men were deeply ashamed of the extravagance into which
they had been committed, but lacked courage to make frank
confession.
Rowley died in 1806, and the regard in which he was held was
manifest in the crowds who flocked to his funeral. In the Roll of
Physicians, Dr. Munk observes—
Dr. Rowley was a determined opponent of Vaccination, and obtained an
unenviable notoriety by his association with Dr. Moseley in opposing every
conceivable obstacle to the reception and progress of that invaluable
discovery.

The obstacles interposed were matters-of-fact, and as matters-of-

fact were recognised and prevailed.
The controversy that followed the introduction of Vaccination
“gave birth,” says the Edinburgh reviewer, “to an infinite number of
publications of all descriptions” from which he could only select the
most characteristic. Among these we find Dr. Squirrel, whose book is
described “as the most entertaining of the whole”—
We will venture to say, though we know it to be a bold assertion, that
there never was anything so ill-written, or so vulgar and absurd, produced
before by a person entitling himself a Doctor of Medicine. There is a
certain nimbleness and agility about him, however, which keeps us in good
humour, and he whisks about with such a self-satisfied springiness and
activity, that it is really enlivening to look on him.

Turning up Squirrel’s pamphlet[178] I find little or nothing to

warrant this description. It is not ill-written, if judged by the
standard of medical literature, and the “springiness” is a conceit of
the reviewer’s to sport with the Doctor’s name. My own impression is
that Squirrel was a dull fellow, jealous of cowpox as injurious to the
trade in smallpox inoculation, and that he availed himself of the
depression in the vaccination business to assert its superiority. He
admits, indeed, that he kept silent during the Jennerian furore, “but
the overwhelming torrent being gradually reduced to a feeble
current,” he reckoned that he “might now promulgate his opinion
with a reasonable hope of success.” He cites Jenner’s account of the
origin of cowpox in the greasy heels of horses, and proceeds to
argue that the disease is scrofula, which its inoculation is certain to
diffuse, whilst affording no protection from smallpox. He then
adduces a number of cases in proof that inoculated cowpox had not
averted smallpox, and had in several instances brought on serious
and fatal ailments.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

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

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebooknice.com