Basic Mathematics

for Biochemists
To Joy and Jenny
Basic
Mathematics
for Biochemists
ATHEL CORNISH-BOWDEN
Lecturer in Biochemistry,
University of Birmingham

LONDON NEW YORK

CHAPMAN AND HALL
 First published 1981 by
Chapman and Hall Ltd
11 New Fetter Lane, London EC4P 4EE
Published in the USA by
Chapman and Hall
in association with Methuen, Inc.
733 Third Avenue, New York NY 10017
© 1981 Athel Cornish-Bowden

ISBN-13: 978-0-412-23010-3

This title is available in both hardbound and paperback editions. The paperback
edition is sold subject to the condition that it shall not, by way of trade or
otherwise, be lent, re-sold, hired out, or otherwise circulated without the
publisher's prior consent in any form ofbinding or cover other than that in which
it is published and without a similar condition including this condition being
imposed on the subsequent purchaser.
All rights reserved. No part of this book may be reprinted, or reproduced or
utilized in any form or by any electronic, mechanical or other means, now
known or hereafter invented, including phytocopying and recording, or in any
information storage and retrieval system, without permission in writing from
the publisher.

British Library Cataloguing in Publication Data

Cornish-Bowden, Athel
Basic mathematics for biochemists.
1. Biological chemistry 2. Mathematics
I. Title
510'.2454 QD42
ISBN-\3: 978-0-412-23010-3 e-ISBN-\3: 978-94-011-6523-5
DOl: 10.1007/978-94-011-6523-5
Contents

Page
Preface ix

1 The Language of Mathematics

1.1 Introduction 1
1.2 Priority rules for operators 3
l.3 The summation sign 6
1.4 Functions 6
1.5 Constants, variables and parameters 8
l.6 Dimensional analysis 9
1.7 Plotting graphs 13
l.8 Precision 18
l.9 Problems 21
2 Exponents and Logarithms 24
2.1 Integer powers 24
2.2 Fractional exponents 26
2.3 Addition and subtraction of exponents 27
2.4 Logarithms 28
2.5 Common logarithms 28
2.6 Negative numbers have no logarithms 30
2.7 Natural logarithms 31
2.8 Logarithms to base 2 32
2.9 Exponential decay 34
2.10 Logarithms as a method of scaling 35
2.11 Products of equilibrium constants 36
2.12 Logarithms of dimensioned quantities? 37
2.13 Redox potentials 39
v
vi I Contents
2.14 Dependence of redox potentials on pH 42
2.15 Problems 46

3 Differential Calculus 48
3.1 Co-ordinate geometry 48
3.2 Slope of a curve 50
3.3 Rapid differentiation 52
3.4 Derivatives of sums and products 54
3.5 Derivative of a 'function of a function' 57
3.6 Derivative of a ratio 58
3.7 Higher derivatives 59
3.8 Notation 59
3.9 Maxima and minima 61
3.10 A note on terminology 64
3.11 Points of inflexion 65
3.12 Sketching curves 68
3.13 Problems 70

4 Integral Calculus 72
4.1 Increases in area 72
4.2 Definite and indefinite integrals 74
4.3 Simple integrals 78
4.4 Integral of Ilx 80
4.5 Differential equations 82
4.6 Numerical integration: evaluating the area under a curve 85
4.7 Problems 88

5 Solving Equations 91
5.1 Linear equations in one unknown 91
5.2 Rearranging equations 92
5.3 Simultaneous linear equations 93
5.4 Determinants 96
5.5 Quadratic equations 99
5.6 Graphical solution of equations 102
5.7 Newton's method 107
5.8 Approximate methods 109
5.9 Problems 113
 Contents / vii
6 Partial Differentiation 117
6.1 Meaning of a partial derivative 117
6.2 Exact and inexact differentials 118
6.3 Least-squares fitting of the Michaelis-Menten equation 120
6.4 Problems 124

Notes and Solutions to Problems 126

Index 132
Preface

Some teachers of biochemistry think it positively beneficial for

students to struggle with difficult mathematics. I do not number
myself among these people, although I have derived much personal
pleasure from the study of mathematics and from applying it to
problems that interest me in biochemistry. On the contrary, I think
that students choose courses in biochemistry out of interest in
biochemistry and that they should not be encumbered with more
mathematics than is absolutely required for a proper understanding
of biochemistry. This of course includes physical chemistry, because a
biochemist ignorant of physical chemistry is no biochemist. I have
been guided by these beliefs in writing this book. I have laid heavy
emphasis on those topics, such as the use of logarithms, that play an
important role in biochemistry and often cause problems in teaching;
I have ignored others, such as trigonometry, that one can manage
without. The proper treatment of statistics has been more difficult to
decide. Although it clearly plays an important part in all experi-
mental sciences, it is usually preferable to treat it as a subject in its
own right and not to try to incorporate it into a course of elementary
mathematics. In this book, therefore, I have used a few examples
from statistics to illustrate more general points, but I have not
discussed it for its own sake. To summarize, the book is directed
primarily towards students taking compulsory courses in mathe-
matics in the early stages of their training as biochemists, but I hope it
will also prove useful as a short revision text at later stages in the
study of biochemistry.
I should like to thank my wife Mary Ann for her encouragement
and support during the writing of this book, and for pointing out
various ways in which it could be made more comprehensible. I am
also grateful to my colleagues Geoffrey Bray, Stuart Ferguson, Baz
Jackson, John Teale and Chris Wharton for reading the manuscript
and suggesting many improvements to it.
January, 1981 Athel Cornish-Bowden
ix
1 The Language of Mathematics

1.1 Introduction
Lord Kelvin once remarked that all of science could be divided into
physics and stamp collecting. Although this rather patronizing
comment of a physicist is not one that will appeal to all biochemists, it
has an element of truth in it. A science can hardly claim to be a science
as long as it remains no more than a catalogue of unrelated
observations. Only when general laws can be proposed and tested by
experiment can it be said to have passed from mere description into
science. In chemistry the transformation from stamp collecting into
science corresponded with the development of thermodynamics and,
later, the atomic theory and theories of chemical bonding; in
biochemistry, the gradual realization that understanding of life
processes requires a foundation of physical chemistry and not just a
list of metabolic reactions has played a corresponding role. It is no
coincidence that mathematics has been central in all of these
developments, and it is now almost impossible to comprehend even
elementary biochemistry without a grasp of elementary mathematics.
Fortunately for non-mathematically minded biochemists, however,
the mathematics necessary for an undergraduate course in bio-
chemistry is nearly all elementary, and nearly all of it has been touched
on in every science student's previous education. Little more is
required, therefore, than to identify the parts of elementary math-
ematics that are important in biochemistry and to reinforce them with
appropriate examples.
In mathematics itself, the transformation from description into
science is paralleled by the development from arithmetic, which is
concerned with numbers and their manipulation, into algebra.
Arithmetic is very useful, but it is much too limited to satisfy all of the
needs of science. In arithmetic, every problem is a new and separate
2 / The Language of Mathematics
problem, and it is difficult to make useful generalizations and hence to
express scientific laws. Let us consider the simple biochemical
example in Table 1.1, which shows a set of rates of a reaction
measured at the substrate concentrations given. As it stands the table
is no more than a description of the results of a particular set of
measurements and as long as we treat the numbers just as numbers we
cannot make a general or useful statement about the enzyme to which
they refer. The table tells us the rates observed at substrate
concentrations of 5 mM and 10 mM but offers no guidance about
what rate to expect at 8 mM; it does not tell us whether the system
studied was behaving in accordance with some general law; it offers no
clue as to what general law there might be. To remedy these omissions
we must move beyond arithmetic into algebra, because only then will
we be able to recognize a pattern or regularity in the numbers, and
express it so that it can be recognized again if it occurs with another
system. If the rates in Table 1.1 are represented as v and the substrate
concentrations as s, then the following equation expresses a law that
defines all of the numbers in the table:
lOs
v=--
4+s
This equation has two advantages over the table: first, it allows a
summary of all of the information while occupying much less space;
secondly, it predicts what v values we might expect to observe at s
values that are not included in the table. For example, it answers the
question above by predicting that v = 6.67 jiM s - 1 when s = 8 mM.
This is clearly more useful than just listing a set of numbers recorded
on a particular occasion.
One can proceed one stage further with this example by noting that
the equation is typical of what is reported for many enzymes, and so if

Table 1.1 Observations from a kinetic experiment

Substrate concentration
(mM)

1 2.00
2 3.33
5 5.56
10 7.14
20 8.33
 Priority rules for operators / 3
we replace the numbers 10 (JlM s -1) and 4 (mM) by V and K m ,
respectively, we have an equation that expresses a generalization
about enzymes:
Vs
v=---
Km+ s
Again, replacing numbers with symbols has increased the generality of
what we want to say. In addition to the symbols v, V, sand K m , which
represent numbers, either particular ones or generalized ones, the
equation contains three operators: one is represented by the addition
sign +; one is shown by the horizontal line between Vs and Km + s;
and the third is implied by the juxtaposition of Vand s, but could have
been made explicit by writing V· s instead of Vs. Each operator
specifies something to be done to the numbers or symbols operated
on: the + sign requires Km and s to be added together; the horizontal
line requires Vs to be divided by Km + s; the juxtaposition of Vand s
(or a dot between them) requires them to be multiplied together.
As long as no ambiguity is possible mere juxtaposition is sufficient
to indicate multiplication, but if several numbers are to be multiplied
together, or if we allow algebraic symbols consisting of more than one
letter each (as in Fortran and many other computer languages), or if
ambiguity is possible for some other reason, multiplication can be
indicated by a dot or a cross. The dot is more common for pairs of
symbols (where no confusion with the decimal point is possible) and
the cross is more common for pairs of numbers, but the two symbols
have the same meaning in most contexts, i.e. V x s means the same as
V . s. (In some specialized applications, such as in vector algebra, it is
convenient to assign distinct meanings to . and x, but these need not
concern us in elementary biochemistry). In current usage the dot
should be written above the line and the decimal point on the line, e.g.
5.1· 8.7 = 44.37 not 5·5.8·7 = 44·37 but this is rather a recent
convention so far as British books are concerned: in older British (but
not American) work one is likely to encounter exactly the opposite
convention.

1.2 Priority rules for operators

To avoid ambiguity it is important to realize that operators have to be
obeyed in a proper order. Unlike ordinary language, equations are
not read from left to right but in accordance with priority rules that
4 / The Language of Mathematics
require certain operators to be obeyed before others. Thus, the value
of 5 x 3 + 2 x 4 - 3 is 20, not 25 or 65, because multiplication must be
done before addition or subtraction. In general, the rule is as follows:
(1) expressions within brackets must be evaluated first;
(2) if brackets are 'nested' (brackets within brackets), 'inner'
brackets must be evaluated before 'outer';
(3) exponentiation must be carried out before multiplication and
division;
(4) multiplication and division must be carried out before addition
and subtraction.
'Exponentiation' is the raising of a number or expression to a power,
as in x 2 , (x + y)a, etc. If a power is itself raised to a power we work
down from the top, i.e.
e- 2x ' means e(-2x'l not (e- 2X)2
As in this example, it is always permissible and often desirable to use
brackets to clarify an expression that might otherwise be mis-
interpreted. This is true even if the expression without brackets is
strictly unambiguous.
There are no rules of priority between addition and subtraction
among themselves, because the result of a sequence of additions and
subtractions is independent of the order in which they are done. This
is normally a matter of convenience only, although occasionally
numerical considerations may make one order better than another. In
principle, the same applies to multiplication and division, but greater
care is needed because thoughtless use of the slash / to indicate
division often results in expressions with meanings that are either
unclear or clear but different from what the writer intended. It is
wisest therefore to use the slash in moderation and to check carefully
that expressions have the meanings intended. Consider for example
the following equation:
Vs
v=------
Km(1 +i/Kj)+s
This is unambiguous, and the priority rule should prevent the
bracketed expression (1 + i/ K j ) from being misread as [(1 + i)/KJ
When there is more than a single term after the slash, however, as in
(i/ K j + 1), misunderstanding is more likely because it is not always
clear whether the slash indicates simple division or whether it is used
 Priority rules for operators I 5
to avoid the typographical inconvenience of a cumbersome fraction
such as _i_. Double slashes are so confusing that they should never
Ki+ 1
be used: this applies not only to ordinary algebraic expressions but
also the units of physical quantities, as in R = 8.314 J/moI/K. Here it
is not clear whether the K belongs in the numerator of the unit with
the J or in the denominator with the mol. To avoid this uncertainty
the definition should be written as follows: R = 8.314 J mol- 1 K - 1.
For reasons that will become clear in Chapter 2, mol- 1 and K - 1 have
the meanings (l/mol) and (11K), respectively. In general, slashes
should only be used in expressing units when there is only a single
term in the denominator.
Certain computer languages do not obey precisely the same
priority rules as conventional mathematics. This fact has generated
rather more confusion about the conventions than existed before
computers became widespread, and the appearance of cheap elec-
tronic calculators has made matters much worse in this regard
because those that use so-called 'algebraic notation' commonly ignore
mathematical conventions altogether and use a 'left-to-right' system.
An expression such as

A * A/B/C/D + AlB * C
would be unambiguous in a Fortran program and would have the
meaning

(The multiplication sign in Fortran is expressed as * and must be

explicit). This unambiguous meaning, which may nonetheless be
different from what the programmer intended, does not imply that
such expressions are acceptable in ordinary mathematics. Similarly,
the fact that simple calculators often disregard priority rules does not
mean that they are obsolete. For example, the expression 5 x 3 + 2 x 4
- 3 must be interpreted as (5 x 3) + (2 x 4) - 3 if the proper conven-
tions are followed, even though most simple calculators using so-
called 'algebraic notation' execute instructions as they are entered and
consequently interpret the above expression as {[ (5 x 3) + 2] x 4}
- 3 = 65 (this is indeed what I get if I key in 5 x 3 + 2 x 4 - 3 = on
my pocket calculator).
6 / The Language of Mathematics
1.3 The summation sign
It often happens, especially in statistical calculations, that we need to
add together a large number of terms of the same kind. For example, if
we have a set of values XI' X 2 , X3 . . . X., their arithmetic mean x is
given by
x = (XI +x 2 +x 3 + ... +x.)/n
In more complex examples explicit representation of the summation
becomes cumbersome and unnecessary, and it is more convenient to
use a special operator called the summation sign I (a capital Greek
sigma) instead:

IX i =X I +X 2 +X 3 + ... +X.
i= I

The limits i = 1 and n written above and below the sign mean 'start
adding at i = 1 and continue until i = n'. If the limits are obvious, as
for example in statistical calculations where one often has to sum over
all of a set of observations, they can be omitted.
Although the summation sign is a considerable convenience when
the underlying calculation is understood, it can sometimes obscure
the meaning when it is not. Indeed, one of the main reasons why more
advanced mathematics than one is familiar with can appear much
more difficult than it actually is, is that it often uses special notation to
express results more compactly. For example, the whole of matrix
algebra is a way of expressing very complicated relationships in an
extremely compact way: very convenient when one is familiar with the
symbolism but baffling when one is not. Whenever obscurity
threatens for this sort of reason it is often helpful to translate the
compact expressions into a more long-winded form, and then their
meanings are likely to become clearer.
A corresponding sign for products also exists, although it is much
less often encountered than the summation sign. It is written as the
Greek capital pi, Il, and is used in an exactly analogous way, i.e .
•
Il
i= I
Xi=X I X2 X 3'" X.

1.4 Functions
A mathematical function can be regarded as a set of instructions to
carry out a series of operations on a variable or set of variables. For
 Functions f 7
example, if we define v in terms of s as
Vs
v=---
Km+ s
where V and Km are constants, then we are defining v as afunction of s,
by defining what operations have to be carried out on s to obtain v. We
can also have functions of more than one variable. For example, v may
be determined not solely by a single concentraion s but may depend
both on s and on another concentration i:
Vs
v=------
Km(1 + if K i ) + s
and now we say that v is a function of both sand i.
Sometimes we may wish to symbolize the existence of a depen-
dence of one variable on another without specifying what the
dependence is. We then often use the symbol I( ) or something
similar, e.g.
v = f(s, i)
which states that v depends on sand i but does not indicate whether
the dependence follows the equation given above or some other
equation.
There are a number of functions that are so often required in
mathematics that they are given special symbols. For example, if y is
always given by multiplying x by itself we say that y is the square of x
and symbolize the relationship as
y = x2
where the superscript 2 indicates that 2 x's need to be multiplied
together. Conversely, x in this example is the square root of y, which
we may write as
x=Jy
or, more commonly and for reasons that I shall discuss in Chapter 2,
we may express the same relationship as
X = Y'l

Other functions of great importance are the logarithmic and exponen-

tial functions, which I shall also consider in Chapter 2, the derivative,
or result of differentiation (Chapter 3), and the integral (Chapter 4).
8 / The Language of Mathematics
There are others, such as trigonometric junctions, that are important in
mathematics generally, but have little application in elementary
biochemistry, and so I shall say little about them. On the other hand,
there are certain functions that have little importance in mathematics
as a whole but which are useful to define for biochemical purposes.
For example, various properties of proteins can be related to the
hydrogen-ion concentration [H +] in terms of the following kind of
expression:

in which y, K 1 and K 2 are constants. This kind of function was first

studied by Michaelis and it is consequently called a Michaelis
function.

1.5 Constants, variables and parameters

Some quantities, such as the number 2.0, have a unique value under all
circumstances and are called constants. Other numbers, such as the
gas constant R ~ 8.3 J mol- 1 K - 1 are found by experiment to be
constant also. Others, such as Km in the Michaelis-Menten equation,

v = Vs/(K m + s)

may be constant for a particular enzyme and substrate under well-

defined and constant conditions, although they may vary with
temperature, pH, etc. These quantities can be treated mathematically
as constants only as long as the physical conditions that determine
them are constant.
We are often interested in quantities that change when the
conditions change. For example, we may find that an equilibrium
'constant' K varies with the temperature Taccording to the van't Hoff
equation:

K = exp(AS O _ AHo )
R RT
where exp( ) is the exponential function (Chapter 2), and ASo, AHo
and R are constants. Thus although K may be constant at constant T
it varies with T and so if we are concerned with changes in
temperature we must treat K as a variable.
 Dimensional analysis / 9
It will be evident that the distinction between a constant and a
variable is not absolute, because any variable becomes a constant if
the conditions that determine its variation are kept constant. In some
contexts the distinction is so difficult or inconvenient to make that we
introduce a third term, parameter, to denote a quantity that is treated
as variable for some purposes and constant for others. This occurs
especially when we use statistical methods to estimate unknown
constants: although we may believe it to represent a true physical
constant we must treat it as a variable during the process of
estimation. Suppose, for example, that we have a measured quantity y
that depends on a controlled variable x according to the following
linear equation:
y = a+bx+e
in which a and b are the physical constants whose values we require
and the third term e on the right-hand side represents experimental
error. This error term prevents us from calculating a and b exactly
from measurements of y at two values of x. Instead, we would
measure y at several values of x and try to find values of a and b that
predicted the observed y values as closely as possible (I shall discuss a
similar problem in more detail in Chapter 6). During the analysis we
are free to try any values of a and b that we like: in other words, we
treat them as variables, even though we may believe that some
physical reality exists in which they have constant, but unknown,
values. To express this duality we use the word parameter for a
quantity such as a or b that controls the dependence between our
observable variables.

1.6 Dimensional analysis

Most of the quantities we measure in biochemistry are not simple
numbers but numbers with units. For example, the concentration of
glucose in blood is not 0.005 but 0.005 mol dm - 3. Put differently, and
very formally, we may say that the blood-glucose concentration has
the dimensions of (amount of substance) (length) - 3. Actually, this
combination of dimensions occurs so often in biochemistry that it is
convenient to define a unit of amount-of substance concentration, such
that 1 molar or 1 M == 1 mol dm - 3; and because amount-of-
substance concentration is the only kind of concentration the
10 / The Language of Mathematics
biochemist is usually interested in we usually abbreviate this cumber-
some term to concentration. Although strictly it is a derived quantity it
is often convenient to treat it as if it were a primary dimension.
Consideration of units and dimensions is sometimes regarded as a
pedantic nuisance, but this is a pity, because it is one of the most
powerful tools that scientists have for detecting mistakes in
algebra - not only their own but also other people's. This is because
there are rules that govern the way in which dimensioned quantities
can be combined and a high proportion of algebraic mistakes cause
these rules to be violated.
The simplest dimensional rule is that one cannot equate quantities
of different dimensions: it is meaningless to assert that a length of
3 cm is equal to a mass of 1 kg, for example. This may seem so obvious
as to be hardly worth mentioning, but it is surprising how many
fallacies in kinetics spring from an inability to appreciate that one
cannot compare a first-order rate constant with a second-order rate
constant, because their dimensions are different. An obvious exten-
sion of the first rule is that one cannot add or subtract quantities of
different dimensions or say that they are greater or less than one
another. On the other hand we can multiply them together or divide
one by the other: if an object has a mass of 1 g and a volume of 2 cm 3 it
is quite acceptable to divide 1 g by 2 cm 3 to obtain a density of
0.5 gcm- 3 .
F or certain mathematical purposes only pure numbers, i.e. numbers
with no dimensions, are admissible. For example, the expression 2i
means that i 2's have to be multiplied together, and this has meaning
only if i is a pure number: we cannot multiply 3 cm 2's together, for
example. In other words, dimensioned quantities must not appear as
exponents, and, similarly, their logarithms cannot be taken.
Sometimes we may wish to define a quantity, such as pH, that
appears to be the logarithm of a dimensioned quantity, namely the
hydrogen-ion concentration in the case of pH. However, as I discuss
in Chapter 2 (p. 37), we can only do this if we first remove the dimen-
sions by dividing by a standard value that has the same dimensions.
The value of all this is that when we make a mistake in algebra we
often introduce a dimensionally incorrect expression. So, if we arrive
at a result that we suspect may be incorrect, the simplest check we can
make is for dimensional consistency. Suppose, for example we are
trying to remember the Henderson-Hasselbalch equation (Chapter 3),
and think that it may be
 Dimensional analysis j II
pH = pK.+log{[salt] [acid]}
Is this likely? We can check by noting that [salt] and [acid] are both
concentrations and therefore their product is a concentration
squared. As this is not a pure number it cannot have a logarithm and
so the expression is wrong. If we are reasonably sure we have the right
ingredients but that we have combined them incorrectly, we can use
knowledge of the dimensions to deduce that we need a ratio of
concentrations, which would be a pure number, not a product. But do
we want [salt]j[acid] or [acid]/[ salt]? Here dimensional analysis
cannot help us, but our knowledge of chemistry can: we should know
that the pH decreases as a solution becomes more acid, and this
should guide us to the correct form of the equation:
[salt]
pH = pK. + logEacid]

If we have a long derivation that begins dimensionally correct but

ends at a dimensionally incorrect result, we know that it must contain
a mistake. How can we find it? One way might be to repeat the
derivation, but then we may well repeat whatever thought processes
led to the original mistake, in which case we shall not find it. A better
approach - better because it is simpler, quicker and requires new
thoughts rather than a repetition of old ones - is to check the
dimensions at various stages until the mistake is located.
Dimensional analysis is also useful for remembering the slopes and
intercepts of graphs. Anything measured along the x-axis, such as the
intercept of a line on the x-axis, must have the same dimensions as x;
anything measured along the y-axis must have the same dimensions as
y. Furthermore, the slope at any point must have the dimensions of y
divided by those of x. The reason for this will become clear in Chapter
3 when we consider the definition of a slope as the limit of a change in
the y co-ordinate divided by the corresponding change in the x co-
ordinate as one moves along the line. For the present we can accept it
as a fact without worrying about the reason.
The rules for applying dimensional analysis to graphs are il-
lustrated in Fig. 1.1. They allow us to check whether we have
remembered the slopes and intercepts of particular plots correctly,
and can guide us to the correct ones when we have not.
Some authorities advocate converting all variables into dimension-
less form before plotting them. For example, instead of plotting [S], a
12 / The Language of Mathematics

x
Intercept has dimensions of x
(al

,'', Slope of tangent has dimensions of !fIx

"

~ Intercept has
dimensions of !:I

Intercept has
dimenSions of X

x
(b)

Fig. 1.1. Dimensional analysis as applied to graphs: (a) straight line; (b) arbitrary
curve.

concentration, in mM, we would make it dimensionless by dividing it

by a 'standard' concentration of 1 mM, so that we actually plot
[S]/mM, a dimensionless variable. Similarly, one can avoid the
problems of defining thermodynamic quantities in terms of the
logarithms of dimensioned equilibrium constants by defining all
equilibrium constants in such a way as to make them dimensionless.
My own view is that the advantages of following these recommen-
dations are very slight and the disadvantages are considerable. If all
variables are dimensionless dimensional analysis becomes meaning-
 Plotting graphs / 13
less and one denies oneself the use of a valuable tool. One can also
very easily lead oneself into absurdities: by insisting tht equilibrium
constants are dimensionless, for example, one prevents the Michaelis
constant Km from being compared with the dissociation constant of
the enzyme-substrate complex unless one defines Km so that it is
dimensionless as well; but doing this causes the Michaelis-Menten
equation to contain a dimensionally incorrect addition of Km to the
substrate concentration - unless of course one is willing to have a
dimensionless substrate concentration too! One recent and widely
used book of problems in physical chemistry for biochemists falls into
this particular trap.

1.7 Plotting graphs

Biochemistry is not a visual subject. Unlike zoology and botany, and
to some extent even such mathematically based subjects as physics
and astronomy, it rarely presents the experimenter with much to see
directly. Nearly always we have to 'see' the properties ofa biochemical
system by way of numbers measured on an instrument such as a
spectrophotometer. The first stage in translating these rather abstract
observations into biochemical knowledge is often the drawing of a
graph to show how the observed variable depends on one or more
controlled variables. It is therefore vitally important for the
biochemist - more than almost any other kind of scientist - to be
familiar with the rudiments of plotting graphs so that they display the
information they contain in the clearest way.
Rather than spend a large amount of text discussing the ways of
plotting data well and badly I have chosen a number of examples to
illustrate common faults in plotting. These are shown in Fig. 1.2. In
(a), some observations are plotted with very poorly chosen scales on
both axes. Less than a fifth of the ordinate scale is actually used, with
the result that most of the graph is uninformative white paper. In (b),
the ordinate scale is much better, but because of the large gap in data
between x = 10 and 50 most ofthe observations are crowded together
on the left and there is still too much empty space. In fact most of the
information is in these low-x data points and the one at x = 50 was
probably measured to give an idea of the limit approached by y. There
is no real need to show the region between x = 10 and 50. We can then
choose a scale that will spread out the low-x points, but break it to
include the x = 50 point, as in (c). Whenever the scale is broken in this
 10

10 20 30 40 x
50

O~--~--~---L--~--~~

(b)
o 10 20 30 40
x
50

!f

(e)

Fig. 1.2. Common faults in plotting graphs: (a) wasted space caused by poorly
chosen scales; (b) crowding caused by poor choice of abscissa scale; (c) satisfactory plot
of the same data;
 (d)

(e)
x

!J3 !J3
•
•
2 •
•
•
•
o~--~--~----~. 00
o 2 3
x
If) (9)

Fig. 1.2. (d) spurious inflexion caused by failing to note a sudden change in scale;
(e) misleading axes suggest a much larger slope than actually exists; (f) straight line
drawn through points that demand a curve; (g) same data without the line;
16 / The Language of Mathematics

3
x
(h)

!I 0-8

0-6

0-4

0-2

0
4 6 8
(i) pH

Fig. 1.2. (Con/d.) (h) systematic lack of fit evident in all three plots; (i) points
connected by meaningless straight lines.

way, the labelling must make it obvious that the break exists, to avoid
giving a misleading impression. To connect the points with a line
across the break, as in (d), is grossly misleading.
In (e), the graph may be quite legitimate, but often graphs that
resemble this one are intentionally misleading and therefore dis-
honest. To the casual eye the plot suggests that changing x brings
about a substantial change in y. Only by checking the labelling does
one see that the change in y is very small. Very much worse than what is
shown here is to present the same sort of graph but to omit some or all
of the labelling. Politicians and advertisers are fond of this kind of
graph but it has no place in science.
 Plotting graphs / 17
The graph in (f) shows a very common fault: a straight line is drawn
through points that demand a curve. The same points are shown in
(g) without the line, and the curvature is somewhat more obvious: a
line tends to bias the eye and it is prudent, therefore, to examine a set
of points carefully before drawing a line through them. It is
illuminating to take any issue of any research journal of biochemistry
at random and search for examples of straight lines drawn through
points that do not fit straight lines. They are not hard to find. A useful
habit to cultivate is to examine all experimental graphs - your own
and other people's - with your eye close to the paper and looking
along the plotted line (Fig. 1.3). This greatly emphasizes any sys-
tematic failure of the points to lie on the line. In real experiments, of
course, there is always some experimental error, so we should not
expect an exact fit. But experimental error should be random, and so
experimental points should be randomly scattered about the line, not
systematically.
When there are very few points on each line, as in (h), it is very
difficult to distinguish between random scatter and systematic lack of
fit. One answer is of course to obtain more observations, but if several
lines show the same sort oflack of fit, as in (h), there is good reason to
believe it is systematic.
Finally, (i) shows a fault that seems to be particularly common in
plots of pH dependences. Although any reasonable model would
place the points on a smooth curve, they have been connected by

~ l~

L-'
(~
Fig. 1.3. How to inspect a graph for lack of fit.
18 / The Language of Mathematics
straight-line segments. This is not only unattractive, it also mis-
represents the form of the dependence, especially in regions of high
curvature: in (i) the maximum ought to be at pH 5.85, not at pH 6.1,
and it should be at a height of 0.82, not 0.79. In experiments where
there has been no attempt to account for the observations mathemati-
cally it may sometimes be justifiable to connect points by straight-line
segments simply to display them more prominently, but this should
never be done in work with any mathematical pretensions.

1.8 Precision
One of the commonest faults in scientific writing is to report results
with an inappropriate degree of precision. For example, if a sample of
100 ml of blood is found to contain 93 mg of sugar, the concentration
in mM (expressed in terms of glucose, of relative molecular mass 180)
is 1000 x 93/(100 x 180), which can be variously written as 5 mM (to
one significant figure), or 5.17 mM (to three significant figures), or
even 5.1666667 mM (to 8 significant figures), etc. Which of these is
appropriate for reporting the result? Common sense tells us that
5 mM is rather more vague than the data would permit, whereas
5.1666667 mM is ridiculously precise, and 5.17 mM seems a reason-
able compromise; but how can we justify such a compromise
logically? If we examine the four numbers used in the calculation, we
see that 1000, the factor for converting from M to mM, is exact and we
could write it as 1000.000 ... if we wished, 93 mg and 100 ml are of
unknown accuracy, and 180 is correct to 3 significant figures, i.e.
the true value is closer to 180 than to 179 or 181. Without knowing the
conditions of the experiment we cannot know how accurately the
mass and volume were measured, but it seems reasonable to guess that
the volume was chosen to be 100 ml by the experimenter and that it is
correct to at least 3 significant figures. This leaves the 93 g as the
probable main source of inaccuracy and the main one we have to
consider: just as a chain is no stronger than its weakest link, the result
of a calculation (in simple cases at least) is no more accurate than the
most inaccurate value used in it. Presumably the true mass was closer
to 93 g than to 92 g or 94 g, but there is no reason to suppose it was
93.00 g rather than, for example, 92.68 g, but this latter value would
give a cakulated concentration of 5.1488889 mM (to 8 significant
figures). Comparing this with the original value of 5.1666667 mM we
see that only the first two digits are reliable, though the third is not
 Precision / 19
worthless: consequently 5.17 mM seems a reasonable way of express-
ing the concentration, with the understanding that the last (or least
significant) digit is unreliable. To include any more than three
significant figures in this result would be to claim that it is more
accurate than it is. As a general rule one should express a result with
no more (or only a little more) precision than the accuracy of the data
from which it was calculated.
Including too many significant figures may seem to be a harmless
fault - after all, the reader can always ignore the unwanted digits. It is
not harmless, however, because it can mislead the reader into thinking
you have measured more than you have, and it can make you look
a fool. Expressing the concentration calculated above as
5.1666667 mM is just as silly as claiming that the fact that grass is
green proves that it contains chlorophyll. The observation and the
conclusion in this case are both true, and they are relevant to ont!
another; nonetheless, the conclusion does not follow from the
observation. Similarly, the true concentration of glucose could
conceivably be 5.1666667 mM, but the available information does
not justify saying so.
Inaccuracy can result not only from inaccuracy in the data; it can
also arise from imprecise arithmetic during the calculation, especially
if several steps are involved. It is advisable therefore to carry out the
intermediate stages of a calculation with more precision (more
significant figures) than one expects to retain at the end. However, it is
not only very tedious to carry out every step with, for example, fifteen
significant figures if only three are to be retained at the end; it is also
unnecessary and it encourages mistakes. A fairly reliable rule is to use
one more significant figure in the calculations than one expects to
retain in the final result. This rule needs to be broken, however, in
calculations that involve small differences between nearly equal
numbers: for example, if a beaker weighs 10.3814 g empty but
10.493 1 g after addition of a sample of a chemical it would be unwise
to calculate the weight of chemical by substracting 10.4 g from 10.5 g
because this would give only one significant figure in the answer
whereas the data would support four.
At the end of a calculation the unwanted digits have to be discarded.
This can be done either by truncation or by rounding; thus 9.358 1 can
be truncated to 9.35 or it can be rounded to 9.36. In truncation we
simply omit the unwanted digits; in rounding we increase the last
retained digit by 1 if the portion of the number to be discarded begins
20 / The Language of Mathematics
with 5,6,7,8 or 9. Rounding is slightly more accurate than truncation,
and is the usual practice in science.
In our own measurements we know (or should know) how precise
they are, but in other people's there may be ambiguity. We had an
example of this at the beginning of this section: the value of 180 given
for the relative molecular mass of glucose is actually correct to three
significant figures, but there is no way of knowing this from the way
the number is written; it would still be written as 180 if the true value
were 177 but for some reason was given to only two significant figures.
If it is important to avoid this ambiguity we can state the precision
explicitly, for example by giving the value as '180 (correct to three
significant figures)', but this is cumbersome for ordinary use and we
normally assume that we can gauge the precision by counting the
number of significant figures. This is somewhat more subtle than it
sounds, because although all non-zero digits are counted, zeroes may
or may not be counted depending where in the number they occur.
The simplest way of counting significant figures (which is easier to do
than to describe) is to proceed as follows: (1) identify the first
significant digit, which is the first non-zero digit encountered on
reading the number from left to right; (2) identify the last significant
digit, which is the last non-zero digit if the decimal point is not
explicitly included, but the last digit of any kind if the decimal point is
included; and (3) count all the digits (including zeroes), starting from
the first and finishing with the last significant digit. This procedure
may be illustrated by the following examples, in which significant
zeroes are shown in italics and the number of significant figures is
shown in parenthesis after each number: 107 (3); 13.064 (5); 0.120 (3);
0.12 (2); 100 (1); 100.0 (4); 003 (1). (Zeroes are not usually written in
front of integers, as in 003, which would normally be written simply as
3, but if they are written they are not significant.)
We do not usually italicize significant zeroes and it is cumbersome
to add a parenthesis specifying the precision. There is, however, a
third way of indicating that a value such as 100 ml is intended to have
three significant figures: we can choose different units that make the
precision clear, e.g. we can write it as 0.100 litre, or we can achieve the
same result by relating the value to a power of 10 (see Section 2.1, p.
25), e.g. by writing it as 1.00 x 102 ml. These representations are often
used in scientific writing.
Statistical methods are not discussed in this book, but a mention of
them is appropriate here because they can provide estimates of the
 Problems / 21
precision of calculated values with less of the guesswork and
plausibility arguments that have characterized much of this section.
When such a precision estimate exists it can be written after the value
prefixed by a ± sign ('plus-or-minus') thus: 5.17 ± 0.08 mM. There is
rarely any point in reporting a precision estimate with more than one
or two significant figures, and the value it qualifies should be reported
with the same number of decimal places, i.e. the same number of
significant digits after the decimal point. If there is no decimal point
the two numbers should have the same number of insignificant zeroes.
Thus we might write 81.031 ±0.006 or 135200 ± 1400, but there
would be little point in writing 135217 ± 1400 because the right-hand
17 would clearly be meaningless and so better expressed as 00.
Finally, I should say a word about the difference between accuracy,
which is concerned with the truth of what we say, and precision, which
is concerned with how we say it. For example, the constant 1! has a
value of about 3.14159, and so the expression 1! = 3.58937 may be
very precise but it is not at all accurate, whereas the expression 1!
= 3.14 is much less precise but much more accurate. Precision is
always within our control, but accuracy often is not. The main theme
of this section has been that we should aim to express results so that
their precision is only a little greater than their accuracy: less precision
than this discards good information; more makes us look foolish.

1.9 Problems
(1.1) If Xl = 1, x 2 = 3, X3 = 2, x 4 = 7, Xs = 6, X6 = 9, what are the
values of the following expressions?
4 6
(c) L ix i ; (d) L xf
i = 1 i =1
(1.2) The sample variance S2 of a set of n numbers, Xl' X 2 . . . X n ' can
be defined as

Show that this is equivalent to

22 / The Language of Mathematics
(1.3) The Nernst equation asserts that
o RT [ox]
E=E + ----or
n.'#'
In -
re[
d]

where E and EO are potentials measured in volts, R

= 8.314 J mol- 1 K - 1 is the gas constant, T is the temperature in K, n
is the number of electrons transferred in the oxidation-reduction
process, fF is a constant with the value 96494 coulomb mol- 1, and
[ox] and [red] are the concentrations of the oxidized and reduced
forms, respectively, of the redox couple. Assuming that 1 V
== 1 J A-I S - 1 and 1 coulomb == 1 A s, show that the equation is
dimensionally consistent.
(1.4) Which of the following equations or statements must be
incorrect because they are dimensionally inconsistent? ([S], [I], Km
and K j are concentrations; v and V are rates, i.e. concentrations
divided by time).
V[S]
(a) v = Km + [S] + [I]/K j

(b) The intercept on the [S]/v axis of a plot of [S]/v against [S] is
Km/ V
(c) The slope of a plot of v against viES] is -1/ Km.
(1.5) Evaluate the following expressions:
(a) 8 +6/3 +4
(b) 8+6/(3+4)
(c) (8 + 6)/(3 + 4)
(d) 2 3 + 32
(e) 23 x 32
(f) 232

(g) (2 3)2
(1.6) Different methods of determining the average relative mole-
cular mass Mr of macromolecules in solution yield different kinds of
average. For example, osmotic-pressure measurements yield a number
average M n' light-scattering measurements yield a weight average M w'
and equilibrium sedimentation yields a Z-average M z. These are
defined as follows:
"n.M.
M =_L._'_'.
n L.nj
" '
 Problems / 23
where ni is the number of mol of a species with Mr = M j and the
summations are carried out over all solute species. Evaluate all three
averages for a solution of protein existing as 27 nM monomer, 17 nM
dimer and 1.3 nM tetramer, assuming that Mr = 72000 for the
monomer.
(1.7) Give the results of the following calculations with appropriate
precision:
(a) 17.3 x 1.38221
(b) 2.571/(6.331- 2.111)
(c) 1.3359 x (35.6579 - 35.6112)
(d) 2.1 x 4.361
(1.8) How many significant figures does each of the following values
contain?
(a) 23.007050
(b) 0.007050
(c) 135000
(d) 1.350 x 105
(e) 10.37
2 Exponents and Logarithms

2.1 Integer powers

In mathematical manipulations one often has to multiply the same
number by itself repeatedly, so that one has such expressions as
2 x 2 x 2 x 2 x 2 = 32
This sort of thing can rapidly become both tedious and difficult to
read, and it is convenient to have a shorthand notation that conveys
the same information not only more concisely but also (once one is
used to it) more clearly. Exponents fulfil this need. The simplest use of
an exponent i is to raise a number to its ith power, i.e. to show that the
number is to be multiplied by itself i times, the exponent being written
a little above the number that it operates on. For example,
22 means 2x2=4
23 means 2x2x2= 8
24 means 2 x 2 x 2 x 2 = 16
33 means 3 x 3 x 3 = 27
and so on. An exponent can have the value 1, but then it is usually
omitted because the number it operates on is unaltered: 21 = 2,
3 1 = 3, 4 1 = 4, etc.
It might seem that an exponent would have to be a positive integer:
if this were so exponents would be much less useful than they are. It is,
however, possible to assign plausible meanings to zero, negative and
fractional exponents and thereby to extend their usefulness very
greatly. Consider the following pair of series:
i = 1 2 3 4 5 6 .. .
2i = 2 4 8 16 32 64 .. .
What rules govern the choice of numbers to be put in each lines? Each
24
 Integer powers / 25
value of i is greater by 1 than the value on its left, whereas each value of
2i is double the value on its left. What happens if we read the series
from right to left? The rules can simply be reversed: each value of i is 1
less than the value on its right, and each value of 2i is half the value on
its right. There is nothing in these rules, however, that requires us to
stop at i = 1, and in fact we can continue indefinitely (reading from
right to left):
-4 -3 -2 -1 0 2=i
1 1 1
- - 1
16 8 4 2

We get similar results for any number, not just 2, raised to the ith
power, and in general we can define:
a3 = aaa
a2 = aa
a1 = a
aO = 1
a- 1 = 1/a
a- 2 =1/a 2
a- i = lid
The relationship aO = 1 must be noted particularly. Although it may
seem surprising at first sight it arises quite naturally from the series
that was discussed above.
The main use of integer powers greater than the second or third in
biochemistry is for convenient representation of very large or very
small numbers. For example, the number of molecules in 1 mol of a
substance is about 600 000 000 000 000 000 000 000, whereas the num-
ber of gram-ions of H+ in a litre of 1M-NaOH is about
0.00000000000001. Numbers such as these are very inconvenient to
manipulate and even to read if they are written in the ordinary way.
(This is perhaps borne out by the fact that in typing the manuscript of
this book I needed three attempts at typing this last number before it
was correct.) In practice, therefore, it is usual to relate very large or
very small numbers to the nearest power of 10: thus, the first of the
two numbers mentioned is 6 times greater than 1023 , and is therefore
usually written as 6.0 x 10 23 , whereas the second is equal to 10- 14 . In
practice in experimental science we usually simplify even further, at
least with dimensioned quantities, by using multiples or submultiples
26 / Exponents and Logarithms
of units: for example, the biochemist often has to deal with
concentrations around 0.000 01 M. This could be written as 10- 5 M,
but it leads to fewer mistakes in arithmetic if we use a smaller unit of
concentration and write it as 10 IlM, where the Il (spoken aloud as
'micro' in this use) is a standard prefix meaning 10- 6.
In principle we could relate large and small numbers in this way to
powers of any number, but it is normally convenient to use powers of
10 because 10 happens to be the basis of our ordinary number system.
To avoid confusion we always use powers of 10 even for numbers that
have simple expressions as powers of other numbers. Thus we would
express 2401 as 2.401 x 103 even though it happens to be precisely 74 .

2.2 Fractional exponents

We have seen that the effect of increasing the exponent in an
expression d by 1 is to cause the value of the expression to be
multiplied by a; we shall now consider the effect of doubling the
exponent. In this case we produce the square of d, i.e. a 2i = (d)2. For
example: 22 = 4,2 4 = 16 = 4 2 ;4 3 = 64,4 6 = 4096 = 64 2,etc. We can
gain an understanding of fractional exponents by considering the
reverse of this process: halving is the reverse of doubling, and taking
the square root is the reverse of squaring. It seems reasonable
therefore to define 1M as 4, 409M as 64, or more generally a~ = a. J
Similarly, if a3 = b, then M = a; if a4 = b, then M = a. In this way the
logic can be extended until any number can be used as an exponent:
for example, 2°·21 = 221/100 = (221)1/100 = 2097152 1/100 = 1.1567
approximately.
If a is a positive number (not necessarily an integer), then a i is
meaningful for any value of i, as we have seen. But if a is negative, d
has a simple meaning only if i is a positive or negative integer, and in
this case the meaning is just the same as for positive a. For example:
(_2)1 = -2
( - 2)2 = ( - 2)( - 2) = + 4
( - 2)3 = ( - 2)( - 2)( - 2) = - 8
(_2)-1 = 1/( -2) = -~, etc.
[I shall not consider fractional powers of negative numbers, such as
( - 2)t, because, although there are circumstances in which it is useful
to define such powers, these circumstances occur very infrequently in
biochemistry: for the purposes of this book, therefore, we can insist
 Addition and subtraction of exponents / 27
that i must be integral for d to have a meaning if a is negative.]
In the above set of examples, notice that ( - 2)2 = + 4 = 22, and in
general, for any a, ( - a)2 = a 2. This raises the possibility of ambiguity
in the definition of fractional powers. If we define 4t as 'the number
that when multiplied by itself gives 4', the definition is ambiguous
because it is satisfied not only by 2 but also by - 2. In most
circumstances it is not at all convenient for a mathematical expression
to have more than one meaning, and with fractional powers the
ambiguity is removed by convention: M is defined as the positive
number a that satisfies the relationship a2 = b. If we want to refer to
the negative solution to the equation a 2 = b we must write - M.
Occasionally we want to retain the ambiguity (as in the formula for
the roots of a quadratic equation: see Chapter 5), and in this case we
use the symbol ± (,plus or minus') to indicate that either sign may be
used.

2.3 Addition and subtraction of exponents

We have seen that d means the product of i a's, and similarly a j means
the product of j a's. If these two products are multiplied together, the
result must be the same as if we had multiplied (i + j) a's together at
the outset, i.e. d + j. In general:

Although this result is obviously correct for integer values of i and j

only, it actually applies for any values of i and j, including both
fractional and negative ones. Ifj is negative - supposej = - k, where k
is positive - then the expression above gives

Thus, just as addition of exponents corresponds to multiplication of

powers, so subtraction of exponents corresponds to division of
powers.
It is important to note that in all of the expressions in this section I
have used the same base a: there are no correspondingly simple
expressions for powers of different numbers. For example, if a f b,
dbi does not have a simple expression in terms of added or subtracted
exponents (unless of course b is a simple power of a, e.g. if b = a 2 then
dbi = d+ 2j ).
28 / Exponents and Logarithms
2.4 Logarithms
Addition and subtraction are much quicker and easier operations to
carry out than mUltiplication and division. The observation that
addition and subtraction of exponents corresponds to multiplication
and division of powers suggests a way in which one might use the
operations of addition and subtraction to obtain the results of
multiplication and division. For example, if one had a list of powers
of 2, one might multiply 4 by 8 by noting that 4 = 22 and 8 = 23, so
4 x 8 = 22+ 3 = 2 5 = 32. This is of course a trivial example as it
stands, because 4 can be multiplied by 8 in the head more rapidly than
one can consult a lists of powers of 2. Moreover, if the list were
confined to integral powers it would be unlikely to include the
numbers we wished to multiply together. Suppose, however, that the
list were extensive and included non-integral exponents: then we
could use the same approach to solve non-trivial problems. Consider,
for example, the product 2.38 x 9.23. An extensive table of powers of2
would reveal that 2.38 ~ 21.25 and 9.23 ~ 2 3.21 ; therefore, 2.38
x 9.23 ~ 21.25 + 3.21 = 24 . 46 ~ 21.97. In this case the direct multipli-
cation is not trivial and would take much more time than the
corresponding calculation in terms of powers of 2 - provided, of
course, that suitable tables were available. In practice it would be most
convenient to have two sets of tables, one showing what exponent i is
needed for 2i to have a selected value, the second showing the value of
2i for a selected value of i. These are known as tables of logarithms and
antilogarithms, respectively. Just as 2.38 can be described as the result
of raising 2 to the power 1.25, so we can describe 1.25 as the logarithm
of 2.38 to the base 2. These relationships are symbolized as follows:
1.25 = log2 2.38
2.38 = antilog 2 1.25 = 21.25
Notice that 'antilogarithm' is simply another word for power: it is
used to emphasize that finding an antilogarithm or a power is simply
the reverse of finding a logarithm.

2.5 Common logarithms

It is not necessary to use 2 as the base for a set of logarithms: any
positive number would in principle be usable, and it might seem that
all positive numbers would be equally convenient to use. In fact,
 Common logarithms / 29
however, the two bases 10 and e (see p. 31, natural logarithms) have
considerable advantages over all other choices and are by far the most
commonly used, although 2 is sometimes used in studies of bacterial
growth, for reasons that I shall discuss later in this chapter. The
practical advantage of 10 over, say, 2 is apparent from the following
comparison of 10glo and logz values for several numbers:
logz 2.741 = 1.454; 10glo 2.741 = 0.438
logz 27.41 = 4.776; 10glo 27.41 = 1.438
logz 274.1 = 8.098; 10glo 274.1 = 2.438
logz 2741 = 11.420; 10glo 2741 = 3.438
Notice that each multiplication of a by 10 increases logza by logz 10
= 3.322, a rather unmemorable number, but log10a by 10glo 10 = 1
exactly. This means that if we use logarithms to the base 10, or
common logarithms (also known sometimes a Briggsian logarithms,
after Henry Briggs, the English mathematician who first published
tables of them in 1617), we need only a limited set of tables oflog 10 a
for a values from 1 to 10, because all other common logarithms can be
obtained from these by a trivial calculation. For example, if we
required log10 274.1, we would look up log lo 2.741 = 0.438 in a set of
tables, note that 274.1 is 100 = lO z times 2.741 and add 2 to the
logarithm of 2.741, so log 10 274.1 = 2.438.
Because of the separate ways of finding the integral and non-
integral parts of a logarithm, they are given different names. The
integral part (before the decimal point) is called the characteristic; the
non-integral part (after the decimal point) is called the mantissa.
Tables of common logarithms provide only the mantissa, as the
characteristic is easy to determine by inspection. Conversely, tables of
antilogarithms provide values only for the antilogarithm of the
mantissa, the nearest power of ten being found by inspection of the
characteristic.
With negative logarithms, which are the logarithms of numbers
between 0 and 1 (not the logarithms of negative numbers, which do
not have logarithms: see below), it is common but not universal
practice to retain a positive mantissa. For example, 10glO 0.2741 =
- 0.562, but if it is written in this way it is not immediately obvious
that it is exactly 1 less than log lo 2.741 = 0.438. Instead, therefore, it
can be written as - 1 + 0.438 or, more concisely, as l438 (spoken
aloud as 'bar 1 point 438'), where the minus sign is written above the
characteristic to show that it applies only to the characteristic and not
30 / Exponents and Logarithms
to the mantissa. Similarly, 10gloO.002 741 = - 3 + 0.438 = 3.438, etc.
The need for this notation has decreased with the widespread
availability of computers and electronic calculators, which have
removed much of the drudgery from arithmetic, and have in
particular made multiplication and division into trivial exercises.
Consequently one now has little occasion to write quantities such as
1:438, and one should especially avoid them in preparing data for
plotting graphs, where they cause many mistakes. Nonetheless the
student should be familiar with the meaning of the notation as it is still
encountered occasionally.
We have seen that aO = 1, regardless of the value of a. It follows that
loga1 = 0, regardless of a. The logic of this is evident from the fact that
o is the only number that can be added to another number without
affecting its value, whereas 1 is the only number that can be multiplied
by another number without affecting its value.
The following equations summarize this section and the preceding
one:
log(ab) = loga + 10gb
ab = antilog (loga + 10gb)
I have omitted the base of the logarithms from the symbols in these
equations, because they apply for all bases - so long as the same base is
used consistently: it would be improper, for example, to mix
logarithms to bases 10 and e in the same equation. The usual practice
in biochemistry is to use the symbol log to mean loglo, and In to mean
log .. Be warned, however, that this practice is not universal, and that
pure mathematicians and others sometimes u~.e log to mean log..

2.6 Negative numbers have no logarithms

Each division of a number by 10 causes a decrease in its logarithm to
the base 10 of exactly 1. No amount of dividing by 10 can turn a
positive number into a negative number, however, and so the original
number remains positive throughout, even though its logarithm can
cross zero and become negative. It is clear, therefore, that negative
logarithms refer to small positive numbers, not to negative numbers.
Negative numbers cannot be expressed as powers of positive
numbers, and consequently have no logarithms. This does not mean,
however, that it is impossible to evaluate products that involve
negative numbers. For example, - 1.274 x 3.859 can readily be
 Natural logarithms / 31
evaluated by use of logarithms by first evaluating 1.274 x 3.859 and
then changing the sign of the result.

2.7 Natural logarithms

The second widely used base of logarithms is at first sight a bizarre
choice, the number e. This number is irrational, which means that it
cannot be expressed as an exact ratio of two integers, and con-
sequently cannot be written exactly as a decimal number. It can,
however, be expressed to any required accuracy as a decimal, and for
most purposes it is more than sufficiently accurate to express its value
as 2.718 28. One reason for choosing e as a base for logarithms is that
it is very easy to calculate its non-integral powers, by means of the
following series:

in which the exclamation point indicates aJactorial, i.e. n! = n(n -1)

(n - 2) ... 3 x 2 x 1; for example, 6! = 6 x 5 x 4 x 3 x 2 xl = 720.
Although the series for eX is strictly an infinite series, with an infinite
number of terms, it is never necessary to evaluate more than a finite
number of them, because the magnitude of the terms always dwindles
into insignificance, after the first few terms, unless x is very large. If x is
less than 1 this decay into insignificance is very rapid and for small
values of x it is sometimes useful to consider only the first two terms in
the series:
eX ~ 1 + x (accurate to within ± 5 % if - 0.28 < x < 0.35)
Although the infinite series provides a convenient way of calculating
eX, i.e. antilogex, rather than logex, a table of eX values can be read
inversely as a table of logex values. It is much easier, therefore, to
obtain logarithms and antilogarithms with e as base than with 10 and
other bases. Nonetheless, once a set of tables with base e is available it
is in principle simple to create a corresponding set with base 10,
because for any number x there is a simple relationship between the
logarithms to the different bases, as follows:

logex = log.10 loglox = 2.303 loglox

As we have seen, however, 10 is much the most convenient base for
32 / Exponents and Logarithms
purposes of calculation: why then are logarithms to base e of more
than just curiosity interest?
The continued importance of e and its functions such as logex and
eX does not, in fact, derive from their almost nonexistent use as aids to
arithmetic but from the fact that they occur naturally in the solutions
to many kinds of problems in applied mathematics. We shall
encounter various examples of this in this book and at this point it will
suffice to indicate two briefly:

(1) In a system at thermal equilibrium at an absolute temperature

T, the numbers n 1 and n2 of molecules in two states with energies E 1
and E 2 , respectively, are related according to the equation

in which k is a constant known as the Boltzmann constant (= 1.38

X 10- 23 JK- 1 = RjN, where R = 8.31 JK- 1 mol- 1 is the gas con-
stant and N = 6.02 X 1023 mol- 1 is Avogadro's number).
(2) In a first-order reaction with rate constant k the extent of
reaction after time t is proportional to 1 - e - kt.

Although I have deliberately chosen rather physical examples here to

illustrate the ubiquity of e, the fundamental reasons for the ubiquity
are mathematical: they derive from the natural occurrence of logex
and e in purely mathematical contexts, as we shall see when
examining calculus (Chapters 3-4).
Powers of e occur so often that it is typographically inconvenient to
write them as eX, especially if 'x' is a complicated expression.
Accordingly the special operator exp is commonly used instead, i.e.
exp (x) == e
the expression being spoken aloud as 'exponential x': it is unnecessary
to specify the base because e is universally understood in this context.

2.8 Logarithms to base 2

We have seen earlier in this chapter that logarithms to base 2 provide a
convenient and easily understood introduction to the idea of
logarithms as an aid to arithmetic. Although for most practical
purposes 10 is the easiest base to work with and e is usually
 Logarithms to base 2 I 33
appropriate in theoretical discussions, 2 has an important use as a
base in bacteriology.
When bacteria grow in ideal conditions in an open system provided
with ample supplies of all necessary nutrients, the rate of growth is
determined only by the metabolic capability of each individual
bacterium to grow and divide. In a homogeneous bacterial culture
there are no qualitative differences between individuals and quanti-
tative differences are slight. Consequently it is meaningful to conceive
of the generation time, i.e. the period between one cell division and the
next, as fairly constant from one cell to another. If the 'initial'
population size (i.e. the number of cells at an arbitrarily defined time
zero) is no and the generation time is r, then at

t = 0, n = no
t = r, n = 2no
t = 2r, n =4no
t = 3r, n = 8n o
and, in general, at
t= Zt, n = 2zno

Taking logarithms (to any base), we have

log n = Z log 2 + log no

or
log (nino) log n -log no
tit = log2 = log2

Although this equation is independent of the base of the logarithms

it assumes a particularly simple form when the base is 2, because
log22 = 1, so

The phase of growth in which the mean generation time is a constant

is known properly as the exponential growth phase. Paradoxically, the
term logarithmic growth phase is synonymous, even though it sounds
as if it means the opposite. The reason for this peculiar terminology is
that during exponential growth the population size is not a linear
function of time but can be plotted as a straight line if its logarithm is
plotted against time (Fig. 2.1)
34 / Exponents and Logarithms

60

40

20

(a) Generations

6
In (nino)
3

2
2

2 3 4 5
(b) Generations

Fig. 2.1. Logarithms to base 2 provide a convenient way of relating population size
to number of generations, for populations that double in size for each generation.

2.9 Exponential decay

Although exponential growth is mainly observed with bacteria,
exponential decay is of great importance to the biochemist in at least
two other contexts: radioactive decay and chemical reactions that
display first-order kinetics. Mathematically, all of these processes are
very similar, differing only in the sign of the coefficient of time in the
exponent, and the mean generation time in a growth process is
parallelled by the half life of a radioisotope or reactive chemical and
the half time of a chemical reaction. In principle, logarithms to base 2
might be used in these contexts but in practice they are not. The
reason for this is partly historical: methods for analysing decay
processes were already well established before 1949, when Monod
drew attention to the usefulness of 2 as a base; but more important is
 Logarithms as a method of scaling / 35
the fact that the mean generation time of a bacterial population has a
tangible physical meaning whereas the half life of a radioisotope is
only a mathematical abstraction, albeit a convenient one. Thus there
is a real interest in knowing the mean number of generations in a
period of bacterial growth, whereas the number of times that a sample
has lost half of its radioactivity is no more fundamental than the
number of times its activity has decreased by a factor of e or by a
factor of 10.

2.10 Logarithms as a method of scaling

In the chemistry of processes in aqueous solution the concentrations
of hydrogen ions that need to be considered extend over more than 14
orders of magnitude, from less than 10- 14 moll- 1 in strong alkali to
more than 1 moll- 1 in strong acid. In the chemistry of living systems
this range is considerably compressed but it is still large enough to
make it arithmetically inconvenient to work exclusively in concen-
trations. In addition to the numerical inconvenience of concen-
trations there is a philosophical reason for avoiding them.
Biochemistry-and indeed much of Chemistry-tends to be more
interested in changes in state than in states, and a particular change in
concentration can have very different effects in different contexts. For
example, in the mammalian stomach at a hydrogen-ion concentration
of about 10- 3 moll- 1 an increase of 10- 5 moll- 1 would be virtually
imperceptible and would be expected to have only slight
consequences; on the other hand the same increase in a cell at a
hydrogen-ion concentration of 10 - 7 moll- 1 might well be
devastating. (This is not a wholly abstract example: the increase in
acidity brought about by rupturing lysosomes has major effects on
the activities of numerous enzymes within the cell but would pass
unnoticed in the stomach). The point is that in the first case the change
is l.01-fold whereas in the second it is 100-fold: in most circumstances
it is equal relative changes that bring about similar effects. It would be
an oversimplification to say that a doubling (for example) of
hydrogen-ion concentration has the 'same' effect when applied to
starting values of 10 - 3 and 10 - 7 moll- 1, but such a statement makes
sense conceptually and is not without meaning. On the other hand it is
difficult to think of circumstances in which it would make sense to
regard increases of 10- 5 moll- 1 as the 'same', whether from 10- 7 or
from 10- 3 moll- 1 .
36 / Exponents and Logarithms
The purpose of this lengthy digression has been to show that there
are reasons other than numerical convenience for introducing
logarithms into consideration of hydrogen-ion concentration. Not
only does the conversion of [H +] into loglo [H +] avoid the need for
inconveniently small numbers; it also provides a scale in which
particular linear changes (in loglo [H+]) have approximately con-
stant meanings. Unfortunately the scale that is in common use is not
10glO [H +] but -loglO [H +], generally written as pH. For the
ludicrously trivial 'advantage' of having a range of positive numbers
around + 7 rather than negative numbers around - 7, chemists have
paid the excessive price of discarding a scale of obvious meaning in
favour of one that has generated endless confusion. The pH scale is
probably here to stay, however, and so one must make the best of an
error of judgement and try to remember that pH is nothing more than
a tiresome way of writing -IOglO[H+]. Similarly pKa means
-loglo K a , where Ka is an acid dissociation constant. Other usages,
such as pKm (where Km is the Michaelis constant) are less common: I
trust that they will remain so and that the time will never come when it
is thought helpful to overcome the 'inconvenience' of having a minus
sign in the definition of the standard Gibbs energy
AGO = -RTlnK
by writing it as AGO = 2.3RTpK. We are fortunate that at least some
extensions of the pH idea, such as the rH scale for application to redox
potentials, have become much less prevalent than they once were.

2.11 Products of equilibrium constants

In a set of equilibria for a sequence of reactions, such as the following
sequence of unimolecular steps:

one can obtain the equilibrium constant for the complete process by
multiplying all of the individual equilibrium constants together:

Metabolic pathways are composed of such sequences of steps

 Logarithms of dimensioned quantities? / 37
(although many are more complex than unimolecular steps) and it is
frequently necessary to carry out such multiplications and cor-
responding divisions in order to assess the thermodynamic feasibility
of particular pathways or importance of intermediates, etc. But it is
much easier to think in terms of addition and subtraction than in
terms of mUltiplication and division and for this reason it is often
useful to express equilibrium constants on a logarithmic scale, so
that
log KAF = log KAB + log K BC + log KCD + log KDE + log KEF
Now it happens, for reasons that are primarily chemical rather than
mathematical, that there is a direct relationship between the equilib-
rium constant of a reaction and the enthalpy and entropy changes that
occur in it. Because of this relationship, it is common practice to use
not the simple logarithm to base 10-which would be best if
combining equilibrium constants were the only concern - but instead
the natural logarithm multiplied by a temperature-dependent
coefficient, i.e. the standard Gibbs energy ~Go, defined as follows:

~GO = -RTlnK = -2.3RTlog 1o K

where R = 8.31 J mol- 1 is the gas constant and Tis the temperature in
kelvins. Nonetheless although ~GO has a fundamental thermo-
dynamic meaning the factor - 2.3RT is irrelevant to many of the
biochemical contexts in which a logarithmic scale is useful. Living
systems hardly ever exploit temperature effects - indeed they usually
try to avoid them altogether - and so in metabolism - 2.3RT is just a
constant, with a value of about - 5700 J mol- 1 (at 25°C). Thus each
- 5.7 kJ mol- 1 in a ~Go value corresponds to a factor of lOin the
corresponding equilibrium constant.

2.12 Logarithms of dimensioned quantities?

It is obvious from consideration ofthe simplest uses of exponents that
they must be pure numbers, i.e. they must not have any dimensions: it
is meaningless to multiply 10 by itself 3 cm times, for example. As a
logarithm is the exponent that the base must be raised to to give a
particular number it follows that a logarithm can have no dimensions.
Moreover, a pure number raised to a pure-number power must also
be a pure number, and so only pure numbers have logarithms.
38 / Exponents and Logarithms
Despite this, there is apparently a glaring exception, one that has
already been considered in this chapter: if we define pH as
-loglO[H+], and we measure [H+] in moll- 1 then we apparently
must take the logarithm of a concentration. The way out of this
difficulty is to define a standard state, in this instance a reference
concentration [H+]O, and to define pH as -loglo ([H+]/[H+]O). In
this way pH becomes the logarithm of a dimensionless ratio, as it
should be. In practice we take the standard state as [H+]O
= 1 moll- 1 and thereby make the ordinary definition of pH
numerically correct provided that [H +] is measured in moll- I.
The fuller definition of pH may seem pedantic and trivial, but it is
important for understanding that although pH differences have a
fundamental meaning, pH itself is measured from an arbitrary datum
and therefore has no absolute meaning. Thus although it makes good
sense to say that a pH change of + 0.2 is twice as large as a change of
+ 0.1, there is no useful sense in which pH 6 can be regarded as twice
as alkaline (or even as twice as 'large', where the definition of 'large' is
kept suitably vague) as pH 3. Suppose, for example, that we chose to
redefine the standard state as [H+]O' = 10- 7 moll-I (a datum that
would make better biochemical sense than the chemists' 1 moll- 1): in
this case pH 3 would become pH' = - 4 and pH 6 would become pH'
= - 1, a quarter as 'large' instead of twice as 'large'!; but a change in
pH of + 0.2 would also be a change in pH' of + 0.2 and would still be
twice as large as + 0.1, because the standard state does not enter into
measurements of change.
Not many biochemists would suppose that pH 6 could mean-
ingfully be regarded as twice as large as pH 3, but equally ridiculous
comparisons are commonplace in discussions of AGo values for
metabolic processes. Standard states do not enter into the definitions
of dimensionless equilibrium constants (as in the reversible uni-
molecular reactions considered in the preceding section), but many
metabolic and other reactions involve changes in the number of
molecules and have equilibrium constants with dimensions that must
be removed by the use of standard states. As in the definition of pH,
the commonest convention is to define dimensionless equilibrium
constants by relating concentrations to standard concentrations of
1 moll- I. As a consequence of this introduction of standard states it
is nearly always meaningless to measure the 'efficiencies' of metabolic
processes by dividing AGo values into one another.
 Redox potentials j 39
2.13 Redox potentials*
A redox couple is a pair of substances such that one, the oxidized form,
can be converted into the other, the reduced form, by accepting one or
more electrons. For example, the Fe 3 + IFe 2 + couple can be written as
Fe 3 + +s- -> Fe 2 +
Now free electrons cannot exist in aqueous solution and so this
reaction, a so-called half-reaction, cannot occur unless there is another
half-reaction occurring simultaneously in the reverse direction.
Nonetheless, certain couples occur in so many different metabolic
reactions that it is useful to be able to express their oxidizing ability
without explicit reference to another half-reaction. This may be done
by means of the redox potential E, which is defined by the N ernst
equation:
RT [ox]
E=Eo+-In--
nfi' [red]

in which EO is a constant for the particular couple known as the

standard redox potential, R = 8.314 J mol- 1 K - 1 is the gas constant,
T is the temperature in kelvins, n is the number of electrons
transferred, fi' = 96 495 coulomb mol- 1 is a constant that allows E
and EO to be measured in volts, and [ox] and [red] are, respectively,
the concentrations of the oxidized and reduced forms of the couple.
For the Fe3+ IFe2+ couple, EO = 0.77 volt and n = 1, so

RT [Fe 3 +] [Fe 3 +]
E = 0.77 + fi' In [Fe2+] = 0.77 + 0.060 log [Fe2+]

where the coefficient 0.060 V is the value of 2.303 RTj fi' at 303 K
(30 0 C). This tells us that the actual potential E is determined by two
separate quantities, not only by the standard potential EO but also by
the ratio of concentrations of the two components: not surprisingly, a
couple becomes more powerfully oxidizing as the proportion in the
oxidized state increases. Rather confusingly, the Nernst equation is

* Strictly the term is oxidation-reduction potential, but this is rather

cumbersome and I shall use the shorter and more colloquial form in this
discussion.
40 / Exponents and Logarithms
sometimes written with a negative sign and the ratio inverted:
E = EO _ RT In[red]
n!!i' [ ox]
It should be evident from the discussion of the properties of
logarithms earlier in this chapter that this form is exactly equivalent.
Confusion can be avoided by remembering that whichever way the
equation is written increasing the proportion of reduced form reduces
the value of the potential.
The meaning of EO can be understood by putting [red] = [ox]
= 1 moll- 1. (Actually the particular concentration is not important
for the mathematical argument, which would be unchanged if [red]
= [ox] at some concentration other than 1 moll- 1. However, this
concentration is specified for defining the standard states and in
practice it is desirable that it should be specified because the Nernst
equation may not be obeyed exactly.) Then the logarithmic term
becomes zero and so
E = EO
Thus EO is the value of E when the components of the couple are in
their standard states. The absolute value of EO is arbitrary because it is
measured from an arbitrary zero. (The value of EO for one particular
couple, the standard hydrogen electrode H+ ItH2 under specified
conditions is arbitrarily defined as zero, and all other EO values are
defined in relation to this standard). Consequently EO values (and E
values, for that matter) for different couples cannot be compared as
ratios but must only be compared as differences (compare the
discussion of pH and dimensions in the preceding section).
The Fe3+ IFe 2 + couple is particularly simple as an introduction to
redox potentials because no protons are involved in the half-reaction.
Consequently its potential is independent of the hydrogen-ion
concentration. In fact EO is defined at pH 0, although for the
Fe 3 + IFe 2+ couple it would have exactly the same value of 0.77 V ifit
were defined at pH 7 or some other pH. This is not the case for most
couples of biochemical interest, because most involve a transfer of
protons as well as electrons. If protons are transferred, but neither
form of the couple ionizes in the pH range of interest, they may be
allowed for by including them in the Nernst equation as reactants. For
example, the acetaldehyde Iethanol couple involves two protons:
CH 3 CHO + 2H+ + 26- --+ C 2H s OH
 Redox potentials / 41
and has EO = 0.258 V. So the potential is given by

Remember that dividing a logarithm by 2 is equivalent to taking the

square root of the quantity whose logarithm is expressed. So if we
remove the 2 from the coefficient we can replace [H+]2 with [H+].
Furthermore, RT/ $' = 0.060V, and so

E = 0.258 + 0.030 log ~CH3CHOj - 0.060 pH (volt)

C 2 H sOH

Thus at any [CH 3CHO]/[C 2 H s OH] ratio the value of E is 0.42 V less
at pH 7 than at pH o.
Of the couples of biochemical importance, some, for example
cytochrome c (Fe3+) \cytochrome c (Fe 2+), resemble Fe3+ \Fe2+ in
having redox potentials that are independent of pH; a few, such as
acetaldehyde\ethanol, show a decrease of 0.06 V for each unit increase
in pH; but many show intermediate behaviour, because one or both
components ionize in the pH-range of interest (as I shall discuss in the
next section). It follows that EO values if applied naively can be a very
misleading guide to redox behaviour at neutral pH values. For
example, cytochrome c(Fe3+)\cytochrome c(Fe 2+) and acetaldehy-
de\ethanol have nearly equal EO values, 0.250 V and 0.258 V,
respectively, but under metabolic conditions the former is a much
more powerfully oxidizing couple than the latter, because of the large
decrease in potential shown by the latter on raising the pH. To avoid
this confusion biochemists commonly redefine the standard as the
observed potential at a specified pH (usually pH 7) when the two
components are at equal concentrations. This potential, often called
the mid-point potential, is given the symbol E 01 , and the Nernst
equation becomes

E = E 01 + RT In [ ox]
n $' [red]
Proton concentrations are not included in this equation and so it is
42 / Exponents and Logarithms
valid only at the pH at which E 01 is defined. At pH 7 we have E 01 =
- 0.163 V for acetaldehydelethanol, but still EOI = EO = + 0.250 V
for cytochrome c(Fe3+)lcytochrome c(Fe2+).
If two redox couples are allowed to interact, whether by being
mixed together in the presence of a suitable catalyst or by connecting
them electrically, they can react stoichiometrically until their poten-
tials are identical. For example, in the reaction catalysed by alcohol
dehydrogenase the acetaldehydelethanol couple reacts with the
NAD+INADH couple:
E 01 = - 0.320 V (pH 7, 30°C)
When the two E values are equal, i.e. at equilibrium, we can write
[CH 3 CHO] [NAD+]
- 0.163 + 0.03 log [CzHsOH] = - 0.320 + 0.03 log [NADH]

which may be rearranged to give

K = [NAD+][CzHsOH] = 1.71 x lOs

[NADH] [CH 3 CHO]
which is the equilibrium constant for the reaction at pH 7 and 30°C.
Like the standard Gibbs energy of reaction AGOI, the difference
between two standard redox potentials is a logarithmic expression of
an equilibrium constant, and in general we can express this difference
AEo as follows:
l

Thus standard redox potential differences and standard Gibbs energy

of reaction are measurements of the same thing but in different units,
volts for AEo and kJ mol- 1 for AG 0 The reason why redox
l 1
•

potentials are commonly given in electrical units is that electro-

chemical cells often provide the most convenient and accurate way of
determining equilibrium constants, especially for reactions in which
the equilibrium position is very far in one direction.

2.14 Dependence of redox potentials on pH

For many redox couples one or sometimes both of the components
ionize between pH 0 and pH 7 and consequently E varies with pH in a
 Dependence of redox potentials on pH / 43
more interesting way than in either of the simple examples of pH
behaviour discussed in the preceding section. Consider, for example,
the couple acetic acidiacetaldehyde:
CH 3C0 2H + 2H + + 2e - --> CH 3CHO + H 20

for which EO = - 0.11 V and the pKa ( = -log Ka) of the oxidized
form acetic acid is 4.73. Omitting water (i.e. following the convention
that the standard state of water is the state that exists in the solution of
interest), the Nernst equation can be written as follows:

i.e. in a form analogous to that used for the acetaldehydeiethanol

couple considered in the previous section. However, for the acetic
acidiacetaldehyde couple this expression is inconvenient to use
because it contains the concentration of protonated oxidized form, i.e.
[CH3C02H], even though experimentally it would be difficult to
maintain this concentration constant as the pH changed. In reality we
would usually keep the total concentration of oxidized form,
[CH 3C0 2(H)] = [CH3C02H] + [CH 3C0 2], constant; so it would
be much more helpful to recast the Nernst equation in terms of this
total concentration, which is given by

and so

E = EO RT I [CH 3C02(H)] RT In [H+]

+ 2ffi n[CH 3CHO](1 +Ka/[H+]) + ffi

= EO RT I [CH 3C02(H)] RTln[H+]

+ 2 ffi n [CH 3CHO] + ffi
RT
- - I n (1 + Ka/[H +])
2ffi
Then the mid-point potential at any pH is given by

= - 0.11 - 0.06 pH - 0.03 log (1 + Ka/[H +]) (volt)

44 / Exponents and Logarithms
The general form of this expression may be found by considering its
limits as [H+] is very large or very small. At low pH, Ka/[H+] is
negligible compared with 1, and so the third term approximates to
0.03 log 1, i.e. zero, and
E 01 ~ -0.11-0.06pH (volt) ([H+] ~ Ka)
At high pH, 1 is negligible compared with Ka/[H +], and so
E 01 ~ -O.lI-0.06pH-0.0310g(Ka/[H+])
= - 0.11 - 0.06 pH + 0.03 pKa - 0.03 pH
= + 0.03 - 0.09 pH (volt) ([H +] ~ Ka)

Finally, one can form an impression of the behaviour between the two
limiting regions by putting [H +] = Ka:
E0 = -O.lI-0.06pKa-0.0310g2= -0.40V
1
([H+]=Ka)

From these last three expressions it is clear that the dependence of the
mid-point potential on pH is as shown in Fig. 2.2: at low pH, EOI
decreases linearly by 0.06 V for each unit increase in pH; at high pH it
decreases linearly by 0.09 V for each unit increase in pH; the two linear

-004

-0·5

-0·6L----L--~----~--~----~---L--~L--
o

Fig. 2.2. Dependence of mid-point potential on pH for the acetic acid Iacetaldehyde
couple.
 Problems / 45
regions are joined by a curve, but the extrapolated straight lines
intersect at pH = pK a•
I have discussed redox potentials and their dependence on pH in
rather more detail than one would usually expect in an elementary
mathematics book because it is a topic that causes a great deal of
misunderstanding and muddle, not only among students but also,
regrettably, among the authors of biochemistry textbooks. Most of
this misunderstanding can be avoided by realizing the following: (i)
the Nernst equation at pH 0 is normally expressed in terms of a single
oxidizing form and a single reducing form; (ii) in many couples of

03

02
(pyruvic aCid)

0-1

,
\~
\\
,
0-0 t\ \~
pka
(lactic acid)

01

0-2~--~--~--~--~--~--~--~
o 2 3 4 5 6 7
pH

Fig. 2.3. Dependence of mid-point potential on pH for the pyruvic acid Ilactic acid
couple.
46 / Exponents and Logarithms
interest ionization of one or both forms occurs between pH 0 and pH
7; (iii) in ordinary experiments one controls the total concentrations of
oxidized and reduced forms not the concentrations of any particular
ionic states; (iv) one therefore needs to re-express the Nernst equation
in terms of total concentrations; (v) this can be done by means of the
usual equations for ionic equilibria. By following these rules it is quite
straightforward to deal with couples in which more than a single
ionization must be considered. For example, the couple pyruvic
acidllactic acid,

CH 3 COCO zH+2H+ +2£- -+CH 3 CHOHCO zH

has EO = 0.29 V, pKa(pyr) = 2.50, pKa(lact) = 3.86 and thus requires

ionizations of both oxidized and reduced forms to be taken into
account. Nonetheless, a derivation along the lines of that shown
above for acetic acidlacetaldehyde is not difficult and leads to

E 01 = 0.29 - 0.06 pH - 0.03 10g(1 + Ka(pyr/[H +])

+ 0.03 10g(1 + Ka(lact/[H +]) (volt)
and the dependence on pH shown in Fig. 2.3.

2.15 Problems
(2.1) Evaluate the following expressions:
(a) 33 (b) 4- z
(c) 271 (d) 4- t
(e) 16° (f) 10- 3
(g) 3.29 6 X 3.29- 4 X 3.29- 3 x 3.29 (h) 2.37
(i) 4! (j) antilog 3 3
(k) log 100 (I) logz 8
(m) antilog 3

(2.2) Without using a calculator or tables write down approximate

values of
(a) eO. I/)
(b) exp ( - 0.077)

(2.3) (a) Rearrange the approximate formula you used in problem

2.2 to give an approximate formula for In (1 + x) that applies when x
 Problems / 47
(positive or negative) is numerically small compared with 1. Then use
it to write down approximate values of
(b) In 1.116
(c) In 0.983
(d) In 1.041
(e) In 0.888
(2.4) Given that In 2 = 0.693, In 3 = 1.099, In 5 = 1.609, In 7 = 1.946,
evaluate the following:
(a) In4 (b) InO.2
(c) In 27 (d) In 0.6
(e) In lO (f) exp ( - 0.693)
(g) exp (1.099) (h) exp (2.708)
(i) exp (0.337)
(2.5) The Boltzmann principle asserts that at thermal equilibrium
the number of molecules in a state with energy E is proportional to
exp( - E / RT), where R = 8.31 J mol- 1 K - 1 is the gas constant and T
is the temperature in kelvins. In a proton magnetic resonance
experiment any nucleus can exist in either of two spin states differing
in energy (in a lOOMHz instrument) by about 4x lO- 2 Jmol- 1 .
Calculate the relative populations of nuclei in the two states.
(2.6) Sketch the form ofa plot of mid-point potential against pH for a
couple of the form
Ox + 2H + + 2£ - --+ RedH
assuming that EO = + 0.100 V and RedH is a monobasic acid with
pKa = 4.3.
3 Differential Calculus

3.1 Co-ordinate geometry

Traditional geometry is concerned with the shapes of constructions
on paper or in space and with the relationships between shapes. It has
two distinct extensions into the more numerical domains of mathe-
matics: trigonometry and co-ordinate geometry. Trigonometry is
concerned with calculating the relationships between lengths and
angles, whereas co-ordinate geometry has almost the reverse objec-
tive: it allows the use of geometrical insight and understanding for
studying problems that are not essentially geometrical at all but
algebraic. It happens that biochemistry is rather little concerned with
real shapes, distances or angles; consequently neither traditional
geometry nor trigonometry occupy the centre of the mathematical
stage for the biochemist. Co-ordinate geometry, by contrast, is crucial,
because many important relationships in physical chemistry and
biochemistry appear in the first instance as algebraic expressions, and
in this form they are too abstract to be immediately comprehended.
As an introduction to the use of co-ordinate geometry let us
consider the following simple equation:
y = 7 +3x
which expresses a relationship between two variables x and y. If we
plot y against x as shown in Fig. 3.1, we are in effect using a
geometrical model to 'map' an algebraic equation, giving concrete
expression to what would otherwise be an abstraction. For example, if
we let x = 2 the equation tells us that y = 7 + 3 x 2 = 13, and we can
plot the relationship by marking a point 2 units from the y axis in the x
direction and 13 units from the x axis in the y direction. We can use the
shorthand (2, 13), which simply means 'the point at which x = 2,
y = 13', to refer to this point. It turns out that all points defined by the
48
 Co-ordinate geometry / 49

30
!I

o~--~--~----~--~----~
o 3 4 5
x=2 x

Fig. 3.1. Plot of y = 7 + 3x.

equation lie on a straight line, as shown in Fig. 3.1, and indeed any
equation of the form
y = a+bx
where a and b are constants, defines a straight line if y is plotted
against x. This more general straight line is shown in Fig. 3.2.
It is obvious that y = a if x = 0, because then the term bx is zero.
Consequently the straight line cuts the y-axis a distance a from the
point (0, 0), the point where the axes intersect, which is known as
the origin. Because of this relationship a is known as the intercept
on the y-axis. The intercept on the x-axis is less obvious but may
readily be found by putting y = 0: this gives a + bx = 0, and so x =
- alb, which is therefore the intercept on the x-axis.

!I=a+bx

!l2
!l1

t o X2 x
-alb

Fig. 3.2. Plot of the general straight line y = a + bx.

50 / Differential Calculus
The meaning of b in the general equation for a straight line follows
from consideration of how Y changes when x changes. Suppose that Y
changes from Yl to Y2 as x changes from Xl to x 2. Then
Yl=a+bx l
Y2 = a +bx 2
If the first equation is subtracted from the second the constant a
disappears:

Thus

Notice that this relationship applies for any values of Xl and x 2 we

could have chosen (other than equal values, x 2 = Xl' which would
introduce complexities that I prefer to avoid at present). In other
words the ratio (Y2 - Yl)/(X 2 - Xl) is a constant equal to b; indeed it is
this constancy that defines algebraically what we mean when we say
that a line is straight.
If b is numerically very small, the straight line defined by the
equation Y = a + bx is almost parallel with the x-axis: y changes very
slowly as X changes. Conversely, Y changes very rapidly as X changes if
b is numerically very large. Thus b has a meaning very similar to that
of the gradient of a hill in everyday life. This term is sometimes used
for the mathematical concept represented by b, but the term most
commonly found in scientific work is slope. As we shall see in the rest
of this chapter, not only straight lines have slopes: we can usefully
extend the term to apply it to any curve, and it is this extension that is
the concern of differential calculus.

3.2 Slope of a curve

Consider the following equation:
Y = 2 + 3x +X2
which defines the curve shown in Fig. 3.3 (the curve is known as a
parabola, but this is not important for our present purpose). We saw
that for the straight line considered in the previous section the ratio
(Y2 - Yl)/(X 2 - Xl) was a constant. Let us now examine the same ratio
 Slope of a curve / 51

!I 60
!I = 2 + 3x + X 2
40

30

O'----"'-'-_ _"'--_--L_ _ -L-_~ _ _...L_j_

o 2 3 4 5 6
x
Fig. 3.3. Plot of the parabola y = 2 + 3x + x'.

in the more complex example given in Fig. 3.3. First, it is convenient to

introduce a new symbolism for (Y2 - YI) and (X2 - XI)' Let us use Ax
to represent a change in x, and Ay to represent the corresponding (not
just any) change in y. Then we can write (y 2 - Y I) as Ay and (X2 - Xl)
as Ax. So if

Then
y+Ay = 2+3(x+Ax)+(x+Ax)2
= 2 + 3x + 3Ax + x 2 + 2xAx + (AX)2
Subtracting, we have
Ay = 3Ax + 2xAx + (AX)2
So
Ay
-= 3+2x+Ax
Ax
As we have said nothing about how large a change in x we are
discussing, we have no grounds for simplifying this expression, but
suppose we specify that Ax is a small change in x, i.e. small compared
with (3 + 2x), and let us emphasize this by replacing Ax with Jx (a
lower-case delta instead of a capital). Then in the similar expression
Jy
- = 3+2x+Jx
Jx
we can justifiably make the approximation of ignoring Jx in
comparison with (3 + 2x):
52 / Differential Calculus
by
- ~ 3+2x
bx
If we make bx so small that it is indistinguishable from zero the
approximation becomes exact. We can express this by saying that the
limit of by/bx as !5x approaches zero is 3 + 2x, or

Lim ~y = 3 +2x
{)X~O uX

This result gives the slope of the tangent to the curve y = 2 + 3x + x 2

at the point where the expression is evaluated, as illustrated in Fig. 3.3.
For convenience we often refer to this simply as the slope of the curve
itself, although this is not a constant, unlike the slope of a straight line.
Now, it may be wondered why we have to refer to the 'limit as !5x
approaches zero' rather than the 'value when!5x = 0', which may seem
simpler and more natural. The difficulty with making !5x = 0 is that
then !5y/8x = 0/0, a ratio that is generally undefined in mathematics
because in principle it can have any value. We have just seen, however,
that !5y/!5x does not approach just any value as!5x approaches zero but
the exact and particular value 3 + 2x. For this reason the more
cumbersome terminology is a better description of the process.
Expressions such as Lim ~y are much too cumbersome for
{)x~ 0 uX

common use and so we define a new quantity :~ as having exactly the

same meaning, i.e. in general,

dy == Lim !5y
dx {)x~o!5x
This is called the derivative of y with respect to x; it can also be
described as the result of differentiating y with respect to x. Quantities
that are arbitrarily close to zero while remaining distinguishable from
zero are called infinitesimal.

3.3 Rapid differentiation

The method we used in the previous section for differentiating
y = 2 + 3x + x 2 is known as differentiating from first principles. It is
q'uite general, i.e. the same approach can be used for differentiating
any function. We shall use it again in this chapter to see how to obtain
 Rapid differentiation / 53
formulae for differentiating products, ratios and other functions. For
everyday use it is rather long-winded, however, and in fact there is a
simple rule that enables us to differentiate any power of x with respect
to x. In general, if
y = AXi
where A and i are constants, then
dy
-=
A.IX i-1
dx
Putting this into words, any constant multiplier remains unchanged,
the index is decreased by 1 and the result is multiplied by the old index.
Applying it to various examples, with A representing a constant in
each case, we have
dA =0
dx
d
dx(Ax) =A
d
dx (AX2) = 2Ax

d
-(Ax 3 ) = 3Ax 2
dx
d d
- (A/x) = -(Ax- 1 ) = - Ax- 2 = - A/X2
dx dx
d L
-(AX2) = rAX-2
1 L
= A/2x't
dx
and so on. With the aid of the rules for differentiating combinations of
terms that I shall discuss in the next three sections, this general rule
permits the differentiation of almost all the expressions encountered
in elementary biochemistry. Only two other simple cases are worth
committing to memory, the derivatives of logarithms and
exponentials:
d
dx (Alnx) = A/x

d:(Ae X ) = A~
54 / Differential Calculus
3.4 Derivatives of sums and products
Suppose we wish to differentiate a sum, such as y = u + v, where u and
v are both functions of x that can readily be differentiated with respect
to x. By the arguments we have used before, a small change bx in x
brings about changes in u and v as follows:
du
bu ~ -bx
dx
dv
bv ~ -bx
dx
Clearly if y = u + v the change in y must be the sum of those in u and v:
by = bu +bv
So
by du dv
-~-+­
bx dx dx
In the limit the approximation becomes exact, i.e.
dy . by du dv
-= Lim - =-+--
dx bx-Obx dx dx
This result can be extended to the sum of three or more functions of x,
and in general the derivative of a sum is equal to the sum of the
derivatives. The derivative of a difference follows directly:
d du dv
dx (u - v) = dx - dx

as a difference is simply a sum in which the second term is negative.

Suppose now that y = uv is the product of two functions of x. Then
y+by = (u+bu)(v+bv)
= uv + ubv + vbu + bubv
so, by subtraction,
by = ubv + vbu + bubv
and, dividing every term by bx:
by bv bu bubv
-=u-+v-+--
bx bx bx bx
 Derivatives of sums and products / 55
In the limit the last term disappears, because it is itself a small quantity
whereas the other terms in the expression are simply ratios of small
quantities:

dy = Lim by = u dv +vdu
dx c!x ~0 bx dx dx
This derivation is illustrated graphically in Fig. 3.4: the change in uv is
represented by the sum of the areas in the three rectangles labelled

v+ ~v ~----------------------~~~~
',t:11:i -.. -. :~~. :~.>
1.I 0V :~ ~I.I ~v "'.
v~----------------------f~:~'~'-
' -"_':~l~'

uv v8u

oL-----------------------I.IL------u-+~o·u----
(a)

uv v v~ u

O~-------------------d'~~J-+-o-u-
(b)

Fig. 3.4. Explanation of the formula for the derivative of a product uv. In (a) the
increments.5v and .5u are appreciable and the shaded area is by no means negligible, but
as the increments become smaller, as in (b), the shaded area decreases very rapidly, not
only absolutely, but also in relation to the areas labelled u.5v and v.5u, which become
smaller much more slowly.
56 j Differential Calculus
u(jv, v(ju and (ju(jv. It is clear from inspection that as (ju and (jv become
small, (ju(jv becomes smaller very much faster, and in the limit can be
ignored.
If we divide the left-hand side of the above equation by y and the
right-hand side by uv (= y) we have
1 dy 1 du 1 dv
--=--+--
ydx udx vdx

This alternative form of the derivative of a product is useful because

each term is 'homogeneous', i.e. it contains only one kind of variable
apart from x, and it can readily be generalized to products of more
than two functions. For example, if
y = uvw
then
1dy 1 du 1 dv 1 dw
-- = --+--+--
ydx udx vdx wdx
and so on.
We can gain some insight both into the ubiquity of logarithms in
mathematics and into why :x (In x) = Ijx by recalling that the
logarithm of a product is the sum of the corresponding logarithms, i.e.
if y = uv then
In y = In u + In v
and so, by the rule for the derivative of a sum,
d d d
-d (In y) = -(In u) + -(In v)
x dx dx

but if we accept (without proof) that :x (In x) = Ijx then we can write
1 1
d In x as -dx, din y as -dy, etc., so
x Y
1 dy 1 du 1 dv
--=--+--
ydx udx vdx

which is the same result as we had before for the derivative of a

product. Although this does not prove that our expression for the
 Derivative of a 'function offunction' I 57
derivative ofln x is correct it does show that it is consistent with what
we know independently and hence that it is plausible.

3.5 Derivative of a 'function of a function'

The bell-shaped curves often found for the pH dependence of protein
and enzyme properties (such as catalytic activity) can be represented
by an equation of the following sort, known as a Michaelis function:
A
y= 1 + (Blx)+Cx =A(1+Bx- 1 +Cx)-1

where y is some property of the protein, x is (usually) the hydrogen-

ion concentration, i.e. not pH but to-pH, and A, Band C are con-
stants. If we write the denominator as
u=1+Bx- 1 +Cx
then the expression as a whole can be written as
y = Au- 1
So it is easy to differentiate u with respect to x:
du 2
- = -Bx- +C
dx
and it is easy to differentiate y with respect to u:

dy = -Au- 2
du
But how can we combine the results to get the derivative of y with
respect to x? This is an example of the problem of how to differentiate
a function of a function - y is a function of u, which is itself a function
of x. I shall now consider this as a general problem before returning to
the specific example of the Michaelis function.
For y = F(u), u = f(x), we have
dy du
by ~ -bu, bu ~ -bx
du dx
and so
dydu
by ~ --bx
dudx
58 / Differential Calculus
and
by :!:: dydu
bx dudx
and in the limit
dy ,by dydu
- = Llm-=--
dx 6x .... obx dudx

In general, if Ji is a function of f2,j2 is a function of h, etc., to j",

which is a function of x, then

df1 dJi df2 df3 dj"

-=---"'-
dx dfi df3 d,i4 dx

This is known as the chain rule. It may seem obvious, as it can be

'derived' by treating all of the derivatives as fractions and cancelling
the like elements. Although such a 'derivation' would not satisfy a
serious mathematician the differences between derivatives and frac-
tions rarely create difficulties in biochemistry. Consequently it does
little harm if the biochemist uses non-rigorous 'derivations' as an aid
to remembering important general relationships,
Returning to the Michaelis function, application of the chain rule
shows the derivative to be

dy dy du -2 -2 ABx- 2 -AC
-d =-d-d =-Au (-Bx +C)=(1 B 1 C)2
X ux +x+x

3,6 Derivative of a ratio

A ratio can be regarded as a product in which the second term is raised
to the power -1. So y = u/v is the same as y = uv- 1 , and

dy= u d- v
- (-1)
+v _ldu
-
dx dx dx
in which V-I is a function of a function, with derivative
 Higher derivatives J 59
So
du dv
v--u-
dy _2 dv _ 1 du dx dx
-uv -+ v - = ---=---
dx dx dx v2

3.7 Higher derivatives

After a function has been differentiated once there is no reason why
the result should not be differentiated a second time by application of
exactly the same rules. For example,
y = 7 + 3x - 2X2 + x3

dy = 3-4x+3x 2
dx

~ (d Y )
dx dx
= -4+6x

This last quantity is usually written more compactly as d 2 yJdx 2

(spoken aloud as 'dee two y by dee x squared' not as 'dee squared y by
dee x squared') and is called the second derivative of y with respect to x.
Just as the first derivative expresses how fast y changes with x, the
second derivative expresses how fast dy/dx changes with x.
Alternatively, if dy/dx expresses the slope of a plot of y against x,
d 2 y/dx 2 expresses the rate of change of slope or in other words the
curvature. If there is no curvature the slope is constant, so the second
derivative is zero. For a general straight line, therefore,
y = A+Bx

dy = B
dx
d2y
dx 2 = 0

3.8 Notation
The notation that I have used hitherto in this chapter is based on that
of the great German mathematician Leibniz. Although it is the most
60 / Differential Calculus
generally useful and the most widely used there are circumstances in
which it becomes cumbersome, and one should have some familiarity
with three other systems that are sometimes used.
(1) In mathematical tables, which one may wish to consult to
obtain unfamiliar results, it is usually obvious what variable one is
differentiating with respect to, and compactness is achieved by writing
d/dx as D. Many people also find this a useful form for memorizing
the standard results for the derivatives of products and ratios:
D(uv) = uDv + vDu
D(u/v) = (vDu - uDv)/v 2
(2) Another system for achieving compactness when it is obvious
what one is differentiating with respect to is to write the first derivative
of y with respect to x as y', and the second derivative as y". This is
especially convenient when one does not want to define a particular
variable (y in this example) as the function, e.g. one can write the first
derivative of f(x) as f'(x), and the second as f"(x).
(3) Calculus was developed independently by Newton in England
and Leibniz in Germany. Newton's dot notation is in general much
less convenient than that of Leibniz, partly because it does not allow
for the possibility that one may wish to differentiate with respect to
more than one variable. Newton was most concerned with functions
of time, or t, and when dealing with time (e.g. in enzyme kinetics) it is
still occasionally convenient to write dy/dt as yand d 2 y/dt 2 asy. Even
in kinetics, however, this notation is much less common than that of
Leibniz.
The various alternative systems are listed in Table 3.1. As indicated
in the table, one can continue differentiating beyond the second
derivative, but the biochemist rarely wants to do this and so I have not
Table 3.1 Notations for expressing derivatives

Function y y y I(x) x = I(t)

dy
1st derivative Dy y' IN x
dx
d 2y
2nd derivative D2y y" f"(x) X
dx 2
d"y in)
nth derivative D"y p")(x)
dx"
 Maxima and minima / 61
discussed it in the text. It will be noticed that although Leibniz's
notation is the least compact it is also the most explicit. For this
reason it is nearly always wise for the non-mathematician to prefer it:
a small saving in space is never justified if it leads to baffiement.

3.9 Maxima and minima

Consider the following equation:
Vs
v = -----::---
Km + s + S2 / Ksi
which represents a simple kind of substrate inhibition in an enzyme-
catalysed reaction. The variables are the rate, v, and the concentration
of substrate, s, and V, Km and Ksi are constants. Unlike the examples
earlier in this chapter, this equation is not written with the typical
symbols of elementary mathematics, x, y, A, B, etc., but with the sort
of symbols most likely to be encountered in biochemical work. This
difference should cause no problems: one does not require a particular
set of symbols to carry out mathematical operations, and it is unwise
to associate particular relationships too rigidly with particular
symbols.
Fig. 3.5 shows the form of a plot of v against s for the equation for

60 slope =0 at maximum
v

°0~-----10~----~2~0-----3~0------4~0~
s
Fig .3.5. Plot of v against s for a system showing substrate inhibition. The plot
illustrates the point that we can define a maximum as a point at which the slope is zero,
but this definition would be incomplete because it would also include minima (see
Fig. 3.6).
62 / Differential Calculus
substrate inhibition. A striking feature of the curve is that instead of
increasing indefinitely with s (as in simpler kinds of kinetic behaviour)
v increases to a maximum and then decreases as s is increased further.
A major use of the differential calculus is that it enables one to
calculate where a maximum will occur. From inspection of the graph
it can be seen that at the maximum the slope is zero. This is true in
general, i.e. at any maximum the slope is zero (although the converse is
not, because a point of zero slope may also represent a minimum).
Accordingly, we can find the maximum in Fig. 3.5 by differentiating v
with respect to s and finding a value of s at which dv/ds is zero. This
can be done by applying the rule for differentiating a ratio:

dv (Km +s+s2/KsdV - Vs(l +2s/Ksd

ds (Km+s+s2/KsY
VKm + Vs+ Vs 2/Ksi - Vs _2VS2/Ksi
(Km +s+s2/KsY
VKm - Vs 2j Ksi
(Km +s+s2/KsY
In principle, an expression such as this can be zero either because the
numerator is zero or because the denominator is infinite. Although
the denominator of this particular expression does increase steeply
and without limit as s increases it remains finite at all finite s. So the
only way the expression can be zero at finite s is for the numerator to
be zero, i.e.

Solving for s, we have

and substitution of this result into the original equation shows the
maximum value of v to be
 Maxima and minima j 63
The Michaelis function that we considered earlier in this chapter is of
exactly the same form as this equation for substrate inhibition,
because if
A
y = 1 + (Bjx)+Cx

then multiplication of both numerator and denominator by x gives

Ax
y=-----=-
B+x+Cx 2

which is exactly the same equation with different symbols (y for v, x

for s, A for V, B for Km and ljC for K,J Consequently the maximum
of the Michaelis function must occur when x = (BjC)l, Y = Aj
[1 + 2(BC)1]. In this case, however, we are more likely to want to plot
y against log x than against x (because the Michaelis function is
a dependence on hydrogen-ion concentration, which is usually ex-
pressed in logarithmic form). The question therefore arises, can we
assume that the same maximum occurs in a plot of y against log x? Put
differently, can we assume that dyjdlogx = 0 when dyjdx = O? The
answer comes from considering the relationship between the two
derivatives, which follows from the chain rule:

~ = dy ~ dlnx =2.303xdy
d log x dx d In x d log x dx
(because dlnxjdx = 1jx and logx = 2.3031nx approximately).
Thus, provided x is finite,a zero value of dyjdx must occur at the same
value of x as a zero value of dyjd logx, and so the maximum in the
plot of y against log x occurs at the same values of x and y as in the
plot of y against x.
Exactly the same criteria can be used to find a minimum of a
function, because this also corresponds to a zero value of the first
derivative (Fig. 3.6). Clearly, therefore, the existence of such a zero
value does not guarantee that a maximum has been found. It shows
only that one has found a stationary point, a more general term that
embraces both maxima and minima. In most elementary biochemical
circumstances this is not a problem because it is usually obvious
whether one is dealing with a maximum or a minimum. In the more
general case there may be ambiguity, but this can be resolved by
examining the sign of slope on either side of the stationary point. At a
64 J Differential Calculus

!I

.~--""-------
slope =0 at minimum

Fig. 3.6. Minimum of an arbitrary curve. Again, the slope is zero at the minimum
(see Fig. 3.5), but it is increasing through zero as x increases.

maximum, the slope is decreasing through zero, whereas at a

minimum it is increasing through zero. Returning to the expression
for dvJds in the case of substrate inhibition, it is clear that the
numerator, VKm - Vs 2 / K si , must decrease continuously as s is
increased through the positive range, and so it must decrease through
zero as s increases through the value (KmKsd!. This is, therefore, a
maximum, as assumed previously, not a minimum. It is not necessary
to consider the denominator of the expression for dvJds because this
contains positive terms only (for positive s) and must therefore be
positive.

3.10 A note on terminology

It is often true that terms in mathematics have exact meanings that do
not precisely agree with the meanings of the same words when used in
everyday language. An example is the term 'maximum', which refers
in mathematics, as we have seen, to a point at which the first derivative
of a function is zero and is decreasing as the abscissa variable
increases. For a complex function there may be several or many such
points, and in such a case we would refer to them as 'maxima' and to
one of them as 'a maximum', not as 'the maximum'. Similar
considerations apply to minima. Moreover, there is nothing in the
definitions of either that requires a maximum to occur at a larger
value of the function than a minimum; indeed one can find functions
 Points of inflexion / 65
that have a unique maximum at a lower value than the unique
minimum, such as
1
y=x+-
x
(If you find this statement puzzling you should plot y as given by this
equation against x, for x between -2 and -0.1 and between 0.1 and
2.) This sort of result would be absurd in everyday language, in which
the maximum of a measurement is the largest value it can have and the
minimum is the smallest value it can have. Moreover, 'it can have' is
usually taken to mean 'under realistic circumstances'.
Because biochemistry was not developed largely by mathema-
ticians, mathematical terms have come to be used in biochemistry
with looser meanings than would be acceptable to mathematicians.
For example, the quantity V in the Michaelis-Menten equation:

v = Vsj(Km + s)

is called the maximum velocity, because v cannot exceed it. However, v

cannot attain the value V at a finite value of s and thus V is not a
maximum in the mathematical sense but a limit. Moreover, when a
true maximum does occur in a plot of v against s, as in the example of
substrate inhibition discussed in the previous section, the value of vat
the maximum should not be called the 'maximum velocity', because
this would cause confusion with the usual meaning of the term as V.

3.11 Points of inflexion

An alternative to the method I have described for deciding whether a
stationary point is a maximum or a minimum is to examine the sign of
the second derivative at the stationary point. This is because the
second derivative expresses directly how the first derivative is
changing with the abscissa variable. At a maximum the first derivative
is decreasing and so the second derivative is normally negative;
conversely, at a minimum it is normally positive. Although I could
illustrate this with the same example as in Section 3.9 we
would then get a rather complicated expression on differentiating
twice; it is easier to consider a function such as the following:
y = 2+9x-6x 2 +X 3
66 J Differential Calculus
In this case

dy = 9 -12x + 3x 2
dx
d2 y
dx 2 = -12+6x

Setting dy Jdx = 0 gives a quadratic equation with the two solutions x

= 1, y = 6 and x = 3, y = 2. As 6 is larger than 2, it may seem obvious
that the first solution is a maximum and the second a minimum.
However, although this conclusion happens to be correct in this
instance it does not follow from the logic, which is unsound (recall the
function y = x + ~ mentioned in the previous section). It is safer
x
therefore to examine the value of d 2 yJdx 2 : this is -6 at x = 1 and + 6
at x = 3, and shows that the point (1, 6) is indeed a maximum and the
point (3, 2) a minimum.
Occasionally dy/dx and d 2 yJdx 2 are both zero at the same value of
x, but this is unusual in the sort of equations that occur in
biochemistry and thus rarely interferes with the method just de-
scribed for distinguishing between maxima and minima. It is,
however, important to consider zero values of d 2 y/dx 2 that do not
necessarily coincide with zero values of dyJdx. These occur at so-
called points of inflexion, which are points at which the slope of the
plot is a maximum or a minimum (Fig. 3.7).
Points of inflexion are important in biochemistry because they
define conditions in which a response (e.g. the rate of a reaction) is
most (or least) sensitive to an influence (e.g. the concentration of a
metabolite). Consider, for example, the ability of a butTer to resist the
changes in pH that might be brought about by addition of alkali. The
equation that defines this (to a first approximation) is the
H enderson-H asselbalch equation:
[salt]
pH = pK. + log [acid]

For example, for a butTer made up by adding x moll- 1 NaOH to a

solution of acetic acid, HOAc, at an initial concentration A, we have:
[NaOAc]
pH = pK.+log [HOAc]
 Points of inflexion / 67

!I 60

40
Point of inflexion

20 Maximum
t

dy 40
dx

20

0r-~~~~~~-----------

O~------~~-------------------

I I ~
o 2 3 4 5
x
6

Fig. 3.7. Point of inflexion. At a maximum or minimum the first derivative is zero;
at a point of inflexion the first derivative is a maximum or a minimum and the second
derivative is zero.

[NaOAc]
= pKa +0.4343ln [BOAc]

x
= pKa + 0.4343ln - A
-x
because as long as x is less than A essentially all of the added OB-
ions convert BOAc molecules stoichiometrically into OAc- ions.
68 / Differential Calculus
(0.4343 is the reciprocal of 2.303, or I/ln 10). Differentiating with
respect to x, we have

dpH -_ 0.4343
d x
. (A -x) [(A -x)-x(2
x (A -x)
-1)J
0.4343 A
x(A -x)

This first derivative is a measure of the sensitivity of the pH to

addition of base. Clearly the buffer is most effective when it is a
minimum (or its reciprocal, known as the buffer capacity, is a
maximum). To find the value of x that makes dpH/dx a minimum we
must differentiate again:

d 2 pH - 0.4343 A(A - 2x)

dx 2 (Ax -x 2 f

By inspection it is clear that this derivative is zero when x = !A, i.e.

when enough base has been added to neutralize exactly half of the
original acid. Substituting back into the Henderson-Hasselbalch
equation, we find that

pH = pKa + 0.4343 In G~) = pKa

Thus a buffer is most effective at the pH equal to the pKa of the acid
involved in the titration.

3.12 Sketching curves

One of the most important uses of differential calculus is as an aid to
sketching the curves corresponding to unfamiliar functions. Suppose,
for example that one had found that the concentration q of product of
a reaction at time I could be expressed by an equation of the following
form, in which all symbols apart from q and t are positive constants:

k+2k+3eO{ I-exp[ - (k+2 + k+3)t]}

(k+2+k+3)2

It is by no means obvious at first glance what sort of curve this

equation defines, but one can deduce a good deal simply by examining
 Problems / 69
two extreme values of t. First, if t = 0 then
q = 0- k+2k+3 eo[1-exp(0)] = 0
(k+2 +k+3)2
So the curve passes through the origin. Second, if t is large enough for
exp[ - (k+2 +k+3)t] to be negligible compared with 1, the equation
simplifies to
k+2k+3 e O
(k+2 + k+3)2
This is the expression of a straight line with a positive slope and a
negative intercept on the ordinate, and so the curve must approximate
to such a straight line at large values of t.
Although this information is useful it tells us little about what
might well prove to be the most interesting part of the time course, the
period before the straight-line approximation becomes valid. Light
can be shed on this period by differentiating twice to reveal the

,
, /

, /

o
t

F.ig. 3.8. Sketch of a complex curve (defined in the text) obtained by considering the
value of q and its first and second derivatives with respect to t at various values of t. The
broken line is a straight line and represents the limit approached by the curve as t
increases.
70 / Differential Calculus
behaviour of the slope:
dq k+2k+3eO k+ 2k+ 3eOexp[ - (k+2 +k+3)t]
dt k+2+k+3 k+2+k+3
k+ 2k+ 3eo {1-exp[ - (k+2 + k+3)t]}
k+2+k+3
d 2q
dt 2 = k+2k+3eOexp[ - (k+2 + k+3)t]
As exp[ - (k+ 2 + k + 3)t] must have a value between 1 (at t = 0) and 0
(for t -> CX)) for any positive value of t, it follows that dq /dt cannot be
negative and is zero only at the origin. Thus the plot of q against t
cannot contain a maximum or a minimum except at the origin.
Moreover d 2q/dt 2 is positive at all values of t, although its value
becomes negligible at large values of t. When all of this information is
combined it becomes clear that the form of the curve must be as
shown in Fig. 3.8.

3.13 Problems
(3.1) Differentiate each of the following expressions with respect
to x:
(a) y = x3 (b) y = X 1 / 2
(c) Y = 5e (d) y = e 5x
(e) y = 31nx (f) y = X 1 / 2 + X- 1 / 2
(g) y = In x - 2X2 (h) y = x 2 ex
(i) y = In (x 3 ) (j) y = (x 2 + 3)1/2
(k) y=(x+l)/(x-l)
(3.2) Two of the derivatives obtained in problem (3.1) should be
identical. Why?
(3.3) Each of the following functions displays one minimum and one
maximum. Identify both:
(a) y = 7 + x 2 - 8x + In (XO)

(b) y = 15-x-l/(x-5)
(3.4) Michaelis and Menten analysed their data in terms of the
equation that bears their names, v = Vs/(K m +- 8), by plotting v against
log s. They used the facts that this plot has a point of inflexion at
 Problems / 71
s = Km and that the maximum slope is 0.576 V. Prove that these facts
are correct.
(3.5) Differentiate the equation v = Vs/(K m + s) with respect to s.
What are the values of dv/ds when (a) s = 0; (b) s = Km; (c) s
approaches infinity?
(3.6) Consider a pH profile defined by the following equation, in
which h represents the hydrogen-ion concentration [H+] and y is a
constant:
y
Y= 1 + (h/Kd + (K2/h)
(a) At what value of h is y a maximum?
(b) Show that this solution is equivalent to pH = (pK 1 + pK 2 )/2.
(3.7) Under certain limiting conditions the binding of a small ligand
to a protein can be approximated by an equation of the following
form:
KXh
Y=------,-
1 +KXh
in which Y is a measure of the extent of binding, x is the concentration
of ligand, and K and h are constants. Putting K = 1 (arbitrary unit),
h = 2, sketch the form of curve given by this equation for a plot of Y
against x (for positive x).
(3.8) For the equation given in problem (3.7), what would be the slope
of a plot of log [Y/ (1 - Y)] against log x?
(3.9) Differentiate the following equation with respect to t, treating p
and t as variables and V, Km and So as constants:
Vt = p + Kmln [so/(so - p)]
4 Integral Calculus

4.1 Increases in area

Fig. 4.1 represents an arbitrary function y of x, and the area of the
region bounded by the curve, the x-axis and the verticals x = Xo and
x = Xl is designated A. What is the increase bA in this area if Xl is
increased to Xl + bx? Apart from the small approximately triangular
area that is shaded in the diagram, all of the increase in area is
accounted for by the tall thin rectangle of height YI and width bx,
which has area ylbx. In symbols,

If X is made smaller, the small shaded triangle decreases rapidly in

area, not only absolutely, but also in relation to the area of the
rectangle. Consequently, the approximation embodied by the above
equation is very accurate if bx is very small, and in the limit as bx
approaches zero it becomes exact. This is not very helpful, however, if
we want to know the value of bA when bx is large! We can overcome
the difficulty by noting that if we divide both sides of the equation by
bx we have
bA
- ==: Y
bx
which in the limit is exact:
dA
-=y
dx
This expresses the important generalization that the derivative of the
area between the x-axis and any curve Y = f(x) is equal to the ordinate
variable y. As we shall see, this relationship means that we can
72
 Increases in area / 73

!f

~ ----------------------

Xo
x

Fig. 4.1. Area under a curve. Ignoring the shaded area, the increase in A as x
increases from XI to XI +"x is YI"X.

calculate the area under any curve whose algebraic expression can be
recognized as the derivative of some other function. First, however,
we must see how a large area can be treated as the sum of many small
areas.
As long as bx is small enough for the relationship bA ~ ybx to be a
good approximation we can write

for the increase in area AA brought about by increasing x by

n
Ax = nbx. At first sight this may seem simple to evaluate, as L bx is
just nbx, or Ax; but to get a good approximation we must not treat y as
a constant but must constantly recalculate it as x changes. If the
approximation is made better and better by making bx smaller and
smaller the number of increments n must be increased concomitantly
to keep Ax the same size. In the limit the relationship is exact when
74 / Integral Calculus
there are an infinite number of infinitesimal increments:

AA = Lim
n- 00
I ybx = fYdX

in which the new symbol S is used to represent this sum of an infinite

number of terms. We call the expression an integral and the process
integration.
Because of the relationship dA/dx = y that we referred to earlier,
we can in principle integrate any function by reversing the rules for
differentiation. (In practice this is often more difficult than it sounds,
because there are many functions that are difficult or impossible to
recognize as the results of differentiating other functions.) For
example, if

then
Sydx = x 3 + Xl + 5x + IX
where IX is a constant, because differentiation of this second function
with respect to x gives the first function.

4.2 Definite and indefinite integrals

We have seen that integration can be regarded as the inverse of
differentiation. Just as the derivative of any constant is zero, therefore,
the integral of zero is any constant and the constant IX that appeared in
the integration at the end of the previous section can have any value. A
constant of this sort is called a constant of integration. Its meaning,
and the reason for its appearance, can be seen by reflecting that in the
previous section we discussed the increase in an area A without
specifying what value of A we had to start off with.
The ambiguity implied by the need for a constant of integration can
be resolved by specifying limits between which the integration is to be
done. Thus, an expression such as
S(3x l + 2x + 5)dx
is ambiguous because although it tells us to add together terms of the
form (3x l + 2x + 5)dx it does not tell us where to begin and end. This
sort of integral is called an indefinite integral, but if we wish to start at
x = 1 and finish at x = 2 we can indicate this by writing the integral as
 Definite and indefinite integrals / 75

r
a definite integral:

(3x 2 + 2x + 5)dx

This can be regarded as the difference between the values of the

indefinite integral at x = 1 and at x = 2:

J(3x 2 + 2x + 5)dx = x 3 + x 2 + 5x + IX
= 7+1X at x=l, 22+1X at x = 2

On subtracting one from the other the ambiguity disappears because

regardless of the value of IX it vanishes from the difference:

It is sometimes convenient, as in this example, to indicate the limits

outside square brackets after the form of the indefinite integral has
been decided. An expression of this sort can be read as an instruction
to 'evaluate the expression contained in the bracket at both limits and
subtract one from the other'.
Notice that an indefinite integral is a function that can have any
value, not only because of the undefined constant, but also because it
can be evaluated at any value of x. By contrast, a definite integral
defines a number. It follows that evaluation of a definite integral
requires two stages: first, determination of the function correspond-
ing to the indefinite integral; and second, evaluation of this function at
the specified limits. In general, the second stage is much easier than the
first: given an indefinite integral we can always evaluate it, but it is not
unusual to begin with a function that we cannot integrate at all.
As a chemical example, consider the rate of a simple first-order
reaction, which can be expressed as

da
-= -ka
dt

where a is the concentration ofthe reacting substance at time t and k is

a constant (the minus sign is to show tha.t the reacting substance
disappears, i.e. its concentration decreases, as the reaction proceeds).
We can rearrange this to show that an infinitesimal change da in a is
76 / Integral Calculus
related to an infinitesimal change dt in t as follows:
da
-= -kdt
a

Suppose we want to know the total decrease in concentration that

occurs between t = 0 and t = t I: we can find this by integrating:

J =tt da
- = -k
1=0 a
ftt dt
0

In this case the indefinite integrals are straightforward because, on the

left-hand side, we know that dd (In x) = .!.. and, on the right-hand side,
x x
d
that dx (x) = 1. So

[ Jt=tt
Ina,=o=-kt o
[Jtt
If we define a = ao at t = 0 and a = al at t = t l , this becomes
Ina l -lnao = -ktl
which can be rearranged to
a l = aoexp( -ktd

and hence specifies the concentration a at any time.

In this derivation I have been pedantic in two respects: first, where
the limits are expressed in terms of a different variable from the one
that appears in the function to be integrated, as on the left-hand side
of the equation above, we should express them as 't = 0', etc., rather
than simply as '0', which, in the above example, would imply that a
was 0 at the lower limit. Second, t I and t were given different symbols
because they were used with different meanings - t was a variable
whereas t I, a constant, was a particular value of t. This distinction is
almost universally ignored in modern chemical practice, i.e. the result
above would normally be written as
a=aoexp(-kt)
This practice does no harm provided one realizes what has been done
fb
and one is willing to tolerate expressions such as dt in which the
 Definite and indefinite integrals / 77
same symbol t is used with two different meanings. It is especially
convenient in applications where one does want to treat the upper
limit as a variable in subsequent analysis.
In kinetic applications it is often clearer not to write definite
integrals at all but to work with indefinite integrals with constants of
integration that can be defined or evaluated. Therefore we might write
the initial equation for the first-order reaction as

f~
da
= -kSdt

and integrate it to give

Ina = -kt + IX
Notice that the same constant of integration IX serves both integrals: if
we had included constants of integration on both sides we could
subtract one from the other to get the constant we have called IX. If we
define a = ao at t = 0, we have
In a o = IX

and so
In a = - kt + In ao
and rearranging gives us the same result as before,

a = ao exp ( -kt)

but without any confusion about whether a and t are variables or

constants.
It is important to note that the constant of integration in this
example was not zero, even though it was evaluated at t = o. With a
few elementary kinds of function-most notably polynomials, i.e.
A + Bx + Cx 2 + Dx 3 + ...-the integral has a value of zero when the
lower limit is zero. However, although these functions are common in
elementary textbook accounts of integration they are not common in
scientific applications of integration and one must not suppose that
there is any kind of general rule that allows constants of integration
to be ignored. In kinetics, which accounts for most of the integrating
that a biochemist is likely to do, it is never safe to ignore con-
stants of integration or to assume that integrals are zero at zero
time.
78 / Integral Calculus
4.3 Simple integrals
In principle, any function can be differentiated, but the converse is by
no means true. The only functions that are easy to integrate are those
that are recognizable as derivatives of known functions. For example,

fAdX = Ax+a.

fAXdX = tAx2 + a.

fAX2dX = tAx 3 + a.

f.
or in general,
Axi+ 1
Ax'dx = - - - + a.
i+ 1
The rule for terms of polynomials is as follows: increase the index by 1,
divide by the new index, and add a constant.
The integral ofa sum (or difference) is also straightforward: ifu and
v are functions of x then
J(u+v)dx = Judx+ Jvdx
Finally, two other simple functions occurred as derivatives in Chapter
3 and consequently can readily be integrated:

dXX = lnx +a.

f

f exp(x)dx = exp(x) + a.

Although rules exist for integrating products and other functions,

these are by no means as easy to apply as the rules for differentiating
the same sorts of function. The need for them occurs so infrequently
in elementary biochemistry that they are best dealt with when they do
by looking them up in standard tables (see below). There are, however,
a few functions that occur so often in kinetics that it is useful to be able
to recognize them. The first is the following:

f _dX---,--
A+Bx
= ~ In (A + Bx) + a.
B
 Simple integrals / 79
This can readily be confirmed by differentiating the right-hand side as
a function of a function. Products of such functions also occur in
kinetics and can be integrated by realizing that they can be expressed
as sums of simple fractions. Thus:

where Al and A2 are unknown constants such that

AI(a2+b2x)+A2(al +b1x) =1
The identity sign ( = ) emphasizes that the relationship must apply
regardless of the value of x, and we can use this requirement to assign
values to the two constants. For example, because the relationship
must apply when x = 0 and also when x = 1 it follows that

and
AI(a2+b2)+A2(al +b l ) =1
Thus we have two simultaneous equations that can be solved for the
unknowns Al and A 2 • The two fractions produced in this way, which
are known as partial!ractions, can be integrated separately as above.
Another integral that is often needed in kinetic problems is the
following:

fAX:~X
The easiest way to deal with this is to define u = A + Bx, so that
du = Bdx and x = (u-A)/B
and then
xdx = _1 f(U - A)du
f A+Bx B2 u

= ;2 [
U - A In uJ + 0:
80 / Integral Calculus

= ~2[ A + Bx -A In(A + BX)] +IX

Although this result is not in itself worth memorizing, the method by
which it was obtained, in which we found a suitable function u that
allowed a much easier integration, is often useful in integration
problems.
The main results of this section are given in Table 4.1.

f.
Table 4.1 Common integrals

AX;+l
Ax'dx=--+a (for any value of i except -1)
i+ 1

f(U+V)dX = fUdX+ f Vdx + a

f dX
~ = Inx+a

f exp (Ax)dx = ~ exp (Ax)+ a

f dx 1
- - = -In(A+Bx)+a
A+Bx B

f xdx 1
- - =2[A +Bx-A In(A+Bx)]+a
A+Bx B

In all cases x is a variable. A. Band i are constants. u and v are functions of x. and cr. is a constant of
integration. All of the integrals in this table apart from the last are worth memorizing.

4.4 Integral of 1jx

d 1
Although I have asserted that -d (lnx) = - and, conversely, that
x x
fd: = In x + IX, I have not proved that this is so. A rigorous proof is
hardly essential in an elementary course but it is useful to invoke
plausibility arguments to increase one's sense of rightness for this
important relationship. Fig. 4.2 represents the function y = l/x. Let
us now define the definite integral between 1 and some number a as

f a

1
dx
- =
X
F(a)
 Integral of l/x / 81

!I

°O~----~--------~a----~------------x~

o h
Fig. 4.2. Plot of the hyperbola y = l/x. The u-axis is an alternative abscissa axis
related to the x-axis such that u = x/a. The area under the curve between x = a and
x = ab is unaffected by calling it the area between u = 1 and u = b.

without specifying anything about the nature of F(a). Then the area
bounded by the curve, the x-axis and the vertical lines x = 1 and x = a
is given by

J-a

1
dx
x
= F(a)

and the area between x = 1 and x = ab is given by

Jab dx = F(ab)
1 X

which can equally well be written as

a dx Jab -dx = F(a)+ Jab -dx
F(ab) = J 1 -+
X a X a X

Let us now define a new variable u such that u = x/a. Then u = 1

J
when x = a and so
ab dx b du
J a -X = 1 -=F(b)
U
82 / Integral Calculus
So
F(ab) = F(a)+F(b)

We see that F(x) has exactly the properties we associate with

logarithms: addition of F (x) values is equivalent to multiplication of x
values. We can more readily believe therefore that

fdX
~ = Inx +IX

(although it does not follow from the argument used here that the
result is In x rather than, for example, loglox; this is true nonetheless,
for reasons that I shall not discuss here).
The logarithm of a constant is another constant. It is legitimate,
therefore, and often convenient, to write the last result in the
following way (with IX replaced by In A):

f dX
~ = In x + In A = In Ax

Accordingly one should not be surprised to see constants of

integration appearing within logarithmic expressions as factors rather
than as constants to be added to the expression as a whole.

4.5 Differential equations

One of the commonest reasons why one needs to integrate ex-
pressions in science is that scientific laws frequently make predictions
about the derivatives of functions of interest rather than about the
functions themselves. Consider, for example, a reaction between two
molecules A and B:
A+B-+P
Elementary kinetic considerations suggest that if the mechanism is
simple the rate of reaction is likely to be proportional to the
concentration of A and also to the concentration of B, i.e.

dp = kab
dt
where a, band p are the concentrations of A, Band P, respectively, at
time t and k is a constant. Although this relationship is easy to arrive at
and easy to understand it does not tell us very much as it stands; in
 Differential equations / 83
particular, it does not tell us the extent of reaction as a function of
time. To convert it into a more informative form we must first
decrease the number of variables from 4 to 2. This can be done by
means of the stoichiometric requirement that (a + p) and (b + p) must
both be constants. If we define these as ao and bo respectively (the
values of a and b when p = 0), the original equation can be written
dp
dt = k(a o - p)(b o - p)

This is an example of a simple differential equation. The easiest way of

solving it is to separate the two variables so that only one appears on
each side of the equation:

f (a o - p)(b
dp = fkdt
o - p)
Solving the differential equation thus resolves itself into a problem of
solving two integrals. The right-hand side is trivial and the left-hand
side may be solved by partial fractions, so that
In (a o - p) In (b o - p) k
- a o -b o
+ ao -b o
= t+1X

Putting p = 0 when t = 0 we can evaluate the constant of integration

and rearrange the equation to
IX

In [ao(b o - P)] = (a o - bo)kt

bo(a o - p)
or
ao(b o - p)
b ) = exp[(a o - bo)kt]
o(a o - P
which is the solution of the original differential equation as it contains
no derivatives.
Not all differential equations permit the variables to be separated in
this way. Ones that do not are not common in elementary
biochemistry, however, and so it is not worth while devoting much
study to methods of solving them. It is much more efficient if one does
not plan to be a specialist in a branch of biochemistry that requires
differential equations to be solved frequently to seek expert help
rather than labour at a problem that may not even be soluble at all. It
84 / Integral Calculus
is always wise to remember that even quite simple kinetic models can
lead to intractable mathematics and there is little point in wasting
time over an impossible problem. For example, the simplest model
commonly discussed in enzyme catalysis is the Michaelis-Menten
mechanism:
k+1 k+2
E + S ~ESE"'E+P

L I k-2
eo-x so-x-p x p

With concentrations defined as indicated under the equation appli-

cation of elementary kinetic laws leads without difficulty to the
following pair of simultaneous differential equations:
dx
dt = k+ I(e o -x)(so -x - p)-L IX -k+ 2X +L2(eO -x)p

dp
dt = k+2x-L2(e O -x)p

Although it is not difficult to remove x from these equations by using

the second one to express x and dx/dt in terms of p and its derivatives
the resulting equation in p and t has no known solution.
If one arrives at an equation that is impossible to solve there is little
point in relying on one's mathematical knowledge and ability.
Instead, one has the choice between trying to solve it numerically
rather than algebraically, an approach that is largely outside the scope
of this book, and restricting the problem so as to make it soluble. This
latter approach is very important and it demands scientific rather than
mathematical knowledge. Thus, a mathematician might point out
that the problem discussed above becomes soluble if k + I = k _ 2' but
the biochemist would most likely regard this as an unreasonable
restriction and would prefer to look for a restriction that could be
made to be true by suitable choice of experimental conditions. By far
the most commonly invoked restriction in enzyme kinetics is to
require So to be much larger than eo and to consider only the time scale
in which dx/dt can be regarded as negligible. This is the basic
restriction used in steady-state kinetics, but it is not the only possible
one: alternatives might be to assume So }> eo and p negligible, the pre-
steady-state condition, or to assume eo }> so' the single-turnover
 Numerical integration: evaluating the area under a curve / 85
condition. My purpose in mentioning these, however, is not to derive
any kinetic results, but to emphasize that the decisions to be made are
scientific, not mathematical: one needs some knowledge of enzymes
and techniques for studying them to know that it is usually easy and
convenient to achieve steady-state conditions but that assuming
k + 1 = k _ 2 is not likely to be useful.
To summarize this section, the most efficient approach to differen-
tial equations in biochemistry is to proceed as follows: first, check
whether the variables can be decreased to two that can be separated
on to the two sides of the equation. If so, it can be solved in principle
by integrating the two sides. If this cannot be done immediately then
tables of integrals or expert advice are likely to prove much more
rapid than blind struggling. If the variables cannot be separated or the
problem appears impossible for some other reason, then try to
introduce biochemically sensible restrictions or approximations that
allow the simplified equation to be solved. If all else fails then seek
expert help, remembering that a mathematician will probably know
much less biochemistry than you do and will need some guidance over
matters that you regard as obvious.
Finally, a word about tables of integrals and differential equations.
Tables of integrals are easy to use, provided you remember two things:
first, constants of integration are usually omitted, on the assumption
that the user knows enough mathematics to supply a constant of
integration without constant reminders; second, the symbol 'log'
means 'In' not, as in biochemical practice, 'logIO" Tables of differential
equations are much more difficult to use as they commonly assume
some understanding of the principles of solving differential equations.

4.6 Numerical integration: evaluating the area under a

curve
It sometimes happens that we need to know the area under a curve but
cannot do the necessary integration exactly, either because we do not
have an analytical expression for the curve or because we cannot
integrate it. For example, the curve in Fig. 4.3 might represent the
absorbance of the effluent from a chromatographic column as a
function of the volume: if the absorbance of a sample is proportional
to the concentration of some substance of interest, the total amount of
the substance is proportional to the area under the curve. If it is not
possible or convenient to integrate the curve analytically, the simplest
86 / Integral Calculus

!J

!/l ---.- .... - - ... ---

3" 4"

!/2 - --- ---- ----

o h 2" 3,1, 4,1, 5h 7h 8" .x

Fig. 4.3. Trapezium method for estimating the area under a curve. The shaded area
h
between x = Oand x = his 2(yo + Yl)' but this is the area of the complete trapezium and
thus significantly overestimates the area under the curve, which ought to exclude the
diagonally shaded segment. Similar inaccuracies result from treating the other strips as
trapezia, and the inset shows an example of significant underestimation. The alternative
method discussed in the text, 'Simpson's rule', is much more accurate and the parabolic
arcs that it assumes are hard to distinguish from the true curve, and for that reason are
not shown in the figure.

way of estimating the area is to divide it into n parallel strips as shown

and replace the short arcs between the ends of the strips by straight-
line segments. It is then a matter of simple geometry to determine the
areas of the trapezia produced. For example, between Xo = 0 and
Xl = h, the area of the rectangle below the curve is Yoh and the area of
the triangle between the rectangle and the curve is (h/2)(Yl - Yo),
so the area of the whole trapezium is (h/2)(yo + Yl)' Similarly, the area
of the adjacent trapezium is (h/2)(Yl + Y2)' and so on. Every yexcept
the first and last appears in the expressions for two adjacent trapezia,
so the total area A under the straight-line segments is given by

A=ho+Yl+Y2+Y3+ ... +Yn-l+hn

 Numerical integration: evaluating the area under a curve / 87
Although simple, this formula is not very accurate (unless n is made so
large that the calculation becomes very tedious) because the straight-
line segments leave appreciable areas unaccounted for, as illustrated
in the inset to Fig. 4.3.
A more accurate method is to assume that over a limited range the
curve can be approximated by a parabola. Thus instead of the
unknown (or known but unintegrable) function Y = f(x) we write a
quadratic
Y ~ a+bx +cx 2
Ifwe evaluate Y at three values of x, Xo = 0, Xl = hand X2 = 2h, and
treat the approximation as exact, we have
Yo =a
y1 = a + bh + ch 2
Y2 = a +2bh +4ch 2
Eliminating a by subtraction, we have
Yl - Yo = bh + ch 2
Y2-Yl =bh+3ch 2
and bh by further subtraction,
Y2 - 2Yl + Yo = 2ch 2
So

and
bh = Yl-Yo-ch 2 = 2Yl -~YO-ty2
Integrating the original quadratic, and substituting these values of a,
bh and ch 2, we have
88 / Integral Calculus
and so on. Notice that every odd-numbered Y appears only once,
whereas every even-numbered Y apart from Yo and Yn appears twice, so
the total area A is given by
h
A =3 [Yo + 4(Yl + Y3 + ... + Yn- 1)+ 2(Y2 + Y4 + ... + Yn - 2) + Yn]
This formula is known as Simpson's Rule. It is hardly more difficult to
apply than the trapezium method I considered first, and it is much
more accurate. For most curves that are not highly complex the
formula is adequately accurate if the number of strips is about ten,
and in most experimental examples inaccuracies in measuring the Y
values is a more important source of error than the formula.

4.7 Problems
(4.1) Integrate the following:

(a) f(X 2+ 3x + l)dx

(b) f(x +~ )dX

(c) f !X3X
2

(d) fex P( - 3t)dt

(4.2) Evaluate the following definite integrals:

(a) f: 3xdx

(b) f2 dx

til
I X

(c) 3x 2 dx

(d) t22 x 3 dx

(4.3) Express the following equation in the form of partial fractions

and use the result to evaluate Jydx:
y(3x+ 1)(x+2) =5
 Problems I 89
(4.4) Integrate the following expression:

f (A + Bx)dx
C+Dx
(Note: this can most easily be done by separating it into two
integrals.)
(4.5) The Michaelis-Menten equation can be written in the following
form:
dp Vs
dt Km+s
in which p and s are the concentrations of product and substrate,
respectively, at time t, and V and Km are constants. Defining s = So
and p = 0 at t = 0, use the one-to-one stoichiometry of the reaction to
eliminate s and solve the resulting differential equation to get an
expression for Vt in terms of p.
(4.6) The equation considered in problem (4.5) takes no account of
the possibility of inhibition by the accumulating product. Show that
the following equation, in which Kp is a constant and the other
symbols are as in problem (4.5), can be integrated to provide an
equation of the same form as the solution to problem (4.5):
dp Vs
dt Km(l + pIKp) + s
What would be the form of a plot of a plot of tlln [so/(so - p)] against
pJln [so/(so - p)]?
(4.7) The relationship between the intensity oflight 1 and the distance
x through which it has passed in an absorbing solution can be
expressed as follows:
dl
- = -kcI
dx
where k is a constant and c is the molar concentration of absorbing
substance. Putting 1 = 10 at x = 0, integrate this expression to obtain
an expression for 1 in terms of x. If the molar absorbance A is defined
such that
log (loll) = Acx
what is the relationship between A and k?
90 I Integral Calculus
(4.8) In a spinning ultracentrifuge cell the outward flow rate of the
mass of solute m is given, ignoring diffusion, by
dm w 2 x(1- Vp)M ADe
-= '
dt RT
where t represents time, x the distance from the axis of rotation, M,
the relative molecular mass of the solute, A the cross-sectional area of
the cell, D the diffusion coefficient, e the concentration of solute in gil,
and the other symbols represent constants that are either known or
easily measurable. The effect of diffusion can be represented by
dm de
- = -AD-
dt dx
After spinning for a long period the system achieves equilibrium and
becomes time independent. Then the net flow rate, given by the sum of
the two expressions, is zero at all points. Obtain an expression relating
e and x at equilibrium and hence suggest a plot that would allow M, to
be conveniently measured.
(4.9) The following values of y at various values of x show
measurements of the height of a recorder trace in a chromatographic
separation. Estimate the areas under the two peaks by using
J J
Simpson's rule to evaluate ~ ydx and ~ 7 ydx. In each case compare
the results with four strips with those obtained with eight.

x y x y x y

0 0.03 7 5.75 14 2.68

1 0.04 8 2.69 15 1.21
2 0.51 9 1.62 16 0.28
3 2.53 10 2.27 17 0.03
4 5.81 11 3.41 18 0.12
5 7.64 12 4.08 19 0.15
6 8.11 13 3.82 20 0.09
5 Solving Equations

5.1 Linear equations in one unknown

An equation such as
4+x = 8-2x
contains only one unknown quantity, x, and is linear, because it
contains no terms higher than the first power of x. Although it is easy
to solve such an equation it is nonetheless useful to examine the
processes involved as they illustrate in a simple way the methods to be
used for solving more difficult equations. The first essential is to
rearrange the equation so that terms of the same type, in this case
constants and terms in x, are brought together. It is often convenient
to collect all constants on the right-hand side and other terms on the
left-hand side. This may be done by use of the principle that we can
apply any operation we like to an equation (other than multiplying or
dividing by zero) provided that we apply it identically to both sides.
So, for example, we can subtract 4 from both sides of the above
equation and add 2x to both sides:

4+x-4+2x = 8-2x-4+2x
i.e.
3x = 4
The operation of subtracting, for example, 4 from both sides is clearly
equivalent to 'bringing it from one side to the other of the equation
and changing its sign'. Although this is a convenient and proper way
of carrying out the operation one should realize that it is no more than
an example of applying the same operation to both sides of an
equation.
Continuing with the above example we may divide both sides of the
91
92 / Solving Equations
last equation by 3 to obtain the solution:
x = 4/3 = 1.33 approx.
Equations that do not seem to be linear at first sight can often be
rearranged into linear form by application of the same methods. For
example, the equation

~=3
5+2x
becomes linear after both sides of the equation are multiplied by
5+2x:
8x = 15 +6x
which can be solved as before to give x = 7.5.
The only difficulty that may arise in solving this sort of equation is
that it may turn out to be singular, which means that it contains less
information than it seems to contain. Consider, for example, the
following equation, which appears to be an ordinary linear equation:
5x+2-3(x+8) = 2x-22
If, however, we collect terms, we find
(5-3-2)x = -22-2+24
i.e. Ox = 0, which is true regardless of the value of x and consequently
tells us nothing about the value of x. Examples as simple as this seem
rather artificial and trivial, but singular equations can be a serious
problem because they can appear in the last stages of solving a more
complex equation or set of equations. This often happens in the
analysis of experiments that have been badly planned so that they do
not in fact provide the information they were intended to provide, and
I shall consider some more realistic examples later in this chapter.

5.2 Rearranging equations

Equations in science often contain the information we require but in
inverse form, i.e. instead of giving the required unknown in terms of
known quantities an equation may express a known quantity in terms
of the unknown. In such cases we can often arrive at a more useful
formulation of the equation by solving it in the same way as we would
if it were expressed numerically.
 Simultaneous linear equations / 93
For example, in studies of individual enzymes it is usual to regard
the rate of reaction v as something that is determined by the properties
of the enzyme and by the concentrations of relevant species such as
the enzyme and its substrate. We might write, for example,
kcateoS
v=---
Km+ s
where eo and s are the concentrations of enzyme and substrate,
respectively, and kcat and Km are constants. When we consider the
enzyme not as an isolated catalyst but as a component of a metabolic
pathway, however, we might argue that it is not s that determines v,
but v that determines s, because v is decided by the flux through the
whole pathway whereas s reflects the capacity of the enzyme to
remove the particular substrate. If the enzyme is very active, s will be
small; if it is not very active, s may be large; but provided it is not
saturated with substrate the value of v will be decided by the rate at
which other enzymes are supplying the substrate. Accordingly, we
may ask, how can s be expressed in terms of v? This is exactly
analogous to the second example in the previous section except that
now we must deal with constants expressed as symbols rather than as
numbers. The method is the same: we multiply by Km + s,
v(Km + s) = kcateOs
collect terms in s on the left-hand side and other terms on the right-
hand side,

and divide by the coefficient of s, (v - kcate o), to give the solution:

5.3 Simultaneous linear equations

If we have two or more unknown quantities a single equation is
insufficient for defining their values. For example, the following
equation:
2x+3y = 12
can be satisfied by x = 3, y = 2, but this is not a unique solution and
94 j Solving Equations
there are an infinity of other possible solutions, such as x = 4.5, y = 1.
In fact regardless of the value of x there is a value of y given by
(12 - 2x)j3 such that x and y satisfy the equation exactly. When there
are fewer equations than unknowns the problem is said to be
underdetermined. For there to be a unique solution the number of
equations must be equal to the number of unknowns. The reason for
this is that each eq uation can be used to express one unknown in terms
of the others but once that has been done no further information is
available from that equation. If one loses an equation for each
unknown that is eliminated it is clear that one can only finish with a
single equation in a single unknown if the numbers of equations and
unknowns were initially the same. Although the approach is general I
shall consider only the case of two equations in two unknowns,
because it is rare in elementary biochemistry for one to have problems
with three or more unknowns. Such equations are called simultaneous
equations because they define two (or more) relationships that are true
simultaneously.
Let us apply the approach indicated above to the following pair of
equations:
2x+3y = 12
2x+ y = 8
Treating the first equation as a linear equation in y with x as an
unknown constant, we have
3y = 12-2x
Hence
y = 4-2xj3
Substituting this for y in the second equation (the first being of no
further use to us) we have
2x+4-2xj3 = 8
which may be solved as a simple linear equation in x to yield x = 3,
and, after substitution back into the expression for y, y = 2. This
approach is superficially attractive because it can be applied
mechanically, i.e. without thought, and it ought always to work.
Nonetheless it is not the best way of solving simultaneous equations
and one can often proceed more simply and with less likelihood of
error by thinking first.
 Simultaneous linear equations / 95
For the particular example we have considered at least two
improvements to the above method are possible. First, we ought to
notice immediately by inspection that the second equation gives a
simpler expression for y than the first does, i.e.
y = 8-2x
and by substituting this into the first equation we can avoid fractions:
2x+3(8-2x) = 12
etc., with the same solution.
Second, we should also notice from inspection that x is multiplied
by the same constant 2 in both equations. Consequently, we can
eliminate x immediately by subtracting the second equation from the
first:
2x +3y = 12
2x+ y= 8
2y = 4
Hence y = 2, etc. The subtraction of one equation from another is
justified by the fact that if 2x + y = 8 then subtracting 2x + y is
equivalent to subtracting 8 and so by subtracting 2x + y from the left-
hand side and 8 from the right-hand side we are applying identical
operations to the two sides of the equation.
This third approach is clearly the simplest and quickest of the three,
but it seems to require the coincidence of finding the same term - 2x in
this example - in both equations. In fact we can always ensure that
such coincidences occur by multiplying by appropriate factors in the
first instance. For example, if
3x-5y = 3
2x+3y=21
then multiplying the first equation by 2, the coefficient of x in the
second equation, and the second equation by 3, the coefficient of x in
the first equation, ensures that x has the same coefficient in both
equations:
6x -lOy = 6
6x+ 9y = 63
from which we can proceed as before to obtain the solution, x = 6,
y= 3.
96 / Solving Equations
An alternative way of achieving the same effect is to divide each
equation by the coefficient of x in that equation (or whatever
unknown is chosen for elimination). In the above example, we would
have:
x-1.667y= 1
x + 1.500y = 10.5
As this approach brings about the appearance of non-integral
coefficients it may seem rather unappealing, but it is nonetheless a
good general approach. One reason for this is that in experimental
problems (as opposed to invented ones in textbooks) the coefficients
are hardly ever integral in the first place. Another is that it tends to
ensure that the calculation is done in terms of numbers not very
different from 1. This tends to decrease numerical inaccuracies,
whether the calculation is done by hand or by computer. Although
these numerical considerations may seem trivial, and indeed usually
are trivial for simultaneous equations in only two unknowns, they
become of great importance in solving sets of equations in several
unknowns.

5.4 Determinants
The methods I have described for solving simultaneous equations
expressed in numerical terms can also be applied to equations in
which the coefficients are not assigned numerical values. Consider the
following general pair of linear simultaneous equations:
A1x+B1y+C 1 = 0
A 2x+B 2y+C 2 = 0
Multiplying the first by A2 and the second by Al we have
AIA2X+A2BIy+A2CI =0
AIA2X+AIB2y+AIC2 =0
and after subtraction of the first equation from the second,
(A 1B 2 - A 2B 1)y + (A 1C 2 - A 2Cd =0
Hence
 Determinants / 97
and, similarly,

A special notation has been devised to allow this last result to be

er IA~~' I~ lA,
expressed in a particularly elegant way, as follows:

IB,X I B, I
B2 C2 A2 C2 A2 B2
The denominators of these expressions are called determinants. Any
determinant is written as a square array of numbers or algebraic
symbols between vertical lines and it has a numerical or algebraic
value that can be found by applying a general rule. Although one can
have any size of square I shall only consider 2 x 2 determinants, which
are not only particularly simple but are also the most important for
our purposes: the value of such a determinant is the product of the
top-left and bottom-right elements minus the product of the bottom-
left

To write the solution of a set of equations in determinant form we

proceed in accordance with the following set of rules, First, the
original set of equations must be organized systematically, so that the
unknowns appear in the same sequence in every equation (with
appropriate gaps or zeroes if there are any unknowns missing from
particular equations) and the constant as the last term on the left-hand
side of each equation. For each equation the right-hand side should
then consist of zero. Each expression except the last in the set of
equations representing the solution consists of a fraction with a
different unknown in each numerator and a determinant in each
denominator. For each unknown the determinant in the denominator
consists of all the coefficients in the original equation, in the same
order, except for those of the unknown in the numerator. The last
(right-hand) expression is also a fraction, with 1 in the numerator and
a denominator consisting of a determinant containing all of the
coefficients in the original equations but not the constants. The left-
hand expression has a positive sign and the remainder alternate in sign
98 / Solving Equations
on reading from left to right. Applying these rules to the following
pair of equations:
2x+3y = 5
y-5 = 4x
we first rearrange the equations into standard form,
2x+3y-5 = 0
4x- y+5 = 0
and then write down the solution in determinant form:

or

=- =- = =
x/1O = - y/30 = -1/14
so x 10/14 5/7, y 30/14 15/7.
It often seems more natural to write each of the original set of
equations with the constant on the right-hand side rather than as the
last term on the left-hand side, i.e. (for the same example):
2x+3y = 5
4~- y=-5
If this is done the solution can be written in the same way as before
except that the sign of the constant is the opposite of what it would be
if the equations were written in standard form. For two simultaneous
equations this means that the last numerator is - 1 rather than 1:
x -y -1

1-: -~I=I~ -~I=I~ -:1

The determinant in the right-hand (constant) expression is especially
important for any set of equations because it appears in the solution
for every unknown. Moreover, its value indicates whether a unique
solution to the set of equations exists: if it is zero there is no unique
~olution and the equations are said to be singular. It is therefore useful
to be able to recognize characteristics of a determinant whose value is
zero (these apply to all determinants, not just those of order 2). First, if
 Quadratic equations / 99
any row or column consists entirely of zeroes the determinant is zero.
This is always true for a set of equations in which the number of
equations is different from the number of unknowns. Second, if any
two rows or any two columns are identical the determinant is zero.
This occurs if any equation is identical to any other, so that the
number of different equations is less than the number of unknowns, or
if the information given about any two unknowns is identical, so that
they cannot be distinguished. Third, if any row or column can be
obtained by multiplying all of the elements in another row or column
by the same constant the determinant is zero. This corresponds to the
case where one equation can be obtained from another merely by
mUltiplying every term by the same constant. All of these charac-
teristics that cause determinants to be zero can be recognized
immediately by inspection, but there is a fourth that is less obvious. It
is not possible at all with only two simultaneous equations but can
render larger sets of equations insoluble: if any row or column is a
linear function of two or more other rows or columns the determinant
is zero. For example, the determinant
237
146
o 1 1
can be seen to have a value of zero by the fact that each element in the
third column is equal to twice the corresponding element in the first
column plus the corresponding element in the second column.

5.5 Quadratic equations

Suppose we have the equation
(x-5)(x-2) = 0
then it is obvious that the left-hand side is equal to zero if either ofthe
two terms in parenthesis is zero, i.e. x - 5 = 0 or x - 2 = 0, because
anything multiplied by zero is zero. Thus there are two values of x,
x = 5 and x = 2, that satisfy the equation. If we multiply out the left-
hand side we have a typical quadratic equation:
x 2 -7x+ 10 = 0
which is identical to the original equation and hence must have the
same pair of solutions, x = 5 or x = 2.
100 I Solving Equations
In principle, a quadratic equation can be solved by reversing the
above process, i.e. by finding a pair of factors that give the left-hand
side when multiplied together. In practice most of the equations we
meet in science cannot be solved in this way because no rational
factors exist. (Strictly, a rational number is one that can be expressed
as a ratio of two integers, but it will do little harm if we think of it
loosely as a 'simple' number.) Nonetheless, we can apply the same sort
of method to obtain a general solution to the problem of how to solve
a quadratic equation. Consider the following equation:
ax 2 +bx+c = 0
which we can write as
x2 + (bla)x + (cia) = 0
Let us compare the left-hand side of this with the square of (x + bI2a),
which is

Apart from the third term this is the same as the left-hand side of the
equation to be solved, so, subtracting the latter (which has a value of
zero) we have
(x + b12a)2 = b 2 /4a 2 - cia = (b 2 - 4ac)/4a 2
and hence
± J(b 2 -4ac)
x+ b12a =-~-2-a--

which rearranges to·

-b ± J(b 2 -4ac)
x=------"--,------
2a
We use the sign ± because either of the two expressions with square
equal to (b 2 -4ac) will satisfy the equation: remember that the
square-root sign is not ambiguous and specifies the positive square
root only. This result is the general solution for a quadratic equation.
Whether it is worth memorizing depends on how often you plan to
solve quadratic equations, but it is important to understand how to
use it even if you do not memorize it.
The quantity (b 2 -4ac) is known as the discriminant of the
equation ax 2 + bx + c = o. It is a useful quantity because its value
 Quadratic equations / 101
indicates the nature of the solutions of the equation. If it is positive,
the equation has two real and unequal solutions; if it is zero there are
two real solutions, but they are identical; if it is positive and a perfect
square, the solutions are rational and the simple factorization method
mentioned at the beginning of this section could be used for solving
the equation; if it is negative there are no real solutions. This last result
follows from the fact that a negative number does not have a real
square root, and it is important because it shows that one can have an
apparently elementary equation that has no solution. For elementary
purposes we can omit the word 'real' from the preceding statements,
because in elementary work only real solutions need be considered as
solutions at all. In more advanced mathematics it becomes convenient
to define so-called 'imaginary' numbers as the square roots of negative
real numbers, and 'complex' numbers as the results of adding real and
imaginary numbers together. However, although these kinds of
numbers have important applications in some scientific contexts - in
quantum mechanics and in the theory of electrical circuits, for
example - it is difficult to think of an application in elementary
biochemistry and so I shall not consider them further in this book.
It often happens when quadratic equations occur in scientific
contexts that although there are two solutions mathematically only
one of them is physically meaningful, and this can usually be identified
by inspection. Consider, for example, the equilibrium between the
three adenylates:
2ADP~ATP+AMP

which has an equilibrium constant of about 1/2 under physiological

conditions, i.e.
[ATP] [AMP]
-
[ADP]2 2

Suppose now we mix 8mM-ATP with 2mM-ADP in the presence of

the enzyme adenylate kinase, which catalyses the equilibration: what
will be the concentrations of the three adenylates at equilibrium? Ifwe
put [AMP] = x, the stoichiometry of the reaction requires that
[ATP] = 8 + x, [ADP] = 2 - 2x, and so the equilibrium expression
can be written as
(8+x)x 1
(2 - 2X)2 2
102 / Solving Equations
or, after rearranging,
x2-12x+2 = 0
The formula we derived above yields the following two solutions:
x = 11.8 or 0.1 69mM. At first sight these are both physically
meaningful if we simply consider them as AMP concentrations, but
calculation of the corresponding ADP and ATP concentrations
shows that
if [AMP] = 11.8mM, then [ADP] = -21.6mM, [ATP] = 19.8mM
if [AMP] =O.169mM, then [ADP] = 1.66mM, [ATP] =8.17mM
As concentrations in the real world cannot be negative, only the
second solution is physically acceptable and the first must be
discarded. As a general rule one should always check that the solution
of an equation makes physical as well as mathematical sense. It is
perfectly possible to have results that are entirely sound mathemati-
cally but are nonetheless physically nonsensical.

5.6 Graphical solution of equations

Sometimes we have to solve equations that are more complex than
quadratics. Exact analytical solutions for certain kinds of higher-
order equations exist, but these are little used and the practical choice
is between graphical solution and Newton's method, which I shall
discuss in the next two sections.
Let us consider the following equation as an example of the sort of
equation that would be very difficult (if not impossible) to solve
analytically:
x 2 +lnx = 7
We can solve this quite easily by a graphical method if we define a
function f(x) as

This function must be zero when x is equal to the solution of the

equation. If we plotf(x) against x (Fig. 5.1), the line crosses the x-axis
at about x = 2.47. As the x-axis is an expression of the relationship
f(x) = 0 it follows that x = 2.47 is an approximate solution to the
equation.
 Graphical solution of equations / 103

fix) 3

Or---4---~----+----+~~r----r--~r---4-----~

-I

Fig. 5.1. Graphical solution of the equation x 2 + In x = 7. The line is a plot of the
functionf(x) = x 2 + In x -7, which must be zero when the equation is satisfied. Thus x
= 2.47 is an approximate solution.

The advantage of graphical solution is that it is easy to understand

and use, but it has compensating disadvantages also. First, it is
laborious, as one has to evaluate the function at numerous different x
values, many of which turn out not to be close to the solution. Second,
it is limited in accuracy by one's draughtsmanship and ability to read
off the solution accurately. In principle, one could improve the
accuracy of the above solution by drawing another graph on a larger
scale covering only a narrow range of x values, say from 2.46 to 2.48.
This would add substantially to the labour required, however, and
one would normally do better to proceed to Newton's method (next
section) if one needed a more accurate solution than the one given by
the preliminary graph.
One can also use graphical methods for solving simultaneous
equations. Although this is not a particularly convenient way of
dealing with straightforward linear simultaneous equations of the
sort I discussed earlier in this chapter, it is nonetheless important for
two reasons. First, it provides one of the simplest ways of solving non-
linear equations. Second, the method illustrates the principle under-
lying certain graphical techniques used in enzyme kinetics, most
notably the Dixon plot for determining inhibition constants. For this
104 / Solving Equations
reason I shall discuss the method briefly. If we have the equations
5x+7y=6
2x- y = 1
we can rearrange them as two expressions for y in terms of x:
y = (6-5x)j7
y=2x-l
Each of these defines a straight line if y is plotted against x; the point
of intersection, the only point common to both lines, provides the
only (x, y) pair that satisfies both equations (Fig. 5.2).
The Dixon plot is a more complex application of the same principle.
Competitive inhibition is the name given to the kind of inhibition of an
enzyme-catalysed reaction defined by the following equation:
Vs
v=------
Km(1 +ijKJ+s
in which v is the rate observed at concentrations i and s of inhibitor
and substrate, respectively, and V, K m , K; are constants. The Dixon

Fig. 5.2. Graphical solution of the simultaneous equations 5x + 7y = 6, 2x - y = I.

The co-ordinates of the point of intersection provide the only (x, y) pair that satisfies
both equations, i.e. x = 0.684, y = 0.368.
 Graphical solution of equations j 105
plot is especially convenient in the common experimental circum-
stance where our main objective is to measure K j , the other two
parameters V and Km being either of less immediate concern or
known independently. If we have values of v measured at several i
values at each of two s values, Sl and Sl' we can determine K j by
plotting l/v against j at both S values. If we define VI for the V values
measured at S = Sl' the two straight lines are given by
l/VI = (KmjVsI)+(ljV)+(Km/VKjSI)j
l/Vl = (Km/Vs 1 ) + (l/V) + (Km/VKh)j
The point of intersection of the two lines defines the value ofj at which
l/vI = l/Vl and hence the value at which the two right-hand sides are
equal:
(Km/Vs 1 ) + (l/V) + (Km/VKjsl)j = (Km/Vs 1 ) + (l/V) + (Km/VKjsz)i
Despite its messy appearance this is just a simple linear equation in i
which may be solved in the ordinary way to yield i = - K j • Thus - K j
is given by the value of j at the point of intersection of the two lines
(Fig. 5.3).

i=-Kj 0 i

Fig. 5.3. Dixon plot for determining a competitive inhibition constant: this is an
application of the method illustrated in Fig. 5.2.
106 / Solving Equations
Another and somewhat simpler application of the graphical
solution of simultaneous equations in enzyme kinetics is found in the
direct linear plot for determining the parameters Km and V in the
Michaelis-Menten equation. If the rates VI and V 2 at substrate
concentrations s I and S2' respectively, are given by the following pair
of equations:
VS I
VI = ---''--
Km+sl
VS 2
V2 =
K m+s 2
then Km and V can be evaluated graphically by rearranging the
equations to show the dependence of Von Km:
V = + (VI/sl)K m
VI

V = V 2 + (V 2 /s 2 )K m
Both of these define straight lines when V is plotted against Km. The
first may most easily be drawn as a straight line with intercepts - S Ion
the Km axis and VI on the V axis, and the second similarly (Fig. 5.4).

-5,

Fig. 5.4. Determination of the parameters of the Michaelis-Menten equation by the

graphical method illustrated in Fig. 5.2.
 Newton's method / 107
The co-ordinates of the point of intersection give Km and V. This plot
is exactly analogous to the solution of simultaneous equations
illustrated in Fig. 5.2.

5.7 Newton's method

Although graphical solution is easy to understand and carry out its
accuracy is limited and it can be excessively laborious. For com-
putational purposes Newton's method is much better. It depends on
the fact that most smooth functions behave over short ranges in
simple ways that can be predicted approximately from knowledge of
the function value and its first derivative at particular points.
I shall use as an example the same equation as we used for studying
the graphical method:
x 2 +lnx = 7
Then if we define a function f(x) as
f(x) = x 2 + In x - 7
we have,as before, solved the equation if we find a value ofx for which
f(x) = O. At any arbitrary starting point Xo the tangent to the graph of
f(x) against x is given by
y = f(x o) + (x - xo)f'(x o)
where!'(x o) is the first derivative off(x) evaluated at x = Xo. As seen
in Fig. 5.5, this tangent is approximately coincident with the curve
representing the true function over a finite range of x values.
Consequently, if we put y = 0 and solve for x we should get a value Xl
that is nearer the solution than x o, i.e. we put
f(xO)+(x 1 -xo)f'(x o) = 0
and rearrange to give
f(x o)
Xl = Xo -!' (x o)
This is the fundamental relationship of Newton's method. Let us see
how it can be used for solving the equation that we set out with. For
that equation, differentiation off (x) with respect to x shows that!' (x)
is given by
!'(x) = 2x + Ijx
108 / Solving Equations

flXl

4 tangent drawn
at x= 3

o~-----------+------------~----~----~'-~~
2

-2

-4
tangent drawn
at x = I
-6

-8
'- flxl=x'+tnx-7

Fig. 5.5. Newton's method for solving the equation x 2 + In x -7 (see Fig. 5.1). The
starting guess x = Xo = 1 provides the tangent that crosses the x-axis at x = XI = 3.
This improved guess for the solution provides a second tangent that crosses the axis at
x = x 2 = 2.51. Further tangents calculated in the same way give, successively,
x = 2.469, 2.46904, 2.4690423 .... This method permits much greater accuracy than
one can hope to achieve with a graphical method.

If we choose an arbitrary starting point of Xo = 1, then

Xl = 1 +6/3 = 3
We can see from Fig. 5.5 that although this is not the correct solution
it is closer to it than Xo = 1. So we proceed with a second
approximation, replacing Xo with Xl and Xl with x 2 , and so on:
X2 = 3 - 3.099/6.333 = 2.51
X3 = 2.51 - 0.220/5.418 = 2.469
x 4 = 2.469 + 0.0002/5.3430 = 2.46904
One can continue until any desired accuracy is achieved.
 Approximate methods / 109
In the above example one could recognize that the successive
approximations were nearer and nearer to the correct solution by the
fact that the numerical values (i.e. the values ignoring the signs) ofJ(x)
in the numerator of the second term were getting smaller and smaller
with each approximation: - 6,3.099,0.220, - 0.0002. Sometimes this
does not happen and instead of getting better and better the successive
approximations given by Newton's method get worse and worse. This
may happen, for example, iff' (xo) is close to zero, so that Xo is close to
a maximum or a minimum ofJ(x). One can usually cure this problem
when it occurs by trying a different value of Xo. In severe cases a
preliminary graphical exploration as described in the preceding
section should reveal a suitable range of values to try. For equations
with two or more solutions, Newton's method does not guarantee to
find all of them; it will usually proceed towards the solution closest to
Xo. Again, graphical exploration is useful in such cases to ensure that
no solutions have been overlooked.

5.8 Approximate methods

In textbook illustrations of methods for solving equations one
commonly deals with examples in which the coefficients are
numbers-often integers-in the range 1 to 10. Real life is often very
different and one may instead have to solve equations with coef-
ficients ranging over several orders of magnitude. This has two
consequences: first, methods that work well with simple examples
may work badly with real ones; and second, there is much more scope
for the use of approximations than one might expect from the study of
elementary examples.
To illustrate these points, let us consider how to calculate the pH of
a 0.1M solution of monosodium glutamate, given that glutamic acid
has three ionizations, with pKa values of 2.3,4.3 and 9.7. These convert
to Ka values of 5 x 10- 3,5 x 10- 5 and 2 x 10- 10 , respectively, and so
the equilibria can be formulated as follows:

where G 1 + represents protonated glutamic acid with a net charge of

+ 1, GO represents glutamic acid with a net charge of zero, etc.
(actually three different states with charge zero can be drawn, but for
110 / Solving Equations
calculating pH values it is not necessary to distinguish between them,
and GO comprises all of them; similarly, G 1- comprises three states
with charge -1). From the definitions of the equilibrium constants
we can write
[GO] = 5 x 1O- 3[G 1+]/[H+]
[G 1-] = 5 x 1O- 5 [GO]/[H+] = 2.5 x 1O- 7 [G 1+]/[H+]2
[G 2-] = 2 x 1O- 10 [G 1-]/[H+] = 5 x 10-17[G 1+]/[H+]3
From the stoichiometry we have the following conservation equation:
[G 1+] + [GO] + [G 1-] + [G 2-] = O.lM
and in addition, from the fact that the total concentration of negative
charge must equal the concentration of Na + ions,
_[G 1+]+[G 1-]+2[G 2-] =O.lM
Substitution of the first three of these five expressions into the last two
gives
1+ ( 5 x 10 - 3 2.5 x 10- 7 5 x 10- 17 )
[G ] 1 + [H+] + [H+]2 + [H+]3 = O.lM

1+ ( 2.5 X 10- 7 1 x 10- 16 )

[G ] -1 + [H+]2 + [H+]3 = O.lM

After eliminating [G1+] by dividing one equation by the other,

multiplying through by [H+]3 and rearranging, we have the follow-
ing cubic equation in [H +]:
2[H+r+5 x 1O- 3[H+]2-5 x 10- 17 =0
On the face of it this is a straightforward cubic equation and we might
expect to be able to solve it by Newton's method. However, the
enormous range of coefficients from 2 to 5 X 10- 17 makes it a very
different proposition from the equations we have considered earlier in
this chapter. If we define
f([H+]) = 2[H+]3 + 5 X 1O- 3[H+]2 - 5 X 10- 17
f'([H+]) = 6[H+]2 + 1O- 2[H+]
and choose [H +]0 = 1M as a (very poor) starting guess, we have
[H+]1 = 6.66 X 10- 1 M
 Approximate methods / 111
[H+]2 = 4.44 X 10- 1 M
[H+]3 = 2.96 X 10- 1 M

Progress is in the right direction but is painfully slow. In fact it is not

until the 30th approximation that we have even one significant figure
correct:
[H+]30 = 1.18 x 1O- 7 M
[H+]31 = 1.01 X 10- 7 M
[H+]32 = 1.00 X 10- 7 M
All subsequent approximations are very close to this last, showing
that the correct pH is 7.0.
We can see by inspection that the answer is equal to the mean of the
two higher pKa values, i.e. 7.0 = (4.3 + 9.7)/2, and we might well
enquire whether this is a coincidence, or whether we could have
arrived at it by a quicker and easier route. We can indeed: the same
enormous range of coefficients in the original equation that caused
Newton's method to perform so badly allows us to introduce
approximations that are virtually exact in this sort of problem even
though the corresponding ones would be quite improper in an
elementary example. To do this it is safest not to regard it as a purely
mathematical problem but to make use of our knowledge of
chemistry. Because of the 0.1 M concentration of Na + ions, we know
that the average charge on the glutamic acid species must be - 1, so it
is reasonable to guess that the predominant species will be G 1- and
that G 1+ will be present at negligible concentration. Ifwe ignore G 1+,
a derivation along exactly the same lines as above leads to the
equation
[H+]2 _10- 14 =0
which yields the solution [H +] = 10 - 7 M, or pH = 7.00,
immediately.
The contrast between these two ways of solving the same problem
could hardly be more striking. By applying a simple approximation at
the beginning we were able to obtain a solution that is so nearly exact
112 / Solving Equations
that it is scarcely an approximation at all. Virtually all pH problems
can be dealt with in this way with little difficulty. Even if several
ionizations are involved there are usually no more than three ionic
states that need to be considered, and these can be identified by
inspection. I shall illustrate this by reference to a more complex
example, the pH of a solution of glutathione partially titrated with
NaOH. Glutathione is a peptide (y-glutamylcysteinylglycine) with
four ionizable groups and hence four pKa values, 2.12, 3.53, 8.66 and
9.12 (corresponding to Ka values of 7.59 x 10- 3, 2.95 X 10- 4 , 2.19
x 10- 9 and 7.59 x 10- 1 °, respectively). Suppose we add 2.5 ml of 1M
NaOH to lOml of 0.1M glutathione: what is the pH of the resulting
mixture? The total glutathione concentration is 10 x 0.1/12.5
= 0.08M, so
[G 1 +] + [GO] + [G 1 - ] + [G 2-] + [G 3 - ] = 0.08M
where the various ionic states are represented by the same sort of
symbolism as used above for the ionic states of glutamic acid; and the
Na + concentration is 2.5/12.5 = 0.2M, so
- [G 1 +] + [G 1 - ] + 2[G 2-] + 3[G 3 - ] = 0.2M
If we continued by means of Newton's method we should obtain a
quartic equation and solving it would be even more laborious than the
cubic equation we had before. Instead, we shall proceed as follows: in
pure glutathione the predominant form must be GO in order to
achieve an average charge of zero; but 0.2M is 2.5 times 0.08M, i.e. the
NaOH added was sufficient to neutralize more than two groups on
GO. We should expect, therefore, that the predominant forms in the
mixture should be G 2- and G 3 -. So let us ignore G 1 + and GO
altogether and write the above pair of equations as
[G 1 - ] + [G 2-] + [G 3 - ] = 0.08M
[G 1 - ] + 2[G 2-] + 3[G 3 - ] = 0.2M
Moreover,
[G 2-] = [G 3 -][H+]/7.59 X 10- 10
[G 1 - ] = [G 2-][H+]/2.19 X 10- 9 = [G 3 -][H+]2/1.66 X 10- 18
Hence
3- ( [H+] [H+Y)
[G ] 1 + 7.59 x 10 10 + 1.66 x 10 18 = 0.08M
 Problems / 113
3- ( 2[H+] 3[H+]2)
[G ] 3+7.59xlO 10+1.66xlO 18 =0.2M

Dividing one expression by the other we can eliminate [G 3 - ] and

rearrange to the following quadratic equation for [H +]:

7.23 x 1016[H+]2 + 5.30 x 107[H+] - 0.04 = 0

which may be solved like any other quadratic to give [H +] = 4.63

x to- 10 M, or pH = 9.33.
This solution is in fact correct to three significant figures, but
suppose we had made an excessively simple assumption at the
beginning: for example, if we had ignored G l - as well as G l + and GO
we would have obtained [H+] = 7.59 x 10- IO M (pH = 9.12), a value
that is appreciably in error: how would we have recognized this?
Whenever one applies an approximation that one is unsure of it is
useful to check its consistency by substituting back into the original
expressions. In the glutathione example we can calculate the concen-
trations of all the species. With [H+] = 7.59 x 10- IO M we would
obtain
[G 3 - ] = O.04M
[G 2-] = O.04M
[G l - ] = 0.0139M
[GO] = 3.57 x 10- 8M
[G l +] = 3.57 x 1O- 15 M

Although the last two of these are indeed negligible, the value for G l -
certainly is not. This alone should indicate that it was wrong to ignore
it, but in addition we can calculate a total glutathione concentration of
0.0934M, which exceeds the correct value of 0.08M by a sufficient
margin to indicate that the initial approximation was not acceptable.

5.9 Problems
(5.1) Rearrange the following equations so that they express x in
terms of y:
(a) y = x + 3 (b) y = 2X2 + 5
(c) y = In5x (d) y = 8 exp (3x)
114 / Solving Equations
x+4
(e) Y = x-I (f) xy+4 = 3x

(g) Y = x 2 + 2x - 4 (h) _y_=_2_

x+l x-I
(i) _y_ _ _l_ = 1
3x+ 1 x+ 1
(5.2) A metabolite B is produced in one reaction at a constant rate vl
and is consumed in a second reaction at a rate V2 [B]/(K m2 + [B]).
Assuming that a steady state is achieved in which [B] does not change,
obtain an expression for [B]. How large would V l have to be for it to
become impossible to establish a steady state?
(5.3) Identify any of the following sets of simultaneous equations that
are singular and solve the remainder:
(a) 2x + y = -1
x+y=-2
(b) x + 6y = 3.5
12y+2x = 7
(c) 8.32x + 1.03y + 39.85 = 0
7.18y - 2.22x = 96.14
(d) 2.34x + 0.63y = 4.31
5.03x + l.35y = 9.27
(e) px +qy = r
Px+Qy = R
(f) na + SI:.x = I:.y
aI:.x + SI:.x 2 = I:.xy
a
(Note: in problem (5.3)f treat and S as the unknowns and n, I:.x,
I:.x 2 , I:.y and I:.xy as known quantities.)
(5.4) (a) Solve the pair of equations in problem (5.3)f after substitut-
ing Xl = 1, Yl = 1.27; x 2 = 2, Y2 = 2.47; X3 = 3, Y3 = 3.62; x 4 = 4,
Y4 = 5.08; n = 4. Interpret I:.x as the sum of Xi for i = 1 to n, and the
other summations similarly. (b) What would be the result of the
calculation if we put n = 1 and used only Xl and Yl?
(5.5) Evaluate the following determinants:

(b) 14.71
6.43
1.281;
5.11
(c) 1-1
3
-21;
-7
(d) I; ~I
 Problems j 115
(e) 0 0 0;
136
(f) I-2.22
2.31 1.1 8 1;
0.47
(g) 1.7 6.4 3.1
2.3 8.1 1.2
4 2 1.7 6.4 3.1

(5.6) Without carrying out a complete solution, calculate the dis-

criminant of each of the following equations and determine whether
there are (1) two unequal real roots, (la) two unequal rational roots,
(2) two equal roots, or (3) no real roots:

(a) x 2 + 3x + 5 = 0 (b) x 2 = 4x-7

(c) 2 + 3x = 4x 2 (d) x 2 = 2x + 12
2 3
(e) 3x(x + 4) = x +6 (f) --+--= 7
x+1 x+2
(5.7) The equilibrium constant for the reaction catalysed by gluco-
kinase has the following value at pH 6.5:

=-[g_lu_c_o--=se---:6_-_ph_o---=s=-=ph_a_te-=]~[_A_D_P-=-] = 230
[glucose] [ATP]

If 5mM-glucose is mixed with 4.5mM-ATP and allowed to equilib-

rate at pH 6.5 in the presence of glucokinase, what is the final
concentration of ADP?
(5.8) A mixture of two biochemicals Band C has an absorbance in a
1 cm cuvette of 0.63 at 460 nm and of 0.52 at 500 nm. The molar
absorbances, in M -I cm -I, of Band Care 1.03 x 10 3 and 4.57 x 103
respectively at 460 nm and 7.12 x 103 and 1.43 x 103 , respectively, at
500 nm. Given that the absorbances of mixtures are additive, and that
the absorbance of a solution with molar absorbance A, concen-
tration c and pathlength d is Acd, calculate the concentrations of
Band C.
(5.9) Use a graphical method to obtain three approximate solutions
to the the equation x 2 + 7x = 4j(3x + 2). Then use Newton's method
to refine these to values correct to three places of decimals.
(5.10) Estimate the pH of a O.IM solution of ammonium lactate,
assuming that the pKa oflactic acid is 3.86, the pKa of the ammonium
ion is 9.26, and pKw = 14.0.
(5.11) Estimate the pH of the solution obtained by mixing 3 ml O.lM
citric acid with 7 ml O.lM NaOH. For citric acid, pKI = 3.08,
pK2 = 4.74, pK3 = 5.40.
116 / Solving Equations
(5.12) For quadratic equations that do not have simple solutions it is
sometimes convenient to consider the sum and product of the two
roots instead, which are usually simpler. (a) Show this by deriving
expressions for the sum and product of the roots of the general
quadratic equation ax 2 + bx + c = 0; and hence (b) write down the
sum and product of the roots of the equation x 2 -7x + 5 = O.
6 Partial Differentiation

6.1 Meaning of a partial derivative

In science we often have to deal with two or more variables that have
no necessary dependence on one another. For example, although
Boyle's law defines the following relationship between the pressure p,
volume V and temperature T of a mole of a perfect gas:
pV=RT
(where R = 8.31 J mol- 1 K - 1 is the gas constant), it does not prevent
us from varying any two of the three variables independently. We can
have whatever pressure we like at whatever temperature we like,
provided we accept whatever volume results, etc. In this sort of
circumstance we cannot define an ordinary derivative, such as d VjdT,
because this has no definite value unless we specify how p is to change
with T. For example, if we decided that p was to be a constant
independent of T we would have

dV R
dT p

but if we set up the system so that p increased in proportion to T, we

would instead find

dV = 0
dT

and other expressions would be given by other relationships between

p and T. This is clearly unsatisfactory, but we can overcome the
difficulty by introducing the new concept of partial differentiation. In
this we define the partial derivative of one variable with respect to
another as the result of differentiating while treating all other
117
118 / Partial Differentiation
independent variables as constants. So, for Boyle's law we would
have:

R V
(~)p p T
V
(~;)T = - ~ = P
R p
(~)v V T

There are two points of symbolism to note here. First, as these are not
ordinary derivatives, we cannot use the ordinary symbols d VldT etc.
for them. Instead of the ordinary d we use a so-called 'curly d', written
as a, for partial differentiation. (This is not a Greek delta, incidentally,
which would be (j and is used in mathematics to represent a small but
finite increment, as in Chapter 3.) Second, we can indicate what
variables are being held constant during the partial differentiation by
showing them as subscripts, as in the expressions above. These are
often obvious, however, and may be omitted when no doubt is likely.
In thermodynamics they are frequently included as an aid to clarity.
In speech we call an expression such as aVI aT 'partial d V by partial
dT', although the second 'partial' is often omitted.

6.2 Exact and inexact differentials

In the previous section we saw that we cannot differentiate one
variable with respect to another if the relationship between them is
incompletely specified. The same applies to the reverse process,
integration, and it is convenient to illustrate it with the same example
of a perfect gas because of its great importance in thermodynamics.
Recalling that a pressure is aforce divided by an area (or length
squared), and work represents the product of a force and a length, we
can easily see that the work dW done (against the atmosphere, or a
piston, or whatever) by a mole of perfect gas at pressure p expanding
by an infinitesimal increment d V must be
dW= pdV
We might suppose, therefore, that we could calculate the work W
 Exact and inexact differentials / 119
done in a finite expansion from V = VI to V = V 2 as

However, just as in the previous section we were unable to evaluate

d VldT without knowing how V depended on T, here we cannot
evaluate pd V without knowing how p depends on V. Further, because
Boyle's law contains T as well as p and V, we must specify something
about T before we can properly express p in terms of V. In this
example pd V is an inexact differential and W can in fact have any
value.
To integrate pd V we must first define the system fully. Suppose we
decide to consider an isothermal expansion, which in mathematical
terms is the same as defining T as a constant. We then have

It is important to realize that we were only able to do this integration,

with T outside the integration sign, by specifying Tas a constant: ifT
varied during the expansion the integration would be invalid. Some
apparently similar functions turn out to have defined meanings
regardless of the way in which the particular change is carried out;
these are called exact differentials. For example, suppose we had set
out to integrate not pd V but (piT) d V. Although at first sight this
seems to be just as incomplete as pd V, in fact it is a function of V only,
because piT = RIV, i.e.
pdV RdV
T V
The factor 1/T, which converts the inexact differential pd V into the
exact differential (piT) d V, is called an integrating factor for the
inexact differential pd V.
Sometimes we find that although two differentials may be inexact, a
simple function of them, such as their sum or difference, may be exact.
For example, ifno relationship between u and v is specified both dulv
and udvlv 2 are inexact differentials that cannot be integrated. Their
difference, however, can be written as
du udv vdu - udv
-;--7 v2
120 / Partial Differentiation
which can immediately be recognized as d (u/v), an exact differential, if
we recall the expression for the derivative of a ratio (see p. 59).
This kind of relationship is of the greatest importance in
thermodynamics. It is possible to convert any system (not just a
sample of perfect gas) from one state to another by adding an
indefinite amount of heat SdQ and by causing it to do an indefinite
amount of work SdW. These amounts of heat and work are indefinite
because knowing the initial and final states does not allow us to
calculate them: their values depend on the path taken between the two
states. Nonetheless, it is a fact of observation that although dQ and
dW are inexact differentials, their difference dQ - dW is an exact
differential. What this means is that although we may observe all
kinds of values for the total heat SdQ and the total work SdW on
passing from state 1 to state 2, we always observe exactly the same
difference between them. This is clearly an important observation and
is an expression of the first law of thermodynamics. The observation
that dQ - dW is an exact differential provides the basis for defining
the energy V of a system, i.e. we define
dU = dQ-dW

6.3 Least-squares fitting of the Michaelis-Menten

equation
Although thermodynamics provides the main contexts in which
students of chemistry and biochemistry encounter partial
differentiation, it is a rather abstract subject that everyone finds
difficult. It is useful therefore to consider a quite different example
that has no obvious relationship to thermodynamics but has nonethe-
less a clear relevance to the practice of biochemistry.
Suppose we have a set of observations (Sj, v;), for i = 1,2,3 ... n,
that fit the Michaelis-Menten equation apart from experimental
error, i.e. suppose we can write

v. =
VS.I (1 + e.)I
I Km+s j
where V and Km are (unknown) constants and (1 + e;) is a factor
representing the effect of experimental error. (We could also have
included experimental error by adding an error term to the basic
expression rather than by multiplying by an error factor. As my
 Least-squares fitting of the Michaelis-Menten equation j 121
purpose in using this illustration is mathematical rather than
biochemical it is perhaps unnecessary to discuss why I think the factor
is preferable other than that it leads to much simpler mathematics.
However, it happens that it is also a more realistic way of representing
the kind of error that commonly occurs in enzyme kinetic
experiments.) In practice we would not know V and Km and would
therefore wish to estimate them from the observed values of Si and Vi;
we might prefer to measure them instead, but the unknown magni-
tudes of the error terms prevent this.
By rearranging the above equation we can readily express each e i as
follows:

= (Km/V)(vds;) + (vdV)-1
= ax i +bv i -l

where a = Km/V, b = IjV, Xi = VdSi. (These substitutions are made

only to make the mathematical derivation simpler and clearer; they
are not essential, inasmuch as we could arrive at exactly the same final
result if we carried out the calculation with Km and V as variables
rather than a and b.) We can define the 'best' values of a and b as those
that make some suitable function of the e i as small as possible. First,
however, we must choose a function. It is not sufficient simply to add
all the ei together, because some are positive and some are negative, so
their sum can be small or zero without the individual ei having to be
small. So we commonly take the sum of squares S, defined as

S= L" e; = L" (ax +bv -l)2.

i; 1 i; 1
i i

as a measure of closeness of fit.

Our aim is now to find values of a and b that make S a minimum. If
there were only one parameter this would be a simple problem in
differentiation of the sort we considered in Chapter 3. However, we
have to find a minimum in S not only with respect to a, but also
simultaneously with respect to b. As we can vary a and b
independently - there is no relationship that specifies b if we know a,
or vice versa - this is an exercise in partial differentiation, which we
must do with respect to a and b in turn. Differentiating with respect to
122 / Partial Differentiation
a we treat b as a constant:

Two points should be noted about these expressions: first, as all of the
summations are from i = 1 to n we can omit the limits from the
summation signs without risk of ambiguity; second, in the first and
second summations on the right-hand side a and b (respectively) are
factors of every term and so can be multiplied once after summing
rather than n times before. We can also partially differentiate with
respect to b in exactly the same way, treating a now as a constant:

as
ab = "[2v.
L.. I
(ax. + bv. -1)]
I I
= 2a~x.v. + 2b~V2 - 2~v.
I I I I

For any value of b, we can minimize with respect to a by finding a

value of a that makes the first expression zero; conversely, for any
value of a we can minimize with respect to b by finding a value of b that
makes the second expression zero. To minimize with respect to both
parameters simultaneously we must make both expressions zero
simultaneously. Let us define a and b, respectively, as the values of a
and b that satisfy this condition, i.e.

2a~x? +2b~XiVi -2~Xi = 0

2a~XiVi+2b~v?-2~Vi = 0

These are now a pair of simultaneous equations in a and b, exactly

analogous to those we considered in Chapter 5 apart from the fact
that we now have rather complicated coefficients such as ~x? instead
of the simple constants we had before. We can therefore use the
determinant method to write down the solutions without further
algebra:

A ~ v? ~Xi - ~XiVi ~Vi

a= . 2 2 2
~Xi ~Vi - (~XiVJ

b ~X?~Vi-~XiVi~Xi
- ~V? ~X? - (~XiVY

Finally we can revert to the original symbols to give a result with a

 Least-squares fitting of the Michaelis-Menten equation / 123
more obvious biochemical significance:
().
1\.. /v=
A LV~Lv·ls.1 - Lv 2/s. LV.
1 1 1 1 1

m LV?/S;LV;-(LV?/SY
IIV = LV?/S; LVj - LV?/SjLV)Sj
LV?/S; LV; - (LV?/SJ2
and it is a simple matter to rearrange these into expressions for Vand
Rm:
v= LV?/S; LV; - (LV?/SY
LV?/S; LVj - LV?/Sj LV)Sj
R LV; LV)Sj - LV?/SjLV j
m - LV?/S;LVj-LV?/SjLV)Sj
These are now the 'best' values of V and Km in the sense that they
make the function S that we defined as our criterion of closeness of fit
as small as possible. Before leaving this result it is advisable to check it
for algebraic errors by the dimensional considerations we saw in
Chapter 1. As Vis a rate, with dimensions of (concentration) (time)-I,
its expression must also be a rate: the first sum in the numerator
contains terms of the form v2/s2, with dimensions of (concentration)2
(time) - 2 (concentration) - 2, or (time)- 2; the second contains terms of
the form v2 , with dimensions of (concentration)2 (time)-2; thus the
product of the first two sums has dimensions of (concentration)2
(time)-4. In the same way the squared sum in the second half of the
numerator also has the dimensions of (concentration)2 (time)-4 and
can therefore be legitimately subtracted from the first product. By the
same sort of analysis we find that the denominator also contains a
dimensionally acceptable subtraction and has dimensions of
(concentration) (time)- 3. Finally, the expression as a whole has the
dimensions of the numerator divided by those of the denominator, i.e.
(concentration) (time) - 1, as it must if it is to be a rate. Thus our
expression for V is dimensionally correct (though this does not
guarantee that it does not contain other kinds of error), and in the
same way we can show that the expression for Rm is a concentration
and is also dimensionally correct.
In this discussion I have treated (concentration) as a fundamental
dimension: this is not strictly correct, as a concentration is a derived
quantity with dimensions of (amount of substance) (length)-3.
124 / Partial Differentiation
Properly therefore I ought to have written (amount of
substance) (length)- 3 wherever I wrote (concentration), but this
would not have made the discussion any easier to follow and indeed
might have encouraged the kinds of mistakes we were setting out to
avoid and recognize. For many purposes there is no harm and a
definite advantage in discussing dimensions in terms of composite
quantities such as concentrations and even rates.

6.4 Problems
(6.1) For the function z = (x 2 + y2)1/2;
(a) determine (:: )y;
(b) determine G: )z;
(c) determine (~~ )x;
(d) show that (::)y (~:)z (~~)x = -1;

a az a az
(e) show that ayax = ax ay"
(6.2) For the function pV = RT (where R is a constant), show that

av aT ap = -1 and that ~ av = ~ avo

aT ap av ap aT aT ap
(6.3) Consider a set of observations (Xi' yJ, for i = 1 to n, that can be
represented as values from the following straight-line relationship
with additive deviations ei :
Yi = a+bxi+ei
Defining the sum of squares as S = 'Lef, find the values of aand bsuch
that S is a minimum when a = a, b = b. [Note: For this problem, in
contrast to the one considered in the text, it is easier to deal with an
additive error than with a multiplicative error (1 + e;). A discussion of
more fundamental reasons than convenience for preferring one
treatment over another would be outside the scope of this book.]
 Problems / 125
(6.4) Show that your solution to problem (6.3) is dimensionally
consistent. (Note: Although no dimensions for Xi and Yi have been
specified, dimensional analysis can still be used, as all that is necessary
is to assume that the dimensions of Xi and Yi are different.)
Notes and Solutions to Problems

(1.1) (a) 28; (b) 5; (c) 41; (d) 180.

(1.4) (a) Incorrect: [I]jKj is dimensionless and therefore cannot be
added to Km or [S], which are concentrations; (b) consistent;
(c) incorrect: the slope must be a concentration not a reciprocal
concentration.
(1.5) (a) 14; (b) 8.857; (c) 2; (d) 17; (e) 72; (f) 512;
(g) 64.
(1.6) (a) 105000; (b) 126000; (c) 153000.Noticehowthesmall
proportion of tetramer has only a slight effect on the value of
M n' but a very large effect on the value of M z. This is because M z
weights the heaviest components of the mixture very heavily.
The progression M n Z M w Z M z applies not only to this
example but in general.
(1.7) (a) 23.9; (b) 0.6092; (c) 0.0624 (although each of the orig-
inal numbers had 5 or 6 significant figures, 3 were lost in
subtracting 35.6112 from 35.6579 to give 0.0467); (d) 9.2.
(1.8) (a) 8; (b) 4; (c) 3; (d) 4; (e) 4.
(2.1) (a) 27; (b) 0.0625; (c) 3; (d) 0.5; (e) 1; (f) 0.001;
(g) 1; (h) -1.63; (i) 24; (j) 27; (k) 2; (I) 3;
(m) 1000.
(2.2) (a) 1.16; (b) 0.923 (= 1 - 0.077). More accurate values are
1.1735 and 0.9259, which show that the approximate
formula gives results accurate to within 1.2 % and 0.31 %,
respectively.
(2.3) (a) In(l+x)~x; (b) 0.116(0.1098); (c) -0.017(-0.0171);
(d) 0.041 (0.0402); (e) -0.112 (-0.1188). In (b-e), appro-
ximate values given by the formula in (a) are followed in
parenthesis by values accurate to 4 places of decimals.
126
 Notes and Solutions to Problems / 127
(2.4) (a) 1.386 (= 2 x 0.693); (b) - 1.609; (c) 3.297 (= 3 x 1.099);
(d) - 0.510 (= 1.099 -1.609); (e) 2.302 (= 0.693 + 1.609);
(f) 0.5; (g) 3; (h) 15 (as 2.708 = 1.099 + 1.609 = In 3
+ In 5); (i) 1.4 ( = 7/5, as 0.337 = In 7 -In 5).
(2.5) About 1.000016:1 in favour of the state of lower energy
(calculated with T = 298 K, i.e. at 2SO C).
(2.6) Midpoint potential is + 0.1 00 V at pH 0, decreases linearly by
0.06 V per pH unit to about pH 3.3, then curves gently to a
smaller slope and decreases by 0.03 V per pH unit above about
pH 5.3.
(3.1) (a) 3x 2; (b) 0.5/x 1/2 ; (c) 5e; (d) 5e 5x ; (e) 3/x;
(f) 0.5 (x -1/2 - X - 3 2);
1 (g) X-I - 4x; (h) (x 2 + 2x)e\
(i) 3/x; (j) 0.5(X2 + 3)-1/2(2x) = X/(X2 + 3)1/2; (k) (x -1)
(1) - (x + 1)(1)/(x _1)2 = - 2/(x _1)2.
(3.2) (e) and (i), because 3 In x is identical with In (x 3 ).
(3.3) (a) Minimum, (3, -1.41); maximum, (1,0). (b) Minimum,
(4, 12); maximum, (6, 8): note that even though there is only one
minimum and only one maximum the maximum can occur at a
lower y value than the minimum. What happens in the vicinity
of x = 5?
(3.4) dv/ds = Km V/(Km + S)2, hence dv/d In s = Km Vs/(K m + S)2,
hence :sd~~s=(Km-S)KmV/(Km+S)3, which is zero if
s = Km. When s = Km the slope of a plot of v against In s is
S2 V/(s + S)2 = V/4. As In s = 2.303 log s (for any s), the maxi-
mum slope of a plot of v against log s is greater by a factor of
2.303, i.e. it is 0.576 V.
(3.5) dv/ds = Km V/(Km + sf. (a) V/ Km; (b) 0.25 V/ Km;
(c) approaches O.
(3.6) (a) h = (K1K2)1/2.
(3.7) Y = x 2/(1 + X2); Y' = 2x/(1 + X2)2; Y" = (2 - 6x 2)/(1 + X2)3.
Y = 0 at x = 0 and approaches Y = 1 at large x. Y' is positive at
all positive x and so Y increases monotonically with x. Y"
is positive at the origin but decreases as x increases, being zero at
x = 1 and negative when x > 1. Thus the curve is sigmoid (S-
shaped) with a point ofinftexion at x =0.577( = 3 -1/2),Y =0.25.
(3.8) 2 (for h = 2); in general it would be h.
(3.9) V = (dp/dt)[1 + Km/(so - p)], which may be rearranged to
dp/dt = Vs/(K m + s), i.e. the Michaelis-Menten equation, if we
define s = So - p.
128 / Notes and Solutions to Problems
(4.1) (a) t x3+ 1X 2+X+ex; (b) tx2+1nx+ex; (c) tln(2+3x)
+ ex; (d) -texp(-3t)+ex.
(4.2) (a) 37.5; (b) 0.693; (c) 2; (d) O. In (d) the implication of
zero area under the curve between x = - 2 and x = 2 may be
puzzling. The explanation is that areas below the x-axis count as
negative and in this example the area of - 4 between x = - 2
and x = 0 exactly cancels with the area of + 4 between x = 0
and x = 2. .
3 1
(4.3) Y = - - - - - , hence Jy dx = In(3x + 1) -In (x +2)+ex.
3x+ 1 x+2
1
(4.4) D2 [B(C + Dx) + (AD - BC) In (C + Dx)] + ex.
(4.5) This problem is the inverse of problem (3.9). Replacing s with
So - P and separating variables we have

(Km+So-P)d P = fVdt.
f
So.-P
The left-hand side can readily be integrated by substituting
A = Km + so' B = - 1,x = p, C = so, D = - 1 into the solution
to problem (4.4). It is, however, much simpler to note that So - P
appears in both numerator and denominator, so that we can
write

f:~d: + fd P = f Vdt

hence - Km In (so - p) + P = Vt + ex. The constant ex is readily

evaluated by putting P = 0 when t = 0, so - Km In So = ex, and
finally Vt = P + Km In [so/(so - p)].
(4.6) In this case the short cut used in problem (4.5) is not available
and we must use the result of problem (4.4), with the same
substitutions as above except for D = (Km/Kp -1) instead of
D = -1. The solution is Vt = (1- Km/Kp)p +
Km{1 +sjKp) In [sj(so- p)].
(4.7) 1 = 10 exp ( - kex), or In (lo/l) = kex. As In a = 2.303 loga for
any a, it follows that kex = 2.303Aex, i.e. A = k/2.303.
(4.8) After setting the sum of the two expressions to zero and
cancelling AD, we have
uix(1- Vp)MrC _ de = 0
RT dx
 Notes and Solutions to Problems j 129
Separating the two variables c and x and integrating, we have
w 2x 2(1- Vp)M
In c = 2R T r + constant
Thus a plot of In c against x 2 should be a straight line with a
slope consisting of known quantities apart from Mr.
(4.9) With four strips, (a) 33.37, (b) 18.51. With eight strips,
(a) 33.99, (b) 18.59. Notice that even with four strips one can
obtain results almost as accurate as those with eight. The results
would be less concordant if the curve were more complicated.
(5.1) (a) x = y-3; (b) x = ±[(y-5)/2]1/2; (c) x = 0.2exp(y);
(d) x = -t In (0.125 y); (e) x = (y + 4)/(y -1); (f) x =
4/(3 - y); (g) x = -1 ± (5 + y)1/2; (h) x = (y + 2)/(y - 2);
(i) x = [y -7 ± (25 - 2y + y2)1/2]j6.
(5.2) [B] = Km2vtI(V2 - Vi). This expression gives a negative value of
[B] if Vi exceeds V 2 , and shows that a steady state cannot be
achieved if the second reaction cannot remove B faster than it is
being produced in the first reaction.
(5.3) (a) x = 1, y = - 3; (b) singular; (c) x = - 6.21; y = 11.47;
(d) singular*; (e) x = (Qr - qR)/(pQ - Pq); y = (pR
- Pr)/(pQ - Pq); (f) a = (I:x 2I:y - I:xI:xy)/[ nI:x 2 - (I:X)2;
b = (nI:xy - I:xI:y)/[ nI:x 2 - (I:X)2]. *Ifthe coefficients in prob-
lem (5.3)d are treated as exact, the equations are not strictly
singular and have the solution x = 2.18, y = -1.26. This
solution is very unstable, however, and would, for example,
become x = 2.53, y = - 2.57 if the coefficient of y in the first
equation were 0.629 instead of 0.63. It is more reasonable to
suppose that the coefficients quoted are accurate only to two
decimal places and that the equations are singular. Equations
that are nearly singular are called ill-conditioned.
(5.4) (a) a = - 0.035, b = 1.258; (b) the equations would become
singular: this illustrates the general point that one cannot
determine two constants from one observation.
(5.5) (a) - 2; (b) 15.84; (c) 13; (d) 0; (e) 0; (f) 3.705;
(g) o.
(5.6) (a) - 11: no real roots; (b) - 12: no real roots; (c) 41: two
real roots; (d) - 44: no real roots; (e) 193: two real roots;
(f) 60: two real roots. Rational roots-whether equal or
not - occur so rarely as the solutions to scientific problems that
it is hardly worthwhile considering them as a possibility.
130 / Notes and Solutions to Problems
Attempting to solve quadratic equations by factorization is
usually therefore a waste of time.
(5.7) 4.37mM. The second solution of 5.17mM is mathematically
correct but physically meaningless because it implies negative
concentrations of glucose and ATP. In this sort of problem
there is no advantage in converting concentrations from mM to
M; it leads to inconveniently small numbers and greatly
increases the likelihood of arithmetical mistakes. In general it is
best to work in units that produce numerical values as close to
1.0 as possible.
(5.8) As in problem (5.7) it is advisable to work in mM rather than M.
If aIM solution of B has absorbance 1.03 x 10 3 at 460 nm, a
1 mM solution has absorbance 1.03, and the other values
convert similarly. Thus we have 0.63 = 1.03[B] + 4.57[C];
0.52 = 7.12[B] + 1.43[C], with the solution [B] = 0.0475 mM,
[C] = 0.127 mM.
(5.9) x = 0.211, -0.908 or -6.970.
(5.10) 6.56
(5.11) There is sufficient alkali to neutralize the first two acidic
groups, so that the significant species are HCit 2 - and Cit 3 -.
Proceeding as for the glutathione example in the text gives
pH = 5.10.
(5.12) (a) sum = - bfa, product = cia; (b) sum = 7, product = 5
(6.1) (a) X/(X2 + y2)1/2 = x/z; (b) - Y/(Z2 - y2)1/2 = - y/x;
(c) Z/(Z2 - X 2 )l /2 = z/y; (e) both derivatives have the value
_ xy(x2 + y2) - 3/2 = _ xY/Z3.

(6.2) OV = R/p = V/T, ~T = VIR = T/p, ~p = - RT/V 2 = - p/V,

aT up uV
hence the product of the three =- 1;
a av a oV . .
- ;;- = ;;- -;-- = - R/p2 = - V/(pT). These relatIOnshIps,
ap uT uT up
like the similar ones proved in problem (6.1), are not specific to
these examples but apply in general.
as
(6.3) S = L(Yi - a - bXJ2; hence oa = - 2LYi + 2an + 2bLX i,

~! = - 2LX iYi + 2aLx i + 2bLX; (note that L1 can be written as

n, because it means the sum of n 1's); setting these to zero for a
= (1, b = b, we have the pair of equations considered previously
 Notes and Solutions to Problems / 131
in problem (5.3)f, which have the solution a = (r.Xfr.Yi
- r.Xir.XiyJ/[ nr.xf - (r.xJ 2], b = (nr.xiYi - r.xir.YJ/[ nr.xf
- (r.XJ2.
(6.4) In the expression for a, the first sum in the numerator has the
dimensions of xf, the second has the dimensions of Yi' so their
product has the dimensions of xf Yi' The second sum also has the
dimensions of xf Yi and can therefore be properly subtracted
from the first, to give a complete numerator with the dimensions
of XfYi' In the same way we can show that the denominator also
contains a proper subtraction and that its dimensions are those
of xf (in reaching this conclusion we should note that n is
dimensionless). Thus the whole expression for a has the
dimensions of xfYJxf = Yi' which is correct for the intercept on
the Yi axis. In the same way we can show that the expression for b
contains no improper subtractions and that it yield dimensions
of yJx i , which is correct for the slope of a plot of Yi against Xi'
Index

Briggs, H., 29
Absorbance, molar, 89, 115
Buffer capacity, 66-68
Accuracy (contrasted with
precision), 21 Calculators, disregard of priority
Acetaldehydelethanol couple, rules, 5
40-42 Calculus
Acetic acidlacetaldehyde couple, differential, 48-71, 117-125
43-45 integral, 72-89
Adenylate kinase equilibrium, Chain rule, 58
101-102 Characteristic, 29
Algebra, scope of, 1-2 Common logarithms, 28-30
'Algebraic notation' (as used in Competitive inhibition, 103-105
electronic calculators), 5 Complex number, 101
Amount-of-substance Computer languages, 3, 5
concentration, 9-10 Concentration, 9-10
Analytic geometry, see co-ordinate Constant, 8-9
geometry, 48-50 Constant of integration, 74, 77, 83
Antilogarithms, 28 Co-ordinate geometry, 48-50
Approximate methods, 109-113 Cubic equation, 110
Area under a curve, relationship to Curvature, measured by second
integration, 72-74, 85-88, 90 derivative, 59
Arithmetic, scope of, 1-2 Curve
Average, different kinds of, 22-23 area under, 72-74, 85-88, 90
Avogadro's number, 32 slope of, 50
Curve sketching, 68-70
Bacterial growth and logarithms to
Cytochrome c couple, 41-42
base 2, 33-34
Bar notation, 29-30 Decadic logarithms, see common
Bell-shaped curve, 57-58 logarithms, 28-30
Boltzmann principle, 32, 47 Decay, exponential, 34-35
Boyle's law, 117-120 Decimal places, 21
Brackets, 4-5 Decimal point, 3

133
134 / Basic Mathematics for Biochemists

Definite integral, 74-77 93, 103-107

Derivative, 52 Equations
function of a function, 57-58 approximate methods, 109-113
higher, 59 graphical solution, 102-107, 115
list of common cases, 53 linear, 91-92
Michaelis function, 57-58 Newton's method, 107-112, 115
notation, 59-61 physically impossible solutions,
partial, 117-118 101-102
ratio, 58-59 quadratic, 99-102, 116
reverse of integral, 74 rearranging, 92-93
second,59 simultaneous, 93-96, 103-107
sums and products, 54-57 singular, 92, 98, 114
Determinants, 96-99, 114-115 Equilibrium constant, 8, 36-37
Differential calculus, 48-71, Equilibrium sedimentation, 22-23,
117-125 90
Differential equations, 82-85 Error, experimental, 9, 17, 120-124
Differential, exact and inexact, Exact differential, 118-120
118-120 Exclamation point (!), meaning in
Differentiation mathematics, 31
from first principles, 51-52 Experimental error, 9, 17, 129-124
function of a function, 57-58 Exponential decay, 34-35
meaning, 52 Exponential function, 32
Michaelis function, 57-58 integral of, 78
partial, 117-125 Exponential growth phase, 33
rapid, 52-52 Exponentiation, 4
sums and products, 54-57 Exponents, 24-47
Diffusion, 90 addition and subtraction, 37
Dimensional analysis, 9-13, 22, dimensionless nature, 10
123-124 fractional, 26-27
logarithms and, 37-38
Direct linear plot, 106-107 Factorial, 31
Discriminant, 100-10 1, 115 Fe 3 +IFe 2 + couple, 39, 41
Dixon plot, 103-105 First law of thermodynamics, 120
Dot notation for derivatives, 60 First-order kinetics, 32, 34, 75-77
Fortran, 3, 5
e, 31 Fractional exponents, 26-27
'Efficiency' of metabolic processes, Function, 6-8
38 Function of a function, derivative
Electrochemistry, 39-45 of, 57-58
Electronic calculators, disregard of
priority rules, 5 Gas equation, 117-120
Enthalpy change, 37 Generation time, 33-34
Entropy change, 37 Geometry, co-ordinate, 48-50
Enzyme kinetics, 2-3, 8, 61-62, 65, Gibbs energy, 36, 37
 Index / 135

comparisons involving, 38 Irrational number, 31

redox potential and, 42 Isothermal expansion, 119
Glucokinase equilibrium, 115
Kelvin, Lord, I
Glutamic acid, ionization, 109-113
Kinetics
Glutathione, ionization, 112-113
enzyme, 2-3, 8, 61-62, 65,
Gradient, 50 (see also slope)
84-85, 93, 103-107
Graphical solution of equations,
first-order, 32, 34, 75-77
102-107,115
steady-state, 84-85
Graphs
faults of, 13-18 Lack of fit, 17
slopes and intercepts, 11-12 Law, scientific, 2
Growth, bacterial, 33-34 Least-squares fitting, 120-124
Leibniz, G. W. von, 59-61
Half life, 34-35 Light absorption, 89, 115
Half-reaction, 39 Light scattering, 22-23
Henderson-Hasselbalch equation, Limit, 6, 52, 65, 74
10-11, 66-68 Linear equations, 91-92
Hill equation, 71 (problem Logarithmic growth phase, 33
(3.7» Logarithms, 7, 24-47
Hydrogen electrode, 40 base e, 31-32
Hydrogen-ion concentration, 8, base 10, 28-30
35-36 base 2, 28, 32-35
Briggsian, 29
Identity sign (==), 79 common, 28-30
Imaginary number, 101 definition, 28
Indefinite integral, 74-77 dimensional aspects, 10, 37-38
Inexact differential, 118-120 integration and, 80-82
Infinite series, 3 I Napieran, see natural
Infinitesimal quantity, 52 logarithms, 31-32
Inflexion, 65-68, 70 natural, 31-32
spurious, 15 negative numbers, 30-31
Integral calculus, 72-89 scaling by means of, 35-36
Integral
definite and indefinite, 74-77 Mantissa, 29
list of common cases, 78 Matrix algebra, 6
numerical, 85-86 Maximum, 61-65
partial fractions, 79 Maximum velocity, 65
reverse of derivative, 74 Metabolic processes, 36-38, 93
tables, 85 Michaelis constant, dimensions, 13
Integrating factor, 119 Michaelis function, 8, 71
Intercept, 11-12, 49 differentiation, 57-58
Ionization maximum, 63
calculations of, 109-113 Michaelis-Menten equation, 8, 65,
effect on redox potential, 42-45 70-71,89,93, 106-107
136 / Basic Mathematics for Biochemists

dimensional aspects, 13 Plots, faults of, 13-18

fitting by least squares, 120-124 slopes and intercepts, 11-12
integration, 84--85 solving equations by means of,
Mid-point potential, 41-42, 47 102-107
Minimum, 61-65 straight line, 48-50
Mistakes, detection of, 9-13 Plus-or-minus sign (±), 21, 100
Molar absorbance, 89, 115 Point of inflexion, 65-68, 70
Molecular mass, determination, spurious, 15
22-23 Powers, 24--47 (see also exponents)
Monod, J., 34--35 fractional, 26-27
Multiplication sign, 3 integer, 24--26
in Fortran, 5 Precision, 18-21, 23
Myokinase, see adenylate kinase, Pre-steady-state kinetics, 84--85
101-102 Priority rules for operators, 3-5
Product, derivative of, 54--57
NAD+INADH couple, 42
Product inhibition, 89
Napieran logarithms, see natural
Product sign (II), 6
logarithms, 31-32
Proton magnetic resonance, 47
Natural logarithms, 31-32
Pure numbers, 10
Negative numbers, logarithms,
Pyruvic aCid/lactic acid couple,
30-31
45-46
Nernst equation, 22, 39-45
'Nesting' of brackets, 4 Quadratic equations, 99-102, 116
Newton, Sir Isaac, 60
Radioactive decay, 34--35
method of solving equations,
Random scatter, 17
107-112, 115
Rate constant, dimensions, 10
notation for derivatives, 60
Ratio, derivative of, 58-59
Number average, 22
Rational number, 100
Numerical methods, 84, 85-88
Rearranging equations, 92-93,
Operators, priority rules, 3-5 113-114
Origin, 49 Reciprocal function, integral of,
Osmotic pressure, 22-23 78,80-82
Oxidation-reduction process, 22, Redox couple, 22
39-45 Redox potential, 36, 39-45
dependence on pH, 42-45, 47
Parabola, 50-52
Gibbs energy and, 42
Parameter, 8-9
mid-point, 41-42
Partial differentiation, 117-125
Relative molecular mass,
Partial fractions, 79, 88
determination, 22-23
Perfect gas, 117-120
rH,36
pH, II, 35-36
Rounding, 19-20
dimensions and, 38
effect on redox potential, 42-45, Sample variance, 21
47 Scale, choice for plotting, 13-16
 Index / 137

Scaling, by means of logarithms, Trapezium method, 85-86

35-36 Trigonometry, 8, 48
Scatter, random, 17 Truncation, 19-20
Sedimentation equilibrium, 22-23,
Ultracentrifuge, 22-23, 90
90
Units, 9-10
Separation of variables, 83
Significant figures, 18-21, 23 van't Hoff equation, 8
Significant zeroes, 20 Variables, 8-9
Simpson's rule, 86-88, 90 Variance, 21
Simultaneous equations, 93-96, Vector algebra, 3
103-107
Weight average, 22-23
singular, 114
Single-turnover condition, 84-85 Z-average, 22-23
Singular equations, 92, 98, 114
Sketching curves, 68-70
Slash (/), care in use of, 4-5
Slope, 11-12, 50-52
Solidus (/), see slash, 4-5
Solving equations, 91-116
Square, 7
Square root, 7
Standard hydrogen electrode, 40
Standard redox potential, 39
Standard state, 10, 12, 38
Stationary point, 63
Statistical methods, 20-21
Steady-state kinetics, 84-85
Straight line, plot, 48-50
fitting by least squares, 124
Substrate inhibition, 61-62
Sum
derivative, 54-57
integral, 78
Sum of squares, 120-124
Summation sign (1:), 6, 21, 22-23
Symbols, 3
Systematic deviations, 17
Tables, use for integrals and
differential equations, 85
Tangent to curve, 52, 107-109
Thermal equilibrium, 47
Thermodynamics, exact and
inexact differentials, 118-120