UC-NRLF

$B

m?

371

DE MORGAN.

CONNEXION OF NUMBER

MAGNITUDE.
Price 4s.

PAULINE FORE MOFFITT LIBRARY

UNIVERSITY OF CALIFORNIA GENERAL LIBRARY, BERKELEY

Oy

^??

THE

CONNEXION

NUMBER AND MAGNITUDE

AN ATTEMPT TO EXl^LAIN

THE FIFTH BOOK OF EUCLID.

BY

AUGUSTUS DE MORGAN,
OF TRINITY COLLEGE, CAMBRIDGE.

La

seule manifere de bien trailer les 616mens d'une science exacte et rigoureuse, c'est d'y mettre toute la rigueur et I'exactitude possible. D'Albmbbrt.

LONDON:
PRINTED FOR TAYLOR AND WALTON,
BOOKSELLERS AND PUBLISHERS TO THE UNIVERSITY OF LONDON,
30

UPPER GOWER STREET.

M.DCCC. XXXVI.

LONDON:
J.

MOYES4 CASTLE STKBET, LEICESTER SQUARE.

GIFT

PREFACE.

This

Treatise

is

intended ultimately to form part of one

on Trigonometry.
sider

The

place in which most students confor the first time,

is

number and magnitude together

in the elements of the latter science, unless they have un-

derstood the Fifth

case.

Book of Euclid

better than

is

usually the

Previously, therefore, to
it

commencing Trigonometry,
upon the consideration of
is,

I consider

advisable to enter
its

proportion in
of Euclid.

strict
is

form

that

upon the Fifth Book

I

There

no other method with which

am

acquainted which gives any thing like demonstration of the

general properties of ratios, though there
is

a dottx oreiller

pour reposer une

tete

Men

faite,

which many of the conshall

tinental mathematicians
stration,

have agreed

be called demonits

and which

is

beginning to make

way
for

in

this

country.
Hitherto, however,
matical students
it

has been

customary

mathe-

among

us to read the Fifth

it.

Book of Euclid;
in

frequently without understanding

The form

which

it

appears in Simson's edition

is

certainly unnecessarily long,

and the tedious

" repetition of
is

AB
in

is

the same multiple of the length

follow.

CD

which

EF

of

G H,"

all

of words,

renders the reasoning not easy to

The use of

general symbols of concrete magnitude, instead of the straight

line of Euclid,

and of a general algebraical symbol

for

whole

307

IV

PREFACE.
to

number, seems

me

to
it

remove a great part of the

difficulty.

Throughout

this

work

must be understood, that a

;

capital

letter denotes

a magnitude
itself:

not a numerical representation,

but the magnitude

while a small letter denotes a

number, and mostly a whole number.

arithmetical proportion^

And by
is

the term

when

it

occurs,

signified,

not the

common and now

useless

meaning of the words, but the

proportion of two magnitudes which are arithmetically related, or

which are cohimensurable.

subject
is

The
the

one of some real

difficulty, arising

from

limited character
as

of the symbols of arithmetic, conof

ratios,

sidered

representatives

and the consequent

that
is,

introduction

of incommensurable

ratios,

of ratios

which

have

no arithmetical representation.
is

The whole
:

number of students
do not
duction
;

divided into two classes

those

who

feel

satisfied

without rigorous definition

and de-

and those who would rather miss both that take a

long road, while a shorter one can be cut at no greater expense, than that of declaring that there shall be propositions

which arithmetical demonstration declares there are

This work
is

not.

intended for the former

class.

AUGUSTUS DE MORGAN.

London f

May

1,

1836.

CONNEXION
01'

NUMBER AND MAGNITUDE.

ERRATA.
Page
35, line 3, for less read greater.
47,
... ...

l,/orv(Q + Z)r<?arfM>(Q +
2,

for V Q read

Z).

tv

Q.

measureilieiit oi umugies^,

wmcu,

lut:
it

wiucsi
will

seiiacj

iii\,uv.uo M.

the applications of algebra to geometry,

be

right to inquire

on

what

sort of demonstration

we

are to pass from an arithmetical to a

geometrical proposition, or vice versa.

Geometry cannot proceed very

connexion was
difficult
first

far

without arithmetic, and the

made by Euclid
it

in his Fifth

Book, which

is

so

a speculation, that
read
it

is

either omitted, or not understood

by

those

who

for the first time.

And

yet this same book, and the

logic of Aristotle, are the

treatises

two most unobjectionable and unassailable

found in the Fifth Book
is

which ever were written.

reason of the difficulty which
Firstly;
It is all
is

The
twofold.

reasoning, unhelped by the senses: most

is

of the propositions have no portion of that intrinsic evidence which seen in

the

" two

sides of a triangle are greater than the third ;" but, at

same

time, the propositions of arithmetic which correspond to

IV

PREFACE.
to

number, seems

me

to
it

remove a great part of the

difficulty.

Throughout

this

work

must be understood, that a

;

capital

letter denotes

a magnitude
itself:

not a numerical representation,

while a small letter denotes a

but the magnitude

number, and mostly a whole number.

arithmetical proportion,

And by
is

the term

when

it

occurs,

signified, not the

common and now

useless

meaning of the words, but the

proportion of two magnitudes which are arithmetically related, or

which are commensurable.

subject
is

The

one of some real

difficulty, arising

from
coit-.

the limited character

of the symbols

of arithmetic,

which arithmetical demonstration declares there are

This work
is

not.

intended for the former

class.

AUGUSTUS DE MORGAN.

London f

May

1,

1836.

CONNEXION
OF

NUMBER AND MAGNITUDE.

When a student has acquired a moderate knowledge of the operations and principles of algebra, with as many theorems of geometry as are contained in the first four books of Euclid's Elements, it becomes
most desirable
that he should gain

some more

exact knowledge of the

connexion between the ideas which are the foundation of one and
the other science, than would present itself either to an inattentive
reader, or to one

whose whole

attention is engrossed

by

the difficulty

of comprehending terms which cannot yet have become familiar to

him.

measurement of

Before proceeding, therefore, to explain Trigonometry (the triangles), which, in the widest sense, includes all
it

the applications of algebra to geometry,

will be right to inquire

on

what

sort of demonstration

we

are to pass from an arithmetical to a

geometrical proposition, or vice versa.

Geometry cannot proceed very

connexion was
difficult
first

far

without arithmetic, and the

made by Euclid
it

in his Fifth

Book, which

is

so

a speculation, that
read
it

is

either omitted, or not understood

by

those

who

for the first time.

And

yet this same book, and the

logic of Aristotle, are the two most unobjectionable and unassailable

treatises

which ever were written.

reason of the difficulty which
Firstly;
It is all
is

The
twofold.

found in the Fifth Book

is

reasoning, unhelped by the senses: most

of the propositions have no portion of that intrinsic evidence which is " seen in " two sides of a triangle are greater than the third ; but, at
the

same time, the propositions of arithmetic which correspond

B

to

2
those of the Fifth

CONNEXION OF
Book
are very evident,

and the student and

fly to that

is

therefore

led to escape from the notion of magnitude,

of number.

Secondly;

The non-existence

of any very easy notation and system of

arithmetic in the time of Euclid,

latively so

made geometrical

considerations re-

much more
The

simple, that the form of his

all

book

is (to

us)

unnecessarily remote from

likeness to a treatise connected with

his lies in
tliis
:

numbers.
in the

difference

between our day and

that

former the exactness of geometry was gained with some degree

;

of prolixity and (to a beginner) obscurity

in the latter, the facility

of arithmetic
sacrificed to

is

preferred,
1 shall

it.

and perfect demonstration is more or less now endeavour to present the Fifth Book of
tlie

Euclid in a form wliich will be more easy than

original, to those

who have some acquaintance with By number is here meant what

algebra.
is

called abstract

number, which

merely conveys the notion of times or repetitions, considered indeBy magnitudey or pendently of the things counted or repeated.
quantity,
is

meant a thing presented

to us, not as to its form, if

it

have form, or as

to colour, weight, or

any other circumstance, but

whole simply as that which is made up of parts, not differing from the in any thing but in being less ; so tliat, if we consider separately a
part

and the whole, we have only two inferences

The

part

is

less than the

is

whole.

The whole
Every thing
our words from

greater than the part.

feel presents to us the notion

we

can see or

of magto pick

nitude or quantity.

And

here

we must

observe, that
use,

we have

among
to

those in

common

which never have very

precise meanings.

English word

which

For instance, we have magnitude, the nearest is greatness ; and quantity, for which the
These words are of
leave
the

word,

if it

existed, should be so-much-ness.

same meaning, and

the

more

indefinite

we now

them (except

considered as applied to any only in assigning that they are to be 7i)ore or less), the better for our purpose ; thing which can be made
since
it

is

the object of this treatise to deduce from that indefinite

method of making mathematical comparisons of quantities, aid of the notion of number. by Upon two magnitudes, our senses will enable us to draw one or
notion a
other of the following conclusions
1.
2.
:

The

first is
first is

sensibly greater than the second.

sensibly less than the second.

The

NUMBER AND MAGNITUDE.

3.

6
that the difit.

The
if

first is
is

sensibly equal to the second

meanmg

ference,
is

any,

so small that our senses cannot perceive

This

is

meant by equality of magnitudes in common life. The English what foot and the Florence foot are equal for common purposes they
:

differ

by about the twentieth part of an inch, which in a foot

a mathematical conception,
;

is

called

nothing.
Perfect equality
is

which never can be

absolutely verified in practice

ceive a certain quantity, be
possible that

for so

long as the senses cannot per-

much

as

it ever so small, so long it must always be two quantities, which appear equal, may differ by as But we are not reasoning upon the imperceptible quantity.

what we can carry into effect, but upon the conceptions of our own minds, which are the exact limits we are led to imagine by the rough
processes of our hands.

The
results
:

following, then,

is

the postulate

upon

which we construct our

An]/ one magnitude being given,

let it be
is

granted that any number

of others
Let
units

may

be found, each

of which

{positively and mathematically)

equal to the first.

represent
it

a magnitude

not as in algebra, the number of

so that if
is
it

which

contains, but the magnitude itself

be, for

instance, weight of which

we

are speaking,

not a

number of

the

pounds, but the weight itself. Let B represent another magnitude of same kind ; we can then make a third magnitude, either by putting
the two magnitudes together, or

magnitude

equal to the less.

by taking away from the greater a Let these be represented by A -j- B and

B, being supposed the greater. We can also construct other magnitudes, by taking a number of magnitudes each equal to A, and Thus we have putting any number of them together.

A+A A+A+A
A-f- A-jand so on.
and
all

which abbreviate into

A-f-A
have thus a
set of

3A 4A
magnitudes, depending upon A,

We

known when

A is

known ; namely,

2A

3A

4A

5A

&c.

which we can carry as far as we please. These (except the first) are distinguished from all other magnitudes by the name oi multiples of A
;

and

it is

evident that they increase continually.

Let the preceding be

4
called the scale

CONNEXION OF
ofmultiples of A. It is clear that the multiples of mul; thus, 7 times 3 A is 21 A, tn times wA is (w?n)A,

tiples are multiples

where

wn

is

the arithmetical product of the whole

numbers

and

n.

The

following propositions
I.

may

then be proved.

Prop.
is

If

A be

made up

of the

made up of B and C, then any multiple of A same multiples of B and C ; for 2 A must be made

up of
of which

B C
B
and

B C
C make
2

B make

2 B,

and

so that 2

is

made

up

of

2B
3B

and 2C.

Similarly, 3

is

made up of

B C B C B C
or of

and 3C.

Corollary.

Hence
less

it

follows, that if

A be

less

multiple of
tiple

A is

than the same multiple of

of C.
;

For, since

up B
2 C. If

therefore, 2

A is less than B by C, A A and 2 C make up 2 B, or 2 A is

by C, any same mulby and C together make

than

the

less than

2B

by
:

The

algebraical representations of these theorems are as follows

A=B +C

If

A=B-C
II.

mA = mB + mC ?wA = mB-wC
3, 4,

m being any of the numbers 2,

Prop.
be, the multiples in the scale

&c
be, or however great

However small

may

B may

A,
will
far

2 A,

3 A,

4 A,

5 A,

&c.

by continuing the scale sufficiently magnitudes of the same kind. This is a proit must be reposition which must be considered as self-evident membered that B remains the same, while we pass from one multiple
:

come

in time to exceed B,

and

A being

of

to the next.

Put

feet together

and we

shall

come

in

time to

exceed any number of miles, say a thousand. But the best illustration of the reason why we formally put forward so self-evident a proposition, will

be to remark, that

it

is

not every way of adding mag-

nitude to magnitude without end, which will enable us to surpass

any given magnitude.

To

a magnitude

add

its

half; to that

sum add

half of the half; to which add the half of the last: and so on.

No

continuation of this process, were

times, could ever double the
first

it

performed a hundred million of

magnitude.

NUMBER AND MAGNITUDE.

is greater Prop. III. If be greater than B, any multiple of than the same multiple of B. This follows from Prop. I. And if be less than B, any multiple of is less than the same multiple of B.

This follows from the corollary, Prop.

I.

And

if

be equal
This
is

to B,
self-

any multiple of
evident.

is

equal to the

same multiple of B.

If any multiple of A be greater than (equal to, or less same multiple of B, then A is greater than (equal to, or less than) B. For example, let 4 A be greater than 4 B ; then A must be greater than B; for, if not, 4 A would be equal to, or less

Prop. IV.

than) the

than, 4

(Prop. III.).
If from a magnitude the greater part be taken
itself

Prop. V.
and
if
:

away;

from the remainder the greater part of

the given magnitude
as

be taken away, and

so on

may

thus be

made

as small as

we

please,

meaning
Let

small as, or smaller than, any

second magnitude

we

choose to name.

and Z be the two magnitudes, and

its

let

A
Let

diminished by

more than

half be B, then 2

by more than half be C ; then and still more less than A. Let
be D, then 2

B 2C

is

less

than A.
than B,

diminished
than 2 B,
its

is less

4C

is less

diminished by more than

is

half

is

less

than C, 8

less

than 4 C, and

still

more

than A. This process must end by bringing one of the quantities A, B, C, D, &c. below Z in magnitude. For, if not, let A, B, C, &c. always remain greater than Z. Then, since 2B, 4C, 8 D, 16E, &c.
are all less than

A (just
But

proved)
this

still

more must 2Z, 4Z, 8Z, 16Z,

;

&c. be

less

than A.

cannot be

therefore,

one of the

set

A,

B, C, &c. must be

less than Z.

called, is a difficult

[The reductio ad absurdum, as this sort of argument is usually form of a simple inference. Suppose it proved

that

whenever

is

Q, then
It is

is

Y.

It follows that

whenever
if

is

not y,

is

not Q.

usually held enough to say, for

in

were

which Euclid argues, supposes an " When opponent ; and the whole argument then stands as follows. X is Y, you grant that P is Q ; but you grant that P is not Q. I say

would be Y.

But the form

that

X
P

is

not Y.

If
it

you deny

this

you must

affirm that

which you admit

that
is

not

be a consequence that P is Q. therefore, you say at one time that

to

X is Y, of But you grant P is Q and that

is

is

not Q.

Consequently, one or other of your assertions B 2

wrong,

S
either
is
'

CONNEXION OF
P
:

is

not

Q'
'

or

'

is

Y/
'

If the

first

be

right, the

second

wrong that is, X is not Y is right." The preceding argument runs as follows
greater than Z, then
all less

when A, B, C, &c.
less

are

all

2Z, 4Z, &c.

than

are all

than

A:

but 2Z,
all

4 Z, &c. are not

greater than Z].

therefore.

A, B, C, &c. are not

Corollary,
i

The preceding
more than
or all of the steps.

proposition

is

equally true

when,

istead of taking

the half at each step,

we

take the half

itself in

some

Prop. VI.
B, and
if

If there be two magnitudes of the

same kind,

A and

the scales of multiples be formed

A,

2 A,

3 A,

&c.

B,

2B,

3B,

&c.

then one of these two things must be true

tiples in the first

scale

; either, there are mulwhich are equal to multiples in the second or, there are multiples in the first scale which are as nearly

scale

equal as
set: that
either

we
is,

please to multiples (not the

same perhaps)
set,

in the

second
shall

we can
it

find

one of the

first

say

m A,

which

be equal
short of

to another in the

second

set,

say

nB,

or shall exceed

or

fall

by a quantity

less

than a given quantity Z, which

we

may name
gB.

as small as

we

please.

Let us take a multiple out of each set, any we please, say pK and Ifj9A and ^B be equal, the first part of the alternative exists; one must exceed the
other.

if not,

Let

pK

exceed ^B, say by

E;

then

we have

;?A

= ^B + E
B, or equal
the
to
;

(1)
B, or greater than B. If if the second, we have
:

Now E
tbiC

is

either less than

let
it

first,

remain
the

for
first

ph.

= + 1) B, or
(^
less

present

alternative exists

if

the third, then

can be so multiplied as to exceed E. Let (^ 1) B be the first of B which exceeds E ; that is, let the next below, or t B, multiple

be

than E, say by G, then

we have

E=^B
or

= ^B + pA=:(^ + OB + G
j9A
than

^B +

G
+ 1) B, +
^ is

Now G
or

must be

less

for

or

^B

+G

is

less than (^

^B + B.

We have

then

made

this first step (observe that

</

NUMBER AND MAGNITUDE.

Either the only some multiple of B; call itrB). we can find jsA and rB, so that
first

7
alternative

exists, or

pA = rB + G
Now G

where

is less

than

B
;

(2)

can be so multiplied as to exceed

let

uG

and

(u

+ 1) G

be the multiples of G, between which

B
say say

lies,

so that

v(j
(v
-{-

is is

less

than

B, B,

1)

greater than

=B K ?;G + G = B + L
vG
(v+l)G
and

and

it

follows that
if

K + L = G;for
lie

since

vG

differ

by G,

a magnitude

between them,

their difference

must be made
its

up

of the excess of that magnitude over the lesser, together with

defect from the greater.

Consequently, either
falls

and

are both

halves of G, or one of

them

short of the half.

Suppose

is

less

than the half of

then take both sides of (2), v times, and

we have

or or

vpA vpA vpA

if

= = =

^;rB

+ vG
K.
I)

vrB-{-B
(i;/-

(K

less

than half
t;

G)

But

be

less than the half of

G, take both sides of (2)

-f 1 times

which gives

or or If

+ l)pA = (v+'i)rB+(v+\)G {v +l)pA = v + i-rB + B + L ?r+l pA = (v+l r + 1)B + L (L

(y

less

than half
if

G)
we

K and L are both

halves of G,

not yet included) a multiple of have then

we may take either. And G, vG, be exactly equal

(a case

to

B,

vpA
which gives the
first

vrB + vG

(vr

+ 1)B
we
either prove the

first alternative.

Consequently,

alternative, or

we reduce

the equation

pA = rB + G
to

(G

less than

B)

an equation of the form

p'A

= r'BG' ^G
as before
;

not greater than the half of

G.

We
may

may now proceed

arise, let

but, to exemplify all the cases that

us take

CONNEXION OF

p'k
If v'G' be exactly B,

= /B-G'
first

we

prove the

alternative, as before

but

if

lie

between v'G' and (v'+l) G',

let

us suppose

= = B-K'i B + L' 3 (y' + 1 G'

v'G'
)

and

K + L' =
as before,

G'

which one of the two, K' or L', will not be greater than the half of G', so that we obtain by the same process, an equation of the form
in

p"k
By

qAG" I

G"

not greater than

the half of G'.

either,

proceeding in

this

way, we prove

l.The

first

alternative

of the proposition; or, 2. the possibility of forming a continued set

of equations

pA = ^BG,
the

j9'A

= ^'BG',
on
this

/)"A

= ^'BG",

&c.

where, in the scale of quantities G, G', G", &c., no one exceeds the
half of the preceding.
first

Consequently, we

may

(unless interrupted

by

alternative) carry
is

process until one of the quantities

G,
the

G',

G" &c.

smaller than

(Prop. V.) that

is,

we have
at

either

first

or second alternative of the problem.

And

exactly the the

same
outset

demonstration
jt>A

= 9B

may

be applied

to

the

case,

where

E.
it

This proposition proves nothing of a single magnitude, but

establishes

two apparently very distinct relations between magnitudes considered in pairs. There may be cases in which the first alternative
is

established at last

established.
is established.

We
it

shall first take the case in

and there may be cases in which it is never which the first alternative

Suppose

ascertained

by

the preceding process that

8A = 5B
Here
fore
is

an arithmetical equation between the magnitudes

of concrete
arithmetic
will

and there-

any processes

40th part
,. ,
.

(8x5 = 40) of both

apply.

Take

the

sides,

which gives
consequently the
or that which
is

8A
fifth

5B
^

A
the

B
as the eighth part of B,

40=4^
part of

5=8same
is

A is

contained 5 times in

also that

which

is

contained

NUMBER AND MAGNITUDE.

8 times in B.

Let

this fifth of

or eighth of

be called

M;

then

A = 5M, B = 8M,
sequently,

and
first

when

the

and Bare both multiples ofM. alternative of Prop. VI. exists, both

Con-

A and

are multiples of

some

third

proved, namely, that

when

magnitude M. The converse is readily and B are both multiples of any third

magnitude, the

first

alternative of Prop.

VI.

is true.

For

if

B=3/M,
of

we have
is

yA = 3^jrM,

xB = xi/Mf

ot
:

yA = xB.

A=^M,
The

term measure

M,

is

used conversely to multiple, thus if A be a multiple Hence in the case we are said to be a measure of B.

now

considering,

be commensurable.
able magnitudes,
first

A and B have a common measurCf and are said to We have therefore shewn that all commensurand commensurable magnitudes only,
to consider,
satisfy this

alternative.

There remains, then, only the second case

which

it is

now

evident contains those magnitudes

(if

any such there be) which

question therefore
is,

have no

common measure
as

whatsoever.

The

On this incommensurable magnitudes ? point the second alternative shews that our senses cannot judge, for let Z be the least magnitude of the kind in question, which they are
Are there such things
capable of perceiving (of course with the best telescopes, or other means of magnifying small quantities which can be obtained) then

we know thoXpA may be made to differ from ^B by less than Z, that is, we may say that all magnitudes are sensibly commensurable.
But
it

evidently does not follow that

all

magnitudes are mathema-

tically

commensurable; and
that take almost

it

has been shewn, by process of de-

* monstration, that there are

incommensurable quantities in such

abundance, any process of geometry we please, the odds are immense against any two results being commensurable. The suspicion that all magnitudes must be commensurable led to
the attempt, which lasted for centuries, to find the exact ratio of the

circumference of a circle to

its

diameter.

And

even now, though the

never tried by those who have knowledge enough to read demonstration of its impossibility, no small number of persons
adventure
is

Legendre, and others before him, have shewn that the diameter and circumference of a circle are incommensurable and the student will
;

Algebra, p. 98, or in the Lib. Useful Know., treatise on the Study of Mathematics, p. 81, proof that the side and diagonal of a square are incommensurables. Also in Legendre's Geometry, or Sir D. Brewfind in

my

ster's Translation.

NUMBER AND MAGNITUDE.

Stand in the

way

of perfect satisfaction, or at least ought to do so.

at the outset is
all

Granting that in the preceding case the error

terial, let

immaincom-

us suppose the student disposed to substitute for

mensurables, magnitudes very near to them which are commensurables,

and thus

to continue his career

till

he comes to the highest branches

set

of applied mathematics.

Let us suppose a

of processes, beginning

in arithmetic, continued through algebra, the differential calculus, &c.,

up to a point in optics or astronomy, in a series of results, embracing, we may suppose, ten thousand inferences. If he set out with an
erroneous method, what security has he that the error will not be multiplied ten

magnitude.
that
it

thousand fold at the end, and thus become of perceptible If somebody acquainted with the subject have told him

will not so

sciences

and receive the

happen, he might as well skip the intermediate result he wants to obtain on the authority of

that person, as study

them

in a

manner, the correctness or incorrectness

of which depends on that person's authority. If he answer that the such multiplication of errors, appears extremely imresult, namely,
probable,
it

may be
;

replied, firstly., that that

is

more than he can

by pursuing his mathematical studies on such a presumption, he makes all the pure sciences present probable results only, not demonstrated results ; more probable, perhaps, than

undertake to decide

secondly, that

many

parts of history, but resting on an impression

which

must, in his mind, be the result of testimony.

It appears,

however, that

we may

expect series of collateral results,

the one for commensurables, the

other for incommensurables, and

;

presenting great resemblances to each other

alteration,

for

we may, by any

however minute, convert the

the former.

But

this

we may

kind of magnitude into prevent, by extending our notions of

latter

arithmetical operations, or rather

by applying to magnitude processes which are usually applied to number only, as follows If we examine the processes of arithmetic, we find, 1st, Addition
:

and substraction,

to which abstract number is not necessary, since the concrete magnitudes themselves can be added or subtracted. 2d, Mul-

tiplication, the raising of

powers and the extraction of

essentially

roots, in all

of

which abstract number

operation.
abstract

is

supposed
it

to

be the subject of

3d, Division, in which

in

is

not necessary to suppose

number

finding the whole part of the quotient, but in

which we cannot, without reference to numbers, compare the remainder and divisor, in order to form the fraction of the quotient. finishing

12
4th,

CONNEXION OF
The process of finding
maimer which
is

the greatest
is

common measure

of two quan-

tities, in

which the remainder

in a

not compared with the divisor, except as applicable to the case of concrete magnitudes

as of abstract numbers.

thod of finding the greatest

this, we shall demonstrate the mecommon measure of two magnitudes. Let A and B be two magnitudes, which have a common measure 6M. Then, it is clear that let A=aM, B

To shew

a;A

+ ?/B

or

{xa-{-yh)M.,
it

xK

yB

or

{xa-^yh)M

have the same measure, unless

should happen that in the latter case Let A be the greater of the two, in which case xA^=i/B. xa=:j/bf 1 times, so that and let A contain B more than /3 and less than /3

A=j8B + B', when

measured by
times
;

B'

is

less

than B.

Then B' being

/3'

A
than

/3B,
)3'

is

M.
/S"

Let

contain B' more than

B''

and

less

+1
And

or let

B=i3'B'+

where B"

is less

than B'.
B"',

Let B' contain

B" more than

times, &c., or let

B'=iS"B" +
&c.

and so on.

B"

or

jS'B'is
:

measured by

M,

We

have then the following

conditions

A
B

is

a multiple of

M M
but
is

A = B + B'
jS

B'

< B,

a multiple of

B=/3'B +

B"

B'=/5"B' +B'" Now, since B B' B"

of

B" <B' B' <B''

are decreasing quantities,
series,

and

all

multiples

M,

they are

all to

be found in the

M,
in

2M,

3M,

4M,

&c.

which continual decrease must bring us

at last to nothing, or

we

must end with an equation of the form.*

that

is,

one remainder

is

a multiple of the next.

;

To

take a case, let

the

fifth

equation finish the process

so that, in addition to the pre-

ceding,

we have

B"

B"
*

= =

/3"B'
iS^^B'^

H-

B^^

When

a letter denotes an indefinite

number of

accents,

it is

dis-

iu brackets, and higher tinguished from an exponent by being placed numbers of accents than three are usually denoted by Roman numerals.

NUMBER AND MAGNITUDE.

In the fourth, substitute B'" from the
fifth,

13

giving

In the third, substitute

B" and

B'" as found, giving

B'

(/5"/3";/3i^

+
B"

/3"

4-/3^0 ^'"

In the second, substitute B' and

as found, giving
/3'/3^^

B =
In the
first,

(/3'/3"^'"/3i^

+ /3'/3" +
and B'

+ /3'"/3^^ +
+i8''^'''0i^

1) B^^

substitute

as found, giving
/3/5'''/3i-

A=

(/3/3'/3''/3'''/3i^

/3/3'/3'' is

+ /3/3'i8i^ +

+ + ^^
/3

Consequently, Bi^
the outset
is

common measure

of

and

but, since

M at
M,
com-

common
M, mon
2 &c.

any common measure we please, measure. Then B'^ must be M, for and were
it

let it
it is

be the greatest

in the series

any other than

M,

there

would be

B*^ a

measure, greater than the greatest.

Hence

this process deter-

mines the greatest common measure, and also the number of times which each of the two, A and B, contains the greatest common measure.
It is here

most

essential to observe, that this

whole process

is

in-

dependent of any arithmetic, except pure addition and subtraction, which can be performed on the magnitudes themselves, without any
numerical relation whatsoever; the only thing required being the

axiom in page
lines.

3.

We

shall actually

exemplify

this

on two

right

Ah
B
h
z

y
"-{

--'

k
(xi/)

A = B + (XT/) B =
Therefore

+ (zk)

{xy)

2{zk)

B =
by
to

3(2^^)

K =

6{zk)

In

this case,

actual

and

A are found

measurement (supposed geometrically exact) be respectively 3 and 5 times zk.

When
greatest

the preceding process has an end,

;

we

therefore detect the

is

common measure and we common measure, the process must

sequently, in the case where there
is

have shewn, that where there

give
it,

with the converse.

Con-

no common measure,

this process

must go on

for ever,

and we have an interminable

series of equations,

A=/3B+B', B=/3'B'4-B", B'=/3"B"+B'",

c

&c., the conditions of

14
which
series,

CONNEXION OF
are, tliat

B, B', B", B'", &c., are a continually decreasing

does not follow that each one was
less

though

it

than the half

of the preceding.
stitutions

We

shall

now examine

the effect of successive sub:

there be

from the beginning, first making the following remark and B, of which any two incommensurable quantities

If

A is

the greater, then there follows an interminable set of whole numbers,

/S, /3',

/3",

.... which are not subject to any particular law, but can

be found when
tities,

and

B
/3

are given

and an interminable

set of

quan-

A, B,

B', B", .... connected with the former

by

this law, that

/3'

contains

between

and

/3

+1

times

contains B' between

and /S'+l times, and so on.

We have B'
or
.

= A jSE B" = B-/5'B' = B-^ (A-/3B) B" = (/3/3'+l)B-/5'A B"'= B'-/3"B" = A-i8B-(/3/3' + l)/3"B + /3'/r'A = (/3'/3"+l)A-(/3/3'/3" + + /3")B
/5

and thus

we go on
and

representing the remainders alternately, in the form

pA

qB

gB
:

pA.
to

We

may

easily find the

law of the co-

efficients, as follows

Suppose we come

Then we have
or

= qB--pA = p'A-^'B = B("+i) B("+2) B^") B("+i> B<'+2> = B<"> = ^B-;?A-i8("+i) (p'A^q'B) = (/3("+i)^' + ^)B-(/3<"+i)y+i>)A
B(")

B(n+i)

/3("+i)

-f-

/3("+^)

or

if,

continuing the preceding notation,

we suppose

we have
so that,
if

f=
we

/3("+i)

-h

p
/S(")

q"

/SC'+i) q'

+q

write the values of B', B", .... with the following no-

tation, putting opposite to in the B(") of the

each the
;

which occurs /or the first time

equation

namely,
/5

B'

=j9iA-^iB

B"

qc^B^p^k

NUMBER AND MAGNITUDE.

B'" B^^

15
^"
^"'

^p,k^q,B
&c.

= q^B-p^k
method of forming p
in succession.

&c.

&c.

we have

the following uniform

and qn

for dif-

ferent values of
i?i

P2

= =

1
/5'

^1

= = ^2
^4

/3
/3'/3

/?4

=
&c.
it

^">3
is

+ /^s
&c.

=
&c.

^"g's

+ qz
&c.

in

which

^j q^

plain, from the method of formation, that />, p^ &c. &c. are increasing whole numbers, so that we may continue,
fail, till

supposing B' B" .... never

pn and qn
....

are greater than

number named.
ad infinitum.

And

since B'

B"

are all less than B,

any and

therefore less than

A, we have the following succession of

results,

pi A.
PnA. PsA.
&c.

is

greater than

^'jB

but

less than

{qi -f
(qo

is less is

than

^^B
^'sB
&c.

but greater than

1)B 1)B

greater than

but

less

than

(g'3+l)B
&c.

Hence,

it

appears that

is

greater than

Pi
less

than

greater than

less than

&cc.

ad

inf.

Now, from
B,
B.

this table of relations,

we can determine whether any

given multiple of A,
3/

To do

this

xAy we must
is

greater or less than any given multiple of

inquire between what two consecutive

multiples of

does

xA

lie.
:

We now proceed as follows 1. We must shew that any fraction, such

r

as

b+n

lies

between

and

16

CONNEXION OF
first

unless where the two latter are equal, in which case the the same. The preceding must be true if

also is

+m

lies

between

^ (5

-f ?^)

and

{b

+ n) m
bm
n

or

a
.

-4-

-7-

and
n

b
O'fi'

-h

or

-\-

m + n-r b
:

am nb

and

-^

m+bm n
-,

b-r b

which
than a
2.

is

evidently true

for if

- be
if

greater than

-,

the

first is

greater

+ m, and
We now
IT P3

the second less

see that

^ be

less

than

vice versa.

''

^^^ i^y^+Pi
^'''^3

l'^

between

^
?-=

and li
Pi

^"p.2

and -^
Pi
92

P2

1
P4

or

+ ya
all

^3
^3

^"'Pz+P2
Consequently, to arrange

P2
con-

and so on.

the fractions thus

sidered, in order of magnitude,

we must

write

them

thus,

li
Pi
3
to
.

il
Ps

^ "" Ps
,

%
Pe

9i

92

P4

P2
;

We
gm

can thus bring two fractions as near together as we please

take three consecutive fractions

prove

this,

gm+i

Pm

Pm+1
to a

/ gm+i \pm+2

^^

fi^"^'^^^gm+i
/S^'^+^^J^m+l

+gm \

+ pm J
and second, and

which reduced

common

denominator, the

first

the second and third, give

gm pm+l

gm+l Pm
Pm-\-l

Pm pm+1
and
in

pm
Pm+1 Pm+l pm+2
is

^^"''^^^9m+l

Pm+l Pm+l Pm+2

is

+ Pm gm+1

^^^

^^'""''^Vrw+1

+ Pm+1
the

gm

which

it

clear that the difference of the numerators

same

in each couple,

but that

if

the

first

numerator be the greater of the

NUMBER AND MAGNITUDE.

first

17
;

couple, the second numerator

is

that of the second

a result we

might have foreseen, having proved that

^^^
Hence
it

lies

between

'till,

and

^.
any two

follows that the numerator

of the

difference of

successive fractions of the set

^
Pi
is

2
P2

ll
Pa
is,

the

same

as that of the difference immediately preceding, that

the difference of

Pn
the difference of

and i^^^ has the same numerator as pn-l

^^^
Pn-l

and ^""^ Pn-2 and

P2
Pi
/3'
;

which has the same

but

numerator as the difference of

9i

^2

_
is

therefore this numerator of the differences

always
^

1, or

^JL

and ^^i^
Pra+1

differ

Pn

by

Pn ^n+l

Hence

the difference

may be made
named by
jon+i-

as small as

we

please, or smaller

than any fraction

us, since

pn

itself

can be made greater

than

rrij

much morej^n

4.

These

fractions cannot for ever lie alternately

on one side and

V
the other of any given fraction -.

For

if this

were possible, then, since

A lies between

^B
Pn
and
since

and
V

^^^B
i>n+l

by the supposition

-B

does the same, and since the couple

just mentioned can be

made

to differ,

by as small a

fraction of

as

we

please, then

we

should have

-B A= X
c 2

18
where
that

CONNEXION OF

may be made
V
;

as small as
either be

we

please.

Now

this is saying

A= B

for

must

-B
but the
latter it is

or

-B
;

ih some

definite

magnitude;
are trying leads to

not

for the supposition

we

=: -

B +

a magnitude as small as

we

please.

Consequently, our supposition that the series of fractions lie alterV nately on one side and the .other of a definite fraction -, leads to the

conclusion that
B'

and

B" ....

finishes,

commensurable, or the process of finding as we have shewn. But it does not finish, by
are

hypothesis; therefore

the series of fractions

cannot

lie

alternately

on one side and the other of -. X

We
is

can

now shew between what

multiples of

jtA must

lie.

It

clear that

xA
now
it

lies

between

-^^B
Pn

and

^^^^^B
Pn+1

is

not possible that any whole

number v should always

would

lie

between

~ and
pn
J^

^""^^
;

for if so, then

pn+l

- always
which has been proved

lie

between

Pn
be impossible.

and ^- pn+l
Consequently,

to

"^^"R
pn
must come
still

A '^'?"+i"R

Pn+i
at last

(which approach each other without limit)

always to

lie

between two multiples of B; and

the7n.

more must x A, which

enough,

lies

between

far

can always find between what multiples of and thence whether xA is greater or less than 3/ B.

we

Hence, by proceeding B lies x A ;

We
1.

have thus divided

all

pairs of magnitudes into

two

classes,

Commensuralles, in which

we can always

say that
tell

-B,

y and

p being whole numbers, and can always

fraction of

or B,

x A exceeds or

falls

short of 3/B.

exactly by what For we have

NUMBER AND MAGNITUDE.

:rA

19

2/B= (x-^y^-\k= (x^^y\^

if

xk>yB
a;A<7/B

(3/B-a;A)= (|y-.:r)A= (y-^|)B

2.

if

Incommensurables, in which

we can

never say

A=

- B, but

P
in

which we can assign a

series of fractions alternately increasing

and

decreasing, but

making

less

and

less

change
I2.

at every step,

1l
Pi

ll
P2

Pa

and such
so that

that

A is

greater than
Pi

B,

less

than

B, &c. ad infinitum:
P2

we can always

assign

A = ^B + K Pn
where
always
exactly

is less

tell

than any magnitude we name; and such that we can by them whether ^A exceeds or falls short of yB, but not

how much.

Let us suppose, as an example, that we have two magnitudes and B, which tried by the process in page 13, give

A= B+
or suppose

B',

B
/3

= =
5
5

B'

+B
/3'

',

B'
/i"

Ij

1,

Hence

the several values

ofp and
6

q are

= B" + B", &c. ad = &c. ad = = P3 = p,

1.,

inf.

inf.

1,

;}2

1,

2,

&c.

as in this table,

12
p
g-

4
3 5

7 13
21

10
55
89

11
.

1 1

2 3

21

34 55

89

12 ^ &c 144

13

34

144 233 &c.

&c.

a>b<2B>|b<|b>|b<^b
Hence

A 2A
3A 4A 5A

lies

between

B 3B 4B 6B 8B

and
..
..

..
..

2B 4B 5B 7B 9B

&c.

If

we wish

to

know between what

multiples of

100 A

lies,

we

find

20

CONNEXION OF

^>W^< TiiB; lOOA > leigB

or 100

<

161

H|b

lies

between 161

and 162 B.
call

We
B

can thus form what

we may

a relative multiple scale made

inserting the multiples of

by writing down
in their

the multiples of

A, and

proper places; or vice versa.

of this scale
is

In the instance just given

the

commencement

B, A, 2B, 3B, 2A, 4B, 3A, 5B, 6B, 4A, 7B, 8B, 5A, 9B, &c.
which we may continue as
scale as follows
far as

we

please by simple arithmetic.

If

the magnitudes in question be lines,

:

we may

represent this multiple

01

)fH

B<

>M

\-^

1^^^

Measuring from O, the crosses mark

multiples of B.

off multiples of

A, and the bars

Thus

=B Oxi = A
01i

Ol2=2B
0x2 = 2A

Ol3=3B&c. Ox3 = 3A&c.

shall now proceed to some considerations connected with a multiple scale, for the purpose of accustoming the mind of the
its consideration. We may imagine a scale like the prebe equivalent to an infinite number of assertions or negaceding tions, each one connected with the interval of magnitude lying between two multiples of B. Thus, the preceding scale contains the following

We

student to
to

list
^

ad infinitum.
1.
2.
3.

Between
Between Between Between

and

4.

B 2B 3B
&c.

and
and

and

B 2B 3B 4B
&c.
1st,

lies

no multiple of

A
A

lies lies
lies

A
no multiple of

2A
&c.
the

&c.

Now, on
series,

this

we

remark,

That the negatives of

above

though they appear prove nothing, yet in reality have each an infinite number of negative consequences. From the third assertion of the preceding list, namely, neither A, nor 2 A, nor
3 A, &c.
following
lies

at first to

between

2B

and 3B, we immediately deduce

2 and 3 B, nor between -

all

the

A does not lie between 2 B

and

NUMBER AND MAGNITUDE.

q

21
3

-B, nor between ' 2

certain

-B
3

and - B, nor between '

3

-B
4

and -B, &c. &c.

4'

2d, Observe that every affirmative assertion in the above includes a

number

of the affirmative ones which precede, and an infinite

number

of parts of the negative ones preceding and following.

For

instance,

we
100

find that

100 A

lies

between 161
is

and 162 B, or

lies

between

and

100

B, that '

between

B
3B

and 2B.

Again, 2

322
lies

between

100

and

324

100

B, or between '

and 4B.

Similarly, ^

3A

lies

between -rr-^ and

-B,

or between

4B

and

5B; and

it

might thus seem

ations preceding

at first as if every affirmation

its

necessary consequences.
it lies

But

made all the affirmif we try 90 A by

the preceding,

we

shall find that

between

^^^^^B
100
so that

and

iM2iB 100
affirm

or between

144 B and 146 B

either

we can only

90 A

to lie

between 144 B and

145 B, or between 145 B and 146 B, but we do not (from this) know which. But we can say that 90 A does not lie between 146 B and

147 B, or between 147 B and 148 B, &c.

Let /cA
between ^B and (Z+1)B
or

The points

at

which any
thus found.

affirmation does not determine those preceding

lie
;

may be

lies

between

-B

and

mk
If
-t;-

-^B
7
^^

and

_1_!_^B
then

and

lie

between

and

t-{-\,

mK lies

between

{t +1)B: but if, in going from the first to the second, we pass through a whole number, or if wj/, divided by /c, gives a quotient t and remainder r, and W2(/ 1), divided by /c, gives a quotient t-{-l

^B and

and remainder

r',

then

we have

or

- ^^~k m = k r+
k
in all cases

k
r'

- + ^+
^

k r

or
is

-\-

m ^=

-\-

and

where r

+m

greater than k, this condition can be

fulfilled.

The process may be shortened, by using

instead of

the

22

CONNEXION OP
/

remainder arising from dividing

by

k.

Suppose, for instance,

it is

required to determine what preceding affirmatives are ascertained by the proposition 10 lies between 33 B and 34 B. have then

We

/=33,

/c

= 10,
t= t=
^

remainder of

/-j-/c

3.

=2 m=S m=4 m=5 m=6 w=7

?W

= 13 = l9 = 23
= 29

t=l6
t
jf

= r=9 r=2 r=5 r=S r= l

r
6'

2A

lies

between

6B
13B 16B

and

7B
14B 17B

4A 5A
7A

and and

23B

and

24B

m=zS

t=z26
^

7'r=4
7'

m=9
By

=7
it

proceeding thus,

will

appear that there

is

no perceptible law

A, among B, 2B, &c., derivable from the sole condition of /cA lying between IB and (/+1)B. Nevertlieless, it is

regulating the places of

2 A,

easy to prove, that

if all

the rest of the relative scale be

given from and after any given point, that the whole of the preceding For, suppose kB to be the commencepart can then be determined.

ment of the
for,

part of the scale given, and let the place of

first

mA be

asked

which precedes A A, the

Multiply

multiple of

appearing in the

scale.

by g, so

that

mg

shall

mgA
(w

appears in the portion of the scale given, say

Therefore

be greater than h. Then between wB and

+ l)B.

mA
and
but
if

lies

between

-B
S
t

and

B
g
1
,

and
o
o

lie

between

and

the question

is

settled

this

must always be

the case, if

we

include the case where

W -

or

is itself

a whole number.
precedes,

g
From
1
.

all that

we draw

the following conclusions

Having given

A and

B, two incommensurable magnitudes of

the same species (both lengths, both weights, &c.),

we can

assign,

by
in

a processes embling that of finding the greatest

arithmetic, the relative scale of multiples of

common measure
lies,

and B, which points

or

out between what two multiples of

vice versa.

any given multiple of A

NUMBER AND MAGNITUDE.

2.

23
we can

Any
it

part of the beginning of this scale being deficient,

construct
3.

by means of the

rest.

can find a magnitude which shall be commensurable with A, differing from B by less than any magnitude we name ; and can
assign the fraction which
it is

We

of

Given the two magnitudes,

but when the scale
it is is

their relative multiple scale is given

given, the
is

two magnitudes are not given.

an
infinite

For

easily

proved that there

number of couples

of

mag-

nitudes which have the same scale with any given one.
of

Let the scale

the same,

and

be given ^

then will the scale

of- A and - B be
9

where p and q are any whole numbers

For
if

9 whatsoever.

kA
;fe

lie

between between

IB

and and

(Z+1)B
(Z ^

then

^A
9

lies

Z^B
9

^ 'q + 1)^B

or

making ^

^A =
q

A'
between
is

^
q

B =

B'

A A'
whence
value of
the scale of
/c.

lies

ZB'
same

and

(Z+1)B'
and B'
for

A and B

the

as that of A'

any

What

is it,

theuy which is given

;

when

the scale

is

given

Not the
it

magnitudes themselves

for if the scale

belong to

A
P

and B,

also

belongs to every one of the infinite cases of

therefore, only defines

-A

and -B.

The

scale,

such a relation between the magnitudes as be-

longs to 2

and 2B,

3A

and SB, &c., as well

as to

and B.

It is

usual to call this relation the proportion between the two quantities in common life, and in mathematics their ratio ; in Euclid the terra is

Two

magnitudes,

and B, are said

to have the

same

ratio as

two other magnitudes, P and Q, when the relative scales of the two are the same ; that is, when the multiples of Q are distributed as to
magnitude among those of P,
are distributed
in the

among

those of A.

nitudes of one kind, two areas, for

another, two lines, for instance.
It is easy to

same way precisely as those of B And P and Q may be two maginstance, while A and B may be of

shew

that this accordance of scales is equivalent to

it

the

common

idea of proportion, such as

would become

if

we

took

24
all

CONNEXION OF

means of companson away, except that of multiples. Let us imagine A and B to be two lines in a picture, and P and Q the two
corresponding lines in what is meant for an exact copy on a larger are in the proper Set an artist to determine whether P and scale.

proportion to each other, without any assistance except the means of

repeating A, B, P, Q, as
as follows
:

many

times as he pleases.
little

He

will reason
it

" If

be ever so

out of proportion to P, though

may

not be visible to the eye, yet every multiplication of the two will
it

increase the error, so that at last

will

become

perceptible.
if it

If there
to lie

be a line 100 A laid down in the

first

picture,

and
lie

be found

between 51

B
if

and 52 B, then should 100

between 51
not
lie

and

52 Q.
51

But

be a

little

wrong, then 100 P

may

between

Q and

52 Q."
remains to see whether
this definition of

It only

proportion will
satisfy

include the case of commensurable quantities.

These

such an

equation as/cA easy to shew that

= /B,
for

being two whole numbers, and it is the whole relative scale is divided into an infinite
/

k and

succession of similar portions.

the whole scale
;

Firstly, this

one equation determines

we have

A
or if
lie

= k JB
t

mA='^B k
^

between

and

K
if

+ 1 m A lies between t B and (^ + 1 )B

,

^^

mA =
for instance,

tB

Let us suppose,

A 2A 3A 4A
From
this

lies

7 A = - B. Then we have B and 2B between 3B and 4B 5B and 6B

is

equal to

7B

or the scale

is

B A 2B 3B 2A 4B 5B 3A 6B 4A l^
point the scale begins again in the same order.
is

Thus,

the second portion

l^ 8B 5A 4A

9B lOB 6A IIB 12B 7A 13B ^t^ 8A

The
arithmetical definition of

and so on ad infinitum.

having the

NUMBER AND MAGNITUDE.

same
ratio to

25

whicli

has to Q,

is

simply that of

being the

same

fraction of

which

is

of

or if

A =
Now,
both
;

/ k

P =

r k
-j-,

Q
it

since the scale depends entirely on

if the

is

the

same

for

conversely,

scale of

and

be

tlie

same

as that of

and Q, then if/cA

= /B,
:

/cP must

= /Q.

Hence

the two defi-

nitions are

synonymous

if

When

the multiple scale of

one applies, the other does also. A and B is the same as that of

P and

Q, we have recognised the proportionality of A and B to P and Q. But these scales may differ. The question now is, may they differ in all possible ways, or how far will their manner of differing in one
part of the scale affect their

manner of differing

in others

Am

I, to

take an instance, at liberty to say, that there

may be
of

four magnitudes

such that 20 A exceeds 18B, while 20

falls short

18Q;

but that,

for

the
?

17 Q

same magnitudes, 13 A falls short of 17B, while 13 Such questions as this we proceed to try.

exceeds

When
is

the complete
is

only two things are possible, which cannot co-exist, each and only contradiction of the other the assertion of
:

one

a denial of the other, and vice versa.

But when

three different

things are possible, one only of

which can be
;

true, the assertion of

one contradicts both of the other two

establish either of the other two.

the denial of

one does not

The want of
that
is,

common

term, which

may simply mean

not

less,

either equal or greater, without specifying which,

causes some confusion in mathematical language.

and so on, To remind the

equal or

student that not

greater,

less

does

not

mean

greater, but

either
less

we

shall

put such words in

italics.

Thus, not
:

and

less,

not greater and greater, are complete contradictions

the denial of

one

is

the assertion of the other.

If

and

be two magnitudes of one kind, and

and

two

others, of the same or another kind, such that

mA
then
it is

is less

than

7i

B,

W2

is

not less than

Q
n'Q,

impossible that there should be any multiples such that

is

m'A
For we

greater than

n B,

ml?

is

not greater than

find,

from the

first

of each pair,

26

CONNEXION OF

A
still

is less

than

B,

A
r m B

n'
is

greater than

more

is

B
less

greater than

or

greater than

m
r

But

P
Now,

is

not

than

Q,

is

not greater than

all the four

combinations of this
greater than
-

latter assertion contradict

is

as follows

=
P P

Q,

=
Q,

Q,

gives

greater than

^,

Q,

gives

less than

-.

Q,

less than

Q,
-,

gives

less

than

greater than

Q,

less than

Q,

gives

less

than

Hence

the two suppositions above cannot be true together

the

happening of any one case of either proves every case of the other to
be impossible.
If

we range
:

all

the possible assertions which can be

made, we

have as follows

As

wz

is

greater than

wP mV
?w'P nfiV
rriV

is

equal

to

is less

than

wQ wQ
w'Q
w'Q
'Q

is

greater than

is

equal to
than

is less

Four of these must be

81

true,

one out of each

triad

and there are

ways of taking one of each, so as to put four together. But we shall take the sets A and a together, and find what inference we can
draw by taking one out of each.

NUMBlSR AND MAGNITUDE.

27
P3 p^

A3

proves

-,

greater than

as does

A3

tti

proves

greater than

as does

P3 p^

Aq
A2 A2

proves

less than

as does

Pg p^

proves

-,

equal to
greater than

as does

P2

p<i

proves

as does

P2 pi

Ai

^3

proves

less

than

as does

Pi

Ai
Ai
Now,
in the

a^ ai
if

proves

less

than

as does

P^ p^ Pi Pi

proves nothing

neither does

we put

these pairs together, or

make

pairs of assertions,

manner already done, we have 81

distinct sets of four asser-

tions, divisible into those

cannot be true together.

Agfla,

which mnj/ be true together, and those which An inconsequential supposition, such as

may

co-exist with

any of the

rest

from the other

set

Fp;

but
in

those which give

the set

7-

necessarily greater, equal to, or less than

A a,
3

can only co-exist either with the similar ones from the

set P/7, or with those

which are inconsequential.

any marked

Thus we have

A3 A3

may be

true with
either

a^ requires

Ps^s,
P3P3,

PsP.,

A3G1

A23
P3P3,

Ag^i
Aifla Aittg

P3P3,

P3P3,

28

CONNEXION OP

them, as a simple induction from the preceding will shew how to Attach an idea of classify those which may and cannot be true.

magnitude to the phrases greater ^ equal, and

greater than B,"
is
is

less;

" say that

this

is

" higher than

less

is

equal to

B," and

again

" higher than

phrases

than JB,"

We

by

the highest numbers.

Say that in AgCfj,

have marked the highest A3 ,, &c. (calling

and a the antecedent clauses of any four marked A, a, P, p), the antecedents are descending; in A3O3, Agtfj, and A, a,, stationary;

and in Aj a^, Aj Og, &c. ascending. Then all the propositions which imply the co-existence of any two antecedents, and any two consequents of the form A a P^,-may be divided into those which may be
true,

and those which cannot be true, by the two following rules Ascending antecedents cannot have descending consequents. Descending antecedents cannot have ascending consequents.

Precisely the

same

rules will apply if

we

take two propositions

and two others ap for consequents ; as we may For if either deduce in the same manner, or by simple inversion.
for antecedents,

AP

Pp, with any numerals

subscribed, do not contradict either of the

preceding

rules, neither will the

corresponding case of
,

A P aj? do
;

so,

and the contrary. Instances, Aj and AgPg Oj j^a? ^^

ascending assertion relative
values of
to

P3P3 and A, P3 a^p^

k^a^

Pg^jg

Let us then take a case of A, B, P, Q, in which we find one

mh, wB,
is less
is

twP,

nQ,

for

some

particular

and n

for instance
*

p ^ r3A
tsP
is

than

greater than

4B 4Q

which, as

we have

seen,

never contradicted in form by any assertion

that can be true of

any other multiples.

These four quantities are

not proportionals: for 3 being less than 4B, and 3P greater than in the same place as in the 4Q, P cannot lie in the scale of P and and B. But to what more common notion can we assiscale of

milate this sort of relation between A, B, P, and Q, namely, that

all true assertions

of the form

(AP)

are either ascending or stationary,

this in

and never descending ? Have we any thing corresponding to the arithmetic of commensurable quantities? Let us suppose

and

commensurable, and also

and

say that

A = iB

^,Q n'

NUMBER AND MAGNITUDE.

Then
t .

29

3-B
,
,

is less

than
t'

4B
.

3-Q
,

is

greater than

4Q
,

IS

less than

V
or

4 -;
3

-,

is

greater than ^

4 3

f
;

-, is

v
;

greater than *

is

a less fraction of

than

is

of

which

in arithmetic is

also said thus,

bears a less proportion to

than

does to Q, or

bears to

a greater proportion than

bears to B.

Hence we

get

the following definitions, in which

we

insert the previous definition of

proportion, and the accordance of the whole will be seen.

When
a
/ess

all

true

assertions

ascending or stationary,
ratio

on (m A, n B) (m P, n Q) are either and never descending, A is said to have to B

always stationary, the same ratio;
or stationary,

than

to

Q; when

when always descending

ratio.

and never ascending, a greater

the

This amounts in

fact

to

the

definition

given by

Euclid,

opening part of whose Fifth Book from, with a few remarks.

we

shall

now make some

extracts

Definition

III.

Ratio

is

a certain mutual habitude {y-xScn

.,

method of holding or having, mode or kind of existence) of two magnitudes of the same kind, depending upon their quantuplicity it means relative {vnXiKornt^ for which there is no English word
;

greatness,

and

is

the substantive which refers to the

is

number of times

or parts of times one

in the other).

In

this definition,
it

Euclid gives that sort of inexact notion of ratio

quantities,

which defines
as to
its

in

commensurable

and gives some

the
its

light

general meaning.

It stands here like

lies

definition of a

straight line,

" that which

evenly between

extreme points"
cannot enclose In most

prior to

the

common
is

notion,

" two

straight

lines

space," which

the actual subsequent test of straightness.

of the editions of Euclid

we

see " Ratio

is

a mutual habitude of two

magnitudes with respect to quantityy^ which makes the definition unmeaning. For quantity and magnitude in our language are very
nearly,
it is

if

not quite, synonymous; or

if

any

distinction can be drawn,

this:

magnitude

is

the quantity of space in any part of space.

is here speaking of magnitude generally (not of space magnitudes only) the words magnitude and quantity are the same.*

But

as Euclid

Euclid again uses the word

settles its

manner which
reader

'^en'kmorni (Book VI. def. 5) in a The more advanced meaning conclusively.

may

consult Wallis, O^era Mathematica, v. 11. p. 665.

D 2

30
Definition IV.
means

CONNEXION OF
Magnitudes are said
to

have a

ratio to

each This

other which can, being multiplied, exceed " one the other."
that quantities have a ratio

when, any multiples of both being

It is usually

taken, the relation of greater or less exists.

rendered

" Two magnitudes

translated, consistent.

are said to have a ratio

multiplied so as to exceed the greater."

when the lesser can be But the above is literally

makes
the next definition

and the sense here given

It is

to ratio

a way of expressing that the two magnitudes must

should be applicable to them.

to

be of the same kind, which requires that the notion of greater and less That this notion should be applicable
the quantities themselves as well as their multiples, being the necessary and sufficient condition of the possibility of the comparison * as the distinction of in the next is here assumed
definition,
ratio.

implied

quantities

which have a

Definition V.
the
first

to the second,
first

Magnitudes are said to be in the same ratio and the third to the fourth when the same
:

multiples of the

and

third being taken,

and also of the second

and fourth, with any multiplication, the first and third (multiples) are greater than the second and fourth together, or equal to them together,
or less than them together.

relative multiple scale of

This amounts to our definition of proportion, namely, that the and B is the same as that of P and Q.

For, take the same multiples of

the same multiples of
relative

multiple scales
it

A and P, namely, and mP, and and Q, namely, wB and nQ. Then, if the be the same, let mA lie between vB and
lies

mA

(u-i-l)B,

follows that

mF
less

between

w be
772

less than

v,nB
is

is

than vB, and

vQ and (v + l)Q. If, then, nQ less than vQ. And

nQ.
to

A being

greater than

vB

must be

greater than tjB, while, for the

same reason,

mP

simultaneously greater than

In the same
be included

way
in

the other parts of the definition

V. may be shewn

that of identity of multiple scales.

Now,

reverse the supposition

* The
definition

common

version

is

several times referred to afterwards, and the

4 expressly alluded to, in the editions of Euclid. But it must be remembered that the Greek of Euclid contains no references to preceding propositions, these having been supplied bj'- commentators. The
reader may, if he can, make Aoyov ip(;,nv t^os SikktiXa //.lyUyi Xtyira,!, a ^vvocrai ToKXa.'TrXcifftu.^ofAiva. ocXXriXuv vTri^i^uv mean, " Magnitudes are said to have a ratio, when the less can be multiplied so as to exceed
the greater."

NUMBER AND MAGNITUDE.

and assume Euclid's definition.
If,

31
between

then,

mA

lie

vB
is

and
the
less

(v4-l)B,
than
fore,

it

follows that

mA

is

greater than

vB, whence, by

assumption

mP

is

greater than
is

vQ.

Similarly, because
less than

wA

(v+l)B, TwP

(by that definition)

(i;+l)Q.

There-

wP

lies

between

vQ

and (v+l)Q, or

in this instance, or for

any one value of m, the scales are accordant, and the same may be proved in any other case. It follows, then, that the two definitions
are mutually inclusive of each other.

The manner

in

which Euclid arrived

at this definition has will

been
nine

matter of inquiry.

But any one who

examine the

first

propositions of the tenth* book, will see that he had precisely the
at it as we have used. But, besides this, he might have come by the definition from a common notion of practical mensuration, as follows. Suppose two rods given, one of which is

same means of arriving

the English yard, the other the French metre, but neither of

them
is

subdivided.

The only

indication which looking at

them

will offer,

that the metre exceeds the yard apparently

by about ten per

cent.

To

get a

more exact notion,

the obvious plan will be to measure

some

Suppose 100 yards to be taken off with the yard measure, it will be found that that 100 yards contains about 91 metres and a half, the half being taken by estimation, and we will
great distance with both.

suppose the eye could not thus err by a quarter of a metre. Then the yard must be -915 nearly of a metre, and the error upon one yard cannot exceed the hundredth part of the quarter of a metre, or '0025
of the metre.
correct, will

But

the mathematician, to

make

this

process perfectly

suppose distance ad infinitum, measured from a point

both in yards and metres, or in fact will form what

multiple scale.

we

call the relative

He
a

a multiple of
this

then looks along this scale for a point at which If yard, and a multiple of a metre end together.
it

happen, and

thus appear that

is

yards

is

exactly equal to

n metres, the question

settled, for a
itself to

yard must be
a

n
of a metre.
is

But

it

will

immediately suggest

mind which
it

accustomed
be no two

not to receive assumptions without inquiry, that

may

may be more
J.

There are two English editions of the xvhole of Euclid, and there that of John Dee (now old and very scarce) and that of
:

Williamson, London, 1788, in two thin quarto volumes. The dissertations in the latter are a strange mixture of good and bad, but the text
is

very

literally Euclid, in general.

32

CONNEXION OP
But
in this

points ever coincide on the multiple scale.

case

it

is

very soon proved

tiiat

mk. may be made

as nearly equal to

nB

as

we

please,

by properly finding

and n; so

that a fraction

m may
Even

be found such that

shall

be as nearly

m
in

as

we

please. ^
it

admitting that this would do to assign

terms of B,

leaves us

no method of establishing any definite connexion between sidered as a part of B, and P considered as a part of Q.

con-

The word

part

usually

means

arithnetical
3 Thus is

part,

namely, the
1

result of division into

equal parts.
parts,

a part of

made by
of
is

dividing 1

mio 7 equal

and taking 3 of them.

The phrase
is

Euclid in the books on number (VII. to X. both inclusive) part of

1,

that

\s

parts of

1.

And

it is

easily

shewn

that, in this

use of

the word, every quantity

is Q\\\iQ\
it.

part or parts of every other quantity

which
tities,

is

commensurable with
is

And

of two incommensurable quan-

neither

part or parts of the other.

is

But

in the original sense

of the word part, any less

always part of the greater.

This notion

of incommensurability, the non-existence of the equation

for

mA = wB,

any values of

or n, obliges us to have recourse to a negative

definition of proportionality, a term

Examine

the definition of a square, namely,

which we proceed to explain. "a foursided

plane
figure,

with four equal sides and one right angle.''

It is clear that the ex-

amination of a

finite

number

of questions will settle whether or no a

?

figure is a square.

Has
?

it

four sides

are they in the

same plane ?
affirm-

are the sides equal

is

one angle a right angle?

Proof of the

ative of these four propositions proves the figure to

be a square. Now, examine the number of ways in which a figure can be shewn to be

not a square.

is

or

is

All propositions are either affirmative or negative ; not B. The affirmative can be proved or the negative

disproved, with one result only, for both give

affirmative can

is

B.

But

the

be disproved, or the negative proved, with an infinite number of results ; it is done by proving that A is C, or D, or E, &c. &c. ad infinitum. Thus there may be an infinite number of ways of
shewing that a figure
shewing that
it is

is

not a square, but there

is

only one

way

of

a square.

This we

call

a positive definition.

Now

examine the definition of

parallel lines,

" those which

are in

NUMBER AND MAGNITUDE.

the

33

same plane, but being produced ever

so far do not meet."

We

are not considering where the lines meet, if they

tinguishing between lines which meet in

do meet, or disone point and in another, but

simply dividing

all

and

intersectors.

Now

possible pairs of lines into two classes, parallels * the affirmative here it is to

impossible

prove

of the proposition, "

finition only,

and

are parallels,"

by means of
relative

the de-

without proving an infinite number of cases.

To

see

this

more

clearly,

remember

that every proposition

is

to the

intersection or non-intersection of straight lines,

an assertion which

which can " Lines intersect" means there " " Lines are is a couple of such points which coincide. parallel means that there is no such couple whatsoever, of all the infinite
either includes or excludes every possible couple of points
straight line.

be taken, one on each

number which can be

taken.
existence

The
parallels

first

proposition in which Euclid proves the

of

(the 27th) does not

shew that the

lines are parallels,

is

but

that the proposition,

"

the lines are intersectors,"

inconsistent with

preceding
it

results.

The

"
proposition,
is

and

are parallels," though

appears affirmative, yet

in Euclid a negative, for his express

definition of parallels does not define

what they

are,

but what they are

not,

" not intersectors."

Now,

a negative definition. to examine further Euclid's definition of equal ratios,

call

This

we

we
has
,

must consider

his definition of greater

and

less ratios.

They amount

to the following. to

A
is,

is

said to have to
all

a greater ratio than

where there

among

possible whole

numbers

and

any
less

07ie

pair which give niA greater than

than

wQ;

or which
fact, in

mP equal to or wA equal to nB, but mF less than wQ give

nB, but
:

which give in
assertion.

And

is

any one case, what we have called a descending said to have to B a less ratio than P has to Q,

when any one pair of whole numbers

and n gives

mB,

but

mP

equal to or greater than

:

Q, or

mA less than mA equal to nB, but

mV

which give in fact, in any one case, what we greater than nQ have called an ascending assertion. Here, to a mind the least inquiOur notions of the terms sitive, appears at once a decided objection.

The
amounts

to another

celebrated axiom of Euclid evades this, and in point of fact and a positive definition of parallels, the assumption

being that the old definition agrees with it. Or rather we should say, that the first twenty-five propositions of the first book establish a part of
the connexion of the definitions, and the axiom assumes the rest.

34
greater and
tity, ratio,
less will

CONNEXION OF
never allow us to suppose that any thing, quan-

or any thing else, can be both greater and less than another

quantity, or ratio;

and

yet,

on looking

at the

definition of Euclid,

we
to
is

see that for any thing

which appears

to the contrary,

one pair of

values of

and n may shew that

A has
that
it

a greater ratio to

than

Q, while another pair

perfectly valid
;

may shew

has a

less.

The
it

objection

the only fault to be found

is,

that

should not

have arisen before, when the definitions of the

posed.

first

book were pro-

How

is

it

then

known

that there can

be such a thing as a

foursided figure with equal sides and one right angle, or as lines

which never meet ?

in the

The confusion The

arises

from placing the definitions

form of assertions, before the possibility of the assertions which

defect

they imply are proved.

may be remedied (we

To

take the

square as an instance) in two ways.

1.

Write

all definitions
it

in the following manner.

define a

" If square, for example,

be possible to construct a plane figure having four equal sides and one right angle, let that figure be called a square."
2.

Omit

the definition of a square, head the 46th proposition of

the

first

book as follows.

" Theorem.
and

On

a given straight line, a four-sided figure can be

all its

constructed which shall have

line,
all its

sides equal to the given straight

angles right angles."

:

Having demonstrated

this,

add

the following definition

Let the figure so constructed be called

sets of four

a square. have shewn that

We

all

magnitudes,

and

of

one kind,

P and Q

both of the same kind with the

classes.

first,

or both of

one other kind, can be divided into three

1.

Those in which simultaneous assertions on

wA

and w B, and

on

mP
2.
3.

and nQ, are

in in

all (for all values of

and n)

either ascending

or stationary.

Those Those

which they are

which they are

all stationary.

all either

descending or stationary.

For we have shewn

that the only remaining possible case a priori,

namely, that in which there are both ascending and descending assertions for different values of m and n, is a contradiction amounting
in feet to supposing

one fraction

to

be both greater and

all the

less than

another.
for

And

it

has been shewn that

three cases are possible,

are

comniensurable quantities at

least.

We

now,

therefore, in a

NUMBER AND MAGNITUDE.

condition to say, let
less ratio to

35

A
P

and

in the

first

case be said to have a

than

has to

Q;

in

the second, the

is,

same

ratio; in

the third, a less ratio.

The only question now

It will

are tliese definitions

properly negative or positive.

of the three, the
first

and

third

immediately appear that, out can be directly and affirmatively

shewn

to

be true of particular magnitudes, and that the second cannot.

is

By which
by a

single instance, establish the

meant, that the comparison of individual multiples may, first or third, but that no com-

parison of individual multiples, however extensive, can establish the

second.

For the second

first

consists in stationary assertions

ad

irifinitumj

and the

and

third are

proved by a single ascending or descending

assertion.

As an

instance, suppose
feet

A=

951

B=

1902 2853 3804 4755

In these
first five

497 994
1491

feet

1300

lbs.

Q=

1988 2485
Thus

2600 3900 5200 6500

679 lbs. 1358 2037

2716 3395

multiples, there are none but stationary assertions,

of twenty five which might be made.

4755 > 994 6500 >1 358

four magnitudes.

2853
1
j

> 2485 1 3900 > 3395

J
is

95 1< 994

1300 < 1358 j

third

^^

but neither of the three definitions

thereby shewn to belong to these

Now,

take the

first

second and fourth 952 times, and

of multiples,

we

498 times, and the have, going on with the series

and

473598 474549

473144
473641

647400

646408

648700

647087

and here the process may close, for we have 473598 less than 473641, while 647400 is greater than 647087. Consequently, we have
proved, by comparison, that 951 feet has to 497 feet a less ratio than

1300

lbs. to

679

lbs.

But

the case in which neither greater nor less ratio exists can never
in the

be established by actual comparison of multiples, except only

case where the pairs of magnitudes are commensurable.
that the

For, remark

mere circumstance of the

relative multiple scale of

and

36
agreeing with that of

CONNEXION OF
P and Q up to any point, is neither proof nor two magnitudes given are actually proportional,
it

presumption
though, as

that the

we

shall see,
if

is

certain evidence that they are nearly

scales

proportional,

the

multiple
is

agree

for

a great number of

multiples.

Proportion
is

not established until the similarity of the

multiple scales

shewn

to continue for ever.

not be remarked at
ditions to

first, this

insertion of

an

infinite

Now, though it would number of conif

be

fulfilled, is

tantamount to a negative definition,

we

wish to make the definition specifically speak of one absolute criterion of disproportion or proportion. Disproportion is where there
is

of the multiple scales.

an ascending or descending assertion somewhere in the comparison Proportion is where there is no descending In the case of commensurable quantities the definition

or ascending assertion.
is

positive,

because there

is

then a single stationary assertion, which, being proved,

all the rest are

shewn
if

to follow.

If

mA = wB;
tiple scales.

then

wP = nQ,
when

A and B be commensurable, let

is proportion; if not, there is proof as to the rest of the mul-

there

disproportion.

See page 24

for the

We

have

said, that,

the multiple scales agree for a long

period, there is proportion nearly;

that the scale of

and it is proved thus: Suppose and B agrees witii that of P and Q, up to 10,000 P and 10,000 Q, but that we have disagreement as follows 9326 A lies between 10,000 B and 10,001 B, whereas 9326 P lies between

10,001

and 10,002 Q.

Or

the scales run thus

10,000
10,000

9326

A
alter

10,001

10,002

Q
must we

10,001

9326

10,002
?

Q
Not

How much

A to

produce absolute proportion

more than would be necessary to make 9326 A greater than 10,002 B, That is, we must or less than would still keep it less than 10,001 B.
so alter

as to

add somewhere between

and

2B

to

9326 A, or

somewhere between
and

7-^B 9^26

to

A
B
to

Consequently, the addition of a small part of

accurate proportion.

would make an

We

might now proceed

to the propositions of the Fifth

Book of

NUMBER AND MAGNITUDE.

Euclid
;

37
the student's

but there are three

difficulties in the

way of

perfect satisfaction with the definition.

terious idea of incommensurables.

1st,

He may

have a mys-

the necessity of departing from

cult to imagine

2d, He may not be satisfied of arithmetic. 3d, He may find it diffi-

how

the existence of proportionals can ever be esta-

blished, with, apparently, an infinite

to satisfy.

number

of conditions of definition

We
is

suppose that the gravity of tone which elementary

inconsistent with the statement of a beginner's diffiin

writers adopt,
culties, in the

words

which he would express them.

Let

We

shall
it

remove

all necessity for

preserving such dignity in a case where

may be

inconvenient,

by a simple supposition.
;

^ be a beginner
mix pre-

in the stricter parts of mathematics

that

is,

a person apt to

viously acquired notions with the meaning he attaches to definitions

which are intended

to

exclude
;

all

but the ideas

literally

conveyed in

the words which are used

much

better pleased with the apparent

simplicity of an incorrect definition, gained either

by omitting what

should not be omitted, or by supposing what cannot be supposed, than with the comparatively cumbrous forms which provide for all
cases,

and distinguish

differences

which

really exist;

and, finally,

predisposed against, than in favour of, the Let be another person, who has subnecessity of demonstration. jected his mind to that sort of discipline which has a tendency to
exists, rather

when a doubt

remove the propensities abovementioned. talking together in this manner

:

We

can imagine them

A.

have been trying to understand the meaning of incommenall

surable quantities, and cannot at

make

out

how

it

can be that one

given line

may be no

fraction whatsoever of another given line,

little

though

both remain fixed, and certain lines ever so

the
first

greater or less than

are fractions of the second.

B.

A little

consideration will teach you, that neither in arithmetic

nor geometry are

we

at all

concerned with

how

things can be, but only

it

with whether they are or not.

that the

Do

you admit

to

be demonstrated
incom-

side
?

and diagonal of a square,

{Algebra, page 98).

for instance, are

mensurable

A.
hensible.

I cannot deny the demonstration, but the result

is

incompre-

Does

it

really prove, that if I

were

to cut the diagonal of a

square into ten equal parts, each of these again into ten equal parts, and so on for ever, I should never, by any number of subdivisions,

38

CONNEXION OF

which cuts

succeed in placing a point of subdivision exactly upon the point off a length equal to the side. B. I take it for granted you have sufficiently comprehended the
definitions of geometry, to

a canal of ink, are not geometrical lines

be aware that a thin rod of black lead, or and that the excavations ;

which you perforate by the compasses are not points. A. Certainly ; I now have no difficulty in imagining mere length intersected by partition marks, which are not themselves lengths. B.
Then,
;

in the case

you proposed, you need not go so

far for

your method of subdivision will never succeed in cutting off so simple a fraction as the third part of the diagonal. J. Why not?
a difficulty
for

B.

You

see that 9, 99, 999, &c., are

all divisible

by

3, so that

10, 100, 1000, &c., cannot in any case be divisible by 3, but must leave a remainder. Your method of subdivision can never put toIf possible, suppose gether any thing but tenths, hundredths, 8cc. one-third to be made up of tenths, a in number, added to hundredths, b in

number, added

to thousandths, c in

number. Then we must have

10

100

1000

Clear the second side of fractions, and

we have

^=
3 or
3
is

axlOO + bx^O + c
is

a whole number, which '

not true.

And

the

same

rea-

soning might be applied to any other case. A. This is conclusive enough ; but it seems to follow that the
third part of a line
is

incommensurable with the whole.

B.
propose

So
is

it is,

as far as the one

method of subdividing which you

concerned.

Let tenths, hundredths, &c., be the measurers^

and one-third and unity are incommensurable. But the word with which we set out implies all the possible subdivisions of halves, thirds,
fourths, fifths,

it

&c. &c., to be tried, and all to fail. is an infinite number of ways of subdividing. Can be possible that no one of them will give a side of a square, when

A.

But here

the di igonal

is

a unit
first

B.

In the

place,

it

would be a

sufficient

answer to

this sort

of difficulty to say, that, for any thing you

know

to the contrary, the

NUMBER ANB MAGNITUDE.

number
in

39
number of

of ways in which you

may

fail is

as infinite as the

which you may try to succeed. In the second place, there ways is also an infinite number of ways of subdividing, which will not give Let your first subdivision be into any number of equal one-third.
parts, except only 3, 6, 9, 12,

&c.; and your second subdivision the same, or any other, with the same exceptions, &c. The same reasoning will prove that you can never get one-third.

A.

But look

at the matter in this

thirds, the fourths,'; the fifths,

way. Suppose the halves, the &c. &c. of a diagonal laid down upon it

nd

infinitum, so that there

is

parts,

how many

soever, but

what

no method of subdividing into aliquot Would not is done and finished.

the whole line be then absolutely filled with subdivision points,

and

would not one of them cut

B.
applied

off a line equal to the side of the square.

You
it

have

now changed your

use of the word

infinite^

and

in the sense of infinity attained, not infinity unattainable.

to signify succession, which might be you pleased, and of which you were not obliged to make an end, the word was rational enough, though likely to be mis-

As long

as

you used the word

carried as far as

understood

but as

it is,

you may

as well suppose

you have got be-

yond

infinite space, at the rate of four miles

an hour, and are looking

back upon the infinite time which it took you to do it, as imagine that you have subdivided a line ad infinitum. But if the idea of infinity attained be a definite conception of your mind, you meet the
difficulty of

nition of the term

incommensurable quantities in another form. The defiincommensurable was shaped in accordance with

the exact notion, that, subdivide a line as far as

stop at some finite subdivision

you may, you must and incommensurable parts of a

whole are those which you never exactly separate arithmetically, stop at what finite subdivision you please. But, if you will contend for infinite subdivision attained, and imagine the line thus filled up by
points, then
it

will

be necessary
off"

to divide all parts of a

whole into two

classes, those

which are cut

by

finite

subdivision, and those which

;

are not attainable, except

to

by

infinite

subdivision

the former answering

parts.

commensurable, the

latter to

incommensurable,

The

diffi-

culty remains then just as before ; in other words, why should the side of a square be not attainable from its diagonal except by infinite subdivision, when the sides of a rectangle, which are as 3 to 4 (instead

of 3 to

3), are attainable by a finite number of subdivisions ? In the next place, you have spoken of a line filled up by points,

40
the infinitude of the

CONNEXION OF
number of points being
the compensation for each
;

of the points having no length whatsoever

see what else you can mean.

at least,

it is

not easy to

A.

Certainly that

is

what I mean

of algebra are in accordance with what I say.

into

and the common expressions For, if I cut aline

of the n parts

n equal

parts,

it is

plain that the

sum

makes up
suffi;

the whole, be the

number n

great or small.

But by making n

ciently great, each of the parts

may

be made as small as I please

and, therefore, allowing

when n

is infinite,

in all

be rational to say that P takes place cases in which we may come as near to P as
it

to

we

please,

by making

n. sufficiently
it

great (which

is

the expressed
say, that the
it is

meaning of
line is

infinite in algebra),

follows that

we may

made up

of the infinite

number

of points into which

cut

when divided
JB.

into

an

infinite

number

of equal parts.

I see every thing but the last consequence.

A.
it

Why,

surely, the smaller a line grows, the

more nearly does

approximate to a point.

B.

How

is

that proved

Suppose two points to approach each other, they continually inclose a length which is less and less, and finally vanishes altogether

A.

when

the

two points come

is,

to coincide in

one point.
is
it

So

that the

smaller the straight line

point.
JB.

the

more near

to its final state

You have

not kept strictly to your

in

own

idea (which

is

which the words nothing and infinite may way be legitimately used. You have supposed a line to be entirely made up of points, each of which has no length whatsoever, because you
correct one) of the

may compose
your

a line of a very large number of very small

is

lines,

each

of which, you say,

nearly a point.

Let us

now

consider whether

one to which we can approach as near as we diminution of a length. Any line, however small, can be please by divided into other lines by an infinite number of different points ; for
final supposition is

any

line,

that there

however small, admits of its halves, its thirds, &c. &c. So is a theorem which is not lessened in the numbers it speaks
or meaning, in
it ;

of, or altered in force

any the smallest degree, by

namely, any line whatsoever
please being laid
the point

diminishing the line supposed in

admits of as

many

difterent points as

we

down

in

it.

^ow
true

of your final length, or limit of length

this is

not

consequently, you throw away a result at the end, which you

NUMBER AND MAGNITUDE.

41

which you
less in the

cannot throw away as nearly as you please during the process by attain that end ; nor will the denial of it, near the end, be

consequence or amount of the error, than if the rejection were made further from the end. Therefore, in asserting that a diminishing straight line approximates to a point, you have abandoned
the condition under which
infinite.

you are allowed

to

speak of nothing or
certainly greater than

Again, the nth part of a line taken twice

the simple nth part, however great

is

n may

be.

Now, what do you

suppose two points to be, which are laid side by side without any interval of length between them ?

A.
B.

They are, of course, one and the same point. But in your infinite subdivision, two nth parts must be

greater

than one nth part, or two of your points must be greater than one ; but these two points are the same point, which is therefore twice as
great as
itself.

Such are the consequences

of points will lead.

to

which the supposition

infinite

of a line

made up
I

A.

have frequently heard of lines being divided into an

number of equal parts. JB. But you never heard those equal parts called points. I can soon shew you that, in the mode of allowing infinity to be spoken of, this fundamental condition is preserved, namely, that no
theorem, limitation, number, nor other idea whatsoever, which forms
a part of any question,
infinite, unless it
is

allowed to be rejected or modified when n

is

may
is

can be shewn that such rejection or modification be made with little error when n is great, with less error when n

greater,

and so on

finally,

with as small an error as

we

please,

by

making n sufficiently great. Now, remark the following truths, and the form of speech which accompanies them, when n is supposed
infinite.

General Theorem.
The
greater the

Terminal Theorem.
If a straight line be divided
into an infinite
parts,

number of

divided, the less line

the parts
:

equal parts into which a line is is each of

so that an aliquot

number
is

of equal

each part

an

infinitely

small line.

part of any line, however great,

may be made

less

than

any
E 2

given line, however small.

42

CONNEXION OF
General Theorem.
Terminal Theorem.

Any may be
as

line,

however

small,

An
please.

infinitely small line

may
we

cut by as

many

points

be cut by as

many

points as

we

please.
straight
line,

No

however

An
points.

infinitely

small

straight

small, ceases to be a length ter-

line is a length terminated

by

minated by points.

Now, taking your notion of infinite subdivision attained, it may be shewn that incommensurable parts necessarily follow. For, however far you carry the subdivision, you do not, by means of the subdivision points, lessen the

number

of points which

may be

laid

down.
infinite

For each

interval defined

by the subdivisions contains an

if

number of
number

points.

Consequently,

subdivision attained, you cannot do

of points

left

will suppose the infinite without supposing an infinite in the intervals, or an infinite number of in-

you

it

commensurable

quantities.

This I intend only to shew that the

proof of the existence of incommensurable quantities is, upon your own supposition, somewhat better than that of their non-existence.

But

it

would be

better to use

nothing and infinity as convenient

phrases of abbreviation, not as containing definite conceptions which

may be employed
A.
if

in

demonstration.

do not see how your objection applies against nothing;

attain infinity

we cannot
B.
So

tainly attain nothing

it

may

by continual augmentation, we can cerby continual diminution. seem at first, and in truth you are right as to one
is
is

sort

of diminution, that which

the place in

implied in the word subtraction.

From

something take away all there is, and you get nothing by a legitimate process. But subtraction is the only process which leaves nothing ; division, for example, never leaves

which thsre

it.

Halve a quantity, take the

half of the half,

and so on, ad infinitum:

that

you

will never reduce the result to nothing.

A.

But however
actually

clearly

you may shew

incommensurable

quantities
finitions

exist, as

a necessary consequence of our debetter

satisfied
if

of length, number, &c., I should feel

you could give something like an account of the way in which

they arise.

B.

If

you

will consider the

are conceived, perhaps the difficulty

way in which number and length may be somewhat lessened. Let

NUMBER AND MAGNITUDE.

a point
set

43

out from another point, and

move uniformly along a

other.
It is

straight line until the

two are a foot distant from each

between

clear that every possible length

and one

foot will

have

been
a

in existence at

some part

or other of the motion.

Now, suppose
first

number of points

as great as

you

off from the please, to set

point together; but, instead of

moving

in the straight line let

them
1

move

off in curves, the first

coming
1
;

to the straight line at

- and
1

of a foot: the second at

foot

12 - - and
3

the third at

3 - - - and 4 4 4

12

of a

and so on, as in

this

diagram.

Can you
will ever

feel

sure

that

these

contacts

of curves

with the

line,

separated as they must always be from each other by

fill

finite intervals,

up

the

whole

line described

by a continuous motion.
be increased to certainty this should be sufficient

If not, this figure will always supply presumption in favour of in-

commensurable
by

parts,

which

will of course

the actual proof of their existence.

And

to overturn a doubt which after all is derived from confounding the mathematical point with the excavation made by the points of a pair

of compasses.

whole

The practical commensurability of all parts with the a consequence of there being magnitudes of all sorts below the limits of perception of the senses (see page 3). A. Granting, then, that there are such things as incommensurable
is

quantities,

yet

on

is admitted, that though A and B are incommensurable, A and B K may be made commensurable, though be insisted that K shall be less than any given quantity, say less than the
it
-|-

it

hundred thousand million millionth of the smallest quantity which the senses could perceive, if they were a hundred thousand million
of million of times keener than they are at present.
sufficient,

Would

it

not be
to

when incommensurable

quantities,

and B, occur,

suppose so slight an alteration made in B as is implied in the above, and reason upon A and B K so obtained, instead of upon and B. Surely such a change could never produce any error which would be

of any consequence
JB.

Of consequence

to

what?

44
A.
useful.
JB.

CONNEXION OF
To any purpose
I

of

life for

which mathematics can be made

am

still at

a loss.

A.
would be

What

process in astronomy, optics, mechanics, engineering,

life,

manufactures, or any other part either of physics or the arts of

vitiated

by such an

alteration, or its consequences, to

any

extent which could be perceived, were the error multiplied a million fold?
JB.

None whatever,

that I

know

of.

What, then, would be the harm of introducing a supposition which would save much trouble, and do no mi^jchief ?
B.
I

A.

am

not aware that I admitted such a supposition would do

I said that it

no mischief, when
I see that

would not sensibly

is

vitiate the appli-

cation of mathematics to

what are commonly called the

very
If they

arts

of

life.

your idea of mathematics

his tools.

shoemaker has of

much like that which a make shoes which keep the

weather out, and bring customers, he need not wish them to do more, or inquire further into any use, actual or possible, which they may or

might have. The end he proposes has sewed the upper leather firmly
serves

to himself is

answered, when he

to the sole.

But whether

his art

whether the possibility of obtaining conany higher purpose veniencies, and avoiding hardships (which it creates in one respect),
excites

industry and

ingenuity,

creates

property

to

equalise

the

and commerce, and prevent the community makes men so defrom undergoing periodical pests and famines pendent on each other that internal war is next to impossible, and
fluctuations of harvests

external

war a grave and serious consideration, &c. &c., are not

;

matters for the thoughts of a working shoemaker

considerations ever enter the

nor will similar

mind of a working mathematician. You have spoken of the purposes of life ; I do not know what the purposes of your life may be, but if among them you count such a discipline

may always render your perception of the force of an argument properly dependent upon the probability of the premises, and the method by which the inferences are drawn, it will be one of your first wishes to propose to yourself, as a standard and a model,
of the mind as

some branch of study

as

in

which the

first

are self-evident, or as evident

any thing can be, and the second indisputable and undisputed. For though you may find no other science which will compete with this in accuracy, yet you will be more likely to infer correctly, when

NUMBER AND MAGNITUDE.

45

you have seen what you know to be correct inference, than you would have been if you had never, in any case, distinguished between demonstration of certainties and presumptions from probabilities. And
still

that

more, will you be qualified to refute, and refuse admission to, which takes the form of accuracy without the reality. If the

better as a shield.

mathematical sciences be good as a weapon, they are a hundred fold I have seen many who were visibly little the

better for their mathematical studies in

what they advanced ; but very

few indeed who were not made sensibly more cautious in what they
received.

A.

But

is

not

my

notion adopted in practice by a great part of

the mathematical world, particularly

JB.

on the continent.
particularly the case with the

It is certainly true,

and

it is

French, who, though they have done more than any other nation, since the time of Newton, to advance the mathematical sciences, have been

by no means anxious
than that

to consider

them as

resting

on other evidence

not of the senses

but of the limits of the senses.

One

of their most celebrated elementary writers considers none but arithmetical proportion, and begins his work by shewing either that two

have a common measure, as in page 12, or that the remainder " echappe aux sens par sa petitesse." All his propositions,
straight lines

therefore, in geometry^ are

either true, or so nearly true, that the

difference

is

imperceptible.
;

The phrase we have quoted

it

is

an honest

and a valuable admission

shews you, that in the opinion of one

of the most useful and extensive elementary writers that ever lived,
arithmetical proportion

makes geometry a science of approximate,

slight shifting of

not absolute, truth.

A.
quantities

I see as

which

much; but cannot the proposed be somehow

one of the

or other corrected, so as to

make a
jB.

strict
?

and useful theory of the proportions of incommensurable

quantities

of Euclid.

Yes, and in a very simple way; by adopting the definition This may surprise you, but I will soon shew that the
correction of your notion leads direct to the definition of

most natural
Euclid.

it be granted that A and B being commensurable, and ?nA=wB, proportion between A, B, P, and Q means that mP=nQ. Now you want, when A and B are incommensurable, to be allowed to substitute B + K instead of B, where K is I excessively small.

Let

suppose you would be perfectly content

if it

could not be made

46
visible

CONNEXION OF
by any microscope. and should not like

Now

am

of a

somewhat more

abstract

geometry to be put in peril by the abolition of the excise on glass ; which it might be by the allowance of experiments for the improvement of that article, which are now
turn,

my

effectually prevented.

I cannot
is

admit

B4-K, where

the magnitude

want

to reason

upon
is

B.

But

as the definition of proportion of

let

incommensurables

not yet settled,

let

us examine this case

A and
and
it

being incommensurable,

P and

Q be

quantities of such a kind

also

that

and B-f

are commensurable,

and

P and

Q + Z,
Let

that the four just

named

are arithmetically proportionals.

be

possible, these conditions subsisting, to

make

and Z as small as
small quantity, but

we

please

not as small as

this, that, or the other

smaller than any whatsoever which

quantities.

I prefer to

may be named, being still some You wish to substitute B + K and Q -|- Z for B and Q use the conditions laid down to ascertain how B and Q
:

themselves stand related to

A and

P.

small magnitudes, K' and Z', of the same kind as

Let us suppose we name two A and P, or B and

Q,

or

please.
that

We

and Z, which we are at liberty to make as small as we can then find K and Z less than K' and Z', and such

A, B K, P, Q -f- Z are proportional. commensurable, and let

Suppose

and

B+K

mA =
whence
it is

n(B + K)
B+K

whence

mP =
as that of

W(Q +
P
is

Z)

easily proved, as in

page 24, that the

relative scale of

multiples of

and

is

the

same

and
the

Q + Z.

say

it

follows, that the relative scale of

;

A and B
uP
less

same

as that

of

P and Q
it

for, if not, the

is

two

latter scales

must

differ

somewhere.

Let

be that i?A

greater than lyB, but

than w?Q.
(it

Then,
as

since

vA

is

greater than

wB,

let

be taken so small

may be

exceed io(B +K), whence, by the proportion assumed in the hypothesis, vP exceeds it>(Q Z), This is while, by the hypothesis we are trying, vP is less than t^Q.
small as
please) that
shall also

we

vA

a contradiction,
at the

for v

cannot exceed

wQ-^wZ

and

fall

short oi

wQ

same

and

it

In the same way, any other case may be treated ; follows that our suppositions, if may be as small as we
time.

please;

amount
If,

lows.

an hypothesis from which Euclid's definition folin the above, we suppose vA less than wB, while uP is
to

greater than

wQ, we

see that

vA

also falls short of v(B -f-K),

whence,

NUMBER AND MAGNITUDE.

by
the proportion,

47
if it

vP

falls short

of

v(Q4-Z), which cannot be

exceed v Q.

A,

But

is

not

this

deduction, namely, Euclid's definition, more

it

cumbrous than
JB.-

the form from which

has just been deduced?

Howso?
Does
it

A.
B.

not involve an infinite

number of considerations, ex?

tending the whole length of the multiple scales

does not your definition do the same thing, unless you and Z ? Is it not necessary, if stop somewhere with the values of we would not be merely microscopically correct, but absolutely cor-

And

and Z may be diminished and diminished ad suppose that ? And what difference is there, as to the number of coninfinitum
rect, to

siderations in question, between

two magnitudes which are

to

diminish

without limit, and a set of increasing multiples of two given magnitudes

?

A.

But Euclid's

definition

seems

to

quantities in question, while the other remains close to them,

wander such a way from the and we

never seem to quit them, except for something very near to them. The actual application of the definition I prefer will require nothing

but the division of

all

magnitudes into aliquot

;

parts.

B.

Your

objection amounts to this

that

you

feel the fractions

of a quantity to be more closely connected in your

mind with

the

quantity
is

itself

than

its

multiples.

This

some reason
there

for preferring the

form to

may be the case ; and, if so, it which you seem most inclined.

a stronger reason for preferring the other ; and, undoubtedly, as long as difficulties exist, every system of science must be a balance of inconveniences. But Euclid is, of all men who ever
wrote, the one

But

may be

who

has a reason for the course which he takes, where

I suppose

there are two or more.

you cannot but admit

that

it

is

better to found a definition in geometry

upon

the result of something

which can actually be done, 63/ the means of geometry y than upon something wliich can only be conceived or imagined to be done, with
to use other

what certainty soever ; for instance, you would not wish to be obliged means than the straight line and circle, or to suppose an
object gained without using any

means

at all

A.
chanical

Certainly not.

That there
that of

shall be

no assumption of meis

power beyond

drawing a

straight line or circle,

the

foundation of pure geometry.

B.

Then

the question

is

settled in favour of Euclid's definition

48
for,

CONNEXION OF

without either assuming more mechanical means, or making a gratuitous assumption, no angle, nor arc, nor sector of a circle, can be
divided into 3, or 6, or 9, &c. parts, unless
it

given half, fourth, eighth, &c. of a right angle.

exceptions

be a right angle, or a There are some other

; but, generally, to cut any angle into three equal parts is a geometrical impossibility, and certain algebraical considerations furaish the highest presumption that it will always remain so.

A.

But

this

difficulty is still left:

how

are

we
?

ever to

shew

that there are such things as proportional quantities

of the process

stumbling-block being so easy and perceptible, that a beginner does not very well see where lies the knowledge he has gained, unless
lies in its

B.

We

can do

this so easily, that the greatest

the

he has paid profitable attention to the definition of proportion. From first book of Euclid it is evident that a rectangle is doubled by
doubling the base, trebled by trebling it, and so on ; and also, that of two rectangles between the same parallels, the greater base belongs to the greater, and the lesser base to the lesser. Now, let B and B'
represent two bases, and

and R' the rectangles upon them, the

If then

altitudes, or distances of the parallels, being the same.

we

take the
is

first

base

times, giving

m B,

the rectangle

upon

that base

mR

if

we

take the second base n times, giving wB', the rectangle

upon that base is wR', the parallels always remaining the same. Hence it follows, that twB and nB' are bases to the rectangles m'R
and wR' between the same parallels; accordingly, is greater than, equal to, or less than n B', so is
equal
to,
it

therefore, as

wiB

mR
all

greater than,

or less than n R'

follows that

and

this

being true for

values of

and

n,

has to B' the ratio of

to R', or the bases

of rectangles between the same parallels, and the rectangles themselves are proportionals.

A.

Am

I to understand then that there are difficulties in the

general, which are not found in

way

of considering magnitude in

arithmetic, or the science of abstract

JB.

number ?
which the
ratio of

Quite the reverse

itself,

the difficulty arises from the deficiences

their being ratios

of arithmetic

and from

number
A.

to

number cannot
is

represent.

But how

that? arithmetic always

considering.

seemed

clear of such

difficulties as
JB.

we have been
so

And
is

would

this subject, if the disposition to

is

be

satisfied

with what

in the

book, which

part and parcel of almost every

NUMBER AND MAGNITUDE.

49

mensurable

beginner had been permitted to rest quietly upon a theory of comBut did you never, in arithmetic, hear of the ratios.

creation of a nonexistent

number

or fraction, in spite of there being

no such

thing,

by agreeing

that there should

be such a thing, and

drawing a picture to represent it?

I do not understand the jest; but I suppose you allude and to quantities less than nothing ? algebra, B. Not at all ; 1 am speaking of pure arithmetic. To me,
is

A.

to

much
2

easier symbol, or picture, than

lie

n/2; and even the

.

difficulties

of \/

as
I

much

in the

as in the

do not understand what you mean by saying that ^2 does not exist; it is the square root of 2, and multiplied by itself it A.

But

gives 2.

You may
If
it

find

it

as nearly as

you

please.
it
is

B.

be the object of arithmetic, commonly so called,

either a whole

number

or a fraction.

Which

of these

is it ?

A. B.

It is a fraction; I did not It

1-4142136, very nearly.

it is

ask you what

veri/ nearly,

but what

it is ?

cannot be given exactly, but we all know there is suclva thing as the square root of 2. B. If the objects of arithmetic were numbers, fractions, and

A.

things,

and the

latter

say.

And

in

concrete
I

term had a definition, I might admit what you arithmetic, where 1 is a things a foot, a
is

pound, or an acre,
that thing
is

admit that there

such a thing as \/2.

But

not attainable arithmetically by taking any aliquot part

it

of the thing 1, and repeating

arithmetic the square root of 2

existence, I
it

is

any number of times. In abstract an impossibility ; and having no

be said
to

do not see how one

fraction can

be nearer

to

than another, except in this sense, that 2-\-z may be made to have a square root where z may be less than any fraction we name. The

independent existence of
difficulty;

v2

is

an algebraical consideration of some

that

is,

belongs to the science which has relations of

its

symbols, under prescribed definitions, for

to their numerical interpretation.

object, without reference

The

difficulties

of \/2 are precisely

is

those of incommensurable magnitudes; in fact \/2

the diagonal

of a square whose side

this

is 1

But

it

is

to algebra that difficulties of

kind should be referred.

The

student, if he use s/2 in pure

arithmetic, must expressly understand it as a fraction whose square is nearly 2, and must consider this part of arithmetic (without algebra

50

CONNEXION OF
some other
science

as a science of approximation, unless geometry, or

of concrete quantity, be supposed to lend

its

aid.

A.

But

cannot divest myself of the idea that

^2,

n/S? and
first

n/6 are really fractions, and that the product of the two I suppose, in some sense or other, you admit the last.
position
?

gives

this

pro-

to exist,

If ^/2-^x and n/s-j-j/ and '^6-\-z be made Certainly. by giving proper values to x, 3/, and z^ which may all be as small as I please, and if, moreover, j:, y, and z be so related that

B.

zz=:3x-\-2y }- xi/j which condition does not I can then admit that

interfere with the last

V2+ar X
But
I

VS-f?/

\/6+2r
times or parts

do not allow myself

to

suppose that (understanding by multior fraction as

plication the taking of one

number

many

of times as there are units or fractions of a unit in another), there can be such a truth as that

\/2

(neither

number nor

fraction) ;7?w////)/i>d

by s/ S (do. do.)

n/6

(do.

do

But

this is

difficulty lies

beyond our subject, except so far as it shews that the more in arithmetical than in geometrical considerations.

A.

make

Might we not then dispense with arithmetic altogether, and a definition corresponding to proportion for geometry?
Yes; but the difficulty would appear in another shape, of same substance. Let four lines be called proportional when,
without alteration
the
first

B.
the very

being straightened
rectangle

of length, if necessary,
is

the

made by

and fourth

equal to that

made by

the

Let areas be proportional when, being converted into rectangles with a common altitude, their bases are proportional. Let angles be proportional, when they are angles at the centre of
proportional arcs of the same circle.
arise this difficulty,

second and third.

But here would immediately

to

make

a straight line equal to a given arc of

a circle

which

is

out of the power of the geometry of straight lines

and

circles.

A.

Is not the reductio

ad ahmrdum (which

is

very

much used
and
if
it it

in the establishment of the theory of proportion) rather a suspected

method.

I have heard

it

called indirect demonstration

to

is

frequently stated as a defective method, not

be used

can

possibly be avoided.

NUMBER AND MAGNITUDE.

JB.

51

The complaints

against this

method of demonstration have

become much more

since the time

frequent, if not entirely

logic
I

made

their

appearance,

when

was a necessary part of a hope

it

liberal education,

as

it

once was, and as

this

will

be again.

I have sometimes

argument would have been considered objec" A is tionable if it had been reduced to the form B, B is C, therefore A is C ;" as follows " Every contradiction of P is a contradiction of the proposition that the whole is greater than its part ; but every

wondered whether

contradiction of this proposition

is false

therefore every contradiction

of

is false

or

is

true."

The

reductio

ad ahsurdum
x)ther

is

as conclu-

sive,
if

argument. And any be good in proportion to the effect upon the mind, any argument where is the affirmative proposition, in geometry or not, which the mind seizes as readily as it recoils from an absolute contradiction in

and may be made

as intelligible, as

terms

Where
is

is

the likeness or resemblance between things

which

are alike, that

so forcible as the unlikeness or

want of resemblance

of two ideas which palpably contradict, such as black is white ? A. Is there then no advantage in the direct over the indirect
demonstrations
?

B.

D'Alembert has said that the former are

to

be preferred

"
is

parce qu'elles ^clairent en

a good description

m^me-temps qu'elles convainquent," which of the difference. But even this must be taken

with some allowance, for there are

many

indirect demonstrations

which are highly

instructive.

Recapitulation.

By

the ratio of

to

further specification at present) a relation

B, we mean (without any between the magnitudes of

A and

B, determined by the manner in which the multiples of distributed, if each be written between the nearest multiples of

A are
B
in

magnitude. That is, if B, 2 B, 3B, &c., be formed, and A, 2 A, 3 A, &c., and if A lie between B and 2B, 2 between SB and 4B,

and so on, the

relative scale

B,

A,

2B,

3B,

2 A,

4B,

&c.

is to be the sole determining element of the ratio, so that there is to be nothing but the order of this scale on which the ratio depends. And if P and Q be two other magnitudes with the same order in their scale, P compared with A, and Q with B, then A and B are to

be said

to

have the same

ratio as

and Q.

But

if

any multi^

le of

52
precede

CONNEXION OF
among
the multiples of

the place which the corresponding

multiple of

occupies

among

the multiples of

said to have to

a less ratio than

has to Q.

Q, then A is to be But if a multiple of

its

come

later in the series

of multiples of

than

corresponding
said to have to

multiple of

in the series of multiples of

Q, then

is

B a greater ratio than P has to Q. It is plain that the ratio B must be greater than, equal to, or less than that of P to also, that in saying A is to B as P to Q, we also say that B
as

of

A to A
do

Q, and
is to

to P.

[We

must remind the student

that

we have now
is

nothing to
its

with the reasons of this definition, or the accordance of

each other, or with any notion of ratio more than

parts with contained in it.

We

are

merely

now concerned

to

know what

follows

from
is

this

defin tion.

The numbering of

the following propositions

that in

Euclid.]

When

A has

to

the

same

ratio as
:

to

Q,

the four are said to

be proportionals, and are written thus

A
which
is

B:: P
is

Q
|

read

A is to B a& P

bo

Q.
:

IV.

If

: :

Q
:

Then
This

mk

TwBr.wP
see that

TZQ)

and w being any whole numbers.

we know when we

order of magnitude, so will be their multiples.

any quantities being arranged in If the scales be

A
P

Q
the following scales

2B 2Q

3B 3Q

.....^ .....^

jwB

mA
wP

2mB

3r/zB

......

wQ
will also

2Q
to

3nQ
magnitude.

be arranged according

Whence

the pro-

position.

VII. IfA, B, C, be three homogeneous magnitudes

all

(all

lines, or

weights, &:c.)and

if

A=B
:

then
::

A
and
for the scales

::

B C

C
Bj

must evidently be

identical,

NUMBER AND MAGNITUDE.

VIII.
to

53

A+M has a greater ratio to B than A has B, and B has A+M than B has to A. Let M be multiplied so many times that exceeds B say wM = B + K then m(A + M) = mA + B + K
a less ratio to
it
;
:

Let

wA

lie

between
(t;

vB

and (u-f-l)

then

w(A + M)

lies

between

uB
of

+ B + Kand
-j-

+ l)B + B + K, and

Consequendy,

in the scales of
to

A+M

beyond (vH-l)B. and B, and A and B, a multiple

certainly

be in a higher place among the multiples of B than the same multiple of A among the multiples of B. Whence, by

is

found

definition,

A+M has to B a greater

is

ratio than

A to

B.

The second
first,

part of the proposition

but another way of stating the

as

appears from definition. Thus less ratio than A has to B.

we may

also say that

has to

IX.
or if

If

+M A C::B
:

C
(viii.), if

A::C
If

C B

then

then

A=B A=B
to

For

A
is

be greater than B,
not true.
is

A has
:

to

C, which

A be

less than

a greater ratio than C, A has to C a less

ratio than

to

C, which

not true

therefore

A=B.

The same

reasoning proves the second case.

X. If

A
if

have to
For

C
if

a greater ratio than

has to C, then
ratios

A
C

is

greater than B.
the

were equal

same:

A
C

were

less than

B, then these (viir.), then would

to

would be
a

have to

less ratio than

if

has to C.

Therefore, A is

B A A than to B, A is less than B a less ratio to A than to B, A is greater than B. XI. If the ratios of C to D and of E to F, be severally as that of A to B, then C has to D the same ratio as E to

A
C

have

to

a less ratio than

ratio to

greater than B. Similarly, And is less than B. has to C,

;

if

have a greater

if

have

the

same
This

F.

answers to a case of the general axiom, that two things which are perfectly like to a third in any respect, are perfectly like each other
in that respect.

The multiples of C
as those of

are distributed

among

those of

D
of

in the

same manner
those of F.

among

those of B, as are ihose

E among

Therefore, the multiples of

are distributed

among

those of

as are those of

E among
as

those of F.

Whence

the

proposition.

XII. If
so
is

A be

to

as

to

D, and
r 2

to F, then as

is

to

A+C + E to B+C+F.

54

CONNEXION OP
For mk. lying between nB and (w-{-l)B, then mC lies between and (w l)D, and wE between nY and (n-|-l)F, and, con-

nD

sequently,

wAH-wC4-mE,or2 (A+C-fE)
Whence
the

between n (B-|-D+F)
has to

and (w+1) (B+D-f-F).

XIII. If

the proposition.
ratio as

A have

to

B
to

same

D, but C

to

a greater ratio than

F, then

has to

a greater ratio than

to

F.
is

This

general theorem

one of a class of propositions which come under this for any ratio, an equal ratio may be substituted,
:

and

consequences of the first ratio are consequences of the second. This, which seems very evident, may appear so upon mistaken eviall

dence.

Ratios, as far as

we have

yet gone, are not quantities, but

expressions of that relation between quantities

upon which

the order

of magnitude of their multiples depends.

substitute other quantity equal to the
relation
stitute
is
first

For quantity, we

may

in

magnitude wherever the

one which depends only on quantity;

we may

not sub-

a triangle of the same area instead of a square, except there be

Ratio, again,

question of nothing but superficial magnitude, or area.

is to

us at present the order of the multiples, so that

if

and

have

their multiples arranged

have the same,

among we may say

is

each other in a given order,

if

P and

that whatever is true of the order of

multiples of

and B,

also true of the order of

and

whatever

connexion the order of multiples of

and

establishes between

and
and

and other magnitudes, the same connexion exists between P and those other magnitudes, because the accident of A and B,
is

which
is

the sole connexion

between them and the consequence hiferred,

also

an accident of

and Q.
its

considerations, arises from

The necessity for going over such never being allowed to be taken for

granted that a mathematician has studied logic.

frequently obliged to reiterate the

Hence Euclid*

is

same

assertions in different forms.

;

To

take the proof of the present proposition

to

say that

has to

Da

greater ratio

than

to

F,

is

to say that

mC

can be found greater than

Euclid was a contemporary of Aristotle, as is generally supposed, and may, therefore, never have seen the science of the latter. It is free to us to suppose that if he had, he would have distinguished between a
purely logical and a geometrical consequence that reiterated the same proposition iu different forms ;
:

is,

would not have

you please,
:

or, if

different cases of the

same verbal

truth as if they

were

distinct truths

and we

will

suppose so accordingly.

NUMBER AND MAGNITUDE.

nD,

55

while m E is equal to or less than wF. But to say that A has to B the ratio of C to D, is to say that whenever wC is greater than nD, wA is greater than wB. Therefore, to say that the ratios of A and B and C and D are the same, but the latter greater than that of E to F,
is to

say that

mA

may be

greater than

nB, while

mE

is

equal to or
to F.

less

than

nF

or that

has to

a greater ratio than

E has
B
is

Now,

let the

student compare this with the following proposition.

A and B

are greens of exactly the

same shade

but

a darker
it

green than C, therefore,

A
?

is

a darker green than C.

it is

Would

be

unnecessary to prove this

preceding.

then

equally unnecessary to prove the

But we

the preceding.

will prove this in the same manner as we prove Let there be a test of greenness, which decides between
is

two greens
darker.

(there

test

of comparison of ratios in Euclid), and

apply the test to

and C.

The

result

is,

of course, that

is

the

But

A being by

hypothesis exactly the

it

same

as

B, the testing

operation would be

self contradictory if

did not exhibit, when

and C, the very same intermediate process by which we were able to compare B and C, with the same result. If the
applied to

above^be unnecessary, then the demonstration of Euclid's proposition

is

unnecessary.

The
is

fact is, that there are in

first

geometry two distinct

sorts of de-

monstration, the

of which

is

only a portion of the second.

The first
terms

the verbal treatment of the terms of an hypothesis,

all assertions

and the devein the

lopement of
of the

which are necessarily included

other

proposition, without drawing upon any

for evidence.

axioms or

theorems

make two
"

assertions

It is the purely logical process, by which we put together shew their joint meaning, and

express what, without deduction,

they

only imply.
it

Thus, from

Every

A is

" no B," and

is

C," we make

evident that in these

assertions is necessarily contained a third, that

it

" no

A
C

is

C."

Thus
more

has been shewn that

we cannot allow simultaneous

is to

existence to the
is

two propositions, "

than

as

is

to

" D," and

to

F," without almost expressing, and certainly implying, " A by the mere meaning of our terms, this third proposition, that
is

to

is to

more than

is to is

F."
that in

The second process

which the demonstration, besides

the purely logical process of extracting implied meanings out of the

expressions of the hypothesis, appeals to propositions which are not in the hypothesis, and which, for any thing the hypotheses tell us to

56
th

CONNEXION OF
contrary,

may

or

may

not be true.

Of course

not logic, but

reason requires that these propositions should have been previously

proved, or assumed on their

following proposition,

own

evidence expressly.

Let us take the

" The sum of

triangle

the circles described

upon the

equal to the circle described upon the hypothenuse." Now, take every notion implied in this " Let there be a hypothesis, right-angled triangle, and let circles be
is

two sides of a right-angled

described on

its

three sides."

The

united faculties of

man

never
to the

proved that the sum of the

circle

circles

on the

sides

was equal

that only one parallel can

line
;

on the hypothenuse, without assuming with Euclid, to the effect be drawn through a point to a given right
is

with Archimedes, to the effect that the chord of a curve

its

shorter

&c. &c.; and various consequences. But are any of these propositions necessary to our complete definition of a right angle, a If not, we have a broad and easily recognised triangle, or a circle ?
than
arc,

distinction

between the

first

and second method of demonstration

an operation of logic, or deduction from the premises of the hypothesis ; the second, introducing premises from without.
the
first,

There are two classes of reasoners whose ideas we recommend

the student closely to examine, before he finally decides: 1.
trical writers in

Geome-

general,
let

who pay no
the
first
it

attention to the
the
fifth

methods which

they are using, but

book and

book of Euclid

contain no difference by which

may be remarked
the engine

that the processes

contained in the two are different acts

think that geometry could be

of mind.

Did they

ever

made

by which the student

could examine certain operations of his own faculties, or did they only imagine that it was a method of making very sure that squares,
circles,

sical writers,

&c. had such and such properties? 2. The class of metaphywho express themselves to the effect that all mathemacontained in the definitions and axioms, in a

tical propositions are

sense in which other results of reasoning are not.

Put them

to the

proof of this assertion as to geometry, and then as to arithmetic.

The whole of
is,

the process in the

fifth

book

is

purely logical, that

the whole of the results are virtually contained in the definitions,

in the

manner and sense

all

in

imagine
truth of

the

results

of mathematics

which metaphysicians (certain of them) to be contained in their

assumption
is

definitions

and hypotheses.

No

made

to determine the

any consequence of this definition, which takes for granted more about number or magnitude than is necessary to understand the

NtJMBEll
definition itself.

AND MAGNITUDE.
being once understood,
its

57
results are

The

latter

deduced by inspection
at

of itself only, without the necessity of looking

fifth

any thing

else.

the preceding books presents

Hence, a great distinction between the itself. The first four are a
;

and
of
is,

series

propositions, resting on different fundamental assumptions

that

about different kinds of magnitudes. The fifth is a definition and its developement; and if the analogy by which names have been given
in the preceding

books had been attended

corollaries
is to

to, the

propositions of that

book would have been called

of

the definition.

XIV.
then
if

If

A be to B

as

C
C,

D,

all

four being of the

A be

greater than

is

greater than

if equal,

same kind, equal, and

if less, less.

must

either

be

wA
C,

is

greater than

mC

>

= or < C.
Let

Let
lie

A be

greater than

mK

between

C then nBand(w+l)B;
;

then will

mQ

lie

between

wD

and {n-\-\)T).

But because

A exceeds

2A

exceeds

2C
;

times as

much

by twice as much, &c., and mK exceeds wC by or rnK may be made to exceed mC by a quantity
named, say greater than
for (to-j-I)D
fall

greater than any one

and

together.

Then

the order of

magnitude of the four multiples

mC
by

(n+l)D, nB,

wA must
much
while
fore,

be as written:

does not exceed

mC
much

by so
as B,

as

D, and
is

wB

does not

short

ofwA

as

wA exceeds mChy more

greater than

than

and

nB

(n+l)D, and

still

put together. Theremore than nT>. That is,

is

greater than

D.

Let be equal to C. If B exceed exceed mT> by more than D, or twB

D at all,
may
That

jB may be made
is,

to

be made, from and

the order of

after

some value oim,

nitude

greater than

(m -f 1) D.

mag'

may be made

mJ)

(m + l)B

mB

(m+l)B

Having gone so far on the scales that this order becomes per^ manent, go on till a multiple ofC (/cC) falls between the two first. Then, by the definition, kK falls between the two last, which is absurd
;

for,

because

A = C,

/cA

= A;C;

therefore,

B does

not exceed
short of

D.
D.

In the same way

Therefore,

it

B = D.

may be shewn
case

that

does not

fall

The remaining

(A

less

than C)
as

may be

proved like the

first.

XV,

is to

mK

is to

mB

58
The
scale of multiples of

CONNEXION OF

A and B

is

nowhere

altered in the order ot

magnitude by multiplying every terra by w. If p A lie between qB and (^ 1)B, {p}n)A which is p{m A) lies between q{mB) and
(y

+ + l)(mB).
XVI.
If
four
if all

be

to

B
C

as

C
B

is to

D
D.

and

be of the same kind,

Then
(iv.)

is

to

as
as

is to

wA
If
;

isto

mB

nC

isto
is

nD
nD,
if

(xiv.)

;A

be greater than nC,

Therefore,

wB
is

greater than
as

equal, equal

if less, less.

A
1)

to

to

D.

XVII.
to

If

A+B
lie

be to

as

D.

If

mA

between n B and (n
Ijetween

+D +

to

D, then
it

is to

B,

follows that

B as C is m A +m B,

orw(A + B)

(m + n)B and {m-\-n-\-l)B. Then, by + D) lies between (m + n)D and (w4-n+ 1)D, ormC + mD lies between mD + wD and wD + (n4-l)D, or wC lies between nD and (n + l)D. Therefore, the scales of A and B,
lies

the proportion,

m(C

and of

and D, are the same

If

whence
is

the proposition.

XVIII.
is to

A be

to

as

D.

proof of exactly the

D, then A + B is to B as C + D same kind as the last should be given

to

by the

student. If

XIX.

A B
:

: :

C
.

A B
:

: :

A
is

C
C
(A

D, C and D being less than A and B, then D. For the hypothesis gives A to C as B to
:

D, and

+ (A
C)

C), and

is

D + (B D + (B B D
as

D), whence,

C+
(xvir.)
(xvi.)

isto

A A
If

C C

isto
isto

C C B B C
>

as

D)
isto

isto

D
A
to

as

D
or as

D
as as

to

D,

XX.
and

A
B

be to
to

D
E
B

to to

E
F
fgreater than J

greater than 1

equal to S less than

C, when

is

equal to

>

U^^s than

j
is to
is

Let

but

A is

be greater than C ; then A is to B more than C to E, and C to B as F to E to B as therefore,

to

B E
;

more than F
other cases

is to

E, or

is

greater than F.

In a similar way the

may

be proved.

NUMBER AND MAGNITUDE.

Hence
it

59

follows, that

is

to

as as

to F. to
to

For,

(vi.)

TwA

is to
is to

wB
wC

mD
wE

twE

wB
therefore,

as

wF
is

wA

is

>=
A
is

or
to

<nC
C

when
as

mD
to

>=

or

<nF

whence,

F.

XXI.

If of the magnitudes

ABC
or

Then
Let
before

D E A> =
A be
is

we ^ have

<
C
than

C
;

when
then
is to

A:B::E:F B:C::D:E D > = or <

is to

greater than

more than
is

is

to

as

to

F more

D,

or

greater than F.

Simi-

larly for the other cases.

XXII.

If there be any

number of magnitudes,

A
P
and
if

D
S

any two adjoining be proportional to the two under or above then any two whatsoever are proportional to the two under or them,
above them.
: :
:

For, since

(xx.)

A B P Q1 -^ ^ ^ ^[Therefore,A:C::P:R-j_. r B:C::Q:RJ ^ _ _ Therefore, A D But, C:D::R:Sj , &c.

:

'

-*

}>

td
:

S,

XXIII. In

the hypothesis of (xxi.),

to F.

by proof

as before in (xx.),

A is

to

as

XXIV.

60
therefore,

CONNEXION OF

is

the least.

Now, prove

that if

be the greatest,
it

is

the least; and that,

by

inverting the proportion, if necessary,

first,

may

always be written with the greatest term

and the

least last.

When A
is

is

the greatest, since

A
A

greater than

therefore, (
is

B:A::C D:D, A B) + B + D greater

is

B
than

(C

D)

+ B + D, or A + D

greater than

C + B.

If there be a given ratio, that of A to B, and another magnitude P, there must be a fourth magnitude Q, of the same kind as P, such that A is to B as P to Q, or Q to P as B to A.
Firstly;

greater than

may certainly be taken so small that (mB being nk) mQ shall be less than wP. Find m and n to
and
let

satisfy the first conditions,

satisfy the second.

Then

K is

to

less than

is to

A.

Now (mB
second.

being less than nA),

may be

taken so that
the
first,

mQ shall be greater than nP.

let

Find

m
to

and n

to satisfy

and

satisfy the

Then,

is

in a greater

ratio than

to B.

And
to

it is

immediately shewn that every magnithan

tude less than

greater than

K
is

is

less

B
to

to

to

P more

than

A.

A, and every magnitude Whence, it is between

is

and

that the fourth proportional

found,
;

if

There cannot be more than one such value of

different
great,

for, if there

any where. be two

sufficiently

magnitudes

and

we may make

mV and m W differ by more than

and

W,

since, then,

by taking

sible that both

mV

mW
B

can

lie

imposbetween the same consecutive

P,

it is

multiples of P, as those of

which contain between them

if

mA. And

the above also evidently shews, that

changing

its

value from

K to

L,

it

we suppose a magnitude Q, cannot during its increase become

of the same kind as L, namely, more to

again become of the same kind
as

than

is

to

A, and then

K.

For, whatever magnitude has

this property of L, every greater

one has the same.

at

There

is

then

only one point between

and

which
:

this

change takes place,

Either G (between K and we have, therefore, this alternative and L) is less to P than B is to A, and everi/ magnitude greater than

G
P

is

more

or, some magnitude

G
all

between

and

is

the

same

to

as

to

A, and
less to
first

is

the intermediate limit lying above all those

which are

P, and below

those which are more.

By

dis-

proving the
let

alternative,

we
B,

prove our proposition.

If possible,

be

less to

than

to

G-fV

more, however small

V may be.

NUMBER AND MAGNITUDE.

Then may
while

61

wG

be made
is

less than

nP

m(G+V)

greater than
at
;

wP.

(wjAbeing greater than nB), For the ascending assertion

Let

must be converted
short of

least

into

a stationary one.

mG

fall

wP

by Z

then

not be so great as Z, or not so great as

that

V may be taken so small that mV shall mG+niY not so great as wzG+Z, that is,
first

wP.

But the

clause of the alternative supposes

w(G+V)

must be greater than

wP, how

small soever

V may
true.

be;

therefore this clause cannot be true, or the second

must be

whether

This fourth proportional to A, B, and P, then, must exist ; but it can be expressed by the notation, or determined by the means
is

of any science,

another question.
are

It
:

when
areas.

and

commensurable

it

can be expressed in arithmetic can be found in geometry

(by the straight line and circle) when But if they be angles, arcs of

and

are lines or rectilinear

it

circles, solids, &;c.

cannot be

assigned by the straight line and circle, except in particular cases.

Let us suppose the

**

ratio of

to

given, that

is,

not

and

themselves, but only the answer to this question for

all

values of w,

Between what consecutive multiples of

lies

mA.

Suppose also

the ratio of

to

given
all
?

how

are

we

to find the ratio of

A to

C, or

can

it

be found at
ratios.

that

is, is it

given or determined by the two

preceding

Take any magnitude P, and determine

to

P B

is

to

Q A

as

B, and then determine

so that
is

to

C.

Then
or

the ratio of
is

to

(page 59)

Q so that Q is to R as that of A to C not

;

that

is

(for they
is

may
C.

even be magnitudes of different

kinds), but

is to

R as A
to

to

The
those of

process by which the ratio of

A to C

is

found by means of

to

B and B
What,

C,
to

is

called

by Euclid composition of these

of the ratios of
ratio

ratios; or the ratio of

is

compounded
Y.

to

and

to

C.

then, ought to be

meant by the

compounded

of the ratios of

to

and

to

Our guide
is

in the assimilation

of processes, and the extension of names,

always the following

axiom.

Let names be so given, that the substitution of one magnitude

equal magnitude shall not change the name of the process ; and, generally, that the same operations (in name) performed upon
for another

equal magnitudes, shall produce the same result. Let X be to as B to N, where N is a fourth proportional to be determined. Then the ratio of to N is that compounded of to B

and

to

N, and

is

what must be meant by that compounded of

62

CONNEXION OP
to

B, and

to

Y.

It

is

proved in Prop. 20, that

ratios

com-

pounded of equal

ratios are equal ratios.

ratios of A to B, by which the ratio of A to D is B, B to C, and C to D, still be called to as C to D, and to N as E to F :

Again, to find the ratio compounded of the

to

D, and

to

let

the process
to

derived from those of

composition.
the ratio of

Then
to

take
that

is

compounded

of the three ratios.

for

In the beginning of this work, we deduced the necessity

sidering incommensurables in

con-

some such manner

as that of Euclid,

from the notion which, as applied to commensurables, admits of a definite representation, derived from the idea of proportion. But the

method of the

fifth

book

is

different.

It is there implied, that

where-

ever two magnitudes

exist," their joint

existence gives rise to a third

is

magnitude, called their

given except what
is

ratio,

of which magnitude no conception

contained in certain directions

less,

how

to apply the

terms equal, greater, and

natural question
is,

to

two of the kind.

is this,

On

this

the

what

sort of

magnitude

and how do we
this

know

that there

is

any magnitude whatsoever which admits of

apparently arbitrary exposition of definitions ? This question is very much to the point, and the want of an answer at the outset is a main

cause of the difficulty of the Fifth Book.

The answer implied

to

in the

work of Euclid

is this

Let us

first

consider what will follow

if

there

be such things as equal, greater, and

ratios, or
less

magnitudes
;

which these definitions of

apply

we

shall then

shew

(in the Sixth

it

Book)
said

that there are different pairs of magnitudes, of

which

maybe

that they have ratios,-

and we

shall never

have occasion

to inquire

what

ratio

is.

We may
straight lines

take a case parallel to the preceding from the First Book.

straight line suggests nothing

The notion of a

but length

that of

two

which meet, suggests a relation, which we may conceive If A, B, C, and D, be straight lines, of which stated in this way.

and B, and C and D, meet ; let A and B be said to make the same angle as C and D, when, if A be applied to C, and B and D but let A be said to fall on the same side, B and D also coincide

make

a greater angle with

outside of

than

with
this

D, when,
it

in a similar case, that

falls

and D, &c.

To

would be answered,

the preceding definitions are a circuitous

way

of saying that the angle

made by two

lines is their
it

which, though

opening or inclination ; an indefinite term, distinguishes angle from length, does not serve to

NUMBER AND MAGNITUDE.

63

compare one angle with another. And just in the same manner, if it were not that the definition is more complicated, and refers to an
abstract, not a visible or tangible, conception,
it

would immediately

be seen that

a term which is sufficient magnitude, to distinguish the thing in question from absolute magnitude, but which does not give any means of comparing one thing of the kind
ratio
is

relative

with another.
If,

The immediate deduction of

and (n
it

this idea is
it

as follows

whenever

wA lies between n B
B,

+1) B,
'

also happens that

wP lies between wQ and

two
the
certain fractions of

(n -j-1) Q,

follows that

A, lying between

B, and

B, then

lies

between

same two

fractions of

Q.

Or,

if

7wA

= nB,
Q

that

is, if

n A= m B,
:

then

is

the

same

fraction of

Q.

Or we may

state
for the

it

thus

if

be

made
units.

unity, for the

measurement of A, and
the

measurement of

P, then

A atid P are

same numbers or

fractions of their respective

Euclid has commenced the subject with a rough definition, as we have seen, p. 29, and the translators have spoiled it, by not distinguishing between quantity, and relative quantity; that is, by so

wording the definition as

relation of

to say nothing more than that magnitudes with respect to magnitude.

ratio is

We now come to
to arithmetic.
relates to abstract
it

consider the application of the preceding notions

first

Let us

separate all that part* of arithmetic which

and

definite

numbers, from the

rest,

and

let

us call

primary arithmetic.
to

little

observation will shew that abstract

is really
is

number
ratio of

as distinguished from concrete,

the

same thing as
It is

magnitude magnitude. an idea which we obtain equally from looking

What

three^ for
at

example?

and

From

putting such concretes together,

we

bring away a notion of there

being the
*

same

relative

magnitudes existing between the individuals

The whole
$

of the First

Book of my

Treatise on Arithmetic, with th

exception of

158, 165-169.

64
of each pair.
in the third,

CONNEXION OF
In the
first,
it

is repetition^ in

the second,

it is

length,

the

first is

reminded of; but in all three, we say Now this word times is, in fact, a three times the second.
it is

opening, we

are

limitation,

which

will not

do

we

will

have no other

ratios

for our present purpose; it implies that except those of line to line in the series

A
B
C

h
h
I

n
,

n
-,

^
-,
1

D
made hy

I-

&c.

repetitions only : but there may be ratios which are not those of line to line in any repetition, how far soever carried.

Here

is

a point at which

we

are

compelled

to pause, to adjust the

well-known terms of number

to the

new

idea

we have put upon them.

Abstract numbers are certain ratios; abstract fractions are certain other ratios : but all possible ratios are not found among numbers and
fractions
so
;

whence

it arises,

that primary arithmetic, though

it

may

be,

far

as it goes, a theory

of

ratios,

is

not a theory of all ratios, nor

are

its opei^ations

such as can be performed upon all ratios.

That

ratios

are magnitudes,

we mu&t have supposed Irom

the

beginning, seeing that they bear the terras equal, greater, and less. But there was still this defect, that our test of being to B more than

to

D, was one which

left

us with no idea

how much more

was

to

than

to

D;

which amounts but

to this, that

we could

not define

the ratio of ratios without having

first

defined ratio.

But, in like

was made the guide to that notion which is so will the properly* called the ratio of incommensurable quantities, ratio of two ratios in arithmetic lead us, after a little consideration, to

manner

as arithmetic

the

meaning of the

ratio

of ratios of incommensurables.
refer to the repetitions of the smaller in
:

When we
ratio of

say two,

we

magnitudes, thus visibly related

When we
It

in our language. say twice two, there is a change of idiom instead of twice two is four, two twos are four ; that is, might be, where there exists that idea of relative magnitude which we signify by

commensurables those operations which apply incommensurables.

of Consistently; so as to couple with operations upon problems to the same problems upon

NUMBER AND MAGNITUDE.

(wo,
let

65

the idea o( relation be coupled with the idea of a larger relation,

in exactly the

same manner

as our idea of magnitude,

at
relative
;

when we look
shall then,

at

is

increased

when we look

and we

by

considering the result as

one of

idea of the relation between

magnitude, be led to the and This,

.

of course, does not give a better comprehension of twice two is four; but what it explains is, that we are using the term ratio in a consistent
sense,

when we

say that the ratio of 2 to 1, increased in the ratio of

as the ratio of

to 1, is the

same

to

and, generally, that the

ratio of

to 1, increased in the ratio of

to 1, is the ratio of

mn

to 1.

And

the notion of relative magnitude contained in the words, ratio of

m
to

top, must be the same as that contained in the words, ratio of

wn

pn

and, conversely, the notion in the latter

is

that implied in the

I doubt if any thing that deserves the name of proof can be given of this proposition, which seems to be worthy the name of an axiom. What idea we form of magnitude as portion of magnitude

former.

from

I imagine these propositions extended to fractions in

and B, the same do we form from 2 A and 2B. Nor can any more funda-

mental manner than by observing, that as

taken

- times
q

is

ng

times (times mean times, or parts of times, either separately or both jn p to 1, altered in the ratio of- to 1, together,) a unit, the ratio of n 9
is

the ratio of

to 1

or that the ratio of

nq
ofp
to 9, is the ratio of

to n, altered in the ratio

mp

to

nq.

These are propositions in which

I

the line between deduction and character of terms

is

mere establishment of the synonymous

very indefinite.

recommend

the student to

idea of what he would have meant by " the proportion of 3 to 2 increased in the proportion of 5 to 4, is the proIf he be a metaphysician, I refer him to his portion of 15 to 8."

examine

his

own

oracle,

on condition only
multiplication of

that the response

shall not contradict the

preceding proposition.

The
of

and n

is,

then, the alteration of the ratio

;

to 1 in the
is

proportion of n to 1

and the

ratio of

magnitudes

mk and nA
oim
to n.

the

same as the

ratio of

magnitudes

mB and
:

wB, and

Hence, to alter mA: nA. (which is m : n) in the ratio of to qB, which is (p : q), is the formation of pB mp nq, or mpA to nqA, or mpB to nqB. Now, this is precisely what Euclid has G 2

66

CONNEXION OP
ratios
:

termed the composition of these

then
:

for, let
is

vB pB compounded withpB

jB,

w A nA::vB pB, vB 9B, or v g. But

:
:

mA:nA::m:n
Therefore

vB pB
:

::

p
1
A

m
V
:

n
is

is

p ^
or

or

n
:

is

V
or

=^pm

q
:

pm ^-'-q

pm

nq

or

pmA nqA
:

pmB

w^'B.

Hence, composition is multiplication of terms, when the ratios are those of number to number. Let, then, composition of ratios stand
for multiplication of terms,

operation in the case of

Prove from

this,

and be considered as the corresponding incommensurable magnitudes. that if U : A and U : B be compounded, giving

U
if

C,
:

that

when
:

A=aU and B=tU,

:

we have

C=6U,
U
:

and that

U A

and
:

B U
:

be thus compounded, giving

:

D, we have
to

D = T U, D U
tiplication

in

which operations, corresponding

mul-

and division.
matter of some curiosity to

It

may be a

know whether Euclid

carried

with him the notion of multiplication of numbers in the composition of ratios. In the Fifth Book, the notion of the numerical magni-

tude of a ratio
TyiXtKovyii (see

is

entirely suppressed, except only in the single

word

page 29.) Composition* is defined to be the taking an antecedent of one ratio with the consequent of another; and it is
not even specified that the intermediate terms are to be the same. But in the Sixth Book we find composition, or collocation of ratios,
to

mean

the multiplication of their quantuplicities (see page 29).

livvhifis

Xoyov

Iffr)

Xi^-^is

rod

riyoiif/.ivov fjtirk

rod Wofjbivov ui

ivog 9r^o(

avTo TO

i^o//.ivov-

V. Def. 15.
^'

tavrks

Aoyos Ix Xoyuv cvyxu(r^ot,i kiyiron orav ai tuv koyuv ^nXixortjrts VI. Def. 5. '^foXXocrXa.ffia.ff^ilffai Totutft rtva,.

editions.

The second of these definitions has usually been omitted in modern But it is worthy of remark, that, in the first, to compound is
; ;

and the second is the word afterin the second, evyxna^on ffuvri^iff^at wards used by Euclid, though in the sense of the first. The reason of the omission appears to have been a disposition on the part of commentators to consider Euclid as a perfect book, and every thing which did not
accord with their notions of perfection, as the
interpolators.

work of

unskilful editors or

NUMBER AND MAGNITUDE.

The
ceived

67

addition and subtraction of ratios can only be primarily conthe latter terms of the ratios are alike.

when

Thus,

A
comthe idea of relative magnitude given by compared with A, to be put together, in pared with A, and by C to A. Addition and order to make up the relative magnitude of

we must imagine

BC

BD

subtraction are, as to ratios, ideas not so simple as multiplication

division.

and
:

Shew

that the preceding

;

is

the only
:

way
1,

in

which

1,

increased in the ratio of w

the

1, will give twto

consistently with

notion of multiplication

of whole

numbers being

successive

additions.

When
to the
:

ratios

have not the same consequents, they must be reduced

same consequents. Thus, D :: taking A B :: P Z and C

: :

A B Q Z,
: :

and
and

C D are added by P 4-Q Z is the sum

: :

of the ratios.

This answers to addition of fractions.

Let

as

and B, meaning that be the mean proportional between B. It may be proved, as in page 60, that there must be
:

such a magnitude as this

that
:

mean
:

proportional, and
:

we can find A P as P Q, and P Q as two mean proportionals. It is readily proved,

also prove B, thus forming that if and

we may
:

A=aU

B=6U,
fraction

then

P=:cU

where cc

= ab.

If,

then, a 6 be a

number or

A
is

which has a square root, P can be found commensurable with and B ; but if ab have no square root, number or fraction, then P

incommensurable with

and B, but

not, therefore, unassignable as

a magnitude, though unassignable as a numerical fraction of

A or B.

when we speak of n/2, it must be with reference to magnitude, and we mean \/2 M, an accurate representative (if we and 2M. choose to define it so) of the mean proportional between Similarly, when there are two mean proportionals, we find P, if
Consequently,

A=aU and B = 6U,

mensurable unless
3
/

to be
a.

cU

where ccc

= ab,

and

this is

incom-

abhe

cube number or
of two

fraction.

But we may

v2M to be the and 2 M and so on

define
;
.

first

mean

proportionals between

definitions,

Are we, then, to use long processes and comparatively obscure whenever the ratios of a problem are incommensurables ?
;

By no means

we proceed

to

shew

that

it

may always be made

pos-

68

CONNEXION OF
be used

sible to let the processes of arithmetic (or rather of algebra)

as if the ratios in question were

commensurable
either

and that we may

strictly at the

thus deduce a result which

may

be interpreted

end of the process, or made

to give a result as near as

we

please to
:

Let us suppose this Problem Two are spent in buying yards of stuff, and as many yards are pounds bought as shillings are given for a yard. Let x be the number of
the truth in arithmetical terms.
yards, then x yards at x shillings a yard, gives

xx

shillings;

whence

xx=i

40, which

is

arithmetically impossible.

Now,
let

turn from

numa

bers of

pounds

to quantities of silver,
;

and

S be the

silver in

be a yard, and Y the length shilling, bought. Then it is required that 40 S should be given, and that X should bear the same ratioto S as Y bears to L. Now, if X be given

that in the price given

let

for

L, what must be given

for

Y?

Take P of such

relative

magni-

tude to X, as

is

to

that

is, let

But as
or

L L

Y::X P = 40S Y S X Therefore

: :
:

40S

make our symbols

must be a mean proportional between S and 40 S. Now, if we general, and let x stand for any ratio, numerically

possible or not, but proceed as

we should do
and find

if it

were arithmetical,
which, being in-

we proceed

as in the

first

case,

x=n/40,
its

terpreted as a magnitude, with reference to the symbols are general,

the

ratio to S,

means, when

and 40S.
suppose 40
then

If
-|-

we wish

for

v/40S, proportional between S an approximate numerical result, we must

mean

a to be the sum, where 40

+ a has

a square root, and

we have j: \/40+a; and since a maybe made as small as we please, we can make this problem as near the given one as we
The
following table should be attentively considered.

please.

In the

first

column, an incommensurable ratio x, of X to U, is given, or a function of it and other ratios, under arithmetical symbols ; in the second
is

the ratio which the function really gives,

side are extended in meaning.

when

the s)'mbols

on the

first

X
z

OT

X
z

'.

the ratio of

X
Y
Z

to
..

y-y:
..

:\

..

U U V

NUMBER AND MAGNITUDE.

69

U X
:

as
\

X
: :

compounded of j:

and x

U X
;

X P
:

tiien

and

or

j;*

is

the ratio which

U and X U. Let or X X U compounded give P U a third proportional to U and X

: :

bears to

U.
ratio
;

xyz

{x:\){y\\){z'.\)
Let
:

compd. of
is
:

X
Z.

U,

U, and Z

V.

X U
:
:

P Y
:

the above

then

P U and Z V. Q V or xyz
:

Let
is

P
:

compounded of The result is then

Q V Y
Z
:

a fourth prop, to U, X, and

Q
xy
-\-yz
a-3/
:

U,P,
1

..
: :

is

yz'.l is Take Q
:

P U when U X Y P Q: V .... U: Y::Z:Q

:
:

: :

M U
:

or

V Q
:

: :

P+M U
:

is

the ratio required.

:

X-

compd. of

X U
:

and

V Z
:
:

Let

X U
:

::

P V

and

is

the ratio required.

Now, we have assumed the operations of finding a fourth proportional, a mean proportional, two mean proportionals, &c. Whether
these can be done, or whether

any or

all

cannot be done,

is

a question

suppose the data arithmetical; a fourth proportional can always be found. In geometry, a fourth proportional can be found to lines or rectilinear
for every particular application.
will

In arithmetic,

we

areas

but not

to

angles, &c.

And

mean

proportional

cannot

generally be found in arithmetic, but can be found in geometry, between two straight lines, or two rectilinear areas. But two mean

proportionals cannot be found in geometry or in arithmetic.

It must be remembered, that while we are here speaking of geometry or arithmetic, we are not speaking of every conception we can form of these sciences, but of the subjects as limited by the definitions of

what

it

has been agreed shall be called arithmetic and

geometry.
fractions
:

Elementary arithmetic means the science of numbers and

elementary geometry, the science of space, so far as the

same has properties which can be deduced by allowing o^ fixed To say that an angle cannot be trisected straight lines and circles.
geometrically f means, that
it

cannot be trisected by means of straight

70
lines

CONNEXION OP
and
circles as defined.

But

there

is

an abundance of curves,
the

the stipulation to
trisecting

draw any one of

wliich

would secure

means of

an angle.
roll

allowed to

And, by simply granting that a circle should be along a straight line, and that the curve described by

one of

its

points should be granted,

we can

either square the circle, or

the same way, if we journey to be 100 miles or less, it would be perfectly true that we could not make a journey from London to York, but that we could from London to Brighton.
arcs.

find the ratio of

any two

And,

just in

were

to define a

It

is

surely time that the verbal distinction between different parts

of the same sciences should be done away with.

Every conception
can be as easily which is itself a

which can be shewn

realised

to

be not

self contradictory,

by assumption

as the drawing of a circle,

perfect geometrical idea,

and can only be roughly represented by

distinctly conceived, exists for
all

mechanical means.
all

Whatever can be
;

mental purposes

whatever can be approximately found, for

while to

practical uses.
It

may be worth
which
exists

make

the student remark the close

between the process in page 64, and that by which we enlarge our ideas in algebra, from the simple consideration of numerical magnitude to that of positive and negative quantities.
similarity

In both, we
results

set out
;

with a notation insufficient to express

in both, this circumstance
is

all

the

of problems

appearance of unexplained results, wider grounds, shews the necessity for attaching more extensive ideas to symbols ; and in both, the partial view first taken is wholly included
in the

marked by the the examination of which, on

more general one

while in both, the processes conducted

under the wider meanings are precisely the same in form and rules as those which are restricted to the original meanings of the symbols.

The

principal difference

is,

that in extending arithmetic to the general

not engaged in interpreting difficulties from contradictions, but from results which are only approxiarising mately attainable. But in both the reason is, that we set out with
science

of ratios,

we

are

our symbols so constructed, that we cannot undertake a problem In beginning without tacitly dictating conditions to the result.
algebra,

we make

quantities indeterminate in magnitude, with symbols

of operation so fixed in meaning, that they cannot be used without

an assumption that we know which

is

the greater

and which

is

the

NUMBER AND MAGNITUDE.

less

71

of two unkr.own quantities.

We

have, therefore, to examine the

different results according

different cases of

problems which present

as one

datum

is

greater or less than another;

and thus we obtain those

and the symbols In beginning arithmetic, we invent no symbols of ratio, except those which represent the ratios of magnitudes formed by the repetitions of a given magnitude. These we find to be not
extensions of meaning which will
the problems

make

equally general.

sufficient to represent all ratios

them represent any

we

please.

; though it is shewn that we can make which magnitudes can have, as nearly as The invention of new symbols of ratio must require

ratio

the generalisation of operations

that

is,

we cannot speak of multi-

plication or division of ratios generally, while these words have a

definition
ratios.

which applies only

to ratios

of repetitions, or commensurable

There

is

a difference between

the impossible
first is

of primary arith-

metic, and that of geometry.

definition, the second

The

unattainable by a restricted

by

restricting the cases of general definitions

which

shall

be allowed to be used.

In arithmetic, we attempt a

science of relative magnitudes,

relative

magnitude to the magnitudes of one certain

by running from the general notion of more precise and easy notion of the relative
set of

magnitudes, A, an arbitrary, A-j- A,

A-f-A-f-A, &c.
not express
think about:

We

are very soon taught that our

symbols will

all ratios, that is, if

we have

a general notion of ratio to

whence our

definitions

are not sufficiently extensive.

But

we
all

geometry, having assumed notions and definitions from which cannot help conceiving an infinite number of different lines and
in

curves,

we immediately proceed
and

to cut ourselves off

is,

from the use of

except the straight line

circle; that

the straight line between

or

beyond two given points, and the circle which has a given centre and a given radial line. Until these demands or postulates are
is

looked upon as restrictions, their sense

the Appendix.)

never understood.

(See

This difference
the

is,

however, not very essential

in too limited a

since

it

is

much

same whether we define

manner, or whether we

limit ourselves to the use of only a part of a general definition.

shall in the

We

sequel discard the restrictive postulates, and suppose ourselves able to draw any line which we can shew to be made by
the motion of a point.

The method by which Euclid

first

exhibits

four

proportional

72

CONNEXION OP

and ingenious, has not the advantage of exhibiting the notion of ratio directly applied to two straight lines. The following theorem is directly proved from the first book, and
Straight lines, though elegant

maybe made

the guide.

If a series of parallels cut off consecutive

equal parts from any one line which they cut^ they do the same from every other. This premised, suppose any two lines OA, OB, and
take a succession of lines equal to

OA and

OB,

drawing through
other line,

every point a parallel to a given

intersecting all the parallels
:

line.

Draw any

O C D,

from which the preliminary proposition ; shews, that whatever multiple O a is of OA, the same is Oc of

OC

and whatever

O6

is

of

OB,

the

same

isOdofOD.
is

And

if

O a be

greater than, equal to, or less than

06, Oc

greater than, equal to,

or less than, Orf.

cisely to the lines

Hence

the definition of equal ratios applies pre-

OA, OB, OC, and OD,

which

are, therefore,

proportionals.

This gives the construction of

to
it.

Book VI. Prop.

12, or

one analogous

The

metliod of finding a

mean

proportional between two straight

Prop. 13; but as we now wish to make the straight line the foundation of general conceptions of magnitude, we shall pass at once to those considerations which involve any number of
lines is given in

mean
first

proportionals.

It

adds considerably

to the interest of this part

of the subject, that

we

are thus brought to the notions

on which the

theory of logarithms was founded .

Let there be any number of

tinued proportion
;

lines,

V, Vj, Vg, Vg,

in

con-

that

is, let all

the ratios of

V
to

to V,, V^j to Vg,

Vj

to

Vg, &c. be the same.

And

let

V, be greater than
the following

in

which case

Vj

is

greater than Vj, &c.

If

Vj were equal

V, then would Vg be

equal to Vj, &c.

And,

first,

we have
little

Theorem.
is

By however

Vj exceeds V,

the series
:

V, Vj, &c.

a series of magnitudes increasing without limit great A may be, a point may be attained from and

so that, however
after

which every

term

is

greater than

but in

all

cases whatsoever,

Vj may be taken

NUMBER AND MAGNITUDE.

SO near lo V, that the terms of the series

73

V,
as

Vp

&c. between which

A lies,
Vj
:

shall

be as near to

in

magnitude

we

please.

Firstly, the

series increases without limit.

For, since

Vj

Vg, and

and V^ are the greatest and

least,

we have

V -f Vg
or

is

greater than greater than

V.2

Vi

is

V^ V^

+ V^

Or, Vg exceeds Vj by more than V, exceeds V. Similarly, Vg exceeds Vg by more than Vg exceeds V^ and so on. But if to V were added continually the same quantity, the result would come in time to
;

exceed any given magnitude

still

more when a

greater quantity

is

added

at every step.

Secondly, since then

we come

at last to

less than

A, while

Vn+i exceeds A, it is plain that A much as they differ from each other.

will not differ

from either by so

But because
:

we have
If then

V V+i-V
Vi

V,^.i
:

V V, V^-V V
: :

be so small that

m (Vi

V)

shall not

exceed V,

neither will

m{Vn+i

V)

exceed

V,

and of course not A,

Let

V be less than the be any given number, however great, and let Vj ?//th part of V ; then will Vn+i Vn be less than the mtln part of A;
or,

by taking

sufficiently great,

may be made

as small as

we

please.

Whence

the second part of the theorem.

Theorem.

In the preceding

series, the selection

V V

V,

V3

&e.
proportionals.

constitutes a similar series of continued

For, since

any two consecutives in the upper to those under them in the lower,

line next given are proportional

V,

V
:

V,

v
V2
:

v V+. V+2 we have (xxii.) V V V,^

:

V2

and so on.
insert the

If between each of the terms of the series

we

same number

of

mean

proportionals, the series thus formed will have the

same pro-

Let us say we insert two mean proportionals between each two terms. Then we have
perties as the original.

74

CONNEXION OF

V K
Now
term
to

K' Vi

L'

V2
:

M
:

M' V3
same
ratio

the only question about the continuance of the

from

term
:

is

in the ratios V,
: ;

L, Vj

M,

&c.

But

I say that since

:

V K
V,
,

K K
:

K'
L'

Vi
^""^

V
V^
is

V^

L'
if

V,

V^
;

that

V K V is to K
:

V, L. For more than V, is

: :

not, let these latter ratios differ

say

to

L.

Then

to

K' more than

is

and hence (presently will be shewn) the ratio compounded of V to K and K to K', or V K', is greater than that compounded of V, L and L L' or V U. Similarly, V to K' and K' to V, being
to
;
:

U
:

more than Vj
which
is

L' and L'

Vg,

we have

V
:

Vj

is

not true.

Therefore
it is

is

not to

more than Vj to Vg, more than Vj to L ; a

similar process shews that

not less
:
:

consequently,
:

V K
;

Vi
is
is

L
uninterrupted.
If

or the continuance of the primary ratio

The theorem assumed

than

in the

above

thus proved.

A B
:

more

we have

7A greater than nB, while

And
is

2P

is less

than

nQ;
:

or any other descending assertion.

if

more than
3^

R,

we have

:rB greater than ?/C, while a"Q

less than

R.

Or we have
greater than
less

mxA greater than nx^, mx^ less than nxQ,

that
is,

nx'Q

greater than
less

nyC,
which

or

mxA

nxQ,
than

than
to

ny^^oxmxV
;

than

nyC wyR

A is

to

C more
insert a

is

is

what we assumed.

If then

we

mean

proportional between

V and A, giving

V
if

M
M

A
we have

between each we

insert a

mean

proportional,

V
If

M'

M"
come

A
at last to a series of the

we proceed

in this

way, we

shall

form

V
in

V.

V_,(V

= A)

We

which no two quantities differ by so much as a given quantity K. can actually insert one mean proportional between any two

quantities ; it is done in geometry between two lines, and (page 60) two magnitudes of any sort may be made (one being given) proportional to two lines. Thus, let A, B, C, be continually proportional

NUMBER AND MAGNITUDE.

lines, or let

75
and C.
it

B
: :

be a mean proportional between

taken proportional to (say angles)

Then

if

A and C were that if A B

:

M and K,

follows

L, that

M,

L, and
in

are continued proportionals,

by a proof of the

sort given

the

lemma

of the last theorem.

Granting, then, that every two magnitudes have one

tional,

we may now shew

;

that they
:

mean proporhave any number of intermediate

first

proportionals

as follows

We

set

out with 2 quantities, and the

insertion adds 1, the

second 2, the third 2^ the fourth 2^

sequently, n complete insertions
1

and the wth 2"-!.

Con-

add

+ 2 + 22 +
;

+ 2"-i
1 in all.

or

-1
+1
not
is

to the

first

giving 2"

-|-

Now,

let

us suppose that 2"

divided by

leaves a quotient g,

and a remainder r which

for the

greater than p.

Consequently,

we have

whole number (V and

A inclusive) after n insertions,

v=pq-\-r
and p
is

which

is

also

p{q

-{-

I)

(p

r)

is

also not greater than

p ; and

Vm=A when m=v and

is

greater or less than

A, according as

greater or less than v.

If then out of the series

Vp(g-j.i))

(the proportion being

continued up to

we

select

V V
We
see

Vg

V23

(Vpq

less

than

V5+1 V2 (q+i)
Vpq, and

C^i>(5+1) greater than

A) A)

V and

and ^p(q+i) each with

mean pro-

portionals inserted between them, namely,

V, Vgg
But from Vpq

^{p-i)q and
to yp(q-\-i) there

Yq+i V2(g+i)
are

V(p_i)(g+i)

passages from term to term

of the complete series, consequently, since each passage

by an augmentation
be made
less than

less than

K, the
call Z.

difference

between the two

may be made may

p K, which
find

Hence we have

the following

Theorem. To
each of which

than A, but differing from

two magnitudes, one greater and the other less it by less than a given quantity Z, between
1

andV, ja
insertion

mean

proportionals shall exist, obtained

proportional, continue the inshall differ

by continual
sertion until

of one

mean

no two successive terms

by so much

as the

76
ptli

CONNEXION OP
part of the quantity

then the quantities required and the

mean

proportionals shall be in the set so found.

Hence

it

can be shewn that there are p

magnitudes (whether

attainable or not with any given

means

is

not the question) which

are

mean

proportionals between

and A.

Let

Pj,

and Qp be mag-

nitudes, one greater

and one

less

than A, which have such

mean

proportionals, namely, let the following be continued proportionals,

V V

Pi

P2

Pp-1 (Pp

greater than
less

A)

Qi

Q,

Q^_i (Q^

than

A)

obtained by the preceding method, from which it is apparent that Pi is greater than Qj. Now, exactly as in page 60, if we assume

Xj
and

to set out in value

p\.\\

=:Q,, so
all

that

Xp more

than

V A
:

(Xp
)

bring the
to

of the set of continued proportionals V, Xj, X^,

possible intermediate magnitudes

change through
or
at
is

up

to

Xi=Pi,
EITHER
OR, there
less

Xp

less

than

V
:

A, there

is

but

this
:

alternative;

some intermediate point

a point at which
is

Xp

as

V Xn

more than

V V

A, or Xp

= A,
latter

A, being always

when Xj

greater

by any magnitude however small.

The

may be disproved, or the former proved, as in the page cited. To resume the original subject. It appears, then, 1st, that
tween

if

be-

and

we

continually insert

mean

proportionals, in such
is

manner

that at every step

one mean proportional

the

inserted

between

every two consecutive results of the preceding step.

2d, If the series

ratio

be continued beyond A, preserving

cutive terms lying between

still

same

between the

consecutive terms of the continuation which exists between conse-

and

then will this process leave us

at last with a series of consecutive proportionals,

having consecutive

terms so near together in magnitude, that every magnitude lying beand any we please to name, shall have a term of the series tween

diflfering

from

it

by

less than Z,

however small Z may be.

NUMBER AND MAGNITUDE.

Let us
line

77

now make
let

OK

and

HL

perpendicular to any chosen

the line

OM,

and

be the line
the

OK, A

HL.

Bisect

OM in
HL.

C, and erect Bisect OC and

CD

mean

proportional between

OK

and

CH,

and

erect the

mean

proportionals between
this process,

OK

and

CD,

and between

CD and HL.

Continue

and we

shall thus get

which

will

to

L.

and L, an increasing number of points between to the eye the idea of a curve line rising from When we have thus divided into 2" parts, by n insoon give

OH

sertions, giving 1^

-\-

\ lines,

we may, by

setting off portions equal

to those intercepted in

OH,

continue that line on one side and the

and thus continue the scale of proportionals and the series of points on one side and on the other of O and H. However far we
other,

may go we can
we

never complete this curve

but

if

we admit

that a

curve exists, wherever a series of points can be laid down, as

please, and consecutively as

near as

we

please, then

many as we have
of'

a right to assume this curve as existing, and, for purposes

soning, as constructed.

rea-

Call this the exponential curve, (exponere, to set forth), which expounds ratios, a phrase to which we shall presently give meaning.

That the student

may

not suppose

we

are using

an

old

word

in a

new

sense,

it is

necessary to inform
illustrated

him

that this curve,

or rather the process

which we have

by

it, is

older than the

it.

algebraical symbol a^,

and

that

gets the

name of exponent from

We shall

presently see the analogy.

given, every line

The exponential curve being

OG has

its

place

M P among
expounds or
it

the ordinates of the curve,

sets forth that place.

and

its

abscissa

O M,

which

From

the nature of the formation,

evident that a given line has but one exponent, and that the order of magnitude of lines (to the right of O), is also that of their
is

exponents.

78

CONNEXION OF
And
the

main property of the curve

lines

is this

that a fourth propor-

tional to

any three

(V being

one),

OK, MP,

(making OM"), and finding the line M"P" expounded by that sum. To prove this, make n sets of insertions in OH, and suppose to lie between

found by adding the exponents

OM

M'P', may be

and

OM'

Vm

and

Vm + i,

while

M'P'

lies

between

MP

and Vm'+i.

Now,

in

the series of continued proportionals,

VV,
^"^

V,

I say that

V V
:

= A)V,V V, V^(V.%,
:

V
^ m'

V,
*

V,
*

V_,
^

V
1

m'+l
:

m'+2
: :

m'+m
&c.

m'-\-m

we have
^vhence
Similarly,

V Vj V V, V V,,+i
: :

Y^'
:

V^.+i
:

&c.

V,,

Y^^^^
:

V^^+i

V,+^.+2

Now, by
tween

lemma we

sliall

presently shew, since

lies

MP
then

lies

be-

Vm

and

Vm + i,

and M'P'

fourth proportional required lies

between Vm' and Vm'+i, the between Vm+m' and Vm+m'+2- Let

K be the
m'K

value of one of the last subdivisions of

to lie

OH

we have
between

supposed OM

between

and (m'-|-l)K.

and (m-|-l)K, and The preceding makes it evident

mK

OM'

that the fourth

proportional has an exponent between {m-\-m')K and (2-}-w'-f-2)K; and also lies between while the sum of the exponents

OM

OM'

and (w4-m'-j-2)K. Since K can be made as small as we please, it must follow that the sum of the exponents is the exponent of the fourth proportional; for two different magnitudes can-

(m+m')K

not

lie

please, as can

between two quantities which can be made as near as we (m -\- m') K and (m m!) K -f- 2 K. If the two approxi-

mating magnitudes approach to each other, keeping one of two between them, they must, at last, leave out the different magnitudes
other.

The lemma alluded

:

to is as follows
:

If

and

A B::C D A B + B' C + C D + D'

: : :

Then
and

if

A,

B + X,

Y are

less than B'

C-f-Y, D+Z, be also proportionals, where and C, then Z must be less than D' ; for A

X
is

NUMBER AND MAGNITUDE.

to
;

79

B + X more than A is to B + B' or (substituting equal ratios), C + Y is to D + Z more than C + C! is to D + D'. Still more is C + C (remember that C is greater than Y) to D + Z more than C + C to D + D' that is, D + Z is less than D + D', or Z less
;

than D'.

The
the last.

following property

we

leave to the student to deduce from

any

lines

Xj Xg X3, expounding Y, Yj Y3, whatsoever greater than V, then the exponent of the fourth
is

If there be any three lines,

proportional

X2+X3

Xj.

These are
{Xoyuv
0.^101^1,01,

all

properties of algebraical exponents, or of logarithms,

ratios).

numbers expounding

We

shall

now make

it

appear, that the line expounded by x is of the form a'. or O be i? ; let that of Let the numerical symbol of

HL

or

be

a.

Then,

if

arithmetical

mean

proportions be continually inserted,

we have
V

{av)^

a
3
1

13
V V
or generally,
serted

a* v^

a^ v^

a^ v^

a
a^v^

a^v^

a^v^
1

a^v^
(say
j9

when 2"

1)

mean

proportionals are inis

between v and

a, the Twth of these

proportionals

(2"
if

= ^)

m
P

m
which
is

av V
Now,
let

i^kp

we suppose a=vk.

us suppose a
this

number

3/

thus ex-

pounded by x; and after n insertions, let ma and {m + iy, a being the pih part of
have then

number x
(let

lie

between
c).

OH,
(m
^

OH be
1)^p
c P

We

X
or

lies

between

mp
ffi

and

+ ^

between

and

m
c
-|-

X
Therefore,

= Cm + P

(^<0
rn a,

= pec
or

Consequently the number expounded by

m-yOX x

/S

is

Vk

'

80
and since
jS

CONNEXION OF
diminishes without limit as the insertions continue, the
X
is

number expounded by x
merical symbols, let

vk*^

That

is, if

we adopt

general nu-

O M = M P =y, and we have

.r,

y
or, if

=:

vlf
and
let

we let OK represent the linear unit(v=l), OK = l, HL = 10, or = 10, we have

/c

0H =

y
or X
is

\0

the

common

logarithm of y.

From
are

the curve

we

see

how

it

is

that magnitudes less than

expounded by negative

quantities, with other

well-known prologarithm of 2

perties of logarithms.

We
distinct

see then, that the assertion " the

common
is

is

30103 very nearly," may be thus made; which

10 V the hundred thousandth magnitude

perhaps the
If

most

view that can be given of a numerical logarithm.

we make

in a series of proportionals,

V, Vi,
equal to 2 V.
If

(Vioo,ooo=10)

Vioo,ooi,

&^-

then will the 30103rd of these proportionals, or V^o^q^, be very nearly

we

chose,

we

might, granting that the exponential curve can be

definition the line

constructed,
for

make V/cX by

MP

where
this

X
it

stands

O M,
X
is

and k

for the ratio of

HL

to

O K.

From

would

readily be deduced, that

when k

represents a commensurable ratio,

and

linear units

where %/ k"^ has an arithmetical existence,

P
the results of this theory are the

same

as those of

And

from hence

it

appears, that the science

common algebra. known by the name of

it is

the application of algebra to geometry (of

which

the foundation,

that a linear unit being given, every expression of algebra

may

be

considered as a length, or at least the symbol of the ratio of a length

to that unit) does, in point of fact,

make

this additional

assumption,

while an application of geometry (with this assumption) to algebra,

would take away

concerned.

all

want of rigorous conception of

the

meaning
is

of algebraical formulae, so far as the

meaning of

the

exponent

The view above given

is

very nearly that by which logarithms

NUMBER AND MAGNITUDE.

were
first

81

calculated, but the

method was not so general. The natural

logarithms (see

my

very large

number

of

Algebra, p. 226) arose thus. If we suppose a and Vj will be very mean proportionals, then

nearly equal.
shall

Let V^

= V + X, then
X
is

if

we assume

O H,

so that
if

X
we
by
or

expound

V+X

when

very small, or more correctly,

suppose the limit of

(V+X)

divided

by

the magnitude expounded

X, as

diminishes without limit, to be unity,

we have

the

first,

Napier's system.

APPENDIX.

ON THE DEFINITIONS, POSTULATES, AND AXIOMS

OF EUCLID.
I

HERE propose

to

endeavour to make such a subdivision of the

definitions,

&c. at the beginning of the First Book, as

may

enable the

student to review the reasoning of the whole.

I shall consider the 10th

and 11th axioms

as

among

the postulates,
;

firstly, because some old manuscripts support this change because the older translations (from the Arabic) support

secondly,
also,

it

and
is

even place the 12th axiom in the same

utterly impossible to place

list; thirdly,

because

it

them
as
its

in Euclid's list of

common
it

notions.

For he uses no such word " the whole is greater than

conceptions of every one.

axiom (Greek though

part,"
Koivn hvota,
is

be), but calls

is

that

which

in the

Now, what
beginner
in

the

probability that

he

considered "

all right

angles are equal," as a truth familiar to the

understanding
{octr'nfActra.,

of

every

geometry?

His postulates

demands) do, according

if

to the

etymology of the word,

kinds of

include those axioms,

1 also place out of

not the 12th also.

all

view the axioms which belong to

magnitude as much as to space, namely, from the 1st to the 9th inclusive. There remains then in the shape of limitation, or assumption,
six postulates,

commonly
11th,

called postulates,*

namely, three which I will call restrictive, being those and three assumptions, being the 10th,
called.

and 12th axioms, so

Some

of the definitions contain assumptions of certain conceptions

existing to
*

which names are

to

be given

namely, those of a point, a

and axiom as a self-evident theorem.

have seen the word postulate defined as a self-evident problem j This definition is derived from
;

the character of the postulates and axioms as usually given other source.

but from no

84
line, the extremities

APPENDIX.
of a line, a straight line, a surface, the extremities

of a surface, a plane surface, a plane angle, a plane rectilineal angle. Others assume the possibility of certain relations existing, as will

appear from the form in which they are put.

I shall

now
in

give the

definitions, classified with the corresponding postulates, in the

manner
]

which appears to me to be most systematic, and placing additions as seem requisite.

1
.

such

A
all

poirit ;
it

an indefinable notion

but two persons, whatever

their idea of

may

be, can reason together in geometry

who deny a
no

point

parts or magnitude.

Let

it

he granted that a point has

parts or magnitude,

and that we are concerned with no other property

of

it,

if

there be any.
also indefinable, but those
it

2.

Kline;

whose ideas of
Let
it

it

allow

it

length, and

deny

breadth, can proceed.

be granted that

all

is to be founded only upon the assumption that have length without breadth. [Thickness should have been they added, but breadth may mean breadth in any direction.] 3. The extremities of a line are points. [If this define any term,

reasoning upon lines

it

must be
it

the term extremities, for the other

two have been defined.

:

To me

appears something like a theorem, as follows That which ends a line cannot have length, for it would be a part of the line it
;

cannot have breadth or thickness, which a line has not

fore the only qualities of a point

it

has there-

on which we reason, or comes within

the definition of a point.]

4.

an indefinable notion, except straight line;

it

idea that

[which is Let it be granted, as a

by the rough does not go on one side or the other of the two points, no definition, because it assumes the thing in question.]

common

notion, that two straight lines do not

enclose a space, or have not

all

intermediate points in

two points in common, without having common. Whatever the idea of a straight

line

may
5.

lencTth

is the only property which will be appealed to. an indefinable notion; those whose ideas give it Surface; and breadth, but deny it thickness, have the means of reasoning

be, this

upon
6. 7.

it

in geometry.

The

extremities of a surface are lines.

(See remarks on 3.)

plane surface; an obvious notion, roughly defined by lying

evenly

between

bounding

straight

lines.

obvious, does admit of a stricter definition.

[This notion, however It is a surface of such

line,
all

kind

that

any two points in

it

being joined by a straight

APPENDIX.
intermediate
points of the straight
line

85
This

are

on the

surface.

property
8.

is tacitly

appealed

to throughout.]

plane angle ; the inclination, or bending towards each other

[This definition
is

of

two

lines in a plane.

superseded by the next;

no angle except one made by

9.

straight lines is ever used.]

A rectilineal angle

(plane), the inclination of

it

two

straight lines.

[An how

obvious notion of opening;

to determine

is

tacitly

assumed

that

we know

when two

angles are equal, or

when one of them

called a

exceeds the other, as in the fourth proposition.] 10. Right angles are those made by a straight
perpendicular, which
falls

line,

on another
;

straight

line,

making equal
all right

angles on both sides.

are equal.
is

Postulate
far

let it

be granted that

angles

seems
in

to

from an obvious postulate ; the reason for it [This have been as follows That two straight lines which coincide
:

two points coincide

sufficiently

in every point between them, has

been admitted

it is

obvious to sense that they coincide beyond or on each

that is, they coincide altogether, ; This seems an infinite assumption ;
all right

side of the

two common points

all

throughout

possible length.

and

if it

be assumed instead that

angles are equal,

it

may be

proved afterwards that no two straight lines have a common segment ; that is, that two straight lines which coincide for any length, never
afterwards separate.
right

But

it

may be shewn,
to the
is

that the assumption of all

angles being equal, amounts

:

as follows

The

right angle

which two
other, as

straight lines

infinity of assumption ; by definition the half of the opening make, when one is the continuation of the

same

AB, BC.

assume

that the doubles

ABC
V
-T
-\
f-

To assume

that all right angles are equal, is to

-r

E
if

F
we
lay

of right angles are equal ; that is, that coinciding with BA, then EF and
precisely the

B
is

on E, with

ED
it

BC

will coincide.

Now
it.

is

same thing

to

assume, that when

AB

made

to coincide

with

D E up to the point E, that the

recommend

two coincide beyond

assumption that two straight lines cannot coincide in two points without coinciding between them, the addition that they also must coincide beyond them.
I should

the student to

make

to the

86
It

APPENDIX.
then be directly proved that
all

may
The

doubles of right angles are

equal,

and thence

that all right angles are equal.]

definitions

11, 12, 13, 14, need no remark, being purely

nominal.
14. The circky a plane figure, having all points of its boundary (15. the circumference) equally distant from a given point (16. the centre) within it. [Here is tacitly a postulate, namely, that this point lies within the figure. It is also assumed in the first proposition, that if any point of a circle be within another, the two circles must
intersect.

There are several assumptions of

this kind,

which shew

that Euclid did not affect that extreme

form of accuracy which sub-

sequent commentators have attributed to him.

circle

The assumption of a

assumes the existence of an isosceles

triangle.]

17.

diameter of a circle

is

a line passing through the centre,

and terminated both ways by the circumference ; it divides the circle into two equal parts, or (18. semicircles). [Here is a demonstrable
theorem positively assumed.
circle to the other (as

The

application

of one

part

of the

as in

by revolution of one-half round the diameter) the fourth proposition, would prove it.]
(19.) to (23.), the definitions are merely nominal.
it

From

24. If there be a triangle having three equal sides, let

equilateral.

be called

[In

this
is

objects of which

to

form I give all definitions, the existence of the be established.] one having two sides equal.
This defini-

25.

An

isosceles triangle is

26.
tion
is

scalene triangle has the three sides unequal.]

never used.]

(27.)

second proposition ; the manner of (24.)

and (28.) are nominal ; (29.) and from (30.) to

two right

tacitly refers

to the thirty-

(33.), should

be written in

(35.) If there be
far
let

lines,

which being produced ever so

on the same side never meet, let them be called parallels. And it be granted, that if two right lines falling upon a third make
angles
together
less

interior

than two right angles, they are not

parallels.

[This bone of contention,

when reduced
:

to

the form in

most palpable to the senses, is as follows Let it be granted that two right lines which meet in a point, are not both parallel to any

which

it is

third line.

This assumed, Euclid's axiom follows.

For he

is

able to

shew

one parallel which he afterwards draws, through a point to a given line, has the property of making the two internal angles
that the

APPENDIX.
equal to right angles there is but one parallel; consequently which have not that property are not parallels.]
:

87
all lines

It remains to

add what

I have called the restrictive postulates.

who appears to assume very obvious even when he might prove them, could have intended to propositions, that a straight line may join two require formally the admissions
I cannot believe that Euclid,
points,

given centre and radius.

and may be continued, and that a circle may be drawn with a If this had been the case, why not assume (Prop. IV.) that two straight lines may be drawn making equal angles a conception more difficult than that with two other straight lines,
of the three assertions

a straight line

may be drawn. I conceive, commonly called

therefore, that the

meaning

postulates,

is

as follows:

it be considered as intended, that no assumption of processes shall be made, except only the drawing of a straight line between two given

Let

points, the continuation of

definite (not given) distance,

any terminated straight line to any inand the construction of a circle with a

given centre and radius.

THE END.

LONDON

PRINTED BY JAMES MOYES,

Castle Street, Leicester Square.

[LIST, No.

3.]

Upper Gower Street March

^

1,

1835.

WORKS
MATHEMATICS, NATURAL PHILOSOPHY,
HISTORY, &c. PRINTED FOR TAYLOR AND WALTON,
BOOKSELLERS AND PUBLISHERS TO THE UNIVERSITY OF LONDON.

**

Some f

the

Books in the following List are in use at the University or

its

School.

THE ELEMENTS OF ARITHMETIC.

Third Edition, royal r2mo, 4s. cloth. a way, that he is not fit to encounter any question unless he sees the head of the book under which it falls. On a very sufficiently convinced me that there exists moderate computation of the time thus a disposition to introduce the principles of arithmetic into schools, as well as the bestowed, the pupil would be in no respect worse ofi', though he spent five hours on practice, I have often heard it remarked children. I that it was a hard book for every page of this work. The method of never dared to suppose it would be otherproceeding which I should recommend, in would be as follows: Let the pupils be wise. All who have been engaged the education of youth are aware that it is taught in classes, the master explaining the article as it stands in the work. Let a hard thing to make them think; so hard, the former, then, try the demonstration indeed, that masters had, within the last on some other numbers proposed by the abandoned few years, almost universally the attempt, and taught them rules instead master, which should be as simple as instead of of principles possible. The very words of the book by authority, demonstration. This system is now passmay be used, the figures being changed, and it will rarely be found that a learner and many preceptors may be ing away ; is capable of making the proper alterfound who are of opinion that, whatever be the additional trouble to themations, without understanding the reason. may The experience of the master will suggest selves, their pupils should always be to him various methods of trying this induced to reflect upon, and know the Such I reason of, what they are doing point. When the principle has been thus would advise not to be discouraged by the discussed, let the rule be distinctly stated failure of a first attempt to make the by the master, or some of the more intellearner understand the principle of a rule. ligent of the pupils and let some very It is no exaggeration to say, that, under simple example be worked at length. The the present system, five years of a boy's pupils may then be dismissed, to try the more complicated exercises with which life are partially spent in merely learning the work will furnish them, or any others the rules contained in this treatise, and which may be proposed." Preface. those, for the most part, in so imperfect
Esq.

By Augustus De Morgan,

" Since the publication of the first edition of this work, though its sale has

THE ELEMENTS OF ALGEBRA,

Preliminary to the Differential Calculus, and
fit

for the higher classes of Schools

in whicli the Principles of Arithmetic are taught. Royal 12mo. 9s. cloth.
III.

By Augustus De Morgan, Esq.

ARITHMETICAL ALGEBRA.
By G.
J. P.

White,

Professor of Mathematics in the University of London. One vol. 8vo. (Prepaiing.)

LESSONS ON NUMBER,
As
given at a Pestalozzian School
at

Cheam, Surrey.
^
S

Second Edition.
1
1

The Master's Manual. The Scholar's Praxis.

12mo. 12mo.

4s. 6rf. cloth.

2s.

/o

bound.

A TREATISE ON ARITHMETIC,
Theoretical and Practical.

By the Rev. Dionysius Lardner, Foolscap Octavo, Qs. cloth.

LL.D.

WORKS ON MATHEMATICS, HISTORY,

ETC.

CONVERSATIONS ON ARITHMETIC,
Adapted
for the

Use of Preparatory Schools and

Part
I.

for

Domestic Instruction.

Foolscap Ovo, 3s. cloth.

VII.

A SYSTEM OF ARITHMETIC,
Practical and Theoretical. Bj' the Rev. William Ritchie, LL.D., Professor of Natural Philosophy and Astronomy in the University of London. 12mo, ds. hound.

PRINCIPLES OF GEOMETRY,
Familiarly illustrated, and applied to a variety of useful purposes. Designed for the Instruction of young Persons. By the Rev. Professor Ritchie, LL.D., F.R.S. r2mo, with 150 Wood Cuts. 35. 6d. cloth. " The practical applications v^rhich are added, must render the study very delightful to the young, since the Exercises oh the Principles will be found as amusing as the ordinary sports of c\nl(.\ho<xA.''Athenceum, Sept. 28, 1833. " Dr. Ritchie's little elementary work is excellently well adapted to its object. It is curious and useful in its application, and brief, plain, and full of all that is necessary other of the kind now existent, in its familiar and distinct explanation of beyond any some of the instruments required in the practical application of the principles laid down and demonstrated." Spectator, Sept. 7, 1833.
;

AN INSTRUMENT FOR TEACHING GEOMETRY,

Convertible into a Theodolite, Spirit Level, Hadley's Sextant, and WoUaston's Goniometer, lias heen prepared, and may he had of the publisher. Price 21. 12s. 6d.

PRINCIPLES OF THE DIFFERENTIAL AND INTEGRAL CALCULUS,

Familiarly Illustrated, and applied to a variety of Useful Purposes. Designed for the Instruction of Young Persons. By the Rev. Professor Ritchie, LL.D.,

F.R.S.

With Wood Cuts.

12mo, 4s.

6rf.

cloth.

X.

CONVERSATIONS ON GEpMETRY,
Adapted
for

the Use of Preparatory Schools and Domestic Instruction. Foolscap 8vo, 3s. cloth.

THE ELEMENTS OF EUCLID,

With a Commentary and Geometrical Lardner, LL.D. Fourth edition, 8vo, 7s.
Exercises.
hoards.

By

the

Rev. Dionysius

AN ANALYTICAL TREATISE ON PLANE AND

SPHERICAL TRIGONOMETRY.
LL.D. Second
'8vo, 12s. cloth.

By

the Rev. Dionysius Lardnkr,

Edition. Corrected and improved.

XIIT.

AN ELEMENTARY TREATISE ON THE DIFFERENTIAL AND INTEGRAL CALCULUS.

By the
Rev. Dionysius Laroner, LL.D. 8vo, 21s.

PRINTED FOR TAYLOR AND WALTON, UPPER GOWER STREET.

XIV.

BARLEY'S SCIENTIFIC LIBRARY,

FOR THE USE OF SCHOOLS, PRIVATE STUDENTS, ARTISTS, AND MECHANICS.
purpose of this Work to furnish a Series of Elementary Treatises on Mathematical Science, adapted to the wants of the public at large. To youth of either sex at public and private schools to persons whose education has been neglected, or whose attention has not been directed in early life to such studies; and to Artists and Mechanics, these little works Avill be feund particularly suited. The principles of the various Sciences are rendered as familiar and brought as near to our commonest ideas as possible the demonstrations of propositions are made plain for the mind, and brief for the memory and the Elements of each Science are reduced not only to their simplest but to their shortest form.
It is the
; ;
;

L A SYSTEM OF

POPULAR GEOMETRY.

Containing in a few Lessons so much of the Elements of Euclid as is necessary and sufficient for a right understanding of every Art and Science in its leading Truths and general Principles. By George Darley, A.B. Third Edition. Price
4s. 6d. cloth.
i

2.

COMPANION TO THE POPULAR GEOMETRY.

In which the Elements of Abstract Science are familiarised, illustrated, and rendered practically useful to the various purposes of Life, with numerous Cuts. Price
4s. 6d. cloth. 3.

A SYSTEM OF POPULAR ALGEBRA,

cloth.

With

a Section on Proportions and Progressions. Second Edition. Price 4s. 6d.

4.

A SYSTEM OF POPULAR TRIGONOMETRY,

Botli Plane and Spherical, with Popular Treatises on Logarithms, and the Second Edition. Price 3s. 6d. cloth. Application of Algebra to Geometry. " For Students who only seek this limited knowledge of these Sciences, there are perhaps no Treatises which can be read with more advantage than Darley's Popular Geometry aad Algebra." Library of Useful Knowledge, Article "Mechanics."

Katural

iStilosoplDB.

THE STEAM ENGINE,

Familiarly Explained -and illustrated ; with its application to the Arts and Manufactures, to Steam Navigation, and Rail Roads ; with Plain Maxims for the Guidance of Railway Speculators. By the Rev. Dionvsius Lardner, LL.D. Fifth One vol. 12mo. Illustrated with Engravings and Wood Cuts. 7s. 6d. Edition.
cloth.
II.

FAMILIAR ASTRONOMY.
By George Darley, A.B., Author of a System of Popular Geometry, &c. 12mo, with Engravings, 5s. cloth lettered. " There is a vast deal of astronomical information conveyed in a most winning and unassuming manner in this delightful little volume, which, not less for the novelty of its plan, tlian the extent of its intelligence, reflects infinite credit on the taste and talents of its projector and editor, Mr. Bailey." Sun, April 5, 1830.

WORKS ON MATHEMATICS, HISTORY,

ETC.

ON THE ORIGIN OF UNIVERSITIES AND ACADEMICAL

DEGREES.
By Henry Malden, M.A.,
of Greek in

Fellow of Trinity College, Cambridge; Professor the University of London. Foolscap 8vo, 35, 6d. cloth.
late

NIEBUHR'S HISTORY OF ROME.

Vol.

I.

Translated by Julius C. Hare, M.A., and Connop Thirlwall, M.A., Fellows of Trinity College, Cambridge. Second Edition, revised, with the Corrections and Additions made in the Third Eldition of the Original, 8vo, 16s.

NIEBUHR'S HISTORY OF ROME.

Vol.

II.

Translated by Julius C. Hare, M.A., and Connop Thirlwall, M.A., Fellows of Trinity College, Cambridge. 8vo, 16s. " Niebuhr undertook to write the His- temples, a half hewn Rameses, as a scholar tory of Rome from the earliest ages of the who should complete the Roman- History of Niebuhr." city to the establishment of the empire of " Here we close our remarks upon this Augustus. Of this great work he accommemorable work; a work which, of all plished only a portion and his History that have appeared in our age, is the best will remain to succeeding ages as a fragment, but it is a fragment which may be fitted to excite men of learning to intelcompared to the unfinished colossal statues lectual activity; from which the most that are found lying in the granite quarries accomplished scholar may gather fresh stores of knowledge, to which the most of Syene, conceived with all the vastness and precision of Egyptian art, which, had experienced politician may resort for they been finished, might have overtopped theoretical and practical instruction, and which no person can read as it ought to the gigantic Memnon, but which, when they were relinquished by the hand that be read, without feeling the better and more generous sentiments of his common first fashioned them, were destined to remain for ever imperfect. We should as human nature enlivened and strengthsoon expect an artist to arise, who should ened." Edinburgh Review, Jan. 1833.
;

elaborate,

and

erect

amid the Theban

A VINDICATION OF NIEBUHR'S HISTORY OF ROME

From
the Charges of the Quarterly Review. By Julius Charles Hare, M.A., Fellow of Trinity College, Cambridge. 8vo, 2*. 6d.

BRIEF OUTLINES OF DESCRIPTIVE GEOGRAPHY,

To which is subjoined a Table of Latitudes and Longitudes. By Henry H. Foolscap 8vo, with Maps, 2s. 6c?. cloth. " The Outlines here presented to the Public, are the results of actual extemporaneous Lessons." Preface.
Davis.

GEOGRAPHY SIMPLIFIED

Being a brief Summary of the principal Features of the Four Great Divisions of the Earth, with a more detailed Account of the British Empire. Also, a familiar Explanation of the Use of the Globes, and an Appendix. By an Experienced Teacher. 12mo, 4s. bound.
III.

OUTLINE MAPS.
Mercator
Three Maps,
Europe.
British Isles.
Single Maps, id. each.
folio, stitched in cover, Is.

GEOGRAPHICAL PROJECTIONS.
Mercator.
Three Maps,

Europe.

British Isles.
Single Maps, id. each.

folio, stitched in cover. Is.

'.jx/<,sv;,--<'i>ffr'

14

DAY

USE

RETURN TO DESK FROM WHICH BORROWED

LOAN
This book
is

DEPT.

due on the last date stamped below, or on the date to which renewed. Renewed books are subject to immediate recall.

a9}{ov'&7fiXX

17

2002

J^Erc'D

us

^AN

9 1958

20Feb'59MR

REC^D LD
tll^RARYluiea

NOV^i

1962

m^
:3i

m;^

f/^^
2001

oc
LD
(C8481sl0)476B

1 ?

mi
General Library University of California Berkeley

21A-50m-8.'57

json.-io-j!|vwAR3~^ %

.A