(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Biodiversity Heritage Library | Children's Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
See other formats

Full text of "The Elements of Algebra Preliminary to the Differential Calculus: And Fit for the Higher Classes ..."


This is a digital copy of a book that was preserved for generations on library shelves before it was carefully scanned by Google as part of a project 

to make the world's books discoverable online. 

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject 

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books 

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover. 

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the 

publisher to a library and finally to you. 

Usage guidelines 

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the 
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing tliis resource, we liave taken steps to 
prevent abuse by commercial parties, including placing technical restrictions on automated querying. 
We also ask that you: 

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for 
personal, non-commercial purposes. 

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine 
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the 
use of public domain materials for these purposes and may be able to help. 

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for in forming people about this project and helping them find 
additional materials through Google Book Search. Please do not remove it. 

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just 
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other 
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of 
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner 
anywhere in the world. Copyright infringement liabili^ can be quite severe. 

About Google Book Search 

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers 
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web 

at |http: //books .google .com/I 




GAK, FraCcBsor of MatheiDBtici in University College, London. 
Fourth Editiaa. Rajal 12mo, it. cloth. 
FIRST NOTIONS OF LOGIC, Preparatorj to the Study of 

Geometry. By Professor De Morgan. Royal 12ma, la. 

THE ELEMENTS OP ALGEBRA, Preliminary to the Differential 

■--■■■■■■'■■ " "-1 — ' — :- nhioh 

or De 

l-Jmo, with DumeroDg Diagrams, 6s. cloth. 
GEOMETRICAL SOLIDS, to illostrate Runsr'- Lbbsons on- 
FoRU, end other vorks on Geometrr. The Set of 19 in a 
Boi, yj. 

LESSONS ON NUMBER, u given in n Peatalozsian School Cheam, 
Surrey. By C. RBi:{Ea. Second Edition, conusting of — 
ThkMastkr'h Mas'ual. 12mo,4j.6rf, cloth. ■( 
Tbb Scholar's Pa axis. I 


38, UPl'ER qoweH BTREET. 











^' WC^ai a htntBtt t]^at onelp ti^png is, to i^anc \^t toiiU toi^cttctr antr 
sf^arpenct), 1 nta^t not traucU to tieclare, sft]^ all men confesse it to tc as 
greate as maie te. 1£xcepte any tDttUssc persons t^inkt J)t mai< tee to totse. 
But ]^e ti^at moste featet]^ t]^at, is leaste in tyannger of it. ^]^erefore to 
condntle, 3E see moare menne to acbnotDle'Dge t]^e beneftte of nomber, tl^an IE 
can espte toillpng to stut^ie, to attaine i^c tenefites of it. i^tany praise it, 
tut fetDe tiooe greatly practise it : onlesse it tec for \^t iiulgare practice, 
concernpng jj^ercj^aunties tratle. Wi^txtin x^t tiesire ant) ]^ope of gain, 
ta&iittif many ixiillpng to sustaine some trauell. Jpor aitie of to^om, 3E tfitr 
sette fortf) t^t firste parte of Arithmetike^ ISut if ti^ei iinetoe f^oto fane t]^is 
secontfe parte, tiooetf) exccll t]^e 0rste parte, t]^ei tDOult) not accoumptr any 
tyme loste, t|)at ioerc imploiet) in it. Yea ti^ei tooultf not t]^infu any tyme 
toell bestotDet), till i^ti l^atr gotten soci^e i^atilitie ty it, t]^at it migl^t te t]^eir 
aitre in al oti^er stutJies."— robert recorde. 





a At si 








In the title-page I have endeavoured to make it clear that 
it will be impossible to teach algebra on the usual plan by 
means of this work. It is intended only for such students as 
have that sort of knowledge of the principles of arithmetic 
which comes by demonstrationy and whose reasoning faculties 
have therefore already undergone some training. 

Algebra, as an art, can be of no use to any one in the 
business of life ; certainly not as taught in schools. I appeal 
to every man who has been through the school routine 
whether this be not the case. Taught as an art it is of little 
use in the higher mathematics, as those are made to feel who 
attempt to study the differential calculus without knowing 
more of its principles than is contained in books of rules. 

The science of algebra, independently of any of its uses, 
has all the advantages which belong to mathematics in 
general as an object of study, and which it is not necessary 
to enumerate. Viewed either as a science of quantity, or 
as a language of symbols, it may be made of the greatest 
service to those who are sufficiently acquainted with arith- 
metic, and have sufficient power of comprehension, to enter 
fairly upon its difficulties. But if, to meet the argument that 
boys cannot learn algebra in its widest form, it be proposed 
to evade the real and efficient part of the science, whether by 
presenting results only in the form of rules, or by omitting 
and taking for granted what should be inserted and proved. 


for the purpose of making it appear that something called 
algebra has been learned: I reply, that it is by no means 
necessary, except for show, that the word algebra should find 
a place in the list of studies of a school ; that, after all, the 
only question is, whether what is taught under that name be 
worth the learning ; and that if real algebra, such as will be 
at once an exercise of reasoning, and a useful preliminary to 
subsequent studies, be too difficult, it must be deferred. 
Of this I am quite sure, that the student who has no more 
knowledge of arithmetic — that is, of the reasoning on which 
arithmetical notation and processes are built — than usually 
falls to the lot of those who begin algebra at school — that is, 
I believe, begin to add positive and negative quantities to- 
gether, — will sooner find his way barefoot to Jerusalem than 
understand the greater part of this work. And I may say the 
same of every work on algebra, containing reasoning and not 
rules, which I have ever seen ; provided it contained any of 
the branches of the subject which are of most usual appli- 
cation in the higher parts of mathematics. 

The special object to which this work is devoted is the 
developement of such parts of algebra as are absolutely 
requisite for the study of the differential calculus, the most 
important of all its applications. The former science is now 
so extensive, that some particular line must be marked out 
by every writer of a small treatise. The very great difficulty 
of the differential calculus has always been a subject of 
complaint; and it has frequently been observed that no 
one knows exactly what he is doing in that science until 
he has made considerable progress in the mechanism of its 
operations. I have long believed the reason of this to be 
that the fundamental notions of the differential calculus are 
conventionally, and with difficulty, excluded from algebra, in 
which I think they ought to occupy an early and prominent 
place. 1 have, therefore, without any attention to the agree- 
ment by which the theory of limits is never suffered to make 


its appearance in form until the commencement of the dif- 
ferential calculus, introduced limits throughout my work : 
and I can certainly assure the student, that, though I have 
perhaps thereby increased the difficulty of the subject, the 
additional quantity of thought and trouble is but a small 
dividend upon that which he would afterwards have had to 
encounter, if he had been permitted to defer the considera- 
tions alluded to till a later period of his mathematical course. 
On those who offer theoretical objections to the introduction 
of limits in a work on algebra lies the (mus of shewing that 
they are not already introduced, even in arithmetic. What 
is k2, supposing geometry and limits both excluded ? 

I have been sparing of examples for practice in ' the 
earlier part of the work, and this because I have always 
found that manufactured instances do not resemble the 
combinations which actually occur. They are but a sort of 
parade exercise, which cannot be made to include the means 
of meeting the thousand contingencies of actual service. 
The only method of furnishing useful cases is to take some 
inverse process, and the verification of literal equations (as 
in the seventh and following pages of this work) is the most 
obvious. With these the student can furnish himself at 
pleasure, the test of correctness being the ultimate agreement 
of the two sides, after the value of the unknown quantity 
has been substituted. 

The only remaining caution which he will need is, not 
to proceed too quickly, especially in the earlier part of the 
work. He must remember that he is engaged upon a very 
difficult subject, and that if he does not find it so, it is, most 
probably, because he does not understand what he is about. 
Wherever an instance or a process occurs, he should take 
others as like them as he can, and assure himself, by the 
reasoning in the work, that he has obtained a true result. 

It was at first my intention to write a second volume on 
the higher parts of the subject. But, considering that of 


these there are s€*veral distinct branches, it has appeared to 
me that the convenience of different classes of students will 
be consulted by publishing them in several distinct tracts, 
which may afterwards be bound together, if desired. 



July ZUt, 1835. 

*^» This Second Edition differs from the first only in 
verbal amendments. Since the publication of the first edi- 
tion, I have carried on the consideration of the principles of 
algebra in two other works. In the first, or Elements of 
Trigonometry^ besides the consideration of periodic mag- 
nitudes in general, I have (Chapter IV.) given a view of 
that extension of the meaning of symbols which must ac- 
company the complete explanation of the negative square 
root. In the second, or Connexion of Number and Mctg^ 
nittuHe, which is an attempt to explain the Fifth Book of 
Euclid, I have entered upon what is in reality the most 
difficult part of the application of arithnietic to geometry. 
Both of the preceding works may be made supplementary to 
the present one, though the latter is altogether independent 
of it. 

Vniversity College, London, 
October 16,1837. 




































It is talcen for granted that the student vfho attempts to read this 
work has a good knowledge of arithmetic, particularly of common 
and decimal fractions. Whoever does not know so much had better 
begin by acquiring it, as the shortest road to algebra.* 

In arithmetic, we use symbols of number. A symbol is any sign 
for a quantity which is not the quantity itself. If a man counted his 
sheep by pebbles, the pebbles would be symbols of the sheep. Our 
symbols are marks upon paper, of which the meaning of every one is 
determined so soon as the meaning of 1 is determined. If we are 
speaking of length, we choose a certain length, any we please, and 
call it 1. It may have any other name in common life, for instance, 
a foot or a mile, but in arithmetic, when we are numbering by means 
of it, it is 1 . We now introduce the sign +, and agree that when 
we write -f between two symbols of quantity, it shall be the symbol 
for the quantity made by putting these two quantities together. 
Thus, if 

1 stand for the length ■ 
1+1 stands for the length ' 

1 + 1 is abbreviated into 2, a new and arbitrary f symbol. Similarly 

2 + 1 is abbreviated into 3,3 + 1 into 4, and so on. 

* The references in this work are to the articles (not pages) of my 
Treatise on Ariihitietict and serve either for the second or third editions. 

Arbitrary^ that is, any other would have done as well. It is S that 
stands for 1 + 1, and not <, 5» OO, or anything else, because certmn 
Hindoos chose that it should be so. See Penny CycXopsdia^ Art, 


When 1,2, 3, Sec., mean 1 mile, 2 miles, 3 miles, &c., or 1 pint, 
9 pints, 3 pints, &c., these are called concrete numbers. But when 
vre shake off all idea of 1, 2, &c., meaning one, two, &c., of any thing 
in particular, as when we say, *^ six and four make ten,'' then the 
numbers are called abstract numbers. To the latter the learner is first 
introduced, in regular treatises on arithmetic, and does not always 
learn to distinguish rightly between the two. How many of the 
operations of arithmetic can be performed with concrete numbers, 
and without speaking of more than one sort of 1 ? Only addition and 
subtraction. Miles can be added to miles, or taken from miles. 
Multiplication involves a new sort of 1, 2, 3, &c., standing for repe- 
titions or times, as they are called. Take 6 miles 5 times. Here are 
two kinds of units, 1 mile and 1 time. In multiplication, one of the 
units must be a number of repetitions or times, and to talk of multi- 
plying 6 feet by Z feet, would be absurd.* What notion can be 
formed of 6 feet taken " 3 feet " times ? 

But in solving the following question, " If 1 yard cost 5 shillings^ 
how much will 12 yards cost?" do we not multiply the twelve yards 
by the five shillings? Certainly not — the process we go through is the 
following : Since each yard costs five shillings, the buyer must put 
down 5 shillings as bflen (as many times) as the seller uses a one>yard 
measure; that is, 5 shillings is taken 12 times. 

In division, we must have the idea either of repetition or of 
partition, that is, of cutting a quantity into a number of equal parts. 
** Divide 18 miles by 3 miles,** means, find out how many times 
3 miles must be repeated to give 18 miles : but " Divide 18 miles 
by 3," means, cut 18 miles into three equal parts, and find how many 
miles are in each part. 

18 miles divided by 3 miles gives 6 ; meaning^ that 3 miles must 
be repeated six times to give 18 miles. 

18 miles divided by 3 gives 6 miles ; meaning, that if 18 miles be 
cut into three equal parts, each part is 6 miles. The answer in ab- 
stract numbers is the same in both cases; 18 divided by 3 gives 6. 

But now we ask, How many times does 12 feet contain 8 feet? 

* In old books the following is sometimes found. " What is 
^99. 19s. llfrf. multiplied by ^99. 19s. ll\d.V* The only intelligible 
meaning of this is as follows : If a stock of money is to be increased at 
the rate of ^99. 19s. llfd. for every £l in it, how much will that be 
when the stock itself is „£ 99. 19s. 11 Jd, 1 Let the student answer this. 


The answer is, more than once and less than twice; which is not 
complete, because we have not an adequate idea of parts of times, 
that is, parts of repetitions. In talking of ^imef, we use a figure of 
speech which we may liken to a machine which works by starts, each 
start doing, say eight feet of work, and which is so contrived that 
nothing less than a whole start can be got from it ; either 8 feet or 
nothing. It is plain that such a machine cannot execute 12 feet of 
work, or any thing between 8 feet and 16 feet. But, let us now sup- 
pose the machine to be made to work regularly, at 8 feet a minute. 
Let us still continue to call 8 feet a start, then 12 feet must be called 
a start and a half. In the same way we say that 12 feet contains 
8 feet a time and a halfy the notion of half a time being equivalent to 
that of repeating not the whole 8, but its half. 

When we speak of dividing one fraction by another in arithmetic, 

2 5 14 2 
this is what is meant; for instance, - divided by ~ gives — ; or ■- 

3 7 15 3 

5 14 
contains => tt of a lime. Let the learner study the following pro- 

'5 2 

If ~ of £1 were gained in a day, then - of a pound would be 
• o 

gained in -T-zofn day. 


If the signification of 1 be changed, so that what was - is now 1, 

then what was - is now — -. 
3 15 

5 2 14 

If - of the line A be the line B, then - of the line A is -~ of the 

line B. 

The want of a proper comprehension of such questions as the 
preceding is a great source of difficulty to most beginners in algebra. 
If the preceding pages be not readily understood, it is a sign that the 
reader Ls not sufficiently acquainted with arithmetic for his purpose in 
reading this work. 

The symbols of arithmetic have a determinate connexion; for 
instance, 4 is always 2 -f- 2 whatever the things mentioned may be, 
miles, feet, acres, &c. &c. In algebra, we take symbols for num- 
bers which have no determinate connexion. As in arithmetic we 
draw conclusions about 1, 2, 3, &c., which are equally true of 1 foot, 
2 feet, &C.9 1 minute, 2 minutes, &c. ; so in algebra we reason upon 



numbers in general, and draw conclusions which are equally true of 
all numbers. This, at least, is one great branch of algebra, and ex- 
hibits it in a view most proper for a beginner. 

But this is a definition in a few words, and can only be understood 
by those who have already studied the subject. No science can be 
defined in a few words to one who is ignorant of it. We shall begin 
by giving an instance of a general property of numbers and fractions. 

Take (8) units,* and the fraction which, taken (8) times, gives a 

unit, t^ is the (8)th part of a unit. Add 1 to both ; this gives 9 and 1-. 


The first contains the second (8) times. Take (^) ^^^ unit, and the 

(2\ ^ 1 

■xj of a time gives a unit, ti^ 1 n* ^^^ ^ ^^ 

2 1 ^2\ 

both, this gives 1- and 2-. The first contains the second [-) of a 

time. Try the following, in which it will be found that the blanks 
may be filled up with any one number or fraction at pleasure. 

Take ( ) units, and the number or fraction which repeated ( ) 
times gives a unit. Add 1 to both ; then the first result will con- 
tain the second ( ) times, or parts of times. 

The following are instances which should be tried. 

Number or fraction 

: ) 


repeated ( ) 


gives a unit. 













1 added to 
the first. 



1 added to 
the second. 


9" " 



Number of times 
which the third 
contains the fourth. 







The connexion between the first and second columns is this, that 
the number or fraction in the second is 1 divided by the number or 

* By putting 8 in brackets, we wish to call attention to the circum- 
stance of the numbers in the different brackets being the same. 


fraction in the first ; or the number of times or parts of times which 
1 contains the first. That is, if we call the number in the first column 

" the number," 

then the number in the second column is 

1 divided by " the number." 

And the coincidence of the first and fiflh columns (which constitutes 
the thing we notice) may be thus expressed : 

Let one more than " the number" be divided by " 1 more than 
the times which 1 contains the number," and the result must be 
** the number." 

The above must be still fiirther abbreviated for convenience. As 
^* the number" means any number we please, and as ** any number 
vfe please" will be better expressed by any short symbol which we 
may choose to make use of, let one of the letters of the alphabet be 
employed, say a. Let the addition of a number be denoted by + as 
before, and let the division of a number by a number be denoted, as 
in arithmetic, by writing the divisor under the dividend with a line 
between them. Let ^ be the sign that what^goes before is the same 
number as what comes aftier. Then the preceding property of num- 
bers is thus expressed : 

We shall now proceed to lay down the definitions of the first part 
of the science. 

I. Algebra is the European corruption of an Arabic phrase, 
which may be thus written, al jebr e al mokabalah, meaning restor- 
ation and reduction. The earliest work on the subject is that of 
Diophantus, a Greek of Alexandria, who lived between a.d. 100 and 
A.D. 400 ; but when, cannot be well settled, nor whether he invented 
the science himself, or borrowed it from some Eastern work. It 
was brought among the Mahometans by Mohammed ben Musa 
(Mahomet, the son of Moses) between a.d. 800 and a.d. 850, and 
was certainly derived by him from the Hindoos. The earliest work 
-which has yet been found among the latter nation, is called the Vija 
Ganita, written in the Sanscrit language, about a.d. 1150. It was 
introduced into Italy, from the Arabic work of Mohammed, just 




mentioned, about the beginning of the thirteenth * century, by Leo- 
nardo Bonacci, called Leonard of Pisa: and into England by a 
physician, named Robert Recorde, in a book called the Whetstone 
of WittCf published in the reign of Queen Mary, in 1557. From this 
vfork the motto in the title-page is taken. 

II. A letter denotes a number, which may be, according to 
circumstauces, as will hereafter appear, either any number we please ; 
or some particular number which is not known, and which, therefore, 
has a sign to represent it till it is known ; or some number or fraction 
which is known, and is so often used that it becomes worth while to 
have an abbreviation for it. Thus the Greek letters ir and e always 
stand for certain results, which cannot be exactly represented, but 
which are nearly 3-1415927 and 2*7182818. 

III. The alphabets used are 1. The Italic small letters; 2. The 
Roman capitals ; 3. The Greek small letters ; 4. The Roman small 
letters ; 5. The Greek capitals. They are here placed in the order of 
their importance on the subject; and as many may wish to learn 
algebra, who do not know Greek, the Greek alphabet is here given, 
with the pronunciation of the letters. 

A tc alpha N y 

BfiQ beta H | 

r V gamma O • 

A ^ delta Tl 7r m 

E 8 epslloa F ^ 

Z ^ ^ zeta S (T 5 

H n heta T t f 

6 theta T v 

1 I iota O 
K K kappa X ^ 
A A lambda Y 4' 
M /M mu ft V 

IV. Under the word number is always included vohjole numbers 



6 micron 









* The young reader may need to be told that the thirteenth century 
does not mean a.d. 1300 and upwards, but^.D. 1200 and upwards. The 
first century is from the beginning of a.d. 1 to the end of a.d, 100 ; the 
second from the beginning of a.d. 101 to the end of a.d. 200, and so on. 

• • 


?ixid fractions. Thus, 2^ is not called a number in common language, 
but in algebra it is called a number; or, if it be necessary to dis- 
tinguish it from 2, 3, 4, &c., it is called a fractional number, while 
the latter are called whole numbers. 

Under the word number the symbol may be frequently in- 
cluded, which means nothing, or the absence of all magnitude or 
quantity. If we ask, what number remains when b is taken from a, 
and if we afterwards find out that b and a must stand for the same 
number, then the answer is, '' no number whatever remains, or there 
is nothing left." If we say that remains, we make the symbol 
an answer to a question beginning, " What number, &c. ? " which is 
equivalent to including under the general word number, 

V. The sign + is read plus (Latin for more), means in correct 
English* <' increased by," and signifies that the second-mentioned 
number is to be added to the first. Thus a-\-b \s read a plus 6, 
and means a increased by b, or the number which is made by adding 
b ijo a. 

+ a by itself can only mean a added to nothing, or 0-\-a, which 
is a itself. 

VI. The sign — is read minus (Latin for less), means ^^dimi- 
nished by," and implies that the second number is to be taken away 
from the first. Thus a — b is read a minus b, means a diminished by 
b, and signifies that b is to be taken away from a. 

When a is less than b, the preceding stands for nothing at all, but 
a direction to do what cannot be done. Thus 3 — 6 is impossible. 
We shall hereafter have to investigate what such an answer means, 
that is, the problem which gave it being of course impossible, what 
sort of absurdity gives rise to the impossibility. - 

VII. The sign x is rendered by the English word into, and 

* Those who have attempted to substitute a short English term 
have never been able to find one which does not shock the ear. We 
have seen "a add b," and ** a more b;'* which should be, ''To a add 
b" or **b more than a." Neither of the last would be convenient, 
and "a increased by b" is too long. The same remarks may be made 
on "a less b" and "o take away 6," for ** a diminished by 6;" while 
*• a save b" used by our oldest writers, is now too uncommon. There 
is no language in which the simplest relations are expressed by the 
simplest terms. 


means ** is the multiplier of/' meaDing that the second number is to 
be taken as many times, or parts of times, as there are units, or parts 
of units, in the first. Thus axb is read a into b, and means 6 taken 
a times. Thus, 1^x6 is 6 taken a time and a half, or 9. The 
expression a 6 is called the product of a and b; the two latter are 
called factors of the product, and coefficients of each other. x a 
and a X must both mean 0; for a taken no time, nor part of a time 
whatsoever, cannot give any quantity, and nothing, however often 
V X/^ repeated, yields nothing. As this is connected with a mistake alwajrs 
made by beginners, we shall (to impress the results on his memory) 
give two problems* the answer to which is self-evident, and put the 
answers in an algebraical form; desiring the learner, of course, to 
remember that no new information is gained, but only an opportunity 
of making certain results occupy a space proportional to their 

There is a number of boxes, none of which contain any thing. 
How much do all together contain? 

If CI be the number of boxes, then repeated a times, or 
a X is 0. 

There is a box full of gold, of which no part whatsoever belongs 
to A. How much belongs to A ? 

If j7 be the number of pounds of gold in the box, then A*s part is 
X p, or 0. 

The reason of the mistake is, that the beginner retains the notion 
that ^* not multiplied*' implies ** not diminished,*' and <' not changed 
at all." But multiplication in the arithmetic of fractions, and in 
algebra, means the taking a number of times, or parts of times, A 
number not multiplied at all yields no number, for no part of it is 
taken; a number which remains the same is multiplied by l,or taken 
once. Thus a is 1 x ^'^ 

When letters are employed, or numbers and letters together, the 
sign X is dropped. Thus a 6 means b taken a times; 3 a means a taken 
3 times. It is only when two numbers are employed that x becomes 
necessary. Thus 3x6 must not be written 36, which already stands, 
not for 6 + 6 + 6, but for 3 X ten + 6. It is more common to place 
points between numbers (thus, 3.6) to signify multiplication. But 

* Problem, any question in which something is required to be found. 


as this may be confounded with 3 + -—, the student should always 

write the decimal point higher up, thus 3*6. 

VIII. The bar between the numerator and denominator of a 

fraction is read '* £»y/' and this is the word for division. Thus -r 


is read ^' a by by* and means that it is to be found what number of 

times and parts of times a contains b. Thus '*-, or 3 by 2, is 1^/' 

means that 3 contains 2 a time, and half a time. That a half is 

written -, or ** 1 by 2," is consistent, because it is the part of a time 

which 1 contains 2. The division of a by 6 will sometimes be 
denoted by a^b, 

- (which beginners usually confound with a) has no meaning. 

How many times does six contain nothing ? The answer is, that the 

question is not rational, a is -, for by a we mean a number of 

units and parts of units, so that a itself is the answer to " how many 
times, and parts of times, does a contain 1 ? '' 

IX. The following are then synonymes for a, which the student 
should repeat till he is very familiar with them : 

o + O axl aXlXl -r- &C. 

fl X 1 
a — IXO IXflXl — T- &c. 

X. The following abbreviations are used for the connecting words 
equals and therefore. By a := 6 we mean that a and b are the same 
numbers : it is read a equals 6. By ,•, we mean therefore, or then, 
or consequently. Thus, a ^ b and 6 = c Z, o = c is read a equals 
b, and b equals c, therefore a equals c. 

XI. Every collection of algebraical symbols is called an expression, 
and when two expressions are connected by the sign =, the whole is 
called an equation. 

An identical equation is one in which the two sides must be 
always equal, whatever numbers the letters stand for, such as the 
equation already described (page iv), 

o + l 


= a 


which cannot be untrue for any value of a; or is true, whatever 
number we may suppose a to stand for. 

The following are very obvious identical equations: 

a + 6 = 6 + a a + l + l=a + 3— 1 

^a + -a = a 3^ + 3^ + 3^ = ^ 

2a + 3a — a + 4a = 8a + 6a + 5a— 11a 

The following are not obvious, but will be found to be true in 
every case in which they are tried. 

1 1 _ 2 + 3a 

l+a"*"l + 2a l+3aH-2aa 

a ax 

An equation of condition is one which is not dtSBJBitrue for every 
value of the letters, but only for a certain number of values. Thus 

i + 1 == 7 and a-3 = 12 

cannot be true unless 6 is 6 and a is 15. 

Again, a ^=i b-^-c cannot be true for every value of a, 6, and c, 
but requires or lays down the condition that a must be the sum of 
b and c. When a number may be written for a letter in an equation, 
and it remains true, that number is said to satisfy the equation. 
Thus, a — 3 = 12 is satisfied by a = 15. 

XII. When an algebraical expression is enclosed in brackets, it 
signifies that the whole result of that expression stands in the same 
relation to surrounding symbols as if it were one letter only. Thus^ 

a — (6 — c) 

means that from a we are to take, not b, or c, but 6 — c, or what is 
left after taking c from b. It is not, therefore, the same as a — b'-^c. 

Example. What is a — (b — [c — d]) when a = 20, 6 ^ 12, 
c = 10, and d = 3. Here c — d is 10 — 3, or 7; b^^^c^d) is 
12 — 7, or 5 ; a — (b — {c — d}) is 20 — 5, or 15. 

Also, (a + ^) (c + (Q means that c -f (/ is to be taken a + 6 times, 
and p{q +r) that ^ + r is to be taken p times. 

These are the first outlines of algebraical notation. Others will 
develope themselves as the work proceeds. We shall now lay down 
some very evident characters of the four principal processes. 


1 . Additions may be made in any order, without affecting the 
result. It is evident that all the following six expressions are the 
same : 

1+2+3 1+3+2 
2+3+1 3+2+1 
3+1+2 2+1+3 

Also a + b + c + d = b+c+d+a = 6+a+rf+c, &c. 

2. The order of additions and subtractions may be changed in 
any way which will not produce an attempt to subtract the greater 
from the less. Thus, if a man lose £20 and gain £50, he has the 
same as if he first gained £50 and then lost £20, with this exception, 
that if his property be less than £20, he may do the second but cannot 
do* the first. Thus 10—20 + 50 is impossible ; but 10+50 — 20 
is possible. Also 8 — 6 + 10 — 11 admits of the following forms : 

8- 6 + 10-11 8 + 10-11- 6 10 + 8-11- 6 
8 + 10- 6-11 10+ 8- 6-11 10-6+ 8-11; 

but does not admit of the following : 

10-11+8-6 -11+ 8 + 10- 6 &o. 

10-11-6 + 8 - 6-11+ 8+10 &c. 

Question. What are the conditions necessary to the possibility 
of a — 6 + c — d? Answer. That a be greater than (or at least not 
less than) b, and that a — 6 + c be not less than d. We shall here- 
after return to this point. 

3. Multiplications and divisions may be performed in any order. 
For instance, abc means that c is to be taken b times and the result 
a times : the rules of arithmetic shew that this -is the same as bac ; 
that is, as c taken a times, and the result taken b times; and this 
whether the numbers be whole or fractional. The student is supposed 
to have demonstrated these rules before he commences algebra. 

It is also shewn in arithmetic that division and multiplication 

* According to the language of common life a man may lose more 
than he has, that is, lose all that he has and incur a debt besides. But 
this is always on the tacit supposition that not only what he has, but 
what he may get, is liable for his debts. No such supposition exists in 
arithmetic until the meaning of words is altered 5 10—20 is, in every 
arithmetical interpretation, impossible. 



may be changed in the order of operation; that is, a divided by 6 and 

the quotient multiplied by c, is the same as a multiplied by c and the 

product divided by h. Or 

a ca 

In fact, owing to the extension of the meaning of words by which 
multiplication includes taking parts of times (page iii), every multi- 
plication is a division, and every division a multiplication. For 
instance, to divide a by six is to take the sixth part of a, or to take 

a one sixth of a time, or to multiply a by -. Similarly, to divide a 

by - is to ask how many fourths of a unit a contains ; the answer to 

which is, four times as many times as a contains units, which 

multiplies a by 4. 

2 3 

Again, to divide 10 by - is to multiply 10 by -. To divide 


10 by - is to find how often 10 contains two-thirds of a unit. Now 

^ 3 

2 1 

a unit is made up of - and -, the second of which is half the 

first; that 1 contains -, a time and a half. Therefore, 10 contains 
/v ^ 3 

two-thirds ten times and ten halves of times, or 15 times. That is, 10 

2 1 
divided by - is 10 taken once and a half or 10 multiplied by 1~, 

3 2 

that is, by -. 

7 2 

Similarly, to divide by - is to multiply by -. How often does 10 

contain the half of 7 units ? Twice as often as it contains 7 units ; 

10 20 20 2 

tliat is, twice -^r of a time or -— of a time. But rr- is - of 10, or 10 
7 7 7 7' 

taken — of a time. 

The learner should practice many different cases of the following 

general assertions. 

a multiplied into - is a divided by - 

9 P 

a divided by ~ is a multiplied into - 

9 P 

pa a q 

9 i S. p 

p « 


The first operation to which the beginner must be accustomed is 
the conversion of algebraical expressions into numbers, upon different 
suppositions as to the value of the letters. For instance, what is 

2±* when a=l and 6 = ? • 
a— 6 2 5 

1 +afl aaa , 1« 

Is —- ; — = - . , true when a s 1 - f ^ 

l+« 2 + ia 2 3 i/ ; 












339 , . , . 9 13 


aa = ~ Xr = T 

2 2 

1 ■ 1 . J' *«^ 

333 27 1 3o.l 11 

aaa =-x— X— = — —a =5— I&H — a ^ — 
2^2 2 8 2 4 ^2 4 

/o . 1 \ 27 11 27 

But (l+tfa)-^(l+a) =H 

therefore the above equation is not true in this case. 

What is aa + h(a + I) when a = 4 6 = 3? 

a + 6 = 7 6(a + i) = 3x7 = 21 aa = 16 
aa + &(a + J) = 16 + 21 = 37 

But the most instructive exercise is the verification of equations 
which are asserted to be identical. For instance, 

aa — hb aaa-\'bbb 

a — 6 aa-\-bb-~^ab 

Let a = 4 6=2, then aa = 16 6t = 4. 

a — b 2 ^ ' 

flOfl + fe66 __ 64 -t- 8 _ 72 _ /ov 
«a + 66 — fl6 — 16+4—8 "" 12 "~ ^^ 

2 ^ 1 
Letfl = - & = -. 



fl« — bb 

a — bb ^^ _9 4_ _ 36 /7\ 

;J3T " !._L "" J. - w 


aaa-f-666 __ 27 ' 8 __ 216 _ /7\ 
^"+6S^Z^ — 1'. i_i "■ il ^ VT/ 

^ 9 ^4 3 36 

The following is a list of identical equations, which the learner 
should verify; first, by giving whole values to the letters, next, by 
giving fractional values. It is needless to give instances of each, 
because the test of correctness is in the two sides shewing the same 
value, no matter what. The only restriction upon the values of the 
letters is that no values must be assumed which will produce in the 
equation an expression for the subtraction of the greater from the less, 
or which will produce in the denominator of a fraction (pages vii,ix). 

{a +x +y)(a + x-^y) = aa-{-2ax + xx-^yy 

(aH-6)(a-fA) = aa + 2ab-\-bb 

(« — h) {a-^b) = aa — 2a6 + JJ 

(mm^nn) (rnm-^nn)+4mmnn = (mm-{-nn) (mm-rnn) 

(a+b) (a + &) + (a— i) (a— J) = 2aa + 2bb 

(a+b) (a + 6)— (a— J) (a— 6) = 4ab 

(a+A+c) (a+i— c) (b+c-^a) (c+a— J) 

= 2aabb+2bbcc+2ccaa--aaaa--bbbb^cccc 

(a + J) (a+b) (a + ft) = aaa+3aab-^3abb-^bbb 

1 1 1 3aa + 2 — 6a 

a a — 1 a — 2 000-^-20 — 3aa 

a'\-b a — b 2aa-j-2bb 

u^b a-^-b aa — 66 


xxxx — yyyy ^^^'\'^3£y'\-xyy •\-yyy 

xxx^^yyy ^x-\- xy-\-yy 


\ X-^^ ] x-f- 6 — jrjr + (6+c)jr + 6c 

The reduction of several terms into one can be performed to as to 
produce a more simple term, when all the terms are alike as to letters. 

3a + 2a = 5a a + 7a — 4a = 4a 

3aJ + 2a6 = 6aft £ + 7? - 4^ = 4^ 

9 9 9 9 

3xx '{-2xx = 5xx aab + 7aab-^3aab = 5aab 

a + 12a — 3a — 6a + 2a — a = 6a 

In the last instance, all the additive terms make up 15 a, from which 
d is to be subtracted three times, six times, and once, or 10 times in 
all; that is, 10a is to be subtracted : and 15a — 10a s= 5a. 

Similarly a-^b — Za + 4 b = 5 b — 2 a. For b and 5b added /y ^ 
give 6b, which is added to a, after which 3a is subtracted. But, 
subtracting 3 a where a had previously been added, is the same as 
subtracting 2 a without previously adding a, which gives 

a+66 — 3a = 66 — 2a 


a-haJ — 2aJ-f 4a4-6a = 11a — aft 
2xx + 6x^4x'-'Xx + c = xx + 2x + c 
3a;— 16 + Ja:-a;-7 = 2ja:-22 
a? + y + a; — y -f 3x = 6x 

If, upon looking through such an expression as either of the 
above, we find the following terms containing the simple product xi/ 
(veith their signs) 

+ 6a;y, — a;y, + 4a:y, +2xy, —Uxy, — 12a:^, 

the single term which represents the result of all those containing jci/, 
is found as follows; the whole number of additions oi xy to the 
rest of the expression amounts to 12, and there are 24 subtractions of 
it in all. Let all the additions, and as many of the subtractions, be 
abandoned, there will then remain 12 subtractions uncompensated by 
any additions, and — 12 xy must be appended to the rest of the 

When two terms are not exactly the same as to letters and number 


of letters which enter them, no sach simplification can take place. 
For example, aa-\-a cannot be reduced. It is not* 2a, nor 2aa, nor 
aaa; it is a taken a times added to a taken once, and is therefore a 
taken a + 1 times, or (a + ^* '^^ ^^ ^^ (<* + 1) taken a times, 
or a(o + 1), page xi. 

Thus a -^ a is algebraically reducible to the single terra 2 a, but 
aa -\' a is not so reducible, nor does it admit of any simplification of 
form, except that which is contained in the formation of its arith- 
metical value, which cannot be found till we know what number a 
stands for. 

How many times is a contained in a -jr b? This question admits 
of no answer till we know how many times 6 contains a; therefore 

I L 

we can only use the algebraical representation , which has been 

chosen to signify the number of times which a-\-b contains a. But 
let the student observe that this is not an answer to the question, but 
a method chosen to represent the answer. 

How many times is a contained in ma — na? Here, though 
we cannot completely answer this, till we know what m and n stand 
for, yet the algebraical meaning of ma and na puts our algebraical 
answer one step nearer to the arithmetical answer, that is, enables us 

to answer otherwise than by directly writing down . For it 

is evident that when n times is taken away from m times, the re- 
mainder is m — 71 times; that is, ma — na contains a, m — n times. 
This must be taken notice of in all algebraical operations. The 
question, "What is 8a + 5a?" cannot be answered until we know 
what a stands for; but to ** What is the most simple algebraical form 
of 8fl -f 5 a?" we answer 13 a. And by the algebraical operations of 
addition, subtraction, multiplication, &c. we mean the methods of 
changing algebraical expressions into others which are Of more simple 
character. For instance, " to add a + 6 to a — 6." The following 

(a + 6) + (a - b) 

is the first form of the result, derived from representing the thing to 
be done under algebraical symbols. But its most simple form is 2a; 
and in the reduction of the preceding expression to 2a, consists what- 

* These are all mistakes to which the beginner is liable. 


we shall call algebraical addition. We shall now go through several 
of these processes. 

Addition. We wish to add a + 6 to c -f e. It to a -\- b we 
add Cy giving a + ^ + <^9 we have not added enough, because the 
quantity to be added is not c, but e more than c. Therefore 
a + ^ + c-|-e is the result, or* 

{a + b) + {c + e) = a + b + c + e .... (1) 

To add c — c to a + 6, we first add c, giving a-j-b + c. But this 
is adding too much, for e should have been taken from c and the 
remainder only added. Correct this by taking e from the result, 
which gives a -f 6 + c — e. 

(a + J)-f (c — e) = a + b + C'-e .... (2) 

In the results of (1) and (2) further reduction is impracticable. 
We shall now try the following. 

To add a -j- b to a — b, 

(a-f J) + (a — ft) = a + b-\-a^b = 2a 
To add 3jr — a to 2a — jt, 
(3a: — a) + (2a ^x) = 3x'-a + 2a — x = 2x -{- a 

To add ab — b to 2ab-\-c — 6b, 

Am. 2ab + c-^6b + ab — b or 3a&+c— 76 

From these, and similar operations, we have the following rule of 
addition : 

Write + before the first terms of all the expressions but one, and 
consider the aggregate of all as one expression. Make the reductions 
which similarity of terms (as to letters) will allow. 

Examples. <X — J + 3c — ab 

Aab^ b +2a — a: 
4a; + 6a + a J — 7 are to be added. 

Anmer, 9a — 2i + Aab + 3c + 3ar — 7 

• When we wish to refer to an equation afterwards, we place a 
letter or number opposite to it, as is here done. 

c 2 







a +ma'^4 



6— a +2ma 


c — 2c? 

12wa— 12 — 3a 


rf — 2a: 

;?+ a — J 

a — x a — J — c — d — 2a: 15»ia — 2a— lOJ +j9 

The following rule is also derived from (1) and (2) : 

Art expression in brackets preceded by the sign + will not be 
altered in value if the brackets be struck out, 

a + (6 + c — e) = a +b +€-^6 

Subtraction. — To subtract 6 -f c from a. If we first subtract b, 
which gives a — b, we do not subtract enough, since b should have 
been increased by c, and the whole then subtracted. Hence c must 
also be subtracted, giving a^b — c, or 

a — (b-\-c) = a — b — c (3) 

To subtract b — c from a. If we now subtract b, giving a — 6, 
we have subtracted c too much ; or a — b is less than the result 
should be by c. Consequently, a — 6 -|- c is the true result, or 

a — {b — c) ^= a^b+ c • • • • (4) 

The following are instances : 

a — (c^a) = a-'C-\-a=:i2a — c 
a^(a^c) = a — a-f-c = c 
3a + i-(2a — i) = 3a + i — 2a + & = a -f 26 
a + ft — (a — i) = a + 6 — a + 6 = 26 
mx^^q — Smx) = mx ^ q + 3mx = 4mx — q 

When an expression in brackets is preceded by the 


— AND — INTO +• This is evident from (3) and (4). 

As the neglect of this rule is the cause of frequent mistakes, not 
only to beginners, but to more advanced students, we have printed it 
thus to attract attention. Still furtlier to impress it on the memory 
of our younger readers, we beg to inform them that to neglect this 


rule is the same as declaring that all debts are gains, and all property 
a loss ; that to forgive a debt is to do an injury, and that the more a 
inan is robbed the richer he grows ; with a thousand other things of 
the same kind. 

Another proof of the preceding rule is as follows. If we wish 
to find 

a — (i + C — p ^q) (A) 

we must remember that if two quantities be equally increased, their 
difference remains the same: thus the difference of a +j: and 6-fx 
is the same as that of a and b. We are told to diminish a by 

b -{-c—p^q 

which is the same as if we diminished «+(/> + ^) by 

(bJtC^p-^q) + {p+q) 
or by i + C — p — y + ^J + J or by 6-fc 

thai is, a — (6 + c — p — q) is the same as 

a+^-fy — (6 + c) or a -{-p-k- q — b — c 
or a — i — c+p+g' (B) 

on comparing (A) and (B) the rule is obvious. 

The rule for subtraction is as follows : 

Consider the first term of the expression to he subtracted as having 
the sign +. ; then change every sign, annex the expression thus changed 
to the expression which is to be diminished, and make all practicable 
red act ions J as in addition. 

From a + i — c — a: + 2z + 3ai — 14 

Take c — 2a + a: + 2r — 4a6 + 2J 

■ -"--I - ■ — ■■■■Jill - - , 

Rem'. 3a + 6 — 2c — 2x + zi^7a6 — 16J 

From a+c a;+y — 3 — a a — fe + c — d + e 

Take 2c — a a: — y + 3 — a a— 26 + c + d— e 

Remr. 2a — c 2y — 6 i — 2d + 2e 

From a + 6 + 2c + 3d + 4e — 5/— e^r 
Take lCd-f4e-3c + 2a + 6-5r+/ 

Rem'. 5c — 9d— 6/— S^r — a 


What is 

a-.(6-(c+^)) + (6-(a;-26)) 

This, by ihe rules for expunging brackets, is 

a — 6 + (c + a?) 4- i — (^ — 26) 
which, by the same rule, is 

a'-b + c + x-j-b^x + 2b or a + c+ 26 
/ , ^ Shew/ that A. - <- -f * 

since all the rules for addition and subtraction are independent 
of the order in which the terms of the expressions are written, we 
shall in future not inquire whether an expression is written in a 
possible or impossible form (page xi), but consider the impossible 
forms as meaning the same thing as the possible ones. Thus, in 
3 — 7 -}- 8, we shall not regard the subtraction as necessarily to be 
performed first, and therefore treat the expression as impossible, but 
we shall consider the order of the operations as immaterial, and the 
above as 3 4*3— 7 ; and the same for other expressions. We oould 
not have done this if the rules for addition and subtraction had 
re€[uired any preliminary inquiry into the possibility of the order of 
the terms. If we have to subtract the preceding, say from 12, we 
find that the employment of the impossible form, namely 

12-(3-7 + 8) = 12-3 + 7-8 = 8 

gives the same result as if the possible form had been employed, 
as in 

12-(3 + 8-7) = 12-3-8 + 7 = 8 

Multiplication and Division. — In the multiplication and 
division of algebraical quantities, it must always be borne in mind 
that the letters may represent either whole numbers or fractions. 
We shall first express the rules which have been found in arithmetic 
for the addition, &c. of fractions which have whole numerators and 

First observe that r- (when a and h are whole numbers) is the 

answer to all of the following questions, which are in effect the same. 


1. If a unit be divided into b parts, and a of those parts be taken, 
how many units or parts of units result? 

2. How many units or parts of units are there in the 6th part 
of a? 

3. How many times, or parts of times, does a contain b ? Thus, 


- representing the seventh part of unity repeated three times, is the 

answer to the questions, " What is the seventh part of three V* and 
'< How many parts of a time does three contain seven?'' 

It is only the method of speaking, or the idiom of our language, 
which prevents our explaining in a similar manner the meaning of 
fractions which h^ve fractional numerators and denominators. For 

a 4 

instance, if we attempt to explain t where a is 2^ and 6 is - in the 

same way as where a is 3 and b is 7, we shall produce a (yet) 
unintelligible idiom. Let us suppose some concrete unit, say a 
mile of length. 

3 3- 2\ 

What is - miles, or ~ of a What is ~ miles ? 

7 7 ^ 

mile ? 4 

Cut a mile into 7 equal parts ^""^ « '^'^ •'•^^ 9 ^^^<^^P^th 

and take 3 of them. and take 2^ of them. 

The words in italics are unintelligible, and not having a meaning, 
may have one given to them.* By altering the manner in which the 
first of the two preceding is expressed, without changing the meaning, 

we may find a mode of speech which shall not become unintelligible 

when - is written for 7, and 2 j^ for 3. As follows : 

"* The same letter may either stand for a whole number or for a 
fraction ^ and we might find out how to group those idioms which belong 
to whole numbers with those belonging to firactions, so as to be able to 
pass from one of the first set to the corresponding one of the second. 
But it will be more convenient to take the modes of speaking which 
belong to whole numbers, and agree that when fractions are spoken of 
they shall be used for the corresponding expressions. We have already 
done this in arithmetic in the word multiplication, which originally 
means, " taking a thing many times,'' but which we have also made to 
signify ** taking a thing a part of a time." Thus we speak of multiplying 
by one half. 


Find the length which taken Find the length which taken 

7 times gives a mile, and take 4 ^ ^ ^^^ ^^^^ ^ ^.j^^ ^^^ 

that length 3 times. ^ 

take that length 2^ times. 

4 45 
The reduced value of 2 J -7- r is — of a mile, or 5% miles, and 


the student may easily shew by the rules of arithmetic that this coin- 


cides with the length, - of which is a mile, repeated 2J times. And 


I : 5i is also the answer to the question, ** How many times and parts of 

' 4 ■ 

times does 2h contain - V* 
" 9 

It is usual in algebra to make use of modes of speaking with 

regard to letters, drawn from our notions of whole numbers ; but as 

the letters themselves may stand either for fractions or whole numbers, 

these modes of speaking are too confined unless they are understood 

to imply the corresponding modes of speaking, with regard to 

fractions. For instance, suppose it asked, ''What is the value of 

one acre if x acres cost y pounds?*' The answer would always be 

given as follows : — Divide the number y into x equal parts ; then 

e€K:h of those parts is the number of pounds which an acre costs. 

Thus, if 18 acres cost £36, since 2 is the 18th part of 36, £2 is the 

: I price of an acre. But if - of an acre cost £2-, and if we speak of 

2 • 3 

dividing 2- into - equal parts, we must mean the same as if we 

1 11 

said "take .2- two times," or divide 2- by - (according to the 

arithmetical rule). Now, though it may appear at first sight almost 

i ': ■ 1 

ridiculous to say that dividing a quantity into — equal parts is the 

same thing as taking it 10 times, we must remember, 1st. that we 
have done something of the same sort, when we said that dividing 

> into 10 equal parts, and taking one part, is multiplying by ---; 

2d. that we do say this, when we say that if x acres cost y pounds, 
; the price of one acre is found by dividing y into x equal parts ; for 
letters may stand either for fractions or whole numbers. 

We should recommend the student to solve various simple 



questions in the rule of three, with fractional data,* 1st. By imitation 
of a question with integral data, 2d. By independent reasoning ; as 
follows : 

I. If 6 yards cost 7 shillings, 
how much do 5 yards cost? 
As 6 yards to 5 yards, so are 

T'X.5 5 
7 shillings to -— • or 5- shil- 
lings, the answer. 

If - of a yard cost - of a 
3 "^ 7 

shilling, how much do - of a 

yard cost? 

2 4 

As - of a yard to - of a 
*i y 

yard so is - of a shilling to 

5 4 

7X0 10 
g or — - of a shilling, the 


2 , 5 

II. If - of a yard costs - 

of a shilling 

5 15 
2 yards cost ^ X 3 or -— 

1 yard costs -=- -7- -& or -— 

^ 7 14 

A J ^ 15 . 30 

4 yards cost 77 X 4 or —- 

4 r A ♦ 30 . Q 10 

- of a yard costs -=- -f- y or -- 

9 "^ 7 21 

We shall now write all the rules of arithmetic relative to fractions 
in an algebraical form^ these rules being those with which .the student 
is already familiar as applied to whole numbers. We shall first 
suppose the letters to denote whole numbers. 

a ma 

a c aa-\- be 

b mb 

6 ' rf ^ bd 

, b ac + h 
c c 

a e ad — be 
6 5"" bd 

b ac '— h 
c e 

a a^be 
6-*=- b 

* Datum, A thing given, that is, a number given as pert of a 
question. ** 2 pints cost 4 shillings, how much, &c. ;" here the data 
are* 3 pints and 4 shillings. 




a c 

I ^5 

— bd 





_ 6c 


c ad 
'^ d^ be 



X - 


The learner must make himself acquainted with each of these 
equations, which he will easily do if he have the requisite degree of 
familiarity with the operations of fractional arithmetic. For ezampley 

in T X J =: 7-}» he will recognise the rule, " To multiply one frac- 
6 d bd 

tion by another, multiply their numerators for a numerator, and their 
denominators for a denominator." 

The preceding rules all hold good when the letters stand for frac- 
tions. We shall take the demonstration of this in one instance, 

namely, t = — r- Let a be the fraction - let 6 be-, and let m be 
•^' 6 tnb q t 


- ; where /?, ^, r, s, x, and y, are whole numbers. 



^ = "7 = — (By the rule.) 



X p xp 1 X r xr 

y 9 y9 y 9 y 

ma ^^ yq ^yp* 

mb il "~ ^y^f 

But '^^' = ^^-^^^P' c= El 

^y9^ {^y)><9^ 9^ 

ma a 1 where o, 6, and m, are frac- 



mb 6 J tional. 

It is shewn in arithmetic that the order in which multiplications 
or divisions are made may be changed without affecting the result, 
both in questions of whole numbers and fractions. We need not, 
wiien speaking of both, distinguish between multiplication and 

division ; for division by a is multiplication by -. The following 


summary should be attended to and repeated on several products. 

The expression abed is the product, 1. Of a, 6, c, and d, in any 


order; 2. Of ab and cd; 3. Of ac and bd; 4. Of ad ana be; 
5, Of abc and(f; 6. Of a6dand c; T.Ofacdoind b; 8. Of ice? and a. 
The product of single terms may be expressed by writing the 
letters consecutively in any order which may be convenient, and 
actually multiplying the numerical coefficients, if any. The sign x 
becomes necessary between two numerals only. Thus, 2ab x ^cd 
might be written in tlie following ways : 

2ab4tcd 2 x Aabcd 8acbd Sabcd^ &c. 

It is most convenient to write the numerical coefficient first, and 
letters of the same sort together, the order of the alphabet being 
generally preferred. Thus, 

2aab x Sabbc is written daaabbbc 

1 3 

12abxx^abxx = 4Saabbxxx 3abcx-ab =-aabbc 

2 2 
When fractions are written side by side of each other, or of single 

letters, multiplication is intended to be denoted. Thus, 

■rC-^ means ZXtXCX:^ and is -r-p- 
of ^ J of 

[Though a, ab, a be, &c. are called integral expressions, and 

fl c ac 

Tf^^-Ti &c. /rac^itms, this refers to their algebraical, not to their 


arithmetical character ; for, since letters may stand for fractions, that 
which is integral considered algebraically, may be fractional con- 
sidered arithmetically, and viee versd. For example, suppose that 

a stands for - and 6 for -, then the algebraically integral expression 

1 /l 

a is the arithmetical fraction - ; while the algebraical fraction -, or 


: — is the arithmetical whole number 2. 


In speaking, therefore, of integral or fractional expressions, we 
refer always to their algebraical appearance, not to their arithmetical 
values. The latter depend on the particular arithmetical value of the 
letters, on which no supposition is made.*] 

* The part of algebra which treats of letters considered as repre- 
senting wluyle numbers only, is generally called the theory of numbers. 
It is very rarely, if ever, of use in the application of algebra to the 
Differential Calculus, and is therefore altogether omitted in this work. 



The multiplication of algebraical quantities depends on the rule 
wliich appears in the following equations : 

m^a + b) = ma + mb 

mia-^b) = ma—mb 

First, it is required to take a-\- b, m times. If a be taken m 
times, it is clear that for every time and part of a time which a has 
been taken, 6 or a part of 6 has been omitted. Consequently, ma h 
too little hy mh, OT ma •{- mb is the product required. 

Secondly, it is required to take a — b, m times. Here, if a be 
taken m times, 6 or a part of b too much has been taken for every 
time or part of a time which a has been taken. Consequently, m a 
is too much by wi 6, or ma — mb is the product required. 

Hence we may prove the following equation : 

as follows: let a + b be called />, and let c-\-d be called q; then 
a -\- b — c — d is p — q (page xix.), and 

m(a4-i— c— d) = m(p^q) = mp—mq 

But mp = ma + mb mq = mC'\-md 

mp—mq = ma-j-mb^{mc + md) 

s= ma + mb-^mc-^md 

The following are applications of the preceding rules : 

3(a+i) = 3/1+36 3^(0-6) = ^\a-^z\b 

ab^a—b) = aab-^abb 2a(a— aa)*= 2aa—2aaa 
3aic(a6— ac + 4) = SaaiJc— 3aaftcc + 12a6c 

2(1+1) =^ + T 6(1 + 1) =3^+2^ 

* The student will observe, that in these examples we do not inquire 
whether the expressions are possible or not, but bow to apply the rules 
when they are possible. For instance, a— a a is impossible when a is 
greater than 1, and possible only when a is less than 1. And let it be 
observed in proceeding, as a matter of great importance in the sub. 
sequent part of the work, that the rules investigated will not serve to 
distinguish between possible and impossible expressions. 


a ^ J 7 ac ad ae 

P9rs{^ "^ J + 7 ""^r) == qrs+prs+pqs^s 

In order to multiply a + 6 by c + rf, let us, for a moment, make 
p stand for a + ^* Then, p multiplied by c -|- </, is the same as c -|- '^ 
multiplied byp, which ispc -{-pd, or 

(a + b) (c + d) = (a + b)c + (a + b)d 

But (a + i)c = ac + Jc, (a'hb)d = ad + bd 

.\ ia + b) (c+d) = (ac + ic)+(a(i+i(i) 

= ac+bc'\-ad+bd 

To multiply o -}- 6 by c — rf, let a + 6 ^ jt?, and we have 
p(c — d)=/>c — prf, or 

(a + J) (c— d) = (a + i)c— (a + 6)d 

= (ac + ftc)— (ad + id) 
= ac + bc^ad^bd 

To multiply a — b by c — rf, let a — b ^ p, and we have 
p (c — d) ssipc — pd, or 

(a^b) {c^d) = (a— J)c— (a— i)d 

= {ac—bc)^{ad^bd) 

= ac—bc-^ad-jrbd 

There are two methods of conducting this process, the first of 
which is recommended to the learner for the present. 

1. To multiply a + 6 — 2cbyrf — a — c. 

Here a times and c times the multiplicand are to be successively 
taken from d times the multiplicand ; that is, a -\-c times is to be 
taken from d times. 


• • « 


Add I 

d times the multiplicand = ad+bd — 2cd 
a times ditto ^ aa-^ab — 2ac 

c times ditto s= ac-}-bc-^2cc 

a-f-ctimes ditto = flo + ac + a6 + ^c — 2ae — 2cc 

{Subtract from d times ad-\-bd+2ac-i-2c<:^2cd 
ditto which gives — aa— ac — ab^-bc 

2. From looking at the preceding instances, it appears that the 
rule of multiplication is as follows : Consider the first terms as having 
the sign + ; multiply every term of the multiplicand by every term of 
the multiplier y and put + before the products of terms which have the 
same sign, and ^- before the products of terms which have different 
signs. The preceding example is here written in the usual way, with 
the proper signs written to every term by tlie preceding rule. 

from d«»»« ad-^bd — 2cdl 

from a • • • • — aa — ai + 2ac ^Product required. 

from c •••• ^ac-^bc + 2ccj 

Where like terms are found in different lines, so that subsequent 
reductions may be made, it is convenient to put like terms under each 
other, as in the following example : 

Multiply xx-^2x-\-l 
By s — 4 

From j: xxx—ixx-^-x 
From 4 

m^m2XX'^X ^ 

. f Product required. 
— 4xx + 8j?— 4 } ^ 

XXX — 6xj:-|-9x — 4 ditto in simplest form. 

But the manner of doing this, which (as yet) we recommend, is 
the following : 

Multiply a?a?— 2a?H-l 

By ar— 4 

„ , . ,. ^ Z '. f Subtract second 

From a? timesmultiphc* a?jra: — Zxx + x ,. . 

'^ <{ Ime from first, 

Take 4 times ditto 4j;j-8j; + 4 | ^ ;„ ^^^^^^ y^^^ 

arxa;— 6a;a;+9a;— 4 Product re<iuired. 




The three instances which follow are of particular importance, and 
the learner should be able to write them (and similar ones) at sisht. 
Mult, a+b a-b a + r> 

By a+b a~h a~b 

To aa+ab From aa—ab From aa + nb 

Add ab + bb Take ab—bb Take ab + bb 


aa+2ab + b 

-2ab + bb 

[The following definitions may be appropriately introduced here, 
A square is a four-sided figure, with sides of equal length, and 
with contiguous aides perpendicular to each other. 

A cube is a solid figure enclosed by six equal squares ; or, a boK 
of the same length, breadth, and thickness. 

A square 4 inches long contains 4x4 squares of one inch long; 
a square of x inches long contains xx squares of one inch long. 
There are x rows of square inches, and x squares in each raw. 

A cube 4 inches long contains 4x4x4 cubes of one inch long ; 
a cnbe x inches long contains xxx cubes of one inch long. There 
are x layers of cubic inches, and xi cubic inches in each layer. 

Owing to this connexion of xx with the square on a line of x 
units, and of xxx with the cube on a line of r units, it has always 
been customary to call xx the tqtiare of x, aod xxx the cube o! x. 
But xxxx is called ihe fourth pmuer of x, xxxxx the fifth power ; 
and so on. So that x itself should be called tlie^tt power ofx, 
IX the tecond power of x, and xxx the third power ol x. Bui llie 
words square and cube are so conveniently short that they have 
never been abandoned.] 

llie last mentioned products may be thus slated : 

1. ia + b){a + b) = aa + 2ab+bb 



or, The square of the sum of two qtiantities is the sura of their sqtiares, 
augmeDted by twice their product. 

or. The square of the difference of two quantities is the sum of their 
squares^ diminished hy twice their product, 

3. (a + J)(a — 6) =z aa — bb 

or. The sum of two quantities, multiplied by their difference^ is the 
difference of their squares. 

Thus let the two quantities be a 6 and 2 a. 

ab X ab = aabb 2a x 2a = 4aa 

2(ab X 2a) = Aaab 
{ab+2a) {ab+2a) = aabb-j-4aab + 4aa 
(fli— 2a) (a6— 2a) = aa66— 4aa6+4aa 
(a6+2a) (a6— 2a) = aabb^-^aa 
The following are examples for the student : 

Square* ofladl-) = aadi2 H 

\ a/ aa 

(a -\ — ) [a 1 =: flfl 
a/ \ a/ aa 

Square of (2ax ^h 6) = 4aaxx it 4abx + bb 
(2ax + b) {2ax — b) ^ 4aaxx ^- bb 

Square of (a + b -\- c) = (a + b) {a -\- b) + 2{a + b)c -^ c c 

=s aa + bb ■}- cc + 2ab -f 2bc + 2ca 

(a + 6 + c) (a + 6 — c) = (a + b) {a -h b) -- cc 

= aa + 66 — cc '\- 2ab 

(c + a — 6) (6 + c — a) = (c + a — 6) (c — a — 6) 

* By + occurring several times in an equation, two equations are 
combined in one. The upper or under sign is to be taken throughout. 

aih6=<^HF<J is either 
a4-6:=c — d or a — 6s=sc4-rf 

The student U recommended never to use this double sign himself ; 
but as it frequendj occurs in books it is here shewn. 



= cc^^(a-^b) (a— 6) 
== 2ab'{'CC — aa-^bb 
^ 2ab — (aa-^-bb — cc) 

From the last two examples, shew that the product of the four 

a+A + c, a + 6— c, h-^-c^Oy c+a^b^ is 
2aabb + 2bbcc-}-2ccaa^aaaa^bhbb^cccc 

Shew the preceding by help of the following (which shew also), 
Squareof (p+y— r) := pp'\-qq-jrrr + 2pq--2qr'^2rp 

The following are miscellaneous examples in multiplication, to be 
done without paper : 

(a + bx) {a'\-cx) = aa + abx + acx-^-bcxx 
(or+a) (a? + 6) = xx-^ax-k-bx + ab 
(x— a) (a:— 6) = xx^ax-^bx + ab 
(x + l) (a?— 3) = a;a?-2a?-3 
(a:— 1) (a;— 3) = ara?— 4a: + 3 
(2a:+l)(a'— 1) = 2a;a;— a:— 1 

The following theorems are given for exercise : 

1 . If a and 6 be two quantities, of which a is the greater, and if 
S be the square of their sum, D the square of their difference, and P 
the product of their sum and difference, then 

8 + = 2(aa + bb) S - D = 4ab 

S + P = 2a{a-{-b) S-P= 2b{a + b) 

D + P = 2a{a-b) P-Ii= 2b{a-b) 

2. If two numbers differ by a unit, the difference of their squares 
is their sum; and if two fractions are together equal to a unit 

(such as -- and - j their difference is also that of their squares. 

3. The sum of the squares of xx — yy and 2xy is the square of 

[According to the rule, the square of a — b is the same as the 
square of 6 — a; for that of the first is ma — 2ab-Ybby and that of 
the second bb-^^ab-^-aa, which two are the same (see page xi). 




But one of the two, a — b or 6— a, must be impossible^ except only 
when 6 = a, in which case, both are nothing ; for in every other case 
either a — b or b — a is an attempt to subtract the greater from the 
less. But for the same reason, one of the two, b — a or a — 6, must 
be possible; therefore aa-\-bb — 2a 6, which is the square of that 
possible one, must also be possible. That is, if either of the two, 
a or bf exceed the other, aa-\-bb must be greater than 2ab. And 
we also see that t^ does not follow that an algebraical process is intel- 
ligible because it gives an intell gible result ; for it appears that the 
algebraical rule of multiplication would apply to (3 — 7) (3 — 7), 
which is absurd, and give the same result as (7 — 3) (7 — 3), or 16. 
This is a defect, the remedy for which we shall afterwards have to 
find ; and we see, that so far as we have yet gone, we can never 
know that any process is correct which is to lead to the value of an 
unknown quantity, until we re-examine the process after the unknown 
quantity has been found by it.] 

Divisions in algebra we shall for the present divide into two 
classes ; those which it is obvious how to do, and those which it is 
not obvious how to do. For instance, to divide ab by a, that is, to 
tell how many times a b contains a, the answer evidently is, that since 
ab means the same as 6a, or a taken b times, therefore ab must 
contain a, b times; or a 6 divided by a is b. In this case the simplest 
rule is as follows : To divide, where there has already been a multi' 
plication by the quantity which is made a divisor, suppress all the 
symbols of the multiplication. 

This will be seen in the following examples : 







































[One of the most common errors of a beginner is a mistake 
between and 1, arising from a confusion of subtraction and division. 
This is partly a result of the idiom of our language, as follows. If a 
begiaaer be asked how many times does 7 contain 7, the answer is 



sure to be, no timet at alii and in one sense this is oorrecty for 7 does 
not contain 7 a number of timet, but one time. But it roust always 
be understood in algebra that tiroes means time, or tiroes, or parts of 
a time, or time and parts of a time, or times and parts of a time.* 
Therefore, though 

X diminithed by x leaves nothings 
X divided by x givet one,} 

Closely connected with the preceding is the theorem in fractions 



that T and — - are the same. For -y multiplied by m gives -7-, and 
o mb " ^ 

this divided by m gives — r. But multiplication, followed by division 

(multiplier and divisor being the same), leaves any quantity the same 
as at first 

From the rule of multiplication it follows, that if any quantity 
contain the same letter or letters in every term, it is the obvious result 
of multiplying another expression by that letter or the product of 
those letters. Thus, ab'\-ae is 6 + c multiplied by a, aab — abc 
is a — c multiplied by ab. Hence, to divide an expression by a 
letter or a product of letters, strike out those letters from every term 
of the dividend. But remember to write 1 where all the letters of 
any term are thus struck out. For instance, a-^-ab divided by a 
gives 1+6, ac-^aac ^ ace divided by ac gives 1— a + c. 

The following are instances, arranged as before: 






a — c + 2ac 



aa^a + 1 

6a6 — 3a + 36 


2a6 — a + 6 



a —6 




* An act of parliament, or any- other legal instrument, always speaks 
of men under the title **man or men,*' &c. If this should happen to be 
neglected, and a single offender should plead that "men" only were 
prohibited from doing as he (one man) had done, it would be called a 
** quibble." If the student, after this warning, should ever say that * 
contains x no timet, it would not only be a quibble, but a very useless 
quibble, because nothing is to be got by it. 


In dividing a & by a we might proceed as follows. The result 
of the process is the fraction — , which is not changed in value if 

both numerator and denominator be divided by a. But this gives -, 

which is h. 

Such a process is of no use in the preceding case ; but suppose 
that a 6 is to be divided by ac. The complete division is here 
impossible until we know what numbers a, h, and c, stand for. But 

— , the symbol of the result, may be reduced to - by the preceding 

theorem. The division here is not completed, but reduced to a more 
simple division. 

26 ^_ 6 aah ^_ ab Zammn mn 

2c c ax X 6aam 2a 

p 1 Sab 3 21vww 3vw 

pq q aab a 2^xw Ax 

The division of an expression of several terms by another of one 
term may also admit of reductions. For example, xy-^yz — zx 
divided by xyy is 

ly^ ^ UL ^ 1!L or 1 + — _± 
xyy xyy xyy y xy yy 

2v — xx-\'Vx 2 * I 1 

vx X V 

a + b-^c _. i. . J_ . JL 
abc be ac ab 

= - H = a -\ — 

aa a aa a a 

x+45/ — 3g + 2 ^ I Hy £. J- i 

All that precedes contains the obvious cases of division ; of those, 
the answer to which requires further process, the following is an 
instance: "How often is x +y contained in xxx ■\-yyyV* As we 
shall not need such a process for some time, we defer it to its proper 
place. In some cases, however, the preceding theorems of multi- 
plication (page xxix) furnish an answer at once. For instance, we 
know that xx'^9 must contain J^ + 3, x -^ 3 times. 


Fractions may frequently be reduced to simpler terms by in- 
spiection, of which the following are instances : 

a -{- ab 1+6 Sx-f 6 jj __ 1 -f 2 j: 

a — ab 1 — b 9xy — 3x 3y — 1 

a — aa 1 — a flfl-+-3fl6 + 36 

2a+fli 2 + ar fl6 + 12a "" 6 + 12 

It is frequently necessary to arrange expressions in a different 
form, without altering their value, by performing inverse operations 
upon them with the same data^ such as addition followed by sub- 
traction, or multiplication followed by division. The four following 
methods of writing x exhibit this process. 

or+a— a ar— a + a -- -xa 

a } a 

Thus a+x = 2a+a;— a = (l + -ja 

aa+2ab—c = aa + 2a6 + 66— (c + 66) 

= (a + 6) (a + 6)— (c + ft6) 

m+w = mn(- -\ ) = w(~ + 1) = ?w(l + — ) 

\n m/ \n / \ m/ 

1 1+ar 1+jr ,_ 

^ = -T = 1 = 1 (See page v.) 

-,0+^) 7 + 1 


The following are instances of those reductions of fractions which 
will occur hereafter. The rules with regard to fractions which are 
proved in arithmetic are here applied in conjunction with the 
algebraical methods of addition, &c. At the head of each section 
stands an example without any complicated expressions, containing 
the arithmetical process used in those which succeed. 


a ma 
b mb 

1+j (l + \)^^ xx-\-x 

r+J "" 7r+J-)^a. ~ ^^ + 1 

* XX \ ' xx/ 



_ y(3^— 1) _ yy—y £ _. j^j'+ax 

" y — 1 y — 1 a ax-\-aa 

\ a—b a+& 4 

a-^b aa — bb fla + 2a6-f-66 4a+46 

X — 4 2j' — 8 4j:— 16 2ax — 8a 

2i 5 "" 10 "~ 5o 

7j— 4 _ ^(7 J— 4) _ 3^J — 2 
10 "" Kio) ~ 5 

5 + -^ ^Ka+Ift) 

_ 6 + 1 

h^a-\-\ a6(6— fl+l) a^fc— aaft + a 
_, . X ay-^x X ay — x x jr — av 

y y y y y y 

. 1 X — 1 1 XX — 1 


_y^^ ^ 2j^ + 2-(3/-l) ^ 5^ + 3 

j^'+i 5^ + 1 y + 1 

aa a6 ab __ aa 

^ a4-6 a + 6 a + 6 a + 6 

aa — 2a6 4a6 + bb 

a + b 

a + ft ■ aH-6 

aa — 2a6 2aa-+-fc6 

a-{-b a-j-^ 

•5lzi± -4- 7/ -4-1 -= • y^-y-y = 7/-2lZ- 

5/— 1 -^ y— 1 ^5^— 


a + 6 — c Q 6 — a-|-c x x — xy 

a — c a — c y y 

ab-^bc-^-ca ^^ ab — cc 

a + 6 +c a + 6 + c 

a J- ^^ ay + 6jr a x ay — b x 

by '~' by by by 

a^h a—b _ (a~\-b){a+b) — (a—b)(a — b) _ 4a6 
a — b a-{-b (a — 6)(o + 6) aa — bb 

1 J- 1 — : ^'^^ - -4- ^ = ''JL^yy. 

X y xy y X xy 

a-^-b a cb — ad a — b a ad — be 

c-{-d c cc-^ cd c — d c cc — c d 



^x y _ xx^2xy-~yy 


9 PP 

?p + 77 

x+y ^—y XX ^yy 


IV. ^ X i^ = ^ 
b y by 

a , X ay 

b'^ y^ bx 

j:— 1 jrH-2 
X — ■ — = 


-l)(-r + 2) _ X4-2 
+ l)(jr-l) j + 1 

2ab a — b 6aab 
a -\-b ' 3o ^ aa'-'bb 

3ax yv 

X = 
yy 2jr 


m 2m 3b 
an ' Zbn 2a 

P^ ^ 3^9 — ^P9 
q cce ce 

The student is recommended to make himself well acquainted 
with every example given in the preceding list, but no more (see the 
Preface); as a better method of obtaining examples will be given. 

All that has preceded is purely arithmetical, and the letters may 
be considered as mere abbreviations of numbers, and all identical 
equations as abbreviations of arithmetical propositions. Thus, 

{a + b)(a + b) = aa+2ab'\'bb 

represents the following sentence : — If two numbers be added 
together, and if the sum be multiplied by itself, the result is the 
same as would arise from multiplying each number by itself, and 
adding to the sum of these products twice the product of the 

An arithmetical problem is one in which numbers are given, and 
certain operations; and the question asked is, what number will 
result from performing the given operations upon the given numbers. 
For instance, what is the fiftieth part of the product of 25 and 300. 

An algebraical problem is one in which numbers are either given 
or supposed to be given (as will presently be further explained), 
and a question is asked of which it is not at once perceptible what 
operations will ^furnish the answer. Such is the following: — The 
numbers 3 and 17 are given; what number is that, the double of 
which will fall short of 17 by as much as its half exceeds 3? And 
the questions asked are the following. 1. Is there any such number? 

2. If there be, by what operations on 3 and 17 may it be found ? 

3. What is the result of these operations, or the number required. 
The answers to which (as the student may afterwards find) will be 



that there is such a number, that it is found by taking two-fifths of 
the sum of 3 and 17, and that in consequence the number is 8. 

If we had contented ourselves with the first two questions, it 
would have been unnecessary to have specified that the numbers in 
question were 3 and 17, for the same problem might have been 
proposed about any other numbers, and the process of solution would 
(as may afterwards be shewn) have been the same whatever the 
numbers might have been. That is, if the following question had 
been asked ; — The numbers a and b are given ; what number is that, 

the double of which will fall short of b (the greater), as much as its 

half exceeds a (the less) ? The answer is - (a -j- f^) and the veri- 


fication is as follows : 

2.4 44 

The double of- (a + o) is - (a + 6), or - a + - b. 
o o o o 

Tin's falls short of & by 6 — (-a + -b] or b a b 

\o 5 / 5 5 

14 2 1 

or T&— -zu; but the half of -(a + &) is-(a + b) 

which exceeds a by - (« + fc) — fl> or r o + - b — a, or ~b a, 

5 5 5 5 5 

the same as that by which the double of - (a -f- b) falls short of 6. 


Which was to be done. Now, observe that the preceding not only 
informs us of the general process by which this problem may be 
solved, but it also shews in what cases the problem is impossible. 

1 4 

For the excess or defect above-mentioned turns out to be - 6 a, 

5 5 

which is an absurdity, unless -r ^ be greater than (or at least not less 


than) -a; that is, unless 6 be greater than 4a. This was the case 


in tlie first instance where b was 17 and a was 3. If it be not so, 
we may pronounce that the problem is impossible ; for instance, let 
the student try to find a number or fraction, the double of which 
shall fall short of 11 by as much as its half exceeds 3. In this 
problem there must be a contradiction ; and when we know there 
is one, and set ourselves to find out how it arises, we see it in the 
following: — The number sought is presumed to have a half which 
exceeds 3 (so that it must be more than 6) and a double which falls 


short of 11 (so that it must be less than 5^). But a number which 
exceeds 6 cannot be less than 5^ ; therefore the clauses of the pre- 
ceding problem contradict each other. 

We see, then, that we may propose a problem which is impossible 
or contradictory, or has no solution. But, on the other hand, we 
may propose a problem which admits of an innumerable number of 
answers. These we will call unlimited problems. And, as in the 
case of impossible problems, there are some of which the impossibility 
is evident, as ** To find a whole number which shall be the half of 
seven i" and others, in which it requires investigation to discover the 
impossibility, as ^* To divide 10 into two parts, whole or fractional, 
of which the product shall be 30 ;*' so in the case of unlimited 
problems, there are those in which the unlimited nature of the result 
shall be evident, and others in which it shall not be so. For instance, 
to the question, *' To find two odd numbers which added together 
shall make an even number?'' it is clear that the answer is, "Ant/ 
two odd numbers ;** and to the question, '< What two numbers are 
those of which half the sum added to half the difference shall give 
the greater number ? " the answer is (but not so evidently), " Any 
two numbers.'' Between these two extremes, we can conceive there 
may be problems which admit of 1000 answers, others of 999 
answers, &c. &c. down to problems which admit only of one answer. 
And even when we find tliat a problem is impossible, we may yet 
think proper to ask, whj/ is it impossible ? what are the two parts of 
the problem which contradict each other, and by how much ? that is 
to say, what sort of change, and quantity of change, in the conditions 
of the problem, will render it possible ? 

To all these questions, arithmetic gives no means of answering, 
and we have therefore to consider algebra as a distinct science, which 
proposes objects of which arithmetic knows nothing, and therefore 
as we may suppose, uses language, finds methods, and adopts inter- 
pretations, of which arithmetic furnishes no examples. 

If the student have read* a little of geometry (a science which he 

* In England, the geometry studied is that of Euclid, and I hope it 
never will be any ether ; were it only for this reason, that so much has 
been written on Euclid, and all the difficulties of geometry have so uni- 
formly been considered with reference to the form in which they appear 
in Euclid, that Euclid is a better key to a great quantity of useful 
reading than any other. 


should begin to study at the same time as algebra, if uot before), f 
knows that all the questions of geometry are made for him, that i^ 
the reasonings, &c. are put together before his eyes, and all he has t« 
do is to comprehend and agree to one step of the process after anotbear 
This is called synthesis {fvfit^ttt a putting together), or the synthetical 
method, in opposition to analysis (AvaXv^it, an unloosing, or bringing^ 
asunder), or the analytical method. The latter consists in taking the 
problem to pieces, if the phrase may be used, that is, reasoning upon 
the whole problem, reducing it to more and more simple terms, and 
so coming at last to those considerations which must be put together 
to make a solution and to verify it. 

We now proceed to establish the principles of algebra analytically; 
and instead of laying down new names or new principles, and putting 
the science together, we begin from arithmetic, such as we know it, 
and leave all additional considerations till the want of them is felt. 
We shall thus see one new result spring up after another, until we 
find the necessity of speaking a new language, and giving interpreta- 
tions to symbols which we did not at first contemplate. How this is 
done, and what it leads to, we cannot otherwise explain than by di- 
recting the student to proceed to the first chapter. 




We now proceed to the solution of equations of the fir$t degree,* 
This term must be explained. 

To find the degree of a term generally, count the letters in it. 
Thus, a 6c is of the third degree; aahc is of the fourth; for though 
there are only three letters, yet one of tbera occurs twice. The fol- 
lowing are examples: 

Of the first degree, a, 6, c, x, z, p, &c. 

Of the second degree, aa, bby ex, hcy pz, &c. 

Of the third degree, aaa, aah, abb, abc, pac, &c. 
and so on. 

To find the degree of a term with respect to any letters, count 
those letters only. Thus Saaxxy, of the fifth degree, is of the third 
degree with respect to x and ^, of the fourth degree with respect to 
a and x, of the second degree with respect to x only, of the first 
degree with respect to y only, and so on. A term which does not 
contain x at all, is of tio degree with respect to x, or is independent of j:« 

The degree of an equation with respect to any letter, is the degree 
of the highest term with respect to that letter. Thus the equation 

XX ~ zxxx 's=^ yz — yyx 

is of the third degree with respect to x, of the second with respect to 
y, and of the first with respect to z, 

* Commonly called tim'ple equations. 



The solution of an equation of condition is the following problem : 
— Given an equation of condition, containing a letter the value of 
which is unknown ; what is that number for which the unknown letter 
must stand, in order that the equation may be true ? Are there more 
such numbers than one ? if so, how many, and what are they ? Or is 
there no such number^ that is, is the equation impossible ? and if 
so, how is that to be ascertained? 

The scope of this will be better seen by some instances, which the 

pupil may verify. 

The equation 

2a:- 1 = 6a:- 19 

is true then, and then only, when x is 6. 
The equation 

2a:— 1 = 6a; + 12 

cannot be true, whatever x may stand for. 

The equation 

16a; s= 48 + ara: 

is true when x is 4, and is also true when x is 12 ; but never in any 

other case. 

The equation 

12a: = 48 + a:a: 

is never true for any value of x. 
The equation 

XXX + 11a; = 6a:a: -f- 6 

is true when a: is 1, when x is 2, and when x is 3 ; and in no other 


As an example of verification, let us try the latter equation when 

jf = 4. Then 

XXX = 64 6a;a; = 96 

ll:r = 44 Qxx + 6 = 102 

XXX + lia: = 108 

But 108 is not = 102; therefore xxx + llx is not = 6a:j7 -f 6, 
or j: = 4 does not satisfy the equation. 

We shall now use the following evident truths : 

1. If equal numbers be added to equal numbers, the sums are 
equal numbers. That is, if a = 6 and c = </, then a -{- c ^ h '\- d. 
If arszb — c and j: = jj — q, then a -\- x = (b — c) + (jd — y) = 
6 -fp — c — q. If a := cT — y, and 6 s^ x + y, then a -f ^ = (•*" — !/) 


+ (*+y) = *— y + '+y = 2j:. If a = 5 4- c, then a + v = 

2. If equal numbers be taken from equal numbers, the remainders 
are equal numbers. That is, if a = 5 and c = d, then a — c^b — d. 
If a ^p — g and b =ip — 2^, then a — b =s (;> — q) — (jf — 2^) =s 
p^q^p^2q = y. If a =s jar+y, tlien a— w = z-^y^m, 

3. If equal numbers be multiplied by equal numbers, the products 
are equal numbers. That is, if a = 6 and c =i dy ac ss bd. If a ss 
6 + c, and z ^=n, then ajet =s n{b-\'c) = nb-^nc. If J = / — v, 
2d= 2/— 2t;. 

2 3 
Then 2|+2|=: 2 or 0:+^ = 2 

3a? + 3y = 6 

or 3a;H-2a: = 6 

4. If equal numbers be divided by equal numbers, the quotients 

are equal numbers. That is, if a ^ 6 and c = (/, then - = ->• If 

c ct 

iR = n, then — =-. If a = b — c and p4-g = x, then — -— ^ 

. If 7x = 14, then -—=---, or x = 2. 

z T T 

The following abbreviations will be used, on account of the con- 
tinual occurrence of the phrases : 

(+)a means, add a to both of the last-mentioned equal quantities, 

which gives .... 
( — ^a .... subtract a from both of the last-mentioned equal 

quantities, which gives .... 
(x)a .... multiply both the last -mentioned equal quantities 

by a, which gives .... 
(-i-)a .... divide both the last-mentioned equal quantities by 

a, which gives .... 
(-f-) ( — ) (x) and (-2-) by themselves I use to denote that the 
itoo last-mentioned sets of equal quantities are to be added. The 
following will explain the use of these abbreviations : 
1. a = 6~c 6— 6 = j + a: 

(+)c a + c = b (+)6 a = g+^ + 6 



2. c— rf = /— m 2jr^3 =: 9 
{+)dT^ c + m=zl+d ( + )3 2a; a: 12 

3. p + q = a— 6 11j:+18 = 100 
(«)y ^ = a-6-y (-)18 lla; = 82 

4. p + y— ^ = 3a+4 

(— )y— * P = 3a+4— gr-H^r 

X X 27 7x 5jr 3 

2 3"*'T "■ "6'"T2"*'4 
, v,rt 12j 12j . 324 84x 60x , 36 

(^)i2 -2 — r+— = -r--TT+T 

or 6a?— 4a: + 81 = 14a:— 6a: + 9 

6. aa? = 6 (aH-6)a: = c 

7. a— fc+2c— 3c? = x^a + b 

64-3c— 2rf = 4a?-a— 26 
( + ) a-}-6c— 6c? = 6x— 2a— 6 

8. 2ax = b^z 

a = 64-2r 

We shall now proceed to the solution of equations of the first 
degree, containing one unknown quantity, by means of the principles 
in pages 2 and 3, and the preceding operations. 

1. What value of x will satisfy the equation 

3a;— 7 = a; + 19 
( + )7 3a; = a; + 26 

(-)ar 2a; = 26 

(-f-)2 a: = 13 

Verification. If a: = 13, 3a:— 7 = 32 

a: + 19 = 32 

2. 3a:+16 = 10a:+9 
(-)3a: 16 = 7a:+9 
(-)9 7 = 7a; 
(-r)7 1 = X 


Verification. If X = 1, 3x+l6 = 19 

10a:+ 9 = 19 
3. 20x— 13 = 102j-ar 

(+)13 20a; = 116J-a; 

i+)x 2U = 116J = HI V 

(-r)21 ^ = 23Sr ^^-"'-^ 

- ?ii-S* 

- w-°^ 

Verification. It x = 6J, 20ar— 13 = 97 

102j-a; = 97 

a: + ^ = 2-^ 

3 2 

2x+y = 4-x 

6x+4x = 12-3jr 

6a;+4a:+3j? = 12 

13j; = 12 

(^)13 x = l| 

Venjicatwn. If ^ = -,-+- or gO; is - of ~, or -. 

And 1 — T is 1 — (-: o^ TTih or 1 — r;:> which is also --. 
4 \4 13/ 13 13 

The same equation might be more easily solved by multiplying 
both sides by any common multiple of 2, 3, and 4. The least com- 
mon multiple is the most advantageous ; why, will appear on trying 
a higher one, as follows : 

2 + 3= ^~4 
36 is a common multiple of 2, 3, and 4. 

(x)36 -2- + — = 36-— 

or 18a; + 12a: = 36-9j; 

(+)9a; 18a; + 12ar+9a; = 36 






that is 


that is 39 X = 36 

( -f" ) 39 X = — , which, reduced to its lowest 



terms, is s 

Now try 12, the least common multiple of 2, 3, and 4. 

JF * t St 


(x)12 lli+!|f = 12-lii 

or 6x + 4x = 12— 3x 

Proceed as in the last case but one ; and no reduction of the result to 
lower terms is necessary. 

5. ab-^-a-^b = 1 

This equation differs from the preceding in' having two unknown 
quantities. The real answer is, that there is an infinite number of 
values of a and 6, which will satisfy this equation. If we choose a 
value of bf we can find the value of a, which, with the chosen value 
of bf will satisfy the equation. For instance, I ask, can 6 be ^ 12? 
Substitute 12 for 6 in the above equation, which then becomes 

12a+a-12= 1 

or 13a- 12= 1 

( + )12 13a =13 

(-T-)13 a = 1 

The answer is, b may be 12 provided a be 1. In this case^ 
ab = 12, and 12 + 1 — 12 = 1. 

Without making 3ny particular assumption about the value of b, 
let us suppose it given ; in what way must we combine this known b 
on the first side with the known unit on the second side, so as to 
point out the manner of finding a so soon as a particular value shall 
have been assigned to 6 ? 

Resume the equation : 

a6-i-a— 6 = 1 
( + )6 ab + a = 1+6 

But ab-\-a is a taken one more than 6 times; that is^ a& + a = 
(1+6) a. 


Therefore (iH-fc)a = 1+6 

(H-)i+6 a = :f^^=l 

The answer, then, is the following : 6 may be what we please, 
provided a be 1. 

Verification. If a = 1, 

ab-i-a—b = 6 + 1 — 6 =s 1. 

We have given this instance to shew how soon the operations of 
algebra lead to unexpected results. We will now take another 

6. xy = ar+y + l 
Knowing the value of y, to find that of x, 
( — )a: xy—x^y + l 

but xy — j: is j: taken once less than y times, or (y — 1) x. Therefore 

(y-l)a:=y + l 

Particular case* Let y ^ 5, then 

^ "■ 5 — 1 "■ 4 ■" 2 

3 15 

Verification, xy =s - x 6 = — 

General Verification, x := 4lII1_ xv =• ^^ ^ / 

__ 5/ + i+3/i^— y+y— 1 


_ yy+3^ _ y(y+i) 
>— 1 y— 1 

7. Two labourers can separately mow a field in 4 days and 7 
days. They begin to work, and on the second day are joined by a 
third^ who alone could mow the field in 10 days. The third remains 


with the former two for a certain time, aAer which he leaves them; 
and it is then found that exactly four-fifths of the field have been 
mowed. IIow many days is this altogether? 

[This question is introduced to shew how very soon algebraical 
symbols may be made to simplify complicated arithmetical reasoning.] 

The fractions of the field which the first and second could mow in 

a day are - and -. Let x be the whole number of days ; or, that 

number being unknown, let x staAd for it until it is known. Then 
the first man, who does one-fourth in one day, two-fourths in two 

days, &c., will in x days do j:-fourths, or the fraction - of the field. 


In the same time the second does - ; but the third, who works one 


day less, at the rate of one-tenth a day, does — — . Therefore, all that 

is done of the field is 

Jf , X , X — 1 
4"^7'* To" 


Biit by the question this is -, which gives 

X , X , X — t 4 

4 "^7"^ lo" "" 5 
The least common multiple of 4, 7, 10, and 5, is 140 (Ar. 103). 

(x)140 !12a:+li2,+^(,_l) = 12O10 

or 36a; + 20a; + 14(a:-l) = 112 

or 36a;+20a: + 14a;-14 = 112 

[because 14(x— 1) = 14x— 14] 

therefore 69j:— 14 =112 

(+)14 69 a: =126 

126 42 ,19 

(^)69 a;=lH£ = lH=ii2 

^ ^ C9 23 23 

In 1 day and -- of a day, there is mowed by the first 1 and -- 
Z6 4 23 


ot a day, there is mowed bv the first ~ and 


^^ 4' ^' 23 ^^ 4' ®' 46 ^^ ^^® ^^^^ ' ^^® second mows || of - or 

6 12 

— , that is -t; and the third, who works one day less than the others, 


19 19 1 19 

or only — of a day, does in that time — r of -— , or --- , that is 
^ 23 ^ 23 10' 230 


—- of the field. But 


21 . 12 . 38 368 4 

46+46 + 460 = 460 =5'^ '^^"*'"^- 

x~r^ ^ — 4 Q * + l 
8. — — — — ■ -, ^ O— ■ ^ — 

2\ 6\ 5 

A common multiple of 2^ and 6} (not the least, which is 95, 
and with which the student should also solve the equation) is 570, 
which contains the first 228, and the second 90 times. It is also a 
multiple of 5. 

( X )570 ^(a:-3) - '^(x-4) = 1710-^°(ar+ 1) 
or 228(a;-3)-90(a:-4) = 1710-114(a;+l) 

But 228(x— 3) = 228x— 684, &c. 


(228a;-684)-(90ar-360) = 1710-(114a;+114) 

or 228a;-684-90x+360 = 1710-114x-n4 

( + ) 684 + 1 14X-360 228a;-90ar+114a: 

= 1710-114+684-360 or 262* =. 1920 
(-r)12 21* = 160 


^=^ 21 = 





X = • 



— 3 



194 194 
105 5X21 

X — 

4 _ 21 
"■ 19 



X — 3 X — - 







2i 6J 





a:+l __ 

- 21 __ 





o— — ; — = - — -Tt the same as before. 
5 5X21' 

From these cases we may lay down the following rules for the 

solution of equations of the first degree. 

1 . To clear an equation of fractions, mtdtiply both iides by a$^ 
common multiple of all the denominaton : generally ^ the least common 
multiple is the most convenient. 

The following are some useful applications of this principle : 

fl __ C , \ 1. _J abd Cbd m m 

J- 5 (x)Jrf -J- = -5-, or arf = Jc 

From the preceding equation the student is left to deduce the 

following : 

cb i ad ^ ad J be 

d c b a 

^— A ^— _1 1— A ^— A 
a cb b ad c ad d'^ be 

From the first of the following equations let the student deduce all 
the rest. 

ah ^^ cd cdxy ahpq A ^^ cdx 

*y ^ pq ~~ pqb " cdy y "~ pqb 

ab dxy « , b ex 

— = — i abpq = xycd -j- = 

c pq ^^ ^ dy apq 

2. Any term of an equation may be removed from one side to the 

other if its sign be changed. If this have not already occurred to the 

student from the preceding examples, it may be established by the 

following : 

Let a + J = c + rf— e 

(— )i a = c+rf— e— i 

( + )e a + e = c + d— ft 

In applying the rule for clearing an equation of fractions, care 
must be taken, when the denominator is removed, to remember that 
the sign which was placed before the complete fraction now belongs 
to the complete numerator, which should, therefore, be placed in 
brackets, or the proper rule for addition or subtraction applied at 
once. The following example will shew what is meant. 

X — a c-\-x J J?— c 
b ab a 

(x)ab aia;-f-a(a;— a)— (c + x) = abd—b{X'^e) 
or aiar+(«x— aa)— (c+a;) = abd^ibx-^be) 

or aiar + aar— aa— c— a: = aJrf— iar + Jc 


The mistake to which the beginner is liable is, to write — c -j- x 
and — bx — fee, instead of — c — x and — 6 jr + 6 e. 
By the second of the preceding rules, 

ahx + aa; -f- ia; — ar = ahd + aa -{• c -{-be 

or {ab-^-a -f-i — l)a; = abd-^aa + c + be 

/ V — T -. abd-^aa-j-c-^-be ,iv 

i-h) ab + a + b-l X= .t^^Jtll (1) 

ao+a +0 — 1 

abd'\-aa-\-c -^-be — a(ab •\'a-\'b — 1) 

"" ab+a + b — l 

abd-^ aa •}- c + be — aab — gg — afr-j-o 

"" ab + a + b-^1 

__ abd + a — ab — gg^-f-c + ^c 
ab -{• a + b — 1 

JT— g abd-^a — ab — aab -{- c •{• be .qv 

"T" 6(g6 + g + 6 — 1) ^^ 

And by similar processes, 

c+jr abc-^ac + bc + abd -\- aa'\- be ,n\ 

"ab ab{ab + a'\'b'^l) ^^ 

X — e abd -\- aa -{-c — abe — ae •}- e .j^. 

~ g(g6H-g4-fc — 1) ^^ 

Reduce (1) (2) (3) and (4) to a common denominator, which can 
be done by multiplying the numerator and denominator of (1) byg6, 
of (2) by g, and of (4) by 6. Then form 

^ + '-T^-Tr' "' (l)+(2)-(3) 

which will be found to be 

gg66<^ + gg6rf-i- abbe -{- abe — ggfe — abd — be — be 

ab{ab + a-\-b—l) 

aabd -\- aad-\-abe + ge — gg — ad — c^^e 

a(flfc+a + 6 — 1) 

the latter of which arises from dividing the numerator and denomi- 
nator of the former by 6. 
By similar processes, 

e?-^ or e?-(4) 

will be found to have the same value as (1) + (2) — (3). 

The student should not pass the preceding solution until he is 
able to repeat the whole on paper without the assistance of the book. 




In the preceding equation, let c and e be each equal to nothing, 
which reduces the equation to 

ub a 

the value of x to 

. ahd-^aa 

/and the value of each side of the equation, as just found, to 

aahd-^- aad-^aa-^ad 
fl(a6 + fl + 6 — 1) 

or — j-^^ — — r — 7— (dividing numerator and denominator by a.) 
afr + a + o— 1 ^ o -f ' 

These the student should find for himself from the equation. 

We now look into particular problems to see what explanations 
may be necessary. Various unforeseen cases will present themselves ; 
and each case will he explained, and a problem will be given for 
each, to shew how it may arise. 

Anomaly 1. Let a = 2 i^3 C?^^ 


The equation then becomes 

jr— 2 X 1 X 

^■^"1 6 "" 6 "" 2 

Then X = ?2L3_2^J±2.>12 = ± = 1 

2x3-1-2-1-3—1 10 2 

On attempting to verify the equation, we see that a contradiction 

appears ; for x is -, and the operation x — 2, which is therefore im- 

possible, appears in the second term. It seems, then, that an equa- 
tion may give a rational tolution ; and on attempting to verify the 
equation by this solutionf the latter may be found to be impossible. The 
question now is, can such an equation arise from a problem ? if so, 
is it the problem itself which is absurd, or the way of treating it ? 
If the latter, how is the method of solution to be set right ? 

Problem in illustration. A enters into this bargain with B, 
that he is to take B's property and pay his debts, taking his chance 
of gain or loss. On examination, it is found that B's property is 
(debts allowed for) exactly the same as that of A, with this excep- 
tion only, that B is in partnership with another, and he and his 
partner have made a similar bargain with C. On examining C's 
afiairs, he is found to be insolvent by £100. The result of the whole 


is, that this transaction falb short by £75 of making A's property 
twice as great as it was. What was A's property? 

Let X stand for A's original property, whicli is, therefore, that of 
B and Co., independent of their share in the engagement with C. 
From the concluding paragraph we might presume that A is benefited 
by the transaction, namely, that the £j: of B and his partner is more 
than sufficient to cover the loss arising out of their engagement with 
C. Let this be so; then x — 100 pounds remains for B and his 
partner, of which B's share, namely, i (jt— 100) is by the bargain 
transferred to A, who has, therefore, 

. JT— 100 
X 4-' 

This doubles his property all but £75^ and is, therefore, the same 
thing as 2x — 75. Therefore, 

x + '^^ix-lb 

(x)2 2a; + a;- 100 = 4a; --150 
4a; — 2a; — a; = 150-100 
a; = 50 

This introduces into the equation the anomaly we are now con* 

sidering, for is impossible. It also contradicts the suppo- 
sition on which it was obtained, namely, that x is greater than 100. 
We cannot, therefore, depend upon this solution. Suppose we try 
the other supposition, namely, that x is not greater than 100. In that 
case, B and. his partner have to pay £lOO, of which they can only 
make good £x\ of the remainder, or 100 — j*, A must, by the 
bargain, make good B's part, or \ (100 -~ j:). Thi» he loses by the 
transaction ; and having x at first, he has now only ^ 

100— jr 

This doubles his property all but £75, or rather we must now change 
this mode of speaking, which may seem to make one part of the 
problem disagree with another, and say simply that A's propeity is 
£75 less than twice what he had before. Hence, 

(x)2 2a; -(100 -a;) = 4a; - 150 


or 2j:-100 + x = 4x-160 

or 2x+ar-100 = 4x-160 

wliicli is now the same equation as before, and not absurd in its 
present form ; for tliough (since it yields x =z 50) s — 100 is absurd, 
yet ftjT -I- JT — 100 is not so. Hence we see that the effects of the 
wrong supposition, which made us write or -f | (x — 100), where 
we should have written x — | (100 — x), disappear in resolving the 
equation, and leave the same result as we should have obtained by 
proceeding correctly. 

The anomaly arises from an error of this sort. If 6 be greater 
than c, we know that 

a— (J— c) = a— i-f-c 

but if we have a — b -{-c, and wish to bracket b and c together, we 
cannot do this correctly until we know which is the greater. If it be 

b, the preceding is d — (6 — c) ; 

but if it be c, this should be 

fl + (c— J). 

One or other of the preceding two is absurd, except only when ft ^ c, 
which makes them a — and a + 0, or a for both. 

From hence we may see, so far as one instance can shew it, that 
any mistake which amounts to no more than writing a — (6 — c) 
instead of a + (c — 6), as the representative of a — 6 -f" c, makes no 
difference in the final result. We here write, side by side, the solu- 
tion of two equations, which only differ as above. 

a— x , X — b , X — a b — x 

Ja:— (a— a:) = JcH-(a:— 6) Ja; + (a;— a) = 6c— (6— x) 
bx^a-hx = bc+x—b bx+x—a = 6c— 6-f-a; 

The remaining part is common to both. 

bx = bc + a — b 
be -|-fl — 6 

^ = — — 

Anomaly (2). Let 

ax + b = cx + d 

ax^cx = d—b 

{a^c)x = d—b 

X = 

a — c 



Let it happen that d is less than hy but a greater than c, as in the case 

There is then an impossible subtraction in the numerator of the 
result ; and it is sufficiently evident, on other grounds, that the equa- 
tion is impossible ; for if a be greater than c^ax must be greater than 
ex ; from which, if b be greater than d^ax '\-b must be greater than 
ex + df and cannot be equal to it. A similar question may now be 
proposed to that which arose upon the last anomaly. (See p. 12.) 

Problem I. in illustra.tion. In the year 1830, A's age was 
50 and B's 35. Give the date at vdiich A is twice as old as B. 

This must either be before or after 1830. Try the second case, 
and let the required date be 1830 + x. 

Then A's age will be 60 + a; 

. . B's 35 + a? 

and 50 + a; = 2(35+ a:) 

or 50 + a; = 70 + 2x 

2x-'X = 50—70 

Here we see an impossible subtraction, and it is also evident that 
2 X + 70 must be greater than x + 50. Now, try a date before 1830 ; 
say 1830 — x. 

Then A's age was 50 — X 

. • B's .... oo — X 

and 50 — 0: = 2(35 — a;) 

or 50 — a; = 70 — 2a; 

2a;-a; = 70-50 
or . a: = 20 

An evidently true answer; for in the year 1830 — 20, or 1810, A's 
age was 30 and B*s was 15. Here then we see that an impossible 
subtraction may arise from assuming a date to be after a certain epoch 
which is in fact before it, or vice versd. 

Problem II. A and B have accounts together. The state of 
their affairs is this : Give A half as much as will make their dealings 
worth £500 to him, and give B £lOO, and they will then, after settling 
their account, have equal sums. How does their account stand ? 

The balance is either in B's favour or in A's. Take the latter, 
fuid suppose A ought to receive £x. Then 500 -» x will make this 
tmosaction worth £500 to him, because 


x + (600 — ar) =b 600 
(Sive liim | (500 — jr) and he m\\ hate (when B pajs him) 

, 500— jr 

Now R, when he gets the £100, mutt pay £r to A, and will 
therefore hafe £(100— x). But they haTe then equal siuns; tbereloac^ 

:c + *2^' = 100-x 

(x)2 2ar + (500-ar) = 200-2x 
or 2x4-500— « = 200 — 2 J? 

2x-\'2x^x = 200 — 500 
or 3x =200 — 500 

which is impossible. Try the other case, and suppose the balance to 

be in B's favour, and that he ought to receive £x. Then, to make A 

worth £500, his debt to B must be paid, and he must receive £500 

besides : that is, he must receive 500 -|- jr. But of this one-half only 

is given, or ^ (500 + x), out of which, when he pays £j: to B, he will 



' —a: 


Now B gets £l00 and also £x from A, and will therefore have 
100 -j- J? pounds. And, since they have then equal sums, 

^2^ -a; =100+* 

( X )2 500 + x^2x — 200 + 2x 

2a; + 2a:— a: = 600 — 200 
or 3a; = 300 and j: = 100 

Therefore A owes B £l00. 

Problem III. A traveller proceeds along a road on which, at 
various intervals, are found direction-posts variously numbered, point- 
ing north or south. As soon as be reaches No. 1, be proceeds in the 
direction pointed out by it till he reaches No. 2, and so on. He finds 
the first direction-post after be has travelled 16 miles north^and he 
finds also that be changes his direction at every post which he meets 
after the first ; * that the distance between every two posts is double 

* The problem does not say whether he changes his direction at the 
first, or not. 


that between the precediDg, and that, at the fiflh post, be is 86 miles 
north of his first position. What is the arrangement and character of 
the posts ? 

After travelling 16 miles north he reaches a post, and we are not 
told whether he is there directed to go on or turn back. Let us 
suppose the former, and that he travels x miles further north before 
he reaches the second post. He will then be 1 6 + ^ miles north of 
his first position. At the next post he has to turn back, and proceed 
Itx miles south. Here two cases arise. If 2jr be less than 16+^9 
the third post will be north of his first position by 16 + j: — 2x 
miles; but if 2 x be greater than 16 + ^) the third post will be 
south of his first position by 2x — (16-f-^) miles. Suppose the 
first; then between the third and fourth post there are ^x miles, 
and he has to go north from the third, therefore he will meet the 
fourth post at 16+j: — 2j: + 4x miles north of his first position. 
At the fourth post he turns south, and after proceeding Qx miles, 
meets the fifth post, his position north of his first position being then 
IC+J? — 2x'\-^x — Qx miles. But this by the problem is 86 
miles: consequently 

16+a:— 2a; + 4ar-8a: = 86 

or 8a;— 4a; + 2x— a; = 16—86 

or 6x = 16-86 

in which there is an impossible subtraction. Let us now try the other 
hypothesis, and suppose that at the first direction-post he has to turn 
south, and that he finds the second after proceeding x miles south, 
or at 16 — X north of his first position, or j:— 16 south, according as 
X is less or greater than 16. By the same attention to the expressed 
conditions of the problem, we find that 

y J/ Z 16— a:+2x— 4a;+8a; = 86 

%x—Ax + 2x^x = 86—16 

6a: = 70 
a; = 14 

Consequently the positions of the posts are as follows : 

South North ^ 

(4)C3- <2)(^ 4^(1) ^(3) (5) 

26 +2 16 30 86 




Under each is marked the number of miles at which it is from his 
first position. 

If we collect together and look at the correct and incorrect equa> 
tions which we ha?e found in the three preceding problems, we shall 
have the following : 

Problem I. ' 
Incorrect, 60-f-ar = 2 (35 + a;) or a; s= 50—70 years after 1830 
Correct, 50— a; = 2(35— a;) or a? = 70—60 . . before . . 

Problem II. 

500— X . ,^^ 200—500 , . , ,5 ^ . 
Incorrect, ' . +a?= 100— a?ora: = r which BowesA 

Correct, — -^ x = 100-f-a; orx = . . A . . B 

Problem III. 

Incorrect, 16 + a: — 2x+4x — 8a? ^ 86 or a: = — ^ — milesnorth 


Correct, 16— a: + 2ar— 4x + 8ar = 86 or a: = — z — miles south 


From which, as well as from other instances, the following principle 
is clear: 

When the value of x, deduced from an equation, contains an im- 
possible subtraction, both the equation and the meaning of x have 
been misunderstood, and require alteration. 

1. To correct the equation, alter the sign of every term which 
contains x once only as a factor, 

[The words in italics are inserted to remind the student that we 
cannot draw any conclusion from the preceding as to equations 
which contain such terms as xx, xxx, &c. All our equations have 
been of the first degree,] 

2. To correct the result, invert the terms of the impossible sub- 
traction (that is, change SO'^TO into 70-— 50), and let the quality of 
the answer be the direct reverse of that which was supposed when the 
incorrect equation vras obtained. Thus, change years after into years 
before ; property into debt; distance measured in one direction into 
that exactly opposite ; and so op. Or, whatever alternatives it may 
be possible to choose between in assuming x, provided one be the 
direct reverse of the other, then, if one alternative produce an impos- 
sible subtraction in the value of x, the other is the one which should 
have been chosen. 


An equation generally obliges us to take a more extensive view of 
the question than the words of the problem will bear, and will fre- 
quently shew that the view taken of the problem is not in every part 
a consistent whole. 

In the preceding questions we have taken care to leave every 
possible case open in the statement of the problem : thus we have 
said (Problem I.), ^ Give the date at which A is twice as old as B,** 
not " How long will it be before A is twice as old as B V* because 
the latter would be tacitly assuming that the event is to come, 
whereas it would be found out that the event is past, and the implied 
statement is erroneous. We write underneath the correct and incor- 
rect mode of enunciating the question. 

In the year 1830, A's age was 
50 and B*s 35. Give the date 
at which A is twice as old as B. 


20 years before 1830, or in 

In the year 1 830, A's age was 
50 and B*s 35. When will A 
be twice as old as B ? 


Never; but A was twice as 
old as B 20 years ago. 

The chance of an impossible subtraction occurs in both; but in 
the first it arises fix>m a question being left open to the student, who 
may choose the wrong alternative; in the second it arises from a 
wrong alternative being already tacitly assumed in the problem. 

The alternatives presented by a problem may generally be ascer- 
tained with ease ; but if not, the equation itself is frequently a guide. 

Anomaly 3. Let 3a;— 10 = 2a;— 8 

3a;-2a? = 10-8 
a; = 2 

On verifying this equation we find each side to contain an 
impossible subtraction, for 3 a: — 10 is 6 — 10, and 2jr— 8 is 4 — 8. 
After what has been said on the last case, we need not dwell upon 
this; the problem in the next page will furnish an instance. 
Somewhat similar to this is a mistake in the process which may 
introduce an impossible subtraction into both numerator and deno- 
minator of the answer. If, in solving ax-^b = cx + d, thus, 

ag^mcx sz if— 6 or jr = -^^, we afterwards find a less than c, and 

fl— c 



d less than h, it is a sign that we ought to have chosen the process 
ex — ax ^h — 4, This is a mistake in the order of operations only, 
and not in the conception of the problem. 

Problem. Divide the number 13 into two parts, in such a 
manner that three times the first may exceed half the second as much 
as the first exceeds 4. 

It will be found from the resulting equation, namely, 
- o 13 — X . 

3a; r— = a;— 4 

(where x is the first part) that jt := 1, and therefore the parts of 13 
required should be 1 and 12. But the problem is then impossible, 
for three times the first does not exceed lialf the second. But if the 
words "fell short of'' be substituted for "exceed" throughout the 
problem, the equation becomes 

13 — j: o a 

the answer to which is j: := 1 as before, and the problem is possible ; 

for 3j: or 3 falls short of i(13 — dr) or 6, as much as x or 1 fells 

short of 4, 

We now ask how it happens than an equation gives a rational 

result, by which, when it is tried, the equation itself is proved to be 

irrational; and are we to conclude that no answer to an equation 

holds good until it has been tried upon the equation, and found to 

satisfy it rationally ? Let us examine our first instance. The 

equation is 

3a:-10 = 2ar-8 

and the answer x ss 2, when applied to the equation, gives 

6-10 = 4-8 
^It so happens that the rules for solving an equation give the same 
answer to both of the following : 

3a:— 10 — 2a:-8 
10— 3a: = 8-2a: 
And the following example will shew how this happens : 

ax—b = cx^d b^ax = d-^cx 

ax— ex ^ b — d b — d = ax^cx 

X = m both. 

fl— c 

Hence, when an anomaly of the kind treated in this article occurs. 


it is the sign of a misconception of the right way of viewing the 
problem, but of a misconception which no way affects the result. 
Anomaly 4, If we first solve the equation 

ax + b = cx-^d 

which gives X ss , and if it should happen 

(without our observing it during the process) that a=c or a— c=sO, 
we shall then find an answer of the following form, 


which is unintelligible ; because there can be no answer to the ques- 
tion, *' How often is nothing contained in d— b?** or, at least, if there 
be any answer, it is, '' Nothing, however often it may be repeated, 
yields nothing, and therefore cannot be repeated often enough to 
yield <^— 6.*' On returning to the equation, we see that the supposi- 
tion of a ^ c gives ax =zcx, so that, as far as the equation only is 
concerned, it is always true if ( = d, and never true if b is unequal 
to d. But in giving further explanation of problems which produce 
such equations, we shall employ the following principle, the propriety 
of which is obvious. 

When any supposition (such for instance as making a^szc in the 
preceding equation), makes the results of ordinary rules unintelligible, 
then, instead of making a exactly equal to c, let it be made very 
nearly equal to c, and observe the result : afterwards suppose it still 
nearer to c, and so on ; the succession of results will inform us whether 
any rational interpretation can be put upon the result of supposing 
a s= c, or not. We shall now try a problem in which the preceding 
difficulty will be found to occur. 

Problem. There are three trading companies, of 4000, 5000, 
and 9000 shares respectively, and all three would, if broken up, pay 
the same dividend upon their shares ; but if every shareholder in the / 
second advanced his company £10 on each share^and every share- ^ ' 
holder of the third advanced £l2 in like manner, then the first two 
companies together would have the same total assets as the third ; 
what is the dividend which each company could now pay ? 

Let X pounds per share be that dividend . Then, after the advances 
supposed in the problem, the three companies could pay x, x-^-iO, and 
» + 12 pounds per share respectively; which, taking their number of 
shares into account, supposes them to be in possession of 4000 j?, 


5000 (jr -f- 10), and 9000 (x -f- 12) pounds respectively : between 
which the last clause of the problem gives the equation 

4000a; + 6000(a: + 10) = 9000(a: + 12) 

(-r)lOOO 4a;-|-5(a:+10) =:9(a;+12) 

or 9a: + 50 = 9a; + 108 

In which we recognise the anomaly which is the subject of this 
article. Nor can we in this case, as in page 18, account for the 
impossibility by supposing that we have mistaken the problem, and 
that the three societies are at first insolvent, and are in debt upon 
each share ; fof , if we make this supposition, and let x be the amount 
they fail for upon each share, we have already seen (and the student 
must make himself sure of the same in this particular case) that the 
resulting equation will be 

50-9a; = 108-9ar 

which is equally impossible with the former. We shall, therefore, 
now try the consequences of a slight change in the conditions, agree- 
ably to the preceding principle. For instance, we will suppose the 
third society to have only 8999 shares, instead of 9000. The equation 
then becomes 

4000ar + 5000(a: + 10) = 8999(a;+12) 
or 4000ar +5000a; + 50,000 = 8999 x + 107,988 

4000a:+5000x-8999a: = 57,988 
or X — 57,988 

The answer therefore is, that each society could at first pay £57,988 

per share. Let us try the effect of a still smaller change, and suppose 

the third society to want only the hundredth part of a share of 9000 

shares, that is, to have 8999 — - shares. Then we have 

4000ar + 5000(a; + 10) = 8999^^(a?+12) 
4000a;+5000:r + 50,000 = 8999^a;+8999^ x 12 
4000a;+5000a;-8999:^a;= 8999:^x12-50,000 

100 100 

_!_,= 2^. 12-50,000 

( X ) 100 X = 899,999 x 12-6,000,000 

X = 5,799,988 


or each society can pay £5,799,988 per share. In the same way, if 

we take the third society at 8999 r^r^ shares, we shall yet have a still 

greater answer, and so on. A similar result would be obtained by 
increasing the 4000 or 5000 shares of either of the other societies. 
Therefore, the answer to the preceding question is, that no number is 
great enough to satisfy the conditions of the question ; but that if these 
conditions be slightly altered, an answer may be found, which answer 
is a greater number the slighter the alteration just alluded to. Dis- 
missing the problem, which we have only introduced to shew that 
such anomalies may arise in the application of algebra, we return to 
the consideration of similar equations. 
The solution of 

ax = bx -{• c 

IS a; = 

a— 6 

in which, should it happen that a ^ (, the answer is unintelligible, 

being --, and the equation impossible, being 

ax ^ ax -f" c 

but if a exceed h by any quantity, however small, the equation and 

its answer are both rational. Let a exceed h by the fraction of unity 

— , then the equation becomes 

\h + -^x — bx-{'C 

or ia; H — = ia; + c 


(—)bx — =c (x)m X = mc 

The same might be obtained from the preceding answer, for 

c c 

X = r =:i -^ = mc 

a — b 1 

To make a exceed ( by a small quantity, — must be small, that 

is, m must be large ; and in this way we may get an equation whose 
answer shall be as large as we please. For instance, let c be 1 ; and 
suppose we want an equation of the preceding form whose answer 


aball be 1,000,000. Let m = . Then such equations as 

the fbllowing hare thjg answer j: = 1,000,000. 

^7 A number which does not satisfy an equation may, we can easily 

conceive, nearh/ satisfy it. But this word nearly is too indefinite for 
our purpose, as we shall now shew. Suppose we have the equation 
7r = 2jr 4- 3, and we try whether j: = 1 will satisfy this equation. 
It will not; for the first side is 7 (x being 1), and the second side is 
5 ; that is, the first side, instead of being equal to the second, is 
greater by 2. The same result applies, in the same words, to the 
equation 7j: = 5 j? -|- 19fl8, if we try x = 1000 upon it ; for the first 
side becomes 7000, and the second 69^8. Shall we then say, jr := 1 
satisfies the first as nearly as x *= 1000 does the second? Is 7 as 
near to 5 as 7000 to 6908 ? If we look at the differences only, we 
must answer yes ; for 

7-5 = 2 

7000 - 6998 = 2 

but in the common use of the word near* (to which it will be conve- 
nient to keep) it would be said that 7000 is nearer to 6998 than 7 is 
to 5. In the first case, the difference is 2 out of 7000 ; in the second 
it is 2 out of 7. Keeping to this meaning of the term, we shall in 
future consider ax and ax-^c as nearer to equality when x is greater 
than when it is smaller. And in this sense we say, that when a 
problem leads to such an equation as 

ax ^= ax -{-c 

the result is, no number is great enough to be an answer to the 

problem ; but the greater any number, the more nearly is it an answer 

to the problem. 

Tried by the common rules (which, in this case, lead us too far) 

the answer to the preceding equation is 

c c 

X = or - 


* In the making of a bargain, ^6998 would be considered as the 
same price, within a trifle, as ^7000 ; but any thing at ^7 would be con- 
sidered dear as compared with the same at ^5. We might multiply 
instances in which the same quantity would he considered small under 
some circumstances, and great under others. 


and it is customary to say, that -- means an injinite numbery and that 

the answer is in/initefy great. Taken literally, such phrases are un- 
meaning, because we know of no number which is infinitely great, 
that is, greater than can be counted or measured. But the word 
is often used; and we shall, therefore, adopt it with the following 
meaning : 

By saying that - is infinite when 9 =: 0, we are to be understood 

as meaning no more than a short way of expressing the following : — 

1 1 

When g is small, - is great ; if g become still smaller, - is still 

9 9 

greater, and so on; so that - may be made to exceed any given 

number, howeyer great, if g be taken sufficiently small. And when 
we say the answer to a problem is infinite, we mean that no number' 
is great enough to satisfy the conditions of the question ; but that 
any great number nearly does so, a still greater still more nearly, 
and so on ; so that the problem may be answered within any degree 
of nearness (short of positive exactness) by taking a sufficiently great 

Anomaly 5. The solution of 

ax -{-b = ex -hd 


IS j: = 

a — c 

Now, if it should happen that, after the solution of this equation, 
it becomes necessary to suppose a equal to c (as in the last case), and 
b also equal to d, the answer 

^=* must be written ° 

which has no meaning. On returning to the equation we no longer 
find any necessity to give x one value rather than another ; for, if a 
be = c, and b =sd, then ax -{-bis equal to ex -^ d, whatever x may 
be. Therefore the answer to the question is, that every possible value 
of ;p satisfies the conditions. 'We shall apply this in the following 

Pboblem. Is there any number such that a times one less 
than the number, added to 6 limes two more than the number, is 
exactly c times the number; a, 6, and c, being given numbers or 
fractions ? 

-; V 



Let X be that number, then 

fl(ar~l) + J(a;-f2) = ex 
ax — a-hbx -{'2b = ex 
ax + bx^cx = a^2b 
(a + ft — c)x = a — 2ft 

a — 2b 

X = —~. 

a -\- b — c 


J^ — 1 = — r-7 — a (a: — 1 ) = — -—. 

a + 6 — c ^ ^ a-|_^_c 

a(r - 1) + i(x + 2) = ^^=l*f « £^ifll*l 

' a + 6 C fl+6 C 

a — 26 

= C X — r-T = CX 

a + 6— -c 

Suppose that o is 8, 6 is 4, and c is 12, or that the problem is the 
following : — What number is that, one less than which multiplied by 
8, added to the product of 2 more and 4, is equal to 12 times the 
number? Here we find a — 26 =: 0, and a-\-b — c = 0; so that the 

preceding answer takes the form - : trying this case by itself, to find 

out the reason, we have the equation, 

8{x-])+4{x+2) = 12a: 
8a:-8+4a: + 8 = 12a; 

12a: = 12a; 

which being always true, the answer is, that every number and frac- 
tion whatsoever, which is greater than 1, satisfies the conditions of the 
problem ; a result corresponding to the interpretation we have already 

seen reason to put upon the form -. 

We shall now proceed to the solution of some prpblems, in which 
the preceding method and principles will be applied. As instances 
will be taken from difierent parts of natural philosophy, we shall 
divide them into sections^ and state at the head of each the facts on 
which the solution depends. 

Examples. — Section I. Specific gravities, — By the specific 
gravity of a body is meant the number of times which its weight is 

iriRST DEGREE. 27 

of the weight of an equal bulk of water. Thus, when we say that the 
specific gravity of brick is 2, we mean that a mass, say a cubic foot 
of brick, weighs twice as much as a cubic foot of water.* A cubic 
foot of water weighs about 1000 ounces avoirdupois.f 

Problem I. One pint of water is added to three pints of milk 
(specific gravity 1*03) : what is the specific gravity of the mixture .' 

If a pint of water weigh m ounces, then one pint of milk weighs 
1*03 x^ra ounces, and the whole four pints of the mixture weighs 
wi + (l-03»i)x 3, or iw + 3'09>» ounces. But four pints of water 
weigh 4m ounces; therefore the specific gravity of the mixture (see 
the preceding definition) is 

wi + 3-09m 409 TO 4*09 , r\c\c\r 

:; or -— — or — -— or 1-0225 

4m 4m 4 

N.B. Here is an instance of a quantity m introduced for conve- 
nience, and which disappears in the process. The question itself is 
too simple to require an equation. 

Problem II. A number of cubic feet (m) of a substance whose 
specific gravity is a, is mixed with n cubic feet of another, having tlie 
specific gravity b. What is the specific gravity of the mixture ? 

It 19 plain that the w -f-n cubic feet of mixture will be as heavy 

as ma-^-nb cubic feet of water. Therefore the specific gravity re- 

. , . ma-hnb 
quired IS ; . 

Exercise. Try to shew that ^^ . ^ must lie between a and 6. 
' m -fw 

Problem III. How much of a specific gravity 2 must be mixed 
with 20 cubic feet of specific gravity 10, in order that the specific 
gravity of the mixture may be 5 ? 

Let J" be the number of cubic feet in that quantity. Then the 

whole 20+^ cubic feet of mixture has the same weight as 

20 X 10 4- J" X 2, or 200 -|- 2 J? cubic feet of water. Therefore the 

. . 200-|-2j: , 
specific gravity is " , and 

?9^±ll = 5 or 200 + 2;r = 5(20+a;) /. x = 33J 

20 + x 
Generalisation of the preceding. How much of a specific gravity 

• Atmospheric air is often taken as the standard of gases. Water is 
tbout 800 times as heavy as air. 

t Easy to recollect, and remarkably near the truth. Let the student 
dedace it from Ar. Art, 217. 



' a must be mixed with m cubic feet of a specific gravity 6, in order 
that the specific gravity of the mixture may be c ? 

Let X be the quantity required. Then x cubic feet of specific 
gravity a weigh as much as ax cubic feet of water, and m cubic feet 
of specific gravity h as much as hm cubic feet of water. Hence the 
whole m -{- X cubic feet of mixture weigh as much as hm-^- ax cubic 
feet of water. Hence, as in the particular instance above, 

hm -^-ax 7 , V 

' — = c bm-{-ax = cCw-f^) 

= cm + cx 
hm-^cm = cx—ax or {b'-'C)m = {C'~'a)x 

X = ,m 

c — a 

This is rational when h is greater than c, and c greater than a ; that 
is, when 6 — c and c — a are possible. It is also rational when h — c 
and c — a are both impossible; since, in this case, the apparent irra- 
tionality arises from our having converted 

bm+ax = cm-\'Cx into bm-'Cm = ex— ax 

instead o( cni'^bm = ax— ex 

c b 

and the rational answer is a: := .m. In this case a is greater 

a — c 

than c, and c greater than 6. That is, this problem is rational when 

c lies between a and b. If c do not lie between a and 6, then a 

rational problem is formed, as in page 1 8, by supposing x the direct 

reverse of what it was last supposed to be ; that is, by supposing the 

m cubic feet of specific gravity b to allow of the substance of specific 

gravity a being subtracted from it, or to be itself a mixture already 

containing that substance. That is, solve this problem : How much 

of a specific gravity a must be taken from m cubic feet of specific 

gravity 6, so that the specific gravity of the remainder may be c ? 

The answer will be found to be 

c — b h — c 

X = ,m or a; = ,m 

C'—a a — c 

according as c is greater than both a and 6, or less than both. But 
here may arise another of the anomalies previously explained. Sup- 
pose, for instance, we ask how much of a specific gravity a(= 10) 
must be taken from m(= 20) cubic feet of a specific gravity 6(= 6), 
in order that the specific gravity of the remainder may be c (= 12). 



(Call this problem A.) Here, though the problem is evidently im- 
possible, the answer will be rational, being 

-=^!'» = ^0-20 = 5-20 = 60 

and the impossibility is detected, not in the form of the answer, but 
on looking at the problem, with which the answer is inconsistent ; for 
60 cubic feet cannot be taken from 20. The equation from which 
this answer results is 

1^5=i2f = 12 or 120-lOar = 240- 12a; 

20— X 
in which, with the answer jr = 60, the anomaly 3 explained in p. 19 
occurs. On correcting this equation, as done in p. 20, it becomes 

]0a;-120 = 12a:-240 or ^^""'^H^ = 12 

X — 20 
which is derived from the following problem : — From how much of 
specific gravity « (= 10) must m (= 20) cubic feet of specific gravity 
*(=:6) be taken, in order that the specific gravity of the remainder 
may be c(= 12)? (Call this problem B.) 

Whether the answer x = 60 is to be called possible or impossib^ftr'' 
depends upon the answer to the following question. Was problem 
B within our meaning or not when problem A was proposed ? that is, 
did we mean to take the one of the two, A and B, which should turn 
out to be rational ; stating A, because we supposed it, before examina- 
tion, to be that one ? or did we mean to confine ourselves within the 
limits of the literal meaning of A ? Because, in the first case, the 
answer is, that we have chosen the wrong alternative, that the other 
should have been chosen, and that the answer is x zs^QO; in tlie 
second case, the answer is that the problem is impossible.* 

Problem IV. — The specific gravities of gold and silver are 
19i and 10|; a goldsmith ofiers a mass of ^ of a cubic foot which 

• I have here stated this purposely, because the matter of convention 
is more obvious in this problem than in that of page 16, where there is 
an even chance of our choosing the wrong alternative at first. The 
problem before us will appear strained, simply because the alternatives 
of north and south of a post come more frequently into practice than 
those of taking a known from an unknown, and an unknown from a 
known, mixture. I state this because some writers on algebra seem 
to imply, by making this sort of extension rest only on the most obvious 
and usual examples, that they wish the student to consider it as not con- 
ventional, but necessary. 



he asserts to be gold, and which is found to weigh 260 pounds. 
Can it be all gold ? if not, may it have been adulterated with silver ? 
and, in that case, in what proportion were silver and gold mixed ? 

Since a cubic foot of water weighs 1000 ounces, and gold is 19^ 
times as heavy as water, a cubic foot of gold weighs 19,250 ounces, 
and ^ of a cubic foot weighs 4812^ ounces, or 300 pounds 12^ ounces. 
Therefore the mass cannot be all gold. Again, a cubic foot of silver 
weighs 10,500 ounces, and i of a cubic foot weighs 164 pounds 
1 ounce. Therefore the mass is heavier than its bulk of silver, and 
lighter than its bulk of gold, and consequently may be a mixture of 
the two. In this case, let x be the quantity of gold, x being a fraction 
of a cubic foot; then \ — x is the remainder of silver, and 19250 j: is 
the number of ounces the gold weighs, and 10500 {\ — x) the same 
for the silver. But the whole weight is 260 pounds, or 4160 ounces ; 

19250a: + 10500(J-.ar) = 4160 

1535 307 3 , 12 . I 

X = = = Tz nearly, or -— of - 

8750 1750 17 ^' 17 4 


/. about 12 parts out of 17 are gold, and the rest silver. 

This is a case of the celebrated problem first solved by Archi- 
medes,* and maybe generalised as follows: — In what proportions 
must substances of specific gravities a and b be mixed, so that the 
specific gravity of the whole may be c ? To one cubic foot of the 
first let there be x cubic feet of the second, to produce the mixture 
required. Then the 1 cubic foot of the first, weighing a cubic feet of 
water, and x cubic feet of the second, weighing bx cubic feet of 
water, the whole 1 + x cubic feet weigh a-^bx cubic feet of water. 
But since its specific gravity is to be c, it weighs c(l-i-j:) cubic feet 
of water; therefore, 

a + bx = c(l + x), or ^^ = -37 
to which the remarks in pages 28 and 29 apply. 

Examples. Section II. The Lever, Generally speaking, a bar, 
when suspended by any point in it, will only rest in one position, 
namely, hanging downwards ; but if properly loaded with weights, it 
may be made to rest in any position, and is then said to be a lever in 
equilibrium, that is, evenly balanced. The rules are, 1. Treat the 

• See the Penny Cyclopadia^ vol. ii. p. 277. 


weight of the bar itself as if it were all collected in its midddle point. 
2. Call the number of pounds* in any weight, multiplied by the 
number of feet by which it is removed from the point of suspension, 
the moment of that weight ; then a bar will be in equilibrium when 
the sum of the moments of the weights on one side of the pivot is the 
same as that of the weights on the other side. If the bar be not 
suspended by its middle point, the weight of the bar itself must be 
taken into account as if it were all collected in the middle point. 
If the sum of the moments on one side be not equal to that on the 
other, the side which has the greater sum will preponderate. 

-Problem I. A bar 18 feet long, weighing 40 pounds, has 
weights of 12 and 20 pounds at the two ends. Where must tlie pivot 
be placed, so that the bar may rest upon it? 

A C D B 

12 lb. 40 lb. 20 lb. 

Let the weight of the bar (40 lbs.) be collected at the middle 
point C. Then AC = CB = 9 feet. We do not know on which 
side of C to place the pivot, which may produce an incorrectness in 
the equation similar to that in Anomali^ 1. page 12. This, however, 
will not affect the result. Let the pivot be at D between B and C ; 
that is, let 12 lb. at A and 40 lb. at C balance 20 lb. at B. Let 
AD=j: feet. Then CD = (j: — 9) feet, DB = (18 — ^) feet. The 
moments of the weights are 12jr, 40 (j: — 9) and 20(18 — x); and by 
the preceding principle, 

12a;+40(a;-9) = 20(18-a:), or a; = 10; 

therefore the pivot is 1 foot to the right of the middle point, and the 
problem has been rightly interpreted, because x — 9 and 18 — a:, tried 
by the result, are both possible. If we had imagined D to be on the 
left of C, and ADz=j; as before, we should have supposed the weights 
at C and B conspiring to balance that at A, and DB =18 — x as before, 
but CD = 9 — X instead of ar — 9. The equation would have been 

12a: = 40(9-a:) + 20(18-a:) 
or . 12a? - 40(9-a:) = 20(18-a;) or a: = 10 

* Any other units may be substituted for pounds and feet j but car© 
must be taken to use the same units throughout the whole of each 


differing froip the first in liaving 

-40 (9 -a:) instead of +40 (a: -9). 

The solution j: =: 10, would have shewn us the Anomaljf 1, page 12. 
To generalise this problem, let the length of the bar be /, its 
weight* W, the weights at the let\ and right extremities P and Q. 
Then let AD = jr, which gives (supposing the pivot right of C), 
CD = j — 4/, DB = / — X. The equation becomes 

Vx+Vf{x^\l) = Q(/-a:) 

_ W/ -f-2Q/ _ W+2Q I 

whence x — 2P + 2Q + 2W " P + Q + W ^ 2 

Exercise. Prove from the value of .r just found that x is greater 
than, equal to, or less than, \ /, according as Q is greater than, equal 
to, or less than P. 

The preceding problem contains the principle of the steelyard. 

Paoblem II. If a bar 20 feet long, weighing 6 pounds, be sup- 
ported at 9 feet from the lefl extremity, how must we place upon it 
weights of 16 and 17 pounds, 7 feet apart, so that the whole may be 
balanced (the 16 lbs. being supposed on the left) ? 

16 lb. 6 lb. 17 lb. 

\ 1 1 T I ' "i 

A E D C F B 

Let C be the middle point, D the pivot, and E and F the places 
of the weights. Then AD =9 feet, BD = 11 feet, AC =10 feet, 
EF = 7feet, DC=1 foot. LetAE=x; then ED = 9 — x, DF = 
AF— .AD=AE + EF— ADrsx-f 7— 9 = x— 2. The system 
therefore consists in 

16 lb. at dist. 9 — j: from pivot; moment 16(9 — x), 

which balances 

6 lb. at dist. 1 foot from pivot; moment 6 X 1 or 6 

17 lb. at dist. x — 2 feet from pivot; moment 17(x — 2). 

Hence 16(9-^:) = 6-fl7(x-2) a: = 5^ feet. 

The equation has been rightly formed, for 9 — x and x — 2 are 
both possible. If we had supposed E and F to fall on the same 
side of D, the resulting equation would have been 

* That is, W is the number of pounds, or other unit, in its weight, 
and / the number of feet, or other unit, in its length. 


16 (9-a;) 4-17(2-0:) = 6 
or 16(9-a;) = 6-17(2-0:) a: = 5^ feet: 

from which the same value of x is obtained, but 2 — j: is impossible. 
Even if we had made the evidently impossible supposition that £ and 
F are both on the same side of D as C (amounting to supposing that 
the weights at E, F, and C are counterbalanced by no weight at all 
on the opposite side), the equation which results, treated by ordinary 
rules, would present no direct sign of impossibility until we came to 
compare the result with the equation. In this case, 

A D C E F B 

ifAE=a:,wehaveDC=l,DE = T— 9,DF = (j:— 9) + 7=j: — 2; 
and the moments of the weights at C, E, and F, are therefore 6, 
\Q(jc — 9) and 17 (x — 2). To trace the effect of the preceding 
supposition, namely, that there is no weight on the left side of the 
pivot D, we must see what will follow from supposing 

6-i-16(a:-9) + 17(a:-2) = 

as if such an equation were possible. This gives 

6 + 16a;-144+17j:-34 = 
33a;-172 = 33 a; = 172 a; = 6^ 

the same as before. But the preceding equation is impossible, since 
the addition of three quantities must give more than nothing. At the 
same time, we see that x — 9 is also impossible. 

We must here make a remark similar to that in page 14. Tlie 

a:— (c — J) = 0, or a: + 6 — c = 

is possible, since it merely indicates that j:=:c — h. But the equation 

x + (J)-^c) = 

is impossible. Nevertheless, when b is greater than c, x-\-{b — c) is 
the same thing as x-^-h — c; and by attempting this conversion 
when h is less than c, we might be led to the impossible form 

x-f (6-c) = 0, 

where we should have adopted the rational form 

o:— (c — i) = 0. 

From this we conclude, that if we meet with 

X -^p = 0, 


it is a sign that, in forming the quantity p, we have inverted the order 
of the terms in a subtraction ; that is, we have supposed p was b — c 
when it should have been f — b. Let us call the latter q ; then if 
we had proceeded correctly, we should have had 

a: — (7 = 0, or a; = j. 

The problem in page 32, generalised, is as follows. A bar / feet 
in length, weighing W pounds, is supported at a feet from its left 
extremity. How must we place P and Q pounds (P being on the 
left) m feet asunder, so that the bar may be balanced ? The equation is 

P(a-a:) = W(J/-a) + Q(a; + m-.a) 

^ "■ P + Q 

Exercise. Supposing the bar to be continued to the left of A 
and the right of B, but the continuation on either side to have no weight, 
explain, as in page 33, the case where W=20 pounds, /=50 feet, 
a=s5 feet, P = 4 pounds, Q^7 pounds, w = 10 feet. 

Examples. Section III. Miscellaneous. Problem I. — A straight 
line AB, 10 inches long (continued both ways), is cut by the point 
C at 7 inches to the right of A. Where must the point D be placed, 
so that AC may bear the same proportion* to CB which AD bears 
to DB? 

■■ " ■■■ ■ I — 1 1 1 — 

A C B D 

Here ACi=i7 inches, CB=£3 inches. The point D must be 
either between A and B, or to the right of B, or to the left of A. Or 
there may be (for any thing we have shewn to the contrary), more 
than one such point; for instance, one such point to the right of A, 
and another to the left. But if we consider the conditions of the 
problem, it will appear impossible that D can lie any where but to 
the right of B. For, suppose D to be placed between A and B, say 
between C and B; then, according to the problem, AD (greater than 
7 inches) contains B D (less than 3 inches), the same times and parts 
of a time which AC (7 inches) contains BC (3 inches). This, the 

* To those who have never used this term mathematically (see 
Ar. 177, 178), we may state, that a bears to b the same proportion 

which c hears to d, when the fraction — is the same as -r* 



least consideration will shew, cannot be. For a similar reason (this 
let the student explain), D cannot lie between A and C. Neither can 
D lie on the left of A; for then, of AD and DB, AD the less roust 
contain DB the greater, aaAC (7 inches) contains CB (3 inches), 
or 2 1 times, which is also impossible. Let us then suppose D on 
the right of B, and let Ap=:jr inches; then BD = x — 10 inches. 
By the problem, x bears to x — 10 the same proportion as 7 bears 
to 3 ; that is, 

_i^=^ (x)3(x-10) 3x = 7(^-10) 

which gives j: ^ — - = 17i inches. 

If we had supposed the point D to lie between A and B (say 
between C and B), then AD being x, DB would have been 10 — j, 
and the equation would have been 

* 7 

= -, which gives x ^ 7 ; 

10— a: 3 

that is, D coincides with C. This is a case which we have not put 

among what we have called anomalies, because the result, though not 

expected, is intelligible without further explanation. It implies, that 

if we would place D between A and B, so that A D should be to 

DB in the same proportion as AC to CB, D must occupy the 

same place as C. But if we suppose D to lie on the left of A, 

and let x stand for AD, we have DB ^ 10-f ^> and the equation 


'^ I, or 3a:-7a; = 70, 

lO+x 3' 

''hich presents the anomaly in pages 14 . . 19. This shews that we 
^ave nqeasured x or AD in the wrong direction, and that if it had 
'>een made to fall to the right of A, the equation would have been 

Ix—Zx = 70, or X = 17J inches, 

which we have found before. 

Now let the problem be altered so that C shall stand in the middle 
^tween A and B. For instance, let AB =:12 inches, and AC = 6 
iQches. Is there a point D, not coinciding with C, such that AD 
tears to D B the same proportion as A C to C B ? Certainly not ; for 
AC contains CB once exactly, but if D be placed right of B or left 
of A, AD is greater or less than DB. But if a point be placed at a 



great distance from A or B on either side, AD contains DB very 
nearly once (see p. 24) ; and by removing D to a sufficient distance, 
A D may be made to contain D as nearly * once as we please. We 
might therefore expect, if we attempt to find D in this case by an 
equation, such an anomaly as that in page 21, explained in page 24. 
If possible, let the point D, on the right of B, satisfy the conditions 
of this problem. Let AD = jr. Then DB = a:— 12 and ^C = 6, 
C B = 6. Therefore the equation of the problem is 

= -7 = 1 or a? = ar — 12 (See page 21.) 

a:— .12 6 

The generalisation of the preceding problem, in supposing AB=ia, 
AC =6, and AD=j:, and placing D on the right of B, gives the 


X b ah 

or a: = 

jr— a a — b 26 — a 

In order to meet all the cases which may occur in the application 
of algebra, we will now take a problem in which the answer will go 
beyond the notion which was formed when the problem was proposed, 
not because the thing proposed to be done is impossible, but because 
the answer is not within the limits of what is usually necessary or 
convenient. For instance, in ordinary arithmetic, a figure placed on 
^ the figbt of another means that it is to be multiplied by ten before 
adding it to the other. Thus 24 is 2 X 10+4. We do not commonly 
use fractions in the same way : thus 2]^ 4 never stands for 2^x10+4, 
or 29 ; but it might do so if we pleased ; similarly 3J 2J^ might stand 
for 3jx 10 4-2 J, or 34^. We now propose the following 

Problem II.' What is that number, consisting of two digits the 
sum of which is 10, and which is doubled by inverting the digits ? 

We see that 91 is not double of 19, nor 82 of 28, nor 73 of 37, 
nor 64 of 46. Therefore, with the restrictions on the decimal system 
usually adhered to in practice, the problem is impossible. What sort 
of answer then are we to expect if we reduce the problem to an 
equation ? Let x and y be the digits in question : then 

a;+y = 10, or y = 10— a:. 

* For instance, place D on the right at a thousand times the distance 

of B from A. Then A D is to D B as 1001 to 1000, or in the proportion 

of 1 + TT™- to 1, which is very nearly the proportion of 1 to 1. 


The number formed by placing tlie digit x before y, is IOjt -|-y 
(just as 24 18 2 X 10 + 4y 58 it 5 X 10 + 8, &c.) ; and when the 
digits are reversed, the number is lOy+x (just as 42 is 4x10 + 2, 
85 is 8 X 1 + 5y &«.). By the problem, the second is double of the 
first; ihatis, 

lOy+a: = 2(10a:+y), or 20a; + 2y. 
Therefore lOy— 2y = 20ar— or, or 8y = 19ar. 
But y = 10—0?, or 8(10— a:) = 19ar; 

therefore, a; = -- = 2--, y = 10— a: = 7--. 
^ 27 27* ^ 27 

The answer therefore is, that if it be understood that none but the 
usual single digits shall be placed in the columns of units and tens, 
the problem is impossible ; but that, if the method of writing numbers 
be estended, so that a fraction placed before a fraction shall be con- 
ndered as meaning 10 times as much as when it stands alone, then 
tiie problem is possible, and the direct and inyerted numbers are, 

2?5 tL and 7^ 2?^. 
27 27 27 27 

^^ ^^ 7~ means 2?^ X 10 + 1^ and is 36^5 
27 27 27 27 27 

7^ 2~ means 7~ x 10 + 2% and is 73;^ 
' 27 27 27 27 27 

^ second of which is double of the first. 

We may then lay down the following : When a problem has a 
"^tional answer, that answer can only be used on the supposition 
^^ any usual method of combining whole numbers, which is neces- 
'^'y in forming the equation, shall also be applied to fractions. 

This principle, when introduced into common life, often induces 

^^ to suppose fractional parts of things which, in the strict and ori- 

S^^l meaning of the terms, have no fractional parts* Thus there is, 

properly speaking, no such thing as half a hone : there may be half 

**** body of a horse (as to bulk or weight), half the power of a horse 

v^hich is but the half of a certain pressure), or there may be a horse 

^ Wf the size, half the power, &c. of another horse — that is, we may 

"^ve any quality of a horse which can be represented by numbers ; 

DQt not the complete idea which we attach to the word, because it 

contains notions which have no reference to number or quantity. 



Nevertheless, we do not speak absurdly when we talk of a steam 
engine of the power of 20} horses, because it is there only the horses' 
power that we speak of, which can be numbered in pounds of weight; 
or of wolves eating half a horse, because we then speak only of a 
weight of flesh. Thus the problem — A horse can draw two tons; 
certain horses drew 5 tons, how many were there ? — is absurd in the 
strictest sense, but not so if we confine ourselves to the only quality 
of a horse which is concerned in the problem, namely, power of 
drawing: and we may either say, there was 2]^ times as much power 
as a horse possesses, or the power of 2]^ horses. The same remarks 
would apply to the following : The reckoning came to £5, and the 
share of each person is £2, how many persons were there ? — which 
cannot be solved in the strictest meaning of the words, but in which 
we may say that the whole reckoning is 2 } as much as that of one 
person, or that of 2} persons.* 

Problem III. There are two pieces of cloth of a and a' yards iik 
length. The owner sells the same number of yards of both sorts at 
b and 6' shillings per yard. If the remainders were then sold, of the 
first at c shillings a yard, of the second at d shillings a yard, the total 
prices of the two pieces would be the same. What number of yards 
was first sold of each ? 

N.B. To avoid using too many letters, it is usual to employ the 
same letter with one or more accents,t to signify different numbers 
which have some common point of meaning. Thus a and a' are the 
lengths of two pieces of cloth, b and b' the prices per yard of the first 
pieces taken firom each, and c and c' the prices per yard of the 
remainders. But a and a! differ as much in meaning as to the num- 
bers they may stand for, as a and b ; either may stand for any number 

Let X be the number of yards first cut off from each piece ; then 
the remainders are a — x and a'— x yards. The sums received for 
the first are therefore bx and b' x shillings; for the remainders, 
c{a — x) and d {d — x) shillings. Consequently, by the conditions of 
the question, 

• So when we say that the. yearly mortality of a country is 1 out of 
40} persons, we mean ^ out of 81. 

t The symbol a' may be read a accented, a", a twice accented, and so 
on. But a dash, a two dash, &c. are shorter, though not quite so correct 
in g^rammar* 


bx + c(a^x) = llx-\'d{al^x) 
bx+ac—cx = h'x'\-a!c'^dx 
bx^cx+cfx—b'x = a'c'^ac 
{b'\'d^V^c)x = dc'^ac 

^^ dt! — ac 
' ~ ft + c/— 6' — c 

Suppose we try to apply this to the following case : Let the num- 
her of yards be 60 and 80; let the prices of the number of yards 
taken from each at first be 10 and 9 shillings a yard ; and let the 
prices of the remainders be 4 and 3 shillings a yard. We have then 

0=60 a' = 80 J = 10 y = 9 c = 4 d ^Z 

— dd -^ac __ 80x3 — 60x4 _ 
^ ■" fc+i/— 6'— c "" 10 + 3 — 9—4 ■" 

an anomaly already discussed in page 25. We have there seen that 
it implies that any value of s will solve the equation, and this we 
shall find to be the case in the present instance. For if we return to 
the equation, we find it becomes 

10x+4(60-x) = 9ar+3(80-x) 
ot lOx+240— 4a: = 9a; + 240— 3a; 

w 6a:+240 = 6a;-f 240 

^Idch is true for all values of x. Hence the answer is, that in this 
piirticular case the total prices of the two pieces are the same whatever 
^^omber of yards be first cut off. 

* hA us now try another .case. Let the pieces be 60 and 80 yards^ 
^ in the preceding ; but let the first pieces cut off be sold at 5 and 
^ shillings a yard, and the second at 2 and 3 shillings a yard ; then 

a = 60 a' = 80 6 = 5 6' = 4 c = 2 c' = 3 

-_ oV — gc __ 80X3 — 60X2 __ 120 __ ^^ 
^""6 + c' — 6'— c 5 + 3 — 4 — 2 "" 2 ■" 

the number of yards cut from both is 60 ; that is, the wh)le of the first 
piece is taken, and 60 yards of the second, which are sold at 5 and 
^ shillings a yard (giving 300 and 240 shillings). Then the re- 
niaibder of the second (we need not mention the remainder of the 
fint, which being nothing, brings nothing), 20 yards, sold at 3 shillings 
a yard, brings 60 shillings. The produce of the first is 300, of the 
^ond 240 + 60 shillings, both the same, as the problem requires. 


The third case we will take is as follows : Let the pieces be 60 
and 80 yards; let the pieces cut off be sold for 7 and 3 '^shillings a 
yard, and the remainders for 5 and 2 shillings a yard, 

= 60 a' = 80 6 = 7 6' = 3 c = 5 c' = 2 

_ 80x2-^60x5 _ 160—300 
^ "" 7 + 2 — 3 — 5 ■" 1 

and contains an impossible subtraction. From the conclusions in 
page 18, we must suppose some wrong alternative has been employed 
in choosing x. But, on looking at the problem, no such thing 
appears ; x yards are to be first sold of each. But the problems in 
page 18 were purposely* put in a form in which the alternatives were 
obvious ; how then are we to widen the expressions used in stating 
this problem so that the problem, as we have given it, may be only 
one case out of two or more ? Remember tliat we alter no number, 
but only the quality of the result. The seller begins with 60 and 80 
yards of the two cloths in his possession, and ends with none of 
either, having in his pocket the same receipts from both pieces. Our 
problem says he first sells a certain number of yards from both, and 
the answer upon this supposition shews the problem to be impossible. 
We have been previously directed in such cases to alter the quality of 
the result; let us do this, and suppose he begins by buying the same 
quantity of both. We must preserve the condition that he begins 
with 60 yards of the first, and ends with none ; therefore, if he begin 
by buying 10 yards more, he must sell the whole 70. If so, he also 
buys 10 yards more of the second sort, and sells the whole 90. But 
as we are to alter none of the numbers, but only change their names, 
if he buy more he buys at 7 (h) and 3 (Jf) shillings a yard ; and 
when he gets rid of all he has (not all he has lefty for that belongs to 
one particular alternative), he sells at 5 and 2 shillings a yard. 
Therefore the wider problem, of which the one proposed is one of the 
alternatives, is as follows : 

* We may here remark that the extension of a problem appears 
datura] or not according to the idiom of the language in which it is 
expressed. Thus, comparing together the case here given, and that in 
page 19, the former appears forced, because we have no very common 
word to denote either buying or selling as the case may be ; the latter 
appears natural, because the words *' give the date," implying asking for 
a time, either before or after a given epoch, are perfectly consistent with 



Extended Problem. 
There are two pieces of cloth 
a and a! ys^rds in length. The 
owner concludes a bargain rC' 
ipecting the same number of 
-yards of both sorts at ( and b' 
shillings per yard. If his stock 
of both were then sold, of the 
first at r shillings a yard, of the 
second at <f shillings a yard, 
the total results of the transac- 
tions in each sort of cloth would 
heihe same. 

Case first proposed. 
There are two pieces of clotli 
a and of yards in length. The 
owner sells the same number of 
yards of both sorts at 6 and b' 
shillings per yard. If the re- 
mainders were then sold, of the 
first at c shillings a yard, of the 
second at d shillings a yard, the 
total prices of the two pieces 
would be the same. 

The case first proposed leads, as before, to the equation 


X = 



6 + c'— 6'— c 

but the general case leads to this equation only when the owner is the 

seller in the bargain mentioned. If he be the buyer, he first pays bx 

and l/x for what he buys of each sort, and then sells the stocks a -f .r 

and rt'-|- X at c and c' shillings per yard. Therefore c(a + x) — bx 

is the balance in his favour from the first, and c'(a'-)-jr) — b'x from 

the iiecond. The equation is 

c{a + x)^bx = cXa+x) — b'x 


X = 

ac — fl'c' 

tlid idiom of our language. But if we translated these two problems into 
a laogaage, in which there was a word in common use, such as trafficking, 
to denote either buying or selling, and in which there was no usual way 
of asking for a date, without implying either before or after some other 
date, then tlie present extension would appear natural, and the one in 
page 19 forced. 

The equations of algebra of course take no cognisance of such differ, 
enoes of idiom, which is generally considered one of the great advantages 
of the science, though some regard it as a defect. The student must 
decide this point for himself, when he has had sufficient experience of 
the advantages and disadyautages arising from such extensions. 




which differs fh>m the former^ consistently with the rules in page 18, 
by an alteration of the sign of every term which contains Vy and ah 
inversion of the subtraction aV— oc in the result. When 

a = 60 a' = 80 J = 7 6' = 3 c = 5 c'=2 

we have already tried the alternative of supposing the owner to be 
seller in the first transaction, and have found an impossible subtrac- 
tion 160—300 in the result. If we now try the second alternative, 
and suppose him to be buyer, we shall get the rational answer 
300 — 160 or 140; which will be found to satisfy the problem. We 
might enter upon various other cases of the same question, but we 
shall leave them for the present to the student^ and state a problem 
which is in all respects analogous to the preceding, and presents 
similar alternatives in a difierent form. The equations of the two 
problems will be the same. 

Problem IV. The paper is a map of a country, of which AB is 
a level frontier. AH the roads rise from left to right, and fall from 
right to left, and all the miles mentioned are meant to be measured 
perpendicularly to the frontier on a level with it. CD is a parallel 
to the frontier (whether right or left of it is not stated), and T and 
V are towns on, above, or below (as the case may be), the line BD 
perpendicular to the frontier. P and Q are two frontier towns; 
R and S two towns, on, above, or below the points R and S, according 
to the position of C D. The roads rise or sink, as the case may be^ 
a number of inches per mile for every level mile from the firontier, 

as follows : 

From P to R 6 inches per mile. 

From Q to S V inches per mile. 

From R to V c inches per mile. 

From S to T c' inches per mile. 

Now T atid V are on the same level, BV is a level miles, and BT 

a' of the same. Required the distance B D and its direction. 

At P 

p< r~~- 

— _.^ 

— :r.::::=— ^^"cv. 



We have given this as an exercise, and shall merely point out the 
resulting equations on all the different suppositions. In all of them 
I stands for BD in miles. 

1. If D lie to the left of By 

either €(a+x) — 6a; = c\a' + x) — b'x 1 according as TV is 
or bx^c{a+x) = b*x — (!{ci+x) ) above or below PQ. 

2. If D lie between B and T (as in the figure), 

cia-^x) +ftx = (f(p!^x) -hb'x. 

3. If D lie between T and V, 

fta?— c(x— a) = Vx + cf{a!^x) when n' is greater than a, 
or bx + c{a^x) = b'x^c'{x — a!) when a is greater than a'. 

In t&e figure a is greater than cf, 

4. When D lies to the right both of T and V, 

bx^e(x^a) = b'x^c\x^a') 
or c{x^d) — fta: = c'(x^a)^b'x 

the first wh^^m« above, the second when TV is below, the level 
of P Q. ^BKKf^ o^ ihesQ seven equations can be altogether true ; 
and as only«j^ alternative in the set marked 3 is directly given in 
the conditions of the problem, six equations may need examination. 
But after the explanation of the anomalies (1), (2), and (3), any one 
of the six equations may be made to give the true answer. The 
anomaly (4) will be found to arise when h 4-c'^6'-f c, and (5) when 
b addition to this, €id=iac. 

The student may make this problem an illustration of every case 
which we have found to arise. 

Before proceeding to other forms of equations, we shall now, 
having found impossible subtractions arise in the solutions of problems, 
and having seen the method of interpreting them, proceed to the 
investigation of rules, by which the interpretation may be deferred to 
any stage we please of the processes, and by which the symbols of 
impossibility may be used as if they were real numbers, without 
creating error. 





Symbols such as 50 — 70 (page 15), 200 — 500 (page 16), have 
received the name of negative quantities. This expression is not a 
correct one, because 50 — 70 is not a quantity, but the contradiction 
which arises out of writing down directions to do that which is 
impossible to be done. But we have seen, pages 14.. 19, that 
50 — 70, when it is the answer to a problem, implies 20 things of 
some sort, as much as 70 — 50, but 20 things of a nature directly 
the contrary of the things first supposed; so that in forming the 
correct equations, these things diminish that which they increased in 
the incorrect one, and increase that which they diminished. 

Conceive a problem consisting of various steps, such as would 
&rise from joining two or more problems together, in such a way that 
the result of the first must be known before we can proceed with the 
second. If we solve the first problem, and find by an impossible 
result that we have chosen a wrong alternative, we should first retrace 
our steps, set the matter right, and when we have obtained an 
intelligible result, proceed to the second problem. But, we may ask, 
are there no rules by which this may be avoided, and by which we 
may continue the process, as if the result obtained had been rational ? 
To try this question, we must examine the consequences of proceeding 
with the symbols of impossible subtraction^ to see what will come of 
applying those processes which have been demonstrated to be true of 
absolute numbers. 

If we look at a problem which presents alternatives, we shall see 
that not only will the answers be of different kinds, according as one 
or other alternative is the true one, but the methods in which the 
unknown answer (jt) must be treated, in order to form the correct 
equation, are different in the two different cases. Thus, in page 15, 
Problem I., when x was years after 1830, the processes into which x 
entered were 50 -^-x and 35 +x; but when x was years before 1830, 


these processes were 50— » and 35— ». And this brings us to 
define what we mean by different problemt as distinguished from 
diferent altemativei of the tame problem. 

This distinction has been already tacitly laid down ; for, when we 
came to an impossible subtraction a— fr, we always looked for that 
modification of the problem, which, not changing absolute numbers, 
but only the sense in which they were taken, would hfve giren (—a 
as the answer. Or when we came to Anomaly I., page 12, in which 
not the answer, but the verification of it, was impossible, we chose 
that modification which, without altering absolute numbers, inverted 
the impossible snbtracticms in the equation. Hence, without per- 
ceiTing it, we have been led to make use of the following definition, 
[in which, however, it must be recollected, that we have obtained it 
entirely from equations of the first degree, and that it must therefore 
be considered as limited to such equations, and possibly capable of 
extension for equations of higher orders.] It is convenient to con- 
sider those problems only as different alternatives of the same problem, 
in which, 1. the absolute numbers employed are the same ; 2. the 
equations only differ in having inverted terms with different signs 

(such as — ' mstead of -\ — in page 13), or else m 

having the signs of the unknown quantity altered throughout (as 
50— X ^ 2 (35— x) instead of 50 + a: = 2 (35 -|- x) in page 15) ; 
3. the answers are either the same in both, or else only differ in the 
inversion of a subtraction (such as 70 — 50 instead of 50 — 70, in p. 15). 

We might propose a problem appearing to have alternatives, but 
which, in this view, is two different problems combined together, 
^or instance, A has £60, and is to receive the absolute balance that 
appears in B's books, whether for or against B ; but C, who has 
£200, is to take B's property, and pay his debts. AAer doing this 
it is found that C's property is three times that of A. What is the 
absolute balance for or against B ? 

If B have x pounds, the equation is 

3(60+:r) = 200+a; or a; = 10 
If B owe s pounds, the equation is 

3(60+ a;) = 200-x or x = 5 

We have here two difierent equations not reducible one to the 
other by any of the changes allowed in the preceding definition \ and 


therefore, so far as the preceding is considered as reducible to equa- 
tions of the first degree, the two cases are distinct problems. 

The step we now make is to apply the rules of algebra to the 
symbols of impossible subtractions, and afterwards to proceed cor- 
rectly with the inverted subtractions, that we may see, by comparing 
tlie results, whether the errors committed may be corrected or not by 
simple and general rules. And since we know that such a subtraction 
as 3 — 7 will, when set right, be 7 — 3 or 4, let us denote it for the 
present by 4, in which the bar written above 4 is not* a sign of sub- 
traction, but a warning that we are using the inverted form 3 — 7 
instead of 7 — 3. Thus we might say, that 10 — 14 should also be 
denoted by 4 ; but here we must stop until we have some further 
assurance upon this point ; for we cannot as yet reason upon such 
symbols as 3 — 7 and 10—14, since they represent no quantity 
imaginable, and we have not yet deduced rules. All we can do is 
to go to the source from whence they came, and see whether, by the 
same means which gave 3 — 7, we might have got 10—14 in its 

Suppose a problem, wrongly expressed or understood, gives 
2x — j: ^ 50 — 70 (as in page 15). If we take the correct equation 
50 — x = 2(35 — x), we may add any quantity to both sides. Say 
vre add a to both sides, which gives 

(50 -fa) — a; = (70 +a) - 2x 
the solution of which is 

2ar-a; = (70+a)-(50 + a) = 20 

If we had taken the incorrect form 50+j^=2(35H-j?), and added 
a to both sides, we should, by what we suppose to be strict reasoning, 
till the result undeceives us, arrive at the conclusion x = (50-|-a) 
— (70 +o). And this is what we want to ascertain, namely, that 
just as in the rational form of an equation, we may by previous legiti- 
mate alterations obtain the answer in the form 70 — 50, 71 — 51, 
72 — 52, &c. : so, in the irrational form, we may, by the same sort 
of process, obtain 51 — 71, 52 — 72, &c., instead of 50 — 70. And 
we will not say, that 51 — 71 is equal to 50 — 70, because our 

* The sign of subtraction is a bar written before the quantity to be 
subtracted. The present is not a sign of subtraction any more than the 
bar between the numerator and denominator of a fraction. 


notion of equal (as far as we haye yet gone) applies only to mag- 
nitode, number, bulk, &c. &c. ; but because any equation which 
gives 50 — 70, might also have been made to give 51—71, &c., we 
will call these equivalent to 50 — 70, meaning by the word eguivaUni, 
that the first may stand in the place of the second, or be substituted 
for it, without producing any error when we come to correct the 
result, or any set of operations for which correct substitutes cannot 
be found. Thus, then, we say, that 0—1,1 — 2, 2 — 3, &c., are all 

equivalent, and may be represented by T ; a — (a -f IjJ and (a +z) — . 
(a 4- c 4- ?) are equivalent, and are represented by c ; the rule always 
being ; — invert the subtraction, and place a bar over the result. 
Thus, if we obtain such an equation as 

x + a-\-b = 

which is the form most obviously impossible of all, we shall, by 
rules (mly, obtain the expression 

a; = — (a + 6) 

which we signify by (a + b) 

In considering equations of the first degree, we may confine our- 
selves to the rational form x — a =z 0, and the irrational form j-f-a = o. 
For to these all others can be reduced. For instance (page 5) 
equation 4 is reduced to 

X = 


and the first equation of Prob. I., in page 15, is reduced to 

a: + 20 = 

We now find equivalent forms for 

a + b a-'b axb ■=■ 


The first will arise from such an equation as the following : 
a? 4- (jp 4- a) — p +(.q + b) ^ q 

from which, if we solve it before observing that it is impossible, we 

X =;> — (;? + a) 4- ? — (? + &) = a + b 

But the same rules of solution will also give 

a: = p+q — (p -j-q + a+b) = {a + b) 


In which the bar over a + i signifies that we have attempted to 
subtract from a quantity another which is greater than itself by 
a + b. Or we have 

a -f- 6 is equivalent to (a + b) 
Similarly the equation 

a:+(p + a)— i>-fy == iq + i) 
gives X = jp— (p-fa)— (5— (y+J)) = a— J 

following rules only. But this equation is not always impossible, 
for it is equivalent to 

or a; + a = 6 which gives x ^ b — a 

Therefore, a — b is correctly 6— a when b is greater than a; when 
b is less than a, it is (a— 6). We may also give the equivalent form 
b-^a, which follows from the attempt to solve correctly. 

x+(p + a)'-p = b 

in the form X = ft + (p— (p -f a)) 

We have as yet obtained no such expressions as 

ab or -=r 

but we shall now shew that these arise from inattention to the equa- 
tion, not to the problem. That is, we shall deduce ab and ab, 

-=r- and T, or similar forms, from the same equation, whether that 

equation be true or false. 

If /) be greater than q, and c greater than d, the multiplication of 
the rational expressions/) — q and c — <f, and the attempt at multipli- 
cation of the irrational expressions g-^p and d-^c, give the same 
result, as follows : 

p-9 g-p 

pc—qc qd^pd 

pd-^-qd qc-^pc 

Subtract pc^qc^pd-i-qd qd^-pd—qc-hpc 


whidi are the tam« in everjr thing but the order of their terms. 
Consequently^ the equation 

x + qc+pd ttx pc-^-qd 
might, by inattention, be thus solved 

X s= pc'^-qd^qc'-pd =» (j— p)(rf— c) 
where the latter aAiould have been 

X^ (p-qXc-^d) 

whence, if p — q be a and c — d be 6, the expression Zb may be 
obtained instead of the quantity a b. 

We have already seen, in page 19, how it may happen that -^ is 


introduced in the place of -. Let p be greater than g, and c greater 

than d; then 

cx + J =s dx +p 

correctly solved, gives 

incorrectly solved, gives 

{d'^c)x ^q-p ^^jETc 

whence, if p — y be a, and d — c be i, -=• is equivalent to ^ 

b b 

We have thus determined either equivalent forms or corrections, 
in all the cases which arise from the equation, considered without 
reference to the problem from which it was derived. We shall 
now consider both, taking as a basis what we have ascertained by 
examples, namely, that the misconception which produces a instead 
of a, also causes us to add where we should subtract, and subtract 
where we should add, in forming our equations. 

Suppose we have obtained a as the result of an equation, and 
that the next step (if all before had been correct) would have been to 
add a to c, er to find c-\-a. 

Correction and Process, The true result is a, but the same mis- 
conception by which a was produced, has made us suppose this 
quantity was to be added where it was in reality to be subtracted, 
and vice versd (pages 15-18); therefore the next step corrected is 
e — a or e^^^a is equivalent to C'-^a, 



Trial of the incorrect Process, a is the representative of x — (M'\'d), 
and c -ho (applying rules only) is c -^ z — (z -f a), or c -{-z — z — a, 
or c-^a, which has been shewn to be correct. 

Similarly, where c — a occurs, we know that the true answer is a^ 
but that the misconception which produced a leads. us to suppose if: 
should be subtracted where it should be added ; therefore c + a is 
the true result. The incorrect process gives c — (« — {z-{-a)\ or 
c — r 4- (^ + a), or c — a: + ;8r + a, or c + a. We have, then, these 
two rules : 

c + a is equivalent to c — a 

c^a c +a 

By similar reasoning (p — a) x ig — b) is equivalent to (j> -\- a) 
(7 4- 6). The incorrect process gives pg — ga — p b + ab. If we con- 
sider a and 6 as resulting from the misunderstanding of a problem^we 
must take some such problem as a guide. Let it be the following : 

A and B have respectively 4 and 5 pounds. One of them loses a 
bet to the other, after which the product of the number of pounds 
belonging to each is 18. Which lost, and how much? 

If we suppose A lost a; pounds, the equation evidently is 

(4-.a:)(6+ar) = 18 
But if we suppose A gained .r pounds, the equation will be 

(4+a:)(5-a:) = 18 

First Alternative. 
4 — X 

Second Alternative. 


5 — X 

20 + 5J: 

4x + XX 

Subtract 20 -}- ^ — ^^ 

20 H- j: — XX = t8 

XX — X ^2 

20 — 5x 


Add 20 — X — XX 
20 — X — XX = 18 
XX -\- X ^ 2 

From this we see, that in this instance (and trial proves the same 
in others of the same kind) the correct equations belonging to different 
alternatives have different signs in the terms which are multiplied by 
,r only once (or contain the first power of a:), or (page 1) are of the 
first degree with respect to x ; but that the terms which contain xx, or 


are of the second degree with respect lo x, have not different signs. 
And the same may be pro?ed of terms containing the product of two 
unknown quantities x and y^ which are not the same ; for instance, 
in the equations 

(4-a:)(5+y) = 18 
^ (4 + a:)(5-3f) = 18 

If we take a term such as abc, vre shall find that, on cor- 
rection, the sign preceding it must be changed ; that in such a term 
^ abcdf the preceding sign is not changed: the rule being, 
c^e the sign where there is an odd number of factors to be 
corrected. It is indifferent whether the fiu^tors to be corrected be in 

^ nnmerator or denominator:' if, for example, we take -=-, we 


^h by deducing - instead of x from the equation which gave r, 


1 ^ 

^ — instead of '=', as follows. Suppose the equation which 
c c 

gives c to be 

2x -^{6 + z) =^ x + z : 

^is> following only rules, may be finally represented either by 

2x^X =i Z-^(c +z) or X = C 
^^7 x — lx or (1— 2)a; = c-f-^:— 2r = c 

that is, (T)a; = c {rr)cx \^\ 

^cing the consequences of this result, we find that the uncorrected 

— ^ /» ^ 

^ i. is equivalent to -=•. Hence the term -=• is equivalent to 

c c c 

af V— or .which contains all the uncorrected factors in the 

c c 


If we now resume (p — a) (^ — ^)> of which the continuation of 
the incorrect process is /? j* — p 6 — 9 a + a ft, we must conclude, that, 
'n the terms of the first degree with respect to a and 6, the signs 
must be altered ; but that in the term a h the sign preceding it must 
not be altered ; so that the correction gives 


But this 18 the lame as would have arisen if the process had been 
corrected one step earlier, as in page 50 ; that is, if (p +a) (9 -f- b) 
had been written at once for (p— . a) (9 — 6 ) • 

It is not necessary to go through all the individual cases that 
might arise. We have found in all that the common rules of algebra 
may be applied without error to the expressions of impossible sub- 
tractions ; that is to say, the correction may be deferred as long as we 
please without introducing error, provided that, when at last the cor* 
rection is made, the following rule be observed : — In correcting any 
term, change the symbols of impossible subtraction by substituting 
the absolute number resulting from the real subtraction [thus, put 
3 for 3> or (ar+3) — z for z — («+3)]; change the sign preceding all 
such terms as require an odd number of such corrections ; keep the 
sign of such as require an even number. Repeat this as ofien as 
may be necessary. If the final result be then rational, the problem 
has been rightly understood, and the mistakes have arisen from 
inattention to the processes which come between the statement and 
the result ; but if the result after correction be still an impossible 
subtraction, then the problem has been misunderstood if there be 
alternatives in it, or is itself a wrong alternative of some more general 

As an example, let the final result of a set of operations be 

abc-{-dd . ab 
ac — a cd 

Let it be then discovered that a, b, and e, are not rational expres- 
sions, and let them be found to be made, a by attempting to subtract 
a number from 1 less than itself, b by attempting to subtract a 
number from 3 less than itself, and e by attempting to subtract a 
number from 5 less than itself. That is, let a, 6, and e, be represented 
by I) 3) and 5. Let ^ and / which are rational, be 2 and 6. Then, 
the preceding expression, before correction, is 

1X3X6+2X2 , 1X3 _ g^g 
1x6-2 6X2 

the first stage in the correction of which is 

18+4 , 3 

— 6 — 2 • 12 

+ :^ +30 


buly —6— 2 (page 47) is signified by 8; and the necessary correc- 
tjoo; according to the preceding i^le, gi?es 

^ + 30-^ or 27J 

which is the result which would have been obtained, had each 
correction been made in its proper place. 

But the preceding form a is not made use of by algebraical 
writers, and is introduced here not to remain permanently, but to 
avoid using the sign of subtraction, to appearance at least, in two 
diflfeffent senses. If we follow rules, without observing where they 
lead us, we should obtain processes of tlie following sort : 

3-8 = 3-(3+6) = 3-3-6 = 0-6 

and, as in + ^9 it is not necessary to the meaning to retain 0, 
we might, by imitation, write — 5. This is what we have hitherto 
written 5. And we shall find, in all the intelligible properties of the 
sign — , a close similarity between —5 (properly placed in an ex- 
pression) and the legitimate rules by which it is treated, and 5? an 
uncorrected misconception, with the rules for obtaining a correct result. 

In &ct, if we apply rules to + ^od — as they are applied when 
the quantities concerned are rational, we find no distinction between 
a and —a, though we have made one until we could establish the 
points of similarity. For instance, b — a when corrected is b-^a. 
And b — (z — a), when z is greater than a, gives b — z-\-a; so that a, 
rationally used with two negative signs before it, gives '\-a in the 
result. Apply this rule to b — ( — a) and it gives b-^a; so that 
b'^ais corrected by the application of no other rule than considering 
a as —a, and applying the rules which, in the Introduction, are 
shewn to apply in possible subtractions. The same will be found in 
every otlier case yet stated . 

In further illustration, \ei ax=zb — c, let c=rf — x, let ar=^ — v, 
and let v^t — z. We have, then, 

a = b^(d—x) = fe--(rf— (y— v)) 

= ^-{^-(y-(^-^))| (A) 

= 6-|rf-Cy-^ + ;2)} 

— b^ |rf—y 4-^—^1 = 6— rf-fj^ — ^ + z 


On looking at this result, we see that z is preceded by + ; in the 
expression (A) it is four times under the negative sign, t is preceded 
by — y and is under three negative signs, y is preceded by +> having 
been under two negative signs. Consequently, those terms are nega- 
tive which are under an odd number of negative signs ; and positive, 
which are under an even number. 

Now, let us suppose that we have obtained from an equation an 
incorrect value — a, denoted by a, as before. Let the result of 
some succeeding process lead to 0— a, which we will denote by 

a. A third process leads to 0— a, which we denote by a, and so on. 
At every step, therefore, a new misconception has been introduced : 
but, as we shall proceed to shew, the repetition of the error any even 
number of times does not shew itself in the result. The first irra- 
tional answer must arise from the attempt to determine x from an 
equation, which reduces itself to jr + a = 0. Let a new process be 
then entered upon to determine y, and let it lead to ^ -f*^ =^0* The 
equations which we should have got by proceeding correctly are 
j:— a = and y— x =: 0, which lead to ^ = a. But this is the same 
as we should get from the incorrect equations jr-|-a = 0, i^+x^O, 
by rules only. For, subtracting the first from the second, we have 
^— a = 0, or ^ = a. Again, proceeding to find 2r, suppose we get 
z -|-^ :=0. From the correct equations in the first column 

x^a = X'\-a = 

• y— a; = y+^ = 

z^y = ^+y = ^ 

we get z^^-a; not so now from the incorrect equations (observe that 
their number is odd). For, from rules only, by adding the first to 
the third we get z-^-x+y-^a^^Oy from which, subtracting the second, 
we find 2r+a=0. And thus we might proceed with more equations. 
Now, observe that the first equation gives x^O — a or a, the second 

^ = — X or — a or a, the third ja: = — y or — a or a, and 
so on. Hence, from the preceding, any even number of errors of 
this kind corrects itself, any odd number requires correction. 
We now return to the expression 

which is rational when t is greater than z, y greater than t — z, &c. 



We hare already applied ^aeie eperatiom to inational cases to see 
what would come of them ; we now apply direct reatonmg knowingly 
to an irrational case, with the same intention. Not that any new step 
is now made ; for in applying operations, we always tacitly employ 
the reasoning by which the rule of operation was demonstrated. It 
is only allowable to write an instance of a — (6— c)^a— 6 + c in 
cases where we can return to the demonstration (see the Introduction), 
and make that demonstration apply to the particular case in hand. 
We place the example of too general reasoning in brackets. 

[Since the preceding equation is generally true, it is true when 
f =0y jrs=0, dssiOy and 6=0. But the preceding equation, omitting 
the term wherever it occurs, because neither increases nor dimi- 
nishes the Talue of an expression, becomes 


therefore z is not altered by being preceded hy four negative signs, 
and the same may be proved of any even number ] 

But the preceding reasoning is obviously incorrect, if seriously 
meant as a proof, and not as an experiment; for the equation 
a — (6 — c)s=fl— 6-|-c is tacitly applied to the case where 6 = 0. 
We will go over the general proof, which ought to apply to the 
particular case, putting the two together. 

General Pboof. 

(True when b is greater than c.) 

^— c is to be taken from a. 
If from a we take away 6, giving 
a—- 6, we have too little, be- 
cause only a part of b should 
have been taken away, namely, 
as much as is left after it has 
been lessened by c. Therefore, 
c too much has been taken 
away, and the true result is 

Particular Application. 

(Not admissible as reasoning in 
any case.) 

0— c is to be taken from a. 
If from a we take away 0, giving 
a — 0, we have too little, be- 
cause only a part of should 
have been taken away, namely, 
as much as is left afler it has 
been lessened by c. Therefore, 
c too much lias been taken away, 
and the result is a — O + c (that 
is, a-\-c). 

The application of the reasoning needs no comment. Nothing 
has been taken away from a quantity, and yet it has been too much 
diminished — only a part of nothing should have been taken away, 



and nothing should have been previously lessened. We will now 
subjoin the version we give of the equation a— ( — c) = a-{'C9 derived 
from the preceding part of this chapter. 

We have examined particular cases, and have always found that 
0— c is the result of a mistake of one particular sort, namely, that 
the quantity represented by c is of the opposite nature to what we 
supposed it to be. And we have also found by examination, that as 
to addition and subtraction, all the operations have been inverted ; 
we have added where we should have subtracted, and subtracted 
where we should have added ; to which additional misapprehension 
we should have remained subject, had we not come to the irrational 
expression — c. Tlierefore, instead of a — (0 — c), we remember 
that — c should be c — or c, and in a — (0 — c) the first — is also 
the result of misconception, and should be +. Therefore a — (— c) 
is correctly written (not is equal to) a-^-c. Thirdly, we have found 
by examination, that if we continue any series of processes with the 
effects of the misconception uncorrected, we shall not introduce any 
errors but those which may be corrected, at any period we please, by 
the use of one simple rule, namely, apply no other rules except those 
which have been deduced in the case of rational expressions. 

It was found out that the rules of algebra might be applied with- 
out error to symbols of impossible subtraction, before the cause* of 
so singular a circumstance was satisfactorily explained. The conse- 
quence was, that many such reasonings as those in page 55 were 
universally received, and a language adopted in consequence which, 
as long as words have their usual meaning, is absurd. 

But, at the same time, every one must see that words are them- 

* I am far from asserting that the view I have taken will be easy, or 
that it is the only one which might have been given as satisfactory to 
those who can understand it. But I think that the matter of it, inde- 
pendent of the method of stating it, must he considered at least of incon- 
trovertible logical soundness. I am aware that many will think the 
connexion of —(—a) and + a to he more necessary than I have attempted 
to shew it to be ; and in the higher view of the subject, which no 
beginner could understand, it may be so ; but I think the exclusion of 
false analogies (to which the student is very subject in this part of the 
science) of more importance than the establishment of true ones. It will 
be easier for the pupil hereafter to acquire new ideas of relation, than to 
get rid of any he now acquires. 


selves sjmbols of oar own makingy over which there is unbounded 
oootrol consittently with reaton, provided onlj that what we mean by 
every word be distinctly known, so that we shall not draw conclusions 
from one meaning of a word, and then apply those conclusions to 
jDther meanings.* 

We have made some additions to common arithmetic^ and have 
found uses ibr symbols which were never contemplated in that science. 
It is suflBcient that we have demonstrated the uses of these symbols ; 
it now remains to find words by which to express the operations we 
diall emplof. 

There are two vniys of proceeding. 

1. Whenever we want to signify an operation which is not wholly 
arithmetical^ we may invent a new term. This would load the 
science with difficult words, which, after all, would only have the 
eflfect of banishing tlie arithmetical words, and substituting others in 
their places ; for we cannot know whether we are proceeding with or 
without the misconceptions explained in this chapter, until the end of 
the process. We should therefore be obliged always to use the newly 
invented terms, to the exclusion of the others. 

2. We may alter the meaning of th^ words already in use by 
extending them; that is, allowing them to mean what they already 
stand for, and more. In common language we are already well used 
to something like this, and, whenever we want a word to express one 
object, are in the habit of using one which belongs to another object 
having some resemblance to, or connexion with, the one we are to 
name. By this means, there is a word in the English language which 
stands for a receptacle, a seat, a small room, a small house, a plant , 
and a blow. But this is forming names by resemblance only, not by 
extension. We might instance the latter in the words arm^ mark, 
plain, &c. ; but it is to the sciences only that we must look for examples 

* As an illuBtration of our meaning : The word square, in algebra 
(as we shall see when we come to it), is made to mean a number multiplied 
by itself; in geometry, it means a well-known species of four-cornered 
figure. How this word came to have two meanings so different, the 
student will see if ever he studies the history of algebra, and will guess 
when he comes to apply algebra to geometry. But in the mean time, 
nothing that is proved of the square in algebra is to be therefore taken 
for granted of the square in geometry. The same word with two different 
meanings is the same as two different words. 


of intentional extension, correctly managed. We take one out of the ^ 
numerous instances vvhich natural history affords. It is found con- 
yenieut to divide all animals into classes, comprising in each class 
those vvhich have certain common arrangements of teeth or other 
members. One particular class of animals contains the cat, which is 
the best known of its class. Instead of inventing a new word to 
signify this class, containing the cat, lion, tiger, panther, &c., they are' 
all called ca^<, and each has a particular additional name to distinguish 
it. Thus, the common cat is felU catus, the lion is felts leo, the tiger 
is felis tigris, the panther is felis parduSy he. Observe that what is here 
done is, to make the word felis (Latin for cat) mean less than in 
common discourse, and imply not the common animal, or any animal 
which agrees in all respects with it, but any animal which agrees with 
it only in those arrangements which are considered the distinguishing 
marks of the class. Consequently, by limiting the ideas which the 
word is meant to imply, the number of objects which come under it- 
is extended. And no mistake could arise by this means when one 
zoologist speaks or writes to another; though a third person, not- 
acquainted with their meaning, might think they believed that a cat- 
could run off with and devour a man. 

Similarly, in algebra, we have terms which are well understood in » 
arithmetic, and processes which we find carry us beyond the object of 
arithmetic, which is absolute number. But, both in the processes of: 
algebra, and in those of arithmetic, there are resemblances which wiU« 
make it convenient to classify together those which follow the same 
rules; and, in giving the names to classes, we shall, as previously 
described, limit the definition of the arithmetical terms so as to name 
the whole class by the arithmetical name. And when we speak of 
the process of arithmetic to which we have been accustomed, we ^hall 
prefix the word arithmeiical. Thus, by arithmetical addition, we 
mean the simple increase of one absolute number by another. But 
this will be, as we shall see, only one case of algebraical addition, or, 
as we shall call it, addition. Andf once for all, observe that in future 
every term has its extended or general algebraic meaning, except when 
the word arithmetical is prefixed. 

We shall now proceed to the limitations of the notions contained 
in the terms, or the extensions of the cases which come under them, 
whichever it may be called. 

1. Quantity is applied in arithmetic to any number or fraction. 


In algebra, it is any symbol* which results from the rules of calcula- 
tion. We have the first effects of this in the following proposition. 

Quantities (as far as we have yet gone) are either positive or nega' 
iive. Arithmetical quantities are all positive. 

The latter part requires this remark, that + 6 is more than the 
symbol of 6 ; it is the symbol of one particular way of obtaining 6, 
namely, + 6, or adding 6 where there was nothing. But 3 -|- 3 is 
not identical with +6 in the operations indicated, but only in the 
result obtained; just as (a-\'x)(a — x) and a a — xx, which are not 
the same operations, always gire the same results. 

Positive and negative quantities are of diametrically opposite signi- 
Rations; if -{-a be a gain qf£a, — a is a loss of the same ; if-^-abe 
a loss of£a, — a is a gain of the same ; if -\- a be length measured 
ftorthward, — a is the same length measured southward ; and so on. 

This distinction has been sufficiently dwelt upon in a different 
form. If we suppose a gain, and the answer to our supposition 
(derived from a rational equation) is that the gain is a, then we were 
right in our supposition; but if the answer be — a, then we were 
wrong in tlie supposition, and we ought to have supposed a loss of 
£a. The extension here made consists in making the symbol of an 
error stand for the correction of that error. On this we must remark, 
that the extension avoids confusion only because there is but one way 
of making the wrong supposition right. If — a might indicate mis- 
takes of more than one kind, it would be wrong to let — a stand for 
the correction of one species, when the problem might oblige us to 
choose another. 

The application of algebra to geometry immediately suggests this 
sort of extension. Let A be a point, in a straight line indefinitely 
extended both ways ; measure A 6 to the right of A, and from B 

C A C B 

* A symbol is any thing which can be placed before the mind as a 
representation of any other thing. The term quantity is not applied 
according to its original meaning, even in arithmetic. This 2 is not a 
qtumtity (unless it be of printer's ink), hut a symbol, which denotes that a 
certain quantity, represented by 1, has been taken twice. Thus, 3 — 5 
is a rational symbol, but not a S3rmbol of arithmetical quantity, because 
it proposes operations which contradict the arithmetical meaning of 3, — , 
and 5* 


mearnre B€ to tbe left of B. If A B be greftter tkao B C, the proper 
description of tbe position of C is ^ at a distance AB — BC on the 
rigbt of A/' But if BC be greater than AB, and we still say the 
same of C, we are warned of some mistake by the impossible subtrao- 
tion, and we see immediately that the proper designation of C's 
position is '< at a distaooe B€ — AB on the left oift'lT^ 

In conversation and writing, figurative extensions are so common 
that they have sometimes been appealed to as a justification of tbe 
processes we have been investigating. Id strict propriety, there is a 
repetition of ideas in the phrase '' to gain a gain/' and contradiction 
in " to lose a gain,*' in which the word '' to lose *' is used in the 
common sense of ^ not to get.'' Bat the most obvious analogy is in 
the words ^' to gain a loss,*' which is an ironies^ term applied to one 
who loses where he thought he should gain. And when we say, 
<' darkness went away," instead of ^ light came," we make a mistake* 
in a matter of fact which bears a close analogy to that which we have 

In the application of matliematics to physics, we are liable to the 
error of imagining that a phenomenon may arise from matter being 
added to other matter, when in fact it arises from matter being taken 
away, and vice versa. This has happened in several instances, of 
which we will cite two of tlie most remarkable* 

1. If glass be rubbed against leatlier in dry weather, both sub- 
stances acquire power to attitict small pieces of matter, which power 
is called electrical attraction, or electricity. It was at first supposed 
that friction made the leather communicate some fluid to the glass, so 
that the glass had more than its natural share, and the leather less. 

* These phrases are not introduced as illustrations or confirmations, 
hut precisely the reverse. They are an impediment, because the student 
may hy them be led to imagine iMt he sees reason in the use of Ae 
negative sign, independently of the proof given that it is merely a con- 
venient method in correction of unavoidable misconceptions. To warn 
him that he has not (from this work at least) any evidence of the pro« 
priety of negative quantities, other than that which he gets from observing 
what will come of using them, is tbe object of this note. If a tteaning 
is to be given to a term, which in its original use it will not hear, the 
more repugnant the phrases employed are to common ideas, the better in 
one respect ; because the less the student can find any thing like them i& 
his mother tongue, the more likely will he be to fasten upon them the 
explanation which they are meant to bear, and no other. 


Consec^uently, the glass was said to be positively^ and the leather 
twegatwely^ electrified ; and the phenomena of the latter were supposed 
to be caused by the subtraction of something from it. But succeeding 
experiments shewed it to be much more likely that the friction of the 
t.wo substances separated a compound fluid, substance, agent, or 
^v?hatever it may be called, into two distinct component parts, having 
this quality, that when united in their natural proportions they attract 
nothing, but that either, when separated from the other, shews it by 
attraction. These were called the vitrcova and rninout electricities, 
because friction gave the first to gla»$y and the second to renn (as was 
Ibund). But many still retained the old names of positive and 
negative electricity; and this produced no inconvenience, because 
what we may call the mathematical phenomena of electricity remained 
the same on both theories, it being exactly the same in calculating 
effects, whether we suppose the cause of the effect to be removed, or a 
sufiicient quantity of something which destroys the efiect to be added. 
2. Id burning a candle in a close vessel of air, it was observed 
that the air soon became incapable of allowing the process of burning 
to continue, and that the air produced was not fit to breathe. That 
an alteration had taken place was then certain ; and it was supposed 
that the burning candle gave out a fluid which mixed with the air. 
This fluid was called phlogiston (thing which makes flame). There- 
lore the effect of burning on air was supposed to be the addition of 
phlogiston. But it was afterwards discovered that in fact something 
is taken from the air when a body bums, which something is oxygen, 
found by other means to be a part of the mixture called air. Hence, 
the efiect of burning is the subtraction of oxygen. And if any chemi- 
cal calculation made on the theory of phlogiston were required to be 
aet right, it might be done on the supposition that -)- a of phlogiston 
is — a of oxygen, with the rules laid down in this chapter. 

2. AdtUtion and Subtraction, The first term means the forming 
two expressions into one, retaining the proper signs; the second, 
forming two expressions into one by altering the sign of the one 
which is said to be subtracted. The following examples will shew 
that arithmetical addition and subtraction are particular cases of the 

3-|-(5-.2) or 3 + (0 + 6-2) = 3+5-2 
8-(6-2) or 8-(0 +6-2) = 8-6 + 2 


But addition and subtraction include such as the following: 
— 3 + (-5) is — 3 — 5 or —8 
-3-(-5) is -3+6 or +2 

3. Equal. Any two algebraical expressions, of which the on^ 
may be substituted for the other without error, are called equals 
and := is written between them. This, as before, includes arith — 
metical equality; for 5 + 3, which is arithmetically equal to 8, may 
be substituted for 8. But the algebraical term also applies to 3 — 7 
and 10 — 14, page 46, to a+( — b) and a — b, and so on ; and the 
term will afterwards apply to still wider cases. For instance, we 
shall come to a species of misconception, which will give 

1 — 1 +1 — 1 +1 — 1 + &c. continued for ever. 

where the true result is 4> This will be thus represented : 

J = 1 — 1+1 — 1+ &c. ad infinitum, 

4. Greater and less; increase and decrease. The extension of 
the words addition and subtraction requires also the extension of 
these. The symbols of arithmetic are, 

1 2 3 4 6, &c. 

and intermediate fractions; and the greater of any two is that which 
comes on the right. The numerical symbols of algebra are, 

I Arithmetical. 
—4 —3 —2—1 I +1 +2 +3 +4 «tc. 

in which the addition proceeds throughout as in the arithmetical 
series by the (here algebraical) addition of +1.. For 

-4 + 1 = -3 -1+1 = 

-3+1 = -2 + 1 = +1 

-2 + 1 = -1 +1 + 1 = +2 

Let the definition of greater and leu remain the same on both 
sides of the line; namely, that of any two quantities, the one which 
falls on the right is the ^greater. Thus —1 is called greater than 
— 2, +2 is greater than -—I, and so on. 

Hence, with the extended meaning of the words, we have the 
following proposition:* 

♦ This is the proposition which has startled so many beginners, and 
not without reason, considering that they have frequently been intrc 
duced to it without any warning that greater and less have not their 


All positive quantities are greater than nothing; all negative 
quantities are less than nothing. Of two positive quantities, that is the 
greater which is arithmetically the greater; of two negative quantitia, 
that is the greater which is arithmeticalli/ the less. 

The extended terms increase and decrease will follow greater and 
less. Quantity is increased when it is made greater, and decreased 
when it i» made less. But the word smaller is always alfowed to 
retain its arithmetical meaning, without extension. 

N.B. We have now separated increase from addition, 8cc. 

Addition of . . {P^S;« } quantity cause, ^^^^} 

Subtraction of {^^ quantity causes {f---} 

The following propositions are also true : 

The greater the quantity added, the greater is the result. 

For example : 

— 7 is greater than — 10 

3+(-7) 3 + (-10) 

for we see that — 4 — 7 

Similarly, the less the quantity subtracted, the greater is the result. 
From 9 subtract —8, the result is +11; subtract less than —8, 
say —12, the result is +15, greater than +11. From — 4 subtract 
7, the result is -^ 11 ; subtract less than 7, say — 3, and the result is 
— 4 — ( — 3), or — 1, greater than — 11. And it will be found that 
all such theorems relative to addition or subtraction as are true of 
arithmetical, will also be true of algebraical, quantities ; which is the 
particular advantage of our new definition. For instance, remark the 
following : 

If a be greater than b, a — b is positive ; if a be less than b, a — b 
is negative. Thus 

_3-(-4)=+l _3-(-2) = -l 

The signs of greater and less are > and < . Thus, 
For a is greater than b, write a > 6. 
For a is less than b, write a < 6, 
The angle is turned towards the greater quantity. In the sign of 
equality ^ there is no angle towards either quantity. 

arithmetical meaning, except when arithmetical quantities are mentioned. 
Those who object to it in its present shape, will of course object to the 
zoological supposition in page 58. 


5. Mtdiiplkaiion and DivUion, These rules are, so ftir as the nu- 
merical quantities are concerned, the same as in arithmetic. The rule 
of signs, as we ha?e seen, is, like signs produce +, unlike signs — . 

+ a6 is both +ax +6 and — a X —6 

—aft is both —ax +6 and +ax —6 

+ « is both ±? and =^ 

— Y IS both -7-7 and -^- 

-f-O — D 

The terms greater and less cannot always be applied to pro- 
ducts as in arithmetic. For instance, 3>2, 5>4, and therefore, 
3x5>2x4; but from 3 >— 2, —3> —4, it does not follow 
that 3x— 3> — 2X— 4, or — 9>8, but the contrary — 9<8. 
But it is seldom necessary to deduce the algebraical magnitude of a 
product from that of its factors ; we therefore leave to the student the 
collection of the different cases. 

6. Proportion, Four quantities are said to be proportional when 
the first divided by the second is equsd to the third divided by the 
fourth. This definition is the same in words as the definition of 
proportion in arithmetic, but the words quantity, cUvided by,vnd equals 
have their extended signification. The words greater and less cannot 

always be applied as in arithmetic. Thus — - being — - we have 

3: — 4:: — 6:8, where 3 is greater than —4, but — 6 is less 
than 8. 

We shall now apply our definitions to a problem, and shall choose 
the cases already noticed in page 45, as being two different problems 
when only equations of the first degree are to be used. 

A has £60, and is to receive the absolute balance that appears in 
B*s books, whether for or against B ; but C, who has £ 200, is to take 
B*s property and pay his debts. After doing this it is found that C's 
property is 3 times that of A. What is the absolute balance for or 
against B ? 

Let X be this balance, positive or negative according as it is for or 
against B ; then A has 60 + x, the positive sign being used* when x 

* For A's property is to be increased on either supposition ; hence, if 
the balance be + 3, he must have 60 + ( + 3) ; if it be —3, he must have 



i« positive, the negative when x is negative. But C ha» 200 + x; 

3(60±x) = 200 + ar 

or ± 3a: = 20 + a: 

Tfais contains two equations, one for the positive, one for the negative 

s^gD. But irom it follows that 

±3arx±3a: = (20 + x)(20 + x) 

in which the first side is -f Qjtjt in both cases, for — 3j:X — 3x 
*Dci -f 3-r X + 3 jr, are the same, namely, + 9 jtjt. Therefore, 

+9a;ar = 400 + 40a; + ara; 
or 8ara;-40jr~400 = 

(•^)8 xx-bx-^bO = 

M^hen we come to the solution of equations of the second degree, 
""^ shall find that this equation can be true only for two values of x ; 
either x = 10, or j: = — 5. That is, the balance is either £lO for B, 
®' •=€ 5 against him ; which are the solutions already found in page 45. 
iTiat either 10 or — 5 will satisfy the preceding, may be shewn as 
follovrs : 

ar = 10 a: = —5 

XX = 100 XX = 25 

—5a: = —60 —5a: = +25 

-50 = -50 -50 = -50 

a:a:— 5a:— 50 = a:a:— 5a:— 50 = 

^ince the extensions of algebra have been so laid down that the 

'"^^^a for managing algebraical quantities when they are not arith- 

'^^tical are the same as those which must be employed when they 

^^ arithmetical, it follows that the arithmetical case of a problem may 

*^ t^en as a guide ; for, to say that certain operations follow the same 

^^^ as in arithmetic, or that we must proceed as if the operation was 

^ntbmetical, is only the same thing in different words. 

Up to page 56 we have considered the symbols of algebra, which 
are not arithmetical, as results of misconception, and have called the 
rules by which they are treated corrections. In page 59, &c., by pro- 
perly laying down definitions, these same symbols are recognised and 
expected, so that the term erroneous no longer applies to them. The 
student should not immediately give over the first method ofcon- 

G 2 


tidering them, but should frequently, while employed upon the rules 
(pages 49y &c.), make himself sure that he understands the con- 
nexion of the preceding method with those rules ; and in future we 
may accordingly employ both methods, it being always understood 
that when the first is used, the extensions required by the second are 
dropped for the moment. 

The following examples of the use of the rules are added for 
practice. Such symbols as a and — a are used indiscriminately, it 
being remembered that they mean the same thing in practice, but are 
referred to two different methods of considering the subject in theory. 

8x4-?-3=10| 8x4-j.3 = -10| or lo| 


= 2 (- 

-6) + (- 

-8) + (- 




ab -\- cd 

cd — ab 



— d 




f X C 

a + 6 









-b) + b{ 




abc = 

abc = 

abc = 








Let there be such an equation as 

x+y = 12 
resulting from a problem which involves two unknown quantities, 
X and y. Such a .one is the following : 

1 1 1 

A C B 

Problem. A is a given point in a straight line, and B and C 
are two other points. From A to half-way between B and C it is six 
feet. What are the distances of B and C from A ? 

Let us take as our principal case that in which B and C are both 
on the same side of A. Let AB = j; feet, AC = y feet; then 
BC = X — y, and from B to half-way between B and C the distance 
is i(x — y) ; whence from A to the same middle point the distance is 

X — Q ( '*' — yj which is therefore = 6 

(x)2 2a;-(a:-3/) = 12 or x-hy =^ 12 

Here the problem is what is called indeterminate, that is, ad- 
mitting of an infinite number of solutions. All that is laid down 
relative to x and y is found to do no more than require that their sum 
shall be 12, which can be satisfied in an infinite number of ways; for 
all the following cases are solutions, and others might be made at 

ar= 1 

y= 11 

_ 1 

^ "" 2 

, = "! 

X rrz 2 

y = lO 


y= »| 

ar = 3 

y= 9 

a; = 3? 

y= 4 





X = 15 y = 


a: = 16 y = 


X = 14 y = 


&c. &c. 


Again, when either j: or y is taken negatively, we have solutions 
such as the following : 

a: = -1 y = +13 
a: = — 2 y = +14 

a:= «ll y= +13i 

&c. &c. 

the first column corresponding, with the explanations before given at 
pages 14-19, to the case in which B only is at the left of A; and the 
second to the case in which C only lies to the left of A. Thus, if B 
be 1 foot to the left of A, and C 13 feet to the right, the middle point 
between C and B is at 6 feet distance from A to the right. Observe 
that we cannot in this equation of the first degree, j;+y = 12, 
include the cases in which the middle point falls to the left of A ; 
for since this quantity 6, given in the problem, has been treated 
throughout as an arithmetical (or algebraical positive, see page 59) 
quantity, the equation formed from it cannot include those cases of 
the problem in which the corresponding line is so measured that its 
symbol ought to be negative.* 

From the preceding and similar cases, we deduce the following 
principles : 

1 . One equation between two unknown quantities admits of an 
infinite number of solutions ; either of the unknown quantities may 
be what we please, and the equation can be satisfied by giving a 
proper value to the other. 

2. A problem which gives rise to such an equation is indetermi- 
nate, or admitting of an indefinite number of solutions. 

Let us now suppose two equations, eacl) containing the same two 
unknown quantities. For instance, 

x+y = 12 3a:-2y = 31 

* On reading the problem again, therefore, we perceive either that 
we have not Bu£ficiently defined it, by omitting to state whether the six 
feet is to the right or the left of A, or else that there are two problems 
involved in it, as in page 45, or that these two must be represented in 
one equation of the second degree, as in page 64. For the student who 
is disposed to try to represent the two cases in one equation, we give the 
result, namely, 

«« + ]/y + 2xi/ «= 144. 


the first (considered by itself) has an infinite number of solutions ; so 
also has the second. As follows : 

Solutions of the 

Solutions of the 

first equation. 

second equation. 

ar =» 10 


ar= 10 


x = 10i 

y = li 



y = 4 

ar= 11 

y = l 

ar= 11 

y= 1 





X = lli 







We have taken the same set of values for x in both ; and we iind 
the corresponding values of y different, generally speaking, but the 
same in one particular case : that is, we find a set of values x = 11, 
^ = 1, which satisfies both equations. The question now is, among 
all the infinite number of sets of values which satisfy one or the other 
equation, how many are there which satisfy both ? There is only one, 
as we shall find from the following process of solution. 

l{x-\-y = 12, it follows that x = 12 — y. Substitute this value 
of Jr in the second equation (which maybe done, since the solutions of 
the second which we wish to obtain are only those which are also solu- 
tions of the first), and we find 3(12 — y) — 2y = 31, or 36 — 5i/ = 31, 
or ^ = 1 . It appears then, that the supposition of both equations 
being true at once consists with no other value of y except 1, or 
(since x-\-y = 12) oi x except 11. 

Let the equations proposed be 

ax-\'by = c poc + qy = t 

where a, 6, c, p, q, and r, are certain known quantities. 

First Method, Obtain a value of one of the unknown quantities 
from one equation and substitute it in the other. The resulting equa- 
tion will then have only one unknown quantity. 

From the first equation, 

X = 


which value substituted in the second gives 

a ^•^ -^ aq — bp 

X = 


To obtain x, find y from the first equation, and repeat the process, 
which gives 

c — ax . Qc-^oax co —> br 

Or substitute the value of ^ first obtained in the previous expression 
for X ; thus, 

c — bt/ , bar^~bcp 

o ' y "■ aq — bp 

1 caq—'Cbp — {bar — bcp) caq — bar 

^ aq — bp aq — bp 

__ fl(c9--6r) . c^by __ cq-^br 

09 — op a aq — bp 

Verification. If x = ^^^, and y = ^^^ 

ay — bp^ •^ ay — 6/> 

ax + by = — ^ — r — i- 

•^ aq^bp aq — bp 

acq — bcp c^aq^^bp) 

aq — bp aq — bp 

cpq — bpr aqr — cpq 

px + qy = -^ — v^ + ^ p- = r 

•*^ ^»7 aq-^bp aq — bp 

Second Method. Multiply both the equations in such a Way that 

the terms which contain the same unlcnown quantities may have the 
same co-eflicient ; then add or subtract the two results, whichever 
will cause the similar terms to disappear. Generally, the shortest 
method is : Multiply each equation by the co-efficient which the 
quantity not wanted has in the other equation. 

To find y. 
ax + by = c (x)j? pax-\-pby =i pc 

px + qy = r (x)a pax + qay = ar 

(— ) oyy— ij^y = ar^cp 

, V ar — cp 

{-r)aq-bp y = j^:^ 

To find X, 
ax-j-by = c (x)y (iqx + bqy = qc 

px + qy = r (x)& bpx + bqy = hr 

( — ) aqx^bpx = cq-^br 

. . cq — br 


Third Method. Obtain a value of one of the unknown quantities 
from each of the equations, and equate the values so obtained. 

c — ox c ^ 6y 
From the first equation y =s — r — x ^ =- 

From the second equation y =s ^-— x ^ 2j_ 

•^ 9 P 

C ^ ax r — px eg — br 

b q a(f — bp 

a p " aq — bp 

Let the student now repeat all the three processes with the fol- 
lowing equations (see page 38). 

^ . __ cb'^bc 

i I y "" ab'^bu' 

(1.) 3a:-2y = 14 (x)2 6a;-4y = 28 

2ar+3y = 100 ( x )3 6ar4-9y = 300 
(-) 13y = 272 y = 20{| 

(x)3 9a;-6y = 42 (x)2 4a:+6y = 200 
( + ) 13a: = 242 j: = 18:^ 

(2.) X'\-y = a x—y = b 

( + ) 1x = a + 6 a: = ''"*" 

(— ) 2y ^ a—b y= 2 

(3.) px-\-y = 1 ^—py == 2 

The first is px+y = 1 

Tlie second, ( X )/? px—ppy = 2p 

(-) y+ppy ^l-^p y = i^ij 


pp p l-^pp 



(4.) 3a:— 7 = 4-|-(a:+y) or 2x— y =11 
2y + 79 = bx or 6j:— 2y = 79 

The first, ( X )5 10a:-6y = 65 
The second, (x)2 10a:— 4y = 158 
(-) y = 103 

From the first, x = ^^^ = 57 


(5.) 3a:-f 4y = 13 4a:+6y = 10 

(x)4 12a: + 16y = 52 (x)3 12a: + 15y = 30 

(-) y = 52-30 = 22 

(x)5 15a:-f-20y = 65 (x)4 16a:-f-20y = 40 

(-) a: = 40-65 = —25 

the problem producing these equations must therefore be treated as 
before described. 

. ax + by = C cq^hr ar^cp 

Havmg given a: = -^ y ^ 

px-i-qy = r ag^bp ^ aq — bp 

we may find the following : 

f ax^by = C ^ ^ cq-^br ^ cp^ar ) 

\px—qy = r aq^bp ^ aq^bp^ 

f ax—by = C ^ _ cq-\-br _ flr+^j 

On looking at the two first sets of equations, we see that they 
differ in this, that + by and + qy in the first are replaced by — by 
and — qy in the second. But we know that ax — by\& the corrected 
form of axH-6y (see page 46, &c.), and px — qy of px + qy; and 
since we have proved that the corrections may be deferred to any 
stage of the process, it follows that we may take the solutions of 

ax -^ by =^ C eg — br ar — cp 

px + qy = r ^q — bp aq — bp 

and conclude, that the corrected solutions will be the true result of 
the corrected equations. This if done will give results as above; for 
the value of x, corrected by the rules in page 52, gives 

--cq-^-br br — cq cq-^br 

— ■ L or A or -^ T (Page 64). 

— aq -\-bp bp — aq aq — bp ^ ° ' 





Or as 
ax + by-\'CZ 
ax-i-by-i-cZy &c. 
ax-\-bi/-i-cZy &c. 

and a similar process must be followed for j^.* In this way, any 
results derived from an expression such as ajr + J^ + cz, may be 
made to furnish the corresponding results which would have been 
derived from ax — bi/ — ez, cz + by — ax, or any other variation 
-which is only in sign. We write underneath the different cases which 
may arise, and the manner of referring them to the one which is 
chosen as the representative of all. 

The ollowing May be considered as 

ax + by^cz ax-^by + cz 

ax + by-^-cz 

ax + by + cz 


Any other case may be taken as the representative of the whole : 
thus, if it were ax — by — cz, then ax-\-by-\- cz must be considered 
as ax— by — cz, &c. 

Problem, f On the intersection of straight lines. 
Principle. It is proved in geometry, that if any angle BOA (for 
simplicity, we use a right angle) be drawn, and a straight line A B 
cutting the sides which contain it in A and B, and if from P, any 
point between A and BX parallels be drawn, PN to OA, and PM 
to OB, and if OA, OB, PM, and PN be measured in inches, or 
teuths of inches, or any other convenient unit (provided it be the 
same for all) ; then (PM meaning, not the line itself, but the number 
of units in it) the product of PM and O A, added to the product of 
PN and OB, will be equal to the product of O A and O B, or 


• Atte:id here to the remarks in the Preface, on the necessity of 
working more examples than are given in the book. 

t If the beginner have no knowledge of the most common terms of 
geometry, he must either acquire it, or omit this problem altogether. 

X From this supposition we start; we shall afterwards make use of 




Let there be two such lines, AB and A'B' (draw a large figure 
and insert A'B'), cutting one another in the angle BO A in P ; having 
given OA, OB, O A', and OB', required the value of PM and PN. 

Let OA = 10 units 0A'= 7 units 

OB = 8 .. 0B'= 15 .. 

PN = a: units 
PM=y .. 

Then, because P is on the line AB (preceding principle), 

OAxPM+OBxPN = OAxOB or 10y+8a: = 80 

Again, because P is also on A'B' (similar reason), 

OA'xPM + OB'xPN = OAxOB' or 7y + 16ar = 105 

Here then are two equations which y and x must satisfy. Solving 
them by either of the preceding methods, we have 

a: 01 PN is 5-3 units; V or PM is 3-= units. 
General Case, Let OA = a units OA' = a units. 

OB = 6 .. 0B'= 6' .. 

PN = a: units 
PM=y .. 

The equations to be solved, which are followed by the solution at 
length, as at page 70, will then be 

ay + bx = ab 
a'y + b'x = ab' 

ady + a'bx = acdb ab'y + bb'x = abb' 

aa'y + ab'x = aa'b' a'by-\-bb'x = a'bb' 

(a'b'^ab')x = aa'(b^b') {a'b'-ab')y = bb'ia'-^a) 

, b — h' , • , a! — a 

ah — ah ^^ ah — ah 

the preceding theory of algebraical symbols to extend our results to the 
case where P is not between A and B. AH words in xialxci point out 
what is peculiar to the first case. 

In the preceding ^ure aie fbur lines, AB, CD, EF,aDd GH, 
cuttiog the axes* in aa many diSerenl ways as is possible, that is, no 
tno inNnecting both OA and B on the same side of O. Tiiere 
are theniore, six points of intersection (as in the figuTe), and (he 
perpendiculara drawn from them to the aies O A and O fi are 
inserted ; but no letters are put to tbem, it being understood that P is 

* Aim, the principsl Ui 
nutntioned, and bngtliened in both diiecUoni. 

■md O B containing the angle liiit 


the point of intersection under consideration,* while P M and P N 
are the perpendiculars, as in the first figure. 

Let the distances at which the four lines cut the axes be as 
follows, in numbers of the unit chosen : 

OA = 3 0C = 8 0E = 4 OG = 2 
OB = 6 0D = 3 OF = 2 OH = 4 

Let us first take the intersection of A B and C D. Place the point 
P at this intersection, and let P M = ^ and P N = j?, as in the first 
figure. Then, because P is on the line A B, we have, from the prin- 
ciple stated at the outset, 

3y + 6ar = 18 

But on looking at the line C D, we observe^ 1 . That the principle 
does not apply directly to it, because C D is not contained in the 
angle BOA, but in the contiguous angle BOC. 2. That OC (in 
length 8 units) is not measured in the direction OA, but in the 
contrary direction. If, therefore, we write 8 instead of 8, and with 
this form the same equation as we should have formed if O C had 
been measured towards A, we shall then, by correcting the equation 
as in page 66, obtain another equation with which to proceed .f That 
is, because the point P is on C D, we have 

8y -f 3a: = 8 X 3 or (page 66) ^8y + 3x == 24 = -24 

or 8y-3:r = 24 

[For, since 3x — 8^ = — 24 denotes that 8^ exceeds 3jr by 24, 
this may be written 8^ — 3 j: = 24.] 

Solving the two equations thus obtained, namely, 

3y + 6a:=18] r^=l^ 

8y-3x = 24j [^ = 3^ 

which, with the succeeding results, may be verified by measurement, 
as well as firom the equations. 

* Let the studeut draw this figure on a large scale in iuk> and mark 
the letters belonging to the point under consideration in pencil, to be 
rubbed out on passing to a new point. 

t We cannot refer to any single page here. This principle is the 
total result of Chapter IL 


To apply the same method to every case^ we must distinguish 
all the lines OA, OB, &c. by the negative sign, which do not fall in 
the directions of O A or O B. That is, 

OA = 3 0C = 8 0E = 4 OG = 2 
0B = 6 0D = 3 0F = 2 OH = 4 

Proceeding with these as in the last case, we have as follows : 
Intersection of Equations. Corrected equations. 

1. ABand CD^--^ — ^ \ 

(8^-|-3a; = 24 83/-3a: = 24 ) 

2. ABandEF^-^ _ - _ ^ \ 

(4y + 2a;=4x2 4j^+2a:= -8 3 

3. AB and GH ' ^ ^ 

2y + 4ar = 2x4 2y 

4-6x=18 ) 
-4ar= -si 

4.CDandEFi^y+^^ = ^^ 8y-3a; = 24 J 

I4y + 2a:=4x2 \y^-1x^ -8 J 

5.CDandGHpy+^^ = ^'5 8y-3a;=24 J 

(2y4-4a;=2x4 2y-4a; = -8 J 

6.EFandGHp2^+^^=^^'^ 4y + 2a; = -8^ 

t2y+4a:=2x4 2y-4a; = -8 S 

These corrected equations are some of them sot arithmetically 
true ; for instance, 4y-f 2i' = — 8. But remember that they are made 
on the supposition that PM(y) and PN {x) are measured in such a 
direction that P falls within the angle AO B; which may not be the 
case. We may, however, apply the rules in page 70 to their solution ; 
of which we shall give one case (the second) complete, as follows : 

3y -f 6j; = 18 
4y 4-2ar = -8 

Multiply the second equation by 3. 

12y-f-6j; = -8x3 or 12^^ + 6.r = -24 

Subtract this from the first, which gives 

H 2 


(3-12)y = 18- (-24) = 18 + 24 = 42 

^"■3 — 12""— 9"" 9"" 3 

From the first equation^ 

_ 18-3j^ __ 18-3(— 4j) _ 18+14 ^ ^1 
6 6 6 3 

On placing the point P at the intersection of AB and EF, we 
see tliat PM {y) is measured in a contrary direction to the one 
supposed, and PN {x) in the same direction, as would be pre- 
sumed from the negative sign of the first, and the positive sign of 
tl)e second. 

Proceeding in the same way, we have the following values of x 
and y in the six cases : 

Intersection of, 

Value of x(PN). 

Value of^ 

1. AB and CD 



2. AB and EF 



3. AB and GH 



4. CD and EF 



5. CD and GH 




6. EF and GH 



[Remember that in such an expression as — IJ, the — refers to 
the whole ; that is, it is — (1 +1) or — 1 — 4, not — l+i-] 

Upon examining the six lines by a well-constructed figure, it will 
appear that whenever an answer is negative, it is measured in a 
direction contrary to that which was supposed in the principle at 
the beginning. 

The corrections might have been deferred to the end of the process, 
in the following manner : 

The equations (see page 74) 

ay -\-bx = ab ci'y + b'x = a'b' 

give X •=^ aa 77 77 y = 00 -n t, 

^ ab — ab •^ ub — ab' 


Take the two following sets (the second being case 6), 
ay '\'bx =i ab 4y + 2a; = 4x2 

a'y + b'x = a'b' 2y -f 4a: = 2 x 4 

which agree if 

a = 4 4 = 2 a'=2 4 = 4 

Substitute these values in the expressions for x and y, which give, 
a: = 4x2x-iz±= .=(«8)x.=l±± 

1X2-4X4 -4-16 

= (-8)x(-^) =i? = 5 

y==2x4x ^_^, _ = 8X"-A = ^2^ 

2X2 — 4X4 20 5 

the same as before. We leave this problem, and proceed. 

We have shewn, page 67, that one equation between two unknown 
quantities admits of an infinite number of solutions; and that a second 
equation must be given by the problem, or it is not reducible to a 
single answer. But this second equation must be independent of the 
first, that is, must not be one of those which can be reduced to the 
first. For instance, x -\-y = 1 2 admits of an infinite number of 
solutions; and if the second equation be either of the following, 

2j:+2y = 24 3a;-f3y = 36 ^a;-f^y = 6 

til tC 

3a:-18 = 18-3y 2x+y = 24-y, &c. 

the same infinite number of solutions will still exist; for if j:4-^ =s 12, 
all the equations just given must be true. That is, instead of giving 
two equations, we have only given the same equation in two different 

Now, we have already found (page 25), that in one case the index 
of an infinite number of results was the appearance of the result in the 

form -. We proceed to see whether this will be the case here. 

,n ^(lX-\-by = c ) , cq — hr ar — cp 

If < ^ > then X = -^ T- y = r- 

IpX-^qy = r ) aq^bp ^ aq^bp 

To try a case in which the second equation is dependent on the 
first, suppose 

p =s ma g = Tnb r = mc 


in which case the second equation becomes 

max-^mby = mc (-r-)m aar-fJy = c 

the same as the first. Substitute the above values of p, q^ and r, in 
the values of x and y^ which gives 

cmb^bmc amc — etna 

amb^~btna ^ amb^-^bma 

in which the same anomaly appears as in page 25, and with the same 

If there be three unknown quantities, by reasoning similar to that 
in page 67y we may shew that there must be as many as three inde- 
pendent equations, or else the problem admits an infinite number of 
solutions. We will shew in one instance the method of proceeding. 


2ar + 4y-3z = 10 (1) 

5x^37j+2z = 20 (2) 

3a:-f2y+5z = 50 (3) 

Multiply both sides of (1) by 2, and of (2) by 3, in the results of 
both of which z will have the same coefficient. 

Equ.(l)x2 4x + 8r/^6z ^ 20 
Equ. (2) X 3 15a;-9y-|-6z = 60 

( + ) 19a:-y = 80 (4) 

Repeat a similar process with equations (2) and (3). 
Equ. (2) X 5 25a:-15y-f 10;^ = 100 
Equ. (3) X 2 6ar+ 4y+10z = 100 

(-) 19j;— 19y = 

(-j-)19 a;— y = or a: = y (5) 

Two equations are thus found (4) and (5), containing x and y 
only, not z. These solved, give 

40 40 

Substitute these values in either of the three given equations, the 
econd, for example. Then 


There is an artifice^ which is useful when one only of the three 
unknown quantities is required. Suppose, for example, that in the 
preceding equations only the value of z is wanted. Take two new 
quantities, m and n, not yet known, but afterwards to be determined 
in any manner that may be convenient, f Since the two sides of an 
equation may be multiplied by any quantity, multiply (2) by »w, and 
(3) by n, and add the results to (1). This will give 

(2+5m-f3w)a; + (4— 3m+2w)y 

+(2m+67i— 3)2? = 10-|-20m-f50w..(A) 

Since m and n may be taken at pleasure, and since the value of 

2 only is wanted, let m and n be such that 

2+5m+3n = or 5wi-f 3w = —2 
4— 3»H-2w = or 3m— 2n = 4 

uien m and n must be the solutions of the preceding equations, which 

give w = — , n = — -—. But in (A) the terms which contain x and 
y disappear, being multiplied by 0. Therefore, 

(2m+57i— 3)z = 10-f 20m-f 50/1 

z = 

2m+5n-3 2x^+6(-|)-3 

(Multiply numerator and denominator of this fraction by 19, and 

reduce it), 

_ 10X19 + 20X8 — 50X26 _ —950 _ 950 _ 50 
— 2X8 — 5X26 — 3X19 "" —171 ~ 171 "" 9 

Anomalies, A problem may give rise to two equations which are 
absolutely incompatible with each other, such as the following : 

x-\-y = 12 x+y = 13 

or it may happen, that if there be three unknown quantities and three 
equations, one of the latter may be impossible if both the others be 
true, though it be not inconsistent with either of the others singly. 
For instance, take 

x-—!/ = 10 y— z = 11 x—z = 12 

• An artifice is a name given to any process by which either the 
prfhciple or practice of any method is shortened, either generally, or in 
any particular case. 

t Such quantities are usually called arbitraiy. 


The 1st and 2(1 are true ifx = 20 ^ = 10 ;?=— 1 
The 1st and 3d .... a? = 20 y = 10 Z = 8 
The 2d and 3d .... a? = 20 ^ = 19 Z=8 

But no values of x, y, and s can satisfy all three together, as will be 
evident by adding the first two. For if x — y = 10, and y — ;; ^ 11, 

(^ — y) + (y — ^) — 21 or a? — ;2; = 21 

which is inconsistent with the third equation. 

It is lefl to the student to examine the general solution in page 
70, and to shew that when the two equations become incompatible, 
the values of x and^ take the form discussed in page 21, namely, 
that of a fraction which has for its denominator. It might also be 
shewn, that the problems which give rise to incompatible equations 

admit of an interpretation similar to that already derived from results 

of the form - in the page last quoted. 

It will be found, that two equations such as are treated in thb 
chapter will in no case become inconsistent except when they can be 
so reduced as to have their first sides identical, and their second sides 
different; such as 

2a: + 3y = 10 2x^Zy « 12 

The subtraction of the first from the second would give as 2, 
an absurdity of the same character as ax ^sz ax -^c (page 23), from 

which the form - was ^st derived. 






The coDtinoal occurrence of the multiplication of the same quantity 
two, three, or more times by itself, has rendered some abbreviations 
necessaiy, which we proceed to explain. 


* multiplied by a*, or XX^ is called the second power of jr. 

F ^^ a:, or XXXy the third power of x. 

* ^^^ • • • • ar, or XXXX, the fourth power of x. 

and so on. Or, n ares* multiplied together give what is called the 

I nth power of x. The second and third powers are usually called the 

^9^^e and cube. Thus jcjt is called the square of x, and is read 

' sgrutre; and xxx is called the cube of x, and is read x cube : and 

a number multiplied by itself is said to be squared ; multiplied twice 

by itself, it is said to be cubed, &c. 

By an extension, x itself is called the first power of x. 
The abbreviated representation of a power is as follows : over the 
letter raised to any power, on the right, place the number of times the 
letter is seen in that power. Thus, 

arj: is written x^ 

n^ fl^ n/> fy» 'W' 

JU *Mj JU %A/ ••».•»». ju 

and so on. Here 2, 3, 4, &c. are called exponents of x. Similarly, 

(a + 6) X (a + b) is written (a + hf 
(a-l-6)x(a + 6)x(a + J) (a + J)' 

and so on. The following results may now be easily found : 

* The beginner's common mistake is, that x multiplied n times by x is 
the Bth power. This is not correct ; x multiplied once by x (^xx) is the 
gecond power ; x multiplied n times by x is the (ft + l)th power. 


XXX = x^ x^xx s= x^ ar* X a: = o;*, &c. 

{a-^xf = a« + 2aa:+a:« 
(a—xy = a^-~2ax+x^ 
{a + x){a^x) = a^^x^ 
{a^+ax+a^){a—x) = a^^x^ 

{a + bf = aH3tt26+3aJ2 + 63 
{a^bf = a-^-3tt2& + 3a6«-ft3 

To multiply together any two powers of the same letter ^ let the 
exponent of the product he the sum of the exponents of the factors. 
Thus, to multiply x^ and j:*, 

x^ is XXX 
X^ is XXXX 

a?xx^ is xxxxxxx or oc^ or a:^+* 

By the extension previously made of calling x the first power of ir, 
and considering it as having the exponent 1, or as being j?*, this rule 
includes the case where x itself is one of tlie factors. Thus, 

x^ XX = a^ 01 x^ xx^ ^=^ a;7+^ = afi 
Examples, a;* x x^^ = x^^ x^ X ar^* = x^'^ 

xxx^xx^xx^ ^ x^^ a^xa^ = a^^ 

3 a 6« X 4 a2 J = Ua^b'' 2abxab = 2a%^^ 

the latter term, written at full length, would be 


To divide a power by another power of a less exponent, subtract the 
exponent of the divisor from that of the dividend. For instance, what 
is x^° divided by j:*? Since 10 is 7 + 3, or since jr*° = jr' x ^ 

(-T-)^ -y = j:' or x^°-\ Similarly, or^ -f- x (or x^) = x^; a-^^-i- ^yii 

= j' orx; j:"-f- x^=:x^; j:«+6 -;-i« = a*; a^b^c^^abc = abc^. 
Anomaly 1. If we apply this rule to the division of x^ by a*, we 
have j«-^ JT^ — j.a-&. If it should be afterwards found that a = 6, 
the preceding result becomes j«-o or x®, a symbol which as yet has 
no meaning. We return, therefore, to the original operation, which 
is, to divide x^ by x^ where i = a, or, which is the same thing, to 
divide x^ by x^. The answer is evidently 1. 


Now why, instead of 1, the rational answer, did we obtain af^, 
which has no meaning? Because we applied the preceding rule to a 
case which does not fall under it. That rule was, '' to divide a 
power by another power of a ksi exponent,'' &c., and was derived 
from the preceding rule of multiplication, which rule of multiplication 
di4 not apply to any cases except where both factors were powers 
of Xf and, consequently, where the exponent of the product was greater 
than the exponent of either factor; that is, where x^ is the product, 
and a^ one of the factors, that rule does not apply unless b be less 
than a. 

When we come to this symbol (j:^ we must do one of two things, 
either, 1. Consider x^ as shewing that a rule has been used in a case 
to which it does not apply, strike it out, and write 1 in its place ; 
or, 2. Let j° (which as yet has no meaning) stand for 1 ; in which 
case the rule does apply, and gives the true result. Therefore, we 
lay down the following definition. 

JB|y any letter with the exponent 0, such as a®, we mean 1 ; or every 
quantity raised to the power whose exponent is 0, is 1. 

Anomaly 2. If we apply the equation j;« -f- x* ^ a^-& to a case 
in which b is greater than a, say i ^ a + 6, the mere rule gives 

a result which has no meaning. The reason is as before ; a rule has 
been applied to a case to which it was not meant to apply. To find 
the rational result, remember that x^-^^ := x** x x^; and 

X** x^ 1 

j:«+6 x^ X x^ 7 

the last result being obtained by dividing both numerator and deno- 
minator of the preceding fraction by x^. The following are similar 
instances ; in the first column is the rational process, in the second 
the (yet) improper extension of the rule : 


X* "" 

3^ 1 
X^ , X X 


x^ 1 


x" 1 

x^ " 

x'\a^ ~~ 3^ 


-5 = a:^-8 -« ^-6 

1 ;y.l7-20 — «.-.3 


To make the rule applicable, we must agree that 

/*»""1 /*»""0 /*»""3 

(which as yet have no meaniDg) shall stand for 

tliat is, instead of x-^ being the sign that the result is to be abandoned, 
and 1 -7- j: written in its place, we agree that ar-^ shall mean the 
same as l-~-x. And we are at liberty to give x^^ any meaning we 
please, because as yet it is without meaning. We lay down, therefore, 
the following deBnition : 

A letter with a negative exponent means unit^ divided hy the same 
letter with the same numerical exponent taken positively ; or, 


X means — 


Our two rules for multiplication and division will now be found 
to be universal. The following are instances, arranged as before : 

1. 1 1 X^ X^ 5 

' »^ ^^— ^^ •— y ^—^ ^^ _^^ ^^ Q* 

X^ ' x^ x^ 1 x^ 

^ 2._^ 1_ 

•^ ^ x^^ x^" x^ 

1-i- 4 — J- J_=:JL = 1 

.r * x^ X* x^+* J"' 

We have now added to our first definition of an exponent in such 
a way that two fundamental rules remain true in all cases where the 
exponents are any whole numbers, positive or negative. These rules 

afxa^ = af-^^ of" -^ a:^ = af'^ 

But we have no meaning for a number or letter with ^fractional 
exponent, such as 

x^ x^ x^ x^^ x'^^ &c. 

Instead of waiting until some improper extension of the preceding 
rules shall force us to the consideration of the manner in which it will 
be most convenient to give meaning to the preceding symbols, we 
shall endeavour to anticipate this step. And first we shall ask what 

should x^ mean, in order that the preceding rule may apply to it? 
In this case, since \'\-\ = 1, we must so interpret x* that 

x^xx^ = a;* "^4 =• x^ or x 


consequently, jr« is that quaDtity which, muliiplied by itself, gives x, 
or is what is called in arithmetic the square root of x. Similarly, 

since i + i + i = l> ^^ must so interpret x* that 

x^xx^xx^ = x*"*"*"*"* = X* or a: 

that is, x* must be the cube root of x. By a root of x we mean the 
inverse term to power; that is, if m is the third power of n, n is called 
the third root of m. As follows : 

Name of n. 

Square root of m 
Cube root of m 
Fourth root of m 
FifU) root of m 

Equation implied in the fore- 
going name, 
nn = m 
nnn = m 
nnnn ^ m 
nnnnn = m 

Thus, 4096 is a number which has exact squart;, cul}e, fburtli, mid 
sixth, and twelfth roots. 

Its square root is 64 because 64 x 64 = 4096 

Its cube root is 16 16x16x16 = 4096 

Its fourth root is 8 .... 8x8x8x8=4096 

Its sixth root is 4 .... 4x4x4x4x4x4 = 4096 

Its twelfth root is 2 

r 2X2X2X2X2X2) ^^^^ 
{ ^ >= 4096 


These results should, if the preceding interpretation can be relied 
on be thus expressed : 

64 = (4096)* 8 = (4096)i 

16 = (4096)i 4 = (4096)* 

2 = (4096)''' 

But, according to the notation best known in arithmetic, they would* 
be thus expressed : 

* -v/ is deriyed from the letter r, the initial of radix, or root. This 
Bymbol is now generally used for the square root, which, in ninety-nine 
out of a hundred of the applications of algebra, is the highest root which 
'veill occur. 


64 = V^4096 8 =: V4096 

16 = 1^4096 4 = V4096 

2 = V4096 

Proceeding with the interpretation of the fractional exponents, 
'^ ought to signify the cube root of x* ; for, if the preceding rules are 
to remain true, we must hare 

and, by the same sort of reasoning, we may conclude that x» should 
stand for the nth root of i^. But here we may shew that we cannot 
decide upon the propriety of the preceding interpretation without 
some further acquaintance with the connexion between roots and 
powers. For instance, 

a^i or a:*+* ought to be a;* x a:* or a^V^x 

But 2| is I; therefore, 

X ought to be ^ or v3? 

Consequently, x^ vx ought to be V^ 

where, by " ought to be," we mean that if it be not so, we cannot,* 
under the preceding interpretation, apply the common rules of arith- 
metic to our assumed fractional exponents. Again, since \-\-\ = |, 

x' X x^ or vx X ^x ought to be a;« or V^ 
but as yet we have neither proved that 

a:* Vx = V^ or that Vx x V^ = VI? 

For the purpose of shewing that the conjectural interpretation as 
yet given leads to no erroneous results, we premise the following 
arithmetical theorems. 

Theorem I. If a be greater than b, than a* is greater than ^, 
a^ than 6^, &c. For aa is then the result of a multiplication in which 
more than b is taken more times than there are units in b ; therefore 
a a is greater than bb: in a' or a^a, more than 6' is tal^en more times 

* The student must remember that we are perfectly free to make x ^ 
imd av mean different things, if we only take care not to confound the 
two. But it would be inconvenient that SJ should any where have a 
meaning different from that of {. 


^^n there are units in h, and so on. By changing the order in which 
^« letters are named, the theorem may be differently worded, thus : 
^b be less than a, then 6* is less than a*, &c. 

Tkeorem II, If a be greater than b, a~' is lets than 6~*, a~* is 

leu than b~*. For if a be greater than b, - is less than j- ; and since, 

in that case, a? is greater than h^, therefore —^ is less than t-,» ^nd 

so on. Similarly, if a be less than b, a"^ is greater than b~^, &c. 

Theorem III, If a be c^ua/ to b, then a' is equal to 6*, a* to 6^, 
and so on. This is evident from page 3. 

Theorem IV, If a be equal to 6, the square root of a is equal to 
the square root of 6, the cube root of a to the cube root of 6, and 
so on. Let m and n be, for example, the fifth roots of a and b (which 
last are equal), then a and b are the fifth powers of m and n; Mm 
were the greater of the two, its fifth power a (Theorem I.) would be 
greater than 6, which is not the case. Similarly, if n were the 
greater of the two, b would be greater than a ; therefore m must be 
equal to n. In the same way any other case may be proved. 

JTieorem V, If a be greater than b, the square root of a is 
greater than the square root of i, &c. (We put the preceding argu- 
ment* in different words). Since a is the square of its square root, 
and b the same ; if the square root of a were equal to the square root 
of 6 (by Theorem III.), the square of the first (or a) would be equal 
to the square of the second (or b), which is not the case. If the 
square root of a were less than the square root of b (by Theorem I.), 
the square of the first (or a) would be less than the u^iSiSsJfff- of the Y' 
second (or 6); which is not the case. The only remaining possibility 
is, that when a is greater than 6, then the square root of a is greater 
than the square root of b. Similarly, if a be less than 6, the square 
root of a is less than the square root of b : and so on. 

Theorem VI, An arithmetical quantity has but one arithmetical 
square, cube, or any other root. For suppose a, if possible, to have 
two different cube roots, m and n ; one of these two is the greater, let 
it be m. Then (Theorem I.) the cube of wi (or a) is greater than the 
cube of n (also a) ; but the last two assertions are contradictory, 
therefore a cannot have two different cube roots, &c. 

* The nature of the argument is supposed to be well understood from 
the last ', it is practice in the use of terms which is here given. 




All whole numbers have not whole square roots, or cube roots ; 
and the higher the order of the root, the fewer are the whole numbers 
lying under any given limit which have a root of the kind. The 
following table will illustrate this. 




)ers which hi 


ive a whole 



Value of 
the Root. 



































































A number which has not a whole root has not an exact fractional 
root. At present we shall only enunciate the following proposition, 
without proving it, leaving the student to try if he can produce any 
instance to the contrary. 

No power or root of a fraction* can be a whole number. 

Consequently, all those problems of arithmetic or algebra are 
misconceptions, which require the extraction of any root of a number, 
unless that number be one of those specified in the preceding table 
(continued ad infinitum) as having such a root. But though we may 
not look for the exact solution of such problems, we shall shew that 
solutions may be found which are as nearly true answers as we 
please ; that is, we shall prove the following theorem. 

Though there is no fraction whose nth power is exactly any given 
whole number, we may assign a fraction whose nth power shall differ 

4 64 
♦ That is, of a real fraction ; -, — -, &c. are whole numbers in a 

fractional form. It is not necessary to prove this strictly here ; because, 
were it not true, what follows would not be incorrect, but only useless. 



from that whole number by less than any quantity named, say '0001, 
or '0000001 y or any other small fraction. 

N.B. With the most conTeoient method of finding such a fraction 
we have here nothing to do, but only with the proof that it can be 
found. This is shewn in arithmetic as to the square and cube root 
only at most. We must enter into the proof of this at some length, 
and shall lay down the following LemmasJ* 

Lemma 1. The powers of 2 are formed by one addition : thus, 

2 + 2 = 2« 2« + 22=2« 2«+23=2* 

or generally, 2«+2'»=2«x2 = 2'»+i 

Lemma 2. The powers of a fraction are formed by forming the 
powers of the numerator and denominator : thus, 

(a\3 a a a aaa o^ 
bJ "^ b^b^b " bbb "^ b^ 

Lemma 3. If j9 be less than g 
Then ap is less than aq 

If p be less than g 

and a 6 

ap is less than bq 

Lemma 4. If v be less than unity its powers decrease con- 
tinually. For instance, if v be one half, its square (i x i) is one half 
of one half, which is less than v; its cube is one half of one fourth, 
which is less than v^; and so on. 

Lemma 5. If one of the positive terms of an expression be 
increased the expression itself is increased, &c. Thus a — 6 is 
increased by increasing a, and decreased by decreasing a ; but it is 
decreased by ina'easing b, and increased by decreasing b. 

Lemma 6. Ifvbe less than 1, then 

(1 +vy is less than 1 ^^v ox \ +(4— l)i; 

(l+w)3 l+7i? or l-(.(8-l)t? 

(1+v)* l^-16^; or l+(16-l)t; 

or (I 4- 1?)** is less than 1 -f (2**— \)v 

Firstly, (1 +t;)* or (1 +t;) (1 -f-v) is 1 -f-Qv+vS which, since 
(Lemma 4) v is greater than v', is (Lemma 5) increased by writing 

* A Lemma is a proposition which is only used as subservient to the 
proof of another proposition. 



V for V*. But it then becomes l + 2t; + v or 1 + 3 v. Therefore 
1 -h 2 V + v* is less than 1 + 3t; ; that is, (I + v)' is less than 1 -|- 3 v. 

Again, (1 + t?)2 is less than 1 + 3 © 

(Lemma 3) (1 -(.y)2(l +v) (1 +3v)(l + r) 

or {\+vy l+4v-\-3vl^ 

Still more, then, (Lemmas 4 and 5 as before) is (1 + vy less than 
1 +4t; + 3v, or 1 + 7v. 

Again, ( J + 1?)3 is less than 1 + 7 w 

Therefore (l+v)* (1 +7t?)(l +t?) 

or l+8t? + 72?« 

Still more is ( 1 -f. ?;)* less than 1 + g w + 7 1? 

or l+15t? 

We might thus proceed through any number of steps, but the 
following is a species of proof which embraces all. Suppose that 
one of the preceding is true, say that containing the wth power ; that 
is, let 

(l+t?)" be less than l-(.(2«— l)i? 

Then (Lem.3)(l +t?)"(l +v) is less than {l +(2"— l)v}(l +r) 

which product may be found as follows : 

l + (2n— l)t; 
1+ V 

l+(2» — 1)|; 

i; + (2»» — l)v« 

Add 1 +2« v + (2« — 1) t;« 

or (1 + vy^^ is less than 1 + 2«i; + (2'»- l)t?« 

Still more (Lemmas 4 and 5) is it 7 , ^^^ ,«^ , ^ 
lessthan { 1 + 2«z, + (2»«l)i; 

or l + (2'» + 2»— l)v 

or (Lemma 1) 1 -j- (2''+^— l)t? 

We have proved, then, that 
if (l + vf be less than l + (2'»— l)i? 

it follows that (1 + vf-^^ is less than 1 +(2"+^ — l)w 
or in the series of propositions contained in this lemma, each one 



must be trae if the preceding be true. But the first has been proved, 
therefore all have been proved. 

Lemma 7. Ifxhe greater than a, then 

(a: + a)^ is less than 0^+3 ax or «*-|-(4— l)ax 

(x+ay x' + Tax^ or x^ + (8— l)ax« 

(x+a)* a?* + 16ax' or x*+(16— l)ax» 


or (x + a)" is less than x" + (2"— l)ax" 

Because x is greater than a, - is less than 1; therefore , 
(Lemma 6), 

(l ^ly is less than 1 + (2'»-l)- 

But 1 +- = ^i-" /. (Lemma2) (l + ^V = (-i±^ 
X X ^ ^ \ xJ .r» 

Therefore, ^-^^^ is less than 1 + (2« - 1)- 
' x^ ^ ^x 

Multiply both sides by x*^ (Lemma 3), which gives 
Cg-ay* is less than x»+(2~— 1) — 


or a;~ + (2~— l)ax«-i 

We proceed to shew the proposition in pages 90 and 91 in a par- 
ticular case. Say the number is 10, the power mentioned is the cube. 
Can a fraction be found whose cube shall be within, say -0001 of 10? 
Since (2)' = 8, and (3)' = 27, 2 is too small and 3 too great. 
Examine the cubes of the following fractions falling between 2 and 3 ; 
2-1, 2-2, 2-3, &c. We have (2*1)' = 9*261, and (2-2)® = 10*648 ; 
whence 2*1 is too small, and 2*2 too great. Examine the fractions 
2-1 1, 2*12, 2-13, &c. lying between 21 and 2*2. We find, 

(2-16)» = 9-938375 (2-16)« = 10-077696 

Therefore 2*15 is too small, and 2*16 too great. 
Proceeding in this way, we shall find, 

(2-154)^ less than 10 (2-156)' greater than 10 

(2-1544)3 .. 10 (2-1546)3 .. 10 

(2-15443)3 .. 10 (2-16444)3 .. 10 

&c. &c. 




The only question now is, shall we thus arrive at two fractions, 
one having a cube less than 10, and the other greater, but both cubes 
so near to 10 as not to differ from it by '0001 ? Observe that in the 
preceding list, 

2' 2 exceeds 2*1 by only •! 

2-16 .. 2-15 .. -01 

2-155 .. 2-154 .. -001 

21645 .. 2-1544 .. -0001 

&c. &c. &c. 

and from Lemma 7, if a be less than x, 

(ar + a)^ is less than x^-^-lax^ 
or {x + ay-^x^ is less than 1 as^ 

Let X be the lower of one of the preceding sets effractions ; then, 
since x^ is less than 10, jr must be less than 3, and its square less 
than 9. Therefore, 7ax^ must be less than 7a x 9, or than 63a. 
Still more, then, will {x-\-(if — a^ (which is less than 7 a a*), be less 
than 63 a. Let x + a be the higher of the fractions in the set spoken 
of; then a, the difference, will at some one step become -OOOOOOl, 
consequently, 63 a will become -0000063, which is less than -00001. 
Therefore x may be so found that 

x^ is less than 10 (a: + -0000001)3 jg ^^^^^ ^i^m 10 
and (a' + -0000001)3-a;3 jg ^^^ ^^^^ -00001. 

But 10, which lies between the two cubes, will differ from either 
of the cubes by less than they differ from each other ; therefore, either 
fraction has a cube within the required degree of nearness to 10. 
The fractions required would be found to be 2*1544346 and 
2*1.544347. In a similar way any other case might be treated. 

Hence, the following language is used. Instead of saying that 10 
has no cube root, but that fractions may be found having cubes as 
near to 10 as we please, those fractions are called approximations* to 
the cube root of 10, as if there were such a thing as V^IO. Thus, 
.2*154 is an approximation to V^10> but not so near an approximation 
as 2*1544346; instead of saying that (2*154)' is nearly equal to 10, 
but not so nearly equal to 10 as (2*1544346)'. 

♦ Approximare, to bring near to. 


The student will now understand the sense in which we use the 
following words : 

Every number and fraction has a root of every order y either exact 
or approximate. 

When we shall have proved that in all cases n/a x V^a*= V^o*, 
what do we mean by this equation in the case where a ^ 10, and 
therefore o' = 100000 ? We mean that we can obtain two fractions 
of which the square and cube are within any degree of nearness 
(say '0001) of 10, which we call approximate values of V^IO and 

V 10, and that, on multiplying these two fractions together, we find a 
product which, being raised to the sixth power, gives a result within 
the same degree of nearness to 10^, or is an approximate value of 

V 10®. We shall anticipate the proof of both propositions, as one 
specimen of the method of passing from the strict to the approximative 
proposition will serve for all. 

Let a be a number which has both a square and a cube root 

(such as 64, or — ). Let x be the square root, and y the cube root. 
-? Then 

a;2 = « therefore {x^Y = a^ or x^.x^.x^ = a^ or x^ = a» 
y^ ^ a therefore {y^Y = a^ or y^ .y^ = a^ or y^ = a^ 
.'. a^y^ = a^a^ or {xyf = a^ 

for 3^^ is xxxxxxyyyyyy, in which the multiplications may be 
performed in the order 

xy.xy.xy.xy.xy.xy giving {xyf 
Consequently, xy is the sixth root of «* ; but x is the square root of a, 
and y the cube root, that is, 

xy = 1?^ or Va X ^^a = "^^ 

Now, suppose a to be a number which has neither square or cube 
root, such as 10. We can find fractions x and y^ such that x^ and y^ 
shall be as near as we please to a. Say that x^ = a-\-p and 
jy* =r a + ^ where p and q may be as small* as we please. We will 
begin by supposing p and q smaller than a. Hence (Lemma 7), 

* As small as we please does not mean that we can choose them 
exactly, but that, name any fraction we may, however small, they may 
be made (we need not inquire how much) smaller. 


96 [law op coNTiKumr.] 

(a+pY is less than a'-f-7/)a« 

(a + j)« a^ + 3qa 

But (x^y or ofi = (a-^pY (y*)2 or t^ ^ (a + qf 

ofi is less than a^ + 7pa^ 

y5 a^ + Sja 

(Lemma 3) afiy^ (a^ + 7/)a*) («« + 3 ja) 

or (a:y)^ is less than a* + (7;? + 3 j)a* + 21;? ja^ 

But since x^ (being a-^p) is greater than a, jfi is greater than a^; 
and since y (being + 9) is greater than a, y* is greater than a*; 
hence j:*^ or (i*^)* is greater than efl,a* or a*. Hence, 

(ary)^ lies between a* and a^ -r (7p + 3q)a* + 2l2?5'a^ 
and therefore does not differ from a* by so much as 

(7p + 3q)a'^'j-2lpqa^ 

Now, since p and g may be as small as we please, Tp-^-Sq and 
2\pq may be as small as we please, and, therefore (however great* a* 
and a' may be), may be so taken that the preceding expression shall 
be as small as we please. Tliat is, {xt/Y may be made as near an 
approximation as we please to a^, or xy is an approximate sixth root 
of a*. 

The preceding demonstration is not frequently given in books on 
algebra, but the result is assumed in what is called the law of con- 
tinuity. This term we shall proceed to explain, as it involves con- 
siderations which will be useful to the student; but as it may be 
omitted without breaking the series of results, we inclose it in 
brackets, and also the heading of the pages which contain it. 

[The word continuous is synonymous y^hh gradual or vnthout sudden 
changes. For instance, suppose a large square, with two of its oppo- 
site sides running north and south. A person who walks round this 
square will make a quarter-face at each comer, that is, will at once 
proceed east or west where before he was moving north or south, 
and vice versd, without moving in any of the intermediate directions. 

* The product mn, if n be given, and m as small as we please, may- 
be made as small as we please ; only the greater n is, the smaller must m 
be taken, in order to give the product the required degree of smallness. 

[law op continuity.] 


In this case be changes his direction discontinuoudy. If the figure 
bad had eight sides he would still have made discontinuous changes 
of direction, but each change of less amount; still less would the 
changes have been if the figure had had sixteen sides, and so on. 

But if he walk round a circle, or other oval curve, there is no 
discontinuous change of direction. If a geometrical* point move 
round a geometrical circle, there is no conceivable direction in 
which it will not be moving at some point or other of its course ; 
and between every two points, it will move in every direction 
intermediate to the directions which it has at those two points. 

The preceding illustration is drawn from geometry, in which 
continuous change is supposed ; and there is nothing repugnant 
to our ideas in the supposition, but the contrary, at least, when we 
imagine lines to be created by the motion of a point. But in the 
application of arithmetic, and eventually of algebra, to geometry, we 
have this question to ask. Can all geometrical magnitudes be repre- 
sented arithmetically? For instance, suppose B to start from A, 
and to move in a straight line till it is 100 feet from A, can we, by 
means of feet and fractions of feet, represent the distance to which 
B has moved from A, in every one of the infinite number of points 
which B passes through ? 

It is clear that between one foot and two feet we can interpose the 
fiactions 1*1, 1*2, 1*3, &c. feet; between 1*1 and 1*2 feet we can 
interpose 1*11, 1*12, 1*13, &c. feet; between 1*11 and 1*12 we can 

* Geometrical, formed with the accuracy which the reason supposes 
in geometrical figures. 


[law of continuity.] 


interpose 1*111, 1*112, 1*11 3, &c. feet; and so on for ever.* But 
assigning B any geometrical position between 1 and 2 feet distance 
from A, we cannot be prepared to say that we shall thus come at last 
to a foot and a fraction of a foot, which will exactly represent that 
position. And we shall now proceed to shew one position, at least, 
assignable geometrically, but not arithmetically. 



It is shewn in geometryf how to assign (by geometrical construc- 
tion, not arithmetically) a position to 6, in which the square described 
on A B shall be twice as great as the square described on 1 foot (as 
in the figure). We now proceed to inquire whether the line AB 
has, in such a case, any assignable arithmetical magnitude (a foot 
being represented by 1). If so, since every fraction of a foot can 
be reduced to a fraction with a whole numerator and denominator 


(Ar. 114,121), let A B be - feet, where m and n are whole numbers. 

That is, let AB be formed by dividing one or more feet each into n 
equal parts, and putting together m of those parts. Let this nth part 
of a foot be called for convenience a *^ subdivision ; '' then a foot 
contains n subdivisions, and AB contains m subdivisions. Then, 
from Ar. 234, it appears that the square on the foot contains nxn 
of the squares described on a subdivision, and the square on A B 
contains m x ni of the same. Hence, since the square on A B is 
double of the square on one foot, we must have 

mm = 2nn 

We now proceed to shew that this equation is not possible under 
the stipulation that m and n are whole numbers. Because n is a 

* The student must not imagine this phrase, or its corresponding 
Latin ad infinitum, to mean more than as long as we please, or as far as 
we please. 

t Euclid, book ii. last proposition. 

[law op continuity.] 99 

whole number, nn is a whole number, and 2 nn is twice a whole 
number, and therefore an even number; but mm equals twice nn^ 
therefore mm is even. Therefore m is even, for an odd number 
multiplied by itself gives an odd number. But if m be even, its half 
is a whole number ; let that half be m', then m = 2 m'. Substitute 
this value of m in the equation, which gives 

2m'x2m's= 2nn Am'm! =^ 2nn or 2mW= nn 

which last equation nn ^ 2mW may be used in a manner precisely 
similar, to shew tliat n must be an even number. Let its half be 
n' (a whole number), then n ^ 2n', and substitution gives 

2n' X 2n' = 2m'm' 4n'n' = 2mW or 2n'n' = m'm' 

which proves as before that r%' is even. In this way we shew that in 
order that the equation mm = 2nn may be true (m and n being 
whole numbers), we must have 

m (ml or half of iw) (m" or half of m') &c. 
n {vl or half of n) (n" or half of n') &c. 

all even whole numbers for ever. But this cannot be ; for if any 
number be halved, if its half be halved, and so on, we shall at last 
come to a fraction less than 1 . Consequently, the equation mmT=2nn 
cannot be true of any whole numbers, and therefore A B cannot be 

represented by any fraction — . 

The preceding equation (if it could exist) would give 

=r2or — X — = 2 or xx = 2 where x = — 

nn n n n 

and we have seen (page 94) that we can admit the equation xx z=z2 
only in this sense, that, naming any fraction, however small, we can 
find a value for jt, which shall give xx differing from 2 by less than 
that fraction. That is, instead of satisfying the equation 

xx^2 = 

we can only satisfy the equation 

XX "^2 = a quantity arithmetically less than ( ) 

where we may fill up the blank with any fmction we please, however 

This is sufficient for all practical purposes ; because no applica- 
tion of algebra which has any reference to the purposes of life can 


100 [law op continuity.] 

require a degree of accuracy beyond the limits of our sight, when 
assisted by the most accurate means of measurement. If we take the 
least possible Tisible line to be that of a ten thousandth of an inch, 
it will certainly be sufficiently near the truth to solve the equation 

XX — 2 = arithmetically less than one millionth of an inch, 
in cases where perfect exactuess demands that xx — 2 = 

The solution which nearly satisfies such an equation as xx — 2 =0 
may be either too small or too great ; that is, xx may be a little less 
or a little greater than 2. See page 94, where sets of solutions of 
both kinds are given for the equation xxx — 10 ^ 0. Hence, though 
X or V 10 has no existence, yet, since we can find two fractions, say 
a and 6, as near to one another as we please, of which the first is 
too small (or aaa less than 10) and the second too great (or hhb 
greater than 10) ; and since we can carry this process to any degree 
of accuracy short of positive exactness, it is usual to make use of 
such forms of speech as the following : 10 has a cube root, but that 
cube root is an incommensurable * quantity, not expressible by any 
number or fraction, except approximately; that is, we can find a 
fraction as near as we please to n/ 10. 

But still the following question remains : though we can, quam 
proximiyf solve the equation xx — 2^0, may there not be processes 
to which it may be necessary to subject that approximate solution, 
and may not those processes have this property, that any error, how- 
ever small, in the quantity to which they are applied, creates an error 
which cannot be diminished beyond a certain extent, however small 
the original error may be ? For instance, when the student oomes to 
know what is meant by the logarithm of a number (for our present 
purposes it suffices to say, that it is the result of a complicated 
process), he might happen to meet with a problem the answer to 
which is " the logarithm of x, where x is to be found from the 
equation xx — 2 := 0." Which of the two following propositions 
will he take ? one of them roust be true. 

1. "In taking the logarithm of :r,.the error committed in finding 

* Having no common measure with 1 ; not expressible by adding 
together any of the Bubdivisions of 1. 

t A Latin phrase frequently applied to this subject, and therefore 
introduced here. Translate it, as near at we plean. 

[law op continuity.] 101 

' may be made so small, that the error in the logarithm shall be less 
than any fraction named/' 

2. ** If an error committed in finding a number be ever so small, 
the error of the logarithm must be greater than (some given fraction, 

To this no answer can be given except the following caution : 
Whenever any new process is introduced, or any new expressioit, 
*t must be proved, and not assumed, that problems involving that process 
odmit o/*quam proximo, if not of exact, solutions. 

The student should now apply himself to prove this of the pro- 
cesses already described. We will take one case at length. 

Let the result of a problem be 

where an error, which we are at liberty to suppose as small as we 
please, has been committed in determining, say b and e. Or (which 
is a way of speaking more consistent with what has preceded), let 
h and e be involved in some equations (such as bb — 2 = 0, 
eee — 3 ^ 0) which only admit quamproximc solutions. Let b' and 
h" be approximate values of b, the first too small, the second too 
great ; let ef and ef' be approximate values of e. Then the substitution 
of the approximate values of the preceding expression gives 

^*: and t±^. 

To compare these expressions, subtract the second from the first, 
which gives 

fl+i' _ a + y' _ {ac-\-a^' -\'Cb' ^-b 'e")-^(ac-\-ae' -i^cb" ->te'b") 

_ a(^^^e')-'c{b"-'b') -f (6V^~e^n 
cc-f-(e'4-e")c + eV' 

And since, page 94, e" may be brought as near to e' as we please, and 
6" to 6', it follows that c" — e' and b" — 6' may be made. as small as 
we please ; whence a(e" — e') and c{b" — b') may be made as small 
^ we please, page 96, note. And so may b'e" — efb'^ for it will be 
found to be the same as 

Hence the numerator of the preceding fraction can be made as small 
as Yte please, because each of its terms can be made as small as we 

K 2 


[law op continuity.] 

please. But since d' is always greater than e', the denominator is 
(page 91, Lemma 5) always greater than 

Here then is a fraction of which the numerator can be made as 
small as we please, but not the denominator; cousequently, the 
fraction can be made as small as we please. That is, the fractions 

; , and ; ,, , different values attributed to ; 
c '\- e c 4- « c + e 

can be brought as near as we please, or be made to differ as little as 
we please. Here, by the same extension of language as before, the 
last mentioned fraction is said to have a definite value, to which we 
approximate by substituting approximate values of h and e. 

The Um3 of continuity which is assumed to exist in algebraical 
expressions, and which must be proved by the student in a sufficient 
number of particular cases, consists in the following theorem : 

General Theorem. 

Let there be an algebraical 
expression P, which contains x, 
and let the substitution of a 
instead of x give to that ex- 
pression the value p. 

Let the substitution of a + m 
instead of x give to the expres- 
sion P the value g. 

Then if a and a-\-m may be 
made as nearly equal as we 
please; that is, if m may be 
made as small as we please, it 
will always follow that p and q 
may be made to differ as little 

Particular Case. 

P = x + x* 
p =z a-^a* 

q =s (a + »i) + (a -f- w»)' 
= fl+o«+(14-2a)m-fm« 
= V +(lH-2a)m-|-m' 
From the last, 

q — p ^ (1 + 2a)m -j-m' 
each term of which, if m may 
be made as small as we please, 
may be made as small as we 

as we please. 

The only question about which any doubt can arise, as regards 
expressions hitherto obtained, relates to the values, real or approxi- 
mate, as the case may be, of n/j7, >/jr, &c. If m can be made as 
small as we please, can the approximate values of v a and w a-^m 
be made as near as we please ? To answer this question, we must 
premise a theorem which will also be useful in many other places. 



[law op continuity.] 103 

It is as follows : 
afi—'f — (ar^y)(ar+y) 

a:"— y" = {x—y)(af^^+af"^y + .... -|-a;y»-*+y**-0 
Multiplication will make any one of these obvious ; for instance, 



or* + x^y +a:^y^ 4- xy^ 
— x^y^x^y^ — xy^ — y* 

a?* + 4- + -/ 

(Observe that in this, the first multiplication which has been given 
at length since Chapter II., we have employed the second of the 
methods in the Introduction, and shall do so in future.) 

If we examine the series of expressions 

x+y^ afi+xy+y^y x^+x^y + xy^+y^^ &c. 

we shall see that each of them may be made by multiplying the pre- 
ceding by y, and adding a uew power of x. Thus, 

x^ + xy+y^ = oi^+y(x+y) 
x^+x^y+xy^+y^ = x^+y(^x^ + xy-]-y'^) &c. 

So that, if we call* these expressions P,, P,, P3, &c. (we shall seldom 
use large letters except as the abbreviations of other expressions), we 

P, = x^+yF^, P3 = a;3+j^P2, P4 = ^^+y^3y &c. 
or generally Pn = a:**+yPn-i 

But we shall also find that the same expressions may be made by 
multiplying by x, and adding powers of ^, as follows : 

sfi-^xy-^-y^ = y^ + x{x+y) 
x^-^-af^y+xy^+y^ = y^ + x{x^ + xy +y^) &c. 

* The figures underwritten most not he confounded with exponents. 
They are used as the accents in page 38, and are read P one, P two, 
P tlvree, &c. 


[law of continuity.] 


Pj = y«+xPi P, = y»+a:Ps Ft=y*+xV»kc. 

or generally, 

P» = y"+JjP«-i 

The theorem we are now upon may be thus expressed : 

af'^y*' = (a;— y)Pn-i 
and it may be proved as follows: 

General Theorem. 

(— ) = j:«— y»— (j— y)Pn-i 
j:»«— y»=s(j: — y) Pn- 1 

Particular Case. 
P, = j:*+yP3 
(-) = a:^-y-(x-y)P, 

Let us now examine n/iO and v 10 4-»»> where m may be made 
as small as we please. Let y and x be approximate values of the 
first and second, so that (pages 94, 95), 

a*^ = (10 -f wi) 4- r ■[ where t; and to may be made 
y = 10 +W 



as small as we please, 
(a: — y) (a?« + xy +y^) = w + i? — ti? 

m -\-v — to 

Now X and y are both greater than 2, since 2' := 8 (less than 10) : 
therefore the denominator of the preceding fraction must be greater 
than 12, while the numerator can be made as small as we please. 
Hence the fraction (which is ^ x — y) can be made as small as we 
please, or x and y as nearly equal as we please. But x and y are the 
approximate values of v 10 + wi and v 10. 

The student may try to prove the following theorem : 

3^ — if' = (x-^y) (x— y) 

a:* — y* = (x -fy) (^oc^ — afly -{• x y^ — y^) 

ofi — y^ = (a; +y) {s^^af'y -{-x^y^-^x^y^ +^y*-~^) 

&c. &c. 
x3 -f y = (ar +^) (^ — xy -f y) 
x^ +y^ = {x +y) (a?* — a;3y ^x^y^ — xy^ +y*) 

&c. &c. 

We shall now proceed with the general theory of exponents.] 

ON BXP0KBNT8. 106 

To 9^€sdse a power qfxto any power, multiply the exponents of the 
tiDO po'^^^crs together for an exponent ; for instance, 

(ofl' == a^x*= or" for (a:^)* = a?.x^.x^.x^ = a^+a+a+s 

In ^ similar manner, 

-^ poioer o/* a product u the product of the powert of the f acton. 

{ahcy = abc.abc.abc = aaahbhccc = a^fc^c* 

T ) == TT 5 

for the first is t X r X r> or ttt* which is the second (page 91). 
bob boo 

A root of a root is that root whose index is the product of the 
inHees of the first mentioned roots. Thus, the fourth root of the cube 
(third) root is the twelfth root. To prove this, let the fourth root of 
the third root of x be called y. Then, 

or y = V ^* but y is also y v a; 

We have shewn, page 89, that x can have but one arithmetical 
cube root, or twelfth root ; and that the cube root can have but one 
arithmetical fourth root. Hence the above process is conclusive ; it 
shews that a fourth root of a cube root of jt is a twelfth root of x ; 
and there is but one arithmetical root of each kind. But when we 
come to consider all the algebraical symbols which are roots of jt, 
both those which have arithmetical meaning and those which have 
not, the student must remember that the preceding does not prove 
that every fourth root of every third root of jr is a twelfth root of x. 
This may be the case, but it is not yet proved. 

In a similar way it may be proved that 

y y/^ = ^x ; y \/x = \/x = y y/x 


The root of a product » found by multiplying together the roots of 
the factors. Thus, 

y/abl ^s/axy/lxs/c (A) 

for these two have the same fourth power, namely, ahc. For by 
definition (page 87), 

(^y/abcf = ahc 
and by page 105, 

That is, each side of the equation (A) is a fourth root of abc. But 
a be has but one arithmetical fourth root; consequently, each side of 
(A) must be that root; and therefore the two sides are equal. 
Similarly it may be proved that 

\/abc = \/a X \/b x >/c s/aV- = \/a x \/^ = h\/a 
t/^;«F? = v/^xv/^x v/c^ t/32 = t/l6xt/2 = 2t/2 

- =s 5-7=; 

for both of these will be found to have the same cube, namely, t-* 

If a power qfxbe raised, and a root of that power extracted^ the 
result is not altered if the order of the operations be changed. That is, 

1^ is the same as {^i}* 
To prove this, observe that (page 105), 
\/^ = \/xxxx = \/xx\/xx\/xx\/x = {v^^l* 

Similarly, ^/7 = (V^)' ; V^ = (v/i)' 

In the expression ^x^,ifiL and b be both multiplied or divided by 
the same number y the value of the expression is not altered. That is. 

For it has been shewn that 

"X/^^ = y V^ which is y V(W = V^ 
because 'V (p^T = !^ 
Similarly, V^ =V^* =V^ = 'X/^ 

OK BXP0VBKT8. 107 

To extract a root of a power, divide the exponent of the power by 
the index of the root, if that diviiion be possible without fractions. 

Thus, V?« = x?z=: X*. For x» = (*»)*; therefore V?« = x». Simi- 
larly, V?« = X*, n/j* = jr'* ; and so on. 

When the expoDent of the power is not divisible by the index of 
the root, as in the case of v jr^, we have (at least as yet) no alge- 
braical mode of operation by which to reduce v^jt^ to any form in 
^hich the sign V does not appear.* It only remains, therefore, 
to 6od what number or fraction x stands for, and then f to calculate 
'^x'' or JT* Vx by the rules of arithmetic. 
The only question that remains is about a mode of representing 

^x^; and this question we have anticipated in page 86, where we 
have found that it would be highly convenient in one respect to let 

J^t ^} &c. stand for >/jr9 V jr, &c. But we stopped our course there, 
because we had no direct reason to know that all the complicated 
relations of roots might be obtained from that notation, K;i^Aou^ the 
necessity of applying rules to fractional exponents different from those 
which are applied to common fractions, or of treating the fractional 
exponents by rules different from those which apply to whole exponents. 
We write in opposite columns instances of the rules which we have 
proved in the case of whole exponents, and those about which we 

• We may proceed as follows. Since x^ — a'.ar, we have >/x7 si 

y/x^x a: =: Va* x >/x = x'>/x ; in which, however, the symbol V 
still remains. 

t There is a certain distinction to he drawn between the processes of 
algebra and those of arithmetic. We cannot be properly said to Jind 
results in algebra, but only to put them into the form in which they can 
be most easily found by arithmetic. For instance, " What is the sum of 
a and b V Of this question a + 6 is not properly a solution, but a repre- 
sentation, and the proper answer to the question must be deferred till 
we know what numbers a and b stand for. But in the question, " what is 
the sum of 8 a and 5 a V* we can go one step further ; for though 8 a + 5 a 
is an algebraical representation, it is not the most simple one which the 
language of the science admits. The latter is 13 a ; but we are still 
without the answer until we know what a stands for. When we come 
to the step at which we must pass to arithmetic to get any nearer the 
answer, we shall therefore say we have arrived at the ultimate algebraical 

form. Tbusa + 6is an ultimate form; 8a + 5a is not. Again wx^, or 

at most a^ V X, is an ultimate form ; v«*' is not. 



wish to inquire as to fractional exponents; the question being, are 
the theorems in the second column true, upon the supposition that 

a:** represents 

ofixa^ ^ a;^+* = 



{sfif = a:^^* = x^ 

\/^ s=z sfi ss: 3fi 

Jxx^=x^^^ = x'l -f 

In the first place we observe that we come by the extension from 

the last rule in the same way as by others. We have found that when 


h is divisible by a, ^ is a correct representation of ^/x^. But when 


b is not divisible by a, j^ has no meaning. We give it a meaning ; 
that is, we say, let it still represent Vj*, whatever that may be. We 
now proceed to investigate the rules which this new symbol requires. 
The first column is the general case, the second a particular case. 
The references ( ) are to the pages in which the rules are contained. 

m p 

What is x^'x af'l 



(106) which 

iP"* means \/ ^ 


(106) which = V ^'^ 

m p 

Therefore a;"xx« 

(106) ='*\/a;"*« X af*P 

(84) ='*\/a:'"« + ''^ 
Which is represented by 

mq + np 



But ""l + np ^m p_ 
nq n q 

m p ^4.£ 
Therefore :r''xa:« = a;'' « 

What is a?' X a:* ? 

x^ means y/a^ 

(106) which = v/^ 

X* means \/x 
(106) which = ^x^ 

Therefore x^ X x* 

(106) =\/x* X X* 

(84) = \/^ 

Which is represented by 




7 _ 2 t 
6 3 "^ 2 

Therefore X^X X^=^X 

4_ J + 4 



The next rule will be more briefly deduced, and without 



Whatis a^ -h 3fit 





But ^9—np _m p 
nq n q 

m p m p 

Therefore af^ -^ ofi = o^ « 

Whatis (a;^)« ? 
'^is means \ | \/^ }' 

or (106) V V(^^ 
w (105) '*v/!x^ 

What is a;' -r- a;* ? 
This is ^Hfi-T-s/x 








Therefore a;'-f- x^= x^ 





mp m p 

nq n q 


(ar«)« = 

What is (a;*)* ? 
This means V Iv^} 

or (106) 
or (105) 



15 — 5 ^ 3 

Therefore {xi)i^xi^i 

The last process contains the answer to both the third and fourth 
inquiries in page 108. 

By looking at page 86, and remembering that the fundamental 
rules there used have now been proved to be applicable to fractional 
exponents, the meaning of, and rules relating to, negative fractional 
exponents may be established. Thus, 

a:""» stands for —- or r-r^s and also for 4/ ( 1^ 

We shall now proceed to inquire how many algebraical roots 
there may be, and of what kind, in the case of the square, cube, and 





fourth roots. To go further would be difficult for the beginner at 

First, as to the square root. It is evident that +1 and — 1 are 
both square roots of -\-if and -j-a and — a both square roots of 
+ aa. For 

— Ix —1 = 4-1 —ax —a = +aa 

+ lx +1 = +1 -fax +a = 4-aa 

The question now is, can there be more than two square roots 
to +1? Let X be any square root of -f-1, then xx must (by the 
definition of the term square root) =: 1, or xx — 1 = 0. But 
jtj: — 1 = (jT +1) (jT — 1) ; therefore {x +1) (jt— 1) = 0. Therefore,* 
either x-^\ or j: — 1 is ^ 0. To jt+I =0, the only answer is 
X = — 1 ; to X — 1 := 0, the only answer is x=i -f-1 ; therefore +1 
and — 1 are the only square roots of 1 . The same process may be 
applied to xx = aa, or {x — a) (j: -fa) ^ 0. 

Therefore a negative quantity can have no square root which is 
either a positive or a negative quantity ; for either of these, multiplied 
by itself, is a positive quantity. We will not therefore say that *s/ — 1 
is no " quantity," because we have agreed to give that name to every 
symbol which results from the rules of calculation. But *s/ — 1 
(whenever it occurs) will be what — 1 was in page 47, the evidence 
of some misconception of a problem, for which the problem must 
be examined, and altered, extended, or abandoned, as may be found 
necessary. But if we look at the steps by which we established the 
meaning of — 1, we shall find them to be as follows if 

1. We met with such combinations of symbols as 3 — 4, &c. 
proposing operations which contradicted the meaning which the 
symbols 3, — , 4, then had. 

2. We examined the problems which gave rise to such combina- 
tions, and found out how to make the correction without repeating 
the process ; so ascertaining what was to be understood by such 
expressions as 3 — 4, &c. that we could either predict their appear- 
ance, or explain them when they appeared. 

3. We examined what would arise from applying common rules 

♦ When a product aft = 0, either a or 6 = 0, for if both have a value, 
the product, by common rules, has a value. 

f The student may make bis understanding what immediately follows 
a test of bis having understood all that precedes. 


to 3— 4, &c.y and what would be the effect of deferring the correction 
of the misconception till any later stage of the process. 

4. From all that preceded, we extended the meaning of the terms 
commonly used, and in such a manner, that what were till then simply 
the results of misconception, became recognised symbols with a definite 
meaning, and used with demonstrated rules, not differing in practice 
from those with which they were used in their limited signification. 

5. And we found that in all cases in which the result produced 
was simply arithmetical (that is, consistent and intelligible when the 
terms had their limited significations), that no error was lefl in the 
lesult by the unintelligible character (with the limited meanings) of 
the preceding steps ; but that the result was the same as it would have 
been if we had retraced our steps and made each step arithmetical. 

Thus we observed that +, which before a number, 3, means 
addition, before — 3 is equivalent to a direction to annex — 3 to the 
preceding part of the expression; and that though we could not dis- 
pense with the extended meaning of + and — in -f (^6) — (+4), 
yet that -|-6 + 3 admitted of arithmetical interpretation, even though 
it were the result of several of the more extended forms of operation. 

From this we are at liberty to conjecture^ that a further extension 
of the meaning of -)- and — might, by precisely the same train of 
operations, give a rational method of using and interpreting v^ — 1, 
>/— 5, &c., which at present are wholly unmeaning and contradic- 
tory. We say conjecture, because it by no means follows that a 
method of the success of which we have only one instance, will be 
universally applicable. Such a method has been given, but it is not 
our intention to explain it here. We shall simply shew that a field 
has been lefl in which the explanation may be looked for. 


If we suppose a person to set out from A, and stop at B, stopping 
first at some other point, and always keeping in the line A B or its 
continuation ; and if we suppose distance measured towards the right 
to be positive, and towards the left negative, we can by our rules 
ascertain tha distance -f-AB through which he goes altogether from 
his first position, as follows : 


1. Ifhestopat C (-I-AC) +(+CB) = +AB 

2. Ifhestopat C (-f AC) + (-C'B) = 4-AB 

3. Ifhestopat C" (-AC") + ( + C'B) = +AB 

Now observe that if he should leave the line AB or its con- 
tinuation, if, for instance, he should go through AP, PB, we have no. 
symbols so connected with AP, &c. that (let If simply denote that 
some yet uninvented symbol is to be in its place) 

f (1[AP)1[(irPB) = +AB 

Therefore we find that in the application of algebra we may yet 
have new symbols to employ, and we also fall upon unexplained 
symbols such as v — 1, &c. May not such extensions be made as 
will make v — 1, &c , with an extended use of + and — , supply the 
place of the yet-to-be-invented symbols ? This is for the student of 
this work a point for conjecture only, but it will make the following 
course advisable. 

1. Apply the rules of algebra to such expressions as n/ — 1, &c. 
merely to see what will come of using them, without placing any 
confidence in the results, or at least more than the experience of a 
great number shall render unavoidable. 

2. Whenever, in the course of a process, it appears. that such 
expressions as v — 1, &c. have disappeared, examine the result and 
see whether it is true. 

We now pass to the cube root. Let x be any one of the cube 
roots of 1. Then x'ssl or jr»— 1 =0 ic«— 1 =(.r— 1)(j?«-|-j:-|-1)=0 
(see page 103, and make y ^ 1) ; 
therefore either x— 1=0 or a;*-fa:4-l = 
The first gives j: = 1, and 1 is evidently a cube root of 1 . The second 
cannot be solved till the student has read the next chapter ; but its 
solutions (for it has two) both contain the yet unexplained symbol 
>/ — a. They are 

r and 

2 2 

and we shall shew of the first of these (leaving the second to the 
student), 1. That it does satisfy the equation ar^'\'X -^-ly if common 
rules be considered as applicable to n/— 3; 2. That it is a cube 
root of 1, on the same supposition. 



Ifar= a:«= ^ 

_ 1— 2^^^ +(—3) _ —2 — 2^/^ _ — 1— Vila 
4 4 "^ 2 


Again ar s= a;* x a; = „ X 5 

^ ^-i?-W^^* _ 1-(-3) _ 1^ ^ 
4 4 4 

. — 1— •>/:^ — i+V=l 

l*t !> = i 9 = 2 

Then the three cube roots of aaa or a' must be a, pa, and ^a. 
For first aXax a=sa^. 

paxpakpa = p^a^ = a^ because p^ — i 
qaxqaxqa = g'^a* = a* because q^ = I 

and the same might be deduced from the equation jr'— a'= 0, so 
soon as we know how to solve x^-^-ax-^-a^ ^=0. 

Let X be one of the fourth roots of 1. Then we have j:* = 1, or 
«♦ — 1 = 0; that is, 

(a;«-«l)(a:«4-l) =s and either a;«— 1 = or x^ + 1 = 

the solutions of a:' — 1 ^ are — 1 anfl +1, as before ; and the solu- 
tions of jr*+l =: are either x = -f^ — 1 or j: = — v — 1. There- 
fore 1 has four fourth roots, -|-1, — 1, -f- v^— ^, and — v^ — 1. This 
will be found true upon the application of cotaoAaon rules; for instance, 

(vc:t)»=-i (>/i:T)«=(N/iir/.vc:T=:->/=r 

The only use which we can logically make of such expressions as 
V— a?, previous to any further inquiry, is the following : Let a, b, c, 
and d, be positive or negative quantities. Then 




cannot be true unless a=zc and b=zd. For suppose a ^ c di e, that 
is, is difTerent from c, then 

c±e + b\^^^ = c + dV — x l/ZT^— ^1 

that is, V — X is a positive or negative quantity, which is absurd. 
Consequently (± e) H- (c' — b) cannot be a common algebraical quan- 
tity. Now if c = 0, and d is not equal to 6, we have >/-«-jr = 
-f- (d — 6) = 0, which is not true ; if c be finite and £^ = 6, we have 
>/ — X ^ :t « -T- 0, which does not agree with page 25, since no 
quantity multiplied by itself is negative because it is great, or nearer 
to negative the greater the quantity becomes. The only supposition 
remaining is that e = and d — 6 = 0, that is, a = c and d = b; 
and the only form at which we have here arrived for >/ — x is 

--. But this simply indicates that the equation from which it is 

derived is always true ; which is the case (so far as such an equation 
can yet be said to be true at all), when a =z c and b =^ d. Observe 
that we do not lay much stress on the preceding ; it only proves that 
the equation cannot be true unless a =i c and b =zd; but it may be 
matter of dispute as yet whether the above is true if a = c and bz=d. 
For we may as yet reasonably refuse our assent even to the equation 
X j^ X unless x represent magnitude of some sort ; we may say that 
ideas are contained in the meaning of the sign = which do not apply 
except to magnitudes. But if any one should say this, we refer him 
to the extended definition of =, page 62, though the beginner must 
recollect that he never will comprehend the force of that definition as 
applied to >/ — x, &c. .^til he has more experience of such symbols. 
But there are algebnd^l quantities analogous to the preceding to 
which we must now/[^«t attention. They are such as >/3, V5, 
V2, &c. which ca^j^rbe exactly found, but only approximately, 
pages 92, &c. w/j^rfJ'^S^t 

cannot be true (if a, 6, c, and d be numbers or fractions) unles^; a z=c 
and b ^ d. The same reasoning applies, word for word, substituting 
n/s for >/ — X, the absurdity dieduced being that we cannot have 

1/3 = -^ 
"^ d^b 

where rf, 6, e, are whole numbers or fractions. 


In a similar way it may be proved that if all the letters stand for 
definite numbers or fractions, and (x and y not being square numbers 

a + 6 Vx = c + dVy (A) 

Then a must = c, and therefore b 's/x ^ d Vy. For if not, let 
(i = c-±.ei substitute, and ( — ) c 

± e + bVx = dVy 
Square both sides, which gives 

(± ef + 2(±e)J \^x + (J V^i)' = (rf Vy)* 
or e«±2eiV^+i^ar = d«y 

therefore ^^x = ^"". /^^ 

tbat is, vjr, which cannot be expressed in a definite fraction, is so 
expressed, which is absurd. The only way of making the equation 
A possible is, therefore, by making a = c and b s/ x = d s/y. 

This principle may, in certain cases, be applied to the extraction 
of the square roots of such quantities as 4-f-2>/3, 21+4 >/5, &c. 
Take such a quantity, say 2 + Vz, and square it. 

(2 + Viy = 22+ 2 X 21/7 + {VlJ 
=:4 + 4V^7 4.7= 11 + 41/7 

Now, suppose such a quantity as 11 +4"/ 7 to be given; how 
are we to find out, 1. That it has a square root of the same form;* 

2. That that square root is 2 + >/?? As follows : if 11 +4 n/z have 
a square root of the same form, let it be x + ^y, so that 

V 11 + 4 1/7 = a: + Vy ^.^(square both sides) 

11+41/7 = a;2+2a:l/y + (V^' 
= a;« + y + 2a; Vy 
Hence x'^-\-y = 11 and 2a: Vy = ^Vl 

Therefore (-) a:^ — 2a; Vy +y = 11- 4 V^ 

• Observe that there is no essential difference of form between 
ll + 4«s/7 and x + n/^ for 4^7 = ^(4)^x7 ssV^Tli; whence 11 + 4n/7 

*ii+>/ nir 


but the first side is the square of jr— -V^, or 

(a:-. V^« = 1 1 -4 V^ lliat is, x- Vy = \/\\ZaV^ 

But a: + Vy= \/ll+4t/7 

Multiply the last two equations together, which gives 

or a:«-y=\/(n+4l/7)(ll-4V^)= 1/121 -112 = 3 
Buta:*+y= 11 

(+) 2a:«=14 a:2 = 7 a; = l/7 

(-) 2y = 8 y = 4 V^ = 2 

Therefore 1^11 ^-^v'? or r + \/y is '^7 + 2, as we saw in the 
method by which 11 + Wj was obtained. 

We shall apply the same process to the formation of ^a + 6>/c. 
Let this be x -|- ^/y. 
Then a + bVc = (a; + Vy)* 

= a:* + y + 2a: Vy 
Therefore a = ar^+y and Jl/c = 2xVy 

Therefore a— 6 V^ = a:2+ y — 2a: V^y = (a;— V^)^ 
or x— Vy = V a — b V~c 

But a: + l/y = V^a + fc V^ 

(x) a:*-y = \/(a-6l/cXa + 6l/c) = V^a2-i«c 
But a:* + y = a 

( + ) 2a:« = flf + Va2— l^c 

X = v/ja + Jl/a2— 62c 
(— ) 2y = a—Va^^lf'c 

and this is the square root of a + 6n/ c. 


Verification. Let } a + J \^a^—b^c be called p 
Let ia-^i^a^—b^c be called ? 
pq = (Ja)«-(jV'a«-6«c)' = ia«-4(a«-J«c) 
= ia«— ia« + 4J«c = 46«c; therefore Vp^ = Jjv7. 
According to the preceding theorem, 

But i^ + ? = ia4-i« = a+2Vj5^= Jl^c 

Therefore |> + y + 2 ^/pq is a -f * '^c ; which shews the preceding 
theorem to be true. 

This theorem is of little practical value, but is very good exercise 
in the use of such expressions as 6vc, &c. It is a simplification 
only when a* — b^c has a real square root; otherwise, it is the reverse, 

for a square root of a square root occurs only once in *^ a-^b Vc, but 
twice in the value found for it. Thus it simplifies the first of the 
succeeding expressions, but not the second, though both are equally 

1/13+21/55 = vTo + v^ 

\/T3+2Vlf = v/ V + Jl/IF 4- v/^'-JV^ 

Anomaly. Apply the preceding result to a case in which 6^ c is 
greater than a*, or a' — 6*c a negative quantity. For example, to 

2 + \/8(a = 2 6 = 1 c = 8), 

\/2+v^8 = v/r+ji/^+\/i3ji7z% .... (A) 

Is the square root of 2 -j- '^S, therefore, of the same unexplained 
character as V — 1, &c.? Certainly not: for it may be found quam 
proximty by the rules of arithmetic, lying somewhere between the 
square roots of 2 + ^^4 and 2 + n/q^ or between the square roots of 
4 and 5. 

Are we then to conclude that the expression (A) is in reality 
arithmetical ? On this we must observe, that a really arithmetical 
expression may, by rules only, be made to appear impossible. For 



X + y = (ji:-\-cV—i) + {j/—cV—\) 
3* + if = x«-(-j(r2j= (a:)*-(yV^)* 

To investigate the expression (A) further, extract the square root 
of each term by the rule. 

Let a = li=Jc=— 4 

The second sides of the two preceding equations still contain the 
square root of a negative quantity ; because, since 1 is less than 
^2, i is less than 4^2, or J — ^^/2 is negative. Add the two last: 

v/l + JV^ + \/l-Jt/Z4 = 2\/i + iV^2 

But this is only the quantity with which we started in another form ; 
for it is 

V^4(JH-JV^2) or \/2+2vl or \/2 + vTx2 
We have, then, by following an applicable rule, simply committed 
the inadvertence corresponding to the intentional alteration made in 

X -\-y above ; and y2-\- Vs, or generally ^a -f b >/7, if a' be less 
than b^c, cannot be expressed in a result of the form 

unless 9 be a negative quantity. 

The extraction of l^a -f-6 Vc is of greater difficulty and less use. 
We shall, therefore, omit it. 

Exercises. Verify the following theorems : 

1. The three algebraical cube roots of — 1, are — 1 and the 
solutions of the equation x^ — x + l =0, which are i + i >/— 3 and 

2. The four algebraical fourth roots of — 1 are 

J \/2(l 4- V^ITT) J |/2(1 - V'^i) 

jV^2(-1+VI:T) j\/2(-1-.VZT) 

3. The eight eighth roots of 1 are 

±1, ±v^^, ±jt/2(i+vcn), ±ji/2(i-vcri) 


Give the reason why this set includes all the values of >/— 1 ? 

Having given a quantity which contains radical* terms of the 
second degree (that is, square roots) to find a multiplier such that 
the products shall be free from radicals. 

1. A simple radical term, such as v 3. Here >/3 is the multi- 
plier, or aVs^ where a is rational ;\ for v 3 x fl'^S = 3a, which is 

2. A binomial, one or both terms of which are radical, such as 
V3H-V2. Since (a'j-b)(a — 6) = a' — b^, if a and b are simple 
radical terms, a' — 6' is rational; therefore a — b is the multiplier for 
a + 6, and vice versd. For instance, V^3 + >/2 multiplied by 
\/3— s/2 gives 3—2, or 1 ; 2>/3— i V? multiplied by 2 \/3 +i V? 
gives 4 X 3— i X 7, or 10^. 

3. A trinomial, containing two or three radical terms; such as 
V3 + >/5— Vt or 's/a+N/^+v'c. Multiply by V'«+\/6— \/c, 
which gives (\^a + v 6)^ — W cY or a •\- 2 '^ ab •\- b — c or 
a + ft — c + 2>/a6. Multiply now by a-\-b — c — 2*^ab; which 
gives (a + 6 — r)«— (2 '^'abf or (a + 6 — c)^— 4 a 6. Therefore 
the multiplier required is the product ofva+V6 — vc and 
a + 6+c — 2>/a6. 

(1 + 21/15) (1- 2 1/T5) = 1-60 = -59 

It will seldom or never be requisite to consider more than three 

The preceding can be applied to find the value of a fraction which 
has radicals in the denominator. Thus to find l-f-Cv's +l) instead 
of extracting the root of three and forming the denominator, multiply 
both numerator and denominator by v 3 — 1 , which gives 

_J ^3"^ _ >/3-l _ , /w^ , 

v/3 + 1 " (V3-M)(\/3-l) "" "Sirr " 5^'''^-^ 

the second side of the equation is evidently the more easily found : 

• Radix, Latin for root ; radical quantities, those which contain roots, 
t Rational, a term used in algebra, meaning, free from radicals ; for 

inBtance, a common number or fraction. Thus 2 is rational, and also v 4) 

though in a radical form. But V 3 is irrational. 



Ve— V5 (V6— V5)(V6+V5) 

The only case of higher roots which is worth consideration is 
where there is a simple radical term^ such as v 8, a V b» The mul- 
tiplier in the first case is (Va)*, or 8^ for 8* X 8* = 8'= 8. Thus 
the following results are obtained : 

Va V2.V3.V3 Via V3 V48 

vi - 


3 ' V2 "^ 2 


or ^ =. 

n— 1 w—\ 

1 n-l 1 n-1 
«" X 6 ** o^ft » 
1 n-l ^ 6 

6" X6 " 


q n J, w 

2 2 V9 V64 

1 1 — 

V3V4 ^ 12 

The following miscellaneous examples on all the matters con- 
tained in this chapter should be carefully verified. 

a X J"* X a"^ X i' = «"*&"* = — r 

m*— m m 

-ybe * 


• \/a = a^j v/a^ = a^ (never used, but might be, why ?) 
n/i60 =n/i6x10=4n/'io; VT60=V8x20 =2^20 
\/3332 ssHn/Tt V3348 = 3 Vl24 V32 = 2V2 

"V 7 f ■" \/2T "V ^ f v^^ 
a : t^aft : : V^ab : b a^ : a^ :: a"* : a"^ 
1 firi/T 4/"^ 1 1 + ^^^ 

va+jr-|-va — j: 2j:\ / 

N/tf+J + v^a— -g 2a + 2'\/a»— x« a /«« T 

/ / = ^-4-V 1 v/ 

va+x — vfl — jr 2j: x ^ x^ 

^dr—x^ = av 1 — ^ = ^V ~2 — 1 == \/ax\ - — ~ 
v/a4-^ = V «V 1 + - = \/x\ ^ + 1 = \/ax\ - + - 


^ ^ X ^ ^ ^ a or 

a-\-h'J — 1 aj: + 6yH-(6j: — ay)s/ — 1 

(H-VZIl) = \/ZT(l— V— 1) \/ir7 = \/7VlIT 

>/— .4 = 2n/^ V "-J = i>/2V^ 

. (a - 6) = (a* - M) (a* + i*) ^ (a* - 6*) («' + a* i* + 6^ 





=z \a ^o )\a +a b -j-a 0+0/ 

a +6 = \a + ^ ;\a —a 6 +a b —a o 4-^ / 

(a'+a* + l) =a*+2a + 3a'+2a* + l 

\a } ^ a \ab c / = a c 

[N.B. It is usual to call quantities of the forms aV — l or v — a*, 
a -jr V — 6, orfl + v6v — 1, impossible quantities. This they are. 
at pre^iient, as not having received any interpretation ; in the same 
manner 10 — 14 was impossible in Chapter I. But, considering 
that they will in due time (if the student proceed so far) receive 
their interpretation, though not in this work, we shall call them 
purely sj/mbolical. AH the phrases of algebra are symbolical; 
but all which contain a letter or numeral, which we have yet met 
with, have an interpretation connected with numbers, making them 

representatives of magnitude. But v — 1 has received no such in- 
terpretation ; it is therefore a pure symbol, as much so as -{- or — , 
or more so, inasmucfr as ^et it indicates neither magnitude nor 
operation. Hence, in performing operations with^pure symbols, we 
can be guided only by experience, or where that foils, we have new 
conventions to make. Since it is understood that such symbols will 
finally be rejected altogether, unless an interpretation present itself 
which brings them under the dominion of common rules, these rules 
only should be applied to them. The only case which requires 

notice is that of forming the symbol which is to represent >/—= a X 
\/ — b. This, by comnaon rules, may be either v — ax — ^ or 
^/ab, or y/a X "^ — 1 multiplied by '>/b X '^ — 1 or ^/ab X — 1 
that is -^^ab. For reasons hereafter to appear, let the student 
always take the latter, that is, let 

x/ZTa X V~b be « V^ not 4. |/^] 

Having two symbols to indicate the nth root of a, Damely^ 
_ 1 

Vaanda«, we shall employ the fii*st in the simple arithmetical 

sense, and the second to denote any one of the algebraical roots, 

that is, any one we please, unless some particular root be specified. 


Thus >/4 is 2, without any reference to sign ; but (4)* may be either 
-h 2 or — 2. Thus Va is the cube root found in arithmetic, while 

(fl) is either V a, ■ va, or Va 

(fl)i Stands for Vfl, — Vo, V — l.\/fl, and — V— 1 Va 

Similarly a-j-^ has two values, namely, either a-^-'vb or a — n/6. 
Hence v 2 (when b is positive) always stands for a positive 
arithmetical quantity. Suppose we wish to represent merely the 
MHmertcal value of b, without reference to its sign ; we may abbre- 
viate the following sentence : '^ the number contained in b, taken 
positively, whether b be positive or negative," by VS*, which* is 

the same thing. Thus b = ±'^b* according as 6 is positive or 

We shall now proceed to the general consideration of expressions 
of the second degree. 

* I have seen a + '9 or what is here signified by a 4- (a^)^> denoted 

by a -f* V a', in which the ambiguity of sign was referred tov . But 
this was only in one place, and though the want of some express 
atipulation as to the distiuction between radical signs and fractional 
exponents, has led to some variety of usage, I think the conventions in 
the text will best agree with the majority of writers. At least, though 
I do not pretend to have made any research expressly on this point, 

:dl>/a is very familiar to my eye, and dt. a^ not at all so. 





The view taken in the First Chapter of equations of the first degree 
simply amounted to their numerical solution; that is, having given 
two expressions not higher tlian the first degree with respect to jr, 
required that value of x which will make the two expressions equal. 
We there saw that all equations of the first degree could be reduced 
to others of the form ax ^ b; thus, in page 3, we reduced 

to 13a; = 12 


It is most convenient to bring all the terms of an equation on one 
side ; thus ax — 6^0 is in a more convenient form for investigation 
than ax = b. In this manner the whole theory of equations is 
considered as involving two fundamental inquiries. The first is; — 
Having given an algebraical expression which contains x, required, 
that value, or those values of x, which make the expression vanish, that 
is, become equal to 0. 

We here (in compliance with common custom) use the term root, 
in a manner different from its use in the last chapter. Every value 
of X which makes an expression containing x equal to 0, or, as the 
phrase is, makes it vanish, is called * a root of that expression. Thus, 

* This might be considered an extension of the former meaning, as 
follows. Let root have the meaning above assigned, then the square 
root of 3 (of last chapter) is the root (as here defined) of x^— 3, the cube 
root of 3 is the root of jr^— 3 ; and so on. But we do not make this as 
an extension, because there is no corresponding extension for the 
correlative term power. And we must confess, that we u^ this new 
meaninjg^ of the term root with some repugnance, in spite of its shortness 
and convenience. Would not the nulUJier do as well? If we were 
inventing algebraical terms anew, we should certainly say that 7 is the 
nuUiJier of 2 j— 14, and that the latter is nullified when x ■=» 7. But it 
is not advisable to introduce words or sybmols which the student will 
not afterwards see in the best writers on the applications of mathematics. 


in page 2, we used language which we may now modify as follows : 
"The root of 2ar— 1 — 5j: + 19 is 6;" "the roots of 16 or— jr*— 48 
are 4 and 12;"" the roots of a:*— 6 j:'+ 11 jr— 6 are 1, 2, and 3.** 

The second fundamental inquiry is as follows : — Having given 
an algebraical erpression which contains x, what values of x make that 
expression po itive, what values make it negative, what values make it 
purely symbolical (See page 122) ? We shall proceed to answer these 
questions for expressions of the first degree. We first make the 
following remarks. 

1. As we wish the student to keep in mind that we consider 
yarious values of <r, and the consequences deduced from them as to 
the sign of the expression, we shall (whenever we may think it 
necessary) use some modification of this latter to denote the rout. 
Thus, instead of saying that 2d7— 14 vanishes when jr ^ 7, or that 
X = 7 is the root, we shall call the root x/, the second root, if there 
be one, x^,, and so on. 

2. We shall always suppose that the co-efficients, a, 6,&c. in such 
expressions as ax -f-6, a-ac^'\-bx + c, 8cc. are either positive or negative 
algebraical quantities; unless the contrary be specially mentioned. 

Thus, we shall treat of cases where a is 1, 2, — 1, — Jv 3, &c. but 
never, without special mention, of the case where a is v — 1, or 
1 ^ V — 3, &c. But we make no such limitation with respect to x. 

3. We shall suppose the student to be familiar with the use of 
the transformations in page 73 ; for instance, that he can make ax '\-b 
coincide with 2x — 3 by writing the latter as 2a: + ( — 3), and sup- 
posing- a := 2, 6 = — 3. 

The general form of the expression of the first degree containing 
xis ax -{- b.' For instance 

X — 5 X — 2 , . X 5 X , 2 , 

— --ar+^» 5-2-3+5 + ^ 
which is (i _ 1 + l) a; - g - I) 

which coincides with ax-\-bf by supposing 

a = 1-1+1 *=-(£-?) 

2 3^ \2 3/ 

The root of ax-\-b is readily found ; for let 

ax -|- 6 = 0, then a; = call this x\. 

' a ' 

M 2 


When ax-\-b vanishes, it may be written ax^-^-b, and x^ denotes 

. Fronj jr, = we get ax, = — b or 6 = — ax,. Write 

a ' a ' 

this value instead of 6 in ax-\-b, which then becomes ax — ax, 
or fl (j — xj. Consequently we have this theorem. If x, be the 
root ofax-^-bf then 

ax -f ft = a{x — X^ for all values of x, 

. The last is not an equation of condition (See introduction) but an 
identical equation. It implies that the two sides are absolutely the 
same, but in different forms: indeed it may be thus immediately 
traced from ax-^-b without any alteration except of form, 

ax -f ft = a\x -{■•-] = a ]x — ^ j> = a{x — x) 

because it has been laid down that x, stands for . 


The preceding will seem an unnecessary repetition, but, on 
coming to expressions of the second degree, the reason of it will 
be seen. 

Exercise. Point out, by inspection, the roots of the following 
expressions 3a: + J, —4a; — 3, Ja;— § 



altered ex- 


— 3 

— 4 

_ 2 





3{-(-y}. -"{-(-!)}. \{'-% 

Theorem. The exprension SL\-\-hisofthe same sign as a, when x 
« greater than the rooty and of a different sign from a when x is 
less than the root. 

For, ax -^-b is a{x — i-,); if j: be greater than x,,x-^x, is positive, 
(page 63), and a x positive quantity, retains the sign of a (page 
64) ; but if x be less than x,, x — x, is negative, and a x negative 
quantity changes the sign of a. 

Examples. 3 or -{--is positive for every value of jp greater than 


; for instance for — - : this we shall try. If j? = — - 

6 8 8 


The same is negative for every value of x less than — - ; for example, 

for jr = — -. For then 

Q , 1 O 1,113 1 

3^ + - = 3x-5+5 = 5-^=--. 

Similarly — 4x — 3 is negative (that is, the same as — 4) for every 

TRlue of X greater than — -, and positive for every value of x less 

3 12 4 

than — -: but -x — - is positive for every value of x greater than -; 

and negative for every value of x less than -. 


This is sufficient to render our notions of expressions of the first 
degree consistent with what we are now going to lay down concerning 
those of the second. 

Lemma 1. Sf x-\-y =/>-j-9> ^^cl xi/ =pq, then x is = one of 
the two, p or q, and y is = the other. 

Square the first equation, and from the result subtract the second 
multiplied by 4, as follows, 

x^ + 2xy +y^ = jp* + 2pq + q^ 

4xy = 4pq 

( — ) x^-'2xy+y^ = p^^2pq+ q^ 

The square root of the first side is either of the two, x — y or y — x; 
that of the second p — q or q — p. Extract the square root, which 
gives therefore one of the four following equations : 

x—y = p—q •••• (1) y—x = p—q •••• (2) 

^—y = 9—p "" (3) y—^ = ?—P •'•• (4) 

But x^y =sj3-f-y; which last combined with the four preceding, 
separately, gives as follows : 

with (1) or (4) X = p y =z q y^ which Was to 
with (2) or (3) X = q y = p ) be shewn. 

(This lemma will certainly seem most superfluous to the student : 
but he must recollect that though, " if j: = /? or q, and y = ^ or p, it 
is most evident that x +y = p + 9> and xy z=.p 9," yet that the con- 
verse, namely, that "ifj:+^ :^p-\-q and xy =pq, then x cannot 



be any thing but p or 9, and y cannot be any thing but q orp/* is not 
equally evident. Thus if 

it is most evident that x ^ b satisfies this equation, but by no means 
evident that nothmg but j* = 6 satisfies it. In fact x = 2a — b will 
be also found to satisfy it.) 

Lemma 2. The product of two expressions of the first degree 
[say ax + b and a'x+b'] cannot be always equal to the product of 
any other two expressions of the first degree [say cx-f e and c'x -\- e'] 
unless the two latter be made fro!H the two former by multiplying 
and dividing by a quantity independent jqf x [that is, unless ax + b = 

m(cx-|-e) and a'x-f-b's=:- (c'x -f- e') where m is independent 
of x]. For 

{ax + b){ax + b') = aa'x^ + (ab' +a'b)x + bb' .... (A) 
{cx + e){c'X'\'e) = c(fx^ + {c€^ + c'e)x-^€€^ .... (B) 

If possible, let these two developements* be always equal, what- 
ever value is given to x. That is, let 

px^ + qx + r =i p'x^ + qx-^r' 

wliere p stands for aa', p^ for cc', &c. (This is merely for abbreviation.) 
Now, these two cannot be always equal unless they are absolutely 
identical, that is, unless j? = p' q =^ q' and r = r'. This we prove 
as follows: — If the two sides of the preceding equation be always 
equal, they are equal when .7' = 1, and also when j: = 2, and also 
when j: = 3. Let t^ t^ t^f be the values of the first side of the 
equation when x is successively made 1, 2, and 3. Then tlie other 
sides will have the same values, according to our supposition ; 
that is, 

when a: = 1 p+q-^r = ^1 p +q'+r' = t^ 
when X = 2 4p'h2q-^r = t^ 4p' + 2q'-^r' = t^ 
when X = 3 9p+3q + r =^ ts 9p -^3q +r' = t^ 
Now, apply the first set of equations to find p, q, and r,^supposing 

* Developement, any expression formed by giving another expression 
a more expanded fond. 

-|- These are distinct quantities ; for the like abbreviation see 
page 103. 



^1 t^ and t^ knowD, as is done in page 80. Then apply the second 
set to find p*, g\ and r'. It is clear that, from the perfect likeness of 
the equations, p, g, and r, are found by exactly the same operations 
on the same quantities which give //, g\ and r^.' Consequently, the 
results will be the same, or we shall have p =p' g ^g' r =s r'. 
As an exercise, we give the three results of the first set, which are 

[The following proof is more simple ; but it involves the suppo- 
sition of a letter being made equal to 0, which we do not wish to use 
till after the discussion in a succeeding chapter. 

If px^ +qx -^-r = p'x^ -{-q'x + r^ always, 

it is true, among other cases, when x ^ 0; but it then is reduced to 

O+O+r = O + O + r' or r = r' 

Therefore px^ + qx +r = p'xi^ + q'x +r always 
(«-)r px^-j-qx ssp'x^+qx 

(-T-)a7 px +y ^ p'x +q' 

This must also be true when x = 0, or 

+ y = H- gr' that is q = q' 

Tlierefore px + q =z p'x + q (— )7 px = p'x or p = p] 

. It has been proved, then, that the expressions (A) and (B), in 
page 128, cannot be always equal, whatever x may be, uuless 

aa* = cc\ ab' + ab = ce -{-c'e^ and bb' = ee' 
Divide the second and third by the first, which gives 

ad '^ ad " cd ^ cd' aa' " cc" 

1/ , b e' t V h d e 

a a c c a a d c 

, ,T -1-^' ^' b e h' e ,6 d 

whence, by Lemma 1, either -:=-,,- = -, or -. = -.and- = -. 
* a d a c a' c a d 

Suppose the first. 

But ax + ft = a\x -^ -\ = a f a; +*-) 

cx + e a. , . 
= a. = -(cx + 6) 



Similarly CLX -f 6' = -;- {ex + e) 

n . • au' -. a a* i # a a' 1 

But since — -. = 1 or - x -r = It " - = iw, -7- = — 
cc • c c ^ c 'dm 

Therefore ax + b = m(cx-\-e) 

a'x + b' = -(c'ar-fO 

If the second assumption be taken, let the student shew that a similar 
result will be obtained. 

To proceed with the numerical solution of equations of the second 
degree, we shall take the most simple algebraical form, ax^-^bx-^-Cy 
to which any other expression of the second degree may be reduced 
(page 125). Thus — \ x^-^-^x — 3 agrees with it, if a = — i 6 = 2 
cs= —3. 

Definition. An expression is said to be a perfect square with 
respect to j:, when its square root can be extracted in a form which 

does not shew x under the sign v • Thus, as may be found by 

Vax^+2a^x-i'a^ = Va{x+a) 

Va^x+2ax^ + x^ = Vx{a+x) 

Hence ax*-\-2a^x-^a^ is a perfect square with respect to jr, but 
not with respect to a; while a*x-^2ax*-\-x* is a perfect square 
with respect to a, but not with respect to x. Observe, that if 
px^ + gx -^r be a perfect square with respect to x, it remains so 
after multiplication by any quantity which does not contain x, for tC 

px^+qx + r = (gx-hhy 
then mpx^ + mqx + mr = |V^(5'a:H-A)|* 

Lemma, The condition which implies that px^-\-qX'\- r is a 
perfect square with respect to Xy is g^ = 4pr. 

Let us suppose that gX'{-h is the square root of px*-i gx'\-r 
(no other form can be, as may be proved by trial). Then 

(gx + hy = px^-\-qix:-hr 
or g^x^ + 2ghx + h^ =^ px^-^-qx+r (always) 

Therefore (page 128) 

g^ = p, 2gh = J, A« = r 



Here, then, are three equations to determine two (yet) undeter- 
mined quantities g and h. If the product of the first and third be 
multiplied by 4, and if the second be squared, we have 

4g^ffi = 4pr and {2ghf or 4g^h^ = j« 

Therefore g* = Apr, which condition must be satisfied, if the three 
equations between g and h are to be true. 

The preceding equations give g = v j5, h ^ >/& or Vpx -f V^ 'J^ 
is (if g* = 4/?r) the square root of px* -{-gx-^-r, ^ 

Some ambiguity may here arise from Vp and v ^, as (from >- 
page 110) g* = p gives g either -f>//? or — >//?, and similarly h 
is either -f n/a or — >/f. CV 

But here observe, that the equation g^ = Apr was obtained 
(partly) by transforming 2gh=:g into 4g'A' = y*. But the latter 
might also have been obtained in the same way from 2gh ^ — y, in 
which — 9 is written in place of y; consequently, the same equation 
implies that both the following are perfect squares, px^ -^gx -^r 
and px^ — g^-^f' And if we take the two values of g and h, and 
combine them in every possible way in the expression gx -{• h, we 
shall have the four following expressions : — 

Vpx + Vr — Vpx + Vr 

Vjx — \^r — X^px — l/r 

each of which is either a square root o^ px^-\-gx-\-r or o^px^ — gx-^-r. 
But, returning to the untmnsformed equation 2gh =^g, which belongs 
to the former expression only, as 2gA = — g does to the latter, 
(both being represented in 4g'A'=92) we see that all the four 
values of gjr-|-A cannot be square roots of pj^-f-yj + r, but only 
those in which g and h have such signs, that the product gh may 
have the same sign as 9. For ihstaitce, if g be positive, g and h 
are either both positive or both negative; because gh must in that 
case be positive : if y be negative, either g is positive and h negative, 
org negative and h positive. 

Thus, tf g be positive, and 9' = 4/)r, the square roots of 
px^-{\gJ!-\-r are 'vpx-{-wr, and -^wpx — v r; if g be negative 
the roots of pr^ — gx-\-r are wpx — vr and — >/px -\-'^ r- 
These agree with page 110, where it appears that the two square 
roots of a quantity differ only in sign ; for 


Examples. 3j;^H-2a: + l is not a square, because (2)', or 4, is 
not equal to 4 (3 X 1), or 12; but 2x^ — 12j:-f 18 is a square, 
because ( — 12)', or 144, is = 4 x (2 X 18), or 144. Its square root 
is either '^2x — >/l8 {q is negative) or — >/2ar-+->/l8. 

The principal use of the preceding theorem is to complete a 

sguar€y 2iS it is called; that is, to supply either of the ternas^^j:', ^x, 

or r, by means of the other two. For instance, to make 2jr*+3j: 

a complete square. Here j9 ^ 2, 9 = 3, r is not given. But that 

the above may be a square we must have 9':= Apr, that is 9 = 8r, 

9 9 

or r = -, and we find that 2i'*H- 3 x + - is a complete square, its 

roots being 

3 _ 3 

either \/2x + ^ or ^\/2x-^^^ 

Generally, Mq^^^pr, r = j-» so ih^i, px^-\-qx is made a complete 
square by the addition of 

(co-efficient of x^^ 

4 (co-efficient of x^) 
Thus, ax^-\- ftj -f -— is a perfect square, and so is 4a^x^-{-^abX'{-b^, 

the roots of the latter being ± (2 a ar -f 6). 

We now proceed to distinguish the peculiarities of different forms 

of the expression 

ax^+bx + c 

If fe2 -— 4af^ ^e have seen that the expression is a complete square. 
We shall then look separately at the cases in which 6' is greater 
than 4ac, and in which 6' is less than 4ac. 
1. Let 6* be greater than 4ac, or let* 

J« = 4ac + e^ or 4ac = J«— e« 

* Why 4 a c + c* rather than 4 a c + « ? Because we wish to signify 
that 4ac is really increased. In 4ac + 6 we do not know whether there 
is increase or decrease, till we know whether e is positive or negative 
(page 63), But e^ is positive, whether e be positive or negative (purely 
symbolical quantities being out of the question). Hence the form of a 
square is a convenient method by which the student may bear in mind 
that a quantity is positive. 



Now aa:»+6x+c = ^°'^'+'°^-^+^''^ = -*"'^'+"''''^+^-^ 

4 a 4 a 

_ (2fl3r + 6)«— c^ _ (2gx + ^^ + 0(2ff.r + & — 
""4a 4a 

or, we have this theorem : 

If €» = fc«— 4ac or e = t/6= — 4«c 

the two being identically equal. 

2. Let 6'^4ac, then aj^-{-bx-\-c is a perfect square, and so 
it 4a'x'H-4a6jr + 4af, which is 4a'j'+ 4a6j + 6'; and 

4a 4a 

3. Let ^ be less than 4ac, that is, let 

i-=4ac — e^ or 4ac = J«H-c2 

_^ , . 4aV+4fl6.rH-4ac 4a«J« + 4a6j + 6»-f «• 

aa:« H- 6a? H- c = = — ; — ' = 

4a 4a 

_ (2aj + 6)'' + g' 
"" 4a 

Previously to proceeding further, we shall apply the preceding 
expressions to particular cases. 

1 . Let the expression be 3 j* — 7x + 4. Here a ^ 3, 6 = — 7, 
c ^ 4. And 6*^ 49, 4ac = 48, whence h^ is greater than 4ac, and 
ft*— 4ac =: 1 . This is c' ; therefore e = +1 , or —1 . Let c = + 1 

^*2 ry^ . .1 - (6.r--7 + l)(6.r-7-l) _ (6i— 6)(6j-8) 

oar- /a? i-^ - ^^^-3 ^2 

= 6(.-l)xn3.r-4) ^ (^.,)(3^.4) 

as may easily be verified by multiplication. Let the student shew 
that the supposition of c = — 1 gives the same result. 

3j — 4 
a: — 1 

3.r'— 4j: 
— 3J-f 4 

3.r3— 7j-|.4 

Now, we ask, what are the roots of this expression, or the values 
of X which make it vanish. A product becomes = if either of its 





factors becomes =0; that is, let x — 1=0, or, let 3* — 4 = 0, 
and in the first case 

3a:2_7a; + 4 = Ox (3-4) = 0: 

in the second, 3 j;=— 7a: + 4= (^ — l)xO = 


But if X — 1 = 0, j: = 1, and if 3j: — 4 = 0, j: = -, therefore 1 



and - are the values of x which make Sj' — 7j: + 4 vanish, or its 

roots (page 124). 

We now inquire what values of x will make 

3j:«— 7a:4-4 or its equal (x— l)(3ar — 4) 

positive or negative. It appears that x — 1 is positive when x is 

greater than 1 , and Zx — 4 when x is greater than - ; while the first 


is negative when s is less than 1, and the second when x is less 

than -, (page 126). 

Value of X. 
Less than 1 

Sign of J" — 1. 

Sign of 3 J — 4. 

Sign of the product 
(j:— l)(3jr — 4) 

f Greater than 1 
1 less than - 





Greater than - 





It appears, then, that the preceding expression is always positive, 

except when x lies between the roots 1 and -. In this manner we 

have determined the following points with regard to 3^:* — 7jr-f-4: 

4 4 

it is + when x is greater than -; when j: ^-•; — when x is less 

than -, and greater than 1 ; when j; is 1 ; + when x is less than 


1 . Follow a similar process with the following expressions. 
2a;2+3a;4-l = (a; + l)(2a: + 1) 
3a;2+4a;-7 = (a:- l)(3a: + 7) 
-2a;2+6a:-4 = (2 - ar) (2 a: - 2) 
Hitherto we have chosen expressions containing no irrational 
results : let us now try Zx^-\-bx — 1. Here we have a s=: 3, 6 = 5, 
c= — 1; h^ or 25 is greater (page 62) than 4flc or —12, and 



b^^Aac = 37 = €*; therefore c = ± ^/37. Let c be + -^37, then, 
page 133, __ _^ 

3a-«+5ar — 1 = (^X^-^ + 5-fV37)(2x3.r+"> — ^37) 

= j^(6x+5+ i/37)(Ca;+6- \/37) 

Hence the roots of the expression, which call x, and x^, are 

>/37 + 5 \/37 — 5 

a; = : — x =1 . — 

' 6 I' Q 

= -1-8471271 = -1804004 very nearly. 

It will be found, as before, that the preceding expression is never 
negative, except when x lies between x^ and x^,, 

2. Let us take Sj:*— 6j + 3. Here 6' or (—6)' is equal to 
4ac, or 4x3x3. Hence the expression is a perfect square, and 
we have (page 131) 

3a;«-6a;H-3 = (t/3a:- V^)' = 3(a:-l)« 

This expression vanishes only when v3j: — v3 vanishes, or 
when dT ^ 1. But, because there are two equal factors, each of 
which is wZx — ^3, and to preserve analogy with the preceding 
case, it is said to have two roots, which are equal. Thus this ex- 
pression has two roots, each = 1. 

This expression is never negative, for {x — 1)' is positive in all 
cases. We can only make it negative by giving a purely symbolical 
▼alueto x: for example, l-\-s/ — 1. Then {x — 1)* (by rules only, 
*ee page 122) will be — 1. 

3. In no case hitherto taken has h^ been less than 4ac. Now 
try 2**—* + 4. Here a =2, 6= —1, c = 4; and b^ is 1, 4ac 
w 32, greater than 6*. Here, as in page 133, let 4ac^^' = e', 
'Hiich is therefore 31. 

Hence, page 133, we have 

9^« x\\- (2X2^-1)« + 31 _ (4j-1)» + 31 
^«a: + ^- 4X2 ~ 8 

Tiiis expression has no positive or negative root, for (4;r — 1)^ being 
always positive, so long as j: is positive or negative, must increase 
31, and, therefore, (4 or — l)*-f 31 can never =0, but is always 
positive. We see, then, that 2j?' — x-{-A is always positive, for every 
positive or negative value of x. The least value of the expression 





under this limitalion is —, for the least value of (4r— 1)« is found 


by making Ax — 1 = 0, or j = -. Consequently^ the above expression 


has the following property : its least value is — , found by making 


jr = - ; for every other value of x it is greater. 

The following cases may also be tried by the student. 

x^ + x + l = l{(2xH.l)«H.3 } 
x^^x+l = l|(2x-l)2H-3 } 

-2a;2+2a:-5 = - 1 |(4x-2)«H.36} 

We can give the preceding expressions purely symbolical roots ; 
for instance, to make 2x^ — x -|- 4 = 0, let us make 

(Ax--\f+31 = (4a;-.l)« = -31 

4ar-.l = 4.l/r3r or 4ar-l = -t/33l 

Call the roots derived from these x, and x^, 

i + v^UsT 

X. = 

X = 

1 — >/— 31 

4 " 4 - 

which will be found to be roots, by rules only, as in page 122. 
We shall now take the more general cases. 

1. ax^-\-})x-\-c = 0, where h^^\ac = e« (page 132) 
and ai? -\-}) X •\- c = — (2aa: + J-f 0(2aa;H-fi— c) 

The expression ax^-\-bx-\-c contains eight different forms^ as 
follows, which we shall distinguish by the eight letters A, B, C, D, 
A', B', C, D'. 

Sign of a. 

f(A) 2a;« + 5a;+l + 

j(A0-2a;2-5a;-l - 

f(B) 2a;«-.5a: + l 

f(C) 2a^-i-5x^l 
j(C')-2a:2-.5a: + l 

f(D) 2a;2-5a;-l 




Sign of b. 



Sign of c. 

- I 

- I 


We shall first consider that whicli is common to all the forms ; 
and then the peculiarities of each. 

The roots of ajr'-f tx + c are found by solving the following 
equations (call x, and x^^ their roots) 

2fla;H-6 — e = 2ax + b + e = 

^6+c — h — e 

X = X ^ 

'2a " 2a 

But e = Vl^—^aCy therefore 

— 6+^62— 4ac ^h^'Jh^'^^ac 

X = ■ X = ^ 

' 2a " 2a 

And, page 126, the expression 2ax-\-h — e is the same as 2a(jr— jr^) 
and 2ax-^h — c as 2a(j: — x^^. This may also be shewn again, 

2ax+b+e = 2a{x + '^) = 2a{x-=^) = 2a{x-x,) 

'.ax^-\-ox-\-c = — ^^ — -^ = a(a; — a:j(^ — ^/,) 

Hence, when the two roots of an expression of the second degree 
are known, and the coefficient of its first term, the expression itself 
is known. For instance, what is the expression whose roots are 2 

and , and the coefficient of whose first term is 4 ? By the pre- 

ceding formula, this expression must be 

4(j-2)(a;-(-J)) or 4(a;-2)(a; + J) or4a;2-6a:-4 
If we develope the preceding expression, we find 

a(x — x^{x — x^ = «a;*— a{x, + x^x + ax^x,, 

which is identical with ax^-\- h x •\- c\ therefore, page 128, we 

h = -a(a;,+ a:J or x,-\' x,, = "" « 

c = ax,x„ or x,x„ = ^ 

_ ^ , Coefficient of j? 

Sum of the roots =r — 7^ — ^-i — r— F-a 

Coefficient of x* 

^ - - , Term independent of x 

Product of the roots = — 7;; — ^ — r— 7-3 — 

Coetncient of r 

N 2 



When a := 1, or the expression is x*-^ bx -{-Cfyre have 
Sum of the roots = — b Product of the roots = c 

Verify this theorem on all the preceding examples. 

We shall denote the preceding forms by placing the signs of 
the terms in brackets : thus A will be denoted by (+ + +), A' by 
( )> &c. This first case, in which 6'— 4ac is positive, in- 
cludes all possible varieties of 

(++-) ( — +) (+ — ) (- + +) 

for in all these, a and c have different signs; ac is negative, and, 
therefore, b^ — 4ac is positive and greater than 6*. Here then, 6* — 4ac 
is positive without reference to the numerical value of a, b, and c. 
This same case may or may not include the following, 

(+ + +) ( ) (+-+) (-+-) 

in all of which ac is positive and therefore the sign of 6'— 4ac 
depends upon the simple arithmetical magnitudes of 6' and ac. 

We shall now examine the cases which have roots; and re- 
mark that either of the expressions in any one pair may be reduced 
to the other, by simple change of sign. Thus — x^ — x -{- 1 = 

— {jc^-^x — 1) or ( h) becomes (+ H ) by entire change of 

signs only. And, since, when A = 0, then — A = 0, the expressions 
in the first of the following columns have roots similar to the corre- 
sponding expressions in the second, in every circumstance which 
depends only upon the signs of the terms. 

(+ + +) 

( ) 

(- + -) 

(+ + -) 
(+ — ) 

( — +) 
(- + + ) 

I. Expressions (+ + +) ( ) ; roots not necessarily existing ; 

and i* — 4ac less than b^. The roots of this expression (when it 
has them) are both negative. For, since ft' — 4ac is less than 6*, 

Vb^ — 4ac is less than v 6% or, page 123, the numerical value of b. 
Therefore — ft + '^ 6' — 4ac and — b — s/b^ — 4ac have* the same 
sign as — 6, or a contrary sign to b. But the roots are 

— 6 + n/6^— 4qc ^h'-^/b^'-'Aac 

2a 2a 

* For instance, both values of —S-±_^ must be negative; —3-4-6 
has one positive and one negative value. 


both of whicli, as to sign, are therefore negative, because b and a 
haye the same signs in both, and the numerator is of a contrary sign 
to, and the denominator of the same sign as, a and b. 

II. Expressions (H 1-) ( j ); roots not necessarily exist- 
ing; and b' — 4ac less than b\ By a process exactly similar to the 
preceding, remembering that a and b have now different signs, we 
prove that both the roots, when they exist, are positive, 

III. Expressions (+ •] ) ( [-); roots always existing; 

t« — 4ac greater than 6*. These two expressions have one positive 
and one negative root, the negative root being numerically the greater. 
For in this case,* since ^6' — 4ac is numerically greater than b, 
— 6+ V 6' — 4ac and — b — Vi' — 4ac have different signs; namely, 
the first is +, and the second — . Therefore, 

— 6 + >/6' — 4ac (which is +) 

— b — >/6' — 4flc (which is — ) 

agrees in sign with a and b. 
differs in sign from a and 6. 


If a and b be positive, the second (which is then — ) is numeri- 
cally the greater (by the preceding note) : if a and b be negative, the 
first (which is then — ) is numerically the greater. Therefore in both 
cases the negative root is numerically the greater. 

IV. Expressions (H ) ( h +); roots always existing; 

(9-.4ac greater than 6*. Here, by reasoning precisely similar, it 
may be proved that there must be one positive and one^negative root ; 
but that the positive root is numerically the greater. Observe that 
a and b have here different signs. 

In all these cases we have also the following theorem. The 
expression flj?'H-fcjp-}-c, when it has different roots, never differs in 
sign from a, except when the value ofx lies between that of the roots, 
(Read page 134 over again, with attention.) For we have 

aar2-f-6a: + c always = a(x—x){x^x,) 

One of the two roots x^ and j,, must be the gieater; let it be or,. 
Then, if x be greater than x^^ it is greater than x,,; and x^x, and 

• Remember that in p + g it is the sign of that which is numerically 
the greater, which determines the sign of the expression ; and that in 
p + 9 that one, either p + g or p— g, is numerically the greater, in which 
both terms +p and i;^ have the same sign. 


X — x^f are both positive. ITierefore a (j— j^) (j?— x,,) has the same 
sign as a. Let x be less than j:^, but greater than x^, (that is, let x lie 
between the two rools\ then x-^x^ is negative, x — x^, is positive ; 
and a(jp — 0(^ — 'O differs in sign from a. Let x be less than x^^, 
then it is less than x,; and j: — x^ and j: — j:^, are both negative; 
therefore a(j — J^,)(,J^ — J?,,) has the same sign as a. A recapitulation 
of these three cases gives the theorem in question. 

2. ax^ + bx-^-c = where i^ = 4ac or fi*— 4ac = 

This case requires that a and c should have the same sign, because 
4ac must be positive. 

The two equal roots are derived from 

2aX'j-b = or a:, = a:,, = — — 

which are positive when b and a differ in sign, that is, in (H h) 

and ( 1 ) ; and negative when b and a agree in sign, that is in 

(+ -f 4 ) and ( ). The other cases are entirely excluded, 

since a and c must have the same sign. 

The expression ax^-\-bx -^c being always a square (a positive 
quantity) divided by 4a, always has the same sign as a; observe that 
X cannot now lie between the roots. 

3. ax^-^-bx-^-c = 4ac— 6^ = e^ (page 133) 

Here a and c must have the same sign, because 4ac is positive, 
being 6' + e', the sum of two positive quantities. 

(Page 133) ao^ + bx^c = j^ |(2aa: + 6)^H-e2 | 

and being a positive quantity divided by 4 a, always has the same 
sign as a. 

The purely symbolical roots (see page 136) are derived from the 

(2aa: + J)2 + e« = or (iax^hf = c^x -1 
or* 2aa; + i = ±eV^ = ± V^4a^ir6« l/HT 

♦ Observe that p^ = 5* or ^:: p » ±5, gives only two distinct forms ; 
for +p » +q and — p «= — g are the same, as also are +/) «= — g and 
-p « +q. 



— ft + '^Aac-'b*^'^ — ft— .>/4ac— ft*^/^ 

' 2a " 2a 

These roots, using rules only, will be found to satisfy tlie equation, 
and also the equations 

_ _ft _ c 

but we cannot at present make an extension of the theorem in 
page 139, because we can attach no notion oi greater or less to x, 
and jr^. 

The numerical solution of equations of the second degree is 
usually performed by a process instead of a formula^ each case by 
itself, as follows: 

Let 2a;2-7a:+3 = 

(-)3 2a:2_7a:-_3 

(-)2 ^-|^ = -5 

Complete* the square, ^ — 5^"^ W ^ \4/ ^2 ^ 16 

7 5 

Extract the root, x — - = ± - 

4 4 

7 , 5 q 7 5 , 

4*4 ' 4 4 * 

But we should, by all means, desire the student to commit the 
following theorem to memory : 

If ax^ + bx-^-c = 

— ft+N/ft*— 4ac — ft_->/69— 4ac 

ar = either ■ — - or 

2a 2a 

Examples. 1 . What are the solutions of 

px^ + q^x = qx^--p^x+p^ 
or (,p — q)x^-\-{p^-\-q^)x^p^ = 

Here a = p^q b = p^-i-q^ c = --p^ 
The roots are the two values of the expression 

♦ See page 132, where it appears that i' + ftx + - is a perfect square, 
namely, that of » + -. 


— 4(p — y) (— p^) = 4|>* — 4p3y 
Therefore the roots are coDtained in 

2. Let asP' — ahx = b^x^h^ 

The roots are contained in 


{ab + b'^y-4caP = a^jg ^ 20^ + 6*-4a6^ 

= a2 6« - 2a63 + 6* = {ab-b'^y 

Therefore the roots are contained in 



! ! = == b one root, 

2a 2a 

1 ■ — = — =s — the other root. 

2a 2a a 


ion, 6 + - = = ^ 

' a a a \ 

5a 53 ( (page 137.) 

fi X - = - 

a a J 

The student should now proceed as follows : 

1. To form examples of numerical equations; — choose two roots 
and a coefficient for the first term, and construct the expression which 
should have those roots, as in page 137; then find the roots of the 
resulting expression by the preceding formula, which should be, of 
course, the roots first chosen. Afterwards take any expressions at 
hazard ; find their roots, and verify them by actual substitution. 

2. To construct literal expressions which shall afford solutions of 
more interest than those taken at hazard, choose any expression which 
is identically = 0, in which one letter has no higher power than the 
second ; such as ^ 

ab^—abc + abc-'ab^ = 
write X instead of b in such places as will create an expression, of 


which it could not be known at first sight that it is made to vanish by 
X =: b. For instance, suppose 

ax^^abc-i-acx^abx = 

one of the roots of this should be b. Find the roots by the formula. 
Or take the following method : Choose any two simple expres« 

sions, one of which only has a denominator; such as — and p. Then 


the roots of 

nx^'-'(m+np)x+mp = 

should be — and p. For instance, take b and . Tlien 

n a 

m = 1— aft 71 = a p =^ b 
m'\-np = l—ab-j-ab = 1 ; mp = b—ab^ 

Therefore the roots of 

ax^-^x + b^ab^ = 

should be a;,= ft and x,=- 

Anomaly. In the expression ax^-\-bx'\-c = let a = 0. It 
then becomes 6j -|-c = 0, giving a: = — -. But if we examine the 
roots ofaj:*-|-6j-f-c = upon the supposition that a = 0, we find 

assumes the form - (page 25.) 

assumes the form (page 21.) 

2a ^^ ^ ^ 

Are we then to say, in conformity to the pages cited, that one root 
is infinite^ and the other what we please ? Apparently not, m the pre- 
sent case; we must therefore examine it further. Instead of supposing 
a s= 0, let us (page 21), suppose it as small as may hereafter be neces- 
sary. The Lemma which we here lay down will be useful in every 
part of algebra. 

Lemma. The expression n/6' + v, may, by supposing t; suffi- 
ciently small, be made to differ from b by as small a quantity as we 
please: and, moreover, the same expression may, under the same 


circumstances, be made to differ from b +r7) not only by as small a 
quantity as we please, but by as small a fraction of v as we please. 


[In explanation, n/i + v, however small t; may be taken, exceeds 
1 by something near the half of v ; but Vl + v may be made to differ 
from 1 + i V by less than the ten-millionth part of v, if necessary.] 

Tlie first part of this lemma is evident enough : the second we 
prove as follows : 

»» ,„ V* 

= ^ + "+46^-^*-" = 46. 

Therefore J -|- iL «. l/fcs 4- 1; = 

4 6' 


6+^^+>/6» + . 

But the last-mentioned fraction has a denominator, which, when 
V is diminished, approaches continually to b -|- v 6* or 26. Let it 
be called 2 6 + 2<;, where, by making v as small as may be necessary, 
we can make w as small as we please. Tlien will 

j + jL_i/jq:^=^ 


26 ^ ^ 46\26-|-w;) 46«(26 + u;) 

that is, 6 + — 7 differs from v 6'*4-t; by a certain fraction of w, namely, 

, .o .^, . . of V. But since v can be made as small as we please, 
4 6* (2 6 + w) r F 

and thence w (see what comes before), that is, since 46'(26 + u;) can 
be brought as near as we please to 46^x2 6 or 86% the fraction of v, 

by which 6 -f -- differs from >/6' + v, may be thus represented : 

V (a quantity as small as we please) 

8 6' (a given quantity) 4- a quantity as small as we please 

and may therefore be made as small as we please. 

Precisely the same sort of demonstration may be given of the 

following; namely, that ^ — ^r may be made to differ from n/6*— v 

by as small a fraction of v as we please. 


We shall now proceed to apply this theorem to the consideration 

of the roots 

— 6 + >/F^^4flc • — 6— '\/6*— 4flc 
■ — s sind — ' 

on the supposition that a may be as small as we please. Hence, 
c being a given quantity, 4ac may be as small as we please; and if 

V ^4ac (by the last lemma) 

V 6' — 4ac may be made to differ from 6 — -^-r- by as small a 
liraction of 40C as we please. 

. 4 A c 

Let, therefore, «s/^ — 4ac = 6 — tt p X 4ac, in which we 

may make p (with a) as small as we please. 
Then the roots are 

^b + b j^ 4pac — 6— 6 + -^^H-4|)ac 


2a 2a 

— 26-1* -T- + 4|? ac 
or «.-_2pc and 

I^ow, diminish a more and more, in which case p is diminished 

in the same way. The first root continually approximates ^o — r-> 


and the second to the form . But the first is the root derived 

from the earlier view of the equation a x^ -f 6 j: -f c = in the 

case where a =: 0, namely, 6jr -f c ^ 0, which gives or = — j. 
The second is yet unexplained. 

Problem in Illustration, a, 6, c, and e, are four numbers, 
the last three of which are increased by a certain number, and the 
first by m times that number. The results are then found to be 
proportionals. What is the number ? 

Let X stand for the number. Then mx-\-a, x-f^, X'\'C and x-f c 
are proportionals. 

^ ^4-6^ "^ 7+e ®^ (»ia?+fl)(a: + e) = (j;+J)(a:+c) 

Perform these multiplications, and reduce the result to an equa- 
tion of the form P =: 0, which gives 

(m— l)a;* + (mc-f-a— 6— c)arH-ae— Jc = 




The values of j? are contained in 

— (tne + a — b — c) ji v(me-i-g — ^ — cy — 4(iw — l)(ae — be) 

2(m— 1) 

and therefore, generally speaking, there are two solutions of the pro- 
blem. But if m = 1, that is, if x must be so chosen that x + a, x -f- 6, 
x 4- c, and x -^d are proportionals, the case we wish to consider 
arises : for m — 1 = ; the equation is reduced to 

(e-f-a — i — c)x -f ae— 6c = 

which gives only one root; and one of the roots just given takes the 


— 2(e + a — 6 — c) 

The interpretation of this form in page 25 was, that any very great 
number would nearly satisfy the conditions of the problem, a still 
greater number still more nearly, and so on. Now, the question 
becomes, will x -\- a, x -\- b, x -^ c, x -\- e, approach more and more 
nearly to proportionals as x is increased ; that is, will 

X -^ tL X "^r C 

. := approach to truth in that case ? 

X *j~ X "J" c 

Divide both numerator and denominator of both fractions by j: ; 

a c 

1+- 1+- 

X X 

this gives t- =: — which may be made as near the truth as we 

1+- 1+- 

X X 

please, by taking x sufficiently great ; for, by so doing, -, -, -, and -, 


may be made as small as we please, and the preceding equatio 
brought as near to - = - as we please. 

Hence it appears, that when a problem which, generally speaking, 
has two solutions, has a particular case in which there is only one, we 
may say that there is another solution corresponding to an infinite 
value of the unknown quantity, in the sense explained in page 25. 

But, though we see a confirmation of the interpretation put upon 

— — in page 25, we also see that --, which is the form of the other 

root, does not admit the interpretation of page 25, namely, that any 
value ofx will satisfy the equation ; but it indicates that the rational 

root is — T, We shall return to this point in the next chapter. 


The case of a = presents new circumstances: let us now' sup- 
pose only c = 0. We have then ax'+ bx := 0, or x{ax -f 6) = 0; 
which is satisfied either byj: = orbyaj: + ^ = 0. That is, the 

roots are and . This would also appear from the general ex- 
pressions for the roots. 

Similarly, if 6 = 0, we have a j^-f c := 0. 

a;*= x=+v or — -i/— i 

a ^ a ^ a . . 

which pair consists of a positive and negative quantity when c and a 
have different signs, and is purely symbolical when c and a have similar 
signs. This also would follow directly from the general expressions. 

We choose one from among many instances of the use to which 
the preceding theory may be put. Suppose we know the sum of two 
quantities (s), and their product (^). Required expressions involving 
nothing but this sum and product, which shall give the sum of the 
squares, or cubes, or fourth powers, &c. of the two quantities. 

By page 138, these two quantities are the roots of the expression 

Represent the roots by x^ and x^,; we have then 

a;»+» _ saT-^' +px'; = 

, , n+2 n+2 / n+1 n+l\ / » . »\ f\ 

Let the sum of the nth powers of x, and x^, be called An ; the 
preceding then becomes 

An+2 — 5 An+i -f p An = 
or An+2=«A„+i — pA„ 

Now Ao = JT,^ + a;,,« = 1 + 1 = 2 (page 85) 

Ai = a;, H- x„ = s 
Therefore, by the preceding equation, 

Ag= 5A1— 2}Ao= s^— 2p 

A3= 5A2— pAi=5(5*— 2p)— ^5 = ^^Z'ps 

A4= 5A3-M2=H^-3;)5)-p(s2-2p) = 5*-4p5« + 2p2 

and so on. 


There are many equations which may be solved 1^ the assistance 
of the preceding theory : the reason being, that though they are not 
strictly of the second degree with respect to the unknown quantity, 
they are so with respect to some expression containing it. For in- 
stance, we wish to find the values of x which satisfy, 

j;«-.3ar-f.l = 2- Vx2-3a?+l 

In order to clear this equation of the radical sign, we should proceed 
as follows. 

Va:«— 3a:-f-l = 1 + 3a:— ar« ; square both sides, 
a:«— 3x+l = l + 6a:+7a;*— ear'-f-a:* 
or a?*— 6a:5-|-6a?«-f-9a; = 

an equation of the fourth degree, for which no method of solution 
has preceded. But, on looking at the original equation, we see 

immediately that it is of the form v*s= 2— v; for if Vx*— 3« + ^ 
be V, then jr* — 3* -fl Js v". Let v = Vj:* — 3 x +1, then 

,^4.v-.2 = i; = 1 or ^2 

First, let t; := 1 

Vofi-^Zx^-l = 1 or a;«— 3a; +1 = 1 /. a? is or 3 

Next, let V = — 2 

V'x«-3a;+l = -2 ar«-3ar+l=4 a: = ^-^-^ 


Therefore the preceding equation is satisfied by the following 
values of x ; 

3 3 + v^gT 3 — \/2T 

2 2 

Again, suppose 2x«— 3 = ^. Here a* is {si^f\ let i; = jr», and 
the equation becomes 21;*— 3 = v, the roots of whidi are —1 and f . 
Hence, 3?*= — 1 or 0:*= |^; that is, any values, real or purely sym- 
bolical, of VHT and >/| are roots of 2j*— 3 = ^. 

We shall close this chapter with some instances of the process 
of clearing an equation of the radical sign. Let it be the following 

Vic -f- Var-f-l -f- V^a; + 2 = 2 
VJT V^JTT = 2- V7+2 

• • 


Square both sides 

x + 2V^V'a7 + l-f.a:-f-l = 4-4l/jcT2 + x + 2 

Square both sides again, 

4«(a: + l) + 16V^a;(a; + l)yJT2 + 16(a:4.2)=25-.10a:+a:« 
«' 16l/^<^TTj(iT2) = - (7 + 30a; + 3»«) 

Square both sides again, 

266a:(a;4.1)(a; + 2) = (7 + 30ar +3a:«)« 

which contains no radical sign, and may be developed. 

The following are more simple instances, which we leave to 
the student. 

The equation Vx + d + Vx — 3 = 4 

gives or — 4 = ^ 

The equation \/x +a + Vx-^- b = c 

gives 4c2a7 + 4a6 — (c^— a--by = 

But we must observe that 

(x + a)* -f (a: + 6)* = c (see page 123). 

gives the same result as the last, and admits of the four following 
forms : 

— Va;+a + Vx+h = c 

— Vx + a — V^a: + 6 = 

Vx-\-a + V)c-^b = c 
V^ar + a — Vx+b = c 
The value of x above obtained, namely 

will only satisfy one of these. Consequently, when we obtain one 
of the preceding equations, we cannot be sure but that the problem 
has been misunderstood and requires an extension of form which 
will give another of the preceding. 

We give the following as exercises : 

1. Shew that a-|-- cannot be numerically less than 2. Prove 

this by shewing that the roots of a + - := 2 — p are purely symbolical 




^hen p lies between 4 and 0. It may also be shewn from (a — 1)* 
being always positive. 

2. Shew that a*+ ^ °^^t he greater than 2 ah, 

3. Prove that if a -f- = »» 


• fl» * a* a 

4. If *, and" x^, be the roots of the expression «*• + 6 jr -f c> then 


±Vg534^ i' = ^=iif :t: — ^6^=4; 

' " a Xff 2ac 2ac 

5. In the expression ax'4- ^' + c, supposing it previously known 
that one root exceeds the other by m, find the roots without the 
assistance of the formula. Do the same on the supposition that one 
root is n times the other. 




Wb have already had occasion to observe the effects of particular 
Oppositions, which make what in other cases are intelligible quan- 

tjties, assume the forms -, -, a^ &c. To these we shall now add the 

^rm itself, as requiring investigation on account of the circum- 
stances under which we may have to use it ; for since we have found 
it convenient to reduce every equation to the form P = 0, we might, 
Vrithout proper caution, be led to such inferences as the following : 
X.(ab =i and ac = 0, then a6 = ac or 6 = c. [We* shall give 
^ striking instance of this, as follows: If J7 — 2 = 0, it follows that 
^ — 4 = 0, and that a?*— 2jr = 0. Are we then to equate a?* — 4 
«md j'— 2j:, and proceed in any manner, previously explained, 

'With the results? If we do so, since jf*— 4 = (j — 2)(jr + 2) 

and or*— 2* = *(*— 2), we have 

(ar-2)(x + 2) = (a?-2)a? (-r-) (ar-2) ar+2- = x 

But a?— 2 = .-. a? = 2 or 4 = 2 

an absurd result, which indicates some absurdity in the process. The 
suspicious step is the division of both sides of an equation by x — 2, 
which is 0. If we go through the preceding process without the con- 
cealment of which takes place by making the supposition ^—2 = 
and then using j:— 2 instead of 0, we shall find a most evident 
&llacy, amounting to the following : 

axO = JxO = .-.0x0 = 5x0 (-f-)0a=6 

that is, we have used as a quantity, have asserted = 0, and have 
divided by 0. Returning now to the principle in page 21, we shall 
suppose X — 2 instead of being =: 0, to be very small,9nd shall put in 

* All that comes between the brackets [ ] is vaguely stated ; the 
object being nothing more than to shew the student how liable he is to 
error in using such terms as nothing, small^ great, nearly equal, &o. 




opposite columns the two analogous processes, each containing its 
own error. 


a;— 2 = 
a;«-4 = 

and a:«— 2j: = 

a;2— 2a: = a;«— 4 

Let X — 2 be as small as we please; 

•*• :r^— 4 may be as small as 

we please 
and 3? — 2a7 may be as small as 

we please 
/. a;*— 2a; and a;*— 4 maybe 

as nearly equal as we please 
and the same of a; (a;— 2) and 


(-f-)(a;— 2), then x andar+2 
may be as nearly equal as 
we please. 

But X may be as near to 2 as we 
please; therefore 2 and 4 
may be as nearly equal as 
we please. 

In the second column there is an error, whichever of the senses 
in page 24 is put upon the word eqwiL If we call quantities n^ly 
equal whose difference is small, then we do not know that because 
«•' — 2x and ar' — 4 are nearly equal, they will still be so after divi- 

or a:(a:-2) = (ar-2)(a;+2) 
-r-(ar— 2) X = a; + 2 


a;-2 = a; = 2 
2 = 4 

sion by x — 2. For if a:— 2 were, for instance. 


then divi- 

sion by X— 2 is multiplication by 1000, or the difference between the 
quotients x and j7-)-2 is 1000 times the difference oi x^ — 4 or and 
jr* — 4. And the less x — 2 is, the greater is the real multiplication 
to which division by x — 2 is equivalent. If it be said that the quo- 
tients X and x-\-% do not differ by a larger proportion of themselves 
than x^ — 2j7 and i^ — 4, and that, agreeably to the sense preferred in 
page 24, a is as nearly equal to 6, as c is to dy when the differences 
of the first two and of the last two are in the same proportion as 
a and c: the answer is, that x' — Ix and s^ — 4 must not then be 
called nearly equal, because they are small, and because their differ- 
ence is therefore small ; for both may be small, and, nevertheless, 
one may be many times the other. An elephant and a gnat are both 
small fractions, if the whole earth be called 1, but they are not 
nearly equal in any sense. 


From the above we gather^ that, calling a and b nearly equal wheo 
they only differ by a small fraction of either, we are not at liberty to 
say that two small quantities are therefore (because they are small) 
nearly equal.] 

There are two obvious tests of the abtoltUe equality of a and b : 

a — 6 = and - = 1, 

We lay down the following definition. Approach towardi equality 
is measured not by the diminution of the difference, but by the approach 
of the quotient towards 1 . Thus 3 is not more nearly equal to 2 than 
20 to 25 because 3 — 2 is less than 25 — 20 ; but 3 is not so nearly 
equal to 2 as 25 is to 20, because 

- exceeds 1 by - — exceeds 1 by - fless than -) 

See page 24, for an anticipation of this use of the words ^^ nearly 
equal,'' and what precedes in [ ] for reasons. 

Theorem. The value of a fraction depends entirely on the 
relative, not on the absolute, value of the terms. The following ei^ 

amples will shew that this is contained in the theorem — r as -. 

'^ mb 

1. Find a fraction whose numerator is 583, and which is as 

„ 1 . 583 

small as r— --. Answer 

1000 583000* 

1 a 

2. Find two fractions, a and b, each less than tttz, so that j may 

be a million. 

Answer a = b = 

2000 2000,000,000* 

Here ^ = 1000,000 * ^ 

6 * a 1000,000 

3. Find two fractions, a and 6, each less than 2, so that -r may 

be =: m. 

Answer, Take any two numbers, p and q (only letp be less than 
2 9), and let 

q mq 

Exercises. (All the letters are positive) p is not so nearly 
equal to p-^q as p-^m is to p -t q -{• m. If a be more nearly equal 
to b than c is to e, then a + c is not so nearly equal to b-^-eBS a is 



to 6, but more nearly equal than c is to e. Again, mx is as nearly 
equal to n j as my is to ny. 

We must now say something upon definite and indefinite terms. 
A word is called definite when there can be no question as to 
whether it is proper or not to use it in any particular case which may 
be proposed. A word is called indefinite when it may be matter of 
opinion as to whether it is proper or not to use it in any proposed 
case. Thus, equal is a definite term. No two opinions can exist 
upon the answer to the question '' Is 4 -f 4 equal to 9 ?'' But great 
is an indefinite term. ^' Is 1000 a great number?'' The answer 
to this is, there is no way of answering the question in any case 
upon which all must agree. We give some examples of each sort. 

Definite — equal, exact, larger, nearer, smaller, greater, less^ 
lai;gest, smallest, &c. as large as, as much as, as great as, &c., quite. 

Indefinite — near, small, large, great, much, nearly, hardly, &c. 
large enough, small enough, &c. 

With indefinite terms we can have nothing to do, unless by 

addition to their meaning, so as to give them a signification which 

will allow us to use them without presenting as mathematical theorems 

propositions which contain matters of opinion. We shall take the 

terms near^ small, and great as instances. Observe that the term 

tmaller is not to be considered as altered in the same manner as 

less in page 62. It keeps its arithmetical meaning. '* If x he 

small, 7+ J^ is nearly equal to 7?^' This is a proposition in which 

all will agree : and the reason is that ^' small'' and " near" have a 

connexion which is independent of what fraction the sp>eaker may 

dioose to think entitled to the term ^* small." A B may be a line 

which one may call small, and another not small ; but all will agree 

that in the meaning of the words, *^ small" and *' near" is implied 

'' If AB be smallf A is near to B." But if we come to ask — 

1 1 
What fraction is small, is it rrrr, T:rrr, &c.? — The answer must 

' 100 1000' 

depend on circumstances. We reject, therefore, the terms small and 
near in their common meaning. But the preceding proposition csUi 
be put in a form which will never render it necessary to inquire 
what is small or near. '' If x may be as small as we please, then 
7 + ^ niay be made as near as we please to 7 ;" or, *' Let me make 
X as small as I please, and I can make 7 -f ^ as near to 7 as you 
please ; or, <^ Name any fraction you please, and let it be such a 


one as you choose to call small, I do not ask why : then, if I may 
make x as small as I please, I can make T+J^ differ from 7 by a 
fraction less than the one named by you;'' and so on. 

Having rejected the terms ^mall^ greatf and near, in their common 
signification, we shall revive them for our own use in algebra, simply 
as convenient abbreviations of <' as small as we please," '* as great 
as we please,*' *' as near as we please," or, ** as small as may be 
necessary," '^ as great as " &c. &c. In this sense we have various 
demonstrable and definite propositions. For instance, *' i(x be small, 

- is great;" that is, if we may make x as small as we please, we 
may make - as great as we please. 


When, under certain circumttances, or by certain tuppositions, we 
can make A as near as we please to P {A being a quantity which 
changes its value as we alter our suppositions, and P a fixed quantity, 
which does not change when we alter our suppositions) then P is called 
the LIMIT of A, A quantity which we are supposing as great as we 
please is said to increase without limit ; one which we are supposing 
as small as we please, is said to decrease without limit. The follow- 
ing theorems will be evidently true. 

If j: decrease without limit, the limit of a-f^ is a; if ^ diminish 
without limit, then - increases without limit ; if x approach without 


limit towards b, then the limit oi a-\-x is a -|- 6. 

The first and third may appear but a complicated method of 
saying that if 2; = 0, a-{-j: = a; and if x = 6, a-fx^a+ft, which 

are perfectly intelligible. But, " if j: := - = - " has no intel- 

ligible meaning; in fact, in page 25, we have already anticipated 
the construction we here put upon that proposition. 

One object of this chapter is, to put interpretations upon those 

forms which would otherwise offer difficulties; such as 0, ~, -, and 


(had we not otherwise found a rational interpretation) a^ But we 
still have the forms 

00 0* G)" &c. 

all of which might occur, if we stumbled upon such expressions as 


ix-a)'-" {x-a)^' (^y &c. 

without observing (what might happen) that j: = a. 

In all these cases, that is, when we get a form which is not a 
direct representation of quantity, we shall not ask *' What is the value 
of that form V* or in any way enter into the question whether it is 
demonstrable that it has a value or not. But the question we shall 
always ask is this : " As we approach the supposition which gives the 
unintelligible form, to what value does the expression which gives 
the unintelligible form approach V For instance, 

when a; = a = - 

j: — a 

But if we examine what sort of change of value takes place in the 

above fraction when x approaches towards a, we find that value to 

approach towards 2 a, as will be afterwards shewn. And it will be 

found that we have the following proposition : '^ If dr may be made 

as near as we please to a, then can be made as near as we 

please to 2 a," or, " If x approach without limit to a, then 

approaches without limit to 2 a." 
Shall we then say that 

when X = a — ^^^— = - = 2a or - = 2a (inthiscase)? 

Whether it will be proper to say so, in the common meaning of all 
the terms, we leave to the student. But we shall not, in this work, 
use such a form, except as an abbreviation of one of the preceding 

It is usual to make the symbol oc stand for -- ; and this is called 

infinity. From what has preceded, and page 25, we shall regard 
j: := oc as an abbreviation of the following : *' Let x increase without 

Again, '^let j: = 0," it will often be most safe to regard as an 
abbreviation of '* let x diminish without limit." We shall hereafter 
return to this. But in equations of the form P — Q = 0, where P and 
Q are certainly finite quantities, this alteration will not be necessary. 

Theorem I. If A and B be two expressions which are always 
equal, so long as they preserve an intelligible form, then the limits 
of A and B are also equal. 


To take a case, suppose that when x increases without limit, A 
has the limit P, and B has the limit Q ; then P must = Q. To 
prove this, suppose that 

A = P + a B = Q4-5 

then, by increasing x without limit, a and h are diminished without 
limit. For, if not, and if a had the limit a, then the limit of A or 
P + fl would be P+«. But it is P; therefore a has not a limit, 
but diminishes without limit. 

But because A = B we have P + a = Q + i. If P be not = Q 
it is greater or less. If possible, let P be greater than Q, and let 
P = Q + R. Then, since P and Q are quantities not containing jr, 
R is the same, and P, Q, and R, do not change when x changes. 
We have then 

Q-{-R + a = Q + & or R =: 5 — a 

Here, then, is the following absurdity : R (a fixed quantity) is always 
equal to 6— a, which can be made as small as we please by in- 
creasing j:, because h and a diminish without limit as x is increased 
without limit. Therefore P = Q + R is absurd. In a similar way 
it may be proved that P = Q — R is absurd. Therefore P = Q. 

Theorem II. When x diminishes without limit, the way of 
finding the limit of an expression is to make j: =0, provided, 1st, all 
the results be intelligible; 2d, that the number of operations be not 

For instance, it is clear enough that 1 -j- 2jr + Sjt*, when x dimi- 
nishes without limit, has the limit 1 -|-0 -|-0 or 1. And, perhaps, the 
student may think it clear that the limit of 

1 -|. X + x^ + X^ + X* + ^c. continued for ever 

is 1 + 0+0+0 + 0+ &c 

or 1, when x diminishes without limit. But here we must make him 
observe, that when we take x small, though each of the terms ^, j^, ar^, 
X*, &c., may be small, yet their number is unlimited. And though 
we know that when a certain number of terms is added together, each 
of which may be made as small as we please, that their sum can be 
made as small as we please, yet we do not know the same of an un- 
limited number of terms. 

Theorem III. When x increases without limit, it is clear that 
such expressions as 



a + bxj a -f bx + ca^f &c. increase without limit: 
and that « + -, a + - +-i>&c. have the limit a. 
But the easiest way, in general, to examine expressions, will be to 
remember that when x increases without limit, - decreases without 



Let, then, t; = - or x = - : substitute this value of x, and, if 

X V 

possible, reduce the expression to a form in which its limit will be 
evident when v diminishes without limit, that is, when x increases 
without limit. 

For instance, what is the limit of — ^ — in such a case. 

3j:— 2 

1+1 (l+i> ^^ 

__ 1 

Let a; — ^ 3 /3 ^\ "" 3 — 2t; 



When V diminishes without limit, the preceding has the limit -. 


Let the student now prove the following cases, which we express 
in the abbreviated form. 

ax^-\-bx -^ c fljr'-l- 6.r + c a 

If X ^ (X. ; =S OC S-, ; = - 

px-^q px'+qx -^r p 

aj:'+ bx -\- c ^ 

px^-^gx -\-r 

Theorem IV. If a be greater than 1, the terms of the series, 
ff, a', a^, a*, &c., increase without limit ; or (abbreviated) a* := oc 

For a2 = a + a^— a = a + a(a — 1) 

or a becomes a* by adding a (a — 1) 

Similarly a^ becomes a^ by adding a^ (a — 1 ) 

Generally «»» becomes a'»+^ by adding a" (a — 1) 

But because a is greater than 1, a — 1 is positive, and the addition 
of a (a — 1) is therefore arithmetical increase. And o* is greater than 
a; therefore a\a — 1) is greater than a(a — 1), or the third power of « 
exceeds the second by more than the second exceeds the first. Simi- 
larly, the fourth exceeds the third by more than the third exceeds the 
second ; and so on. But if to a the same quantity be sodded as many 


times as we please, the result may be made as great as we please j 
still more if the same number of additions be made, with a greater 
quantity each time than the last. Whence follows the theorem. 

Theorem V. If 6 be less than 1, the terms of the series by b*, b^, 
h\ &c. decrease without limit, or (abbreviated) 6* ^ 0. 

It 1 

Let 6 ^ -, then 6» = — . But because b is less than 1, a or 7- is 
a a^ b 

greater than 1 ; therefore a^ may be made (Theorem IV.) as great as 

we please. Hence, — or ^ may be made as small as we please. 

9 1 

Find out the first of the powers of —— which is less than ^ ^^ j^^^ 

^ 100 looqooo. 

Am, The sixth. 
It is hardly necessary to notice, that if a ^ 1 the terms of the 
series a, a% a% &c. neither increase nor decrease. 

Theorem VI. If x be positive, and less than 1, the series of 

(1 + a?) {l-^-x+a?) (1 + a; + a;^ + a;5) &c. 

increases, but not without limit. The limit is ; that is, no term 

1 — X 

of the preceding, how many powers soever it may contain, can be as 
great as , but may come as near to it as we please. The abbre- 
viation is as follows : 

^ "— I "7" vC "f* X "I" %c» "7" • • • • "f" X 

1— jr 

more generally written 

= l-f-a;4-^* + ^+ &c. ad infinitum, 

i^^" X 

As this series is a most important part of the groundwork of all 
that follows, we shall try to establish the proposition from the method 
of its formation. We remark that when x is positive, the terms 1, 
1 4- jr, 1 -f j; 4" ^> &c. evidently increase ; and that each term is 
formed by multiplying the preceding by x, and then adding 1 . Thus, 
l + *-f** is \'\'X{1-Jtx)y and 1 + x + j:*+j^ is 1 + j'(1 + ^+^'0> 
and to on. If A stand for any term, and B for the next; then 

B = 1+Aa? 

Nowy B is greater than A ; therefore, adding 1 more than com- 
pensates the diminution which A undergoes by being multiplied by x 



(Remember that x is less than 1). But Aar =: A + Ax— A = A — 
(1— jr)A; or multiplication into x diminishes Aby(l — jr)A. This 
the addition of 1 more than compensates ; that is, 1 is greater than 

(1— j:)A. Divide both of these by 1 — x\ and is greater 

than A. But A is any term we please of 1, 1 + j, 1 -fcT 4-a:', &c.; 

therefore every one of these, how far soever we go, is less than . 

1 X 

Now we have to prove that, though we cannot find A as great as 
, we can come as near to this as we please. Remember that the 


1— j' 

way of forming the next term is always, " Multiply by x and add 1." 

Let A differ from by p ; so that 

A == », the next term isl4-Aa: = l4- px 

1 — X -^ X 1 *u * * • 

=: ■ VX ^ ^'VX ; the next term is 

1 — X ^ 1. — X ^ 

X 1 
1 -| .^px^ or »Z*; the next term is 

pa^; and so on. 

X ^^x 


Hence we can find a term which differs from r by px% where 

1— X 

n is as great as we please. But /) is a given quantity, and jr» 
(Theorem V.) diminishes without limit when n increases without 


limit; therefore /)a:»* may be made as small as we please, or- pj* 

as near to as we please. But, by continuing the preceding terms 

1 — X 

from A, we shall at last come to par». Therefore, by con- 

tinning the terms, we come as near to as we please. 

We shall now try to find whether (j: being less than 1), 

1 — X '\'X^ — a;^ -f- JT*-" &c. continued ad infinitum^ 

has a limit; or what is the nature of the increase or decrease ofl, 
1 — j:, 1 — X -\-x^y &c. Here we see alternate increase and decrease : 
but still under a simple law. To find the next terni, multiply the 
last term by x^ and take the result from 1. Thus, 


l^x + x^ = 1 — a;(l— a:) 
1 — a: + a;*— a;^ = 1 — x(l ^x -{-x^) and so on. 
Or if A and B be two consecutive terms^ 

B = l-Aa; or l+A-A(l + a;) 
That is B = A + 1 — A(l + x) 

then the next term C = B + 1 — B(l + x) &c. &c. 

But the results alternately increase and decrease; that is, suppose 
B greater than A, then C is less than B. Or, suppose 1 greater than 

A (1 + x)f then 1 is less than B (1 + j) : or is greater than A, 

and is less than B. So that the results are alternately less 

and greater than . 


Thus we have 1 is greater than 

1 — X is less than 

1 +x 



1 — X -{• X^ is greater than 
&c. &c. &c. 

Now, since x^ diminishes without limit as n is increased, we can 
take n so great that two consecutive terms 

l—x+x^— he, ±af-^ 

and 1— a; + ^*— See. ±0^'^ T^c" 

shall differ by a quantity as small as we please (for they differ by 
^x^). But we have just proved that one of these terms is greater 

than and the other less. And they differ less from any quan- 

1 -\~x 

tity which falls between tliem, than they do from each other; con- 
sequently, either may be (if n be taken sufficiently great) as near to 

as we please. 

1 -)- j: '^ 

These two results we express as follows : 

^ 1+^4-^ + ^^+ &c. ad infinitum. 

1 —X 


= 1 — X-\-X^ — x^ -\- &CC, ad infinitum. 

1 H-j: 






and r is called the sum of the first infinite series, meaning, the 

limit towards which we may come as near as we please by continual 
addition of the terms 1, or, x^, x^, &c. 

We shall proceed with this subject in chapter VIII. 

Theorem VII. If the numerator and denominator of a fraction 
diminish without limit, the limit of that fraction may be nothing, 
finite or infinite : that is, the fraction may diminish without limit, 
may have a finite limit, or may increase without limit. Neither of 
these suppositions is inconsistent with the unlimited diminution of 
both the numerator and denominator. 

Take the three fractions 

3^ — o^ a:'— a' (jr — a^ 

(■r — ay X — a x^ — a* 

By supposing j: := a, all the three assume the form -. By sup- 
posing X to approach as near as may be* necessary to a, we may 
diminish the numerators and denominators without limit. The reason 
of this is, that x — a is a factor of every numerator and every deno- 
minator, and X — a diminishes without limit as x approaches to a. 
For the three fractions are 

(■r — a)(a'-f-a) {x — a)(j:-f-fl) {x^^d){x — o) 

* {x — a){x — a) {x — o) {x — a){x-\'a) 

Divide both terms of each fraction by x — a, which gives 

x-4-rt . X — a 

X -{-a — — 

X — o x-^a 

Which are always severally equal to the first, except where a: = a, on 
which we give no opinion (see page 156). But as x approaches 
towards a, the first is 

A quantity whose limit is 2 a 

A quantity which diminishes without limits 

and therefore increases without limit. The second is 

A quantity whose limit is 2a, 

and therefore approaches without limit to 2 a. The third is 

A quantity which diminishes without limit 
A quantity whose limit is 2 a, 

and therefore diminishes without limit. Consequently, when we see 


a fraction under circumstances in which the numerator and denomi- 
nator diminish without limit, we have no right to draw any conclusion 
as to the value towards which that fraction is tending, but must 
examine the fraction itself to see whether it diminishes or increases 
without limit, or whether it tends towards a finite limit. 

Theobem VIII. The same caution is necessary as to a frac- 
tion whose terms increase without limit, or which approaches the 

oc A 

form — . For let -rr be a fraction whose terms increase without 

a B 

limit. We know that 

A __ B_ 

B " 1 


and when A and B increase without limit, -r- and -^ diminish with- 

Out limit. Therefore the same circumstances which make -^ approach 

the form — , make the same fraction (in a different form) approach 

to -. Whence the last theorem applies. 

Theorem IX. The same caution applies to the value of a pro- 
duct, in which one of the terms diminishes withput limit, while the 
other increases without limit. Let AB be such a product, which ap- 
proaches the form x oc : that is, while A diminishes without limit, 
B increases without limit. We know that 

AB = -^ 


and when B increases without limit, -^ diminishes without limit. 

Hence, as in the last case, AB, under a different form, approaches to 

the form -. 

Thus we see that the three forms 

a rk 

a 0^« 

are so connected, that any expression which gives one, may be made 
to give either of the others. 

We now take the form a°, considered, not in the absolute and 
defined sense of page 85, but as the representative of 

The limit of a" when x diminishes without limit. 



If j: = -, then when x diminishes without limit, y increases 

without limit. Let ^ be a whole number, then 

a' =^ c^ =• y/a 

Firstly, if a be greater than 1, all its roots are greater than 1 (for 
all the powers of less than 1 are less than 1). Let 

\/a = 1 +w or a = (1 +vy 

Now, y can be taken so great that i; shall be less than any fraction 
we may name, however small. For, if not, say.it is impossible that y 
should be so great as to make v less than k,- Then, v being always 
greater than fc, whatever may be the value of y, 1 + v is always 
greater than \-\-k. But, Theorem IV., y may be taken so great that 
{X+ky shall exceed any quantity we may name, and shall there- 
fore exceed a. Still more, then, will (1 -f v)* exceed a (u is greater 
than k). But (l+v)y eqtmh a: here then is a contradiction. Con- 
sequently, the supposition that v can never be made less than a given 
fraction k is not true: that is, v can be made less than any given 
fraction, or 1+v can be brought as near to 1 as we please. But 

1-)- v ^ V a, therefore, if^ increase without limit, we see that 

V^ or a*' or a* has the limit JL 

Or a^:= 1 where is used in the sense given in page 156. 

Secondly, let a be less that 1, whence - is greater than 1. By 

the last case y - can be brought as near to 1 as we please; but 

this is 1-T-Vfl, therefore Va can be brought as near to 1 as we 

Theorem X. In any rational* integral expression with respect 
to Xy if X may be increased without limit, the term which has the 
highest power of x may be made to contain the sum of all the rest 
times without limit. That is, in the expression 

ax^-\-hx^-^cx-\'e^ for example, 
let a be any given quantity, however small, and 6, c, and c, any given 

* Look at the beginning of the next chapter. 



quantities, however great, yet x may be taken so great that ax* 
shall contain 6jr*-|-cjr-i-cas many times as we please. 

The number of times and parts of a time which ax* contains 
tjr*-f cr4-c is expressed by the fraction 

axi* ax^-i~x* ax 

or .. . . : — : s or 

X 3r 
C 6 

Let — I — 3^P. Then, by increasing x without limit, p is 

X X 

diminished without limit, and will, if x be taken sufficiently great, 

become less than 1. That is, ax* contains 6j' + ca:-i-c, 


b + p 

times, or more than -r--- times. But ri—- or rr-r X se increases 

o-fl 6-j-l 0+1 

without limit, when x increases without limit. The same then does 
the number of times which as^ contains bx^-^-cx-^-e. 

For example, how great must x be, in order that we may be 
certain that the millionth part of x^ contains 1000a:*+ 500 j: +1000 
more than a hundred thousand times ? 

one millionth of a-^ ona millionth of j- 

1000 j:« + 500 X +1000 "" ,^^^ . 500 . 1000 

^ ^ 1000+ +— 2- 

X x^ 

Now, if dp be 1000 or more, 1 -r- is less than 1. There- 


fore, in this case, the preceding fraction is greater than one millionth 
of x-r-1001, or than 



If we take j: = 1001000,000X100,000 or 100100,000,000,000, 
the preceding fraction becomes 100,000. Hence, the millionth part 
of x* is greater than 100,000 times ( 1 000 x' + 500 j;'+ 1000). We 
do not say that this is the least value of x which will answer the 
conditions, but that this, or any thing greater, will do so. 

Theorem XI. In any integral and rational expression with 
respect to x, if x may be diminished without limit, the term con- 
taining the lowest power of x may be made to contain the rest of 
the expression as many times as we please. For instance, in 

1 1 

——j: +1000,1^+100^:3 we may take x so small that x shall 

contain 1000 r' + 100 x^ as many times as we please: or in 



ax*-\-bx^-^cx'\-e, x may be taken so small that e (or cj®, which 
is the terra containing the lowest power* of x) shall contain 
ax^'j-bx''\-cx as often as we please. This last is evident in this 
particular case, because e remains the same when x diminishes with* 
Out limit, and ax^-\-bx^-^cx diminishes without limit. Therefore, 
the second may become less than any given fraction of the first. 
Now, take a case in which there is no power of x so low as j:'' ; such 
as ax^-^-bx'+cx. Here x may be taken so small that ex shall 
contain ax^-^bx* as many times as we please. For the number of 
times and parts of times which ex contains a j:* + ^^ is 
ex c fixed qiiantity 

. QJ« — 

ax*'\-bx* ax'-j-bx one which diminishes without limit 

which latter increases without limit when x is diminished without 

Hence it follows, that when x increases without limit, x^y x^^ x*f 
&c. not only increase without limit, but each of them increases without 
limit with respect to the preceding ; by which is meant that a^ in- 
creases so much faster than x^, that x^ will come at last to contain jr* 
as many times as we please. Similarly, when x diminishes without 
limit, we find that x*, x^, x^, &c. not only diminish without limit, but 
each of them diminishes without limit with respect to the preceding ; 
by which is meant that s^ diminishes so much faster than x*, tliat a^ 
will come at last to be as small a fraction of x* as we please. These 
notions are sometimes abbreviated into the following phrases, which 
it must be remembered are not intelligible, except as abbreviations. 

Abbreviated Phrases. 

1. Of two infinitely great 
quantities, one may be infinitely 
greater than the other. 

Rational Meaning. 

1 . Of two quantities which 
increase without limit, one may 
increase so much faster than 
the other, as not only to increase 
without limit absolutely speal^- 
ing, but to increase without 
limit in the number of times 
which it contains the other. 

* The algebraical series of whole powers of x is 
.... x-^ x-^ x"^ x^ x^ x^ x^ . . 
answering to 

"^ ^ X ^ xar'a:^ (see page 85.) 



Abbreviated Phrases. Rational Meaning. 

2. Of two in6nitely small 2. Of two quantities which 

^^^tities, one may be infinitely diminish without limit, one may 

^ than the other. diminish so much faster than 

the other, as not only to dimi- 
nish without limit absolutely 
speaking, but to diminish with- 
out limit in the fraction which 
it is of the other. 

Let the student now try if he can explain the following Problem. 




If A and B move together towards the line CD, and if A move 

^^1 such a way that A Q is ^ of an inch when BQ is i an inch ; 

^bat AQ is J of an inch when BQ is | of an inch ; or generally that 

-^Q is a:' inches* when BQ is j: inches: and if a microscope be 

1)Iaced over the figure which grows in magnifying power as A moves 

towards Q, in such a way that the increase of magnifying power just 

compensates the real diminution of AQ, so that AQ always appears 

of the same length ; then B, instead of appearing to move toward Q, 

will appear to move away from Q. 

* It is usual to say x inches, when x is less than 1, or when x inches 
is really a fraction of an inch ; in which case x of an inch would be more 
agreeable to analogy. Returning to the consideration discussed in the 
note to page 40, it will be useful to observe that the idiom of our 
language makes the connexion between multiplication by 4 and multi- 
plication by ^, less obvious than it might have been. We only say 4 of 
before an article or pronoun, " four of the men," " four of them." But 
we always say, " one fourth of six," " one fourth of an inch." If it had 
been idiomatic to say ** four of six," or ** a four of sixes," for 24 ; and 
" four of an inch " for '* four inches," the propriety of extending the 
term multiplication to fractions would have been much more obvious. 






Previously to commencing this subject, we shall make a classi- 
fication of the expressions we have obtained. All the terms in use 
are usually relative to some particular letter; in what follows we 
shall suppose this letter to be x. We shall proceed to explain the 
following table. 


Common Algebraical. 




Integral. Frac- 
' » tional. 




Integral. Frac- 
* » tional. 


Inverse Circular, 

Any expression which contains a: in any way is called a Junction 
ofx: thus a-^-Xf a + 6r^ 8cc. are functions of x; they are also 
functions of a and b, but may be considered only with regard to x. 
All expressions which contain only a finite number of such ex- 
pressions as are treated in the preceding part of this work, are called 
common algebraic* functions, except only where x is an exponent. 
Thus s/a-\-x^, a^4-6, &c. are common algebraic functions, but 
a' is not. And we do not know whether 1 -j- a: -j- j;*' -f- &c. ad in- 
finitum (page 1 59) is a common algebraic function of x or not, until 
we have found that it is the same as l-f-(l — x"). All other functions 
of j: are called transcendental functions; such are a' and all functions 
containing it; such will be (when we come to define them) the 

* Usually, algebraic functions ; but transcendental functions are 
certainly also algebraic, that is, considered in algebra. 


^^^garithm of Xf the sine and cosine of x in trigonometry, and many 
^^^thers. A function which contains a* is called an exponential* 
^^m^inction of Xf one which contains a logarithm, a logarithmic func- 
^- ^on, &c. 

Common algebraical functions are divided into rational, which 
mtain only whole powers of j:, as a+^j ax-^-j-ft, &c. ; and irro- 
^ionalf which contain roots or fractional powers of Xy as ax^-\'hy 

"^Vo'-j-J^* &c. 

Rational and irrational functions of x are both divided into 
"^integral, which contain x only in numerators, as 

, x^ . a 4- wx 

X A — and r- : 

^ a a + 6 ' 

and Jractional, which contain x in denominators, 

as 7 — ; — 5 and i-r=^ 

bx-j'x' c + Vj* 

Integral functions are divided into monomiaUy which contain only 
one power of x, as x^y ax*y 's/bxy {a-\-b)x*; binomialsy which con- 
tain two distinct powers of x (j^ included) as a-{-bx or ax'^-^bxy 
cjr'-f-vx, mx*-}-nx^, &c. ; trinomialsy containing three distinct 
powers; quadHnomials containing four, &c. The two latter terms 

are little used : all expressions of more than one term are called 


Integral and rational functions are divided into those of the first, 

Second, third, &c. degrees, according to the exponent of the highest 

power which is found in them. Thus 

Ct -h 6 a: is a rational integral function of x, of the first degree. 

U'\-hx-\- ca^ of the second degree. 

&c. &c. 

The term a, if written ax^y is of no degree with respect to x. 

rp, . . n-f V^ log c-4-a'c* 
The expression t — ■ ^^^r-^ 


* The letter principally considered is an exponent, 
f The meaning of log. c, or logarithm of c, will be afterwards 




is a rational integral trinomial function of €t 

.... rational fractional ^ 

.... irrational integral O 

.... irrational fractional ^ 

.... exponential ........ ^ 

.... logarithmic C 

Rational and integral functions are generally arranged, so that 
the powers of x may rise or fall continually in going from left to 
right. Thus, ax + 6 — cx^ is never so written, but either 

-^cjd^-^ax + b or b + ax—car^ 

in the first case it is said to be arranged in descending^ in the second 
in ascending, powers of x. 

Thus a—bx^ + cx—x^—x^ 

should be a + ca;— (& + l)x^— ar* 

or — a;^— {b-{-])x^ i-cx + a 

The most important class of functions is the rational and integral, 
contaming those which appear rational and integral, but of which it 
cannot be known whether they are algebraical or transcendental, 
owing to their containing an infinite number of terms. Such are 
the forms 

a + bx + cx^+ -^-paf'^-^qx^ 

where a, b, c are not functions of Xy and n is a whole number ; and 

a + bX'\'CX^+ ex^ + + Sec od infinitum. 

The reduction of expressions to such forms is one of the principal 
branches of the subject. We shall call the first generally a polynomial, 
the second an infinite series. 

Definitions. In multiplication of polynomials, the several pro- 
ducts formed in the process may be called subordinate products. Thus, 
in multiplying a-\-x by b'\-x, the subordinate products are ah, ax, 
bx, and x^, A term of a polynomial is all that contains any one 
power of a:; thus, the preceding product at+aar + ^.r + j:* is not 
said to be of four terms, but of three, namely, abjia-^- b)x, and x'. 

Theorem. In the product of two polynomials, there must be at 
least two terms, which are subordinate products, and not formed by 
two or more subordinate products. 

Suppose we multiply ax-^bx^-^cx^ and px*-\-qx^. It is plain 



that no other subordinate product can contain so high a power of x 
as cj*xg^9 or cqj^f nor so low a power of a? as axxpj^i or apjc^, 
because, in these, both exponents are the highest in their expressions, 
or the lowest. Consequently, these must be terms of the product, 
which is, in fact, 

apar^+{aq + bp)3fi+(bq + cp)x'' + cqafi 
having four terms, two of which are the simple subordinate products 
already noticed. 

Algebraical division differs from arithmetical in this, that in tlie 
latter we wish to ascertain whether a whole number P can be made 
by taking another whole number Q a whole number of times; in the 
former we inquire whether a polynomial function }f x, P, can be made 
by multiplying another polynomial Q by any third polynomial. For 

instance, to divide Bx^-^1 by 2x4-1 > is the following: To reduce 


*- — --— if possible, to a simple polynomial. The way of treating this 

iw X "^ X 

question will also give that of treating any other. 

If possible, let So^ + l be made by multiplying 2x'\-l by 
«i4-6j;4*cjr' + ej?*+&c» Now, firstly, this latter expression cannot 
go higher than ex* ; for, if it did, say to ex^, we should have, by the 
last theorem, ex^x^x or 2ex* in the product. But that product is 
SjF'-i-l, in which x* does not appear; consequently, ejr^ and higher 
terms are not in the polynomial required, which is therefore of the 
form a-\'bx-\-cx^. We have then, if our question be possible. 

We have proved that 2xxcx^ must be a term* of this product; but 
it can only be 8 a?*, therefore 2 j: X c j?' := 8 x^, or c j' = 8 x^-~2 x = 4 x*. 
Consequently, Bx^-\-i = {2x +l){4x^ -^ bx + a) 

= (2x+l)4r' + (2x+l)(6x + a) 
8^4-1— (2x+l)4x« or — 4ar«4-l = (2xH-l)(6x4-«) 

of the latter product, 2xxbx must be a term; but it can only be 
— 4*', therefore bx = — 4x^-7- 2x = — 2x, or 

-4ar^+l = (2x + l){--2x+a) 

= -^2x(2x + l) + (2x + l)a 
-4a;2 4.H-2a:(2a:-f 1) or {2x + 1) == {2x +l)a 

* This being a subordinate product, which cannot be altered or 
destroyed by any other subordinate product. 


This last equation is made identical by a=l ; therefore 4 jr' — 2x+l 
is the polynomial by which ix-^-i being multiplied^ the product is 
8 j' + l : as will be found by trial. 

The steps of the preceding process may be arranged after the 
manner of division in arithmetic ; the only difference being, that in- 
stead of finding a new term in the quotient by trial of the left hand 
figures of the divisor and dividend, it is found by dividing the left 
hand term of the dividend or remainder by that in the divisor. As 
follows, in which the same question is solved in the two different 

2ar + l)8a;3 + l(4x«-2a;+l 1 + 2a;) 1+8x^(1- 2a; + 4a:« 
8a;» + 4x2 l-i-2a7 

— 4a;«— 2a7 — 2a?— 4x« 

2a: + l 4a;2 + 8a;* 

2a; + l 4a?«+8a:» 

Great care must be taken to preserve the same order of arrange- 
ment throughout, either in ascending or descending powers of x. 
The following is the general theory of this process : 

First, it is evident that the sum, difference, and product of rational 
polynomials, are rational polynomials. Let P and Q be two rational 
polynomials, from which it is desired to obtain V in such a way that 
P ^ QV. Here P is the dividend, Q the divisor, and V the quotient 
to be found. Assume any convenient polynomial or monomial A, 
multiply it by Q, and subtract the product from P, which gives 
P— AQ. Call this R, so that 

P-AQ = R (1) 

Assume any other polynomial A' ; repeat the process with R (instead 
of P) and Q. Call the result R'. 

R-A'Q= R' (2) 

Assume a third polynomial A", and let 

R'-A"Q = R" (3) 



The object is to simplify the remainder at every step, so as to 
reduce it at last to a form in which we may see one of these two 
things ; either how to find a new polynomial which shall reduce the 
next remainder to 0, or that this is impossible. Suppose, first, that a 
new polynomial or monomial A!" can be found, which will reduce the 
remainder to 0. 

R-«A"Q = (4) 

We have, then, from the different equations 

P = AQ + R = AQ + A'Q-hR'=AQ+A'Q + A"Q + R'' 

= AQ + AQ + A"Q-i-A"Q = Q(A + A'-f A'+A"0 

so that A -f A'-j- A"4- A''' is the polynomial required. Suppose that 
instead of (4), we have 

and suppose it to be evidently useless to attempt to continue the 
process further. We have then 

P = AQ + A'Q + A"Q + A "Q + R" 

(-f-)Q g = A + A'+ A"+ A"' + J- 

, , . , . V a more simple 

= rational polynomial 4- < ^ , p 

^ ^ ^l fraction than ^ 

Example. To reduce « . ^ to a more simple form 

x«+2a?*= AQ 

-2ar*Hrl = R, A = ^=4^ = - 2j:« 
-2a;*-4a;» = A'Q 

4x^ + 1 = R', A"=^= 4x 



-8a;« + l=R", A"'=-^=-8 

16a: + 1 = R 





It is useless to carry the process further, and we have 

From the preceding, we may deduce the following theorem, which 
is useful in many parts of mathematics. 

If P and Q be two rational polynomials of which P is of the 


higher degree (or dimensionf as it is frequently called), then jz 

can be reduced to the form O -f -pr where G and H are rational 

polynomials, and H of at least one dimension less than Q. 

£xERCiSE. If, in the preceding process, the remainders be 
severally multiplied by B, B', B'', &c. before using them, then 

Q " ' B ^ BB' ^ BB'B' ^ BB'B"Q 

The preceding process may be used in an infinite number of 
different ways; for though it is only convenient to employ it as in 
the preceding example, yet in the reasoning P, Q, A, A', &c. may be 
any quantities whatsoever. As in the following example, 

P = 1 Q = 1+a: 

l + x) I (let A = ar 

x + x^ 

1 — ar — j;«= R, 

let A'=a^ 



TT = = X + X^-{ 

-j: — 2x»— j:* 

But this would in most cases amount to no more than an arbitrary 
method of adding or substracting fractions. When, however, the 
process of dividing the left-hand term of the remainder by that of the 
divisor is followed, the result will generally be a symmetrical, and 
often a useful, developement. For instance, we thus obtain 

^ ^^l — x + x^—x^+x^— ^ 

l + x ' ' 1-f JC 

--1- = l + ar + ar^ + or^H- -^ 
1 — j: 1 — jt 


_L. = 1_1+J._J. + ±._L 

jr + 1 X X* x^ x^ x^ X'\-\ 

^ r= 1 j-2a?4-3a;g+4g^+ ^^'""^^ 

By this method we can tell whether either of the preceding series 
of terms continued without limit will approach a limit or not. For 
instance, we see that 

\-\-x^3t^-\-a^+ ^3if (A) 

becomes when : is added to it. If then x be less than 1, 

1 — X 1 — X 

since j^+^ diminishes without limit (page 159) when n is increased 
without limit, the sum of the terms in (A) continually approaches to 

as was shewn in page 1 60. 

Let us, for the present, denote by (P) that P is a rational poly- 
nomial ; and by (P) + (Q) = (P + Q), that P and Q are rational 
polynomials, whence their sum is a rational polynomial. We have 
then, always, 

(P) + (Q) = (P + Q), (P)-(Q) = (P-Q), 

(P) X (Q) = (P Q) ; and ^ = (-^^ in certain cases. 

Every polynomial which divides (P) without remainder is called 
2L factor o( P; thus x* — 1 =(jr + l)(ar — 1) and j: + 1 and x — 1 
are factors of j?* — 1 . It is evident, from page 1 71 , 

1. That no polynomial can have a factor of a higher dimension 
than its own dimension. 

2. That if the polynomial be of m dimensions, and one of its 
two factors of ^ dimensions, the remaining factor must be of m — p 

Thus in 

x*—l = (x—l){a^ + x'^-\'X + \) 
= (a;2-.l)(a:2 + l) 

the polynomial being of the fourth degree, its factors are in one case 
of the ^rs^ and third (1 + 3 = 4), and in the other of the second and 
second (2 + 2 = 4). 

In what follows we speak only of rational polynomials, and by 
rational division we mean tliat the preceding process leaves no 


remainder. The following theorems are evident consequences of 
what goes before. 

1. Rational division is impossible unless the dividend be at least 
as high in degree as the divisor. 

2. Where the dividend is higher than the divisor, and rational 
division is still impossible, the remainder is of a lower degree than 
either the divisor or dividend. 

3. If the dividend be of the mth and the divisor of the nth degree, 
the quotient is of the (m — n)th, and the remainder not higher than 
of the (n — l)th degree. (For so long as the remainder is as high or 
higher than the divisor, the process can be continued). 

4. The dividend being P, the divisor Q, the quotient* A, and the 
remainder R, then 

P = AQ+R or I = A+l 

5. Every quantity which rationally divides M and N rationally 
divides their sum, difference, and product. Let Z be the divisor; 

y = (A) -2=(B) — Z_ = (A + B) 

51=:^ = (A-B) ^ = ABZ = (ABZ) 

6. Every divisor of P and Q (in 4.) divides R, and every divisor 
of Q and R divides P, &c. so that no two of the three has any 
rational divisor which the third has not. For instance, let Z divide 
P and Q rationally, then 

T " '*"'"' °"= (I) f = (f ) 

= ( y — j; but this is -=•, which is therefore rational, or Z 

also rationally divides R. By similar reasoning the other cases 

7. The highest common divisor of P and Q is therefore the 
highest common divisor of Q and R. 

* Not strictly a quotient, unless the remainder he nothing. It is what 
comes in the quotient-part of the process in trying this point. 



8. If one factor of a product be not divisible by Xy then those 
powers of Xy and those only, which rationally divide the other factor, 
will rationally divide the product. For instance, in (jr* -f a) (■'' + 6 x*)y 
since the lowest term is Qax^, and since no power of x higher than 
that in the lowest term will rationally divide an expression, it follows 
that j:^ is the highest power of x which rationally divides the product. 
Bttt jr* is the highest which rationally divides x^-\-Qx*. 

9. If an expression be rationally divisible by a power of j:, the 

quotient is divisible by every divisor of the first, which is itself 

indivisible by any power of jr. For example, x^ — x divided by x 

gives 3^ — l,jr— 1 is a divisor of the first, and, therefore (were it not 

known otherwise), is a divisor of the second. 

To prove this theorem, let jf*P (for example) be an expression, 
vvhich is divisible, say by j: + ^ 9 that is, let 

^ = (A) ^(P) = (A)(x + l) 

consequently (8.), (A) is divisible by x^ ; 

^r A = (B) that is (A) = X^{B) 

Therefore, X^F) = x^{B)(x + l) 

(P)= (B)(a:+1) or ^^ = (B) 

that is, P (as well as x^F) is divisible by x -f- 1, and the same may be 
shewn of any other divisor of x^'P, 

The method of finding the highest common divisor of two rational 
polynomials is now exactly similar to that of finding the greatest 
common measure of two whole numbers in arithmetic. For example, 
required the highest common divisor of x^ — x and 3x^ — 3x*, First 
separate the monomial factors; that is, put the expressions in the form 

x{x^^l) and 3j;4(a;4-l) 

Neglect the monomial factors for the present, and proceed to find the 
highest common divisor of 

afi^l = p and a^^l = Q 

a;4— l)ar^— l(a; 
Rem. or— l)ar*— l(^a^ + x^-\-x-i-} 



By (7.) the highest divisor of j:*— 1 and j:* — 1 is also that of 
X* — 1 and X — 1; and, since x — 1 divides the former, and also is 
the highest divisor of itself, it is the highest divisor common to the 
two, and, therefore, the highest divisor of .r*— 1 and x* — l. The 
original expressions have also the divisor x; consequently, j:(j? — 1) is 
their highest common divisor. 

Any power of x may be thrown out of a remainder by division, 
from theorem 9. For instance, in finding the highest common divisor 
of 1 — j:3 and 1 — x*; the first remainder is x^ — jr* or x^(l — x^); 
but 1 — x' has all the divisors of x^ — x*, except those which are 
powers of x. But 1 — j:* and 1 — a:* are neither divisible by a power 
of X ; consequently, 1 — x* contains all their common divisors, as well 
as x^ — X*, and the former may be used for the latter in any division. 

The divisor may be taken any number of times which may be 
convenient, before using it as a new dividend. For instance, in 
finding the greatest common measure of x' — 2x4-1 and x* — 1, the 
first remainder is 2x* — x — 1; before dividing by this,* it will be 
convenient to multiply x* — 2x + l by 2, and no new common divisor 
will thus be introduced. The whole process in the latter case may be 
as follows, wiiich, though not the shortest possible, will illustrate the 
methods to be applied to more complicated cases. 

X*— 2x4-l)x' — l(x 

x^ — 2 x' 4- * 

2j.«—jr — l)2x«— 4x4-2(1 
2x* — X — 1 

— 3x4-3 
Divide by —3 x— l)2x*— x— l(2x + 1 

2x«— 2x 

X — 1 
X— 1 

Therefore x — 1 is the highest common divisor. 

* The division might be carried one step further before using the 
remainder; but either method answers equally well. 




We have already seen (page 160) that the sum of the terms (j being 
less than 1) 

how far soever it may be carried, never can exceed or come up to 
l-r-(l — x). This expression is called an infinite series, and the 
fraction 1 -f- (1 — x) v^^hich (page 150) might be called the limit of 
the sum, is called the mm. 

Definition. By the sum of an infinite series is meant the limit 
towards which we approximate by continually adding more and 
more of its terms. 

A convergent series is one in which such a limit exists, that is, 
in which we cannot attain a number as great as we please by summing 
its terms : a divergent series is one in which there is no limit to the 
quantity which may be attained by summing its terms. The follow- 
ing series are divergent, and all but the last, evidently so. 

1+1 + 1 + 1+ &c. 1+2 + 3+4+ &c. 

1+2+4 + 8+ &c. l + ^ + l + i+ *^*^* 

In an infinite series, we must know the connexion which exists 
between each term and the next, otherwise we cannot reason upon it. 
For, as we cannot write down all the terms, it is only from knowing 
the connexion between successive terms that we can be said to know 
of what series we are speaking. So that an infinite series with no 
law of connexion existing between its terms, has no existence for the 
purposes of reasoning. 

The student might perhaps imagine that the law is immediately 
perceptible when the first four or five terms are given, and an obvious 
connexion exists between them. For instance, he would suppose 
that the following series 

1+1 + 1+1 + 1+1+1+1+1+ &c. 

* See Arithmetic, article 197. 















if it be to be continued according to a law actually existing among 
the given terms, must be a succession of units added together. But 
this is not the case ; the preceding series might be continued in an 
infinite number of different ways, each different method following 
a law which actually exists among the given terms as they stand. 
For instance, the preceding series might be thus continued : 

9th term. 10th. 11th. 12th. 13th. 14th. I5tb. 16th. 

1 1 2 3 4 6 7 10 &c. 

the law of which is, that the (n + 0^^ exceeds the nth by the tens 
figure of the sum of the first n terms, which requires that the terms 
should remain equal until their sum has a figure in the second 
column. The following series of terms have laws which we leave 
to the reader to detect. 

26 3? 36 42 adinfin, 3'^'"^'^^"^'^ 

10 9 10 9 a(/f«>.-h^^/,-t'/,-^^/y 

15 21 30 39 43 62 61 70 
94 103 109 109 109 adinfin. 

If we attempt to deduce general propositions from a few par- 
ticular cases, as, for instance, the law of a series from that of a few 
of its terms, we are liable to error. All that can be derived from 
observing a few cases is a strong presumption, high probability, or 
great likelihood, that the law observed is always true. But that 
which is beforehand very likely, does not always turn out, on ex- 
amination, to be true : as in the following instance. Take the series 
of numbers 1, 2, 3, 4, 5, &c. multiply each by the next higher, and 
add 41 to the product, as follows : 

1 X 2 -f- 41 = 43 5 X 6 H- 41 = 71 

2x3 + 41 =47 6x7 + 41 =83 

3x4 + 41 =53 7x8+41 =97 

4x5+41 =61 8x9+41 = 113, &c. 

On examining the series of results 

43, 47, 63, 61, 71, 83, 97, 113, &c. 

we see that all of them seem to be prime* numbers, and hence we 

* A prime number is one which does not admit of any divisor except 
1, and itself. The series of prime numbers is 

1, 2, 3, 5, 7, 11, 13, 17, 19, &c. 



have a very strong reason for presuming tbat tliis will continue to be 
the case, tbat is, we suspect the following to be true : if jt be any 
whole number, i'(jr + 1)-f 41 is a prime number. And on conti- 
Duing the series, we actually find prime numbers, and nothing but 
prime numbers, up to 39 x 40-f 41 or 1601. But, nevertheless, the 
next term, or 40 X 41 +41, is evidently not a prime number ; for it 
is (40 + 1)41, or 41 X4l. 

To avoid the continual necessity of expressing the law of a series, 
we always mean, in future, that, where a simple law appears among 
the few terms which are written down, that law is to be the law of the 
series, unless some other law be mentioned. Thus 1 + jr +x' + &c. 
implies that the succeeding terms are r* + j* + j:* + &c. 

Definition. The general term of a series is the algebraical 
expression for the nth term, as will be better understood from the 
following cases. 

First few terms. 

nth, or general term. 

1 + 1 + 1 + 1 + &c. 


1+2+ 3+ 4+&C. 


2+3+ 4+ 5+ &c. 

n + 1 

+ 1 + 2 + 3 + &c. 

n— 1 

1+4+ 9 + 16+ &c. 


4-|.9 + 16 + 25+ &c. 

(n + l)» 

* + drJ^r» + jr*+ &c. 


1+jr +jr2 + a:»+ &c. 


gm ^ jm+1 _|- ^+2 ^ jm+3 ^ 


j>m+n— 1 

*+i + T+T + '^- 


^ + '+2<3 + *"=- 


1.2.3....(n— 1) 

In the last series, the first terra 

is not included in the g( 

rm. HR eriven. For if n = 1. 

the Q 

reneral term becomes 

. „ - '"h 

which is not true. Properly speaking, the general term is n factors 
of the following product : 

^ Su X *Mf X 


Theorem. The series a + 6 + c + e +y+ &c. is the same as 
the following: 


i a oa c a e c a ) 

The student will have no difiSculty in proving this. Let the ratio* 
of each term to the preceding term be denoted by the capital letter of 
tl)e numerator. 

^ = B i = C - = E ^=F&c. 

a b c e 

Then a + b-\-c+e +/+ &c. is 

a{l+B + CB4-ECB + FECB + &c.} (1) 

If every one of the ratios, B, C, E, F, &c., be less than some 
given quantity, say P, then 

a(l + B) is less than a(l + P) 

a(l+B + CB) a(l+p + pp) 

&c. &c. 

or the sum of any number of terms of (1) is less than that of the same 
number of terms of 

a(l+P + P2 + PH&c.) (2) 

If, then, P itself be less than unity, (1) must be a convergent 
series; for no number of terms of 1 + P -f P* 4- &c. can then exceed 
1 -J- (i«-P), consequently, no number of terms of (2) cfe exceed 
a -f- (1 — P) ; still less can any number of terms of (1) exceed the 
same, because the terms of (1) are severally less than those of (2). 

Consequently, a series is always convergent when the ratio of 
any term to the preceding term is less than some quantity^ which is 
itself less than unity. It is sufficient that this should happen after 
some certain number of terms : for, say that the first hundred terms 
are increasing terms, yet if no summation of terms after the hundredth 
will give a result exceeding, say 50, and if the sum of the first 
hundred terms be, say 1000, then no summation whatever will give a 
result exceeding 1050, or the series is on the whole convergent, or, 
properly speaking, begins to converge after the hundredth term. 

* r- is the algebraical synon3nne for what is called in Euclid " the 

ratio of a to 6," and the geometrical term, which is a highly convenient 
one, is frequently adopted. 


Exampk. l + l+l+^^+^^ + kc. 

U coDTergent. For here we have 

5- = 1 i=l 1=1 t]^==l&c 
a b 2 c 3 e 4' 

90 that each ratio after the second is less than -, which is less thau 



[As the limit of this series is an important number in algebra, we 

shall proceed to find it as fiir as 10 decimal places, using eleven 

places to insure the accuracy of the 10th. Let the terms be called 

^1) f^v &c., then we have 

111 1 1 , 

fli = J fifg = 1 fls = -a£ a^ = -cfs % = -a4 &c. 

Ojssl 1 00000000000 

02=1 1-00000000000 

03 = -Oj 0-50000000000 

04=^03 -16666666667 

Os = icf4 -04166666667 

Og = -05 -00833333333 


Wj =3 iflg •00138888889 


Og = -^7 -00019841270 

Oq as -^8 -00002480159 


fllO"* -«9 -00000275573 

ail= iflio -00000027557 

ai2=5= — «ii -00000002505 

ai3= JL^jg -00000000209 


ai4= r^-aia -00000000016 


ai5= j^flM -0000000000 1 



This is correct to the last place ; in fact, the sum of the series 
lies between 


and 2-71828182846 

but nearer to the latter. The letter c (and sometimes e) is used to 
denote the limit of this sum ; or we say that 

8=1+1+^+ :rT+ ^^' (= 2-71828182846verynearly.) 

The above series is said to converge rapidly.^ 

Theorem. The series a + 6 -f c -f &c. is always divergent, when- 

b c 
ever its terms are so related that -, 7-, &c., are all greater than unity, 

a b 

or continue so from and af^er a given term. 

As the demonstration of this theorem is very like that of the last, 
we leave it to the student. 

Theorem. The series a — 6-|-c — c + fitc. is convergent when- 
ever the terms decrease without limit ; that is, when a is greater than 
h, b greater than c, &c., and when some term or other of the series 
must be less than any fraction we may name. 

Let a, b, c, e, &c. be a series of decreasing terms, as in the 
theorem, then 

(a — i) + (i — c) + (c — e)+ &c. 

must be a converging series, for the sum of the first two terms is 
a — c, of the first three, a — e, and so on. Now, since a,b,c,c .... 
decrease without limit, a — c,a — e, &c. is a series of increasing terms 
which has the limit a. Consequently, the series made by taking 
only alternate terms of the preceding, must have a limit less than a. 
But that series is 

a — i + c — e + &c. 

whence the theorem is proved. 
Hence we know that 

1 — - + i — i+ &c. 

is convergent, with a limit less than 1. 

Theorem. If any given quantity P be greater than any one of 
the series of ratios 

* ^l-^f &c. 
a b c e J 


then the series 

a'hbx'{-ca^+ex^+fx^+ffr^+ &c (A) 

is convergent whenever x is less than r^. 
For, since the preceding series is 

a f 1 + * a; + ^^ .«« + - . X • -;r» + &«.} 

\ * a b a c b a^ j 

b e 
and since P is greater than -, r, kc, the preceding series will be 

b c 
increased by writing P instead of -, ?-, &c. But it then becomes 

a{l+Pj: + P*a:«+P^a;*+ &c. } 

or a{l+(Px) + (Par)« + (Px)^+ &c.} 

which (page 159) is convergent if Pj: be less than 1, or j;* less 

than -5-. Still more is the original series convergent under the same 

circumstances, because its terms are severally less than those of the 
last series. 

If P be greater than any one of the ratios af^er some given ratio, 
the series converges from and after the term which gives that ratio, 
livhenever Pj? is less than 1. Suppose, for instance, that the thou- 
sandth and following terms of the series (A) are 

Ax»9+Bar'«» + Ca;i<»'+ &c. 


Aj999 |H. |.;P+ ^ |.a;2+ &c.} 

Then, by the preceding reasoning, if P be greater than any of the 

'B C 1 

ratios -j-, -^j &c. the preceding series converges if j: be less than ~. 

For instance, take 

l+2ar + 3a;«+4a:»-f &c. 

2 3 4 5. 
ratios T « « T fitc. 

12 3 4 

2 is greater than any of these ratios after the first; consequently, 

this series converges from the second term if x be less than -. The 
hundredth and following terms are 

100a:^ + 101xi«^ + 102a:^<>^+ &c. 

R 2 


101 102 103 p 

100 101 102 

of which — - is greater than any except the first. Consequently, 

this series converges from the hundredth term, if x be less than 

1 "T- rrr: or --— . Similarly it may be shewn that this series is con- 
100 101 

vergent whenever x is less than 1, though the term at which con- 
vergency begins may be made as distant as we please, by making 
X sufficiently near to 1. 

As a second example, take 

•I ** x^ X* 

X -r ^ T 2 ^ 2.3 ^ 2.3.4 ^ 

ratios 1 x - - &c. 

2 3 4 

Since these ratios continually diminish, and without limit, a point 
of the series will come, after which they will all be less than any 
given fraction m, however small it may be. But if m may be made 

as small as we please, — may be made as great as we please. 

Therefore this series is convergent for every value of x, however 
great; though the greater x is taken the more distant will be the 
term at which convergency begins. 
As a third example, take 

1 +2ar+2.3a:« + 2.3.4a:3+ &c. 
ratios 2, 3, 4, 6, &c. 

and as these ratios increase without limit, there is no quantity which 
is greater than them all. Consequently, no value can be assigned to 
X for which this series must necessarily be convergent. The following 
theorem may easily be proved in the same manner as the last. 

b c 
Theorem. If P be less than any one of the ratios - y, &c.; 

^ a ' 

then the series 

a + 6a; + ca;2+ ^^ 

must be divergent for every value of x greater than — . 

In this way the series in the last example may be shewn to 

diverge from the second term for every value of x greater than -, 



from the third for every value greater than -, and so on; so that 

there is no fraction so small that the series shall not diverge from and 
after some term by giving x that value. 

[Such necessarily divergent series never will be found in practice : 
tliey are introduced here as a warning against applying to all series 
general conclusions drawn from series which may be made con- 

In future, unless the contrary be specially mentioned, in speaking of 
the series a-f-bx4-cx*-|- &c., we only mean to speak of series which 
may be made convergent. We suppose all the terms positive. 

Theorem. Every series of the form a-\- bx-\-cx^'{- &c. has this 
property, that x may be taken so small, that any one term shall 
contain the aggregate of all the following terms as often as we please. 

For instance, by taking x sufficiently small, we may make ex* 
more than ten thousand times ex^ -\-fx* + &c. Let x^ be the greatest 
value of X which makes e +fx-{- &c. convergent, and let the sum in 
that case be S. Then, for every value of x less than x^, e -{-fx + &c. 
is less than S. Now, cx^ contains ex^ -\-f^* + &c. 

,2 ^ 


or . ^ „ . » — or 

ex^-\-fx*-\-&cc, ex -\-fx^ -\-^c. x(e -{-fx-^-hc) 

times or parts of times. Take x less than x^, so that S is greater 

than e -{-fx + &c., or 

c - . c c.r* 

— ^ less than — ; — —r. — r-z — ; or 

xS x(e -^fx + &c.) ex^ -\-f^* -\- &c. 

Now, c and S being fixed quantities, x may be taken so small 
that c ^ xS shall be as great as we please; and still more 
cx* 'T' {€x^ +fx* + &c.), which is greater than c ^i- xS, Whence 
the theorem is proved. 

Example. How small must x be taken, so that we may be sure 
the fourth term of 

l+2a: + 3a;2+4a:^ + 5a:*+ &c. 
contains the sum of all that follow 1000 times at least. 

The whole of the series after the fourth term may be written thus : 

5a:*|l + ja;+^.|a^+&c.J (A) 

and -r is greater than any of the succeeding ratios; consequently, we 

increase the preceding by altering it to 


6ar*|l-f Tar+ -.-a:*4- &c.| or — (B) 

(see page 182.) We have then to take s, so that 


4x^ is greater than 1000 


. , 1 — -a; must be greater than 260 X 5a: or 1260ar, 
4jr /' 5 

which is certainly true if 1 — 2x be greater than 1250 jr, or 1 greater 

than 1252 J, or -— - greater than x. In this case 4j:* is greater than 

1000 times (B); still more then is it greater than 1000 times (A). 
Theorem. If the two series 

ao + ai^x + a2a^+ &c. and 60+^1^ + 62^^+ &c. 

be always equal for -every finite value of x, then it must follow that 
a^ ^ Iq, flj = 61, (7, = 62, &c. or the series are identically the same. 

Let these series be called Aq + A and b^ -j- B, in which, by what 
has just been proved, we can make A and B less than the mih parts 
of Oq and 6^. If possible, let Oq and Bq be different numbers, and let 
Cq = bQ-{-t. Then, since the series are always equal, we have 
(io-f-A = 6o-f-Bor6o-f ^ + A = 60 + B; that is, t= B— A. But 
because Oq and 6^ are fixed quantities, their difference is the same ; 
and we have t, a fixed quantity, equal to the difference of two quan- 
tities, each of which may be made as small as we please, which is 
absurd. Hence a^ = 60 + ^ cannot be; and by the same reasoning, 
OqZs: Bq — t cannot be; therefore Qq ^ 6^. Take away these equal 
terms from the two equal series, and divide the equal remainders by 
ar, which gives 

a^-^-azX + a^x^-i- &c. always equal to ii -[-620:4-^3^*+ &c. 

from which the same species of proof gives Aj =: 6| ; repeat the process 
of subtracting equal terms and dividing by x, and the repetition of 
the proof gives a, = ^2 > ^^^ ^^ ^^' Hence, if one or more terms be 
wanting in either series, the same must be wanting in the other ; for 
instance, if a — jt be always equal to aQ-^-a^x -■\-a^xi^-\- &c., we must 
have a = Aq, — 1 = ap = a,, = a^, &c. 

[The preceding process amounts simply to shewing that we may 


make jr = in a series, and take the ordinary algebraical con- 
sequences, in the same manner as if the expression were finite in 
its number of terms. For instance, if ff^ + (X|X -{- &c. be always 
s= 60 -{- 6, JT + &c., we have proved that the consequence of making 
4r = 0, namely, a^ = b^y is true. But, as we have sufficiently seen, 
it is not safe to say that when x = 0, P = Q, except in cases where 
we may say that by making x sufficiently small (or near to nothing), 
P may be brought as near to Q as we please. The difficulty which 
we have avoided is as follows : In the series 1 ■■\-2s '\-2.3jc^ -{- &c. 
we have seen that, take x as small as we may, the sum of the terms 
can be made as great as we please. Are we, then, entitled to say, 
that when x = 0, the preceding becomes 1 + + + &c. or 1 ? If 
the number of terms were finite, there could be no doubt of the pro- 
priety of answering in the affirmative; but when the number of terms 
is infinite, nothing that has preceded will enable us to give an answer. 
The student will remember that we have confined the demonstration 
entirely to series which admit of being made convergent. 

It is usual to prove the preceding* by saying, that when the two 
series are always equal, they are equal when J7 = 0, and consequently 
Oq ^ b^y and so on. This is avoided in the present case ; and we 
may say that we have proved the following theorem. If two series 
(which can be made convergent) are always equal when x is finite, 
then they are also equal when jr = 0.] 

* On this point the student, when he is more advanced, may consult 
Professor Woodhouse, Analytical Calculations, &c. Preface, p. viii. note. 
It is there objected that to make x » and thence to deduce a^ = b^, is 
the same as arbitrarily making a^ — b^. This I conceive to he not the 
true point of difficulty. All mathematical consequences are necessarily 
contained in the hypotheses fron^ which they spring : so that to invent 
any hypothesis is necessarily to invent all its consequences, some of 
which may be so near as to appear nothing more than the hypothesis 
itself, others so little perceptible as not to seem necessary attendants of 
the hypothesis. The real objection to the proof on which this note is 
written, I conceive, is this, that having frequently found the passage 
from X as a symbol of magnitude, to x as the symbol not of magnitude 
but of the absence of all magnitude, to be attended vrith consequences 
which require a special examination, it is not allowable to enter upon any 
new ground, without either establishing the accordance of the conse- 
quences of X as 0, with those of x «= some magnitude, or distinguishing 
and explaining the discrepancy, if any. 



The following are a few instances of the method by which we can 
obtain the limits of the sums of many series. First, let us take 

P =r l4.a;-fa:«-i-a;»+a?*+ &c. 

in which we wish to determine a finite algebraical expression for P. 
It is plain that 

1 +x-fj:«+ &c. arftn/"- = 1 +^{l +* + **+ &c. fldtn/:i 

that is, P = l+a;P or P = -J- 

1— J? 

a result previously ascertained. Now, let us take 

V P = l-^2x-\'33fi+4x^+ &c. 

\ 5^ = 2 + 3a;4-4a;« + 6a;»+ &c. 

^^ — P = l+X + X^ + X^-i- &C. =7-^ 

whence P«j 1> = h - = -.t 

{x J 1 — X X X 1— J? 

Next, let P = l+3x + 5a^-^7x^+ &c. 

— P = 2 + 2a;4-2a;«+2a;»+ &c. = j-^ 


1 +jr 

whence P = 

Next, let P == l-\'4x+9a^-i'l6x^'h &c. 
^"^ = 4f^9x + l6x^-^25x^+ &c. 


p— 1 


— P = 3 + 5a? + 7a;«+9a;*+ &c. 

r. {^^'P)x + l =l+3a;+6a;«+&c. = ^-l±^ 

whence P = 

(1 -*)» 

The same method might be applied to finding a more simple 
algebraical expression for the sum of any finite number of terms of 
the preceding series. For example, let 


P 5= \+2x+3a^+ +(n— l)a;"-« + na:"-* 

?=1 = 2H-3x+4a;«+ +naf"^ 


— P = l+a:+j:«+ +a:"-"— nx^' 

(page 103) = na:"-* = ^^ — -r-^— ■ 

^ ^ ' 1 — X 1 — * 

p _ n.r"+^ — (n + l)3:» + l 

The student may endeavour to prove the following : 
l+3a: + 5a:*+..+(2n— l)a:~-^ = ^q 5- — -i---i-- 


The inquiry which we have most frequently to make is the in- 
verse of the preceding; not, having given the series to iind its sum, 
but, having given an expression, to find the series of which it is the 
sum, or to develope it in powers of some one of the Utters contained 
in it. ' Let us suppose, for example, that we want a series of powers 
of X with coefficients, &c. which shall be, in all cases in which it is 
convergent, equal to (1 -j- a:) -7- (1 — x)'. Suppose the^eries to 
be Oq + a^x -{- a^x* -{- &c., so that we have 

Multiply both sides by (1— jt)' or 1— 2jr + x' 

1+a: = fOo + ai x + az x^+az a;3+ &c. "j 
— 2^0^— 2aia*— 2a2a;^— &c. \ 

-i-flfo ar^+fli a;3+ &c. J 

= ao+(ai— 2ao)^ + (a2— 2ai+flfo)a:*+ &c. 

The two sides of this equation being equal for every value of x, 
the theorem in page 188 gives 

aQ=^\ flfi— 2e/o =1 or ai = 3 

ag — 2ai + ao = or og = 5 

«3 — 2a2 + fli = or 03 = 7 

&c. &c. 

So that the series is 1 -^Zx •\-5x^-\- &c. as already determined. 

The first term of the preceding might have been found imme- 
diately : for, since we have proved that the results of or = may be 


employed, and since (1 +x) -f- (1 — jt)* becomes 1, and the series 
is reduced to Oq, when x = 0, we have Oq = 1 . 

We shall now ask what is (1— j:*) -f- (1— J?) expanded in a 
series of powers of x. The first term is 1, found as in the last 
sentence; let us suppose 

1 —J* X 

= l+aiX'\-aaX^'\-as3:^-^a43r-^ &c. 

1 X 

— X— ttio:*— aga:*— as^— fi^4^+ &c. 

As there is no first power of x on the first side, we must have 
a^ — 1=0, or a, = 1. Similarly, a^ — a,=0, or a^=saiZ=l ; a^ — 03 = 0, 
03 = 1. But the coefficient of x* on the first side being — 1, we must 
have 04 — 03= — 1, ora^ — 1 =s — 1, thatis, a4:=0; again, a^ — a^zszO, 
or a^ =s 0, a, — a^ == 0, or a^ = 0, and so on. Hence the series is 

l+2; + x2 + x34-0xar*+0xar^+ &c. 
that is l-\'X + x^ + a^ 

as might be found by simple division, or from page 103. Thus, we 
see thaA'when we assume an infinite series to represent a quantity 
which is in fact a finite expression, the method of determining the 
coefficients of the series will shew the coefficient for every term of 
the series which does not exist in the finite expression. 
Again, to develope 1 -7- (1 + a:*) assume 

o = ao + «i^ + fif2^^ + as^'+ &c- 
the preceding process will give 

^0 = 1 az'\-aQ = OT 0^=^ — 1 04 + flfi = or ^4 = 1 
a^ = ^3 + fli = or ^3 = ^5 + 03 = or ^5 = 

so that the series is 

l+Oxx—x^ + Oxx^ + x^ + Oxx^Sic. 

or '^—X^'\-X^^x^+ he. 

If it happen that, by the preceding process, an equation is 
produced of which the two sides cannot be made identical by any 
suppositions as to the value of the coefficients, it is a sign that the 
expression cannot be developed in a series of the form proposed. 
If we try to develope 1 -J- (1 -far) a: by assuming 


we shall find 


= aQ + aiX'\-a<iX^'\-asa:^'\- &c. 

1 = aoX-^(ai'\-aQ)x^+(a2-^ai)a^'\- &c. 

the two sides of which cannot on any supposition be made to agree ; 
for there is no term independent of x on tlie second side which may 
be made ^ 1 . In fact, we should find 

T— ; — = 1— a;4-a;*— a;^-i- &c. 

1 -f-x 

(-f-)a: -7tV-n = — l+a;-.a;2+ &c. 

^ ^ x{l-\-x) X 

so that the fraction proposed cannot be developed entirely in whole 
and positive powers of Xy without the introduction of a negative 
power (in this case jt-^). 

The following are given as exercises: 

1. If P = ao'haiX+aQX^+ &c. then 



2. ^ a = 1— a: + a:3— x*-f^— a:'^+ &c. 

3. iif =i_(J^_l)^+a_i);,«_&e. 

c+jp c \c' J 'Vc* c/ 

'p-\-qx p \p^ pi \ jD* p^ i 




In pag6 62, among the extensions of terms, we notified that the word 
equal was to be considered as applicable to any two expressions of 
which one could be substituted for the other without error. Hitherto 
we have only applied this extension to the case of definite algebraical 
quantities, either positive or negative : the numerical value of the 
quantity determining its magnitude, the sign determining only which 
of two opposite relations is intended to be expressed. We now pro- 
ceed to consider the word equal, or its sign =, not in a sense wider 
than any which the definition will bear, but wider than any in which 
we have yet had occasion to use it. 

Two expressions are said to be equal when one can be substituted 
for the other without error. The whole force of this definition lies 
in the answer to the question. What is error ? The answer is, any 
thing which leads to contradictory results, or which may in any 
legitimate way be made to lead to contradictory results. 

Results are contradictory when, both being intelligible, or capable 
of interpretation, the two do not agree ; but they are not necessarily 
contradictory merely because one or both are unintelligible, for where 
the meaning of any part of an assertion is unknown, we cannot say 
whether it be true or false. For instance, in page 23, the expression 

c c 

X ^ •-• was no contradiction of any thing which had preceded, for - 

had no meaning. Having ascertained this, our object was to give 
it a meaning which should not contradict any thing deducible from 
preceding principles. 

We have shewn that if x be less than unity, the summation of 


1 -\- X ■\- x^ '\- 8tc. will continually give results nearer to , which 

however can never be absolutely reached by that process. Hence 
we used the sign = in the equation 


= 1 +x+a:^+a^ + &c. ad infinitum. 


1— « 

Arithmeticalfy speaking, we can make the preceding as nearly 
true as we please, by taking a sufficient number of terms from the 
second side. Algebraically speaking, the above may be considered 
as absolutely true ; but it must be remembered that this is only oii 
a supposition which is arithmetically impossible, namely, that all the 
terms on the second side, expressed and implied^ shall be considered 
as included. For instance, we know that the multiplication of the 
first side by 1 — x* gives 1+^; the multiplication of the second 
side by the same gives 

l+X+X^+x^+ai^ + x^+ &c. adinf. 
— a:2— a;*— ar*— or^— &c. adinf. 
or l+a? + 0+0 + + 0+ &c. adinf. 

in which we can prove, if necessary, that every succeeding term 
shall be 0; not by actually looking at every term, which is im- 
possible, but by a deduction from what we know to be the law of 
the series. Similarly, we shew that jr^^xar^ss a!^+f^; not by 
looking at evexy case, which is impossible^ but by what we know of 
the meaning of ^. 

If we proceed arithmetically, taking any number of terms, how- 
ever great, say as far as a**^ we then have 

{l+x+x^+ .... +a:^-^ + af»)x(l-a:«) 
=z a+x+x^+x^-i- .... +af^^ + af' 

in which, since x is less than 1, n can be taken so great that x*^-^^ 
^nd ^+^ shall be as small as we please. Here is algebraical equa- 
lity whicfi can be made arithmetical equality, guam proximt. 

Let us now suppose that x is greater than 1 ; say x =i2; the 

l-i-X + X^+X^-i- kc. ad infinitum 

is then, arithmetically speaking, infinite, since there is no limit to 
the magnitude we may obtain by summing 1 +2 + 4-|-8+ &c- 

Should we then assert algebraical equality between - and the pre- 



ceding series? We have no right to do so from pages 21, &c. for 

tliough it was there shewn how to make it clear that -- is above all 

arithmetical numeration , the converse does not therefore follow^ that 
whatever is above arithmetical numeration is properly represented 

by -. What, then, is the proper algebraical representative of the 

preceding series ? If it have one, let us call it P ; then, since the 
series itself is the same as 

l-i-x|l+a:-fx2+ &c. ad infinitum^ 

P cannot be used for the above unless 1 +xP may be substituted 
for P. Or we must have 


P = \-\-xV which gives P = 

the same result as when x is less than unity. And we may shew, 
in the same way as in the last page, that any algebraical operation 

performed either upon , or upon 1 +x -f^-h &c. gives but 

one result: indeed, since there is in algebraical multiplication no 
stipulation that the quantities employed must be positive, the process 
there employed is equally applicable in the case where x is greater 
than 1. But in this case we cannot obtain (as in last page) any 
approach to arithmetical equality, but the direct reverse ; for j:«+i 
and j^+3 increase, instead of decreasing, as n increases. 

The method of expanding algebraical quantities in pages 191, &c. 
does not require that x shall lie within the limits of convergency : 

but the process is equally conclusive in all cases. To expand -, 

for instance, we ask what expression of the form flo + OiJr-l- &c. 


will be the same as t-- — . All we know of the latter lies in its 

1 + x 

definition, that, multiplied by 1 + jt, it gives 1. The series 1 — .j? -f- 
X* — x^-\- &c. has this property, independently of the value of x. 

We shall now ascertain, 1st, that so far as instances can shew it, 
we may prove that this general use of the sign == will not lead to 
inconsistent results : 2d, that we are not obliged to depend upon 
such a species of proof, but that the result is one which necessarily 
follows from the nature of our primary assumptions. 


To try tlie first, let us assume that, in all cases, 

-i- 3s 1— a: + a;«— x3+a:*+ &c. ad inf. 
We have then, if x = 1, 

I r= 1 — 1-f 1 — 1 + 1— 8tc. adinf. 

a result which is in no sense the expression of an arithmetical 
equality ; for the above series continued to an even number of terms 
must be 0, and to an odd number of terms, 1. We shall now try 
whether alternate addition and subtraction of a quantity to and from 
itself, ad infinitum, will or will not, when treated algebraically, give 
the half of the quantity in question. Let us suppose 

P z= l+x+x^ + x^+ &c. 1 

— P = -l-a:— a;«- &c. 

+ P = +1 +x + &c. 

-P = - 1 - &c. J 

.% P-P+ P-&C. = l+{x^l)+a^-x+l + kc. 

x+l :r»— 1 ^•-l-l 
(pagelOS) =-+^ + __. + ^ + &c. 

_ jr-i-r'-i-3:» + &c . i— i-fi^&c . 
X'\-l ■*■ x + 1 

But P = -i-,and x+3^+x*-\- &c. = x(l-\-x+x*+^.) =— ^; 
if, therefore, the preceding process be algebraically equivalent to 

halving and if 1 — 1+1 — 1 + &c. may be changed into -, we 

have the following equation : 

1 1 _ -1=7 1 1 


2 1— X x-j-1 ' 2 X'\-l 

1 1 _ _£_ • 1 ^ 
^^ 2 1 — J ^ l-.x«'*"2 x + l 

which will be found to be true. 

The above contains the algebraical artifice of writing the series (A) 
in the form 

] '^X-^X^+X^'\- &c. 

— 1— a:— j;«— &c. 
&c. &c. 




instead of 

l+x + x^-\-a^+ &c. 

— 1— a:— j;*— x'^ &c. 
&c. &c. 

or (1 — r+l4- &c.) + (a:— a; + a;+ &c.) + &c. 

But it is to be observed, that we do not say that every algebraical 
use of as will produce arithmetical equalities, but only that when- 
ever an algebraical use of =s does produce an arithmetical equation, 
we shall find that equation to be arithmetically true : an assertion as 
yet uncontradicted. 

We now proceed to the second point. 

We have previously so constructed the meaning of the funda- 
mental symbols of algebra, that algebra, in certain cases, coincides 
entirely with arithmetic ; and, more than this, the rules which follow 
from the definitions are so constructed, that when the result only 
is arithmetical, and preceded by algebraical steps, the alterations 
necessary to make these steps arithmetical, produce no alteration in 
the result. This being the case at the outset, and it being shewn that 
the number of steps through which we pass by algebraical process 
does not affect the preceding statement, we then know, 1st, that all 
arithmetical results so deduced may be depended upon, as much as if 
they were arithmetically deduced ; 2d, that all results which are not 
explicable arithmetically, are such as are perfectly consistent with the 
definitions laid down ; and, if not always arithmetically true, cannot 
produce a result which shall be arithmetical and false. 

The reason why we have appealed to instances is, that the pre- 
ceding argument, being general and abstruse, will not be thoroughly 
understood by the student until he has a degree of exact compre- 
hension of the words employed, which can only be gained by 
familiarity with the use of algebraical language. There are also 
principles of reasoning,* independent of algebra, which are difficult, 

* Kemember that mathematical reasoning means reasoning applied to 
mathematics, and is not a different kind of reasoning from any other. 
The art of reasoning is exercised by mathematics, not taught by it ; nor is 
the mathematician obliged to use one single principle which is not 
employed in every other branch of reasoning. In fact, an opposite of 
this is the case ; there are principles in other branches of reasoning which 
are not employed in most branches of mathematics. Those who are not 


and which the beginner will therefore not conquer by algebra alone. 
Such is the following : that if assertions which are not inconsistent 
with each other are rationally and logically used, the conclusions 
cannot be inconsistent with each other. But though this, in its full 
extent, be a very difficult proposition for a beginner to understand, the 
difficulty is not an algebraical one. 

We have said, or at least implied, that when an arithmetical 
result is produced, the steps by which it was produced, if not already 
arithmetical, may be made so. Under arithmetical equality we in- 
clude not only the absolute notion of equality which we recognise in 
4 + 5 = 9, but what we have called the quam proxime equality,* 
which we see in 

-1^ or 2=1+1 + 1+1 + &c. adinf. 



Let us now suppose that a is less than i , but that we may sup- 
pose it as near to 1 as we please. Also, let x be less than 1 (which 

aware of this talk of mathematical demonstration as if it were distinct 
from all other kinds. 

But this is not the case ; mathematical demonstration is, so far as it 
goes, the same thing as any other demonstration ; the superior safety of 
mathematics lying, not in the method of arguing, but fii the reasoner 
knowing more exactly what he is talking about than is usual in other 
discussions. The mathematical sciences are concerned in whaterer can be 
counted or measured ; and whoever talks of mathematical demonstration^ 
as applied to any thing else, either means merely logical or certain 
demonstration, or does not know what he is talking about. 

The opponents of mathematics are, of all men, those who pay this 
undue respect to mathematical reasoning ; which they invariably do, by 
asserting as a discovery and a triumph, that those who are only mathe- 
maticians frequently reason ill on other subjects. We recommend the 
student to believe this whenever he meets with it, and to act accordingly ; 
for it is an important truth, though not either a great or recent discovery — 
having been, in point of fact, ascertained immediately after the fall by our 
common ancestor, who, having till then been nothing hut a gardener, 
must have foimd himself but an indifferent tailor. 

* We are not here stepping even beyond the bounds of ordinary 

arithmetic. For vlO, v 11, &c. have no other but a quam proxime 
existence ; we can find fractions which, multiplied by themselves, shall 

be as near as we please to 10 ; we can sum "^t a> 1* ^^* ^^^^ ^® come 
as near as we please to 2, 


is necessary to the arithmetical existence of the final equation), and 
we have 

1 "* > arithmeUcally 

j-^ = l+ar + x^ + a;« + &c.J ^2) 

— a — aa; — aa;* — &c. 
+ a^+a^x+ &c. 
— a*— &c. 
= 1 — (a — a;) -h (a;* — aa? + a^) — &c. supposing a > a; 

= — 1^ 1 1 — — &c. 

a-^-x a-\-x a-j-a? 

a — a^'\-a^ — &c< x -|- jp* -j- jt* -f- &c. 

fl + J" O + J? 

a ^ I ^ 1 

l-j-a'o + J^ 1 — x'a-\-x 

So long as we suppose a less than 1 (no matter how little), the 
preceding process is arithmetical; but the moment we suppose a = 1, 
the equation (1) loses all arithmetical character. But still, the last 
equation reduces itself to the one given in page 197. 

We have chosen a very simple case, but we might, with operations 
of sufficient length, and by various artifices, reduce any algebraical 
equality to an arithmetical one. But methods of doing this, at once 
short and general, cannot yet be understood by the student. 

Let X be first supposed positive; let it then become 0, and after- 
wards negative. The following and similar tables may then be 

Sign of a? + — 

Sign of - + oc — 
a;3 + - 


a;« + + 

i + « + . 


These, and other instances, give the following principle: If^ 
when X changes from a to b, passing through all intermediate values, 
the sign of a Junction of x change from positive to negative, or vice 
▼ersft, the point at which the change takes place is marked by its value 
being either nothing or infinite; but the converse is not true, that a 
function always changes its sign when its value becomes nothing or 


The infinite here spoken of is the form - ; but, as we have seen, 

all methods of obtaining an arithmetical infinite (we should rather 
say all arithmetical methods of increasing number without limit), are 

not properly represented by -. If we take the equation 

— — = l^x+X^+X^-^- hQ.adinf 

we see that when x is greater than 1, the second side is, arithmetically 
speaking, only an indication of a method of obtaining number with- 
out limit, while the first side is negative. There is then a change of 

sign when x passes, say from ■- to 2, and the change takes place when 
x=: 1, giving, 

2=l. + l+l+l+&c. 

i = 1+1+1 + 1 +&C. 

-1 = 1 +2 + 4 + 8 + &C. 

We therefore see that a divergent series may be the algebraical 
representative of a negative quantity : but the series 

T^3^2 = 1 +2x + 3x^+4x^ + ^,0. 

which is divergent when x is greater than 1, gives a positive result in 
all cases. From this we see that when an equality specified is purely 
algebraical, we are not at liberty to compare magnitudes by any arith" 
metical comparison, if infinite series be in question. For instance, 
if a be greater than a', b greater than b', &c., we may say that 
a + b+ &c. is greater than a'+ b'+ &c. : 1st, so long as the number 
of both is finite; 2d, if a, b, &c. be so related that a + b+ &c. can 
never exceed a given limit. But we may not draw this conclusion 


whan a + 6 + &c. increases without limit ; nor may we say ths 
algebraical representative of 1 -f 2 +4 -f &c. is greater than tl 

1+5 + 5+&C. 

There is no liability to trangress this rule, because the qus 
in question must cease to be the object of arithmetic befor 
reasoning upon it can fail. Let it then be observed, that with r 
to divergent serieSf we admit no results of comparison except 
which are derived from their equivalent Jinite algebraical expressio 





We have already defined what is meant by a function of x (see 
page 168): the present chapter is intended to exemplify the method 
of denoting a function, either for abbreviation, or because we wish to 
allot a specific symbol to some unknown function. 

When we have to consider an expression, such as x^ + ax^ only 
with reference to the manner in which it contains x, that is, to reason 
upon properties which depend upon the manner in which x enters, . 
and the value of x, and not upon the manner of containing a, or its 
value, we call the expression a function of x, and denote it by a 
letter placed before jt, in the manner of a coefficient. But to avoid 
confounding this functional symbol with that of a coefficient, certain 
letters are set apart always to denote functional symbols, and never 
coefficients. These letters will be, in the present work, F, f, 0, -v^. 
Thus, by Vx,fx, ipx, ^x, we mean simply functions of x, given or 
not, as the case may be. By Tt^x we mean the same function of 0x 
which Fa: is of a: : for instance, if F x =i x -j- x^, F^jr is 0j? + (jp^T' 

Examples. Let <px = l-^x^ Fx = 1— a:* 
-Pfx = 1-(1 +0:2)2 = -2x2-a;* 
<p¥x = 1+(1— a;2)2 = 2'-2x^+x* 

Let (px = Of then 9(1 +x) = (I +xY <p(2x) = (2xy 
(p{(i) = a« ^i = J« &c. 

A functional equation is an equation which is necessarily true of 
a function or functions for every value of the letter which it contains. 
Thus, if 0x = ax^ we have ^(6^) = ahx = 6 x <px, or 

is always true when (^x means ax. 

Thus the following equations may be deduced: 

If <px '=z x^ <pxx<py =^ <p{xy) 

<px = a' <pxx(py ^ 9(^+y) 



(px = ax'\-h \ ^ = — ^ 

r ^x — ^Z X — Z 

(px = ax <px + <py => pC^+y) 

We can often, from a functional equation, deduce the algebraical 
form which will satisfy it : for instance, if we know that ^{xy) =x x ^y, 
supposing this always true, it is true when y = 1, which gives 
^{x) zsz XX ^(1)* But ^(1) is an independent quantity, made by 
writing 1 instead of x in ^(a:). Let us call it c : it only remains to 
ascertain whether any value of c will satisfy the equation. Let 
^x =i cx'y then ^{xy) = cxy, and xx^y =^ xxcy :=> cxy; whence 
0(x^) =s x<l>y for all values of c, and <l>x being ex, ^(1) is c x 1 
or c, as was supposed. Similarly, if 

(p{xi/) = (<pxy, by making a: = 1 we have 

(py = I ^(1) Y = C (px =-c* 

^ixy) = (f^^ i(fy ^{<pxr 

and ^(1) = c* =sc, as was supposed. This, with the following theorem, 
will be sufficient for the investigations connected with functions which 
we shall need in this work. 

We have seen that if ^x = c* we have ^jrx^y = ^(j^ +y); ^"^ 
we do not know whether there may not be other functions of x which 
have the same property. We shall now, however, prove this con- 
verse, namely, that the equation ij>xxi>y = 0(^ H"^) ^^ ^® satisfied 
by no other function of x, except those of the form c*. 

Let us suppose ^x to be a function of such a character that 
whatever may be the values of x and y, the equation 

<pxx<py = <p{x+y) (A) 

is true. For y write a + 6, which gives 

<pxx<p{a + b) = (pix + a + b) 

But equation (A), as described, gives ^(a + fe) := 0aX0i, whence 

<pxx(pax<pb = 9(a: + a4-6) 
For either letter, a, for instance, write c + e, which gives 

(pxx<p{C'\-e)x<pb = 9(a:4-c4-€ + J) 
but (pic-\-e) = (pcx(pe 

whence (pxx <pcx (pex <pb = (p^x+c+e + b) 


and so on : that is, if tliere be n quantities of any value whatsoever, 
namely, o^^ a^ a^ On-'U ^9 ^^ have 

Let these f» quantities be all equal to each other and to a, which 
gives . 

{fax^ax •••• X pax pa) __ /a+a+ .... +« + a\ 
) there being a factors. 3 ^ there being n terms. / 

or if>ay = p(»a) 

where n is any whole number.* 

Similarly, if we had supposed m quantities each equal to 6, we 
should have had 

Now, m and n being whole numbers, let us suppose mb z=z na, 
which gives f 

(p(jnb) = <p(na) or (pS)** = (pa/* 

or fb=s (pa)"* but J =* — a, whence 

p(-^a) = (pa)' 


or the equation ^{pa) = (^a)** is true when p is a whole number, 
or a commensurable fraction (page 98). 

In the original equation (A), let jr = and y = 0, whence 
* -j-y =0. Let ^(0) be called c; we have then cXf== c, or c = 1. 
Theu let ^ = — x, or j: +^ = 0> which gives 

pa:xp(-a:) = p(0) = 1 or K"^) = ^' 

which being true for every value of x, is true if for x we write ;>«,/) 
being a whole number or commensurable fraction. This gives 

p(-/'«) = ^ = (^ = (?«)" 

* We cannot conceive a fraction of the preceding process. (See 
page 37). 

t Similar operations performed upon equal quantities must give 
equal results. 



or the equation 

is true Up be a negative whole number or commensurable fraction. 
Hence, as in page 204, by making a = 1, we determine 

(f>{p) = c^ 

where (provided p be commensurable) c may be any quantity 

The preceding equation will also be true when p is an incom- 
mensurable quantity, such as *J2y *>/ Ay &c. ; but the method by 
which we shall treat the consequences of this equation will render a 
proof unnecessary. 

Exercise. Shew that the equation 

is satisfied by 

(px = -^y^'+or') 

for every value of a : and also that 

<p{x-\-y) = (px + (pt/ 
can have no other solution than 

f)X = ax 





The binomial theorem is a name given to the method of expanding 
(fl-^by into a series (finite or infinite as the case may be) of powers 
of a and b^ n being either whole or fractional, positive or negative 
commensurable or incommensurable. 

The preceding case may be reduced to that of expanding {i-^x)* 
in a series of powers of x : for 

o + & = a(]+y (a + 5)"=a-(l + ^)" 

Let '=-9 and (1 + dr)* is the function to be expanded. 

As this theorem is of particular importance, we shall give two 
investigations of it : the first analytical, inquiring what is the series 
which is equivalent to (1 4*^)*; the second synthetical, shewing that 
the series so found is the one required. 

If it be possible, let (1 + j:)" be a series of whole powers of x of 
the form 

(1 + xY = flo + «ia? + ^aX^ + flj^ + &c. 
in which a^, a,, &c. are functions of n and not of x, 

a* — ff» 

Lemma. Whatever may be the value of n, the limit of -, 

a — ft 

to which it approaches as 6 is made more and more nearly equal 

to a, is na*"^. 

[When a = 6, observe that this fraction takes the form -, 

page 162.] 

First, let n be a whole number. Then, page 103, 

a — 
as 6 and a approach to equality, the limit of the second side is 



or Hid! 

or a"-'+a"-^ +0**"^ + .... 4-0""' +a" 


[That there are n terms in the preceding is evident from this, that 
there is a term for every power of 6 from 1 ton — 1, both inclusive, 
and one term independent of 6.] 

Secondly, let n be a fraction ^ where /) and q are whole numbers. 

We have then 


a— 6 a — 6 

af — h? 


ai — h^ 

now, as a approaches to ft, a, approaches to 6,, and as p and q are 
whole numbers, the limits of the numerator and denominator of the 
preceding arepo/"* and qaf"^; whence the limit of the fraction i 

C^or ?<- or Ha)'" or lJ~' or „„— 

Thirdly, let n be negative, and let the corresponding positive 
quantity be p, so that nsc~^p. Then 

a^—h'* _ a-P — b''P aP bP _^ 1 bP—aP 
a — b "~ a — b a-^b <i'6»'*fl— ft 

— ^ ^^—^^ 

■" aPbP ^ a^b 

as fl approaches towards ft, the limit of the first factor is or 


— a—'Pf and that of the second (/? being positive) is, as already 

proved, paP-K Hence the limit of the preceding product is 

--a-'P X paP"^ or — pa-P-^, which, since « = — j9, is na"^-^. 

We now resume the assumed series 

(1 + xf^a^ + aix + a^ -f a,^ -i- &c. 


Let ^ be a quantity which may be made as near to x as we 
please ; the above series being supposed true for all values of jc, we 
may put^ in the place of x, which gives 

(1 + y)" = a© + fliy + Ost/^ + 8tc. 
(l + ^y-(l+y)'*=eii(a;-y)+a2(a;'-y«) + &c. 
Divide both sides by x — y, or (1 +x) — (1 +y)» 

(i+jf) — (i+y) * j^—y •r— y 

vrbich two sides being always equal, the limits to which they 
approach, as s and y approach to equality (in which case 1 +' ^^^^ 
1 +y approach to equality), are equal ; or 

w(l + ar)"-^ = ai + 2a2X + Saj^x^ + 8cc. 

Multiply both sides by 1 + x. 

n(l + xy = fli + ^OicX + Sasx^ + 4a^x^ -f &:c. 

+ aix + 2afiX^ -f 3a^x^ + &c. 

w(l + x)*" = woo + nuiX + nusx^ + na^x^ + &c. 
therefore, page 188, 

«i = noo, 2a2 + «! = wai or a^ = -— — Oi = w ^- Qq 

3. rt n — 2 fi — 1 »i — 2 

a^-\- ZOi^nOi or a3= —— a2 = 7i. -——.——, ao 

o 2 3 

A , o « — 3 n — \n — 2n — 3 

4a4 + oa3 = wa3 or 04 = — j— a3 = w-^ ^ ^ a^ 

from which, by substitution and observing that a^ is a factor of all the 
terms, we have 

(1 +a;)"=ao(l +«a: + w'^a;« + w^-^'^a;3^g^ \ 

In which a^ is not yet determined. To determine it, we must first see 
whether the preceding series is ever convergent. 

But previously, we must observe that what we have proved is 
not, properly speaking, the truth of the above equation, but only that, 
if {\ +x)° can always be expanded in a series of whole powers qf\, 
the preceding series is the one. For we have assumed that (1 +x)* is 
flo + ^i^+^a^ + ^c., when, for any thing we know to the contrary, 
it might be a series of fractional, or negative, or mixed powers which 




should have been chosen. 'Whenever, in our previous investigations, 
we have made an impossible supposition, we have always been 
warned of it at the end of the process by the appearance of some Dew 
and unexplained anomaly. As we have not, before this one, made 
any assumption of the form of an expansion other than we had either 
reason or experience to justify, and as we have neither for the one we 
have actually made, we know neither whether we are right, nor what 
is the index of our being wrong. It may be that die appearance of 
an always-divergent series would follow from a mistake, if ^ere be 
any. We therefore examine the preceding series. 

The ratios of the several terms to those immediately preceding are 

,1—1 fj— 2 fi— 3 , 

nx ~17~^ a? X &c. 

46 o 4 

the general form of which is 

(/?-t-2)term ^ n— /? 
(/) + !) term p+1 

First, we observe, that if n be a positive whole number, the series is 
finite : for the (n + 2)th term will contain n — n or as a factor, and 
so will all the following terms. We therefore take the case where 
n is fractional or negative. When p has passed n, the preceding ratio 
will be always negative, shewing that the terms become alternately 
positive and negative ; for that the ratio of two quantities is negative 
indicates that they differ in sign. Neglect the sign of the preceding, 
and make it positive. This we may do, as our object is to see 
whether the series can be made convergent, which a series of terms 
alternately positive and negative always can be, if the correspond- 
ing series of positive terms can be made convergent. We have 

p — n px nx X nx 
' X or -^ — '^^- or ■ » ■ — 

as we take higher and higher terms, the second term of the preceding 
diminishes without limit, and the first has the limit x. Consequently, 
if X be less than 1, the ratio above mentioned will, after a certjaiv 
number of terms, become less than unity, and wiU afterwards apr 
proximate continually to the limit ar. Tha( is, the aeries gt^taiQed is 
always convergeq^ when x is l^s than 1. 

This bein^ the c^se, we m^y, page 189, use the result of making 


jr = 0, winch gives (l)^=c: a^. If ti be a whole number, a^ = 1 ; but 

if n be a fraction, as ^ we hare (1) • = o^, a: (l**)' = (1)' , or a^ may 

be any 9th root of 1; see page 113. If we choose the arithmetical 
root, we find a^ss 1 ; and this series is (if the doubtful assumption be 
true) the arithmetical nth power of 1 + j*, when x is less than 1 , or 
the algebraical equivalent of (1 + x)** in all cases. 

We shall now shew that the series is correct whenever n is a 
whole number. Suppose it correct for any one whole number, 
say m. Then (gq being 1), 

Multiply both sides by 1 + j:. 

4- ar + i» x^+m-^^ x^-{- &c. 

^ 1 • / . i\ . / ^ — 1 . \^ , f ^ — 1 '" — 2 , m — 1\ - , ^ 
^l+(m+l)x+[m-j-^m)jfi+[m— 3- + '^ -"2"/ ^ "^ ^^* 

n ^ 1 . ^W 1 . l\ WI + 1 , . Ix'W 

But m-^4-»i = w(^-^+lj=i»-^ = (m + l)2 

m — 1 m — 2 . tn — 1 m — 1 /w? — 2 . ,\ , , -,.mm — 1 

(l+xr+'s !+(>» + l)x + (»M + l)|x« + (m+l)|^^ar'+ &c. 

which, if we now write w for wi -|-1, or n— 1 for wi, becomes the same 
series, or follows the same law as 

/I . \« 1 . ^ — 1 o . w — 1 ^ — 2 • , - 
(^l^xT^l-tfix + n—^x^+n.— ^^+ &Ct 

Hence, if this expression be true for any one whole value of n, 
it is true for the next. But it is true when 9i = 1 ; for 

(l+a;)'=l+la; + l.^a;«+l.~.~a;»+ &c. 

therefore it is true when n =s 2 ; but it is therefore true when n = 3, 
and so on, ad infinitum. 

If we consider (1 + ^)* as a function of it, and call it 0n, we 
tee that 


(1 + xY X (1 + xT = (1 + a:)"+'* 

but when n is a whole number {\-\-xY is the series in question; 
therefore, calling the above series 0n, we have, when n is a whole 

<pn X <pm = p(w + m) 

^l + nj:+».^^x»+n.^^ .23_a:«-|- &c. j 

= 14-(»w+n)j:4.(»i4-n) ^ j:«+(»i+n) 1-_ j»^ gcc. 

This may be verified to any extent we please, by actual multipli- 
cation ; for the two first series multiplied together give 

1 4-(m + n)x-\- fw.^^ + wm + m^^Ja:* 
. / n — In — 2 . n — 1 m — 1 . m — \m — 2\ - 

„ , n — 1 , , m — 1 n' — »i + 2nfii + ni* — m 
But n.— |-»»i+ »!•— ,7— = ■ ^ — 

« <6 2 

= '—^ = (n + w) ^ 

n — 1 n — 2 . n — 1 m — 1 . m — 1 m — 2 

2 3 2 2 ^ 2 3 

w^— 3w*-h2n4-3w*w?— 3n>w + 3 nw*— Znm-\-Tf? — 3m*4-2m 


2X3 ^ ' ' 2 3 

and so on. We now lay down the following principle: When an 
algebraical multiplication^ or other operation^ such as has hitherto 
been defined, can be proved to produce a certain result in cases where 
the letters stand for whole numbers, then the same result must be true 
when the letters stand/or fractions, or incommensurable numbers^ and 
also when they are negative. For we have never yet had occasion to 
distinguish results into those which are true for whole numbers, and 
those which are not true for whole numbers ; but all processes have 


been, as stated in the introdnctiofi, true whether the letters are whole 
numben or fractions. There has been no such thing in any process 
as a term of an equation, which exists when a letter stands for a 
whole number, but does not exist when it stands for a fraction. If, 
therefore, ^mx^n will give ^(fii-f-") by the common process of 
multiplication when m and n are whole numbers, that same process 
will also give ^(m^n) when m and f» are fractions, and also when 
they are one or both negative : consequently, the series 

l+nx + n'-^a^+ &c. 

has this property, that, considered as a function ofn, and called 0n, 

it satisfies 

^nx<pm = ^(m + n) 

in all cases. But in page 205, it has been proved that any solution 
of the preceding equation must be 0n=rc" where c = 0(1), and 0(1) 
we find to be 

l + l^ + l^Y^a:*+ &c. or 1 + x 
therefore 0n is in all cases (1 +!•)*, which is the theorem in question.* 

* Every proof which has ever been given of this theorem has been 
eontested ; that is, no one has ever disputed the truth of the theorem 
itself, but only the method of establishing it. And the general practice 
is, for each proposer of a new proof to be very much astonished at the 
want of logic of his predecessors. The proof given in the text is a 
combination of two proofs, the first part, making use of limits, given 
(according to Lacroix) in the Phil, Trans, for 1796 ; the second, the 
well-known proof of Euler. The objection to the first part lies in the 
assumption of a series of whole powers ; to the second, in its being 
lynt^tical, that is, not finding what (l+a:)" is, but only proving that a 
certain given series is the same as (l + x)\ But each part of this proof 
answers the objection made to the other part ; in the first part analysis 
is employed, but only so as to give strong grounds of conjecture that 
l + nx+ &c. is the required series; in the second part this conjectural 
(not arbitrarily chosen) series is absolutely shewn to be that required. 
The proof of Euler may be condensed into the following, of which the 
several assertions are proved in the text. 

1» If l + ni + n.— =-«*+ &c. be called ^n, then ^(1) is 1+x, and 
^nx^tn is found to be f(n + m), 

2. If ^nx^m SB f{n + fn) then ^n must be {^(1)}** 

S. Therefore, l + nx+ &c. is (1 +x)* 

In 18^7, Messrs. Swinburne and Tylecote published their d^mon- 


To use the preceding series, the readiest way is to form the 

several factors n, — — , , &c. previously to proceeding further; 

for instance, let n = ^, or let it be -n/i +j:, which is to be expanded. 

— 1 ^— ^ — ^ «— 2 __ 1 n~3 __ 5 
^ ■" 2' "1 4' "T" "■ " 2' "T" 8 ^'^' 

+(i)(-i)(-5)(-i)^+ ^- 

^2 8 ^16 128 ^ 

If » = — 1, 

n = «l, -_=-.!, -_=-.!, _-==«i &c. 

O+X)"' or ^= l+(-l):c+(-l)(-])a;« 

+ («1)(-1)(-1)^+ &c. 

= 1— a:+<a;*— -2^*+ &c. 
as has been proved before. 

stration of this theorem, the last original one with which I am acquainted. 
It is much too laborious and difficult for a beginuer, but is as unobjec- 
tionable in point of logic as I conceive the one given in the text to be. 
Their general objections to the theory of limits are not, I conceive, to 
its logical soundness, but to its applicability in algebra ; it being more 
frequently than not a sort of convention that limits shall not be introduced 

into algebra. But, until \/]0 is explained in arithmetic without limits, 
I shall hold that limits are, and always have been, introduced into 
algebra. Nay, until the symbol is well explained in all its uses, with- 
out limits, the same may be said. With regard to the objections made 
by the above-named gentlemen to Euler's proof, in their page 35, they 
evidently hang upon the assumption of a finite number of terms of 
the series, which is not Euler's supposition, and which he therefore 
" keeps out of sight,** not " dexterously shuffles out of the way," 
for it does not come in the way, any more than lines which meet come 
in the way when we are expressly talking of parallel lines. After 
quoting the last-mentioned phrase, it is but fair to those gentlemon to 
state, that, judging by the general tenor of their work, it is not to be 
presumed that they meant to accuse Euler of intentional deceit, which 
the phrase " dexterous shuffling '* generally means, and which, therefore, 
was not the proper phrase to use. 



If f» = 5, 

f' n — I gy n— 2 I n — 3 1 n — 4 1 w-^5 

"" 2 "" 3 — 4 ""2"5"""5 5 









1 xj 


a:)*=l+6ar+/ 6\a:«+/ 6U-«+/ 6\ar*+ / 61ar»+/ 6\a^ + 

x2/ [x2] x2 x2 |x2' 





= l+5x-f 10a;*+10a;*+6ar*+ar^+0+0+ &c. 

It is usual to prove the binomial theorem in the case of a 
whole exponent, as follows: Multiply x + a by x-\-b, which gives 
s*-\'{a-\'b)x-\-ab, and this again by j: + c, which gives 


From which it appears, that if there be n quantities, a^a^ a^, 

and if the sum of all be called Pp the sum of the products of every 
two, P, &c., that is, if 

Pi = «1 + «2 + fls + 

Pjj = ttia^ -^ a^a^ + OiGs -j- 

Ps = aiOtta^ -i- ttiQ^a^ -\- 

Pfi-i = 

it follows that 

{product of all 

►fall 'I r product of all') 

a, j \ except Cj J 

product of all; 

The number of terms in P^ is n ; in P, as many as there are 

n— 1 
combinations of 2 out of n quantities, or (Ar. 210) n . > ; in P, 

18 many as there are combinations of 3 out of n quantities, or 

f» — -— ; and so on. Hence, if a^ a^ &c. be all supposed 

equal to each other and to a, we have 



(^ containing n factors. J 2 

Id which, if we suppose j: = 1, we have 

(l+a)" = l4.na + »^a2+ &c. 

See Ar. 211, for the reason why the coefficients are the same, 
whether we begin from the one end or the other of the series, as will 
also appear in the following cases. 

{l+xY = l+2a;4' x^ 

(] ^xy = l+3x+ 3x^-\- x^ 

(1+0:)*= l+4a?+ 6a;2-H 4a;»+ a^ 

(l+a:)*= l+Sx+lOa^-hlOx^^ 5a^+ x^ 

{l-hxf = l'{-6x+}5a^'h'20x^+i5ai*+6x^-hafi 

In order to find (1—- j:)% change the sign of x in the series; that 
is, write — x for x, leave x' unaltered ; write — x* for x®, and so on ; 
which gives 

(1— a:)** = 1 — »a: + w-^a;*— &c. 

We may write the general series in the following way when n is 
a whole number^ where Pn signifies tlie product of all whole numbers 
between 1 and n, both inclusive. 

which shews the similarity of the coefficients above alluded to. 
We give the following as exercises : 
1 . When n is a whole number, 

2« = l+w + n-^— + /1-2 J^-^^^' 

r, I ■ n— 1 n — 1 n— 2 . « 


2"-*= 1 4-n--r- +»-;; ;; :; h &c. 

2 2 3 4 

2. (a4-i)n— afi4.„an-l64-»l^a«-«6«+ &c. 

3. If n be a whole number ' 

(^ + l)'=^ + l+3(x + l) 

ending with ya"-l)(2n-2) ■ • ■• (»+l) 

{x + l)*""* = a*«+> + 1^+1 +(2n + 1) (a*-i + 1^.,) + . . . . 
,. .. (2n+l)(2n) (.n + 2) / , t\ 

^ —^nx + n-j-- — x^+ &c. 

5. ITie student may provide himself with examples and verifi- 
cations in the following way: choose any exponent n, whole or 
fractional, positive or negative, and expand (I + jt)** and (l + .r)»»+i 
by means of the theorem; then multiply the first series so found 
by l+^> which should give the second series. 

6. (f''^l/' + n(a'-b)lf"^+n~-{a^byb*'-^+ &c. 

The following cases will afterwards require particular notice : 

(IN*** 1 nx — 11 njT— Inj:— 21 

nJ ' n ' 2 n' 2 3 w^ 


But nx-=^x nx — - — x -. =:a:-T— 

n 2 w* 2 

1 2 

nx — 1 nx — 2 1 

n n 

nx — :: , = X — &c. &c. whence 




-. 1 12 

X— - ar— - X — 

(\\ns , X — - ar— - x 

1+1) =l+^ + a;-5^ + x^^+&c. 

In the preceding let ar = 1, which gives 

But (page 105), (l + 1)"' = |(l + 1)" 1' 

or the first series is the arth power of the second, and 

(l+l-f — 2-^ + &c.y = l + a: + a;-^^+ &c. 

If n = 0, (1 + j)« = 1, or (1 + a:)« — 1 = 0, or the fraction 

^ i assumes the form -. We now ask whether this frac- 


tion has a limit when n is diminished without limit (page 162). 
From the general theorem^ 

n 2 2 3'' 

and when n is diminished without limit, the limit of the second side is 

x^ , X^ X* , ^ 

That is, 

As n diminishes without limit, *| 

(H-^)« — 1 u -.u J x^ , x' -J^ . p 
^^ ' approaches without ^ x r- + -^ T '^ ^^' 

limit to J 

If ar = z — 1, then the limit of (n diminishing without 

limit) is 

('^-'i^)-l(z-if + l(z-iy- &c. 


Of this series we know no property as yet, except that it is 
convergent when x or z — 1 is less thani, or z less than 2. We 
shall proceed to examine the expression from which it is derived. 
If in it we write z*^ for z, we have 


or m 

n mn 

Let m be a fixed quantity and let n diminish without limit : then 
mn also diminishes without limit. Now, if 0n have the limit N 
when n diminishes without limit, <p{mn) must have the same limit. 
The only difference is, that (say m = 6) for any very small valae of n, 
^{6n) will not have come so near to its limit as 0n. For the nature 
of the limit being, that, by taking n sufficiently small, we may make 
^n within any given fraction (say A;, which may be as small as we 
please) of N, we may, by taking the sixth part of the requisite value 
for n, make 0(6n) within the same degree of nearness to N. Hence 
the limits of the two^ 


n mn 


are the same. But if the first be called ylz, the second is — -^P (z"^) ; 

m ^ ^ 

and we have 


limit of -d/Z = limit of — -4/2:'* 
^ m ^ 

= —limit of >L 2;'" 

m ^ 

^-.l-.l(2-l)2+&c. = ^(^'»-l-|(z'«-l)«+8cc.) 

a property of the series which will serve to verify succeeding 

We have not yet included in our results the case in which the 
exponent is incommensurable, such as 

but since we look upon \/2 as the limit to which we approach nearer 
and nearer in the series (obtained from arithmetical extraction in this 

1 1-4 1-41 1-414 b4142 &c. 

we must regard (1 + x) "^^ as the limit to which we approach by 
taking the successive expressions 

14 141_ 1414 

l-fiT {y^-xY (l + ar)i«» (l-far)»«» &c. 

so that, whatever approximation k may be used for V2 or the limit 
for which this symbol stands, then 


l + kx+k^^x^+ &C. 

is the corresponding approximation to (1 + x)'^. When the series 
is convergent, it is evident that if each term be found, say within its 
thousandth part, the sura of the whole series is found within its 
thousandth part. Suppose k and k-\-m to be two approximations to 

v2, as in page 101, the first too small, the second too great, and 
suppose we compare the pih terms of the corresponding approxi- 
mations to (1 + a") '^^ , or 

L7i....^-g±£^-iand(A + »»)^±^.... '^ + '"-f+V -i 
2 p^l \ ' / 2 p — 1 

the ratio of which is 

k-^m k-j-m — 1 k-j-m — p + 2 

k * A:— 1 k^p + 2 

Since m can be made as small as we please, it is evident that 
each of these factors can be made as near to unity as we please, and, 
consequently, their product can be brought as near to unity as we 
please. That is, for any given number of terms, the terms of the two 
approximations may be made as nearly equal as we please. Now, 
suppose X less than 1, so that (page 210), both approximations are 
convergent. Hence so many terms (say q terms) may be taken that 
the limit of the sum of the remainder shall be as small as we please : 
and as m may be taken so small that all the q terms of one approxi- 
mation shall be within, say their millionth parts, of those of the other 
approximation, it follows that we may reduce the two approximations 

to the following form : 

{A remainder less 
than the millionth 
of the preceding 

(A remainder less 
timn the millionth 
of the preceding 

where a fi w are severally less than one millionth. Let 

a-\'b-{'C'jr .. , , -jr z (which we may call the approximation to the 
first approximation) be called P; aa-f6/3-{- ,,., -}-;?» is less than 
one millionth of P, and the first approximation complete is 


P + X P (X is less than one millionth) 

and the second 

P(1+V)+YP(1+V) (V and Y do. do.) 
the difference of which is 

PV+P(Y-X) + PYV 

which is less than three millionths of P; because V, Y — X and YV 

are severally less than one millionth of P. But the limit (1 + x) *^ 
must lie between these approximations, and therefore does not differ 
from P by so much as the approximations differ from each other, 
that is, by three millionths of P. 

It is not possible to shew, by the preceding method, the same 
result in the case where x is greater than 1, or the series divergent: 
but it must be remembered that in this case algebraical equality only 
is asserted, not arithmetical equality : and all that is said is, that 

1 + \/2x'b t/2^— a:« + &c. 

may be substituted without error for (1 -f j:)'^^; so that the proof of 
this case falls within the general proof, as deduced from the principle 
in page 212. What is done in the preceding process shews the 
approximation to arithmetical equality, when the series is convergent. 
The same arguments might be applied to any other case, with any 
other degree of approximation. We now proceed to develope some 
further consequences of the binomial theorem. 




An algebraic symbol may have different names, according to the 
relation in which it stands to different symbols, or combinations of 
symbols. Thus in ab, a, with respect to b, is called a coefficient ; 
with respect to a 6, it is called a factor. Similarly, in o^, 6, with 
respect to a, is called an exponent ; with respect to o^, it is called a 
logarithm,* and a in reference to b is called the base of the logarithm. 
Thus, in 3^ 4 is the logarithm of 3^ or 81, to the base 3; in 
a*, X is the logarithm of a* to the base a. This we shall denote by 
4 = log^Sl and x := logo a' : the letters log being an abbreviation of 
logarithm, and the underwritten figure being the base. 

Examples. lO^ = 1000 3 = logiolOOO 

If a* = y a: = logoy 

If p^ = l^z y = logp(]-z) 

To construct a system of logarithms to a given base, say 10, we 
must solve the series of equations 

lO' = 1 10' = 2 10* = 3 10* = 4 &c. 

and find the value of x in each. This can, generally speaking, only 
be done by approximation : that is, the logarithm is generally incom- 
mensurable with the unit. By saying, then, that log|o2 =:-30103, we 
mean that 

lO-*®^^^ = 2 very nearlt/, or ^"ow^jQaoioa _ 2 nearly, 

and that a fraction k can be found such that 10* shall be as near to 2 
as we please, to which fraction *30103 is an approximation. 

[In all the following theorems one base is supposed, namely, a.] 

Theorem I. Whatever the base may be, the logarithm of \ is 0, 
This is evidently another way of expressing a°^l, and may be 
written logo 1=0. 

* From Xoyc/v u^ififMs, the number of the ratios, an idea derived from 
an old method of constructiug logarithms, which cannot be here explained. 




ToEOREM II. The logarithm of the hate itself is 1. This is 
contained in a^ := a, and is expressed thus : loga a ^ 1 

Theorem III. The logarithms ofy and - are of different signs, 
but equal numerical value. For if ^ := a' or x ss. ^o^y, we have 
- ^ or-* or — or := log©-; that is, loga- = — loga^. 

Theorem IV. If ?l be the base, and a number or fraction lie 
between a™ and a°, the logarithm of that number or fraction lies 
between m and n. 

For if a*, the number, lie between a»» and a*» ; then x, the 
logarithm^ lies between m and n (see page 89). 

Base 10. 

Number , Logarithm 

has its 
.between between 

1 and 10 



10 and 100 

1 and 


100 and 1000 

2 and 




1 and ^^ 


— 1 

10 ^"^ 100 

— 1 and 


1 , 1 

100 ^" 1000 

— 2 and 




Base -. 

Number Logarithm 

between "*^ ^^^ between 

1 and - 


- and - 

2 4 

4 8 


1 and 2 

2 and 4 

4 and 8 



and 3 
and —1 

—1 and —2 

—2 and —3 

Theorem V. The logarithm of a product is equal to the sum of 
the logarithms of the factors. Let a be the base, and j9, q, and r, the 
logarithms of P, Q, and R. Then 

P = aP Q = a« R = a*- 

PQR = a^+«+'' or log(PQR) =/> + ?+ r = log P + log Q -flog R 

Theorem VI. The logarithm of a ratio, quotient, or fraction, is 
the difference of the logarithms of the antecedent and consequent, 
dividend and divisor, or numerator and denominAtoi: 


P P 

For — = a^^^ or log ^ = p—^. = ^^g P^^og Q 

Theorem VII. The logarithm ofF^ is found by multiplying the 
logarithm of F by m ; that is, if P =: aP^ P"* := a^P, or log P»» = 
mp = m log P. 

Theorem VIII. A negative number has no arithmetical logarithm : 
nor is a system of logarithms with a negative base within the limits 
of algebra as hitherto considered. Tliis is rather a definition than a 
theorem,* and it amounts to this : on account of certain anomalies, 
of which the explanation cannot yet be understood by the student, 
we are obliged to defer the consideration of the logarithms of 
negative quantities, and all logarithms of positive quantities, except 
only those which have arithmetical meaning. For the equation 
a* := 6 has not been proved to have only one solution, though it has 
only one arithmetical solution. 

Theorem IX. The logarithm ofO is infinite; by which we mean 
that as y diminishes without limit, its logarithm (being always 
negative on one side or other of 1), increases numerically without 

limit. To diminish f -J without limit, x must undergo numerical 

increase without limit, and be positive : to diminish 2' without 
limit, X must also undergo numerical increase without limit, being 
negative. Hence (page 156) the meaning of the proposition stated ; 
and it must be observed that the symbol oc has either sign in 

algebra, which may be explained as follows. If 3^ = -, then y and x 


must have the same sign : if x diminish without limit, it approaches 
a form (0) in which it has no sign, being the limiting boundary 
between positive and negative quantity. Consequently y, which 
increases without limit, approaches a similar boundary; for since ^ 

has the same sign as -, if there be any form of x in which either sign 

may be supposed, the same may be supposed for y. But such a 
comprehension of this point as can be derived from frequent instances 
must be reserved until the student has seen more examples of the 
application of algebra to geometry. 

* Being usually considered as a theorem, we have stated it as such. 


The following examples will illustrate the preceding theorems. 
xxl = X loga + logl = logo? (logl = 0) 
x^ = X 1 X logo? = logo? 

logaax = logaX + logatt = logaX-\-l 

\ogxVy = log a:+logy* = loga:+-logy 
log j^ = log x^ I logy -log;? -2 logy 

log (^)''= -l{loga:-f31ogy-log;?-(-l)logy} ^ 

= — logar— 31ogy-f logp— logy 

We shall now proceed to the series connected with logarithms. 
In page 218, we have proved the following theorem for all values 
of n and Xy 

In which both series, being forms of (l + -} > will (page 210) be 

convergent whenever - is less than 1, or n greater than 1. Now, 


let n increase without limit, in which case the limit of 
l_i i_Li_l 

and that of 

1 12 

or— - X X X* *.3 

1 +^+ -2^ + -2^ -3^ + ^^- ^^ 1 +^+ T "^ 2"T3 + *'''• 

But 1 + 1 + -+ &c. has been calculated approximately in page 

183, found to be 271828182 , and called e. Therefore, 

(page 183), 

g'= \+x+- + — ^ + &c. 

Consequently, if i he the base, the number which has the logarithm x 

is 1 + a: H 1- &c. Hence a system of logarithms, having e for 

the base, being arrived at in the course of investigation, has been 


called the natural system of logarithms ; it is also called the Naperian* 
system of logarithms, and the hyperholic\ system of logarithms. 

And here let it be remembered, that in algebraical analysis^ the 
letters log, by themselves, imply the logarithm to the base c, any except 
tion being always specially mentioned. 

The preceding equation being always true, we have 

But 6** is («*)*; and if we take k to be the logarithm (base c) of a, 
we have ^ ^:s a, and 

(g*)* or a* = ] +(loga)a:+ i-^Y — + 2.3 "*" 

From which we find that will, if or be diminished without 

limit, have the limit log a. 

But (page 218) we have already found the series for this limit; 

logo = (a_l)- l(a-l)« + i(a-l)»- &c. 

or, if a =: 1 -f- bf we have 

log(l+ft) = J-|'+^-8cc (1) 

Substitute — b for b, and we have 

2 " 3 

log(l-6)= -6-.^*-^'-8w5 (2) 

Subtract (2) from (1) flog (1 + 6) — log (1—6) =log ({3^)]] 
log(S) = 2{6+f +3 +&0.} .... (3) 

, ^ 1 + 6 l + x .... . 1 

let - — J- := which gives b := - — — • 

1—6 X ® 2j:+1 

* John Napier, commonly called Lord Napier, though not a peer, 
or otherwise entitled to the appellation of lord than in the way in which 
many landed proprietors are lords (of manors), was born in 15.50, at his 
lordship of Merchiston, near Edinburgh, and died in 1617. He published 
the invention of logarithms (his method also leading him to the natural 
system) in 16 14. 

t So called from a supposed analogy with the curve called the 
hyperbola, but which analogy belongs equally to all systems of logarithms. 



l^g rzi = ^^S ^-^ = l^g (1 +^) -logo: 
The last series gives 

X = 21og3 = log2+2{i + 1^ + 1 3^5+^ 
X = 3 log4 = log3 + 2{l + 1 3I3 + 1 ji, + &c.} 
:r = 4 log5 = log4+2{l + 1^ + I35L5 + &*.} 

Thus the logarithms of whole numbers may be successively cal- . 
culared with tolerable readiness, as the first example (which contains 
the least convergent series), here given at length to ten places (that is 
to eleven for the sake of accuracy), will shew. 
















































log 2 = '69314718056 very nearly. 



By this means the logarithm of 3 may be found from that of 2, 
that of 4 from that of 3, and so up to any given whole number. It 
would be desirable,* as an exercise of arithmetic, that the student 
should calculate them up to 10 inclusive; the results of which (to 
eight places) would be as follows : 

log 1 = 0-00000000 log 6 = 1-79175947 

log 2 = 0-69314718 log 7 = 1-94591015 

log 3 = 1-09861229 log 8 = 2-07944154 

log 4 = 1-38629436 log 9 = 2-19722458 

log 5 = 1-60943791 log 10 = 2-30258509 

But the series need be employed only for prime numbers, and the 
first tables of logarithms were thus constructed, as follows. Suppose 
the logarithm of 59 to be required, or of 58 +1. Now 58 is 2 x 29, 
both factors being prime numbers; if, then, we have the logarithms 
of 2 and 29, we have that of 58 from the equation 

log 68 = log 2+ log 29 

and that of 59 from 

log 59 = log68+2{^-L + 1 _i_ + &C.} 

Beginning, then, with log 2, we have the following : 

log 2 = a given series log 6 = log 3 + log 2 

log 3 =: log 2 + a given series log 7 = log 6 + a given series 

log 4 = 2 log 2 log 8 = 3 log 2 

log 5 = log 4 -{- B. given series log 9 = 2 log 3 

log 10 = log 2 -|- log 5 ; and so on. 

♦ HaUey (PhiL Trans. 1695) thus expresses himself, after having 
described the preceding method. '* If the cariosity of any gentleman 
that has leisure, would prompt him to undertake to do the logarithms of 
all prime numbers under a hundred thousand to twenty-five or thirty 
places of figures, I dare assure him that the facility of this method will 
invite him thereto ; nor can any thing more easy be desired." Without 
insisting upon any thing that would take up so much of a gentleman's 
leisure as the preceding, I should strongly recommend the student 
always to work one example of every moderately convergent series 
which occurs. 



In page 219^ we proved independently that 
a_l_l(2r_l)«+&c. =-l(a"-l-i(3^-l)«+ «tc.) 
wliich isy as we now see, the same thing as 

losz = — log;2'* 
and in it we also find further elucidation of the equation, 

Umit of i^!l=i = 2r-l— 1(0— 1)«+ «cc. = logz. 

The diminution of m without limit, or the supposing of m to be a 
smaller and smaller fraction, implies the extraction of higher and 
higher roots of z. By extracting a sufficiently high root of z, we can 
bring z"^ as near to 1 as we please, or make Z^ — 1 as small as we 
please; that is (page 187) s'" — 1 may be made as nearly equal to 
the sum of the whole series as we please. It was from this principle, 
by continual extraction of the square root of z, that logarithms were 
once calculated, by means of the formula 

logz = {^^^^-^^ l) X 


very nearly,* the number named being 2^', equivalent to 47 extractions 
of the square root. 

The logarithm of any whole number being found, as in the last 
page, that of a fraction can then be found by the subtraction of the 
logarithm of the denominator from that of the numerator. 

We also notice the following result : when x is a large number, 

log(ar -f- 1) = log 4?+ 5jq-j nearly (page 227). 

The rest of this subject will be reserved for the next chapter, 
on the practical use of logarithms in shortening arithmetical com- 
putations. We now proceed with some uses of the preceding series. 

Lemma. IffQc) be such a function of jr, thaty*(j:-f-y) can be 
expanded in a series of the form 

Ao-hAiy + Aif-h &c, 

* Halley, in the memoir already cited. Each square root was ex- 
tracted to 14 places, 



where Aq, A^, &c. are functions of x onfy, then 

/(a + bVT-i) +fia - J V'^D 

treated by common rules, will always represent a quantity, either 
positive or negative, that is, all purely symbolical or impossible* 
quantities disappear ; while, on the other hand, 

/(a + 61/3T) _/(« _ ftl/TT) 

will be of the form (possible quantity) x ^ — 1 

Write the value ofJ'{s-\-f/) and afterwards that of /(x — y)y 
made by changing the sign of ^; 

fix+y) = Ao + A,y + A2y+A3y^+ &c. 
fix^y) = Ao-Aiy+A2y«-A3y3+ &c. 
ivom which we find, 

f(^ + y)'hf(x^y) = 2Ao +2A2y«+2A4y*+ &c. 
/(a:H-y)-/(x-y) = 2A,y+2A3y«+ 2A53/^-f- &c. 

for X write a, whence A^, A,, &c. become functions of a only ; for 
y write b >/ — 1, that is, suppose 

y = 6VCT y^ = i^l/HT 

/= -J3l/ZTx 6V=1 j^= 68 &c. 

= -.64x_i= i* 


/(a +i VHT) +/(a_5 v'lIT) = 2 Ao-2 A^ft* +2 A*^*- &c. 

which is a possible quantity : and 

/•(rt+ jV':rr)«/(a-6V'^)==2Ai6l/3l--2A36^vCT+ ^ 

= {2Ai6-2A36*+ &c.| VCT 

which is (a possible quantity) x \/— 1. 

* We shall now begin to make use of this common phrase : to the 
•tudent it must mean " impossible till further explained/* See page 110. 


EX.MP..S. L_+ ' 2« 

o + Jn/—! 0—6^—1 a' + b* 

(a+J^/— l)»_(a_JV'=T)» = 

As the preceding theorem is true for all values of a, it is true 
when fl = ; that ii, for 

Examples. Let n be a positive whole number. 

r + 2 6" when n is evenly even.* 
(61/171)'*+ (— 6 vCn)'* = < when n is odd. 

[^ — 2 6" when n is oddly even.f 

(0 when n is even. 
2 6- n/Z:! when n is 1, 5, 9, &c. 
— 2 6" \/^^ when, n is 3, 7, 
11, &c. 

If we apply the same process to €* ^^ and c"* , we find 

2 2.3 2.3.4 

2 2.3 2.3.4 

i^ = i-.^+--4— - ^ . . . ^ +&c (A) 

2 2 2 .3.4 ^ ^ 

€* '''^ — e-' "^^ x» ;r» ,. 

2>/Z4 2.3^ ^ ^ 

We have left the use of the symbol n/— 1, in this work, to be 
justified by experience only (see page 111); we have now an oppor- 

* Divisible by 4 ; 4, 8, 12, &c. are evenly even, 
f Divisible by 2, but not by 4 j as f , 6, 10, &c. 


tunity of examiDing the result of a loDg series of deduction, with a 
view of ascertaining how far we shall produce consistent results. 
The algebraical sign of equality is placed between the two sides of 
the preceding equations ; the question is, Will any relations we may 
discover to exist between the two first sides also exist between the 
two ucond sides ? 

£ "r ' 

Let = 2 ^ called ^X 

and 7== be called -i^X 

we have then 

{(ftxj = J 

(4,x)' = ' -^' ^, +' 

{fxf + (>|.ar)* = t^— -J = ,• = 1 

(pa;)* - (.^a;)* = ' ^ = <f (2x) 

of which three relations, namely, 

2(pxX'>^x = 4(2a:) 

it is asked, are they true of the second sides of (A) and (B) ? 

The multiplication of a series by itself will be found to amount to 
using the following rule : square each term, and multiply all that 
follow by twice that term. Thus, the square of the series in (A) is 

+2»- 271:4 +««=• 

+ &c. 


that of the series in (B) is 

^- 3 +37*75 -*'<=• 

— &C. 

The first square increased by the second is 

= !+{ } -{ } +&C. 

The first diminished by the second is 
•* ^l3.4^2«^3i^ ^2.3.4 ^3. 4X5 ^2«3«)**' ^ 

2 ^ 2.3.4 ^ 

The third relation may be similarly verified by multiplication. 
The following results may also be proved in the same way both of 
the first and second sides of equations (A) and (B) 

(f>{x+y) = (f>x<py'-^x-^y 4(a:+y) = '^x<py-\-(f>x^y 
^(^— y) = <px(py + '^x^y -^(x^y) = -vl/orpy— ^ar-^/y 

Let c*'*^-* be called d, and let -^-- be called a; ^9 then (equations 


A and B) 

, T- = -^ = yX or ^-TTT S= V — 1 YX 

whence p* = -^^ and, page 226, 

1— V— l;t;J^ 

But ^=: g2*V^ QP logp* = 2a; V^^l, therefore 



The series A and B are always convergeDt, as may be proved by 

the test in page 1 85 ; but the convergency may be made to begin at 

any term, however distant, by making x sufficiently great. Thus, if 

X were 1000, the first series would not begin to converge before 

the term 


<6«0«4«0 •••••••••• <60O t <w04 

but, notwithstanding the magnitude of the first terms of the series 
(which it must be remembered, are alternately added and subtracted) 
the actual value of ^j: and ^x can never exceed 1 ; for if either were 
arithmetically greater than 1 (positive or negative), the equation 
(0 x)' -\- (ypx)* ^ 1 could not be true. 

In trigonometry, the properties of the preceding series are con- 
nected with geometry in the following way. Let a circle be drawn ; 

and from A let the point B set out until it has described an arc equal 
in length to x times the radius OA, going round the circle again if 
necessary. Draw BD perpendicular to CA; then it will be proved 
that BD is the fi'action ^j: of O A, and OD the fraction ^j^ of OA. 

Exercise 1. By means of the equation 
,- s= X, prove the equation 

3. Prove the following, 

j-xv^^ par- V^ 4-0; 
(pa;+ Vdl4'a:)'" = (p(mx) + V^'^imx) 





The natural system of logarithms, already explained, has ibis defect 
as an instrument of calculation, tliat there is no method of finding 
the logarithm of a fraction more simple than subtracting the loga- 
rithm of the denominator from that of the numerator. For instance, 
the logarithm of *3 can not be more shortly found than by taking 

log 3 — log 10 or 1-09861229 — 2-30258509 = — 1-20397280 

We proceed to find another system in which there shall be some 
more obvious connexion between the logarithm of 3 and those of 
-3, -03, &c. 

The fundamental definition of log«;r gives 


ar = a a = a; * 

We have then x = a ^ o 

But b^a'""' .'. X = a"^''"^'' 

Similarly x = ^"'^•'^"^•^^"e" 

log« e log* c log« b logb X 

= a 

which last result may be also thus verified : 

loga e log« C lege b logft X ( ^^STa «llog« C loge b logft X 

a = la 3 

log« c log« b log6 s logc b \0%h X ,logft X 

= e ^ c =6=a; 

Now, there is but one arithmetical exponent which applied to a 
will give x; for, if possible, let there be two, p and g, and let 
aP ^ a:, a? = x, whence aP = fl«, and aP— «= 1 ; therefore j? — y = 0, 
or p = g, that is, p and g do not difier, as was supposed. Hence, 


loga X loga b log& X 

a? = a = a , we have 


log;, a; = loga* log^x = logaC lege 6 log^x 8cc. 

these results may be remembered by means of the identical equations 

X xb X X b c 

a "^ b a a b c a 

g;iving the following theorem : the series of equatiom 

a a c a e c afec „ 

b c b e c b J" e c b 

remains tme, if for each fraction be substituted the logarithm of its 
numerator to the denominator as a base. 

As an example, by means of 
T - = - we remember that log^a loga6 = logout = 1 

We also have 

or; to convert a given system of logarithms having the base a, into 
another which shall have the base b, divide every logarithm given by 
the given logarithm of b. 

Seeing that in practice it is convenient to reduce all fractions to 
decimal fractions, the base chosen should be one in which the 
logarithms of 10, 100, &c. are whole numbers, that is, it should be 
10. For in that case we have 

loglO = 1, loglOO = 2, loglOOO = 3, &c. 

And if we call/) the logarithm of any number, say 25, we have 

log 2-5 = log25— log 10 = ;?— 1 
log -25 = log26-log 100 = p-2 
log -026 = log26— log 1000 = p—3 &c. 

log 260 = log26-hlog 10 == p + 1 
log2500 = log26-hlog 100 =p + 2 &c. 

So that, when the base is ten, any alteration of the place of the 
decimal point in the number requires only the addition or sub- 
traction of a whole number from the logarithm. 



The system of logarithms to the base 10 U deduced from the 
natural system by the following equation, 

•<«.-^ = Itro = JlSsS = '°S'^ X -4342944819 

This system oriogarilhms is called the common, tabular, decimal, 

or Brigg't sjaten, and -43429 .... is called its moduluM, aod gene- 

rally l-i-1og, a, or logs t is called the modulus of the system whose 

All the logarithm! in the reminder of this cAopter art c 

The folloning aie examples of the arrangements of some tables 
of logarithms, for the purpose of eKpIaiiiing how to find the logarithms 
of given numbers, or vice veri6. 

1. Lalaade.' 




























S. Sherwin, Hutton, Babbage.-\ 





















































































* Tabtei dt Loptrithtaa, &c. pat J4rame de li Lahdi, Edition slirf- 
Otyp«, pw FiHMiH Didot: Paris, chez Finnio Didot, &o. Rue Jacob, 
No. 34, 1805 (tirage de 1831). This work ia sufficient for most pur- 
poses, but those who order it should remember to insist on having one of 
the later liroga. 

t All these works are well known in this country. The first (an old 
work) is fieqoently to be found with second-hand booksellers. Hie last 
two oan b« obtained in the nsnsl way fron any bookaeller. 



A logarithm usually consists of a whole number, followed by a 
decimal fraction, both or either of which may be negative : but in 
fviot^ tables nothing is put down but positive decimal fractions. We 
proceed to shew how this arrangement includes all cases. 

1. Take a negative logarithm, such as — 3*16804, which is 
— 3 — -16804, or — 4 +(1 —-16804), or —4+ -83196. This is 
usually written 4*83196, in which the negative sign over the four 
means that that figure only is negative. [According to analogy, 13 
would mean — 10 + 3, or — 7 ; 136 would mean 106 — 30, or 76.] 
Thus, every negative logarithm may be so converted as to have a 
positive decimal part. 

2. When the number corresponding to the decimal part is known, 
that corresponding to the whole can be immediately found by the 
following table, 

^ or log 10-3 ^ _3 





'^g To 

or log 10-* = -2 
or log 10-^ = —1 

log 10 or log 101= 1 
log 100 or log 102 = 2 

log 1000 or log 10* = 3 

Thus, the number to *30103 is 2 very nearly ; that is, -30103 

= log 2 : therefore, 1-30103 = log 10 + log2 = log 20, 1-30103 = 

1 2 

log— -h log 2 = log— = log -2, and so on. 

If D simply stand for a decimal fraction less than unity, we have 
(page 223) the following table : 

The log. of a No. 
lying between 

Must lie 

And must there- 
fore be of the form 

Instances of 
such numbers 

1 J 1 
1000 ^" 100 

—3 and —2 

— 3 4- D 



100 ^"^ 10 

—2 and — 1 

— 2 + D 



To "°^ ' 

—1 and 




1 and 10 

and 1 

-h D 



10 and 100 

1 and 2 

1 +D 



100 and 1000 

2 and 3 


2 -+-D 


159* 108 



Definition, The whole part of a common logarithm, whether 
positive or negative, is called the characteristic of the logarithm. 
The decimal part is called the mantissa,* We now give some 
theorems which obviously follow from what precedes. 

1. No alteration in the place of the decimal point (in which is 
included the annexation of ciphers to a whole number) alters the 
mantissa of the logarithm, but only the characteristic, 

2. When the decimal point of the number is preceded by signi- 
ficant figures, the characteristic of the logarithm is positive, and a 
unit less than the number of these figures. Thus, the logarithm of 
12345-67 is 4 + mantissa ; that of 6*9 is -(- mantissa, 

3. When the decimal point of the number is not preceded by 
significant figures, the characteristic of its logarithm is negative, and 
a unit more than the number of ciphers which precedes the first 
significant figure. Thus the logarithm of '00083 is ^-^^^ mantissa, 
that of '83 is — \'\' mantissa, 

Lalande's table first mentioned gives the characteristic, on the 
supposition that the number mentioned is a whole number; thus, the 
logarithm of 1081 (as given in the preceding specimen) is 3*03383. 
But, to take the logarithm of 1*081 from this table, the mantissa 
'03383 is all that must be taken, and the characteristic applied. 
Thus, the logarithm of 1*081 is 003383, and, from the rule laid 
down, we have the following table : 

log 1081000 




1-081 = 0*03383 

log 108100 




•1081 = r-03383 

log 10810 




•01081 = 2-03383 

log 1081 




•001081 = 3-03383 

log 108*1 




•0001081 =s 4-03383 

log 10-81 




00001081 = 5-03383 

The second table gives the logarithms of numbers of five places 
of figures. From the first table we might have found the logarithms 
of 5153 and 5154, or (the diff*erence being only in the characteristic) 
of 51530 and 51540; from the second specimen we can find the 

* This word is now seldom used, though there is not another single 
word which means the same thing. 



logarithms of 51531, 51532, .... 51539, intermediate to the two 
numbers j ust cited . 

. Now we must observe, that when a figure is changed in a 
number, the first figure which changes in the logarithm will be nearer 
to or further from the left hand, according to the figure changed in the 
number. This amounts to saying that the smaller (in proportion to 
the whole) the change of the number, the smaller the change in the 
logarithm, and is shewn by the following theorem. Since (pages 227, 
237) the common logarithm of 1 + ' is (M =r -43429 . . . •) 

logO +x) = logj:+2M ( 

+ 1 


2x + l • 3 (2jr-+-iy 

+ &C.) 

the greater x is, the less the addition to log x by which log (j: +1) is 
formed. The following instances will ■ illustrate this, in which the 
figure undergoing change is marked with an accent. The columns 
succeeding shew, 1st. By how much of itself the number is changed ; 
2d. By how much of a unit the logarithm is changed (nearly). 

'.log 2' 





log 10' 
I log 11' 





' log 100' 
log 101' 





' log 1000' 
log 1001' 





log 10000' 
' . log 10001' 





flog 100000' 
Clog 100001' 





Proportion to 

the whole of the 

change in the 





in the 




















Hence, roughly speaking, the absolute change in the mantissa of 
the logarithm is something less than one half of the relative change 
in the number. Let the student try to ascertain this from the series 
given above. Thus, if a number increase by its thousandth part, 


the increase in the logarithm is less than the absolute fraction , 

and so on. Hence we can ascertain with sufficient precision to how 
many places of logarithmic figures it will be necessary to carry any 
table. Let us suppose, for instance, our table is to give every 

number of five places, from 10000 to 99999. At the end of the 


table the relative increase of the number is about r, the 


absolute increase of the logarithm is therefore about -— -, or 

^ 250000 

•000004. Consequently six places of logarithmic figures are abso- 
lately necessary. With less than six places distinction would b^ 
lost; for instance, 

log 99846 = 5-9993307 

log 99847 = 5-9993350 

which only differ in the sixth place. In the first part of the table, 


where the relative increase is little more than r— --,, the absolute 


increase of the logarithm is nearly -00004, or five places only would 
be sufficient. But the tables must, for reasons of practical con- 
venience, be of the same number of figures throughout, and, therefore, 
must be at least of six places of figures. 

In the second specimen,^ve places of figures in the number are 
accompanied by seven places of figures in the logarithms. But, as 
the three first places of the logarithm continue the same for some 
time, even in the most changeable part of the table, they are placed 
by themselves in the first column, at the point where a change takes 
place : which saves much room, but is subject to this inconvenience, 
that as a change in the third figure of the logarithm will seldom take 
place exactly at the beginning of a line, the new third figure cannot 
be shewn till after it has really made its first appearance. The fol- 
lowing instances, taken out of the second specimen, will shew both 
the arrangement of the tables and this new difficulty, better than ny 
verbal explanation. 





Mant of the log. 


Mant. of the log 


•711 9759 


•712 0264* 


•711 9843 


•712 0349* 


•711 9927 


•712 0433* 


•712 0011* 


•712 0517* 


•712 0096* 


•712 0601 


•712 0180* 


•712 0686 

In this list it will be observed tliat each logarithm differs from 
the preceding either by -0000083, 0000084, or -0000085. In fact, 
the whole diflR?rence between the logarithms of 51520 and 51520+10 
is '0000842, giving for each increase of a unit an average increase 
of -0000084. We have then, at and near 51520, the following 
equations : 

log (5 1 520+1) = log 51520+ -0000084 
log (51520 +2) = log 51520 +'0000084 x 2 

or, if h be not greater than 10, 

log (51520 + A) = log 51520 + -0000084 X ^ (A) 

or, for a small part of the tables, the logarithms of numbers increasing 
by a unit increase in arithmetical progression very nearly. Now 
(M being -43429 .... pages 226 and 237), 


= M - very nearly, when - is small (page 187) 

or log(a? + A) = logo? + M-, very nearly, 

an equation of the same form as (A) ; whence it follows that (A), 
which is nearly true when h is 10, is even still more nearly true 
when A is a fraction of a unit. Hence we have 

In this case x = 51620 and ^ or *i51?£l£ = .0000084 

X 51520 

* In all these, the first three figures of the mantissa must be looked 
for below. There are various devices in different tables for reminding 
the reader of this, which we need not explain, as tbey are evident on 


log51260i = log51520 + -0000084 x i 
log51250-36 = log51520 + -0000084 x -36 

The column marked Diff'. (for difference) is meant to expedite 
the multiplication which the last equation shews will become neces- 
sary when the logarithm of six or seven places is sought. It consists 
of the tenths of 84 to the nearest whole number : thus, 

one tenth of 84 is 8'4, nearest whole number g 

two tenths .,.. 16'8, 17 

three 26-2, 26 

and so on. The hundredths of 84 may be got by striking off one 
figure from the corresponding tenths (adding 1 where the figure 
struck out is 5 or upwards), thus, 

one hundredth of 84 is about *8, nearest whole number 1 

two hundredths 1*7, 2 

three 2*5, 3 

and 90 on. Hence by inspection of the column of differences we can 
immediately determine the tenths or hundredths of the difference in 
question. And now let us determine the logarithm of 51*53946 from 
the second specimen in page 237. 

The mantissa is the same as that of the logarithm of 51539*46 and 

log(51639 +-46) = log61639 +-0000084 x -46 

= log51539 + -0000084 {± + 4) 


•0000084 (± + 4) = -0000001 (± x 84 + ^ x 84) 

(from the table) = -0000001 (34 + 6) 

But multiplying a whole number by *1, *01, '001, &c. is the 
same as removing its unit*s place to the first, second, third, &c. place 
of decimals : from which we have 

Mantissa of log 51539 from the table = '7121360 

Addition on account of the 4 which follows the 9 = 34 

Ditto on account of the 6 which follows the 4 =s 5 

Sum -7121399 


This sum is the mantissa of the logarithm of 51539*46, which has 
tlie same mantissa as that of 51-53946; therefore, taking the proper 
characteristic for the last, we have 

log51-53946 = 1-7121399 

The following are other instances derived from the same rule, and 
falling within the limits of the specimen. 

log -51527 





log 5 150008? 

log 5-1500 





log -5152748 


log 5-150008 


log 5152768000 log -00005 154899 

log5152700000 9-7120349 log000051548 5 7122118 

6 50 9 76 

8 7 9 8 

Iog5l52768000 9-7120406 log -000051 54899 5-7122202 

The inverse of this question is done as follows : Suppose it 
required to find from the table the number corresponding to the 

logarithm 1*7118366. Rejecting the characteristic we look in the 
table for the mantissa which is nearest to 7118366 (but below it). 
This we find to be 7118325, which is the logarithm of 51503; so 
that the number required is (as to its significant Bgures) within a 
unit of 51503. Let it be 51503 -f-^y then we know that A is to be so 
taken* that 

log (51603 + A) = 4-7118366 

But (page 242), 
log (51503 + h) = log 51503 + -0000084 X h very neariy. 

= 4-71 18325 + -0000084 x h 

* We neglect the characteristic, or rather we make our logarithm 
4-7118366. But au alteration of the characteristic is only an alteration 
of the place of the decimal point. 


__ 4-71 18366 — 4-7118325 _ -0000041 _ 41 
" -0000084 "^ -0000084 "" 84 

Now, from the table of diflferences we see that 

34 is ^ of 84 nearly, 

76 is ^ of 84, or 

7 is — — of 84 nearly; 

1 00 

so that 34-1-7 or 41 is — -+- -^ of 84; that is, ^ =a '48 = A: 

whence 51503 -f-A = 51503*48 and 

4-7118366 is the log of 5150384 
1-7118366 -5150384 

But the readiest method of putting this into practice is by making 
an inverse process to the method of finding a logarithm. We take an 
instance from another part of the table. 

What is the logarithm of 217483-6 ? 

log 21748 (*) 5-3374193 

3 60 

6 (t) 12 

log 217483-6 is 5-3374265 

What is the number whose logarithm is 5*3374265 ? 

Nearest log in table, belonging to N© 21748 '3374193 

Difference 72 

Nearest N'^ in table of Diff. belonging 

to N° 3 60 

Annex a cipher, because a figure was struck 

off, and therefore a figure (we do not 

know what) must be annexed. (See 

subsequent remark). Opposite to this 

we find 6. 

* Observe that we pat in the right characteristic at ouce. 
f Here, as before, we look opposite to six, and find 120, from which 
we strike off one figure. 





Tlierefore the number to the logarithm required is (making six 
places before the decimal 'point, on account of the characteristic 5) 

The student will better understand, by forming a number of 
logarithms and inverting each process: 1. Why a figure must be 
annexed to the last difference, which comes after the table of differ- 
ences has been used once. 2. Why that figure cannot be known. 
3. Why is most likely to be right. The above 12 might have been 
the result of any tabular difierence between 115 and 125, the mean 
number of which is 120. The following are examples of the process, 
without explanation : 

What are the numbers to the logarithms 2*1183214 and 





Ans. -01313171 




Ans. 92*22140 

The only unusual circumstance with which the student will now 
meet is in the multiplication and division of such quantities as 2*9. 
To multiply this by 5, carry to the negative term according to the 
common algebraical rule of addition. Thus, five times 9 are 45, set 
down 5, and carry 4; five times —2 are —10, and 4 are —6. The 
answer then is 6*5. Or 

6(-9-2) = 4-6-10 = 4-.10 + -6 = 6-5 

To divide 2*9 by 5, make the negative term divisible by 5, and 
correct the expression by a corresponding addition. Thus, 2*9 is 
5 + 3*9, and 

2*9 __ _5_ 3*9 

5 " 5 "^ 5 

14-. 78= 1-78 

The following are further instances : 


3-46 T-417 

8 10 

6)21-68 5)6-170 

4 613 .. 2-834 

The following is an example of the use of logarithms in multipli- 
cation and division, &c. What is the result of -5729578 multiplied 
by 20-62648, and the product divided by 7853982, after which the 
tenth root of the ninth power is taken ? or what is 


r -5729578x 20-62648 T 
I 7853982 J 

log -5729578 T-7581226 

log 20-62648 1-3144251 

log 7853982 


6 8950899 




57505 -7597056 

8 60 

5 38 

Answer -000005750585 

Tlie student should furnish himself with examples for practice by 
verifying such equations, for instance, as a (a + 6) = «• + a 6, where 
a and b n^y be any given numbers. The logarithms of a and a + 6 
being added together, and a^a-^-b) thus found, a' and ab should be 
separately found ; and if the whole be correct, the sum of the two 
last will be equal to the first. 


The following equations may be thus used : 
(a + ft) (a— 6) == a«— 6^ 

l/^ = V^x Vft 

(aft)" = a" X ft" 

Nothing but practice will enable the student to work correctly 
with logarithms, and most treatises on that subject contain detailed 
examples of all the cases which arise in practice. 





Page viii, line 9, The mistake alluded to is the saying tbat a multiplied 

by nothing is a, 

X, ... 11, omit the word a/ti;a^«. 

zii, ... 16, for that 1 contains, read that is, 1 contains. 

XX, ... 7i for shews, read shew. 

3, ... 4 from the end, /or added, read added, &c. 

21, ... 8 from the end, to on each share, add which he holds. 

24, ... 1 1, 12, and 14,/or 1988 and 6988, read 1998 and 6998. 

3d, ... 4, for as, read as often as. 

36, ... 20, for right, read left. 

47, ... 9, for (a + 6), read (a + c). 

60, ... 6, for 9, read A. 

89, ... 13 from the end, /or square root, reac^ square. 

103) Though the student omit what is said on the law of 

eontinuily^ he should not omit either this page, or 
what immediately relates to it in the next. 

109,- ... 4, for n/^, read V^, 

109, ... 10, /or d?i, read J^~^. 

131, ... 7) /o** 9) read r, in both places. 

131, ... 9 and 11, for g, read r, 

1 V 

144, ... 7 from the end, /or b + g^, read h + -^, 

Upper Gower Street. 





The Elements of Arithmetic. By Augustus De Morgan, 
Professor of Mathematics in University College, London. Third Edi- 
tion, royal 12mo, 4s. cloth. 

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 
moderate computation of the time thus 
bestowed, the pupil would be in no re- 
spect worse off, though he spent five hours 
on every page of this work. The method 
of proceeding which I should recommend, 
would be as follows : — Let the pupils be 
taught in classes, the master exphining 
the article as it stands in the work. Let 
the former, then, try the demonstration 
on some other numbers proposed by the 
master, which should be as simple as pos- 
sible. The very words of the book may 
be used, the figures being changed, and it 
will rarely be found that a learner Is 
capable of making the proper alterations, 
without understanding the reason. The 
experience of the master will suggest to 
him various methods of trying thu point. 
When the principle has been thus dis- 
cussed, let the rule be distinctly stated by 
the master, or some of the more intelli- 
gent of the pupils; and let some very 
simple example be worked at length. Tlie 
pupils may then be dismissed, to try the 
more complicated exercises with which 
the work will furnish them, or any others 
which may be proposed." — Preface. 

** Since the publication of the first edi- 
tion of this work, though its sale has 
sufficiently convinced me that there exists 
a disposition to introduce the principiet of 
arithmetic into schools, as well as the 
practice, I have often heard it remarked 
that it was a hard lxx)k for children. I 
never dared to suppose it would be other- 
wise. All who have been engaged in the 
education of youth are aware that it is a 
hard thing to make them think ; so hard, 
indeed, that masters had, within the last 
few years, almost universally abandoned 
the attempt, and taught them rules in- 
stead of principles ; by authority, instead 
of demonstration. 1 his system is now 
passing away ; and many preceptors may 
oe found who are of opinion that what- 
ever may be the additional trouble to 
themselves, their pupils should always be 
induced to reflect upon, and know the 
reason of, what they are doing. Such I 
would advise not to be discouraged by the 
failure of a first attempt to make the 
learner understand the principle of a rule. 
It is no exaggeration to say, that, under 
the present system, five years of a boy's 
life are partially spent in merely learning 
the rules contained in this treatise, and 
those, for the most part, in so imperfect a 


The Elements of Algebra, preliminary to the Differential 
Calculus, and tit for the higher classes of Schools in which the Principles 
of Arithmetic are taught. By Professor De Morgan. Second Edition, 
royal 12mo. 9s. cloth. 


Elements of Trigonometry and Trigonometrical Ana- 
LTSis, preliminary to the Differential Calculus : fit for those who have 
studied the Principles of Arithmetic and Algehra, and Six Books of 
Euclid. By Professor De Morgan. Royal 12mo. 9s. cloth. 


The Connexion of Number and Magnitude : an attempt to 
explain the Fifth Book of Euclid. By Professor De Morgan. Royal 
12mo. 4s. cloth. 

• • 

This Work it incktded in the Elemente qf Trigonometry, 


Lessons on Number, as given at a Pestalozzian School at Cheam, 
Surrey. By Charles Reiner. Second Edition. 

The Master's Manual. 12mo. 4*. 6d. cloth. > /« u i x 

The Scholar's Praxis. 12mo. 2s. bound. J ^^***^ separately.) 



/ • -. ' - ;l 


Published by Tdylor and Walton^ Upper Gower Street. 


Lessons ok Form, at given at a Pestalozzian School at Cheam, 
Surrey. By Charles Reiner, Author of '^ Lessons on Number.*' 
With numerous Diagrams. 12mo. 6s. doth. 


Principles of Oeome-try, familiarly illustrated, and applied to 
a variety of useful purposes. Designed for the Instructitm of Young 
Persons. By the late Rev. William Ritchie, LL.D., F.R.S. Second 
Edition, revised and enlarged. l2mo. with 150 Wood Cuts, 3«. 6d. cloth. 

** The practical applications which are added, must render the study very delightful 
to the young, since the Exercises on the Principles will be found as amusing as the 
ordinary sports of ch\]dhood."—Athetiaum, Sept. 28* 1833. 

<* Dr. Ritchie's little elementary work is excellently well adapted to its object. It is 
brief, plain, and full of all that is necessary ; curious and useful in its application; and 
beyoim any other of the kind now existent in its familiar and distinct explanation of 
some of the instruments re(|uired in the practical application of the principles laid down 
and demonstrated."— Spectof or. Sept 7* lt^> 

An Instrument for teaching Geometry, convertible into a 
Theodolite, Spirit Level, Hadley's Sextant, and Wollaston's goniometer, 
Las been prepared, and may be had of the publishers. Price 21. 12«. 6d. 


Principles of the Differential and Integral Calcu- 
Lus, familiarly Illustrated, and applied to a variety of Useful Purposes. 
Designed for the Instruction of Young Persons. By the late Rev. Wil- 
liam Ritchie, LL.D., F.R.S. With Wood Cuts. 12mo. 4s. 6d. cloth. 


The Elements of CSuclid, with a Commentary and Geometrical 
Exercises. By the Rev. Dion r si us Lardner, LL.D. Fifth Edition, 
8vo. 7«- boards. 


An Analytical Treatise on Plane and Spherical Tri- 
gonometry. By the Rev. Dionysius Lardner, LL.D. Second 
Edition. Corrected and improved. 8vo. I2s. cloth. 


An Elementary Treatise on the Differential and 
Integral Calculus. By the Rev. Dionysius Lardner, LL.D. 
8vo. 2ls. 



1. A System of Popular Geometry. Containing in a few- 
Lessons so much of the Elements of Euclid as is necessary and sul^cient 
for a riglit understanding of every Art and Science in its leading Truths 
and general Principles. By George Darley, A.B. Fourth Edition. 
4^. 6d. cloth. 


Elements of Al»stract Science are familiarised, illustrated, and rendered 
practically useful to the various purposes of Life, with numerous Cuts. 
4^. 6d. cloth. 

3. A System of Popular Algebra, with a Section on Propor- 
tions and Progressions. Third Edition. 45. 6d. cloth. 

4. A System of Popular Trigonometry, both Plain and 

Spherical ; with Popular Treatises on Logarithms, and the Application 

of Algebra to Geometry. Second Edition. 3^. Bd. cloth. 

*" For Students who only seek this limited knowledge of these Sciences, there are, 
peih ps, no Treatises which can be read with more advantage than Darle/s Popular 
Geometry and Algebra."— Lt&ran^ qf Ust^ul Knowledge, Artide ** Me<^anic8," 






^,^J 7^ 

3 6105 033 238 366 !_- 


ELEMENTS OF EUCLID, with a CommeDtarf and Geometricnl 

Eierdm s to which ire anneied a TreetUe on Solid Geometry, 

and Short Esaays on the Ancient Geometricul Aoaly ua and the 

Theory of TnairerMls. By Dr. Lakdnbb. Eighth Edition. 

8va, 7(. boards. 
DIAGRAMS IN WOOD, to iUaatrate Dr. Labdheb's Edclid. 

Solid Gaometry, Book I. The Set of 9 in ■ Box, price It. &d. 
PRINCIPLES OF GEOMETRY, famUUrly Illoetrated, and qiplied 

to a Tariety of naefnl purpoaes. Deaigned for the Inatniction 

of Young Peraona. Seiiand Edition, reiiaed and enlarged. 

By Dr. Ritchie. I2mo, with 150 Woodcnts, 3>. Bd. doth. 

into a Theodidite, Spirit Level, Hadley'a Seitant, and Wol- 

laston'a Goniometer. Price 21. 12>. Sd. 



Stanford University library 

Stanford, California 

Id order that others may use this hw 
please return it as soon as possible, I 
not later than the date due.